Parties, Ideologies, and Budgets: A Study of Budgetary Trade-offs

8/6/2019 Parties, Ideologies, and Budgets: A Study of Budgetary Trade-offs

http://slidepdf.com/reader/full/parties-ideologies-and-budgets-a-study-of-budgetary-trade-offs 1/236

P A R ~ I E S , IDEOLOGIES AND B U D G E ~ S : A STUDY

OF B U D G E ~ A R Y TRADE-OFFS

B y

Joe Todd Adams

D i s s e r t a t i o n

Submitted t o th e F ac ulty o f t h e

Graduate School o f Vanderbilt U n i v e r s i t y

in par t i a l f u l f i l l m e n t o f t h e r e q u i r e m e n t s

f o r t h e degree o f

DOCTOR OF PHILOSOPHY

in

Pol i t ica l Science

December, 1992

N a s h v i l l e , Tennessee

Date

'0-1 v -1'1 : > - 7 D - ~ ?

10-10 - t i l -



@ Copyright by Joe Todd Adams 1992All Rights Reserved



ACKNOWLEDGEMENTS

While doing this study I incurred many debts. I would

l ike to express my gratitude to several people who contrib-

uted thei r time, expertise and support.

For the i r contributions to the collect ion of data on

pol i t i ca l part ies , I would l ike to thank Professors Marina

Arbetman of Tulane Univer si ty , F ranc is G. C astles of the

Australian National Univers ity , Robert Cox o f th e U niversity

of Oklahoma, John Logue of Kent State Universi ty, Eric

Einhorn of the University of Massachusetts a t Amherst,

Charles F. Delzel le of Vanderbilt Universi ty, M. Donald

Hancock of Vanderb il t Univers ity, David Wilsford of Georgia

Ins t i tu te of Technology, and John Williams of Emory Univer-

s i ty . I would also l ike to thank Lieutenant Colonel Clayton

R. Newell and his s ta f f a t the Department of the Army's

Center for Military History for providing data on American

combat deaths.

I would l ike to thank Professor M. Donald Hancock for

h is encouragement and patience during the past three years.

I am especial ly indebted to Professor Jacek Kugler who

provided many useful criticisms and suggestions as well as

several motivating inquiries regarding my progress.

I owe many other intel lectual debts. I am part icularly

indebted to Professors William D. Berry and David Lowery for

i i



the i r work on conceptualizing budget t rade-offs . This study

i s bu i l t on the i r effor ts to provide a more f rui t ful ap

proach to budgetary analysis.

Most importantly, I would l ike to thank my wife, Drewe

Ann Knight, for her pat ience and understanding in l is tening

to my ideas as they developed. She endured far too many

nights alone while I worked. Her encouragement, pers is-

tence, and affect ion i s great ly treasured. Without her

support th is project would not have been completed.

i i i



TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS i i

LIST OF FIGURES v i

LIST OF TABLES .................•.......................v i i i

Chapter

I . TRENDS ANDISSUES IN

GOVERNMENT EXPENDITURES1

An Interest ing Question ..........•..............1

Issues and Recent Trends 9Economic Determinism .•.........................22

American Exceptionalism ....•...................24Party and Policy .....•.....•..•................26

I I . THEORY AND METHODS .....•.....•..•..•............. 30

Introduction .........•........•................ 30A Theory of Ideology .•...........•............. 31A Spatial Arrangement of Ideologies •..........• 33Ideology and Distr ibutive Just ice 43

Policy Character is t ics 48Strategic Opportunities 55A Theory of Tradeoff Behavior ......•........... 59

Americans: Right and Left ..•..•..•............. 62Conceptualiz ing Tradeoffs 69

The General Hypothesis 75

Data Collection and Measurement .............•..76

I nspecti ng t he Poli t ical Variables 87I nspecting t he Dependent Variables ...•......... 98

I I I . CROSS-NATIONAL COMPARISONS 106

Introduction ....•....•.............•.•........ 106

Cross-National Patterns 107Summary and Conclusion 113

IV, TRADE-OFFS IN THE UNITED STATES 116

Introduction 116Ideology, Structure and Process 121Hypotheses ......................•...........•. 135Control Variables ...........................•.137

iv



Preliminary Observations 139Defining the Variables 146

Specifying the Basic Model 153Explaining Trade-offs 174Surnrnary 181

V. SUMMARY AND CONCLUSION 183

Principal Findings 183Conclusion: A Peace Dividend? 185

Appendix

A. DATA SOURCES 190

B. QUESTIONNAIRE CONSTRUCTION 193C. MEASUREMENT: AN AGENDA FOR RESEARCH 194

D. RELATED FINDINGS 199War, UnemploYment and Trade-offs 199

BIBLIOGRAPHY 207

v



LIST OF FIGURES

Figure Page

1. Labor Strength and Percent Moderate Left 90

2. Labor Strength and Percent Moderate Right 91

3. Labor Strength and Anchored Ideological Scores 91

4. Labor Strength and Weighted Ideological Scores 92

5. Relationship between ADA Scores and Party Unity 96

6. Relationship between ADA Scores and Party Unity 97

7. Comparisons of T rade -o ff Measures 100

8. Comparisons of Data Sets for the Heal th Trade-off . . 101

9. Difference between Budget Shares 104

10. Difference between Budget Shares 104

11. Components of Defense Spending: Selected Items 121

12. Defense Outlays and Batt le Deaths 140

13. Defense Outlays and Batt le Deaths (vietnam) 142

14. Education as a Percentage of F ederal Outlays 143

15. Education versus Defense Outlays 148

16. Difference and Proportion Measures 149

17. Difference and Proportion Measures (1940-1986) 150

18. Comparisons of Domestic Trade-off Measures 152

19. Domestic Trade-offs and GNP 163

20 . Democratic Unity During the Vietnam War 166

21. Comparison of Reagan Years with Earl ie r Period 171

22. Unemployment and the Domestic Trade-off 200

v i



23. Model 2 Predictions of Unemployment 204

v i i



LIST OF TABLES

Table Page

1. Policy and Ideology 60

2. Left and Right in the Libera l Ideology 68

3. Corre la t ions with u.s. Budget Data 102

4. Cross-National Regression Resul t s 111

5. Average Pres ident ia l Success Rates in the Senate 130

6. Reagan's Success Rate in th e Senate 131

7 . Pres ident ia l Par ty and War: 1940-1990 164

8. Pres iden t ia l Party and the Domestic Trade-of f 165

9. Explaining the Domestic Trade-of f (Po l i t i cs Alone) . . 168

10. The Domestic Trade-of f (With Par t i t ioned Variance> .170

11. Explaining the Domestic Trade-off (The Basic Model) .173

12. Explaining the Health Trade-off 174

13. Explaining the Education Trade-off 175

14. The Education Trade-of f (With Par t i t ioned Variance) .176

15. Differences in Trade-offs : Before and Afte r 1981 178

16. Explaining the Civi l ian Research Trade-of f 180

17. Explaining UnemploYment: Mode l l 202

18. Explaining UnemploYment: Model 2 203

v i i i



CHAPTER I

'!'RENDS AND ISSUES IN GOVERNMEN'l'

EXPENDI'l'tJRES

An Interest ing Question

The Cold War has ended. With the dissolution of the

Soviet Union, the poss ibi l i ty of a so-called "peace div i -

dend" in the United States and abroad i s widely discussed.

The question raised by recent changes in the internat ional

context of superpower pol i t ics i s whether the f isca l re -

sources of advanced industr ia l ized democracies wil l be

redirected toward domestic purposes.

Perhaps there is no precedent for such a change, and

therefore, no sol id basis for predicting how the budget may

be sp l i t in the future. Conventional wisdom and a good deal

o f em pirical re sea rch suggest that budgetary distr ibut ions

are re la t ive ly stable in th e sho rt -r un . In fact , few people

have suggested that changes wil l occur rapidly. But the

ra t ionale for the exist ing dist r ibut ion of resources seems

to be fading and defense cut-backs have already begun in

some areas. Does th is mean that other areas of the budget

will benefit? I f so, which areas? Can we predict a t rade

off of th is kind?

1



2

Even during the Cold War public expenditures changed.

The most noticeable changes included a large expansion of

public services and welfare programs. Not only have bi l -

l ions been spent on defense, but much more has been spent on

the combined categories of heal th, education, and various

kinds of income maintenance programs. Ironical ly , the

al leged arms race was occurring at the same time as the

welfare s tate was approaching i t s zenith.

In effec t , the q ue stio n of whether a peace dividend

should be expected shares the same theoret ical ground as the

issue of the guns-versus-bu tte r t rade-off . Thus, the logi-

cal place to seek an answer to the question of how a poten-

t i a l peace dividend might be distr ibuted i s in the l i t e ra -

ture on budgetary t rade-offs .

The number of papers, ar t ic les and books on the subject

of the g ~ ~ s - v e r s u s - b u t t e r issue has increased markedly

during the l as t two decades. 1 Despite widespread in teres t

and a considerable amount of empirical analysis , theorists

do not agree on the elements of a guns-versus-butter trade-

off . In general , scholarly work on the subject has deal t

with two basic kinds of t rade-offs .

The f i r s t kind of t rade-off is constructed as a hypoth

es is tha t well-being, or quali ty of l i f e , i s affected by

1 See Steve Chan, "The Impact of Defense Spending onEconomic Performance ," Orbis (Summer 1985): 403-34.



3

mili tary expenditures. S tud ies of this kind of t rade-off

generally consider the effect of defense spend ing on the

economy. 2 Empirical studies on this topic suggest tha t the

influence of defense spending on the economy is not uniform

across a l l countries. Apparently, effects dif fer according

to a country 's level of development, the mixture of i t s

industrial base and whether i t s armaments are imported or

domestically produced. 3 The l i terature suggests tha t what

ever dividend might be achieved in so-cal led quali ty of l i f e

or economic performance i s clearly dependent on several

kinds of qualif icat ions and contingencies.

2 On this point there i s a considerable differencebetween developed and developing countries. For a very goodoverview, see Chan, "The Impact of Defense Spending," 40334. For background on the debate in the United States , seeSeymour Melman, Our Depleted Society (New York, Dell Publ ishing Co., 1965); Bruce M. Russett , What Pr;ce Vigilance?The Burdens of National Defense (New Haven: Yale Universi tyPress, 1970); Morton H. Halperin, Jacob A. Stockfish, andMurray Weidenbaum, The Pol i t ica l Economy of th e M i lit ar y -Industrial Complex, eds. Warren F. Ilchman and Joe S. Bain(Berkeley: Universi ty of California Press, 1973); SeymourMelman, The Permanent War Economy; American Capitalism inDecline (New York: Simon and Schuster, 1974); Bruce Russet,"Defense Expenditures and National Well-being," AmericanPol i t ica l Science Review 76 (September 1982): 767-77.

3 Cf. Chan, "Impact of Defense Spending;" David Dabelkoand James D. McCormick, "Opportuni ty Costs of Defense: SomeCrOSS-National Evidence," Journal of Peace Research 2(1977): 146-54; Saadet Deger, "Economic Development andDefense Expenditures," Economic Development and SocialChange 35 (October 1986): 178-96; Steve Chan, "DefenseBurden and Economic Growth: Unraveling the Taiwanese Enigma," American Pol i t ica l Science Review 82 (September 1988):913-20.



4This study focuses on a second kind of t rade-off: the

t rade-off between categories of government expenditures.

The most common example of th is kind of t rade-off is the

often hypothesized f iscal t rade-off between defense expendi-

tures and various categories of domestic expenditures.

Many studies that examine th is kind of t rade-off indi-

cate tha t the relat ionship between changes in defense expen

ditures and other kinds of expendi tu re s i s re la t ive ly

weak. 4 These studies suggest that expenditures for dif fer -

ent categories appear to be on relat ively independent t ra -

jectories , responding to different pressures and demands

arising from changing economic, demographic and diplomatic

conditions. Changes in international affai rs are frequently

credited with changes in defense expenditures. Shifts in

the demographic composition of the population and economic

performance engender shi f t s in the distr ibut ion of funds

across categories of domestic spending. s

4 Cf. Russett , "Defense Expenditures and NationalWell-being," 767-77; William Domke, Richard Eichenberg, andCatherine M. Kelleher, "The I l lusion of Choice: Defense andWelfare in Advanced . Industrial Democracies , 1948-1978," TheAmerican Pol i t ica l Science Review 77 (March 1983) :19-35; - - Alex Mintz, "Guns Versus Butter: A Disaggregated Analysis,"

American Pol i t ica l Science Review 83 (September 1989): 128593.

S For example, see James L. Clayton, "A Comparison ofDefense and Welfare Spending in the United States and theUnited Kingdom, 1946-1976," Journal of Sociology and SocialWelfare 4 (1978): 401-18; Glenn Palmer, "Alliance Poli t icsand Issue Areas: Determinants of Defense Spending," AmericanJournal of Pol i t ica l Science 34 (1990): 190-211 .



5

Occasionally, some scholars mix the issues of national

well-being and f iscal distributions in a more complex model

of t rade-offs . In a widely cited study, Bruce Russett

concluded an analysis of the impact of U.S. defense spending

on national well-being with an especial ly inciss ive apprais-

a l of f isca l t rends in the post-war era:

One remarkable fact about th is growth in federaleducation and health spending i s that i t continuedthrough several major periods of mili tary buildup.

Expenditures for education rose through the cold warrearmament and Korean War, and during a l l but two ofthe Vietnam War years. Federal spending for healthcontinued to r ise throughout the Vietnam period. Evenduring the four costly World War I I years from 1941through 1944, these civi l ian federal programs continuedto expand. Sacrifices surely were imposed, but theywere largely sacrifices imposed on current consumption,not on the kind o f long-run social investment requiredto build a healthier and better-educated populat ion.Indeed, the problems o f igno rance, i l l -heal th , andmalnutrit ion among many potent ia l draftees during WorldWar I I helped to convince even the most defense-mindedlegis la tors and off ic ia ls that the federal government

had to take remedial action.A second remarkable fact is that support for

inc reased federa l health and education programs wasreasonably non-part isan. Federal spending for educat ion, e.g . , the National Defense Education Act, went upsharply during the Eisenhower years and continued tor i se during a l l but one year of the Nixon and Fordadministrations. Federal health expend itures passedunscathed through the Nixon and Ford years. When theRepublican presidents fe l t compelled to raise defensereadiness, they kept the i nc reases within l imits andaccepted the taxes necessary to maintain social programs. 6

6 Russet t , "Defense Expenditures and National WellBeing," 776.



6

Russet t ' s observations reveal three dis t inct problems

in analyzing budgetary t rade-offs . Firs t , Russett res t r ic t s

his evaluation of t rade-off behavior to those instances

where the mili tary expenditures increase. Pro-domestic

t rade-offs are not seriously considered. Russett dismisses

such cases because:

Experience with reduc ti ons in m ilita ry spending i sre la t ive ly l imited, occurring in j u s t 16 or th e 39

years in our sample. [Evidence] suggests tha t there i s

some t rade-off in th ese y ea rs , and i n p rin ci pl e we

might analyze y ears w ith mili tary upswings in separateequations from those with downswings. However, thesamples--especial ly for years with mili tary down-swings--would be very small for adequate multivariateanalysis , and th e te chnic al problems of analyzingyearly data for years that were often not adjacentwould be formidable. 7

Russet t ' s remarks are reminiscent of many other schol-

ars who have writ ten on the issu e. A close reading of the

l i tera ture on th is subject reveals tha t there a re uns ta te dassumptions made about the way defense versus domestic

t rade-offs operate. I t appears tha t the majori ty of studies

posi t a zero-sum game where an increase in defense expendi -

tures occurs a t the expense of other programs. I f th is is

t rue, then an equally possible alternative tha t increases in

domestic expenditures occur a t the expense o f defense spend

ing should exis t . This poss ibi l i ty i s dismissed in

Russet t ' s study because the number of cases were too small

7 Russett , "Defense Expenditures," 775, n.7 ( i ta l icsadded) .



7

for s t a t i s t i ca l analysis . In most studies, the quest ion i s

not addressed at a l l because the concept of a t rade-off i s

i t se l f not clearly defined. I t is p re sent ed a s a hypothe-

sized negative relat ionship between the growth ra tes of two

or more categories of expendi tu re s. Such an approach does

not capture the ful l range of f isca l possibi l i t ies .

Second, the influence of internat ional factors on

budgetary behav ior i s not clear . The widely debated concept

of an arms race w ith i t s emphasis on escalat ion seems to add

very l i t t l e to our understanding of how budgetary outcomes

are determined. Both Soviet and American defense expendi -

tures increased during the per iod Russet t analyzes, yet many

categories of domestic expenditures did not apparently

suffer . Do changes in Soviet mili tary spending ef fec t

t rade-offs in federal outlays as im plied by the arms race

l i te ra ture?

Thi rd , according to Russet t , the succession of

Republican presidents appears to have had l i t t l e influence

on the budgetary changes. Unfortunately, Russet t does not

develop the kind of measures required to t es t adequat ely the

hypothesis that ' ·bipartisanship accounted for the growth in

domestic spending; i t is not included in his model.

Perhaps an adequate measure of partisan influence could

account for the apparently odd combination of events Russet

describes. Of eleven independent variables used in



8

Russet t ' s study only one dummy variable direct ly measured

the in fl uen ce o f a pol i t ical factor: the pres ident ' s party

af f i l i a t ion . The questio n o f whether part isan pol i t ics

influenced the outcome is a matter of part icu lar in teres t to

stu den ts of pol i t ics , but the l i te ra ture on the effect of

pol i t ics on f i sca l outcomes is inconclusive. A substant ial

l i terature suggests, as does Russett , that economic and

demographic factors account for most of the variance in

budgetary behavior.

In using a predominantly economic and demographic model

to explain trends in expendi tu res, Russe tt is not alone.

This i s the norm in scholarly work on government spending.

One might even say that th is general approach represents a

paradigm. This approach is accepted by a wide range of

scholars in several disciplines. In part , the analyt ical

approach used in Russet t ' s study stems from the kind of

research which preceded i t .

Unfortunately the conceptual and analytical methods em-

ployed in Russet t 's study, and in most of th e re se arc h on

th is issue, cannot adequa te ly address the t ra de -o ff i ss ue .

For a varie ty of reasons, most of the research in th is area

is simply misdirected. This study wil l employ an al terna-

t ive approach. Using direct measures of t rade-off outcomes,

I t es t the hypothesis that part ies have an effec t on t rade

offs in the United States and other advanced industr ial



9

democracies. A proper t es t of th is hypothesis might yield a

bet te r understanding of condi ti ons requ ir ed for obtaining a

peace dividend.

Issues and Recent Trends

Turning to the l i terature on government expenditures,

one f inds a profusion of studies claiming to explain why

expenditures change. Comparatively less research deals

di rec t ly with the issue of t rade-offs between specific kinds

of expenditures. Instead, there i s a preponderance of

l i terature dea li ng wit h the determinants of one or another

kind of expenditure, usually in isolat ion from other kinds

of spending. Findings in th is body of research are not

unanimous. In part , the dispari ty between findings can be

at t r ibuted to the scope of studies, differences in levels of

analysis , and the level of budget aggregation as well as the

methodological techniques employed. These factors cause

some dispar i t ies , but the l i terature also reveals a variety

of parochial disciplinary preoccupations.

Research on government spending i s the providence of

several academic discipl ines including economics, pol i t ical

science, sociology and public administration. Each disc i -

pl ine brings i t s own questions to the subject o f exp la in ing

the dynamics of th e pub lic purse. A fragmented and special-

ized l i terature on publ ic expenditures has emerged. As a



10

resu l t , there is l i t t l e common ground on what needs to be

explained.

For economists, the central questions concern the

l imits of government spending and i t s effec t on productiv-

i ty , in teres t rates, inflat ion, international trade, unem-

ploYment and the l ike . For t he s oc io log is t, the centra l

questions re la te to how government expenditures w ill a ffe ct

the quali ty of l i fe fo r various elements of the populace and

relat ions between classes of people. For the specia l is t in

publ ic admini st ra ti on and public finance, scholarly in te r -

es ts revolve around predict ing budget surpluses or defici t s

as well as measuring the eff ic iency of public ly produced

services . Pol i t ica l sc ient is ts have tended to focus on

budgetary process, pol i t i ca l accountabili ty, responsible

f isca l behavior, and the coherence of po li cy ob je c ti ve s. 8

8 The study of the budget process has a long heri tage,but remains t ied to legal and procedural modes of analysis .The centra l finding of th is body of research i s tha t budgetprocesses promote incremental t inkering, as opposed tocomprehensive budgeting. Cf., Avery Leiserson, "Coordination of Federal Budgeting and Appropriations Proceduresunder t he Legi sl at ive Reorganization Act of 1946," NationalTax Journal 1 (1948): 118-25; Charles E. Lindblom, "TheScience of Muddling Through," Publ ic Admin is tra tion Review19 (1959): 79-88; Richard F. Fenno, J r . , The Power of the

Purse: Appropriations Poli t ics in Congress (Boston: Li t t l eBrown and Company, 1966); Aaron Wildavsky, The Pol i t ics ofthe Budgetary Process (Boston: Li t t l e Brown, 1964); OttoDavis, M.A.H. Dempster, and Aaron Wildavsky, "Toward aPredictive Theory of Government Expenditures: U.S. DomesticAppropriations," Brit ish Journal of Pol i t ica l Science 4(October 1974): 419-52; Louis Fisher, Presidential SpendingPower (Pr ince ton: Pr inceton Unive rs ity P re ss , 1975); ThomasW. Wander, F. Ted Hebert, and Gary W. Copeland, eds. , Con-



11

Some of these concerns in tersect at certain points .

The question of growth in expenditures, what drives i t and

what l imits i t , is one area where the various disciplines

have converged. In part icular , at tent ion has been directed

a t the growth of government expenditures on those items most

closely ident i f ied with the so-called welfare s tate .

Most of th e re se arc h on domestic spending on welfare

and socia l services has focused on demographic and economic

determinants. Accounts of how domestic programs developed

f requent ly descr ibe the his tor ica l sequence of events or

conditions leading to the establishment of social programs

and the corresponding growth in th e pub lic sector for a

s ingle country or se t of countries sharing more or less

s imilar character i s t ics . 9 Other studies emphasize the

aressional Budaetina: Pol i t ics , Process and Power (Baltimore: Johns Hopkins Univer si ty P re ss , 1984).

9 Descriptive, narrat ive analy ses o f the developmentof the welfare s tate can be found in Hugh Heclo, ModernSocial Pol i t ics in Britain and Sweden: From Relief to IncomeMaintenance (New Haven: Yale University Press, 1974); PeterFlora and Arnold J . Heidenheimer, eds. , "The Historical Coreand Changing Boundaries of the Welfare State," in The Development of Welfare States in Europe and North America (NewBrunswick, N.J. : Transaction Books, 1981); Ann Shola Orloffand Theda Skocpol, "Why not equal Protection? Explaining the

Polit ics of Public Social Spending in Britain, 1900-1911,and the United States , 1890s-1920, " American SociologicalReview 49 {December 1984}: 726-50; and Norman Furniss, ed. ,Futures o f the Welfare State {Bloomington, I l l inois : IndianaUniversity Press, 1986}; Richard Rose, "How Exceptional i s

the American Poli t ica l Economy?" Polit ical Science Quarterly104 (Spring 1989): 91-115.

Quanti tat ive studies include: Morris Beck, "PublicSector Growth: A Real Perspective," Public Finance 34



12

differences between countries, taking a sample of very

different countries for the i r analysis . lO

Most studies find that there i s a general trend toward

expansion of social programs. Expansion has two dimensions.

Firs t , the p ropo rtio n o f th e populatio n targeted fo r p ro vi-

sion of services has generally increased. Second, the range

of benefi ts ava ila ble t o recipients has also increased. One

of the reasons given for the expansion of benefi ts i s tha t

there is a tendency for governments to index benefi ts in

many domestic welfare programs, such tha t increases occur in

a quasi-automatic manner, following the rate of inflat ioL.

Indexing, and expansion in the target population has raised

(1979): 313-55; Morris Beck, Government Spending: Trends andIssues (New York: Praeger Publishers, 1981); Frank Gould,liThe Development of Public Expenditures in Western, Indust r ia l ized Countries: A Comparative Analys is ," Pub li c Finance38 (1983 ): 38-69 ; Balbir S. Sahni and Balvir Singh, "On theCausal Directions Between Income and Government Expenditurein Canada," Public Finance 39 (1984): 359-93; Tom W. Rice,"The Determinants of Western European Government Growth,"Comparative Poli t ical Studies 19 (July 1986) :233-57 ; F red C.Pampel and John B. Williamson, "Welfare Spending in AdvancedIndustrial Democracies 1950-1980," American Jou rn al ofSociology 93 (1988): 1424-56.

10 The classic example is Philips Cutright , "Pol i t ica lStructure, Economic Development, and National Social Security Programs," American Journal of Sociology 70 (March 1965):537-50; cf. Sohrab Abizadeh and John Gray, "Wagner's Law: APooled Time-Series, Cross-Sectional Comparison," NationalTax Journal 38 (June 1985): 209-16; Bruce E. Moon andWilliam J . Dixon, "Poli t ics , the S tate , and Basic HumanNeeds: A Cross-National Study," American Journal of Pol i t i -

cal Science 29 (November 1985): 661-94.



13

questions about how growth in government expenditures can be

controlled.

After a relat ively rapid expansion of social programs

during the 1960s and 1970s pol i t ic ians and scholars began to

consider whether continued expansion was possible, or had

the growth of th e public sector in the new welfare s ta te

reached i t s apex? With the economic shocks of the mid-

1970s, producing economic crises in many industr ia l ized

countries and the recent phenomenon of stagf lat ion with the

dual problems of unemploYment and inf lat ion, f isca l resourc-

es seemed to be reaching thei r l imi ts . Several countries

began to sh i f t away from expansion of social programs to

programs of auster i ty during the late 1970s and 1980s.

In Bri ta in, Margaret Thatcher 's Conservative Party

began a process of privat izat ion. 11 Meanwhile, the United

States began a defense buildup, and began trimming or post-

poning the implementation of programs designed to combat

poverty or improve opportunit ies for the disadvantaged.

Many observers pointed to th e doc tr in es of pol i t i ca l par t ies

as the cause of these changes.

Whether pol i t ical part ies have a signif icant effec t on

government expendi tu res con tinues to be an item of contro-

11 For a detailed description of programs targeted bythe Thatcher government see Anthony J . Harrison, The Controlof Public Expenditure 1979-1989 (Oxford: Transaction Books,1989), 18-140 , 263-277.



14

versy among students of government spending. Research

concerning the effect of part ies on expenditures has also

spawned a debate on whether th e obje ct iv es of redistr ibut ive

programs have been achieved. For instance, debate continues

on the subjec t of whether socia l is t governments tha t espouse

ega li ta r ian po l ic i es actual ly achieve redist r ibut ion through

expenditures, or whether these pol ic ies merely provide a

baseline of income protect ion. Research on th e in flu en ce of

part ies and pol i t i ca l structures has deal t a t len gth w ith

the question of whether pol i t ica l factors improve the aggre-

gate sta nd ard o f l iving, improve economic performance, or

even extend l i fe expectancy and reduce infant mortality.12

Much of this research has focused on government expenditures

12 This topic continues to be a matter of controversy, cf. Cutright , "Poli t ical Structure, Economic Development," 537-50; and Heclo, Modern Social Pol i t ics in Bri tainand Sweden, 227-283; and Richard M. Coughlin and Phil ip K.

Armour, "Methodological Issues in the Comparative Study ofSocial Securi ty: Qual it at ive vs. Quan ti ta tive Analysis andthe Appropriate Use of Data," Internat ional Review of ModernSociology 10 (July-December 1980): 25-48; Douglas Hibbs,"Poli t ical Part ies and Macroeconomic Policy," American

Pol i t ica l Science Review 71 (December 1977): 1467-87; JamesL. Payne, "Inflat ion, Unemployment, and Lef t-Wing Poli t ica lPart ies : A Reanalysis," American Poli t ica l Science Review 73

(March 1979) : 181-85; Douglas Hibbs, "Communicat ions,"American Poli t ica l Science Review 73 (March 1979) : 185-90;Robert W. Jackman, "Soc ia li st Pa rt ie s and Income Inequalityin Western Industrial Societies," Journal of Poli t ics 42

(February 1980) : 135-149; Moon and Dixon, "Poli t ics , theState, and Basic Human Needs," 661-94.





16

duce these policies a re gener al ly populated by un-elected

bureaucrats and other c iv i l servants unaff i l iated with major

pol i t i ca l par t ies . While the issue of the relat ive impor

tance of f iscal , monetary and regulatory pol ic ies on econom-

ic performance and social change is very interest ing, and

worthy of continued research, th is s tudy add resses the more

l imited que stio n o f whether pol i t ical par t ies act to al lo-

cate resources in the manner commonly assumed in popular

pol i t i ca l discourse and debated in scholarly journals.

The question posed here i s whether pol i t ical part ies

systematically change the direct ion of government spending.

This quest ion i s intimately t ied to the question of whether

par t ies control public pol icy. I f a party cannot change the

al locat ion of government r esou rc es , t hen i t i s not able to

implement i t s policy objectives. Events during the l a s t

decade seem to in dic ate tha t a correla t ion exists between

changes in the part isan composition of government and subse

quent spending pat terns. Recent pol i t i ca l history seems to

support the idea that part ies influence the allocation of

expenditures.

In the United States during the 1980s, social workers

and advocates fo r social programs in the areas of education

and other social services complained about the Reagan

administrat ion 's lack of financial support for the i r pro

grams. Many social programs experienced cutbacks, and



17

f inancial support for proposed programs were not forthcom-

ing. Two theoret ical ly important issues are a t stake in

determining whe ther such complaints are properly directed a t

the Reagan administration. Fir s t i s the que stio n o f whether

budgetary changes are products of a general pat tern of

part isan pol i t ics , whereby increases in the power of one

party, in th is case the election of a president (and senate

majority) from the Republican Party, predictably foreshad-

owed the demise or expansion of specific kinds of programs.

Put different ly, is there something about the Republican

Party that determines how budgets are divided among policy

al ternat ives? Or, is th is an economically driven or idio-

syncrat ic occurrence without any pol i t ical ly generalizable

features? In short, was Reagan personally responsible, qua

Ronald Reagan, for these changes, or was he responsible qua

Republican president? Is there something inherent to the

nature of pol i t i ca l p artie s th at provides a foundation for

predict ing government policy? Some prel iminary observat ions

on th is matter seem espec ia lly relevant .

During the 1980s, the similar i ty of events in United

States and the United Kingdom evoked many commentaries on

the lead ers of those countries. The personal aff ini ty of

President Ronald Reagan and Bri t i sh Prime Minister Margaret

Thatcher i s widely recognized. In the United Kingdom under

the Thatcher government, a program of privat izat ion and



18

f iscal austeri ty led to the d ismant li ng o f a number of

ins t i tut ions tha t had received suppo rt under previous Labour

Party governments. In part icular , nat ional ized industr ies

were targeted for privatizat ion. This led to a coal miner 's

s t r ike in p ro te st of the government's pol ic ies . 1s

Similar changes occurred in the United States during

the Reagan administration. What Anthony Harrison terms

"off- loading" became a key component of Reagan 's domestic

policy agenda. 16 In the United States, under the Reagan

administrat ion, grants- in-aid to the s ta tes were cut, under

the guise of providing more "local control ," and l i f t ing the

weight of centralized decision-making from the shoulders of

s ta te and local off ic ia ls . u

Changes in both countries seemed to be made a t the

expense of labor in terests and other social groups a t the

lower end of the socio-economic spect rum, which prompts some

15 See Harrison, Control of Public Expenditure, 6,passim.

16 Ibid, 18-83. Cf. Harold Seidman and Robert Glimour,Pol i t ics , position and Power, 4th ed. (Oxford : Oxford Univers i ty Press, 1986), 119-35.

17 Cf. R ichard Nathan and Fred C. Dooli t t le , "Changesduring Reagan's Firs t Term," in Reagan and the States , eds.Richard Nathan and Fred C. Dooli t t le , (Pr ince ton: Prince tonUniversity Press, 1987) 44-66; John E. Chubb "Federalism andthe Bias for Centralization," chap. 10 in The New Directionin American Poli t ics , eds. John E. Chubb and Paul E. Peterson (Washington, D.C.: The Brookings Ins t i tut ion, 1985),273-306. Seidman and Gilmour, Poli t ics , posi t ion and Power,133-6.



19

sense o f c ompa rab ility . But i s th is a general p atte rn th at

applies to a l l administrat ions that share a so-called con-

servative ideology? Moreover, how do nat ional legislatures

f igure in the equation?

Numerous scholars have found relat ionships between the

part isan composition of national parliaments and growth in

specific kinds of budget allocations . Others argue tha t

social , demographic and economic factors determine govern-

ment expenditures. In cross-nat ional s tudies , i t i s widely

theorized tha t par t ies of the l e f t increase the size of the

public s ec to r, e sp ec ia ll y with respect to domestic expendi

tures on welfare programs .1 8

Some scholars have expressed skepticism that changes in

expenditures result from party influences. They argue that

18 See Douglas Hibbs, Economic In teres t and the Polit i cs of Macroeconomic Policy (Centre for InternationalStudies: MIT, 1975); Hibbs, "Pol i t ica l Part ies and Macroeconomic Policy," 1467-87; David Cameron, "The Expansion of thePublic Economy," American Poli t ica l Science Review 72 (De-cember 1978): 1243-61; Tufte, Poli t ica l Control of theEconomy; Francis G. Castles and R. McKinlay, "Does Poli t icsMatter: An Analysis of the Public Welfare Commitment inAdvanced Democratic States ," European Journal of Pol i t ica lResearch 7 (1979) :169-86; Walter Korpi, "Social Pol icy andDistr ibutional Conflict in the Capital is t Democracies: APreliminary Framework," West European Pol i t ics 4 (1980):

296-316; Francis G. Castles, "The Impact of Part ies onPubl ic Expenditures ," in The Impact of Part ies , ed. FrancisCastles (Beverly Hills: Sage, 1982), 21-96; Gould, "TheDevelopment of Publ ic Expend itures ," 38-69; Paul Whiteley,Poli t ica l Control of the Macroeconomy: The Poli t ica l Economyof Public Policy Making (Beverly Hil ls : Sage Publications,1986); Duane H. Swank, "The Pol i t i ca l Economy of GovernmentDomestic Expenditures in th e A fflu en t Democracies , 1960-80,"American Journal of Poli t ica l Science 32 (1988): 1121-50.



20

var ia t ions in expenditures are not his tor ica l ly significant

and tha t pol i t i ca l influences are subordinate to other

factors, part icularly economic and demographic trends. 19

There are several reasons for skepticism. One impor-

tan t reason i s tha t government expendi tu re s t end to change

more slowly than the pol i t ica l environment, especial ly with

respect to changes in public opinion.

Moreover, it is widely recognized tha t changes in

government expenditures from one year to the next are highly

correla ted. Changes in expenditures generally occur in

small increments. 2o For th is and other reasons, the pre-

ponderance of research has focused on socio-economic factors

tha t also change incrementally, or very slowly over tim e.

Indeed, much of the research on the origins of the welfare

s ta te has emphasized the his tor ica l context of the welfare

s ta te . Additionally, certain aspects of a country's po l i t i -

19 See Cutright, "Poli t ical Structure," 537-550; HaroldWilensky, The Welfare State and Equality: Structural andIdeological Roots of Pub li c Expenditu re s (Berkeley : Univers i ty of California Press, 1975); Richard Rose, Do Part iesMake a Difference?, 2d ed. , rev and enl . (Chatham, New

Jersey: Chatham House Publishers, 1984). For a less assuredaccount of non-part isan influences see Beck, "Public Sector

Growth," 313-55; Beck, Government Spend ing: Trends andIssues.

20 See Charles Lindblom, "The Science of MuddlingThrough," 79-88; Aaron Wildavsky, The Poli t ics of the Bud-getary Process; and Otto Davis, M.A.H. Dempster, and AaronWildavsky, "Toward a Predict ive Theory of Government Expenditures: U.S. Domestic Appropriations," Bri t i sh Journal ofPol i t ica l Science 4 (October 1974): 419-52.



21

cal cu ltu re are sometimes credited with promoting expansion

of welfare programs and increases in the s ize of the pu blic

sector . Among the factors considered to be in f luent ia l in

promoting expansion in the welfare s ta te are: th e exis te nc e

of a strong labor movement (combined with a strong democrat-

ic soc ia l i s t movement), rel igious aff i l ia t ions (such as

Catholicism), and economic rel iance on the export sector . 21

These perspectives on government spend ing pose al terna-

t ive hypotheses. On one side of the issue are those who

argue tha t pol i t i ca l part ies have an important impact on

expenditures, especially those expenditures related to the

size and complexity of the welfare s ta te . On the o ther sid e

are those who argue tha t government expenditures are econom-

ica l ly determined, as a resul t of industrial izat ion, demo-

graphic composition, economic affluence, and even the

country 's vulnerabil i ty to f luctuations i n i nt er na ti on a l

trade. 22 Economic theories comprise the bulk of scholar-

21 Harold L. Wilensky, "Leftism, Catholicism, andDemocratic Corporatism: The Role of Pol i t ica l Part ies inRecent Welfare State Development," in The Development ofWelfare States in Europe and North America eds. Peter Floraand Arnold J . Heidenheimer, (New Brunswick, N.J. : Transac

t ion Books. 1981) , 345 -82. See also, Flora andHeidenheimer, "The Histor ical Core and Changing Boundaries,"17-34. For a very good overview of t h is l it er at ur e, seeHannu Uusitalo, "Comparative Research on the Determinants ofthe Welfare State: the State of th e A rt," European Journalof Pol i t ica l Research 12 (1984): 403-22.

61.

22 See Cameron, "The Expansion of the Public," 1243-



22

ship on government spending, but can th ey add re ss the ques-

t ion of how guns versus but ter choices are made?

Economic Determinism

By far the most frequently ci ted and most widely dis-

cussed theory of the growth in public expenditures i s the

theory f i r s t espoused in the 1880s by a German economist,

Adolph Wagner. He hypothesized tha t as populations become

more affluent , the i r demand for public services inc reas

es . 23 This theory has received widespread acceptance among

economists and pol i t ical sc ient is ts alike. 24 A closely

related school of thought suggests tha t increases in govern-

ment spending resu l t from the pressures of powerful in terest

groups.25

Some theor is t s have argued that extending government

benefi ts generates a rachet effec t whereby, once offered,

23 For a concise descr ipt ion, and empirical t es t ofWagner's Law, see Abizadeh and Gray, "Wagner's Law: A PooledTime-Series," 209-16.

24 See James Buchanan and Gordon Tullock, "The Expanding Public Sector: Wagner Squared," Public Choice 31 (1977):147-50; Rice, "The Determinants of Western European Government," 243.

25 Cf. Lowi, The End of Liberalism, 42-63 ; Morr isFiorina, Congress: Keystone of the Washington Establishment(New Haven: Yale University Press, 1977): 29-49. Theseaccounts are firm ly r oo ted in the ear l ie r work of David B.Truman, The Governmental Process (New York: Alfred A. Knopf,Inc. , 1951); and v.o. Key, Pol i t ics , Part ies and PressureGroups Fourth Edition (New York: Thomas Y. Crowell Company,1958) .



23

public welfare programs are pract ica l ly i r revers ible . 26

This dynamic i s based on the observation that such benefi ts

are ra re ly r ol le d back to previous levels . 27 The authors

of this theory, Peacock and Wiseman, argue tha t growth in

such expenditures occurs in the wake of war. They theorized

tha t popu la tions sub jected to the s t ra ins of war expect a

payoff for the i r deprivation during the war. In the i r view,

a kind of quid quo pro between pol i t i ca l e l i te and the

population tends to in cre ase expenditures. This theory is

supported by the expansion of social expenditures in many

Western democracies following the Second World War.

Other theorists argue tha t the very nature of demo-

cratic government i t s e l f mil i ta tes against f isca l re -

s t ra in t . 28 The t rad i t ional form of th is argument i s tha t

office holders increase payoffs to voters to influence

elec tora l outcomes. A closely related theory is tha t , in

addit ion to the elec tora l needs of pol i t ic ians , in te res t

26 See Alan Peacock and Jack Wiseman, The Growth ofPublic Expenditure in the United Kingdom (Pr ince ton: P rince

to n Univ ers ity P res s, 1961).

27 See Dirk Wolfson, "Controlling the Welfare State: ACase Study of Retrenchment in the Netherlands," PublicFinance, 42, no. 2 (1987).

28 See Anthony Downs, "Why the Government Budget Is TooSmall in a Democracy," World Pol i t ics 12 (1960) : 541 -563 ;Buchanan and Tullock, The Calculus of Consent.



24

groups conspire, sometimes in collaboration with bureau-

crats , to expand public expenditures. 29

According to t he se per spec ti ves , part ies contributes

l i t t l e or nothing to the explanation of changes in expendi-

tures . Together, such theories const i tute what has been

termed a "par ti san inva r ian t hypothes is ," suggest ing that

part ies do not affect expenditures. Instead, expenditures

are subject to the influences of in teres t groups, pol i t ical

cycles, or economic condit ions, regardless of the party in

power.

I f these t heor ie s a re correct with respect to the

influence of pol i t i ca l part ies, then there should not be a

correlat ion between the strength of certa in kinds of part ies

and cer tain kinds of expenditures. The l i terature , however,

indicates that some researchers have found jus t tha t (even

when controlling for economic and demographic variables) .

American Exceptionalism

The controversy over the part isan impact on expendi-

tures has important implicatio ns for stu dents of American

pol i t i cs . For decades , American po l it ic a l s c ie n ti st s have

cr i t ic ized both the Democratic and Republican part ies for an

29 Cf. Fiorina, Congress: Keystone of the WashingtonEstablishment, 87-93; Lowi, The End o f L ib er al ism.



25

apparent inabi l i ty to provide voters clear al te rna t ives .

I t i s well known that American part ies do not maintain

the level of party discipline found in most other democra

cies. This fact is at t r ibuted to a variety of influences

including the structural fragmentation of party organiza-

t ion , party members who act as policy entrepreneurs for

the i r own electoral advantage and the tendency of part ies to

become enmeshed in interest-group pol i t ics a t the expense of

coherent policy making.

Coupled to these characterizations of the two major

par t ies , students of American pol i t ical values have s tressed

the non-ideological character of the electorate . Indeed,

th is i s an enduring theme in most studies of American pol i t -

ica l thought.

For more than 150 years, one of the most frequently

c it ed au tho ri ti es on the character of American pol i t i ca l

thinking has been Alexis De Tocqueville. His comments about

the United States in the 1830s have been recognized for

thei r enduring relevance. Among his many comments on the

character of Americans, he wrote tha t "I think no country in

the civil ized world is less attention paid to philosophy

than in the United States . ,,30 Moreover, De Tocqueville

noted tha t Americans tended to be more occupied with busi-

30 Alexis de Tocqueville, Democracy in America, ed.Phil l ips Bradley, Vol. I, pt.2 ch. 10 (New York: Knopf,1945), 3.



26

ness than people in Europe. This description f i t s well into

other accounts of Americans being individual ist ic , se l f -

interested, and pragmatic. This character izat ion has domi-

nated most subsequent analyses of pol i t i c s in the United

States . Ideology is not considered to be an important

feature of American pol i t ics .

Pragmat ism, and a commitment to a general notion of

l ib er al ism a re consis tent charac ter iza t ions of the American

pol i t i ca l cul ture . 31 In comparison with other most

European democracies, t he id eo logi ca l differences between

part ies in the United S tates a re relat ively minute. In a

manner of speaking, many scholars consider the United States

to be "exceptional" in i t s deviation from the ideological

dist inct ions found in other Western democ racie s. L ikewise,

i t has been suggested that U.S. expenditures patterns are

equally excep tional, lagging behind other advanced indust r i

alized democracies in the development of a number of social

programs. This study will examine the empirical val idi ty of

th is assessment of American pol i t ics .

Party and Policy

Much of the previous research on the influence of

par t ies on expenditures has included references to par ty

31 The two most prominent works on th is su bjec t areLouis Hartz, The Liberal Tradit ion in America (New York:Harcourt, Brace, 1955); and Richard Hofstadter, Liberalismin America (New York: Vintage Books, 1958).



27

ideologies. A host of di f fe rent coding methods have been

employed in this f ield with mixed resu l ts . At the root of

this kind of analysis i s the presupposi t ion that par t ies

dif fer in thei r preferences for part icular kinds of pol i

cies . Exactly how or why t he se d if fe rences should obtain i s

frequently l e f t to the imagination of th e read er. Even

those studies that suggest tha t part isan ideologies play a

par t in determining expenditures devote l i t t l e time explain

ing the connections between ideology and the kinds of ex

penditures under consideration.

To the extent that generalizations can be made from

th is area of research, there appears to be a scholarly

consensus tha t lef t - leaning part ies tend to support, a t

leas t rhetorical ly, expenditures for social programs, while

more conservat ive, or right-leaning par t ies tend to support

expenditures for defense and minimize the amounts spent on

social programs.

Several gaps in the explanations presented in the

l i terature are apparent. Few accounts describe what i s it

about the ideology that promotes one kind of bias as opposed

to another. I f there i s an underlying principle operating,

i t i s rarely made expl ic i t . In short, the content of what

i s being measured when l e f t - or right-wing parl iamentary

seats are being counted i s not f ul ly spec if ied. What i s



28

missing i s a defini t ion of the theoret ical content of the

terms l e f t and r ight .

Giacomo Sani and Giovanni Sartor i describe the l e f t -

r ight continuum as being in the "social change/ soci al equal -

i ty domain." This i s consistent with many descriptions of

ideologies found in the philosophically oriented l i terature ,

but rarely mentioned in empirical, quant i ta t ively oriented

studies . 32 Sani and Sar tor i ' s comment on the content o f

the lef t - r ight continuum suggests that the Left favors

social change and social equali ty more than the Right.

Colloqui al cat ch phrases, such as "bleeding heart ," or "hard

r ight ," capture much of the sense tha t Sani and Sartori

intend in describing th is domain.

Popular and scholarly perc ep ti on s o f t he ideolog ical

lean ings o f pol i t ical groups are often ar t icula ted in spa-

t i a l descriptions which note the depth and degree of those

leanings. Pol i t ica l groups on both sides of the l e f t - r igh t

continuum are percieved as being moderate or radical in

thei r adherence to ideological posi t ions.

Par t ies of the le f t are often descr ibed a s advocates of

"progressive," social programs and adherents of social

change. Part ies of the r ight are characterized as being

32 Giacomo Sani and Giovanni Sartori , "Polarization,Fragmentation and Competition in Western Democracies,"Western European Party Systems, Edited by Hans Daalder andPeter Mair, (Beverly Hills: Sage, 1983), 310.


http://slidepdf.com/reader/full/parties-ideologies-and-budgets-a-study-of-budgetary-trade-offs 38/236.. .

29

le ss s en sit iv e to the issue of inequali ty and generally

opposed to social change. As mentioned, there i s a wide-

spread perception that these leanings are a matter of de-

gree, where those part ies at the extremes are considered

radical as opposed to more moderate, or even centr is t par-

t i es . These perceptions of the pol i t ica l landscape are not

frequently ar t iculated systematically. For this reason, i t

i s adviseable to inspect the more cogently art iculated

philosophical ground latent in the perception that a po l i t i -

cal party is more or less support ive of s oc ia l e qu ali ty and

social change (A more extensive treatment of th is matter,

and a d is cu ss ion o f re la ted measurements, is presented in

the fol lowing chapter ) .

These perceptions are used to t es t the very general

hypothesis that policy al ternat ives stem from the ideolog-

i ca l commitments of pol i t ic ians . Moreover, it i s theorized

that pol i t ic ians associate themselves with other like-minded

people in organized pol i t i ca l part ies. Explaining the

connection between ideology and public policy i s the subject

of the next chapter.

The following discussion draws on the work of scholars

in the f ield of pol i t ical philosophy to relate three major

ideological persuasions, or orientations, to specif ic kinds

of pol ic ies and then to a theory of budgetary t rade-offs .



CHAPTER :I I

THEORY AND METHODS

Introduction

This chapter develops a theory of how ideologies are

re la te d to budgetary t rade-offs . The f i r s t step in this

process i s to establ ish the theoret ical content of the l e f t -

r ight continuum which wil l be used to measure the principal

independent variable. The "social equal i ty/social change

domain" that Sani and Sartor i mention, provides the focus

for the discussion of the theoret ical construct of the

continuum. This i s a necessary step in explaining why

part ies should matter to budgetary outcomes.

What i s developed in the following sections i s a theory

of what i s inside the "black box" so frequently referred to

in systems models of pol i t ics . By developing a description

of the concep tua l con tent o f a l e f t - r igh t dimension, I am

postulat ing that the in ternal mechanisms of government

operate according to a set of generalizable rules . To

define these rules, and thereby the content of the dimen-

sion, I refer to the work of pol i t ical philosophers, whose

area of expertise is the definit ion of ideological content.

In the in teres t o f b rev ity and s imp li ci ty t hr ee "ideal

types" of the most preval en t i deo log ie s are developed below.

30



31

The ideologies considered include socialism, class ica l

l iberalism, and conservatism. Drawing on the cr i t i ca l

an aly se s o f scholars who specialize in the f ie ld of moral

and pol i t ical philosophy, a select ion of key character ist ics

are ident i f ied. More variety of opinion exists than can be

catalogued coherently in a single study, so some narrow

issues are not t reated here. With these l imitations in

mind, what is being asserted is the plaus ibi l i ty of the

soci al equa li ty /soci al change dimension suggested by Sani

and Sartor i across a spectrum of pa rty ideo log ie s. These

characterizations are intended to i l lus t ra te why certa in

kinds of biases toward policies should be expected by par-

t ies that fa l l more or less to the r ight or l e f t .

A Theory of Ideology

Since the word was coined in the eighteenth century,

the concept of ideology has been defined in many di f fe rent

ways. The most common way o f descr ib ing ideology is as a

bel ie f system or "world view," through which social real i ty

is interpreted. This part icular understanding of ideology

emphasizes the cognit ive effects that a publ ic ly o rien ted

phi losophy has on an individual 's interpretat ion of percep-

t ions, pol i t i ca l at t i tudes and public values. In th i s

respect, the analogy of a "prism" is frequently used to

describe the way that an ideology affects cognitive opera-

t ions. Thi s ana logy is similar in many ways to what has

been conceptualized as an "operational code," or ra ther a



32

se t of predisposi t ions to select ively receive and respond to

events . According to th i s view, ideology generates a sensi-

t iv i ty to part icu lar kinds of s t imul i , and blindness to

other kinds of s t imul i , structuring the kinds of consider

at ions which inform decision making. 33 In essence, di f fe r -

e nt id eo lo gie s provide al ternat ive frames of reference, o r

ways of thinking,34 and thereby constra in the menu of con-

ceivable pol icy a l te rna t ives avai lable to c erta in actors .

Pejorat ively , the term "ideology" i s sometimes used to

mean a kind of stubborn adherence to a se t of soc ia l objec-

t i ves . In th i s view, an ideology fu nctio ns as a kind of

secu lar t heology, p rov id ing the basis for moral and po l i t i -

ca l judgements. Alternatively, ideologies are sometimes

viewed as socia l conveniences, as ra t ional izat ions or j u s t i -

f ica t ions fo r po l it ic a l a c t iv i t ie s . In this view, ideology

does not t ru ly provide a cause for act ion, but is a cloak or

33 See Milton Rokeach, The Open and Closed Mind (NewYork: Basic Books, 1960), for an excel lent treatment of howb elie fs a ff ec t the interpretat ion of events . The r e la t ion

ship of core b el ie fs to ideology i s explained qui te well inElinor Scarbrough, Pol i t i ca l Ideology and Voting: An Exploratory Study (Oxford: Clarendon Press, 1984), 24-49. There la t ionship of the cognitive view to cybe rne ti c theo ry i squi te in teres t ing . Cf. Karl W. Deutsch, The Nerves of Gov-

ernment (New York: The Free Press, 1962); Bradford P.Keeney, Aesthet ics of Change (New York: The Guil fo rd Press ,1983), 12-109 . Theoretical metaphors di ffer considerably.I n c yb ern et ic theory, what I am describing is a matter ofthe cal ibra t ion of receptors.

34 See Karl Mannheim, Ideology and Utopia, t rans .

Louis Wirth and Edward Shils (New York: Harcourt BraceJovanovich, Publishers, 1985), 70-108.



33

instrument of pol i t i ca l power, serving more basic in teres t -

seeking motivations. 35

This study tentat ively posits former view tha t ideolo-

gies are art iculated "world views" having significance in

determining pol i t i ca l values and, thereby, how decision

makers formulate and choose pol i t i ca l al ternat ives . The

u t i l i ty of employing ideo log ica l descr ipt ions to predict

behavior will determine th e e xte nt to which th is ten ta t ive

commitment i s warranted.

A Spatia l Arrangement of Ideologies

The essence of the no tion of l e f t and r ight employed in

th is analysis i s a spat ia l re la t ionship. This section will

describe ordinal re la t ionships between ideologies by refer-

ence to certain core values, such tha t one ideology i s "more

than" ano ther with respect to i t s menu of jus t i f ica t ions for

supporting certain kinds of policies. The following charac-

terizat ions should not be taken as arguments tha t ideologies

are themselves s t r ic t ly unidimensional. Each of the major

ideological movements being descr ibed has , within i t s own

system o f v alues, certain con tr ad ic tions , o r tens ions , which

produce some inconsistency and, in some cases, incoherence.

Thus, the following descriptions are meant to i l lus t ra te the

35 See Eric Voegelin, Science Poli t ics and Gnosticism{Chicago: Regnery, 1966}i see also Alasdair MacIntyre,Against the Self Images of the Age {Notre Dame: Notre Dame

University Press, 1970}.



34

reasons for a general bias , or predisposition toward a

part icu lar viewpoint.

The method employed below is speculat ive, re lying as i t

does on c r i t i ca l commentaries which frequently display

ul ter ior aims in presenting opposing views. Nevertheless, I

believe tha t t he charac te ri za tions are val id insofar as the

assert ions concern the l imited number of issues addressed.

I t seems appropriate to begin with the ideologies of

the Left, which have received the most scholarly at tent ion.

I t i s widely acknowledged tha t s oci al is t p a rt ie s take inspi-

rat ion, i f not specif ic po li cy ob jec ti ves , from Marxist, or

neo-Marxist theories . Contemporary Marxists theor is t s tend

to shy away from the systematic, quanti tat ively styled

arguments presented in Das Kapital, preferr ing instead to

focus on the normative humanistic values of Marxism. The

famous dictum, "From each according to his abi l i ty , to each

according to his need," o r v aria nts th ereo f, has continued

to be the cornerstone of socia l is t thought. 38 The rea l iza-

t ion of a more jus t distr ibut ion of economic and socia l

benefi ts the centerpiece of socia l is t values.

38 This quotation is found in Marx's "Crit ique of theGotha Program," [See Robert Tucker, The Marx-Engles Reader,2nd ed., (New York: W.W. Norton & Company, 531.] but somescholars at t r ibute i t to the French social i s t , Louis Blanc,in dating as early as 1848 [See George Seldes, The GreatThoughts, With a Foreword by Henry Steele Commager. (NewYork: Ballantine Books, 1985), 274.].



35

Among socia l is ts , debates have raged for over one

hundred years about what aspects of Karl Marx's writ ings are

relevant to the practical circumstances of the socia l is t

movement. Marx's theory of revolution has been modified and

amended by many of his followers, who hoped to see a t rans

formation of society through peaceful means. 39 Orthodox

Marxists have tended to adhere the bel ief that the contra-

dict ions of capital ism cannot be resolved peacefully and

tha t the real izat ion of the communist society depends on the

violent overthrow of exist ing pol i t ical ins t i tu t ions , with

the dictatorship of the prole tar ia t presiding over the

t ransi t ion to the s ta te less society. Despite the i r stated

aim, Marxist part ies have been less revolutionary in the i r

act ivi t ies than the i r rhetoric would suggest. 40

Soc ia li st p a rt ie s , on the other hand, tend to be more

moderate than the i r communist counterparts in adhering to

Marxist doctrines. 41 Moreover, in many in stances th e dif -

39 Karl Kautsky is credited as being a central figurein the schism tha t developed within the soc ia l i s t movement.Lingering tensions between Communists and Social Democratshave continued to this day. For a concise narrat ive discussion of this problem, see Robert C. Tucker, The Marx-Engels

Reader, 2d ed., (New York: W.W. Norton & Company, 1972),xxxiv-xxxvi.

40 The classic commentary on th is subject is RobertMichels, Poli t ica l Part ies , t rans . Eden and Cedar Paul(Glencoe, I l l ino i s : The Free Press, 1915, reprinted 1949) .

41 See Charles Hobday, Communist and Marxist Part iesof the World (Santa Barbara, California: ABC-Clio, 1986).



36

ferences between communists and socia l is ts over th eir in te r-

pretat ions of Marx have re su lte d in enduring b i t t e r antago-

nisms. 42 In g en eral, socia l is t part ies tend to balk a t the

more radical methods and aims of the communist part ies ,

which tend to be relat ively more doctr inaire and mili tant in

the i r approach to pol i t ics . Nevertheless, soc ia l i s t part ies

have tended to advocate highly egal i tarian object ives ,

endorsing redistr ibut ion of income in part icular, and wealth

in general . 43

One important aspect of soc ia l i s t thought has been the

extension of egal i tarian values beyond nat ional boundaries.

This in ternat ional i s t or universal is t ic perspective is

congruent with a concern for human equali ty, and di ffers

from the re la t ive ly more par t icu la r i s t i c concerns of other

ideologies to which we now turn .

42 The or1g1ns of such disagreements begin with theFir s t Internat ional , with the disagreements betweenProudhon, Marx, Lassall and others. A number of works provide detai led narrat ive accounts of the spl intering ofcontemporary soc ia l i s t part ies . See Thomas T Mackie andRichard Rose, The International Almanac of Electoral History2d ed. (New York: Facts on File, Inc . , 1982); V incent E.

McHale, ed . , Pol i t ica l Parties of Europe (Westport, Conn. :Greenwood Press, 1983); Hobday, Communist and Marxist Part i es of the World; see also Keesing's Contemporary Archives(London: Keesing's Limited, various issues) ; Facts on File ,(New York: Facts on File , Inc. , various i ssues) .

43 Social is ts have also advocated the national izat ionof industr ies . See J.A. Hobson, Imperialism: A Study (London: Allen & Unwin, 1938).



37

I f communism, or radical socialism, i s viewed as one

pole of an ideological continuum, conservatism is often

described as the opposite pole. In essence, the common

notion of conservatism includes a par t icular is t ic , i f not

parochial aspect. 44 For conservat ives, the realm of ac-

ceptable policy is l imited to what has been t r ied success-

ful ly. Conservatives tend to view the world as f ini te , with

few poss ib i l i t i e s for posi t ive change. 45 While socialism

emphasizes the transformation of society and the confl ic t

between classes, conservative theorists emphasize the pres-

e rv atio n o f social order and domestic tranquility.46 In

44 For an excellent treatment of conservative thought,as well as socia l is t and l ibera l thought, see Karl Mannheim,Ideology and Utopia, 229-39, passim.

45 The theme of l imited and unl im ited "vi sions " devel oped by Thomas Sowell, Conflict of Visions (New York: Will iam Morrow and Company, Inc. , 1987), corresponds closelywith the writings of Alasdair MacIntyre, Af te r V ir tu e. 2d ed(Notre Dame: University of Notre Dame, 1984); idem WhoseJust ice? Which Rationali ty? (Notre Dame: University of NotreDame, 1988). MacIntyre argues against the universa l is t ic ,post-enlightenment moral philosophies of l iberalism andsocial ism, largely from the per spec tiv e o f a par t icular -i s t i c philosophy that emphasizes and embraces the notion o fpractices embedded in the his tor ica l ground of a t radi t ion( i .e . way of l i fe ) with purposes internal to that way ofl i f e , outside of which moral discourse becomes confused forlack of an in te l l ig ible point of reference, resul t ing in the

decay of moral agency.This perspective i s highly reminiscent of Mannheim's

treatment of conservative thought. See Mannheim, Ideologyand Utopia, 229-30.

46 For a cri t icism of conservative thought generally,and Edmund Burke in part icu lar , regarding a lack of purposeand the tendency to blind adherence to convention, seeMacIntyre, After Virtue, 221-22; idem, Whose Just ice , Which



38

contrast to the socia l is t vision of the perfect ibi l i ty of

society, conservat ives tend to s ee inhe rent l imits to the

potent ia l benef i ts of change. For conservatives, socia l

change is a matter of prudence. 47

With i t s preoccupation with the r isks and dangers of

change, conservative ideology should not be confused or

equated with capi ta l i s t ideology ( treated here as an aspect

of l iberal ism), which lauds the r isk -t aking ind iv idual and

the social benef it s a ssociat ed with entrepreneurship. Under

a good many circumstances, la issez fare capitalism and

soc ia l l ibe rta riani sm i s a threat to conservative values.

In fact , conservative rhetoric and theory advocates the

prohibi t ion of certain economic and social ac t iv i t i es in the

in teres t of public decency and preservation of the moral

community. 48 For similar reasons, there is no inherent

prohibit ion of redist r ibut ion in conservative thinking.

Some modicum of redist r ibut ion of income or wealth may be

Rationality?, 217-218; Cf. Mannheim, Ideology and Utopia,229-33.

47 See Sowell, A Conflict of Visions, 25-27; ManheimIdeology and Utopia, 229-39 . .

48 The debate between Sir Patrick Devlin, "The Enforcement of Morals," Proceedings of the Brit ish Academy(London: Oxford Univer si ty P re ss , 1959) and H.L.A. Hart,"Social Solidari ty and the Enforcement o f Mora lit y," Univers i ty of Chicago Law Review 35 (1967): 1-13; cf. Michael J .Sandel, "Mora lity and the Liberal Ideal ," The New Republic(May 17, 1984): 15-7; Cf. Mannheim, Ideology and Utopia,238-9, on the issue of moral harmony.



39

advisable from time to time as a paternal i s t ic manoeuver in

maintaining social order. But as a matter of prudence,

there i s a strong t endency toward maintaining the s tatus

quo. Accordingly, conservative theorists do not promote the

idea of changing the dis tr ibutions of goods, but express a

preference for maintaining and protecting exist ing dis t r ibu-

t ive arrangements , whatever those arrangements may be. 49

Law and order, t radi t ion, re l igios i ty and preservation

of social o rde r tend to be the touchstones of conse rvat ive

rhetoric and play a central ro le in defining what conserva-

tism i s about. L iberalism, on the other hand, appeals to

pr incip le in ways that run counter to both conservative and

social i s t values.

Within the t rad i t ion of l ibera l pol i t i ca l thought, the

question of d is tr ibu ti ve ju st ic e has been a persistent area

of disagreement and debate. With an emphasis on principle

and r ights , l ibera l theorists have been ambivalent toward

redist .r ibutive policies. While most l ibera l theorists

recognize a need for providing an equitable means of oppor-

tunity, the general thrust of l iberalism eschews governmen-

ta l i nt er vent ion for the purpose of adjusting economic

outcomes. One of the core values of the l ibera l ideology is

49 See Lowi, End of Liberalism, 45; cf. Mannheim,Ideology and Utopia, 230-9.



40

a preference for minimal government intervent ion. 50 For

the l ibera l , the Lockean caveat holds t rue tha t "government

is best when i t governs least ." A redistr ibut ive policy,

l ike government i t se l f , is viewed negatively.

At best , government is seen as a necessary evi l and

something to be minimized i f a t a l l possible . Indeed, many

l ibera l theorists argue tha t red is t ribu tive po l ic i e s harm

society by in terfering in a semi-automatically se l f -adjus t -

ing social marketplace (though not a l l l ibe ra l theor is ts

maintain th is view) .5 1 In th is sense, l iberal ism i s the

intel lectual bulwark of capital ism. However, l iberal ism

also supports the ideal of equal opportunity, which offers

l imited support for redist r ibut ive policies in cases where

the viabi l i ty of economic and social opportunities are at

stake, or as adjustments for market dis tor t ions . 52

50 See Robert Nozick, Anarchy, State and Utopia (NewYork: Basic Books, 1974). See also James S. Fishkin, Just ice , Equal Opportunity, and the Family (New Haven: Y a Univer si ty P re ss , 1983) for a discussion of l ibera l tensionsand ambivalence toward the use of the s ta te in preservingcore principles . For a discussion of why debate withinl iberalism fa i l s to e st ab li sh c le ar boundaries for the useof the s ta te , see Willmoore Kendall, "The 'Open Society' andI ts Fallacies ," American Pol i t ica l Science Review 54 (Decem-ber 1960): 972-79; cf. MacIntyre, Whose Just ice?

51 Cf. M ilton Friedman, There's No Such Thing as aFree Lunch (LaSalle, I l l inois : Open Cou rt, 1974 ); JohnRawls, A Theory of Just ice (Cambridge: Harvard Universi tyPress, 1971); James S. Fishkin, Just ice, Equal Opportunity,and the Family (New Haven: Yale University Press, 1983).

52 See Fishkin, Just ice, Equal Opportunity, and theFamily; cf. Nozick, Anarchy, State and Utopia.



41

By comparison to the socia l is t and the conservative

visions of dis t r ibut ive just ice , l iberalism fa l l s uneasi ly

in the middle. While i t i s possible to construct an argu-

ment for temporary pol ic ies of redis t r ibut ion within a

system o f conse rvat ive values, as a matter of prudence in

support of maintaining social order, a recurrent theme in

l ibera l discourse is the recommendation of a consistent

modicum of redis t r ibut ion as a matter of principle . 53 On

the other hand, the thrust of socia l is t thought is toward a

t ransformation of society in which redis t r ibut ion of soci-

eta l benefi ts i s a central pol i t i ca l objective.

Implici t ly , this l ine o f r ea soning has influenced many

studies on budgetary allocations, where l e f t party control

i s hypothesized to be re la ted to increased welfa re expendi

tures . Alternatively, r ight party control is hypothesized

to depress welfare expenditures , while increasing expendi-

ture on defense. While most studies are fa i r ly clear about

why l e f t of center, or s oci al is t p a rt ie s favor welfare

expenditures over defense, few s cho la rs exp la in why part ies

of the r ight would tend to favor defense expendi tu res per

see Perhaps it i s taken for granted that the r ight sees i t s

53 A good example of this concern i s found in Rawls, ATheory of Just ice. Rawls' v ersion o f redistr ibut ive just icei s to provide the leas t well off in society with a bonuswhen anyone else in society gains an improvement in thei rown condit ion. This standard has been widely cri t ic ized,for obvious reasons.


http://slidepdf.com/reader/full/parties-ideologies-and-budgets-a-study-of-budgetary-trade-offs 51/236..

42

nemesis in communism and therefore endorses foreign policies

of containment.

The right-favoring-defense hypothesis i s supported, in

some studies, by empirical regulari t ies , but the philosophi-

cal basis for such a hypothesis i s rarely art iculated and

remains speculative. According to Will iam Schneider, the

af f in i t ies o f right-w ing poli t ic ians transcend national

boundaries because of a shared in teres t in confronting

communism. He writes:

Conservative in ternat ional ists are anti-detente andpromilitary. They al ign ideologically with the in te r national Right, which i s to say that thei r primarymoral commitment is to the free world in i t s confrontat ion with communism. Liberal in ternat ional i s t s are prodetente and ant i -mil i tary. S4

This statement seems to crystal ize the conventional

wisdom r egarding the intent ion of the l e f t toward the mil i -

tary, but few studies expl ici t ly de fin e th e relat ionship of

l e f t i s t part ies to defense policies. The connection may l i e

in the way tha t defense policies re la te to other pol ic ies .

Certain types of policies change the exis t ing dis t r ibu-

t ion of societal benefits more than others . The archetypi-

cal redist r ibut ive policy i s the welfare t ransfer payment.

While the impact o f wel fa re payments on income inequal i ty i s

S4 William Schneider, II Public Opinion, II in The Makingof America'S Soviet Policy, ed. Joseph S. Nye (New Haven:Yale University Press, 1984), 16-17; cf. Jerrold E. Schneider, Ideological Coalitions in Congress (Westport,Connecticut: Greenwood Press, 1979), 59-80.



43

not firm ly established, 55 it i s certainly a pol i t ic ized

issue with di rec t relevance to ideological values. 56 In

fact , the issue of poverty, i t s causes and remedy is an area

where ideological arguments and pol i t i ca l at t i tudes are most

clearly ar t icu la ted . Policies aimed dealing with pover ty

ra ise ideological ly charged questions --questions related to

d i st ri but iv e j us ti ce .

Ideology and Distr ibutive Just iceThis study posits the view tha t ideologies provide

jus t i f ica t ions for public pol ic ies . Two dis t inc t issues are

raised in debates about the jus t i f i ca t ion fo r using govern-

ment to change the distr ibut ion of goods, one re la t ing to

individuals in society and another relat ing to the nature of

society i t se l f . Ideologicial proponents define th ese issu es

according to the i r own frame of reference, which is deter-

55 Cf. Jackman, "Social is t Part ies and Income Inequali ty ," 135-49; Moon and Dixon, "Poli t ics , the State, andBasic Human Needs ," 661-94.

56 See Darrel Montero, "The Guns and Butte r Is su e:Trends in American Public Opinion," Journal o f Sociology and

Social Welfare 5 (1978): 135-149; Philip Arthur Au Claire ,"Public Attitudes Toward Social Welfare Expenditures,"Social Work 29 (March-April 1984): 139-144; Larry J . Griff inand Kevin T. Leicht, "Poli t icizing Welfare Expenditures inthe United States ," in Futures of the Welfare State, ed.Norman Furniss (Bloomington: Indiana Univer si ty Press ,1986), 320-356; James R. Kluegel, "Macro-economic Problems,Beliefs about the Poor and Attitudes Toward Welfare Spending," Social Problems 34 (Feb ruary 1987) : 82-99.



44

mined by a set of core values which are frequently re la ted

to a sense of his tor ica l purpose.

Firs t , is the question of whether or not the recipient

i s des erving o f public support. The selection of a c r i t e r i -

on for making this kind of judgement is inescapably ideolog

ica l in nature. 57 Second, i s the question of how a pol icy '

w ill affe ct the structure and pract ices of social ins t i tu -

t ions, or rather , the nature of socie ty.

In general, the cr i t e r i a for a rec ip ient ' s worthiness

i s increasingly more res t r ic t ive , or rather less encompass-

ing, as one moves across the spectrum from l e f t to r ight .

Cons iderat ions of a rec ip ient ' s worthiness tend to be based

more on characterological cri ter ia for t ho se holding r ight-

wing views, whereas left-wing ideologies tend to profess a

universal concern for humanity. In the la t te r view, the

speci fi c cha rac t er is t ic s of the individual are not impor-

tanto Generally, for the r ight the recipient i s considered,

qua individual, bearing specific claims according to custom,

class, or merit. Alternatively, for the l e f t the recipient

57 I t i s doubtful tha t the cr i t e r i a of one ideology

would even be in te l l ig ible to adherents o f certa in otherideologies. Because such cri ter ia are inherent to thesystem of bel iefs of which they are a part (in a senseendogenously derived), they may not be t ranslatable to com-

peting systems of th ought. Neverth ele ss, adherents ofcompeting ideologies do disagree about pol i t ical means andends, but frequently "talk past one another." For a moreextensive treatment of th is issue see MacIntyre, Whose

Just ice , Which Rationali ty?, 349-403.



45

comes bearing human r ights , needs, or circumstance. For the

conservative, governmental responsibi l i t ies are prescribed

by considerations of the re la t ive significance of the

claimant 's evidence. For the socia l is t , governmental re-

sponsibi l i ty is l imited, i f a t a l l , by the existence of

exper t ise or resources to provide remedies. s8

On the ques tio n o f how public policies affec t the

society as a whole, the considerations of each ideology are

equally divided. For the conservative, the matter i s clear-

ly one of s tab i l i ty and order. For the soc ia l i s t i t is a

matter of meeting human needs, where the order of society i s

rearranged to provide an ega li ta r ian d i s tr ibu t ion . For the

l ibera l , social order i s considered important insofar as it

affec ts the individual and his or her l i fe chances, his or

herl iber ty

toengage in self -def ined pursuits , and whether

i t is being main ta ined as a fa i r e n v i r o ~ ~ e n t fo r economic

and socia l in teract ion. Stabi l i ty is valued for i t s role in

assis t ing the advancement of these principles through a

reasonable expectation that the rules of the game wil l be

observed, but new rules are necessary when the playing f ie ld

becomes too uneven.

Ideo log ica l considera tion of the effects of a public

policy on social order follow a gradation from an in t r insic

58 For a discussion of the role of ideology in determ ~ n ~ n g the l imits of governmental intervention see Sowell, AConflict o f V isio ns, 18-39.



46

value to an instrumental value. For conservat ives, poverty

and soc ia l i nequa li ty are par t of a natural order, an un

avoidable character ist ic of a f ini te world. Poverty needs

no part icular explanation for the conservative, i t jus t i s ,

as a matter of immutable social fact . Charity is jus t i f ied

as a means of prese rv ing the order of society . For the

l ibera l , poverty i s i t se l f not a problem, unless i t is

persistent , that i s , when i t is inter-generational and l i f e

long. Then i t raises questions about the opportunit ies for

individuals who get stuck in poverty.

According to the classical l ibera l view, in a fa i r

market people r i se and fa l l according to rules of fa i r

competition and exchange. To th e e xtent that any group or

class of people do not r i se (and fal l?) according to the i r

meri ts , the ques tio n o f fairness becomes an issue for the

l iberal . Is there discrimination? Are opportunit ies being

made ava ila ble to everyone? Have car te l s or monopolies

blocked part ic ipat ion and advancement? These are the con

cerns of the l ibera l .

For the l ibera l , keeping the playing f ie ld level , and

the rules fa i r is the key to the problem of (persistent)

poverty. For the social i s t , poverty i s a product of a

maladjusted society, of unresolved contradictions and dis

to rtio ns, th at once corrected, will remedy the pl ight of the

poor. According to socia l is t thought, economic and pol i t i -



47cal ins t i tut ions cause poverty, and t ransforming those

ins t i tut ions to meet human needs is the answer to the prob-

lem of poverty. Thus, the tendency to support

redistr ibut ive pol ic ies is affected two ways: by the l imits

that the ideology places on the val idi ty of individual

claims and by the relat ive importance of soci al s tr uc tu r es

vis-a-v is individual needs. These l imits , and the range of

concerns that jus t i fy them, become more narrow as one moves

from l e f t to r ight .

Returning to the question of defense policy, i t seems

reasonable t o pos tu la te that , where the s tab i l i ty of the

social "regime" is a central concern, there should be more

concern for i t s preservation from outside th rea ts . Con-

versely, when the structures of society are viewed more or

less as instruments of pol icy, there ought to be less impor-

tance attached to the i r preservation. In ei ther case, there

i s always a need to provide securi ty . The difference might

jus t be the margin o f par anoia that one id eo logy en joys over

another concerning what i s a t stake. But when the in t r ica-

cies of t radi t ions, the r i t es of re l igion and subt le t ies of

family re la t ions are a t stake, small changes are fundamental

changes. While dyed-in-the-wool socia l is ts see t radi t ion

and r i tua l as v es tig es o f ignorance and superst i t ion, con-

servatives see in them "a way of l i fe ." Their interpreta-

t ions of what i s at stake and what should be done to pre-



48

serve the status quo should also dif fer sharply. Percep

t ions of danger outside of so cie ty a re quali tat ively dif fer -

ent for socia l is t and conservatives. Domestic policies are

generally less apt to produce the kind of radical change in

social inst i tut ions that might be expected by an invasion or

defeat in war --which might be seen as the ultimate social

change, shor t o f revo lu tion. 59

Against these concerns, expectations of policy conse-

quence should provide a basis for predicting how the b iase s

described above should play out among part ies holding views

more or less to the right or l e f t . Some policies , by the i r

very nature, should be more appea li ng than others.

Policy Characteristics

Welfare policies have typically provided the focus of

r esea rch dea li ng with the part isan impact on expenditures.

However, many other kinds of p ublic policy a re redis-

t r ibutive as well . While welfare t ransfer payments may be

immediately redis tr ibutive, heal th and education policies

are potent ial ly more lasting in the ir redist r ibut ive impact.

The public provision of hea lt h s ervi ce s has the p ote ntia l to

redistr ibute the chances of l iving longer. Access to

educat ion prov ides ind iv iduals wi th l i fe- long opportunit ies

with potent ial ly redis tr ibutive consequences. These conse-

59 See Mancur Olson, The Rise and Declin e o f Nation s(New Haven: Yale University Press, 1982), 75-117.



49

quences may be re la ted to the range of pol ic ies tha t might

be endorsed by adherents of one or another ideology. Ideo-

logical considerat ions may also inform the way th at p olic ie s

a re spe ci fi ed .

Pol icy opt ions are not l imited to the questio n o f

whether or not a part icu lar kind of service or benefi t wil l

be provided, but also the terms by which they are made

available , and to whom. Whether a service i s provided on a

means-tested basis , or a non-means tes ted basis i s one such

policy opt ion.

Expenditures on higher education may subsidize those

who might not re ce iv e t he se s er vic es i f they were not pub

l i c ly produced, provided tha t the service i s made available

to everyone, regardless of the abi l i ty to pay. In essence,

expenditures on higher education redistr ibute goods tha t

would otherwise be available a t market prices only to those

members of society who could bear the cost of purchasing

them. These kinds of pol ic ies are rarely conceptua li zed as

being economic intervent ions , but they are. The i ssue of

economic intervention has been, and continues to be, an

issue of continuing controversy in ideological debate.

These concerns are related to whether a policy wil l

provide symmetrical or asymmetrical outcomes. A symmetrical

outcome i s one where a l l members of society share equally in

the re su l t . In an asymmetrical outcome, some members take a



50

la rger share of the resul t , be i t posit ive or negative, than

others . Symmetrical outcomes are less l ikely to change the

socio-economic status quo than asymmetrical outcomes.

Redistr ibutive pol ic ies are defini tely asymmetrical, but may

not necessarily benefi t the lower classes .

The difference is more than whether policies are viewed

as zero-sum or positive-sum. Even with an asymmetrical

outcome, both sides may real ize a posit ive gain, jus t as the

beneficiar ies of a symmetrical outcomes. The dist inct ion i s

whether the "relat ive," or marginal posi t ion of class of

beneficiary is changed vis-a-vis other benef ic iar ies .

The l ikelihood of a policy producing asymmetrical or

symmetrical outcomes has been ident i f ied as a cr i t ica l

factor in determining how tha t policy wil l be promulgated

and implemented. Distr ibut ive pol ic ies , where nearly

everyone gets a piece of the pie , a re cha ract er iz ed as

consensual policies. Conflict is low and implementation

relat ively easy. Redistr ibutive, or asymmetrical policies

are characterized as being more conf li ctua l, e speci al ly when

those policies are implemented quickly.60

60 See Paul Berman, "Designing Implementation to MatchPolicy Situation: A Contingency Analysis of Programmed andAdaptive Implementation," Paper Delivered at the AnnualMeeting of the American Pol i t ica l Science Association, 1978,11; Gordon Chase, "Implementing a Human Services Program:How Hard Will I t Be?" Public Policy 18 (Fall 1979) : 385-6;cf. Robert F. Durant, e t a l , "From Complacence to Compliance: Toward a Theory of Intergovernmental Relations,"



51

The question o f symmetry has part icu lar relevance to

th is study, as i t raises t he que sti on of social equity and

social change, both of which a re r ele vant to notions of

d i st ri but ive ju st ic e .

One of the key dis t inctions made between pol ic ies i s

the difference between those tha t distr ibute goods gener

al ly , tha t i s to almost everyone in the society, and those

poli cie s th at provide goods to a select group. Distr ibut ive

and redist r ibut ive policies have received considerable

scholarly attention as a resu l t of growth in the public

sector . Popular , consensua l pol ic ies , had apparently gotten

"out of control" during the l as t three decades. Theorists

who take this view, l ike Mancur Olson or Theodore J . Lowi,

argue tha t distr ibut ive poli cy a reas tend to be dominated by

in teres t groups and that such programs are not checked by an

ideological brake. 61 A similar argument i s presented by

Glenn Palmer.

Writing on the difference between the pol i t i c s of

domestic policy and foreign policy, Palmer ut i l izes a pol i -

cy-issue typology to ana lyze defense and domestic t rade-

offs . His categorization re l ies on the dist inct ion between

poli cie s th at have a high or low potential for redis-

Administration and Society 17 (February 1986) 433-59.

61 Cf. Lowi, The End o f L iberal ism; Olson, The Rise andDecline of Nations.


http://slidepdf.com/reader/full/parties-ideologies-and-budgets-a-study-of-budgetary-trade-offs 61/236.

52

t r ibut ion. The dist inct ion he employs depends on whether a

government policy provides benefits tha t are divisible and

individually consumed or goods that cannot be easi ly divided

and consumed individually.

For the purposes of this study redistr ibut ive policies

include t ransfe r payments, and services that would normally

be available only through private vendors. Transfer pay

ments are clearly l iable to redis t r ibut ion, though th is is

not a necessary characteris t ic of t ransfer outlays.62

Services are more complicated and require closer inspect ion

to determine whether they are redistr ibut ive.

Welfare t ransfers allows recipients to consume goods

and services which are divis ible and exclusive. With such a

policy, there i s a c le a r po ten ti al for a redis t r ibut ion of

benefi ts among groups of individuals. On the other hand, as

Palmer a rgues, expendi tu re s for public safety and defense

provide services that , at l eas t in theory, are not divis i -

ble. 63 For example, one person's enjoyment of law and

62 According to William D. Berry and David Lowery, II An

Alternative Approach to Understanding Budgetary Trade-offs,"American Journal of Pol i t ica l Science 34 (1990) : 690, "Lib

era l and conservatives each have the i r preferred t ransferprograms and specif ic consumption p re fe rences , but l iberalsand conservatives are unlikely to have di f fe rent preferencesabout the re la t ive u t i l i ty of consumption and t ransfersexpenditures per se, as these two categories of spending areto a considerable extent ' subs t i tutable . ' "

63 See Palmer, "Alliance Poli t ics and Issue Areas, II

191-3, 208.



53

order does not dim in is h ano th er person's enjoyment of the

same good. Likewise, national defense is not con sid ered an

individual ly divisible good and i s generally not considered

to be redist r ibut ive in i t s effec ts . Of course, th ere are

potent ial "side-payments" to specif ic groups under certain

circumstances. This is obvious where employment and govern-

ment contracts are concerned. Palmer asser t s that domestic

benefi ts of mili tary expenditures may be asymmetric under

c onditio ns o f l imited war. 54 But, ge nera lly , th e benef i t

of nat ional securi ty i s not considered to be div is ib le among

ci t izens of a country. The question of whether defense

spending i s properly conceptualized as a benefi t has i t s e l f

s t i r red much controversy.

Some scholars have argued t ha t m i li ta ry spending

fu nctio ns as a surrogate employment policy or as a method of

industrial development, part icularly for developing nat ions.

There has been a l ively debate surrounding the economic

impact of mili tary spending. Many s chola rs b el ieve tha t it

diminishes economic growth by consuming valuable economic

resources without contributing to the supply of marketable

f inished products. 55 Whether defense spending can be gen

era l ly describ ed a s a benef i t i s subject to considerable

doubt.

54

55

To th e e xte nt that one adheres to the theory of

Ibid, 197-8.

See Chan, "The Impact of Defense Spending."



54

deterrence, appropriations for this purpose might be loosely

described as a non-divisible benefi t .

These characterizations of government policies are

subject to certain excep tion s as i l lustrated by the case of

side paYments in defense spending, but some public policies

are more l ikely to result in redist r ibut ive outcomes than

others .

According to Olson, governmentally produced benefi ts

that are divisible are subject to the dynamics of col lect ive

action. Benefits shared within the class of recipients are

divisible and exclusive relat ive to other groups in society.

Generally, the concept of collect ive goods applies to those

goods result ing from the efforts of organized in teres t

groups who receive those goods direct ly . With respect to

welfare policies , one could argue tha t welfare recipients

const i tute a potential interest group in support of in -

creased welfare paYments; however, many people have noted

tha t aff luent people sometimes advocate increasing benefi ts

to the poor. Because welfare recipients are not generally

well organized, some scholars have argued that the bureau-

crats who administer welfare and other social s erv ic e p ro

grams become surrogate lobbyis ts f or those programs.66

66 See William Niskanen, Bureaucracy and Representat ive Government (Chicago: Aldine and Ather ton, 1971 );Fiorina, Congress: The Keystone; cf. Beck, Government Spending, 3-6.



55

For Olson, and many other public choice theor is t s ,

parliamentary bodies are condu it s through which in teres t

groups achieve the i r objectives. Members of parliament do

l i t t l e to mediate or modify these demands in the public

choice perspective. Bureaucracies, in teres t groups and

elec tora l ly motivated poli t ic ians are responsible for ex

panding benefi ts as widely as possible. I f this i s t rue,

ideology shou ld not matter and benef i ts should expand unt i l

the median vote r obje cts . I f ideologies do matter, then

they should help predict shifts in spending for di fferen t

programs. Under what conditions might we expect such a

shif t?

Strategic Opportunities

The opportunity for a government to make major changes

in p olicy i s not present at a l l times. As Machiavelli

noted, luck, timing and good judgment play a part in the

careers of poli t ic ians, but the most important factor i s

power. 67 While i t i s poss ib le to develop hypotheses re -

garding the judgements and timing of pol i t ic ians , good

67 See Niccolo Machiavelli, The Prince and the Dis

courses. With an Introduction by Max Lerner, (New York:Modern Library, 1949) 23-30, 91-4. Machiavell i 's commentsabout changing inst i tut ions might be generalized to thecontext of developing and implement ing changes in policy aswell . For Machiavelli , p roper t iming is an in tegra l par t ofany pol i t i ca l stra tegy. Whether i t i s waiting for the ful lforce of one 's mili tary, planning a counteroffensive, ord is po sin g o f one's henchmen, Machiavelli f requen tly notesthat there are proper and improper moments to take act ion.



56

fortune i s not so easi ly operationalized. What Machiavelli

points out i s that there are more or less opportune moments

to take act ion. This sentiment i s echoed in the work of Ira

Sharkansky. He suggests that , "the availabil i ty of new

federal programs, technological change, or a combination o f

pol i t ical happenings that generate a 'r ip e' l eg is la ti ve

opportunity, may precede major changes. ,,68 Whether a pro-

gram i s cut from the budget or expanded may be a function of

the correlat ion of forces opposing and supporting that

program, as determined by the abi l i ty of the opposition to

ra l ly i t s members.

A weakness on the par t of one party i s the advantage of

the other . P arties in disarray might be circumvented by

more organized par t ies . These conditions offer pol i t i ca l

pa rt ie s s tr at eg ic opportunit ies for changing the shape of

the budget. 69

The s t ra teg ic opportunit ies of a party a re la rg ely

dependent on the number of seats , or legis la t ive votes, that

the party can control . I f coali t ion governments must be

formed, then the power of any part icular par ty i s checked by

68 Ira Sharkansky, Spending in the American States(Chicago: Rand McNally & Company, 1968), 16 ( i ta l ics added).

69 For a more detai led discussion of "strategic opportu nities" in ideological pol i t i c s , see Jane H. Bayes, Ideologies and In teres t Group Poli t ics (Novato: Chandler & SharpPublishers, 1982), 14-7.



57

the members of other part ies in the coal i t ion. In the

United States Congress, achieving par ty discipl ine is a

challenge. Maintaining party unity in congressional voting

is s imilar to maintaining coal i t ions formed in multiparty

systems. Of course there are other constraints .

Structural obstacles may play a par t in providing

strategic opportunities. For example, in president ial

systems, where executive and legislat ive functions are

separated, ins t i tut ional arrangements provide checks on the

power of part ies to con trol po lic y. Controlling both

branches is a potential advantage. In other systems the

pr incipal obstacles may l ie in t he bureaucat ic structures.

In general, democratic policy making is a cumbersome

process. Successful implementation of a policy change often

req uires th e oversight and protection of legis la t ive members

who check on i t s progress and development from time to

time. 70 Without legis la t ive "fixers" who, in conce rt w ith

other a l l ies in government, shepherd i t s development, a

program can die in the post-legislat ive implementation

phase, result ing in reduced outlays for tha t purpose. While

there is considerable variety in the pol i t i ca l p roce ss es o f

th e countr ie s sampled in th is study, I would expect tha t new

program in the areas of public services, and changes in

70 See Eugene Bardach, The Implementation Game (Cam-bridge: MIT University Press, 1984).



58

expenditures fo r e xis ting programs should take a t l eas t one

year , and probably two or more years, to become ful ly imple-

mented. Of course, some programs can be implemented more

rapidly than others. For example, t ransfer paYments may be

increased or decreased almost immediately, but the implemen-

ta t ion of a social services program i s a much more complex

and time consuming endeavor.

Policy making i s not a discrete moment in time, but a

process, which does not t ermina te with the passage of l eg is-

la t ion . Most studies on th e su bje ct of the government's

effec t on spending posit a one-year lag, assuming that what

happened l as t year wil l effec t this year 's outcome. Such a

mechanistic view of policy making ignores the fact tha t many

programs are reviewed, investigated and "pushed" throughout

the l i fe-span of th e polic y. I f neglected during the next

year , there may be not a program in operation.

In the United States, members of Congress are

cont inual ly inves tigat ing the implementation of pol ic ies and

hearing tes timony regarding the i r progress. 71 Perceived

weaknesses in a program's performance might be exploited by

opponents of tha t policy, in Congress or in t he admin is tr a-

t ion , to bring i t to a hal t , or slow i t s progress. Thus, a

71 For an excellent treatment of the subject of par t i san involvment in congressional hearings, see David R.Mayhew, Divided We Govern: Party Control, Law Making andInvest igat ions 1946-1990. New Haven: Yale University Press,1991.



59

commitment at one point in time i s not sufficient for a

program's survival . As Max Weber put i t ,

Poli t ics i s a strong and slow boring of hard boa rd s.I t takes both passion and perspective. Certainly a l lhis tor ica l experience confirms the t ruth - - tha t manwould not have attained the possible unless time andagain he had reached out for the impossible. 72

The idea that the effects of part isan influences should

emanate from a discrete point in time seems untenably naive.

This consideration i s incorporated into the models con-

s tructed in the following s ta t i s t ica l analyses. Exact

specif icat ions are discussed in more d eta i l in Chapters I I I

and IV.

A Theory o f T rade-o ff Behavior

From the foregoing discussion o f id eo lo gie s one can

construct a ta ble th at describes the hypothesized linkage

between part isan ideologies and biases toward certain kinds

of pol ic ies . Based on the previous discussion o f i deolog ie s

and types of public policy, we might expect part ies of the

l e f t and r ight to have different incl inations toward public

pol ic ies . Some ideologies provide more jus t i f ica t ions for

cer ta in kinds of policies than others . 73 A very crude, but

72 Max Weber, "Poli t ics as a Vocation," in Gerth andH.H. Gerth and C. Wright Mills, From Max Weber: Essays inSociology, t rans. H. H. Gerth and C. Wright Mills, (NewYork: Oxford Univer si ty P re ss , 1970, repr in t ) , 128.

73 Cf. Lowi, End of Liberalism, 45-9.



60

reasonable matching of ideological biases toward di fferent

kinds might appear as fo llow s:

TABLE 1

Policy and Ideology

Party Ideology

Policy Issue ] ;&l l . Right

Education Services POSITIVE NEGATIVE

Hea lth Servi ces POSITIVE NEGATIVE

Welfare Policies Posit ive Negative

Unemployment Benefi ts posi t ive negative

Retirement Benefi ts posi t ive negative

Military Expenditures Negative positive

These pol ic ies are ordered in a gradation. At the top

of the l i s t are those kinds of policies which would have the

most enduring effec t on the dist r ibut ion of social benefi ts .

Toward the bottom of the l i s t are items in which

redistr ibut ive consequences dissipate in a re la t ive ly short

t ime, and defense which i s general ly not associated with

redis t r ibut ion a t a l l .

According to the overwhelming scholarly consensus,

part ies toward the l e f t tend to endorse the expansion of

social services. On the other hand, part ies of the r ight

a re charac te ri zed as mil i tar i s t ic and f iscal ly conservat ive.



61

The hypothesized bias of l e f t part ies toward t ransfer pay-

ments must be qualif ied with considerations of whether the

t ransfers are l ikely to be redist r ibut ive. Many programs

tha t re ly on t ransfer payments are not means-tested. That

is , not a l l t ransfer payments are made on the basis of

economic need. For example, many pensions, veterans bene-

f i t s and retirement supplements do not provide money to

lower income people. In fact , one of the frequently ci ted

reasons for the expansion of t ransfer payments has been tha t

they are pol i t ica l ly popular and most effect ive for the

electoral success of incumbent off ic ia ls when they are

distr ibuted as widely as possible . 74

In g en eral, the chart reflects the postulate that the

agendas of le f t leaning part ies dif fer most from r ight

leaning part ies where redist r ibut ion i s a t issue. Conven-

t ional wisdom indicates tha t left-wing part ies favor labor

unions, while right-w in g part ies favor capi ta l i s t or busi-

ness in teres ts . These social ins t i tu t ions provide the basis

for class s t ra t i f ica t ion and social different iat ion in

indus tr i al i zed soc ie t ies .

Accordingly, part ies leaning to the r ight , with an

in teres t in the preserva tion of societal order, would be

more l ikely to promote programs to defend soci al s tr uc tu r es ,

74 See Tufte, Poli t ical Control, 105-136; Lowi, End ofLiberalism; Fiorina, Congress; Berry and Lowery, "An Alternat ive Approach," 690.



62

such as defense and perhaps, police protection. While the

preservation of exist ing social structures i s not entirely

neutral (but pro te ct s exi st ing e l i te ) , there is l i t t l e

evidence tha t defense pol ic ies a re gener all y l iable to

downwardly red is t r ibute payoffs.

Within the l iberal t radi t ion, there are grounds for

leaning s of a l e f t versus r ight nature. In l igh t of certain

t ensions wi th in l iberal ism, i t is possible to discriminate

between a lef t - leaning and right-leaning schools of thought.

These relat ionships might be t ransla ted into the context of

American pol i t ics by examining how l ibera l tensions might be

resolved.

Americans: Right and Left

The terms l e f t and r ight are usually reserved for

discussions of European pol i t ics , while discussions of

American pol i t ics are framed in the language of l iberal

ver sus conserva ti ve , denoting a narrower range o f id eo lo gi

ca l differences in the l a t t e r context . The dimensional

nature of "l iberal versus conservative" disputes within the

l iberal t radi t ion i s approximate to the l e f t versus r ight

differences found in the context o f European pol i t ics , but

fa r more narrow by comparison.

Liberal ism has come under attack from several quarters

during the la s t two decades. Even within the ranks of

l iberal thought, theor is t s have become troubled by the



63

ideology's fai lure to provide coherent responses to emerging

social questions. While many cri t iques of l iberalism have

focused on the problems of equality and fairness , there has

been relat ively l i t t l e attention paid to the his tor ica l

aspect of l iberalism.

Liberalism i s fundamentally grounded in adherence to

"rat ional" principles . In i t s preoccupation with pr incip le ,

l iberalism is ahis tor ica l . Principles, as conceived by

l iberals , do not lend themselves to his tor ica l exception,

but are considered to be expressions of what is universal ,

regardless of time or space.

Liberal principles are frequently expressed in the

idiom of individual r igh ts . Rights are most clearly ar t ic -

ulated in expressing principles that l imi t governmental

author i ty. I n dev ia ting from p rin cip le , o r r at he r v io la ti ng

individual r ights , government subjects i t se l f to sl iding

down one or another "slippery slope."

In c on tra st to socia l is t movements, which seek some

kind of end-state objective, se t in the context of his tor i -

cal change, or conservatism, which i s bound to his tor ica l

experience, l ib e ra li sm f loa ts in the realm of ideas. 75

75 See Manheim, Ideology and Utopia, 219-29; IsaiahBerlin, "Two Concepts of Liberty," in Michael J . Sandel, ed.Liberalism and I t s Crit ics , 15-36; Rawls, A Theory of Just ice ; Ronald Dworkin, "Liberalism," in Sandel, ed. Liberalism and I ts Crit ics , 60-79.



64

What i s missing in the l iberal t radi t ion i s a vision of

history. Without such a v isio n, the re i s no overarching

social aim. Moreover, with i t s extensive array of arguments

in favor of individualism, tolerance, and the sancti ty

representation (pluralism), there is no basis for defining a

h ie ra rchy o f values by which to discriminate between compet

ing pol i t i ca l claims. By default , a few commonly shared

preferences substi tute for intel lectual ly coherent social

objectives . This can be seen in the kinds of debates in

which l iberals engage.

TYPically, debate among l iberals degenerates into

ut i l i ta r ian assertions, which cannot be resolved by appeals

to commonly shared conceptions of the "good," precisely

because there are no commonly accepted defini t ions of what

consti tutes "the good. ,,76 Moreover, many of the fundamen-

ta l tenets of l iberalism are mutually incompatible as James

S. Fishkin has pointed out. 77

Fishkin argues that three tenets of American l iberalism

pose a trilemma, whereby unstab le sequen ti al solutions may

be constructed on an ad hoc basis , but not s imultaneou sly .

76 See Willmoore Kendall, "The 'Open Society' and I t sFallacies ," American Polit ical Science Review 54 (December,1960): 972-9; cf . MacIntyre, After Vir tue, 6-35.

77 See Fishkin, Just ice, Equal Opportunity, and theFamily.



65

ly . The three horns of his trilemma are the principles of

equali ty o f oppor tuni ty , autonomy of the family (in i t s

al locat ions of i t s own resources), and selection on the

basis of merit . Government intervention has the potent ia l

to affec t a l l three principles .

Taxation (for non-essential purposes) cuts into the

autonomy of the family, by taking a share of i t s resources

and allocating those resources for public use. Equality o f

opportunity i s promoted or discouraged by the governmental

intervention in making education available, enforcing ant i -

discrimination regulations, etc . Select ion on the basis of

merit deals with rewarding individuals based on successful

competition. This principle i s supported by market forces,

where individual advancement depends on personal

acheivements. Fishkin argues that a l l three principles are

held to be immutable and universal by the great majo rity o f

Americans --and inconsistent .

In pract ice , t he se val ue s are joint ly incompatible.

Any two, but not a l l three, principles can be maintained a t

the same time. Autonomy of the family and select ion on the

basis of merit may be compatible, but there is no guarantee

of equality of opportunity. Equality of opportunity and

autonomy of the family may be compatible, but not with

select ion on the basis of merit, for the playing f ie ld must

be level led by adjustments to markets. Select ion on the



66

basis of merit and equality o f oppor tuni ty may both be

compatible, but not w ithou t doing violence to the autonomy

of the family, since i t s resources must be allocated to

provide the opportunit ies tha t sustain select ion on the

basis of merit .

Fishkin believes that t he se tens ions might be addressed

sequentially, one or two a t a time, in his tor ical ly mapped

pol ic ies . This remedy, however, must do violence to the

immutable qual i ty claimed by the principles .

An al ternat ive to Fishkin 's suggestion is th at l ib era ls

do not, in pract ice hold to a l l three as immutable pr inc i

ples . While granting that these principles are widely

shared, I believe tha t i t i s typical of l iberals to discount

one or another of these tenets in forming a personal public

philosophy. While the philosophy may require ad hoc solu-

t ions, individuals are noted for ta i lor ing thei r own in ter-

preta t ions of p rin cip le s to f i t thei r personal biases . 78

In doing so, there i s a poss ibi l i ty that one of the immuta-

b le ten ets becomes mutable ( i .e . discounted), via personal

interpretat ion. Moreover, I postulate tha t there i s an

underlying l e f t and r ight dimension in the select ion of

which principle to discount. The key i s the principle of

equality o f opport un it y.

78 See James W. Prothro and Charles M. Grigg. "Fundamental Principles of Democracy: Bases of Agreement andDisagreement." Journal of Pol i t ics . 22 no.2 (1960) :276-294.



67

A posi t ion "to the rig ht" w ith in l iberalism tend to

discount equality of opportunity in favor of autonomy of the

family or select ion on th e b asis of meri t . This reinforces

ex is ti ng soci al inequal i t ies . Liberals fur ther "to the

le f t" tend to discount both selection on the basis of merit

and autonomy of the family in favor of equal i ty of oppor-

tuni ty. This i s the most red is tr ibu t ive a l te rna tive .

Between these posi t ions is a moderate view that discounts

(within some l imits) the autonomy of the family o r s ele ctio n

on the basis of meri t (but not both) in favor of equal i ty of

opportunity. The logical properties might be expressed a a

series of t ransit ive relationships, where t he se t en si on s are

resolved by means of a kind o f d is coun ting , or emphasis.

Consider t he posi ti on s A, B, and C, where A is equality

of opportunity, B is selection on the basis of merit , and C

is the autonomy of the family. One may rank one principle

over ano ther in a t ransit ive ordering of the kind used by

ra t ional choice theor i s t s .79 Six t ransit ive orderings of

these principles are possible (see Table 2) :

79 See William H. Riker and Peter C. Ordeshook, &L

Introduction to Posit ive Poli t ical Theory, ( E n g l e w o o ~Cli f fs : Prentice-Hall , Inc, 1973), 16-9 .



68

TABLE 2

Left and Right in theLiberal Ideology

Rank

Firs t

Second

Third

LEFT

A A

C B

B C

CENTER

C B

A A

B C

RIGHT

B C

C B

A A

Thus, underlying tensions within l ibera l thought might

yield a l e f t / r igh t dimension by vir tue of cer t ian kinds of

interpretat ions, or personal solutions to l iberal ism's

i nheren t t ens ions.

I expect partisan divisions between Democrats, who are

typically described as being to the center- le f t , and Repub-

l icans, who tend to be to the c en te r- righ t, to have an

impact on the way budgets a re a ll ocat ed . By implicat ion,

th is impact should also af fect the shape of t rade-offs

between c atego rie s o f government spending.

The following section deals w ith the concept of a

t rade-off and specif ies i t s operational defini t ion. Drawing

heavily from the work of William Berry and David Lowery, I

present a cri t ique of previous work on the subject and

emphasize the necessity for t rade-offs to be defined in

directional terms.



69

conceptualizing Trade-offs

One of the best cri t iques of t rade-off s tudies i s

presented by William D. Berry and David Lowery. 80 They

raise three object ions to the t rad i t ional approach to study-

ing t rade-offs . Firs t , the l i terature has focused on a

l imited number of potent ia l t rade-offs . As mentioned in

Chapter I , the t rade-off between defense and domest ic expen

di tures i s the primary focus of the trade-off l i terature.

Trade-offs between categories of domestic expenditures may

also be made, but comparatively less at tent ion has been

devoted to choices between categories of domestic spending,

or for tha t matter, different c atego rie s o f spending within

the defense budget. 81

Second, the conventional modeling strategy used in

t rade-off studies does not explain budgetary outcomes. This

is the i r most important cri t ique, for i t points to the

essence of what a t rade-off i s . In short, the use of re-

gression analysis, with one category regressed on another

does not reveal the direction of the t rade-off . Regression

analysis only describes the magnitude, but not the direct ion

80 See also Chan, "Impact of Defense Spending onEconomic Performance," 433-434.

81 See also, Mintz, "A Disaggregated Analysis," whichdeals with the level of aggregation in studies of t radeoffs .Berry and Lowery not that o ther cat egori es of expendituresmay be rel evant sub ject s for analysis, but ignore the aggregation problem.



70

of the t rade-off . Coefficients produced in regression

analysis des cr ib e th e slope of th e r eg re ss ion l ine . A

negative slope ( i.e . s ta ti st ic a ll y significant negative

coeff ic ient) i s interpreted as evidence of a t rade-off .

This method does not include a measure of the actual

division of an identif ied amount between the two categor ies .

A negative coeff icient does not describe whether A gained or

B gained, but rather, whether A and B are generally moving

in opposite directions . 82 For this reason, regressions and

correla t ions between budget categories are not very useful

in analyzing the most important aspects of t rade-offs . A

conceptualization that accounts for both the direction and

size of a t rade-off i s required.

Third, Berry and Lowery point out that "the concept of

a t rade-off is poorly def ined in the l i t e ra ture . ,,83 The

concept of a t rade-off is grounded in the notion tha t an

increase or dec re ase o f one item occurs a t the expense or

benefi t of another item.

A s t r i c t definit ion might require that a l l of the

increase or decrease in budget category A be accounted for

by an equal decrease o r in crea se in another category B.

82 Moreover, when variables are defined in percentageterms, regression analysis can produce spurious resul ts .See William D. Berry, "Testing Budgetary Theories withBudgetary Data: Assessing the Risks," American Jo urnal ofPol i t ica l Science 30 (1986): 597-627 .

83 Berry and Lowery, "An Alternative Approach," 700 .



71

This i s empirical ly and theoretically unlikely. Such a

conceptua li zat ion requi res the assumption that the re la t ion

ship between A and B be independent from changes in any

other budget category, such that funds for A and Bare

subject to subst i tut ion with one another, but not with any

other category. This seems to be an unreasonable assump

t ion, because the thing being allocated is money, which is a

quintessential ly fungible commodity.

Nowhere in the l i te ra ture is such a conceptual izat ion

proposed. Instead, t rade-off ' s are presented as hypotheses.

That is , t rade-offs are not d ef in ed as direct ly observable,

empirical ent i t ies , but rather as educated guesses about the

persistence of a relationship between two or more catego

r ies .

This approach dances around the issue o f exp la in ing why

al locat ions change. The que stion o f whether a t rade-off

exis t s , i f dwelt on long enough, becomes absurd. I t i s

obvious tha t a t rade-off must be made i n a ll oc at in g f ini te

resources among different purposes. What t rade-of f s tudi es

intend to "get at ," i s how the dist r ibut ion of budgetary

al locat ions are determined. Thus, any dist r ibut ion of

r esou rces involves some kind of t rade-off .

Unfortunately, most studies do not direct ly examine the

dist r ibut ion i t se l f . This is the bas ic cont ribu tion that



72

William Berry and David Lowery have made to research in th is

area. They actually look a t the dist r ibut ion i t se l f .

Berry and Lowery ut i l i ze a conceptual izat ion that

re l ies on the ident if icat ion of an amount, or base, which is

sp l i t between two categories.

I n exp la in ing the i r conceptual izat ion, a number of

conside ra tions a re introduced that might impinge on the

defini t ion of the base, which is described as a "pool." The

f i r s t consideration concerns th e n atu re of the budgetary

process. Recognizing that the l i te ra tu r e is inconclusive on

the question of whether budgets are determined in a "top-

down," or "bottom-up" manner, and noting that budgetary

changes are incremental in nature, they identify the "pool"

as the increment o f in cre ase or decrease in the budget.

They fur ther res t r ic t the ident i ty of th is pool as being

that amount of the budgetary increment that i s eventual ly

sp l i t between A and B. This quantity (A+B) is what they use

as the base to measure the trade-off between A and B.

A second considerat ion concerns the sequence of events,

or processes, by which amount of th is pool is determined.

Noting the work of other scholars who conceive of increments

as a set of expectat ions, or "minimum aspirat ions," (about

the expected t ra jectory of spending) for each category,

Berry and Lowery briefly discuss the poss ibi l i ty of using a

"f loat ing pool" as the base.



73

This i s an in tr iguing idea. In brief , th e n otio n i s

that the process of budgeting i s grounded in the ra t ional

expectat ions tha t various agencies have about future levels

of funding. These "aspirat ions" are based on a varie ty of

considerat ions, including the minimum needs of exist ing

programs. Amounts above or below what i s required to main

ta in exist ing program levels i s d efined as the pool to be

sp l i t between programs. They also speculate on the possi

bi l i ty that decisions might be made in a sequence, whereby

minimum al locat ions are made, and what i s "leftover" from an

i n i t i a l al locat ion i s then sp l i t among categories.

Acknowledging tha t the measurement of a "f loat ing pool"

increment i s exceedingly complex, they re t rea t to the more

t ractable defini t ion of a pool as the actual increment of

spending which A and B jo int ly share. Thus, the ident i ty of

the pool i s the sum of increments for A and B. Increments

(INCREM) for each category are determined by subtracting

th is year ' s spending from l as t year ' s spending:

I N C ~ = SPEND t - SPENDt _1 (1 )

Postulating a hierarchical sequence of decisions, which

include both top-down and bottom-up cons iderat ions l eading

to tra de -o ff choices, they suggest that a t rade-off between

the objectives of f isca l policy ( i .e . balancing the budget)

and programmatic objectives i s made s imultaneously with the



74

determinat ion of the to ta l budget increment. At a la t te r

stage in the process a decision regarding the sp l i t between

categories is made. The amount of the increment for each

category (AMT) is calculated in the manner described above

for the to ta l spending increment. The pool i s thus defined

as the sum of these amounts:

POOL = AMT(A) + AMT(B) (2 )

Where:AMT(A) = SPEND (A) t - SPEND (Al t-l'

and, AMT(B) = SPEND (Bl t - SPEND (Bl t - l

I t i s important to note that an increment may be ei ther

posit ive or negative. As Berry and Lowery explain:

While our theory of the budget process assumestha t t rade-off choices are always focused on the a l location of increments from the base, we must allow forthe possibi l i ty tha t decision makers may occasionallycut into the base of one category to finance the increment received by the other84

After considering a number of al ternat ives for measur-

ing the sp l i t between two categories, they se t t l e on two

methods. The f i r s t method i s based on a ra t io between A and

B. This proportional measure i s calculated as follows:

PROP(A:Bl= (3)[AMT(Al-AMT(B)]/[AMT(A}+AMT(Bl], i f INCREM(A+Bl>O; or[AMT(Bl-AMT(A)]/[AMT(A)+AMT(Bl], i f INCREM(A+B)<085

A second measure i s based on the differences between

the c urren t year 's expenditures and expenditures the year

84 Ibid, 684.

85 Ibid, 685.



75

before. Noting that the absolute d if fe rences a re extremely

large, they adjust thei r measure by use of square roots,

thus:

DIFF(A:B)= (4)~ A M T ( A ) - ~ A M T ( B ) , i f AMT(A»O and AMT(B»O, orIAMT(A) I - IAMT(B) I, i f AMT(A)<O and AMT(B)<O, or

~ A M T ( A ) + ~ IAMT(B) I, i f AMT(A»O and AMT(B)<O, or- [ IAMT(A) I + ~ A M T ( B ) ] , i f AMT(A)<O and AMT(B) >0 86

These calculat ions resu l t in measures of t rade-offs

that are in te l l ig ible in terms of both direction and magni-

tude. Instead of conceptualizing t rade-offs as hypotheses,

these theoret ical cons truc ts p resen t t rade-offs as measur-

able ent i t i es . With such a conceptualization, i t i s possi-

ble to t es t hypotheses about the direction of the t rade-off .

This kind of t es t is simply not possible using the conven-

t ional method of analysis where one category i s regressed on

another.

The General Hypothesis

The gen eral h ypothesis o f th is study is tha t American

par t ies affec t budget allocations , and thereby, t rade-offs

between categories of expendi tu re s. I hypothesize tha t

par t ies further to the l e f t favor social spending more than

par t ies further to the l e f t and tha t part ies further to the

r ight favor defense spending more than part ies to the l e f t .

r es ultin g in distr ibut ions that ref lec t the power of the

par t ies .

86 Ibid, 686.



76

H1 : Trade-offs f avor defense spending more as the strengthof the r ight increases.

For reasons explained below, sufficient ly precise data

for trade-off measures based on the Lowery-Berry formula

could not be constructed for cross-nat ional analysis .

Difficul t ies in compiling data sufficient ly precise for

use in the calculat ions of small margins required tha t the

measurement of trade-offs be modified f or c ro ss -n at iona l

analysis . The following sections deal with the way tha t thedata were compiled and transformed into variables sui table

for analysis . Following the descript ion of the sources and

methods of measurement, an inspection of the variables i s

presented. Based on the findings of these inspections,

modifications to the in t ia l formulation are presented.

Data Collect ion and Measurement

To t es t the hypothesis presented in th i s study a number

of data sources were consulted. A complete l i s t of sources

i s reported in Appendix A. This sect ion describes the

calculat ions by which the principal concepts were opera-

t ional ized.

Two separate sets of data were constructed. One set

contains cross-national data for 18 advanced indust r ia l

countries. The second data se t contains more detai led and

precise data on United States.



77

In construct ing the data se t for the cross-national

analysis, a number of obstacles were encountered. The f i r s t

problem i s comparability. Definit ions of categories di f fe r

from one country to another, thus there is some conceptual

ambiguity in taking data from national yearbooks. By com-

parison, data reported by in ternat ional organizat ions such

as the Organization for Economic Development and Cooperation

(OECD), or for that matter the International Monetary Fund

(IMF) are gathered with a special eye t o c ro ss -nat iona l

comparison.

Highly disaggregated, such as welfare paYments to

select groups of people, are frequently incomplete and not

directly comparable to that of o ther coun tr ie s. Moreover,

expenditures on welfare programs and other cash t ransfer

programs do not appear to be comparable in terms of the

requirements of e l ig ib i l i ty . Considerable differences

between welfare programs (some of which are means-tested

while others are not) in different countries seemed to

s t re tch the comparabi li ty o f expendi tu res in th i s policy

area. On the other hand, functional expenditures for educa

t ion and health appear to be more comparable. For th is

reason, I re ly on aggregate data on expenditures for educa

t ion and health .

Unfortunately, the precision (and re l iabi l i ty) of

cross-nat ional data i s not very high. As i t stands, I have



78

ra ther good data on 18 countries of a theoret ical universe

of approximately 24 to 25 countries (depending on how one

views the level of development and comparative democratiza-

t ion of countries in Southern Europe, e.g. Greece, Turkey,

Spain and Portugal, which were deleted from this study).

Also, I have excluded very small countries, such as Liech-

tenstein and Iceland, which, for reasons of comparability,

did not seem appropr ia te subject s of th is a na ly sis .

The principal source for data on expenditures for

health and education expendi tures is the OECD publication

Social Expenditures, published in 1985. 87 Series fo r most

member countries a re a va ila ble for the 1960 to 1981 time

frame, though a good many points are missing (e.g. Denmark,

Switzerland, and of course France). Additional data were

taken from the OECD publication National Accounts, Volume I I

(various years) . The definit ion of health and education

expenditures were roughly comparable between these two

sources. Thus, the ear l ie r series were spliced to the

l a t t e r ser ies . The spl ic ing technique was based on percent-

age increases or decreases in constant units . 88 A few

87 In personal conversat ions wi th the s taf f of theOECD in Paris , I have been to ld th at an update of th is datawil l be available from the OECD in 1992 or 1993.

88 There i s some error in this method, owing to thefac t that the ear l ie r series had been harmonized, while thel a t t e r ser ies was based on f i sca l years, which dif fer fromcountry to country . .



79

missing data were averaged from exist ing points, where that

seemed reasonable. Otherwise, missing data were simply

coded as m issin g. Percentages of government expenditures

were then calculated for each category.

Two separate sources for defense expenditures data were

consulted. The f i r s t source was taken are taken from a data

se t of IMP s ta t i s t ics provided by Professor Marina Arbetman.

This data i s reported in per centage s o f Gross National

Product (GDP) and converted to per cent ag es o f government

spending using OECD data.

Another source of data for mili tary spending was con-

sul ted, World Military Expenditures and Arms Transfers

(various years) published by the United States Arms Control

and Disarmament Agency (ACDA) provided a second source of

data. ACDA figures were reported in percentages of Gross

National Product (GNP) for the years between 1963 and 1988,

and defense expendi tu res as a percentage of Central Govern-

ment Expenditures (CGE) for the years between 1967 and 1988.

Earl ier edit ions in cluded per centage s o f GNP, but not

percentages of central government expenditures. Data from

the OECD on the l a t t e r were used to back-f i l l estimates to

1964, based on the ra tes of increase of both indicators in

constant units . This procedure did not reso lv e th e discrep-

ancy between ACDA est imates and the est imates based on IMF

figures. Selecting a rel iable data source for defense



80

spend ing posed serious problems. Considerable differences

between dat a r epor ted in subsequent issues of the ACDA

publication were encountered. The l a t es t f igures were

20btained from each issue, as they reputedly contained the

bes t a va ila ble information. These d if fe renc es a re examined

in more deta i l at the end of this chapter.

For c ros s-nat ional analys is , c a lcu la tions of t rade-offs

between defense expenditures and other expenditures on other

expenditures were based, in i t i a l ly , on a modification of the

formula p roposed Ber ry and Lowery using percentage t erms . 89

These measures are posi t ive when the sp l i t in the pool

favors defense and negative when the sp l i t favors domestic

or social expenditures. The in terval of th e d if fe renc e

measure has no boundary except for the zero point, where the

sp l i t i s even. Dependent var ia bles a re constructed for

t rade-offs between defense expenditures and health and

education expendi tu res .

89 Obviously, measures based on monetary units wouldhave created a severe problem in t he c ros s-nat ional analys i s . Conversions to common units would not have solved theproblem of comparing budgets that di f fe r a great deal ins ize . While there are s t i l l problems associated with di f -ference in bases (denominators) from one country to another,

percentage terms provide the best al ternat ive to a currencybased approach. In selecting an appropriate indicator, Ichose the difference-based model.

The proport ion based measure generates more extremeerrors than the simple difference based measure. Moreover,the proport ion based measure i s more sensi t ive to the basethan th e d if fe renc e measure, which would further complicatecross-national analysis comparisons. This matter is dis-cussed in more deta i l in Chapter IV.



81

Measures of partisan ideology were taken from a data

set developed with the assistance of several country

specia l is ts .gO The measurement of t he i deologi ca l composi-

t ion of governments are based on the placement of a l l pol i t

ica l part ies holding parliamentary seats on a five point

l e f t - r igh t continuum. A series of revisions to an i n i t i a l

collection were made for the f inal computations used in th is

study. The scale position of pol i t ical part ies were based,

in i t i a l ly , on a comparison of descriptions found in Vincent

McHale's Pol i t ica l Part ies of Europe, Keesing's Contemporary

Archives, Facts on File and Mackie and Rose's The Interna-

t ional Almanac of Electoral History. As mentioned, a number

of country special is ts , contributed revisions to the or igi -

nal placements. In revising the original placements, con-

t r ibutors were allowed to move the placements of part ies

over time to account for some degree of ideological dr i f t .

These placements were used to calculate ideological scores

gO I would l ike to acknowledge the advice and ass is tance of a number of contributors to th is port ion of the

data se t . The collect ion is part of a jo int project amongseve ra l scho la rs including Professors Marina Arbetman,Francis G. Castles, Robert Cox, Eric Einhorn, ConstantineDanopolous, M. Donald Hancock, John Logue, John williams,and Barclay and Joan Ward and David wilsford. ProfessorCharles F. Delzelle , though not a con tr ibutor t o the broaderproject , a ss is te d in the collect ion of data on I ta l ianpar t ies . These data are to be used in country-specificstudies of the influence of part ies on budget outcomes.



82

based on the dist r ibut ion of seats in the lower chamber of

parl iament. 91

The number of seats held by each party were also ob-

tained from the sources above. The occurrence of mid-year

elections posed some problems for the annualized ser ies . To

provide consistency, distr ibut ions of seats resul t ing from

elect ions held af te r June 1, are credited to the next year.

Changes due t o e le ct io ns held before June 1 are credited to

the year in which the elect ion was held. Lags are based on

these estimates.

These dist r ibut ions were used to construct two indica-

tors of part isan influence. There i s no general ly agreed

upon coding scheme for analyzing th e in flu en ce of par t ies ,

but most measures depend on aggregating the ideological

posi t ions, weighted by the proportion of seats held by

par t ies . Some schemes base calculations on cabinet seats ,

while others base calculations on the dist r ibut ion within

the lower chamber, reg ardless o f cabin et representa t ion.

Posit ive empirical resul t s have been obtained using both

kinds of indicators. Without an expl ic i t basis for asser t -

ing the super ior i ty of one method or another, I constructed

two measures. One measure is based on a s imple averaging

91 See Appendix B for a sample o f th e q ue st io nn airesent t o con tr ibutor s.



83

technique and the other is anchored in the posi t ion of the

category that has a plura l i ty of seats .

The f i r s t indicator, i s based on a weighted, or

geometric mean. To calculate this score, each of the posi-

t ions was assigned a numerical value ranging from 1 to 100,

with radical l e f t occupying 1, moderate l e f t a t 25, center

at 50, moderate r ight a t 75 and radical r ight a t 100. The

number of seats held by part ies at each posi t ion i s mult i

plied by the assigned value of that posi t ion and summed. 92

This sum is then divided by the number of seats to provide a

weighted mean. The formula for these calculations i s as

follows:

= (5)[(RL X l)+(ML X 25)+(CTR X 50)+(MR X 75)+(RR X 100)]

SEATS

Where:RL=ML=

Number of seats held by radical l e f t part ies

Number of seats held by moderate l e f t par t ies

92 This method corresponds with the procedure used byMichael John Morgan, "The Modelling of Governmental Coalit ion Formation: A Policy Based Approach with Interval Mea-surement" (Ph.D. Diss. , Univers it y o f Michigan, 1976); Seealso Francis G. Castles, and Peter Mair, "Left-Right Pol i t i -cal Scales: Some 'Expert ' Judgements," European Journal ofPol i t ica l Research 12 (1984): 73 -88. The method i s employedby Palmer, "Alliance Poli t ics and Issue Areas," 203-205.Palmer notes tha t "there i s no account o f bargaini ng power.

This may be especially troublesome in s ta tes where minorpart ies a re p ivot al, that i s , they can determine which ofthe major part ies i s a member of the rul ing coal i t ion . . . . Atechnique was experimented with that weighted the par t ies byhow many cabinet seats they held, or al ternat ively, how many' pres t ige ' positions they held (e .g. , foreign ministry,defense ministry). Such weighting systems provided l i t t l ed iffe re nc e in the value of the coali t ion mean and was dis-carded." [Ibid, 204].



84

CTR = Number of seats held by center part iesMR = Number of seats held by moderate r ight part iesRR = Number of seats held by radical r ight part iesSEATS = Total number o f par li amentary seats

A second scale, AMEAN, was constructed, which anchored

the sca le in the s ca le po si ti on having a plura l i ty of par-

liamentary seats . This scale was calculated as follows:

AMEAN = ANCHOR +/ - COMP (6 )

Where:

ANCHORWhere:

COMP =

Where:Where:

= The position containing a plura l i ty of seats

The weighted value of the scale position (B)containing a plura l i ty of remaining seats ,calculated as:[POS X (XPER), i f POS > ANCHOR; or

-[POS X (XPER), i f POS < ANCHORXPER = Percentage seats in sca le pos it ion BB<A, and B>C, and B>D, and B>E; and C, D, and Eare the remaining s ca le po si ti on s

For the sake of consistency in cross-nat ional analysis,

these measure were applied to a l l countries in the sample.

However, the comparative lack of party discipl ine in the

United States, l imits the appl icabi l i ty of using percentages

of seats as a measure of part isan s t rength.

For a more p rec is e ana ly si s of t rade-offs in federal

outlays, a surrogate measure of partisan ideology was de-

vised. Following the work of Keith Poole and R. Steven

Daniels , I used measures of party unity voting to est imate

the ideological effec t of part ies . 93 Poole has argued tha t

93 Other researchers have analyzed part isan cleavageswith similar resu l t s . See Schneider, Ideo log ical Coal it ionsin Congress; W illiam R. S ha ffe r, P arty and Ideology in the



85

a l iberal-conservative dimension in Congressional voting

approximates party unity voting. 94 Using this conceptual-

izat ion of the ideological strength of American part ies ,

unity data were taken from Norman J . Ornstein, Thomas E.

Mann, and Michael J . Malbin's Vital Sta t is t ics on Congress.

I pos tu la te th at the higher the percent of ro l l -ca l l

votes where a l l the members of one party voted tog eth er, the

higher the degree of ideological consistency within that

par ty. 95 Thus, the effect of the ideological tendency of a

United States Congress (Washington, D.C.: University Pressof America, 1980); Poole and Stevens' argument tha t congressional voting is highly unidimensional has been challengedby Kenne th Korford, "Dimensions in Congressional Voting,"American Pol i t ica l Science Review 83 (1989): 949-62, whoargues that the lef t - r ight dimension accounts for much l essthan Poole and Stevens' analysis repor ts . He argues thatth ere are at leas t three major dimensions to congressionalvoting.

94 See Keith Poole and R. Steven Daniels, "Ideology ,Party and Voting," 380.

95 This i s a ten ta t ive proposit ion. Some evidencesuggests that exogenous factors influence levels of par t i sanship, especial ly in the House. Cf. David Brady, JosephCooper, and Patr icia Hurley, "The Decline of Party in theU.S. House of Representa tives, 1887 - 1968," in Congress:Structure and Policy, eds. Matthew D. McCubbins and TerrySullivan (Cambridge: Cambridge University Press, 1987), 250257; Samuel C. Patterson and Gregory A. Caldeira, "partyVoting in the United States Congress," Brit ish Journal of

Pol i t ica l Science 18 ( January 1988) : 117-29.Data on par ty unity are taken from Norman J . Ornstein,Thomas E. Mann, and Michael J . Malbin, Vital Stat i s t ics onCongress 1989-1990 (Washington, D.C.: Congressional Quarterly Inc. , 1991); see Tables 8-3,8-4; 208-209.

The measures are defined: "Party Uni ty Votes in Congress, 1953-1988; (percentage of a l l votes)": "Data indicatethe percentage of a l l recorded votes on which a majo rity o fvoting Democrats opposed a majo rity o f voting Republicans."



86

party should be more pronounced when it votes en bloc.

Holding tha t Democrats are to the l e f t of Republicans,

re la t ive ly speaking, I hypothesize that the more tha t Demo-

crats vote together, the more t ha t t ra de -o ff s should move

away from a defense and favor various kinds of domestic

expenditures.

A very l imited analysis of the re la t ionship between

evaluations by the Americans for Democratic Action (ADA) and

par ty unity suggests that p ar ty unity votin g ta ps into the

ideological dispositions of the part ies . Average ADA scores

for House Democrats and Republicans for the years 1965 to

1976 were taken from William R. Shaffer 's study of dif fer -

ences between congressional party members. 96 The resul ts

[Ibid, 208]; "Party Unity Sco re s in Congressional Voting,1954-1988; (percent)": "Data show percentage of membersvoting with a majority of the i r party on par ty uni ty vote s.Party unity votes are those ro l l calls on which a majorityof a party votes on one side of the issue and a majority ofthe o th er p arty votes on the other side. The percentagesare normalized to eliminate the effects of absences, asfollows: par ty unity=(unity)/(unity + opposit ion)." [Ibid,209].

96 See Shaffer, Party and Ideology in the united

States Congress, 49. For a description of the l imitat ionand cri t icisms of ADA scores, see Congressional Ouarterly'sGuide to Congress (Washington, D.C.: Congressional QuarterlyInc. , 1985) and Ornstein, Vital Stat i s t ics on Congress.This index of congressional voting tendencies has beencr i t ic ized for i t s lack o f conceptual clar i ty , but remainsone of the leading sources o f s yst ematic data on members ofcongress. Repeated effor ts to obtain annual data direct lyfrom the ADA were f rui t less .



87

of a l imited analysis are reported in the next section,

where measurements are inspected more closely.

Data on federal expenditures are taken from Histor ical

Tables, Budget of the United States Government, for f i sca l

years 1987, 1990 and 1992. These data are organized by the

superfunc tion , func tion and subfunction. All data were

converted to constant 1'982 dollars prior t o c alc ul ati ng

t rade-off measures. The spec if ic t rade-of fs analyzed using

th is data are discussed in deta i l in Chapter IV where t rade

offs in the United States are examined.

Inspect ing the Pol i t ica l Variables

The use of a l e f t - r igh t continuum is fam iliar to jour-

nal is t s , historians and pol i t ical sc ient is ts . Moreover,

empi ri ca l r esearch suggests tha t the public understands the

terminology of an ideological continuum arrayed from l e f t to

r ight . The practice of using a Left-Right terminology to

describe the p olicy orientat ions of pol i t ic ians i s quite

old, dating from the French Revolution.

This approach to c la ss if yi ng pol i t ical part ies has been

widely used and empirical studies indicate that the continu

um has meaning to a large p ropo rtio n o f th e populatio n. 97

97 See Hans Daalder, liThe Dutch Party System: FromSegmentation to Polarization- And Then?1I chap. in PartySystems in Denmark, Austria, Switzerland, the Netherlandsand Belgium, ed. Hans Daalder (London: Frances Pinter, 1987)where cross-national survey resu l t s are reported th at in dicate an underlying s c al ar p ri nc ip le in voter self-placements



88

Moreover, the concept of a l e f t - r igh t continuum appears to

have val id i ty across cul tu ra l and pol i t ica l boundaries. As

Sani and Sar tor i point out:

As it stands, our evidence only says tha t , in a l l thecountr ies under consideration, samples of respondentsa re w illin g and apparen tly able to locate themselves ona l e f t - r igh t continuum in sat isfactory percentages(67.6 i s the lowest percentage of responses in theEight-nat ion Study, and a surprisingly high one considering tha t it i s th e U .S .). The important underlyingquestions are, f i r s t , what is the c ro s s- cu lt ur al t rav el l ing capacity of the l e f t - r igh t imagery; and, second,

how do we reckon with the i r inevi table, context-basedre la t ivi ty? Cross-cultural equivalence i s a generalproblem of a l l comparative analyses; yet we seem to bebe t te r off -on t ravel l ing grounds - with ' l e f t - r igh t 'than w ith ' l i bera l - conservat ive ' . As to the i ssue ofre la t iv i ty - how r ight i s ' r i gh t ' - it should be c leartha t in our analysis the problem i s minimized by thefac t tha t the primary concern i s the comparison ofgroups of part isans within the i r respect ive countries.Thus, no assumption needs to be made about equivalenceof the l e f t - r ight scale in different countries. 98

Sani and Sartor i succinct ly s ta te the essent ia l prob-

lem of using the l e f t r ight continuum to make generaliza-

t ion s c ro s s- na ti ona ll y. The fundamental question i s whether

r ight in country A i s equivalent to r ight in country B.

The way tha t the continuum i s being used in th is analysis

r0quires th is kind of comparison. More extensive collec-

t ions of th is kind of data have resul ted in very high levels

on a l e f t - r igh t continuum.

98 Giacomo Sani and Giovanni Sartor i , "Polar izat ion,Fragmentation and Competition in Western Democracies,"Western European Party Systems, ed. Hans Daalder and PeterMair (Beverly Hil l s : Sage, 1983), 309.



89

of re l i ab i l i ty , but some kind of independent t es t seems

warranted by the comparat ively smal l number of contr ibutors

to the data collect ion. 99

Some anecdotal evidence suggests tha t the placements of

part ies on the continuum used here are cross-nat ional ly

val id . Placements of European part ies on the continuum

correspond closely to the par t ies ' pre fe rr ed seat ing ar

rangements the European Parliament. This indicates some

symmetry between the values ass igned to part ies in terms of

thei r selections of coali t ion partners .100 Comparisons

with the s trength of labor unions a lso d isp la y a pat tern.

Labor unions are considered the natural constituency of

l e f t i s t par t ies . 101 Where labor unions are strong, l e f t i s t

part ies should also be strong. Regression analyses o f labor

organization strength display a fair ly consistent pat tern in

99 Cf. Francis G. Castles and Peter Mair. "Left-RightPol i t ica l Scales."

100 See Stanley Henig, ed. Pol i t ica l P arties in theEuropean Community (London: George Allen & Unwin Limited,1978), 276-277; Francis Jacobs, Richard Corbett, and MichaelShackleton, The European Parliament (Boulder: Westview Press1990), 54-83.

101 See Hibbs, "Poli t ical Part ies and MacroeconomicPol ic y; " Tuft e, Pol i t ica l Control, 83-7. In fact , i t i scommon to find high col l ineari ty in models tha t include bothle f t -par ty control and measures of unionization or unionstrength. Cf. Michael R. Alvarez, Geoffrey Garrett , andPeter Lange. "Government Partisanship, Labor Organization,and Macroeconomic Performance," American Pol i t ica l ScienceReview 85 (June 1991): 539-556.



90

the categor ical components of the indices WMEAN and AMEAN,

as well as th e in dice s themselves. Using th e p er cent of

moderate l e f t seats averaged over the 1960-1988 period as

the dependent var iable and union organization strength102

as the independent variable, a pos it iv e r e la ti onsh ip i s ob-

ta ined. For moderate r ight par t ies , the relat ionship i s

weaker, but n eg ativ e as expected. Both aggregate measures,

AMEAN and WMEAN, produced s ignif icant resul t s as well , with

coeff icients in the direct ion implied by the l i t e ra ture (See

Figures 1-4) :

A-SqulII'8 =.59, SiGnificant at p < .000.

50 50...,cIIII

-L

II IC L_ lD

D. 30~ ~ lD L

1i;1 20LO

-8 ....02:...,

10Iii

0L..

0 0 0

1.21.41.61.8 2.22.42.62.8 3.23.43.6lSIar IfI!eXLabor Strength and Percent Moderate Left

FIGURE 1

102 Ibid, 553.

.



91

•

•

•

o

•

•

•

•

•

•

so

A U U ~ U U ~ U D U ~ l 8 b a r ~ . Index

Strength and Percent Moderate Right

FIGURE 2

R-Sql8'e =.G , S:lQniftClnt at p < .001



92

~ =.31, --.flCIftt lit P < .01

70

I 60 00

~ 0

•0

at0

0 0i ....,0

-I...

3D

The question of whether party unity in the U.S. House

of Representatives i s indicative of a par tis an b ia s is more

complex. I postulated that part y unity is indicative of a

kind of coalescence of members around core values. I t seems

reasonable to believe that when a party is united in i t s

voting patterns i t i s displaying unity, or agreement about

i t s values.

The fact that a party varies in i t s cohesiveness, by

i t se l f , says nothing about the d ir ec tion o r i nte ntion of the

bias. I am not aware of any use of party unity to measure

party ideology. However, several p iece s of evidence suggest

tha t part isan unity i s related to ideology. The most sug-

gest ive piece of evidence i s found in an ar t ic le by Keith



93

Poole and R. Steven Daniels on party voting in the u.S.

Congress .103

In the i r analysis of rol l -ca l l voting, Poole and Dan

ie ls as sert th at there is a high degree of unidimensionality

in congressional voting. Using factor analysis, they argue

tha t , on average, a l iberal-conservative dimension accounts

fo r 81 percent of the variance in the in teres t group rat ings

of congressmen. 104 Moreover, they note that :

[T]here are dis t inctions between part ies : the Democratic party i s much less homogeneous than the Republicanpar ty. The Republicans are concentrated a t centerrigh t to far r ight . The bulk of the Democratic partyi s concentrated a t center l e f t to fa r le f t , but subs tant ia l numbers of Democrats are located a t centerr igh t and far r ight .

Party unity scores increase with movement along theparty dimension . . . 105

Basing thei r assessment of an ideological dimension in

congress ional voting on th e r atin gs of in teres ts groups,

ing. "

103 See Poole and Daniels, "Ideology, Party and Vot-

104 Ibid, 379. For a cri t ique of the Poole and Danielsar t i c l e , see Kenneth Korford, "Dimensions in CongressionalVoting." Korford issues a methodological c ri ti ci sm tha t thetechnique used by Poole and Daniels overstates the dimen

sional i ty in voting.

105 See Poole and Daniels, "Ideology, Party and Voting," 380-1, for a visual i l lus t ra t ion of the s pa tia l r el at ionship of par ty unity and the l ibera l conservative dimension. In thei r i l lus t ra t ion, the spat ia l relat ionship i snot presented as para l le l dimensions, but d ispla yed as twol ines, crossing at a 20 to 30 degree angle at an zero point[Ibid, 380].



94

Poole and Dani el s' fi nd ings f i t well with the work of other

scholars who find that differences between par t ies on in ter -

est group ra t ings.

The most widely used in teres t group ra t ing i s reported

by the in teres t group, Americans for Democratic Action

(ADA). In ADA rat ings, members are scored in accordance

with the percen tage o f votes where th ey vot ed in favor of

b i l l s supported by the ADA. ADA scores may be interpreted

as a measure of l iberal ism, though there i s some scholarly

skepticism regarding the content of ADA evaluations. 1oG

High scores indicate agreement with l ibera l values, w h i l ~low scores indicate relat ive conservatism.

William Shafer has conducted an extensive study of the

difference in ADA scores between Democrats and Republicans.

He has found tha t average scores for both par t ies indicate

tha t there i s a difference between the Democra tic and Repub-

l ican par t ies . Shaffer found tha t Democrats tend to be more

l ibera l as measured by ADA scores, while Republicans tend to

be comparatively more conservat ive. While regional dif fer -

ences between scores are apparent, within each region Demo-

crats are s t i l l more l ibera l , on average, than Republicans.-

lOG See Sha ffer, Party and Ideology, 31-36 for a discussion of ADA scores general ly and as a measure of ideologyin par t icu lar . Cf. Schneider, Ideological Coalit ions inCongre ss , 102.



95

107 Shaffer 's study is based on ADA scores from 1965 to

1976. Shafer reports average scores for each party in each

chamber of Congress (and regional breakdowns by region) for

each year . Using th is data, I t es t the simple hypothesis,

implied by Poole and Daniels, that par ty unity is associated

with ideology. The hypothesis i s twofold. I f unity is

associated with ideology in the way that I suspect, Demo-

cra t ic party unity should be pos i t ively associated with ADA

scores, and Republican par ty unity should be negatively

associated with ADA scores.

With only twelve years of data, test ing the re la t ion-

ship between part isan unity and ADA scores stretches the

val idi ty of s t a t i s t i ca l analysis , so resu l t s are very ten ta-

t ive . Nevertheless, bivar ia te r eg ress ion analys is of the

relat ionship between the average ADA scores for House Demo-

cra ts , and the p ercen t o f ro l l ca l l votes where party unity

obtained, yields a s ta t i s t i ca l ly s ignif icant , posi t ive

coefficient . This f inding supports the hypothesis that

par ty unity i s indicat ive of an ideological push toward more

l ibera l posi t ions when Democrats vote in unison. D isunity,

on the other hand, i s associated with lower levels of ADA

scores, indicat ing that party members, on average, are not

voting in favor of positions endorsed by the ADA. The f i t

of this bivariate model i s shown in Figure 5.

107 See Shaffer, Party and Ideology.



96

80

• 75

64.1

•

•

•

•

•

Lr-----r------,r-----y---.,-. 70

80

.... DlIIar:NtIRelationship Between ADA SCores and party U'1i.ty

>-..§>0

-\ICL

I.7!i

>

'5

I.

-!CL

FIGURE 5

A structural ly identical model for Republican House

members is tested. This analysis yields a weaker, but

s ta t i s t ica l ly significant , negative relat ionship (Adjusted

R2=.35, significant a t p<.02). That i s , the more that

Republicans House members are united in the i r voting, the

lower the i r average ADA scores, while the more united Demo-

crats are, th e hig he r the i r ADA scores. This supports the

hypothesis that party unity is associated with the ideologi-

cal leanings of Republican House members (See Figure 6).

These findings are encouraging. Both par t ies move in

the directions that I hypothesized as they become more

united in the i r voting.



97

85

I'Ii --

I

80

7!1

••

•

• ••

7Q 10L . w - - - - _ - - - - . , . - - - ~ r _ _ - - - ~

"

IS a tHue- AlpubltClllll"

Relationsh1p Between ADA" Scens Rt P8"ty lkI1ty

FIGURE 6

These fi n d i n g s s u g g e s t t h a t changes in u n i t y v o t i n g a re

i n d eed i n d i c a t i v e o f movement toward, o r away from, c e r t a i n

k i n d s o f v al u e s . The d i r e c t i o n o f th e r e l a t i o n s h i p s between

u n i t y v o t i n g by b o t h p a r t i e s and ADA s co res a r e a s I had

p r e d i c t e d . In t h e co n t ex t o f American pol i t ics , Democrats

a r e more l ibera l , i n a manner o f speaking lito th e le f t" as

compared t o Republicans. In sum, t h e p o s t u l a t e o f th e

i d e o l o g i c a l d i m en s i o n al i t y o f p a r t y u n i t y i s supported by

comparison to t h e interest group r a t i n g s o f th e ADA. As

Democrats become more u n i t e d i n the i r v o t i n g b eh av i o r, t h e

more l i k e l y they a r e to s u p p o rt t h e l ibera l p o s i t i o n s o f t h e



98

ADA, while the more united Republicans are, the less l ikely

they are to support those same positions.

Inspecting the Dependent Variables

The re l i ab i l i ty of cross-national data posed a serious

problem in construct ing the t rade -o f f va riable . The formula

proposed Berry and Lowery is quite sensi t ive to year to year

change. Modifying the t rade-off formula to employ percent-

age terms magnifies error in year to year changes.

Seemingly small errors in reports of percent expendi-

tures generate large errors in the es timates o f the t rade-

off . Inconsistencies in the data reported by the U.S. Arms

and Disarmament Agency raises serious doubts about the

ac cu racy o f that data. Corrections to prev iously report ed

figures varied widely from one issues of World Mili tary

Expenditures and Arms Transfers to the next . Figures on

West Germany provide the best example of how much "correc-

t ion" was necessary in some issue. Estimates of the per-

centage of central government expendi tu res spen t on the

mili tary differed from between 10 to 15 percent!

On the other hand, IMF f igures appeared to be less

vola t i le in thei r correct ions to ear l ie r reports . Checking

raw figures, I noticed very l i t t l e change from one volume to

the next . Generally, corrections were confined to a two-

year span prior to publication in IMP publ ica t ions . By

comparison, ACDA f igures are continually re-based, and



99

est imates of current dol lar expenditures vary over ent ire

ten-year spans from one issue to the next with no explana-

t ion other than the blanket asser t ion that est imates are

based on the most accurate data available. Suspecting that

th is data might contain an unacceptable amount of error , I

designed a t es t to confirm or re jec t my suspicion that IMF

data are superior to ACDA data.

To do th is , I f i r s t converted U.S. budget data to

percentages in order to repl icate the procedure employed in

calculat ing the modified t rade-off measures to be used in

the cross-national analysis . U.S. budget data are much more

precise, less l iable to revision, and subject to consider-

ably more public scrutiny than the data reported by ei ther

the IMF or the ACDA. For these reasons, I believe that the

data source which f i t s be t te r with calculations based on

U.S. budget data i s the more re l iab le source.

Measures of defense versus health and education t rade-

offs were repl icated appeared to be the most t ractable

t rade-offs to repl icate. I compared measures using a l l

three data sources for these two t rade-offs . I also com-

pared modified and un-modified calculations using only U.S.

budget data to see whether the effect of modifying the

Berry-lowery formula to adapt i t to percentage calculations

warrants concern.



100

The correla t ion between the modified t rade-off measure

the unmodified t rade-off measure i s .85 fo r the education

t rade-off , and .87 for the heal th t rade-off . Comparisons of

the series for the heal th t rade-off are shown in Figure 7:

o Unmodified Calculation A Modified calculation

1960 1970 1980

U.S. Health Trade-off

Comparison of Trade-off MeasuresFIGURE 7

c: c:.<1

00

.. ....

IU ofJ.... 10: I ....

U :::J

.... UID 0

....

tJ 10U

'D"CIII

004 CD.... ........ ....'D ....

Q "C

! ! 0

c: :E::>

Both ser ies appear to move in the same direct ion, but

there are some points tha t diverge. Similar resul ts were

obtained fo r the education t rade-off . The re la t ively high

correla t ion between the ser ies does not indicate tha t there

i s cause for much concern in using the modified formula,

provided tha t the data are of suff ic ient precis ion. Howev-

er, data reported by the IMF and ACDA produced considerable



101

d i f f e r e n c e s i n th e e sti m a t e s o f t h e t r a d e - o f f s . A compar-

i s o n o f e s t i m a t e s f o r t h e h e a l t h t r a d e - o f f u s i n g each s o u r c e

i s p r e s e n t e d i n F i g u re 8 .

o ACDA·'DataD U.s. Budget Data

II. DF Data

o

1960 1970 1980 1990

yearComparison of Data Sets for the Health Trade-off

FIGURE 8

In comparing t h e d i f f e r e n c e s between d a t a drawn from

t h e IMP an d d a t a from th e ACDA, t h e s u p e r i o r i t y o f IMF d a t a

i s a p p a r e n t . However, n e i t h e r measure f i t s p e r f e c t l y w i t h

u.s. budget d a t a , c o rr el at io n s i n di ca t e t h a t t h e IMF d a t a

a r e c onsi de r a bl y c l o s e r t o t h e mark t h an ACDA d a t a . Table 3

p r e s e n t s t h e Pearson product moment c o r r e l a t i o n s between t h e

t r a d e - o f f measures c a l c u l a t e d u s i n g u.s. budget d a t a and



102

t rade-off measures calculated using the other two sources.

TABLE 3

Correlations with U.S. Budget Data

Heal th Trade-of f

Education Trade-off

ACDA Data

- .13

- .25

IMP Data

.73

.51

My suspicion of the comparatively low comparabil ity of

ACDA data i s supported by these correlat ions . Even using

the IMF data, a direct ana ly sis o f trade-offs using cross

nat ional data under the Berry-Lowery formulation i s doomed.

Measurement error i s amplified by the formula, which re l ies

on very small increments.

A somewhat less p re ci se solut ion to the problem i s to

use a sl ightly different conceptualization that i s l ess

sensit ive to very small changes: the difference between the

percentage spent on each category. In constructing th is

kind of measure, the base has changed from the budget incre

ment to the budget as a whole. While such a measure i s

conceptually less precise, i t i s an elegant solut ion to

using less precise data. Moreover, i t is s t i l l superior to

regressing one category on another because i t provides a

directional measure of the distr ibut ion among categories



103

instead of presentin g the dist r ibut ion as a hypothesis.

Interpretat ion of th is measure i s quite simple.

I f we subtract th e p ercen t of defense expenditures from

the percent of health expenditures, the result may be posi

t ive , negative or zero, depending on which category is

larger or i f they are equal. I f the resul t ing value of the

next observat ion i s h igher h ighe r, health has fared bet ter

( relat ive to defense) than i t did in the l as t observat ion.

I f i t i s smaller , health fared worse than i t did in the l as t

observation. In some ways this measure i s superior to the

Berry-Lowery formula in tha t i t i s a more substantive indi-

cator of the change in the structure of the budget.

Their measure is adequate for analysis of marginal

changes over time, but does not p rovi de a basis for making

cross-nat ional comparisons concerning the st ructure of

government spending. By basing the t rade-off calcuations on

percentages of the total budget, one can see very clear ly

how dist r ibut ions between two categories dif fer between

countries. S plits in incremental increases and decreases

from one year to the next might hide larger diferences

between countries. Using th is procedure, new t rade-off

measures were const ruct ed . C ross -nat iona l comparisons of

these calculations are shown in Figures 9 and 10.



104

Percentage Health minus Defense Expenditures

Difference between Budget SharesFIGURE 9

Percentage Education minus Defense Expenditures

Difference between BUdget SharesFIGURE 10



105

What is in terest ing is tha t both measures display a

high degree of s tab i l i ty in most countries. The series for

the United States i s one of the most volat i le . Indeed the

United States i s one of the most dev iant c as es in the sarn-

pIe, which makes i t an especial ly in terest ing case for

analysis .

With t hese prel im inar ie s out of the way, I turn to the

s ta t i s t i ca l tes ts of the hypothesis tha t par t ies influence

t rade-offs . Chapter I I I examines the generali ty of the

theory across countr ies . Chapter IV t es t s the th eo ry u si ng

a wider range of tr ad e- off s in the United States , with more

precise budget data .


http://slidepdf.com/reader/full/parties-ideologies-and-budgets-a-study-of-budgetary-trade-offs 115/236,

CHAPTER I I I

CROSS-NATIONAL PATTERNS: TRADE-OFFS

IN EIGHTEEN DEMOCRACIES

Introduction

This chapter presents a br ief s t a t i s t i ca l analysis of

t he hypothesi s that party ideologies effect t rade-off out-comes. Two trade-offs are examined: the t rade-off between

spending on defense and health, and the t rade-off between

defense and education. Following t he d is cu ss ion of the

measurement problems presented in Chapter I I , t rade-offs are

defined in terms of th e d if fe rences between the percentage

of outlays for each category.

In th is chapter, the dependent variables are on an

interval , bounded by zero. Budget dist r ibut ions (al locat ive

trade-offs) increasingly favor the c ate go rie s o f health or

education as values of the dependent variables become larg-

er. Positive values indicate tha t the health or education

percentage of the budget is larger than the defense share of

the budget. Negative values indicate tha t defense takes a

larger share.

Eighteen members states of the OECD were selected to

t es t the generality of the hypothesis tha t party ideology

affects budget allocations for health and education. These

106



107

countries included: Austral ia, Austria, Belgium, Canada,

Denmark, Finland, France, West Germany, Ireland, I ta ly ,

Japan, The Netherlands, New Zealand, Norway, Sweden, Swit-

zerland, the United Kingdom and the United States .

Cross-National Patterns

No general pattern of part isan i nf luence across a l l

countries in the sample was found, but part isan influences

were detected in some countries. In fact , the evidence

suggests tha t the impact of part ies , or whether part ies

influence budgets, varies from one country to another, b ut

not always in the di rec tion an ti cipated by the hypothesis.

In i t i a l ly , the data were examined using simple ordinary

l eas t squares regression (OLS). While th is approach i s

generally not advised in pooled t ime-series analysis, closer

inspection of resu l t s , by country, do not indicate tha t

using the much more complex procedures required for General-

ized Least Squares (GLS) regression i s warranted.

The appeal of GLS i s that it provides more robust

estimates in face of violat ions of the assumptions of re-

gression analysis . 108 I t i s widely recognized tha t analy

s is of pooled t ime-ser ie s , e spec ia lly of cross-national

data, i s plagued by the twin devils of autocorrelation and

108 See Jan Kmenta, Elements of Econometrics, 2d ed. (NewYork: Macmillan Publishing Company, 1985), 607-757.



108

heteroskedast ic i ty. Most researchers resort to the use of

GLS to r es olv e th es e problems. Unfortunately, computer

programmed GLS routines are in tolerant of missing data .109

Gathering a seamless data set , without any missing

observations for a l l variables (including appropriate con-

t ro l variables) on a l l cross-sections proved impossible.

Even with exis t ing sources, construction o f expendi tu re s

d ata r equire d a considerable amount of spl ic ing. Attempting

the same procedure with variables that are more vola t i le

than government spending (e.g. unemploYment) would stretch

th e a lre ady s trained comparability to unacceptable extremes.

Nevertheless, enough data of fair ly good quali ty and

consis tency110 on health and education and defense spending

were avai lable to construct a complete series for 17 of the

18 countr ie s se lec ted for the years 1960 to 1985. 111 For

109 See Eichenberg and Stol l , II Straightjacket: Supportfor Defense Spending, II for a good discussion of these problemsand a simple, non-technica l explana tion of how to deal withthem in analyzing data with missing cel ls .

110 Published longitudinal series are r i fe with adjustments, re-basing and redefini t ions . Perhaps the most severecase of redefining var iables i s the data on unemploYment. Thebest source, the International Labour Organization, based inGeneva, Switzerland, does a f ine job of gathering emplOYment

data . However, each country defines unemploYment i n d i ff er en tways. Allegedly more re l iab le survey data a re a va ila ble formore recent years, but are not comparable to earl ier datareported by government agencies. [See Year Book of LabourSta t i s t i cs , (ILO: Geneva, various years)]

111 Some data could be imputed from s imilar ly definedseries found in other sou rces, b ut using multiple sources isnot generally advised.



109

some countries data on the dependent variable were available

to 1988. Complete expenditures data f or Ir eland, however,

were available up to , but not beyond 1981.

Using OLS as an exploratory method f or const ructi ng the

most l ike ly specif icat ions for models of t rade-offs between

health and education, I experimented with several specif ica-

t ions . One and two-year lag st ructures produced confounding

resul ts in the cross-nat ional as well as the single country

analysis presented in the next chapter. 112 A moving aver-

age t ransformation of the pol i t i ca l variable seemed to

remedy the problems of seasonali ty, and accounted for the

theore t ica l postulate that the pol i t i c s of the budget i s a

continuous, not discrete process. Such a t ransformation i s

also theoret ical ly consistent with the notion of policy

shepherding discussed in Chapter I I . A three-year moving

average of WMEAN was used as the principal independent

variable .

A s ta t i s t i ca l ly signif icant model, with apparently

signif icant coeffic ients , was constructed using the weighed

mean score (WMEAN). No s ta t i s t i ca l ly signif icant resul ts

were obtained in cross-nat ional analysis using the anchored

score (AMEAN). The multivar iate model presented below

includes the most common control variables.

112 See Chapter 4 for a more thorough discussion ofspecifying the lag st ructures .



110

Following the findings of those who have found empiri-

cal regulari t ies between government spending on social

programs and economic dependence on exports, in f la t ion ,

populat ion size, aging and even time, I use these variables

as co ntro ls in a mult ivaria te equation. 113 The signi f i -

cance of the coeff icients reported in Table 3 i s probably

(almost definitely) over st at ed a s a resu l t o f autoco rr e

l a t ion . No global t es t of autocorrelation in pooled time

series i s curren tly avai lab le and Durbin-Watson s ta t i s t ics

are not applicable to a pooled design. These resu l t s are

interest ing only insofar as they display what appears to be

a signif icant ly negative relationship between the dependent

variables and the weighted ideological score. This con-

forms to the theorized relat ionship. That i s , as the mean

ideological score of the parliament moves to the r ight ( i ts

value increases), the difference between health or education

and defense decreases. Tabel 4 displays the resul ts of the

i n i t i a l analysis:

113 See Cameron, "Expansion of the Public Economy;" Beck,"Public Sector Growth;" Clayton, "A Comparison of Defense andWelfare Spending;" Swank, "The Poli t ical Economy of GovernmentDomestic Expenditures." Time i s a very crude control forsecular t rends, see Rose, Do Part ies Make a Difference.



Table 4

Cross-National Regression

Results

111

Variable

Constant

WMEAN

% Population 65+ l t - 1 )

% Population 15 ( t - l l

Government Spending/GDP

Exports/GDP

Inf lat ion

Log of Population

Time

HealthTrade-off

30.51**(10.427)

-0.25**(-6.32)

-0.17**(-3.30)

-0.18**(-2.18)

0.03(0.57)

0.12

(3.00)

-0.32(-6.55)

0.58(11.95)

EducationTrade-off

26.7**( ... )

-0.31**(-6.31)

0.08(1.45)

-0.53(-8.86)

0.19(3.14 )

0.18

(4.06)

-0.20(-3.54)

0.42(7.37)

Adjusted R2

N

=

=

.45

427

.35

427

Standardized regression coeff ic ients are reported.T-scores are reported in parentheses.* Signif icant a t the .05 level** Significant a t the .01 level



112

This would indicate that the r ight has the effect of

reducing t he d if fe renc e between the categories. This could

occur i f the r ight depressed social spending down, or in

creasing defense spending vis-a-vis health o r education.

This supports the hypoth esis , but longitudinal , per country

analyses casts doubt on these resu l ts .

Examining the coeff icients for each country, a very

d if fe re n t p ic tu re emerges. Of the 18 countries, only five

had signif icant coeff icients for the relat ionship between

WMEAN and the health t rade-off . Of these, three were posi

t ive and two were negative. For the education t rade-off the

picture is even more murky. Regression results for only s ix

countries produced s ig ni fi ca nt c oe ff ic ie nt s f or WMEAN, and

five of these were posi t ive. For both equations, only a

third, or less of the sample produced signif icant results

and most of those were in the "wrong" direct ion.

Clearly, the hypothesis of a general part isan influence

holding across countries, in the direct ion hypothesized,

must be rejected.

The one country where a negative coeffic ient obtained,

in both equations, was Japan--which is l imited by i t s con

s t i tut ion in i t s defense spending. The only other country

which had signif icant (but posit ive) resul ts in both equa

t ions was the Netherlands.



113

Summary and Conclusion

The s t a t i s t i ca l tes ts of th e gener al hypoth es is fa i led

to provide consistent support for the theory. In fact ,

there i s more support for the view tha t the r ight increases

social spending in health and education. Obviously, the

theory does not have strong generali ty, a t leas t not with

the indicators used here to measure ideology.

Many of the problems encountered in determining the

proper placement of part ies , especially those that sp l in t -

ered, or merged with other part ies (especially in I ta ly and

France) may have contributed a substant ial amount of impre

cision in the measure.

Numerous r e- spec if ica tions to the i n i t i a l model, in

cluding the addit ion of the ideological positions of the

party holding the presidency or prime ministry, fa i led to

improve upon the resul ts in i t i a l ly obtained. C alcu la tin g

year- to-year changes did not improve the f i t of the model,

or of the resu l t s by country. Analysis using components of

th e i deol og ic al scores, l ikewise, fai led to yie ld s ign i f i -

cant resul ts .

I am also confident that the fa i lure of the hypothesis

is not the resul t of inadequate specif icat ion of the func

t ional form of the equations. Residuals were examined ad

nauseam. No non-linear patterns appeared to be present .



114

A more di f f i cu l t problem to address is whether a single

method of aggregation i s applicable across countries having

a larger or smaller number of par t ies . Two-party and multi-

party systems are in the sample. Some have experienced

h ighly uns tabl e coali t ion governments, while others have

been dominated for long periods of time by one party or

another.

A more fundamental question is whether some of these

countries have so inst i tut ional ized thei r patterns of spend-

ing tha t to change might be pol i t ical ly unthinkable. Re

viewing the pat terns described a t the end of Chapter I I , the

majo rity o f coun tries experienced very l i t t l e change over

the la s t two decades.

Some of t he countr ie s in the sample appeared to be

almost s ta t ic in thei r dist r ibut ion of allocations. I t may

be that the ins t i tut ional strength of the parliament to make

the kinds of changes hypothesized is not very great . Much

more research needs to be done on th e n atu re of budget

processes in advanced industr ia l democracies.

Comparative research on th e s ub je ct has not y ie ld ed th e

kind of conceptual tools needed to specify, i n qual it at iv e,

much less quanti tat ive terms, differences between the pro

cesses of f iscal control. Research in th is area might focus

on the relat ive strength of the bureaucracy and parliamenta

ry bodies in set t ing budget pr ior i t i es , or the number of



115

"veto points" tha t a policy must t raverse before implementa-

t ion, etc .114

The weakness of results obtained in the cross-national

analysis requires a re-evaluation of the approach, as well

as th e th eory . In the next chapter I examine t he in fl uence

of part ies in the United States using more subt le measures.

Results obtained in an analysis of fe de ra l out la ys in the

United States using a re-conceptualization of the measure-

ment of part isan strength suggest that more precise measures

are needed in this type of analysis .

Turning to an examina tion of trade-offs in federal

outlays, Chapter IV presents an analysis of the impact of

par t isan unity on budget al locat ions in the United States .

114 Robert Cox suggests that the budget process in theNetherlands i s dominated by the bureaucracy and tha t theparliament has l i t t l e control over allocations, [Robert Cox,"Do Part ies Matter? Lessions from the Low Countries," in eds.Joe Adams, Marina Arbetman, and M. Donald Hancock, Pol i t ica lPart ies and the Ideology of the Welfare State, (Forthcoming),14. ]



CHAPTER :tV

TRADE-OFFS :IN THE tJN:ITED STATES

Introduction

This chapter p re sents s ever al t es t s of the hypothesis

that the part isan i deologi es o f the Democratic and Repub

l ican par t ies have an impact on the direction of t rade-offs

in federal outlays. This hypothesis i s applied to several

kinds of t rade-offs in federal outlays.

The principal explanatory variables used in this chap

te r are based on a re-conceptualizat ion of part isan influ-

ences, using party unity in voting as a surrogate measure

for the ideological i nf luence o f par t ies .

Three separate sets of dependent v ari ab le s a re exam-

ined. Each se t of dependent variables deals with a dif fer-

ent level of aggregation. I begin with the highest level of

aggregation and descend to more disaggregated t rade-offs

between narrower components of federal spending.

The f i r s t se t of dependent variables include t rade-offs

between highly aggregated program categories. Following the

c en tr al th ru st of the l i te ra ture , a measure of th e tr ad e-o ff

between outlays for d iscret ionary domestic programs and

defense i s constructed. This variable serves a dual pur

pose, as a thing to be explained in i t se l f , and as a basis

116



117

for establ ishing the fonnal structure of a general model.

Many essent ia l parameters are unknown from the outset .

For example, the exact nature of the lag s tr uc tu re fo r a

po ten ti al pa rt is an impact, cannot be established a priori .

Based on my reading of the l i te ra ture on policy development,

I speculate that the impact of part isan influences might be

expected during a span of about two years . In analyzing the

aggregate t rade-off between defense spending and discret ion

ary domestic spending, th is parameter, and others are iden

t i f ied through a ser ies of experiments. These experiments

culminate in the specification of a general model.

Once ident i f ied, the generali ty of the model is tested

by subst i tut ing the in i t i a l dependent variable (the aggre

gate t rade-off between defense and discretionary domestic

spending) with other dependent variables. Appropriate

control variables are added to assess the robustness of the

principal components under al ternat ive specif icat ions.

Continuing to use th is framework, t rade-offs between

outlays for defense and health and education programs are

introduced. A second set of dependent var ia ble s a re con

structed using more disaggregated data. Trade-offs between

components of defense and domes ti c outlays, with functional

s imilar i t ies , are examined. Two specif ic kinds of outlays

are examined. The f i r s t deals with outlays affect ing em-

ployment. The second deals with outlays for research.



118

Following the cri t icisms of several economists and

policy special is t , who argue that defense spending drains

the economy of ta lent and expertise, I also examine t rade-

offs between outlays for defense research and development

and civi l ian research. ll S Spending for mili tary research

has been cr i t ic ized for creating a "brain drain" on the

civi l ian economy. Add iti onal ly , t he scholarly consensus i s

tha t posit ive economic spi l l -over effects are less l ikely

for mil i tary research than civil ian research. 1l6

Federal spending on mili tary research is d i f f i cu l t to

separate from civi l ian research. Federal outlays for re-

search purposes are itemized under a number of functions,

i nc luding ou tl ays in the areas of health and education as

well as in the catego ry o f science, which also includes

expenditures for the National Aeronautics and Space Adminis-

11S The t radeoff between defense expendi tures and researchand development i s discussed in Russett, "Who Pays forDefense?" 421; Szymanski, "Military Spending;" and Mintz, "A

Disaggregated Analysis."

116 A more general view i s tha t , in addit ion to thesenegative effec ts , mil i tary expenditures do not producemul t ip l i e r effec ts as large as those obtained through domesticspending. For a review of these arguments, see R.P. Smith,

"Mili tary Expenditures and Capitalism." Cambridge Jo urn al ofEconomics 1 (1977): 61-76; Friedland and Sanders, "The PublicEconomy and Economic Growth," 434; Whiteley, Pol i t ica l Controlof the Macroeconomv; Marion Anderson, "The Soft Spot: How toAttack the Pentagon." Journ al o f Sociology and Social Welfare4 (1977): 597-610; Chan, liThe Impact of Defense Spending,"415-22.



119

t ra t ion (NASA). However, the persis tent bureaucratic st rug

gles between the mili tary and civi l ian administrators within

NASA, and the number of secret space missions conducted on

behalf of the military, suggest that some outlays for space

f l ight serve mili tary in teres ts . Expenditures for space

f l ight have, therefore, been excluded from th e fig ures used

in calculating the t rade-off between science and defense

spending. In other areas, such as health and education,

spending i s more clearly divided between mili tary and c iv i l -

ian research.

Three categories of civi l ian research are c o n s t r u c t ~ d .Each category represents a different level of aggrega ti on .

The f i r s t category includes aggregate outlays for the func

t ion of science (minus space f l ight outlays). The second

category i s composed of outlays for the subcategory of

general science and basic research, and a th i rd category i s

constructed which includes expenditures for research in the

areas of heal th, education and the catego ry o f general

science and basic research previous ly mentioned.

In the f inal se t of hypotheses, t rade-offs within the

military budget are examined. Following the work of Mintz,

who suggests tha t trade-offs within the mili tary budget have

occurred during the l as t decade, I examine the distr ibut ion

of expenditures for pe rsonnel a s opposed to outlays for

procurement. Mintz argues that very spec if ic t rade-of fs



118

120

have occurred during the 1980s, under the Reagan administra

t ion. His finding raises the question o f whether such

t rade-offs follow a consi st en t , p r ed ict ab le pat tern.

The impact of part isan pol i t ics on allocations within

the mili tary budget has not been widely discussed in the

l i t e ra ture . Mintz' f inding of a trend toward h ighe r outlays

for procurement i s sugge st iv e. In cre as ed expendi tu re s on

mili tary procurement and research and development may repre-

sent a kind of "capi ta l izat ion" of the defense port ion of

the public sector . Private business in teres ts are heavily

involved in procurement programs, and many large corporate

benefi t from gaining defense contracts. 117 Republicans are

widely perceived as favoring large business in teres ts , while

Democrats are perc ei ved a s having more concern for labor

in teres ts (which may be posi ti v ely a ff ec ted by outlays for

personnel) .118

The hypothe sis o f a part isan influence on changes

within the defense port ion of the budget seems to be sup-

ported, prima facie, by r ecen t tr ends in defense spending as

reported by Mintz. Given the apparent sh i f t within mili tary

117 See Larry J . Griff in, Joel Devine and MichaelWallace, "Monopoly Capital , Organized Labor, and MilitaryExpenditures in the United States , 1946-1976," AmericanJournal of Sociology, 88 (1982): 113-53.

See Baran and Sweezy, Monopoly Capital; Larry J .Griff in, Joel Devine and Michael Wallace, "Monopoly Capital ,Organized Labor, and Military Expenditures i " Chan, "Impact ofDefense Spending," 419-20.



121

outlays, an examina tion of ear l ie r periods for seems war-

ranted. 119

20

o Military R&Do Military Personnel

A Military Constructiono Procurement

20It )

>-lU.....

..:::Jo.....

lU

c..QJ

'CQJ

IL

-U....o....-....

E::QJ

Uc..QJ

Q.

15

10

5

o

1960 1970 1980 1990selected Itells

Components of Defense Spending

FIGURE 11

15

10

5

o

.....

lU

c..QJ

'C

CD

L1..-U....o....-..CQJ

oc..QJ

Q.

Both Mintz and Russet noted that there have been wide-

spread shi f t s in federal expenditures during the past decade

tha t cannot be explained by the use of exist ing theories of

budgeting. I f party ideo logy has an impact on these chang-

es, how has i t worked? Examining government st ructures and

processes may provide answers to th is question.

Ideology, Structure and Process

The quest ion of whether ideology plays a par t in con-

119 See Mintz, "A Disaggregated Analysis."



122

gressional voting has been a subject of spir i ted debate

among students of legis la t ive behavior. Much of the debate

concerns the existence of a l e f t - r igh t dimension in ro l l -

cal l votes . Several scholars claim that congressional

voting follows part isan- ideological pat terns. 120 In teres t

group rat ings based on key congressional votes indicate tha t

the congressional part ies dif fer on sa l ien t issues. 121

Addit ionally, several studies of federal spending support

the hypothesis tha t the Democra tic and Republican part ies

effect outlays according to the i r proc l ivi t ies to support

welfare programs and other public services .122 This study

tentat ively accepts the th esis th at the Democratic Party

represents "the le f t" in American pol i t i c s and tha t Republi-

can Party represents "the r ight ."

As discussed in Chapter I I , i deo log ica l d i ffe rences in

the United S tates are confined to debates within th e l ibera l

t rad i t ion . By comparison to the par t ies of European democ-

120 See Patterson and Caldeira, "Party Voting ;" Schne ider ,Ideological Coali t ions; Poole and Daniels, "Ideology, Partyand Voting." For a cri t ique of the claims of Poole andDaniels, see Korford, "Dimensions in Congressional voting."

121 See Sha ff er , P ar ty and Ideology. Shaffer also f indstha t r eg iona l a s well as part isan differences account for somevaria t ion in ADA rat ings, but the pattern of Democraticmembers having higher average rat ings than Republican membersi s consistent across regions.

122 See Davis, Dempster and Wildavsky, "Toward a Predict ive Theory," 431, Browning, "U. S. Social Welfare Expenditures,"206-7.



123

racies , ideo log ica l d i fferences in the United S tates are

more narrow and less systematic. The lack of coherence in

l iberal ism makes i t possible for opponents on one policy to

be a l l ies on another. Moreover, the two part ies have a

strong tendency to converge on many issues. Nevertheless,

broad differences between the two part ies are evident.

The Republican and Democratic part ies are known to have

di fferen t views of the proper role of government. Defense

is seen as a necessary governmental function by Republicans,

while other kinds of programs must be jus t i f ied more r igor-

ously by thei r impact on the principle of minimal govern-

ment. In g eneral, Republicans take a class ica l l ibera l view

tha t government should be minimized as much as possible . 123

While t hey d isag ree, in principle, with the use of govern-

ment to re gu late the economy, the l i terature suggests tha tfor Republicans national defense i s a legitimate, even

fundamental, governmental function. Nat iona l defen se and a

strong mili tary, is one of the principal ideological con

cerns of Republicans. According to Schneider, Republicans

are more l ikely to perceive danger in international affai rs ,

while Democrats are less l ikely to perceive a ho st il e i nt er -

nat ional environment.124 By comparison, Democrats view

123 See Schneider, Ideo log ical Coa li ti ons , 159-90; cf .Keefe, P ar ti es , P o li ti cs and Pub li c Pol icy, 211-9.

124 See Schneider, Ideo log ica l Coal it ions , 59-80.



124

government more posi t ively, as a legitimate instrument for

social and economic change. In short , for Republicans,

domestic programs must pass the tes t of whether the program

fa l ls into the legitimate domain of governmental ac t iv i ty .

For Democrats the issues are different . With respect to

international affa i rs , there i s a general bias toward dis-

counting the importance of defense. Democrats tend to be

less incl ined to accept the domino theory, and perceive less

threat from international opponents. 125 Democrats are more

l ikely to see domes tic needs and more readily accept govern-

ment actio n to remedy socia l problems. These procl ivi t ies

should affect the way each par ty effects the distr ibut ion of

federal outlays.

Of course, nei ther party completely con tr ol s pub lic

expenditures. As intended by the framers of the consti tu-

t ion, ins t i tut ional and pol i t i ca l divisions can frustrate

policy making. In the United States, budgetary power is

shared among three ins t i tut ions , the U.S. House of Represen-

ta t ives, the U.S. Senate and the presidency, with r ival r ies

develop ing a long ins t i tut ional as well as par ty l ines .

While part ies , and the differences between them, are the

p ro ducts of his tor ica l developments, grounded more in custom

and t radi t ion, ins t i tut ional divisions are firmly estab-

lish ed in the formal s tructure of the federal government

125 Ibid.



125

i t se l f .

The United States Consti tut ion provides the House of

Representatives with the l ion 's share of power in the area

of f isca l policy, as the source for policy making in the

area of taxat ion. 126 The Senate has long claimed an equal

prerogative in the in i t ia t ion of appropriations bi l l s , but

has t rad i t ional ly followed the lead of the House in financ-

ing new programs .127 with the ascendance of president ial

leadership in th e twent ie th century, the ch ief executive has

become a major source of policy in i t ia t ives and program

proposals. Presidents also exercise the formal power to

veto appropria tion and tax b i l l s , but these are not thei r

only means of budgetary control. Aside from the recommenda-

t ions and requests for funding that presidents rout inely

send to Congress, presidents have also exercised more direct

control over expenditures. As Louis Fisher notes, pres i -

dents have t radi t ional ly exercised some degree of discre-

t ionary spending power, especial ly in the area of de

fense .128 And, unt i l 1974, presidents could influence ex-

126 Article I , section 7, par t 1, of the United StatesConsti tut ion reads, "All Bil ls for rais ing Revenues shall

or iginate in the House of Representatives; but the Senate maypropose or concur with Amendments as on other Bills ."

127 See Fenno, Power of the Purse,

128 Much of pres ident ia l spending power i s de facto incharacter . Fisher makes pointed reference to TheodoreRoosevelt 's "daring" Congress not to fund the return of thef lee t on i t s circumnavigation of the globe. As commander-in-



126

penditures through impoundment. This power was check by the

Congress in the Congressional Budget and Impoundment Act of

1974. With the passage of Gramm-Rudman-Hollings, the t ide

seems to have reversed to an extent. Determination of

conditions fo r seques tr at ion has the potent ia l to become the

fu nc tio nal equiv alen t o f impoundment. 129

Budgetary power in Congress i s dominated by committees.

The Appropriations, Ways and Means, and Budge t Committees in

the House, which are the key committees on the budget, are

also among the most conf l ic tual of the congressional commit-

tees , which ref lec ts the salience of the issues th ey h andle ,

many of which deal with budget al locat ions .130 Confl ic t

over the budget is not ult imately se t t led in committees, but

spi l l s onto the f loor i t s e l f . Budget votes ten d to be quite

close. As Gary W. Copeland explains,

The part isan nature of budgeting in the House i s unusua l for Congress. Throughout the twentieth century

chief of the armed forces, t he p re si dent has typical ly had astrategic advantage in promoting defense expenditures. Thistoo has changed with the passage of the War Powers Resolution.See Louis Fisher, Presidential Spending Power (Princeton:Princeton University Press, 1975).

129 The Office of Management and Budget (OMB) and theCongressional Budget Office (CBO) issu e rep orts regardingexpenditu re s, but executive orders implement t he s eque ste r.President Bush exempted outlays for mil i tary personnel fromsequestrat ion, as he is allowed to do by law [See StevePhil l ips , "Budget Negotiations and the Sequester ," World ofPoli t ics (October 1990): 6].

130 See Smith and Deering, Committees in Congress, 70.



127

partisanship in Congress has declined, and issues thatclearly and neatly divide the part ies on a continual

basis are rare . But vast majorities of Democrats haveopposed vast majorit ies of Republicans on every resolut ion to date . . . That voting on House budget resolut ions is highly partisan is incontrovertible, but i t isless c lear why those votes divide the part ies so clearly and consistently (par t icular ly in l igh t of thebipart isan nature of budgeting in the Senate).

Budget debates have every reason to be part isan;they involve the issues that are fundamentally import an t to the part ies . . . The budget includes both adream of where our country should be headed and ablueprint for getting there-- i f anything wil l dividethe part ies , that should. 131

Copeland continues to note that ideological extremes

have played an import an t part in the budget process in

recent years. Part isan div isions with in the Congress are

reinforced a t times by partisan divisions between the execu-

t ive and l eg is la t ive branches.

Since World War II , Republicans have had a majori ty in

the U.s. House of Representatives for only 4 years, from

1949 to 1950 and from 1953 to 1955 or rather , the 80th and

83d Congresses. Since then, Democrats have enjoyed a major-

i ty . Republicans have con trol led the Senate for ten years

between 1940 and 1990. Only three of the five terms of

Republican control of the Senate have been consecutive.

These in clude th e 80th, 83d, 97th, 98th and 99th Congresses,

131 Gary W. Copeland, "Changes in the House ofRepresentat ives Af te r the Passage of the Budget Act of 1974,"in Congressional Budgeting: Poli t ics, Process and Power. W.

Thomas Wander, F. Ted Hebert and Gary W. Copeland, eds.(Baltimore: Johns Hopkins University Press, 1984), 70-71.



128

or ra ther , from 1981 to 1987. This sh i f t plays an important

part in explaining budget behavior during the 1980s.

While Republicans have held the White House for 26

years between 1940 and 1990, just over half of the ent i re

period, they have rarely shared power with a Congress also

having a Republican majori ty. I t is widely recognized that

confl ic t between the two branches tends to increase when

control i s divided. 132 With the election of 1980, Republi-

cans gained a majori ty in the Senate.

As mentioned, this is the f i r s t time since World War I I

tha t the Republicans have been able to maintain control of a

chamber of Congress for consecu tive terms. This sh i f t in

control of Congress marked a fundamental sh i f t in the bal-

ance of power between Democrats and Republicans.

I t has also been noted that exploit ing th is opportunity

was well orchestrated. Some scholars have pointed to the

way tha t the Reagan administration's implemented control .

As Harold Seidman and Robert Gilmour note:

In many respects Ronald Reagan i s unique among recentpresidents . His agenda i s highly select ive and wasestablished before he was inaugurated. His goal i s toreverse the federal government's direct ion by cutt ingdomestic programs and spending, reducing taxes, andl imiting federal regulations. To gain control of the

132 Republican presidents have a s ta t i s t i ca l ly s igni f i -cant lower success ra te than the i r Democratic counterpartsbetween 1953 and 1988. In a regression analysis o f p re si dent i a l success ra tes , I also found that the number of times apresident takes a posit ion on a b i l l i s negatively correlatedwith his success ra te .



129

vast federal establishment and harness i t to h ispurposes, he has had to devise a new approach and a

radical ly different organization strategy .133

Reagan's strategy depended on congressional assistance.

This was provided by an unusually sympathetic Senate. 134

Reagan clearly enjoyed more support in the Senate than his

l as t two Republican predecessors. In fact , h is success rate

approached that of Democratic presidents who had a Democrat-

ic majority in the Senate. Available data on the success

ra tes of presidents in Senate voting indicate tha t , on

average, President Reagan had a higher success ra te than

Dwight D. Eisenhower. Table 5 shows the percentage of

votes, on which th e p re sid en t has taken a posi t ion, that

resul ted in a president ia l v i cto ry . A pa rt is an pa tt ern is

evident.

Democratic presidents have had much higher success

ra tes than Republican presidents during the l as t few de

cades. This supports the common perception tha t shared

party af f i l i a t ions are an important factor in securing

legis la t ive success.

133 Harold Seidman and Robert Gilmour, Pol i t ics , Posit ionand Power: From the Positive to the Regulatory State . 4th ed.(Oxford: Oxford Unive rs it y P re ss , 1986), 127.

134 The Senate voted to extend president ia l veto power toalmost a l l agency regulations in 1982. This was l a t t e r ruledunconstitutional by the Supreme Court. [Ibid. , 129].



130

TABLE 5

Average Presidential Success Ratesin the Senate

President Percent of Votes Won

Mean Std. Dev. Freg.

------------+------------------------------------tEisenhower I 73 12.9 7

Kennedy I 83 3.5 2Johnson I 81 10.5 6

Nixon I 63 10.6 6

Ford I 68 4.8 2

Carter I 79 5.2 4Reagan I

7711.6

8------------+------------------------------------Total I 75 11.7 35

Source: Vital Sta t is t ics on Congresst Data on Eisenhower's f i r s t year in office are not

reported. Republican presidents are in i t a l i cs .

In fac t, his success rate in the Senate during his

f i r s t year (88.3 percent) is the h igher than any other

president for which data a re ava il ab le , even higher than

President Johnson's success ra te in his f i r s t year in o ffice

(87.5 percent) . This pattern began early and i s even more

clearly evident when his success rates during the s ix-year

dominance by Republicans are compared to the l as t two years

of th is administration, when Democrats regained control of

the Senate. Reagan's success ra te dropped precipitously

during th is period. For the l as t two years of his adminis-

t rat ion, h is success rate was more than 20 percent lower

than it had been during h is f i r s t year .



131

TABLE 6

Reagan's Success Ratein the Senate

Year

198119821983198419851986

19871988

Success Rate RepublicanHeld Seats

88.3 5383.2 5385.9 5585.7 5571.6 5381.2 53

56.4 4564.8 45

Source: Vita l Stat i s t ics onCongress.Democratic control i s in i t a l i c s .

The unusually h igh succes s rates during the f i r s t s ix

years of the Reagan presidency supports the th esis th at the

s t ruc ture of the pol i t i ca l arena is quite different during

th is period than it had been when the Democrats controlled

both chambers. Thus, th is period should have a sl ightly

di fferent effec t on budget outcomes.

The fact tha t Republicans have been a minority party

during the bulk of the years under examination in th is study

ra i ses the q ue stio n of whether one should expect Republican

party unity to have an effec t on budget outcomes. However,

par ty unity becomes a l l the more important in the pol i t i c s

of budgeting when one recognizes that disunity in one party

might be exploited by the unity of the opposition.

Consider t he hypo the ti ca l case where, on a part icu lar



132

vote, Republicans are completely united, but Democrats are

divided. Republicans stand a much bet ter chance of winning.

When Republican House members align themselves with re la-

t ively conservative Democrats, usually Southern Democrats,

the probabi l i ty of gaining a voting majority, or preventing

an overrid e o f a (Republican) presidential veto, should also

increase. Such circumstances are not precisely measurable

using aggregate par ty unity data, but are not precluded by a

Democratic majori ty. 135

Most scholars use a single indicator or two indicators ,

one for Congress and one for the president , to measure

part isan influences. 136 The measures of par ty ideology

used in the previous chapter, as those used in most analyses

of American party influences, lack the capacity to d is cr im i-

nate between countries where party unity i s almost assured

and those c O Q ~ t r i e s where party unity varies over time.

By comparison to the disciplined part ies of European

parliamentary systems, the relat ively low level of party

discipline in congress ional voting raises doubts about the

potent ia l for systematic part isan influences. However,

several studies support the thesis that Democratic and

135 See Shaffer , Party and Ideology, 135-69; Keefe,Part ies , Poli t ics and Pub li c Poli cy , 218-9.

136 See Russett , "Defense Expenditures," 772; Berry andLowery, "An Alternative Approach," 689-90; c f . Browning, "u.S.social Welfare," 205.



133

Republican congressmen are dis t inc t ly different in thei r

ideological commitments and that th i s aggregate difference

i s re la t ive ly s table over tim e. 137 Additionally, there i s

evidence to suggest tha t the ideological differences between

the part ies have an impact on budgetary outcomes.

In a recent study, Ian Budge and Richard I . Hofferbert

argue tha t Democratic party platforms a re clo se ly associated

with federal expenditures. Their f indings support s the

hypothesis tha t American part ies abide by the i r stated

objectives, with consequences for federal outlays.l38 Ear-

l i e r studies have also found that Democrats tend to spend

more on welfare and social programs than Republicans. 139

Similar pat terns have also been found in the budgetary pat-

terns of municipal governments. 140 At the s ta te level ,

137 See Schneider, Ideological Coali t ions: Shaffer,Party and Ideology; William J . Keefe Part ies , Pol i t ics andPublic Policy in America. Fifth Edition (Washington,D.C.:Congressional Quarterly Inc, 1988), 211-230, c f . Budge,Robertson and Hearl Ideology, Strategy and Party Change, 3940, 44-5, 69, 392-7

138 See Ian Budge and Richard I . Hofferbert . "Mandatesand Policy Outputs: U.S. Party Platforms and Federal Expenditures ." American Pol i t ica l Science Review 84 no.1 (March1990): 111-31.

139 See Davis, Dempster and Wildavsky, "Toward a Predict ive Theory," and Robert X. Browning "U.S. Social WelfareExpenditures, 1949-1977" American Journal of Pol i t ica lScience Vol. 29, no. 1, 197-216, 1985 .

140 See Richard Koven, Ideological Budgeting: TheInf luence of Pol i t ica l Philosophy on Public Policy (New York:Praeger, 1988).



134

however, t he i nf lu ence of part ies on budgetary behavior has

not been clearly established. 141

Despite the relat ively undisciplined behavior of Ameri-

can par t ies , party aff i l ia t ion continues to provide the

principal basis for organization in the United States Con-

gress. The level of pa rt is ansh ip , however, has varied over

time. In general the level of partisanship has de

clined. 142 Much of th is decline has been at t r ibuted to

ins t i tut ional and electora l changes. Even with the general

decline in party discipline, the best s ing le p red ic to r of

voting behavior i s s t i l l party aff i l ia t ion. 143

The use of party unity as a measure of part isan values

may a t f i r s t seem strange, but i t clearly indicates the

strength of the party to mobilize and discipline i t s mem-

bers, which is essent ial for the legislat ive success of the

par ty. To the extent that unity is achieved, the members of

the party must be in some kind of agreement. In Chapter I I ,

I posi ted that the object of this agreement i s a se t of core

141 Cf. Sharkansky, Spending in the American States;Thomas R. Dye, "Taxing, Spending and Economic Growth in theAmerican States," Journal of Poli t ics 42, no. 4 (November1980): 1085-1107 i idem, "Party and Policy in the States ."

Nathan and Dooli t t le , Reagan and the States , 355-63; DavidMayhew Placing Part ies in American Poli t ics (New Haven: YaleUniversity Press, 1986), 257-307.

142 See Brady, Cooper and Hurley, "The Decline of Party i ..

Patterson and Caldeira, "Party Voting."

219.

143 See Keefe, Parties, Poli t ics and Public Policy, 211-



135

bel ie fs t ha t c on st it ute the party ' s ideology. Where unity

i s low, the impact of a par ty ' s ideology on budgetary t rade

offs should be less than when i t s unity i s high.

Of the pol i t i ca l influences on the budget, the party

dominating the House should have the most impact on outlays,

with the par ty af f i l i a t ion of the president having a s ignif -

icant , but weaker effec t on outlays.

Hypotheses

Following the discussion of ideology presented in

Chapter I I I hypothesize tha t increased unity voting among

Democrats will in cre as e th e budget shares of domes ti c spend-

ing re la t ive to defense expenditures. Given the Democratic

dominance of the House, I expect to find tha t Republican

party unity to have a comparatively weaker effec t on budget-

ary t rade-offs . The following hypotheses apply to the f i r s t

two sets of t rade-offs , where defense versus civi l ian out-

lays are concerned, but are generalized to other t rade-offs

in s imilar fashion:

H1 : As Democratic party unity increases the defense shareof the t rade-off pool decreases.

H2 : As Republican party unity increases the defense share

of the t rade-off pool increases.

With respect to the in flu en ce o f president ial pol i t ics ,

I expect to f ind tha t Republican presidents would decrease

spending on domestic programs re la t ive to defense expendi-



136

tures. This hypothesis is stated as follows:

H3 : Domestic shares of trade-off pools increase under Demo-

cra t ic presidents and decrease under Republicanpresidents .

Hypotheses regarding trade-offs within the mili tary

budget follow the same form:

H4 : As Democratic party unity increases the procurementshare decreases and the personnel share increases.

Hs : As Republican party unity increases the procurementshare increases more than the personnel share.

And,

H6 : Domestic shares increase under Democratic presidentsand decrease under Republican presidents.

Using both of the formulas described in Chapter I I , I

constructed measures of t rade-offs . These interval- level

measures o f con tain both positive and negative values. All

independent v ar ia ble s a re p os it iv e intervals , bounded by

zero.

Trade-offs favoring defense were arbi t rar i ly assigned

posi t ive values and those favoring domestic categories are

ass igned negat ive values. Accordingly, a posi t ive re la-

t ionship between an independent variable and a dependent

variable indicates a pro-defense relat ionship between two

variables . Conversely, a negative coeffic ient indicates

that the variable is associ at ed wi th decreasing defense

shares, or rather , an increase in the re la t ive share of

domestic outlays. Put different ly, posi t ive s ta t i s t i ca l



137

coeff icients indicate "more" defense and less domestic

spending.

In analyzing trade-offs between components of mili tary

spending, the same s tructure i s used, but o the r c at ego ri es

are substi tuted in the calculat ions.

Control Variables

Discontinuities in spending are observed during periods

of warfare. For this reason, the annual number of combat

deaths (soldiers kil led in action) i s introduced as a con-

t ro l variable.

Demographic and economic changes are also widely cred-

i ted with effecting budgetary outcomes. Other control

variables include: the percentage of Americans under the

poverty level , the percentage of peop le ove r the age of 65,

the percent of the population aged 1-14, the ra te of inf la-

t ion, percent change in gross national product, population,

th e p erc en t of the population under the poverty level and

the percent unemployed. 144 Following the work of Mintz,

Russett and others, I include a dummy variable for the

144 •Percentages under the poverty level are from Sta t ls -t i ca l Abstracts of the United States (various years) .Estimates pr ior to 1959 are based on income distr ibut ions.Percent unemployed were taken from the ILO, which are ident i cal to Bureau of Labor Stat i s t ics data. These data were backf i l led to 1950 [See Donald J . Bogue, populations of the UnitedStates . (New York: The Free Press, 1985), 582-600.



138

period 1981-1988. 145 This variable i s explained in more

deta i l below. Control variables are included in those

models where the l i terature sugges ts re levance.

Specifying the appropriate lag s tr uc tu re to t es t the

theory i s a b it more complicated. I expect tra de -o ffs to be

influenced by long-term pol i t ical commitments. Legislat ive

outcomes take time to develop and implement. I t i s expected

that t rade-offs between outlays for defense and public ser-

vices should occur in the implementation phase, one or two

years af te r the l eg is la t ive in i t i a t ives . This reasoning is

based on the idea of shepherding discussed in Chapter I I ,

where policy makers involve t h a ~ e l v e s in protect ing poli-

cies throughout the l i fe of a program.

I assume tha t t ra de -o ffs a re simultaneous. Assuming a

common lag between the dependent variables and the compo-

nents of a t rade-off i s a strong assumption, given the

pos si bi li ty t hat some programs may be implemented more

quickly than others . For example, changes in t ransfer

programs can be implemented almost overnight. Other pro-

grams are l ikely to take longer to implement and to produce

measurable outlays. In assuming a common lag, I am assuming

tha t the processes tha t generate defense outlays are identi-

cal in duration to those producing domestic outlays.

145 See Russett, "Defense Expenditures and National WellBeing," 776; Mintz, "Guns Versus Butter ."



139

Prel iminary Observations

This section brief ly descr ibes t rends in federal out-

lays since 1940, with a special emphasis on mili tary spend-

ing since 1960. The focus of th is study is the distr ibut ion

of al locat ions among programs. For th is reason, percentages

outlays are used to describe trends.

There are disadvantages to using percen tage terms to

describe expenditures. As the federal government has taken

on more functions, the denominator of to ta l expenditures has

changed in two ways. Tota l expend itures have increased in

real and nomina l terms, but the number of budget items

funded by th e fed eral government has also increased. I f

expenditures for exist ing items were to remain co ns tan t, th e

addit ion new items wil l necessari ly decrease the percentage

of outlays allocated for existing i tems. For t he se r ea sons ,

percentage terms can also mislead and should be read w ith

caution.

To begin, I show that most large increases in defense

outlays are re la ted to wartime needs. One of the best

indicators of the presence o f mili tary confl ic t i s b at t le

deaths. This v ar ia ble in dic ate s th e intensi ty of confl ic t

and the needs of the mili tary f or add it iona l resources.

As bat t le deaths in crease, th e percentage of the budget

going for defense increases as well . This pat tern i s clear-

ly evident in Figure 12 below:



o Defense OUtlays..

'. A Killed in Action

140

UI::0-

III-r- I..:::to

Q)

tI'Ic:......111C

90

80

70

60

50

40

30

20

10

1940 1950 1960 1970 1980year

Defense Outlays and Battle Deaths

FIGURE 12

150000

140000

130000

120000

110000 c:

1 ס ס o o 0 0.....

90000()

<

80000 c:....

7 ס ס o o "0

60000Q)r -Ir -I

50000....

40000

30000

2 ס ס o o10000

0

1990

..

The effect of the shocks of World War I I , the Korean

War, and the Vietnam war on military expenditures are easily

ident if ied. However, increases in percentage defense expen-

di tures during the 1980s is not explicable by reference to

combat deaths.

Taking a closer look at the las t three decades, the

impact of the Vietnam War is clearly visible, but the trend

in defense expendi tures does not uniformly decrease, as i t

did in the wake of the two previous large military con

f l i c t s .

The nadir of defense expenditures as a percent age of

federal outlays occurs in 1979. The trend then swings



141

upward. In fact , increases in the percentages of the feder-

a1 budget going to defense during the 1980s were almost as

large as the increases during the Vietnam War, but the

United States was not engaged in any large-scale combat

act ivi t ies (see Figure 12).

A large body of research on arms races has developed

tha t suggests defense expenditures respond to changes in the

mili tary expenditures of the Soviet Union. Much of the

evidence on th is question indicates that , i f a relat ionship

exis t s a t a l l , i s negative. 146 Nevertheless, enough ernpha-

s is has been placed on the influence of Soviet expenditure

to take th is fac to r in to consideration in analyzing poten-

t i a l t rade-offs in f eder al outlays .

With the exception of the past decade, the s ingle most

useful var iable in predict ing large percentage increases in

146 See Ostrom and Marra, "U.S. Defense Spend ing and theSov iet E stimate." Ostrom and Marra mistakenly suggest aposit ive relationship exis ts between Soviet expenditures andU. S. defense spending. Despite thei r verbiage, reporteds ta t i s t i ca l results show a s ignif icant negative relat ionship.I have taken a look a t this , with similar resul ts .

Using changes in percent ages o f U.S. defense spending,my resu l t s for the 1963-1988 period, using OLS and control l ingfor bat t le deaths, produced a negative coeff ic ient , s igni f i -cant a t p< .0000 (R2 =.65) . Some co ll in ea ri ty ex is ts between

bat t le deaths and changes in Soviet mili tary expenditures.When using CORC adjus ted regress ion , t he s ign if ic ance of thecoeff icient for soviet expenditures i s p < .18 (R2 =.54) withbat t le deaths s t i l l s ignif icant a t p <.0000) .

Paradoxically, Berry and Lowery, "An Alternative Approach," 694, however, find that Sov ie t defen se spending hasan effect on the domestic tradeoff . I also find tha t it i sre la ted to larger defense share.



142

defense expenditures has been armed confl ict .

o Defense outlays ' : l . , ' to Killed in Action

100

90

BO

70

60

50

40

30

20

191963-1990

Defense Outlays and Battle Deaths

FIGURE 13

2 ס ס o o

15000c:0........

uc (

1 ס ס o o c

-CCD....

....

....

SOOO

1000

As in the area of defense, trends in domestic spending

have also reversed in r ec en t yea rs . The most closely

watched categories are health and education expenditures.

Education outlays have f luctuated more than health expendi-

tures , which tend to be relat ively stable from one year to

the next. Percentage outlays on education and health de-

clined sharply a t the beginning of the 1980s. Both catego-

r ies d eclined, but education received the hardest hi t .

In much of the work on trade-offs between domestic and

defense spending, education is singled out as being one of



143

the most sensi t ive categories to changes in defense spend-

ing. This pattern seems to be evident during the 1980s.

201918

..17

>- 16CD

15-... 14::J

013

- 12D....

11

10.... 90

8....c 7GJ

6L

5D

a.4

3

2

1

0

1960 1970 1980 1990yeaI'

Education as a Percentage of Total Federal Outlays

FIGURE 14

A preliminary examination of the data on federal

spending indicates that outlays for defense and out lays for

domestic programs do not respond equally to increases and

decreases in the size of the to ta l budget increment. In

analyzing the expansion and contract ion of the budget, I

have found that defense outlays tend to grow more than most

other kinds of outlays during periods of expansion. De

fense outlays also tend to absorb most of the decrease in

when the budget shr inks.



144

This pat tern, however, is c lose ly a ssoci at ed with

combat act ivi ty . Large increases in budget increments

generally occur a t the beginning of combat act ivi ty and

large decreases at the end of such act ivi ty .

I t i s often assumed that discretionary categories of

spending are squeezed between military spending and large

enti t lement programs, but th is is not necessarily the case.

Enti t lement programs are quite res i l i en t to changes in

to ta l spending. When the to ta l budget increment shrinks,

both defense and discretionary domestic outlays tend to

decrease, but when the to ta l budget increment increases,

defense expenditures appear to take the larger share. Thus,

increases in the size of the to ta l budge t increment is

s igni f icant ly re la ted trade-offs between defense and the

categories of heal th and education dur ing times of war, when

the budget increases the most.

The overal l trend in federal spending contradicts th is

assessment. Non-defense outlays have generally increased

re la t ive to defense spending, except for the 1980s.

Mili tary confl ic ts , which tend to inf la te the to ta l

budget and decrease the share spent on domestic programs, do

not necessar i ly decrease domestic spending in real terms.

Many of the Great Society programs were implemented during

the Vietnam War. In real terms defense spending increased

on top o f increased domestic spending. This "piling on,"



145

may create a misleading impression about how defense spend

ing effec ts domestic spending.

As Russe tt observes, the growth in ou tlays fo r social

programs occurred during a period of mili tary confl ict . At

the same time that the programs of the Great Society were

f i r s t being implemented, the country was also escalating i t s

involvement in the Vietnam War. In Russet t ' s seminal study

on the effects of defense spending, he f inds that , "In the

aggregate, there is no very strong impact on civ i l public

expenditures. ,,147 However, some components of domestic

expenditures, however, fared bet te r than others when defense

spending increased.

Russet t and others have noted that inst i tut ional ized

programs such as farm subsidies and other well-entrenched

enti t lement programs appear to be relat ively insula ted from

the effec t of mili tary spending, but tha t education, health

and welfare programs tend to be the comparatively more

sensi t ive to f lu ct ua ti on s i n defense spending.

My prel iminary review of the data supports Russet t ' s

generalization, but also indicates tha t the overal l growth

or decline in to ta l budge t increments has an impact on

t rade-offs . Large budget increases a re u su ally associated

wit h i nc re ased defense spending during times of war when a

greater share of the budget i s devoted to wartime needs.

147 See Russet t , "Who Pays for Defense?" 419.



146

War, then, is a central factor in determining the direction

of t rade-offs .

Defining the Variables

A number of problems were encountered in tes t ing the

hypotheses presented. The f i r s t problem is that data are

not uniformly ava ila ble fo r a l l the variables of in terest .

Well def in ed dat a on aggregate (superfunction) budget

items are available beginning in 1940. Data on important

subfunctions, however, are not generally available before

1963. In fact , many of the components of domest ic spending

did not e xis t p rio r to the 1960s.

Party unity data a re a va ila bl e for the years between

1953 and 1988. 148 Consistent annual demographic data are

available from 1960, while data on Soviet mili tary expen-

148 A few missing points in th e in te rim (e. g . , 1961) areimputed from exis t ing data. This procedure was executed withStata software, using the "impute" command. Missing pointsthus generated were substi tuted for missing points within theexis t ing range (1953-1988).

The variables used in imputing the data includ ed theperc en tage o f seats held by each party in both the House andSenate, the number of to ta l unity votes in each chamber, par tyuni ty votes for each party in both chambers, the party unityvotes for Southern Democrats in each chamber, the pres ident ' s

party, the success ra te of th e p re sid en t in each chamber, inCongress, and the number of b i l l s on which he took a posit ion(omitting the dependent variable) . See Stata ReferenceManual, Vol. 2, 347-51.

This procedure produced resu l t s t ha t d if fe red very l i t t l efrom averaging between contiguous data in the ser ies , butprovided a more objective procedure for handling the problemthan mean, "best-guess" estimates. No data were imputed outside the span of years for which data were available.



149

.

147

di tures a re a va ila ble beginning in 1963.

The select ion of the best e stimate o f the dependent

var ia ble (s ) a ls o presented a challenge. Two al ternat ive

indicators for trade-offs are proposed by Berry and Lowery.

The two measures do not behave in the same way, which would

produce different resul ts depending on which calculat ion is

used. To avoid ad hoc specification, a decision had to be

made between proportional and difference based calculations

of the dependent variable(s) .

In the i r discussion of the i r measures, Berry and Lowery

make the following observation:

We believe that the modified difference measure, DIFF(A:B), i s superior to the pure proport ional measure,PROP(A:B), for several reasons . Firs t , t he p ropor ti on al measu re can yield an extremely high value when thepool i s very small. For example, i f the pool i s $1 ,where AMT(A) = 100 and AMT(B) = -99, the indicatorPROP(A:B) takes on the value 199. But jus t a s l ight

change in the outcome to (102, -97) r es ult in g in a hugedrop in the score on PROP(A:B) to 39.8. Consequently,empirical analyses relying on PROP(A:B) for measuringt rade-off outcomes have the potent ia l to be seriouslydistorted by a few extreme, and perhaps misleading,out l ie rs on the dependent variable . 149

In l ight of the i r comments, I constructed measures

based on both formulae for comparison. 150 In a l l , nine

t rade-offs were examined, which yielded 18 pairs of ser ies .

I found tha t the proportion based measure did produced

Berry and Lowery, "An Alternative Approach," 687.

150 To provide longitudinal comparability, a l l outlaysare converted to 1982 constant dollars before any furthercalculat ions are made.



148

extreme outl iers in several instances.

For example, in the tra de -o ff in education values for

proport ion t rade-offs for defense ver su s educa ti on exploded

between 1986 and 1988. This measure seems to exaggerate the

t rade-offs between the two categories.

The following i l lus t ra t ions show how the component ele-

ments of the education t rade-off are related as percentages

of the federal budget and are transformed into measures of

t rade-offs using both the formulae.

o Education Total o National Defense: Total.

80 CD

>-m

70 -...:::J

0

60

-o

....

500.......

40 0

....

c:30 CD

Co)

t-

20CD

Q.

10

0

100

90

I I I I I I

--

-• -

-

- T r e

-- '": '"

- f-

1 9 ~ 1 O 1950 1960 1';'0 1980 1990

100

90

GJ 80>-lU- 70...:::J

0

-

60m....0 50......0 ~ ....II

30:U

t-Il20.

10

0

yearDefense Versus Education OUtlays

FIGl.TP£ 15

The t rade-off measures calculated for the these two

categories i l lus t ra te the problem with t he p ropo rt ion mea-



149

sure. Values for both measures are shown in Figure 16

below. At f i r s t glance, i t appears tha t the proport ion

based measure does not real ly vary unt i l 1986-87. This

appearance i s misleading. Values for the pre-1986 period

are compressed for presentat ion on a common axis w ith the

difference based measure , which has extremely high values in

the l a t t e r port ion of the ser ies (See Figure 16). Closer

s imi lar i t ies in the variance of both measures during the

ea r l ie r per iod is hidden by these extreme out l ie rs .

347.145

o Difference Meastre to ProporUon Mea5U'e

-197.546

1960 19 0 1980 1990

Defense versus Education

Difference ancI Proportion Measures

FIGURE 16

When the l a s t four years are removed, and both vert ica l

axes are rescaled, values for the previous period are more



150

clear ly discernible . As in the case of the heal th expendi-

tures , the proportion based measure varies much less than

the difference based measure. Wnen the two measures for the

t rade-off pr io r to 1986 are examined, th e v aria nc e of the

proportion based measure is more vis ib le . Both measures

resemble one another more closely during th i s period, but

the p res en ce o f extreme out l ie rs for the l a t t e r period has

the potent ia l for " se rio us ly d is to rt in g " th e analysis .

Similar dis tor t ions might be expected in measures of other

t rade-offs as w ell.

o Difference '" Proportion

0

GJCD

L '-j

::J

II ICD

lUlU

CDCD

:z

lUc:

U 0

c: 0 -..,lIJ L'- 0lIJ C.- 0-... L

0 0..

1 1960 1 0 1980 1 0Defense versus Education: 1940-1986

Oifference and Proportion Measures

FIGURE 17

The implications of choosing one indicator ra ther than

another, for the subsequent analysis , a re obvious. I agree



151

with Berry and Lowery concerning the superiority of the

difference based measure. Comparing the two indicators, the

difference based measure appeared to be a more rel iable

indicator . For th is reason, th e d if fe rence based measure i s

used throughout the following analysis .

With th is matter set t led, I turned to the thornier

problem of defining the t rade-off between defense and domes-

t i c expenditures. The concept of "domestic" spending i s

widely used but infrequently defined. 151

Two definit ions of discretionary domestic spending are

examined. Following work by Russet and Peterson, a t rade-

off measure was constructed that aggregates several subfunc

t ions . 152 These categories include the subfunctions of

energy, commerce and housing credi t , health research, educa-

t ion & t raining of health care workers, education, general

government, veterans benefi ts , science, and t ranspor ta t ion.

Data on most of these subcategories are available beginning

in 1963 (Data for the independent pol i t ica l varia ble s a re

avai lable beginning in 1953).

Using th is highly disaggregated def ini t ion of domestic

151 Berry and Lowery do not define th e term by referenceto any par t icu lar items, but leave simply refer to CITIBASEdata used in the analysis . See Berry and Lowery, "An Alternat ive Approach," 694.

152 See Russett , "Who Pays?" 419; cf. Paul E. Peterson,"The Rise and Fall of Special Interest Poli t ics ," Poli t ica lScience Quarterly. 105 (Winter 1990-91): 548; Berry andLowery, "An Alternative Approach," 694-5.



152

spending would enta i l losing a decade of observations. For

th is reason a second def in i t ion i s used to provide a longer

ser ies . The second def in i t ion re l ies on a s l ight ly higher

level of aggregation. To construct a longer ser ies for

domestic spending, I use the superfunction for the above

items instead of component subfunctions. Data on a l l re le

vant superfunct ions are avai lable beginning in 1940 as are

data on to ta l defense outlays. Using each of the defin i -

t ions of domestic spending, two measures of t rade-offs were

const ruct ed for comparison (see Figure 18) .

o

A Aggregated Data

1960

yearComparison of Domestic Trade-off Measures

FIGURE 18

o

I use the second definit ion of domestic spending be-



153

cause i t provides a longer series and th e emp iri ca l dif fer

ence between the two indicators i s negligible. 153

Specifying the Basic Model

This section explains how I arrive at the specif icat ion

of a basic t rade-off model. I use the aggregate domestic

t rade-off as the dependent var iable . Several preliminary

analyses are s cr ut in iz ed to construct the most parsimonious

model possible. The model constructed is the building block

for analyzing t rade-offs between defense and heal th and

education, between functionally s imilar programs in defense

and civ i l ian programs, and trade-offs between components of

defense budget.

The s tructure of the equations tes ted i s quite simple.

Linear models taking the following the following form are

constructed:

Y = a + Xl + X2 + X3 + Zl + Z2 + Z3 + e (7 )

Where y is a t rade-off , X i s a pol i t i ca l variable and Z i s a

demographic or economic var iable .

I began my analysis with an examination of bivariate

re la t ionships with simple (t-1) lag s tructures . No s igni f i -

153 Sta t is t ica l ly , the two measures are almost ident ical .The Pearson product moment co rr el at ion coef fi ci en t obtainedbetween the two measures i s .99, and an R2 of .98, s ignif icanta t p < .0000 was obtained using ordinary leas t square regression.



154

cant bivariate associat ions were found for any of the pol i t -

ica l variables, but some of the demographic variables ap-

peared to be s ignif icant . 1s4 Bivariate analysis , however,

does not adequately address the relat ionships that I have

hypothesized.

The interaction of part ies competing for control of

government p o l i ~ requires a more complex form of analysis .

Mult ivar ia te analys is , however, was confounded by the exis -

tence of autocorrelat ion (serial correlat ion) ,

mult icoll ineari ty and the unavailabil i ty of data for impor-

tant control v ar ia ble s fo r certain time periods.

The presence of ser ia l correla t ion tends to inf la te the

significance of coeff icients . Stat i s t ical ly s ignif icant

ser ia l correlation was detected in almost every model es t i -

mated, as indicated by Durbin-Watson s ta t i s t i c s . OLS re -

gression models were re-estimated using a Cochrane-Orcutt

(CORC) regression procedure, a common general ized regression

technique f or adj us ting for ser ia l correlat ion. 1ss The

CORC adjus ted regress ion routine i s not applicable when the

series i s broken because th e ro utin e requires contiguously

154 Unity among Southern Democrats was s ignif icant a t t ,but was not s ignif icant when lagged one year , or when used inmoving averages (MA) models discussed below.

155 The method i te ra tes t he reg ress ion unt i l rho converges(to a tolerance of .0005) or 99 i t e ra t ions , whichever occursf i r s t . This procedure was executed with Stata 3 .0 . SeeComputing Resource Center, Stata Reference Manual Vol. 2(Santa Monica: Computing Resource Center, 1992), 200-3.



155

ordered data for the calculation of rho.

Preliminary analyses of the data produced confounding

resu l t s , which i s not uncommon in models have some degree of

seasonal i ty. Noting tha t almost every model estimated

contained s ignif icant levels of autocorrelation I returned

to the problem o f spe cif yi ng the lag s tructure of the model.

McDowall, McCleary, Meidinger and Hay provide a useful

comment on the problem in the i r work on interrupted time

series analysis :

Trend and seasonali ty, which are quite common in socialscience time series , and random error tend to o b s c u ~ eany intervention. I f the model does not account forthese types of noise, the analysis will be confounded. 156

To f i l t e r noise without over-complicat ing the analysis ,

I smoothed the pol i t ica l variables by means of three-year

moving averages. 157 Converting the dependent var iable in

the same way produced superior s ta t i s t ica l resu l t s , but made

in terpretat ion more di f f icu l t and causal tes ts more tenuous.

The three-year average i s used for the e st imat ion o f the

156 See David, McDowall, e t ale Interrupted Time Series

Analysis, Quanti tat ive Applications in the Social SciencesSeries . (Beverly Hills : Sage Publications, 1980), 14.

157 The formula used was the sum of the two l as t yearsand th e cu rren t year ' s values divided by three . This procedure may seem odd, but I believe tha t pol i t ical control i s notwrested away in the year that the outlay i s made. Thus, somedegree of unity in tha t year should, theoret ical ly , bere la ted .



156

basic t rade-off model. This procedure also proved to be a

parsimonious so lu tio n to the theoret ica l problem of model

l ing the impact of pol i t i ca l unity on subsequent outlays.

The three-year moving average i s based on a theory of how

policy makers are involved in program development and imple

mentation.

As mentioned in Chapter I I , some policy makers act as

"f ixers ," or what I have termed "shepherds," involving

themselves throughout the l i fe of a policy. Such people

follow a policy through i t s development, enactment as law,

and implementatio n. For th is reason I expected part isan

in flu ence s t o span a number of years. I had original ly

postulated that , af te r l eg is la t ive passage, shepherding

should be fe l t after a two-yea r span. This approach incor

po rates the level of par ty unity during that p erio d as well

as the year in which the outlay is made, because the success

of policy shepherds should depend on the i r abi l i ty to mobi

l ize support of the i r favored policies.

This procedure seemed to eliminate some of the noise in

the models. No seasonality in the measures of party unity

was detected; however, there appeared to be some seasonal i ty

in the percentage of ro l l -ca l l votes in the House where

pa rty ma jori ti es opposed one another ( to ta l unity votes) .

Stat i s t ical ly s ignif icant differences in to ta l unity votes

were found in elect ion years, perhaps as part of an underly-



ing effect of electora l cycles .158 Despite the elimina-

157

t ion o f su sp ec ted seasonality in the independent var iables ,

autocorrelation remained a problem, but was corrected

through the use of the CORC regression procedure.

Another problem encountered, was to control for major

changes in the party dominance of Congress. For two, non-

consecutive Congresses in the la te 1940s and 1950s Republi-

cans enjoyed a majority in both the Senate and House. Each

period las ted only two years. However, as previously dis-

cussed, a more last ing sh i f t occurred in the 1980s, when

Republicans controlled the Senate from 1981 to 1986 (with a

Republican president) . For analysis of the ent i re series

(1955-1988), a dummy var iable was used to "control for" the

change in context. For the sake of consistency, th is vari -

able was also converted to a three-year moving average. 159

A second problem, with theoretical significance, was

the presence of multicoll ineari ty among some of the indepen-

dent var iables . Party unity measures for Democratic and

158 ANOVA models showed s ignif icant differences betweenelection year to ta l unity votes and non-election year votes.A pseudo R2 of .11, s ignif icant a t p<.02, was obtained in a

bivariate model. President ia l election years, however, didnot appear to be re la ted to the differences.

159 Results were also tested using a four-variable modelthat part i t ions the variance for each var iable according totime period ( i .e . Democratic unity before 1981, Democraticunity af te r 1980, Republican unity before 1981, Republicanunity af te r 1980). Results of th is procedure are robust andconsistent with the basic model .



158

Republican House members a re cor re la ted, which tends to

inf la te the "explained variance" as indicated by the R2 , and

makes assessing th e d ire cti on of the coeff icients di f f icu l t .

Despite these problems, th ere are compelling theoret ical

reason for u sing bo th variables together in the same model.

Theoretically, the two variables are inextricably

l inked. I f Republicans are not able to produce discipl ine

among thei r members, they are le ss lik ely to exploi t dis-

unity among Democrats and v ic e v ers a. The presence of

col l inear i ty between measure of Democratic and Republican

p arty u nity might confuse the signs of the coeff icients , but

subsequent analysis assured that the signs reported in the

models are in proper order.

Multicoll ineari ty was ident i f ied between several con-

t ro l var iables . Col linea ri ty ex is ted between the var iables

related to foreign affa i rs and domestic pol i t i ca l variables ,

and between demographic variables . The re la t ionship between

the two demographic variables i s not surpr is ing. The per-

cent of the population aged 65 and over and the p ercent of

t he populat ion aged 15 and younger , were highly correlated

(negatively). This correla t ion confirms common demographic

t rends in the distr ibut ion of the populations in developed

countr ies . The re la t ionships between domestic pol i t ical

variables and measures pertaining t o in te rn at io na l r el at io n s

is more complex.



159

Changes in Soviet spending and bat t le deaths were both

significant during the Vietnam war. For the series spanning

1942 to 1988, bat t le deaths have a significant ly posit ive

relat ionship with on domestic t rade-off . Conforming to the

findings of other studies, large spikes in the series for

both var iables account for this finding. 160

The relat ive effects of war (bat t le deaths) and changes

in Soviet mil i tary spending on t rade-offs are di f f i cu l t to

assess, given the data avai lable . Unfortunately, data on

Soviet mili tary expenditures prior to 1963 are not reported

by the ACDA. A direct comparison of the effects of bat t le

deaths and changes in Soviet mili tary spending during the

1950s is , therefore, not possible using these data .

Regression resul ts based on different periods obscure,

ra ther than i l luminate important differences between his to r -

ica l periods. During the Vietnam War, changes in Soviet

mili tary spending peaked a t the same time American bat t le

deaths peaked. One might suspect tha t a similar pattern

would be found during the Korean War i f data were avai lable .

I t i s also very l ikely tha t Soviet mil i tary expenditures

were increasing a t a high ra te during World War I I , as

American bat t le deaths also increased. What is troublesome

about such patterns is that the pol i t ical significance of an

160 See Russett , "Defense Expenditures," 775; cf . Berryand Lowery, "An Alternative Approach," 694.



160

increase in Soviet mili tary spending may also be t ied to the

context of his tor ica l periods.

Obviously, the United States would have preferred that

the Soviets spend more on the i r mili tary during World War

II , and less thereafter . The peak in the percent change in

Soviet mili tary spending during the Vietnam War hardly needs

comment, except to note tha t i t resu l t s in col l ineari ty with

American bat t le death during that period. The two var iables

are both measuring aspects of the Vietnam War and are not

independent of one ano ther dur ing that period. After the

Vietnam War, American bat t le deaths are pract ica l ly non

exis tent , while Sov ie t spending continues to vary. I t i s

not surprising that the effect of bat t le death on t rade-offs

washes out when Soviet spending is introduced to the equa

t ion, because the period in which bat t le deaths are most

closely associated with t rade-offs is the 1941-1975 period.

Most of this period is dropped (due to missing data) when

Sovie t expendi tu res are included in the model. However,

when the post-1975 period is examined separately, changes in

Soviet defense spending are not significant ly associated

with trade-offs in federal outlays.

Significant bivariate coeff icients were obtained for

only three in terval-level var iables: changes in GNP, percent

of the population over age 65, under 15 (and thei r lags) ,



161

and unemployment were s ignif icant .161 No other bivariate

relat ionships were found to be s ignif icant .162 However,

none of the variables having s ign if ic ant b iva r ia te re la t ion-

sh ip s cont ri bu ted very much to th e o ve ra ll f i t or the s ig-

nificance of other variables in multivariate models.

Moreover, there is an unexpected posi ti ve r el at ionsh ip

between the percent of the population aged 65 and the domes

t i c t rade-off . Put different ly , as the percentage of the

population aged 65 and over grew, the more tha t the t rade

offs favored defense. Closer examination revealed that the

two var iables moved closely in the same direct ion during a

shor t period during the la te 1970s and ear ly 1980s. Unless

one i s will ing argue that there is a causal reason for a

posi t ive re la t ionship between the variables, which seems

unlikely, the re la t ionship must be spurious. For th is

reason, the var iable was dropped from consideration in the

development of a basic model. The percent of the population

aged 15 and younger had no s ignif icant impact in

multivariate analysis .

161 This re la t ionship will be examined further in ChapterV. I bel ieve that including unemployment as an independent

variable is a misspecification. The poss ib i l i ty that unem-ployment i s a dependent var iable i s considered in subsequentanalyses.

162 Signif icant differences between tradeoffs weredetected using categor ical , nominal-level variables for warand the Reagan presidency. In ANOVA models con ta in ing bothvariables , the Reagan presidency accounts for more of thevariance in domestic tradeoffs than war.



162

Changes in GNP contributed very l i t t l e to o vera ll f i t

of the models tested and washed out completely when the

Reagan administrat ion i s in cluded a s a variable .

Extreme out l ie rs , or "influent ia l observa ti on s, " a re

responsible for producing signif icant regression coeffi

cients between percent changes in GNP and the aggregate

domestic t rade-off . Extreme values for both variables oc

curred during World War I I and the Korean War. Changes in

GNP were dramatical ly affected by the dynamics of the war

time economy and subsequent reconversions. Thus, a huge

port ion of the variance in both the t rade -o ff v a ri ab le and

the change in GNP i s at t r ibutable to the shocks of war.

Jus t four or five observations account for th e o ve ra ll

significance of the associat ion. Figure 19, below, shows

how the two variables have varied over time. The large

"ou tl ie rs " a re obvious. Analysis of the period af te r 1955

did not produce signif icant coeff icients for changes in GNP.

In fact , changes in GNP washed out completely in multiple

regression models c orr ec te d f or autocorrelat ion. Alterna

t ive est imators fared no bet ter . Lags of f i r s t and second

order, moving averages, and logarithmic transformations also

fai led to survive CORC r eg ress ion analysi s a t signif icant

levels . Moreover, I detected no signif icant effec ts with

other variables in multivariate models.



163

25.477618.066

Q.

...

z...

(,!)

a cICD -D •1J a tCo C.... eu....

u

...III.. c:

EQ)()a c-o •.

1960 1970year

Domestic Trade-off and GNP

-13.8283

FIGURE 19

The occur rence o f large trade-offs during wartime, and

s hif ts a fte r wars also obscures the p o t ~ n t i a l effect of the

partisan aff i l ia t ion of presidents.

Presidential party aff i l ia t ion was not s ign if ic an t i n

any model specified. In part , the differences between Demo-

cratic and Republican presidents i s confounded by the coin-

cidence with war.

Franklin Roosevelt, Harry S Truman, and Lyndon Baines

Johnson a l l presided over wartime periods, as measured by

bat t le deaths. Republican presidents held office during the

post-Korean and Vietnam confl ic ts , when spending shif ted

away from defense. Thus the presence of war seriously



164

contaminates any analysis of president ia l influences.

Table 7 below descr ib es t he pattern. War i s defined as

any year where u.s. bat t le deaths occurred as reported by

the Department of Defense. 163

TABLE 7

President ia l Party and War1940-1990

President ia l PartyDem. Rep. Total

-------+----------------------+----------Peace I 12 21 I 33. War I 13 5 I 18-------+----------------------+----------Total I 25 26 I 51

In any case, no s ta t i s t ical ly s ignif icant direct re la-

t ionship was detected between president ia l party and the

domestic t rade-off . Table 8 i l lus t ra tes that domestic

t rade-offs do not correspond to the part isan control of the

presidency in the way that I had hypothesized. This also

163 Data on bat t le deaths was compiled from Sta t i s t i ca l

Abstract of the United States, Meid, Pat and James M.Yingling. Operations in West Korea: U.S. Mar ine Operations inKorea 1950-1953, Vol. V, (Washington, D.C.: HistoricalDivision Headquarters, U.S. Marine Corps, 1972), 757; and"Army Batt le Casualties and Nonbattle Deaths in World War I I :Final Report," Prepared by Sta t is t ica l and Accounting Branchof the Adjunct General, (Program Review and Analysis Division,Office of the Comptroller of the Army, C.C.S., June 1953,mimeo) .





166

Causation is not clear . House Democrats deferred to

pres ident ia l l eade rship i n. escalating the war, with the

infamous Gulf of Tonkin Resolution. But resistance and

discord ensued as the war dragged on. Defections in ro l l

ca l l votes among Democrats 'increased during th is period.

Figure 20 shows the relationship between Democratic

unity and the domestic t rade-off during th is period. Data

points are labelled by year. . The change in Democratic unity

seems to confirm what has been in tui t ive ly understood about

the effec t of the war in Vietnam on pol i t ical ins t i tut ions

and po li cy d ir ec ti on s of the par t ies .

Years N i Reported Battle Deaths

...- ...

0 0

• •J II I' 0 ' 0

to IVt . t .to- t -

U U.... 0 0

......ft IS)

U 41)-

e e0 0c 0

DeID. tJn1ty: Mov1ng Average

Democratic lJnity During the Vietnam War

FIGURE 20



167

I feel that the data a ccur ate ly por tr ay th e indecisive-

ness and d isco rd within the Democratic party toward the end

of the war, and a certain aimlessness af ter the war. What

i s worth noting about these pat terns in the data i s that

these t hr ee va ri ab le s: Democratic party unity, bat t le

deaths, and changes in Soviet mili tary expenditures are a l l

closely correlated between 1965 and 1975. While Democratic

p arty u nity i s negatively associated with changes in Soviet

mili tary spending, Republican party unity is posit ively

associated with Soviet mili tary spending.

With these underlying re la t ionships in mind, I estilttat-

ed a multivariate model containing only the pol i t ical var i -

ables. Data for th is portion of the analysis a re a va il ab le

from 1955 to 1988. The data were analyzed f i r s t with OLS,

then sub je ct ed to the more rigorous CORC procedure to con-

t ro l for ser ia l correlat ions .

The most inf luent ia l var iable in the model i s the dummy

variable for the Reagan administration, which, i f l e f t out,

destroys the model. This might seem to put the model i t se l f

in doubt.

However, i f th e p erio d before 1981 i s examined, s imilar

resul t s obtain, but w ith a substantial reduction in the

expla ined variance and robu st nes s o f th e e stima te s. I take

this to mean that the pol i t ic iza t ion of budget allocations

reached a new plane under the Reagan administrat ion.



168

The resul ts of a regression analysis of the effects off

part isan unity, using only the po l i t i ca l var iables , on the

t rade-off between defense and domestic outlays a re re po rte d

in Table 9 below.

TABLE 9

Explaining the Domestic Tradeoff

(Poli t ics Alone)

1955-1988 1956-1988

Variable t OLS CORCRegression Regression

Constant -309.71 -317.38(260.64) (259.64)

Democratic Unity -11. 01*** -10.91***(2.77 ) (2.77 )

Republican Unity 14.82*** 14.82***

(4.44) (4.38)

Reagan Shif t 193.82*** 192.85***(0 = 1955-81) (22.30) (22.18)(1 = 1981-88)

Adjusted R2= .71 .72

Durbin-Watson = 2.17

N = 33 32

Unstandardized regression coeff icients a re r epor ted.

Standard errors are in parentheses.t All variables a re th re e- ye ar moving averages (MA).* Signif icant at the .05 level** Significant at the .01 level*** Signif icant a t the .001 level

Upon closer examination, the s tructure of the relat ion-



169

ships i s clearer , the relationship between the party unity

variables and the domestic t rade-off i s apparently stronger

than i t had been during the previous period. Table 10 show

nearly ident ical relat ionships when th e v aria nce i s par t i -

t ioned between the two periods --before and af te r the st ruc

tura l shi f t , or what I have labelled the "Reagan Period."

Spli t t ing each of the p ar ty unity variables into two

separate variables broken down by period, I subst i tuted

zeros for al ternat ive periods. This "revised model" was re-

est ima ted using the procedures employed above.

The direct ion of the par ty unity coeffic ients b e f o r ~and af te r the 1980 elect ion are similar, and in hypothesized

direct ions, I graphed the effect of Democratic party unity

on the domestic t rade-off to show tha t the sh i f t i s upward

in the in tercept for t rade-offs during th e 1980's when

Republicans take the Senate. The resu l t s of th e rev ise d

regression model are reported in Table 10.

The extent of the effect of the s t ructura l sh i f t on the

domestic t rade-off i s evident in Figure 21 , where Democra tic

party uni ty ( th re e-ye ar moving average) i s graphed against

the domestic t rade-off . Regression l ines have been placed

in the graph to show the relat ionsh ips between the var iables

during each period.



Table 10

The Domestic Trade-off(Using Part i t ioned Variance)

1955-1988 1956-1988

Variable tOLS CORC

Regression Regression

Constant -308.22 -313.57(267.42) (267.42)

Dem.Unity (before 1981) -11. 04** -10.69*(4.24 ) (4.25)

Dem. Unity (af ter 1980) -10.85** -10.95**(3.85 ) (3.81 )

Rep. Unity (before 1981) 14.82** 14.56*(5.75) (5.67)

Rep. Unity (af ter 1980) 17.09** 17.27**(5.13) (5.08)



N = 33 32

Standard errors are reported in parentheses.t Unity var iables a re t hr ee -y ea r moving averages (MA).* Signif icant a t the .05 level** Signif icant a t the .01 level

170



171

1955-1980 aM 1982-1988

...."t -

oICD

' 0«tr.....

U...c

+ 'CD

CDIi ;

ocolI63

1110· t----

10571-

i_

t969 i_877

1970 :lSl7a

o

...

...

oI..'0tar.....

u

+ 'en

..

..oD

so 85 90DetL Unity: MoV1ng Average

Comparison of the Reagan Years with EarHer Period

FIGURE 21

The re la t ionship between Democrat ic unity and the

domestic t rade-off is obv ious ly not as strong prior to 1981

than it i s a f t e r 1981. I in terpre t the pattern a f te r 1981

as a Democratic counter-offensive a ga in st th e gains made by

the Republicans.

These resu l t s are fa i r ly robust against al ternat ive

specif icat ions . Adding changes in Soviet mili tary spending

and ba t t l e deaths, I obtain a sl ight ly higher coeff icient of

multiple det ermina ti on , but I suspect that col l inear i ty i s

in f la t ing model estimates.

The var iable "bat t le deaths," though not s ignif icant

i t s e l f , might conceivably be par t of the explanation of



172

certa in other disaggregated t rade-offs , or act as an in te r -

vening variable. For theoret ical reasons i t is worth keep

ing for the sake of controlling for the br ief wartime period

at the beginning of the ser ies , to e lim in at e th e poss ibi l i ty

that wartime distort ions might be responsible for the re-

su l t s . Demographic and economic variables contributed noth

ing, or spurious cor re la tio ns , t o the model and were dropped

from the basic model.

The elimination of so many of the convent ional exp lana

t ions for changes in government spending, because of the i r

lack o f exp lana to ry power, merits further comment. As noted

in Chapter I , characterizations of the budget have long been

dominated by the school of thought that budgets change in

smal l increments .

Budget changes are usually minor, and typical ly associ

ated with long-term trends. The pat tern among t rade-offs ,

as conceptualized here , shows l i t t l e evidence of the kind of

s tab i l i ty suggested by the notion of incremental budget ing.

Admittedly, t he d if fe rences found in th is ana ly sis a re

bui l t upon the idea that changes occur at the margins, but

the variance i s much wider than what an incremental model

would imply. The percentage changes during the 1980s are

not indicat ive of marginal adjustments. Changes during th is

period were widespread and comprehensive, which i s not the

sor t of behavior that can be predicted according to a theory



173

of incrementalism.

Moreover, these resul ts support the hypothesis tha t a

po l i t i ca l par t ies do influence the way budgets are al locat-

ed. Turning to the analysis of the other t rade-offs , I

employ the basic model (Table 11) to t es t the hypotheses.

Table 11

Explaining the Domestic Trade-off(The Basic Model)

1964-1988 1965-1988

Variable OLS CORCRegression Regression

Constant -2210.29*** -2397.60***(578.95) (522.27)

Democratic Unity -10.04* -8.79*(4.52) (3 .94)

Republican Unity 25.26*** 25.98***

(5.55) (4. 65)

Reagan Period 198.99*** 183.93***(39.24) (35.75)

ASoviet Spending1t _1l 953.01** 992.56***(280.12 ) (253.10)

Batt le Deaths -0.00 -0.00(O.OO) (O.OO)

Adjusted R2 = .75 .82Durbin-Watson = 2.75

N = 24 23

Standard errors are reported in parentheses.* Significant a t the .05 level** Significant a t the .01 level*** Significant at the .001 level



174

Explaining Trade-offs

Following the same explanation, the t rade-offs between

defense and outlays fo r h ealth and education are substi tuted

in the models below. Tables 12 and 13 re po rt th e reg re ssion

resul t s fo r the 1964-1988 period:

Table 12

Explaining the Health Trade-off

1964-1988 1965-1988

Variable OLS CORC


Constant -1984.02* -2538.66***(834.26) (673.74)

Democratic Unity -20.83** -15.56**(6.51) (5.33)

Republican Unity 44.90*** 49.82***(8.00) (6.43)

Reagan Period 214.00*** 131.34*(56.54) (50.29)

ASoviet Spending ( t - l l 33.62 -143.20(403.65) (320.73)

Batt le Deaths 0.00 -0.00(0.00) (0.00)



N = 24 23

Standard errors are reported in parentheses.* Signif icant a t the .05 level** Signif icant a t the .01 level*** Signif icant a t the .001 level



175

Table 13

Explaining the Education Trade-off

1964-1988 1965-1988

Variable OLS CORC


Constant -1513.26* -2157.65**(636.00) (532.75)

Democratic Unity -9.92 -6.56(4.96) (4.00)

Republican Unity 21.17** 24.37***(6.10) (4.71)

Reagan Period 178.81** 139.74**(43.10) (36.18)

ASoviet Spending 1t_11

591.28 738.77**(307.72) (258.53)

Batt le Deaths -0.00 -0.00(0.00) (0.00)

Adjusted R2 = .61 .76


N = 24 23

Standard erro rs are reported in parentheses.* Signif icant at the .05 level** Signif icant at the .01 level*** Signif icant at the .001 level

Results for the health t rade-off are consistent with

the results for the domestic t rade-off . However, the educa-

t io n t rade -o ff model does not produce s ignif icant coeff i -

cien ts for democratic unity.

In analyzing the resu l t s by means of the par t i t ioned



176

variance model constructed along the l ines of the domestic

model ( i .e . , divided before and af te r the Reagan period),

none of the variables was significant . Co ll in ea ri ty i n this

model made resul ts almost un-interpretable. The signs of

the coef fi ci en ts i n the CORC ad jus ted regres sion analysis

a re c on sis te nt with the theory, bu t the signif icance of each

variable cannot be determined . Standardized regression

coeff ic ients for the variables are reported in Table 14:

Table 14

The Education Trade-off(Using Part i t ioned Variance)

1964-1988

Variable t

Constant

Dem. Unity (be fo re 1981)

Dem. Unity (after 1980)

Rep. Unity (before 1981)

Rep. Unity (af ter 1980)

Adjusted R2 =CORC Adjusted R2 =Durbin-Watson =N =

OLSRegression

-332.91(0.329)

-6.19(0.187)-2.20(0.417)6.73(0.205)3.63(0.471)

.52*

.442.34

24

Standardized regression coeff icients are reported.T-scores are in parentheses.t Unity varia ble s a re th re e-y ea r moving averages (MA).; The model is s ignif icant at the .003 level using OLS,

and a t the .013 level using CORC regression.



177

Jointly, the variables appear to be significant . This

is an ar t i fac t of the shif t in context, or what I have

termed the "Reagan Period." The variables are not s ign i f i -

cant precisely because they are not s ignif icant . When the

period between 1955 and 1980 i s examined separately, no

significant coeff ic ients are obtained.

What is s ignif icant is the sh i f t in context. Analysis

of variance indicates that the part isan sh i f t in control of

government during the Reagan period accounts for over 50

percent of the variance in the education t rade-off between

1964 and 1988 (ANOVA Adjusted R2 = .54, Significant a t

.0003, with 22 degrees of freedom for the three-year moving

average, i . e . , 0, .5, 1). Put dif ferent ly , the sh i f t toward

Republican dominance during this period is what accounts for

the model's significance. Part i t ioning the variance by

breaking the un ity var iabl es in two, according to period,

makes th i s clear .

This pat tern i s so consistent that an analysis of

variance of the difference between the periods for a l l of

the dependent variables is warranted. Table 15 reports

resu l t s of the analysis of each dependent var iable .

There are problems in using analysis of variance mod-

els , with time series, since the procedure offers no remedy,

or t es t for autocorrelat ion. Moreover, the la t te r port ion

of the series contains only eight observations i f both



178

admini st ra ti ons a re counted, six i f only the sh i f t in the

Senate i s counted. For these reasons, the following resu l t s

are extremely tenuous. I have coded the years 1982 to 1988

as 1, and 1964 to 1981 as zero to match the period in the

above analyses as closely as possible without using a moving

average.

TABLE 15

Differences in Trade-offs

Before and After 1981

Dependent Variable

Defense versus:

ANOVA R2 SignificanceLevel

Domestic Trade-off

Heal th Trade-o f f

Education Trade-off

Defense:(Functional Trade-offs)

Personnel vs. Training*

.40

.15

.40

.21

0.0005

0.03

0.0005

0.01

R & D vs. Total Science**

R & D vs. Basic Research t

R & D vs. Civil ian Research;

Mili tary Trade-off:

Procurement vs. Personnel

N = 24

.006 0.3

.23 0.01

.28 0.004

.04 0.7

* Employment t raining and education** Total Science minus space f l ight and related outlays.; Subcategory of science for basic research.t Health and education research.



179

Overall there is a signif icant change in the way that

outlays were divided between categories. These patterns are

apparent in the multivariate analyses as well .

The resul ts of t he f unct iona l trade-off between outlays

for mili tary personnel and emploYment t raining does not

indicate Democratic unity played a part in determining how

th is part icular division occurred. Under more scrutiny,

using part i t ioned variance, Democratic uni ty and Republican

unity before 1981 appeared to be signif icant . Analyzing

th is period separately ( i . e . , excluding observa tions before

and af ter 1964 and 1981), only changes in Soviet spending

and bat t le deaths were signif icant . Again, the resul t s for

the parti t ioned variance model were driven by the shi f t in

1981. A similar pat tern was found in the science t rade-offs

in non-space science outlays and basic research.

However, in the t rade-off between outlays for defense

research and development and research in he alth and educa-

t ion, the findings a re con si st en t. Democratic unity and

Republican unity are signif icant even during the period

before Reagan.



180

Table 16

Explaining the Civil ian Research Trade-off

1964-1988 1965-1988

Variable OLS CORC


Constant -485.77 -520.90(270.16) (272.29)

Democratic Unity -7.22** -6.89**

(2.11 ) (2.10)

Republican Unity 11.11*** 11.32***(2.59) (2.51 )

Reagan Period 91. 44*** 87.50***(18.31) (19.42)

ASoviet Spending 1t _1l 142.53* 139.15(130.71) (130.91)

Batt le Deaths -0.00 -0.00(0.00) (0.00)


N = 24 23

Standard errors are reported in parentheses.* Signif icant a t the .05 level** Signif icant at the .01 level*** Signif icant a t the .001 level

In the t rade-off between cat egor ie s o f defense spend

ing, the model has no explanatory power a t a l l . Neither the

e ntir e s er ie s, nor th e p erio d before the sh i f t produced

s ignif icant resu l t s .



181

Summary

This study has p roduced three important findings.

Firs t , I have found some evidence of a part isan influence on

budget allocations , though the predict ive capacity of the

model does not apply to a l l of the t rade-offs examined.

Pa rt is an un it y has proven to be a useful means of measuring

the s trength of the par ty and i t s abi l i ty to implement

programs, though more precise measure could be constructed.

Second, the effect of presidents is not clear because

of the coincidence between Democratic presidents and war. I

bel ieve tha t th is fact has been overlooked in previous

research and may account for the fai lure to find a relat ion

ship between party and budget outcomes. 165 Nevertheless, I

must re jec t the hypothesis tha t presidential party has an

effec t on allocations .

Third, the effect of a s t ruc tura l change in the domi-

nance of a par ty can have a large impact on budget alloca-

t ions. Granted, I only have one such change, but the ensu-

ing sh i f t s in al locat ions are in accordance with a general

pat tern of part isan pol i t ics occurring prior to the shi f t ,

and i s exaggerated afterwards.

I also find evidence of an organized pol i t i ca l counter-

offensive on the part of the House Democrats in the wake of

165 See Russett , "Defense Expenditures and National Well being;" and Berry and Lowery, "An Alternative Approach."



182

a s t ruc tura l shi f t in power toward Republican dominance.

Democrats a re i nc reas ingly united in the i r voting

during the 1980s. Given th e d ef ea ts tha t they suffered in

electoral and policy matters, the upswing in party unity i s

consis tent with a poli cy o ri en ted interpreta t ion of that

measure. 166

To conclude, the findings presented in th is chapter,

using a re-designed measure of part isan influences, support

the theory tha t ideology has an important influence on the

way that budget allocations are determined. In the next

chapter, I summarize these findings, discuss the theore t ica l

s ignif icance of the findings and offe r conclusions regarding

the potent ia l for a peace dividend in the United States .

166 Cf. Brady , David , Joseph Cooper, and Pat r ic ia Hurley.liThe Decline of Party in the U.S. House of Representat ives,1887-1968."



CHAPTER V

SUMMARY AND CONCLUSION

principal Findings

Using the most common method of measuring part isan

influences, based on the number of seats and ideological

posi t ions of part ies to the l e f t and r ight , the hypothesis

that part isan in flu en ces are responsible for trade-off

outcomes i s not generally supported. However, when party

i nf luences a re operat ional iz ed in terms of party unity in

the United States, part isan influences are apparent and the

hypothesis that part ies effect trade-offs is supported.

In the case of the United S tate s th er e is evidence of

part isan influences on t rade-off behavior. These influences

are manifest a t two levels . At a st ructural level, the

model I have presented re l ies on the dominance of inst i tu-

t ions. When control of the major ins t i tu t ional players,

House, Senate and Presidency, shif ts from one party to

another, as i t did during the 1980s, budget trade-offs also

sh i f t toward the favored c at egor ie s o f the newly dominant

party . In th is case, Republican gained the upper hand and

t rade-offs shif ted sharply toward defense.

At another level , that is within a pa rt icu la r s tr uc -

tu ra l arrangement, the unity of part ies takes over. The

183



184

more united a party i s in voting, the more t rade-offs wi l l

move, ceteris paribus, in the direction of i t s favored

categories. The model i s essent ia l ly a unity driven model,

with ins t i tut ional dominance playing'an important role.

The part isan unity model succeeds in explaining aggre

gate t rade-offs in federal outlays, but i s not generalizable

to more narrowly defined t rade-offs . I t works best in

explaining the defense versus domestic t rade-offs , but does

not fare as well in explaining the spending mixture for de

fense programs or more narrow constructions of the defense

versus domestic spending t rade-off .

Given the relat ive success of the revised conceptual

izat ion and operationalization of part isan influences used

in th e a na ly sis of American federal outlays, a similar

approach may yield bet te r resul ts when applied cross-nation

al ly . Such an operationalization might re ly on very subtle

changes discernible to scholars who deal intimately with the

ins t i tu t iona l and procedural in t r icac ies of a given country.

The unity of part ies, or party so l idar i ty , might be

measured by the s tabi l i ty of cabinet posts , party membership

and candidacies, o r th e l ike . According to my theory of the

influence of part isan ideologies on budgetary t rade-offs ,

dissention with in th e ranks of a party should constrain i t s

achievement o f id eo lo gical g oals. I f members are not w ell

organized (or lack the majorities needed to p rev ail in

l eg is l at ive ba t tl e s) , the par ty ' s impact on budget outcomes



185

should be less than when members are well organized. 167

Conclusion: A Peace Dividend?

I have found that the unity of American part ies makes a

difference in the direct ion of budgetary t rade-offs . While

I have found a part isan influence on t rade-offs in the

United States, I do not f ind that president ia l party a f f i l i -

ation helps to explain t rade-offs . In f ac t, p re sid en tia l

party af f i l i a t ion , as a variable, i s severely contaminated

with the coinc idence o f warfare. Nor does the Reagan admin-

i s t ra t ion, by i t se l f e xp lain th e shi f t s in outlays during

the 1980s. I f my thesis is correct, i t was not Reagan, the

man, or even Reagan the president tha t produced these chang-

es. Instead, i t was the general bias of the Republican

party tha t changed the direct ion of the t rade-offs .

Firs t of a l l , Reagan was not alone. Noting the coming

proposals for increased defense spending, Russet t closes his

1982 ar t ic le stat ing that :

None of the other variables in our model account forthese sh i f t s . There i s no shooting war, and the otherv aria ble s a re n eith er very vola t i le nor, in our equat ions, very powerful anyway. Most important, the data

167 In parliamentary systems one might look a t thes tab i l i ty of membership in the party, the number of resignat ions or other signs of discord for clues as to how th is couldbe measured. I n cou ntrie s where party voting varies in thesame way as i t does in the United States, voting s ta t i s t icsmight be equally useful.

For a discussion of how executive party aff i l ia t ion mightalso be reconceptualized, see Appendix c.



186

show that i t i s not merely the pres ence o f a Republicanin the White House which made the difference; the

Reagan presidency i s different . 168

What Russet t does not notice that Reagan is also pre-

siding over a peacetime ascendance in the Republican party

(not associated with the end of a war) with the Senate being

controlled by the Republicans. By comparison, Dwight Eisen-

hower presided over the end of a war, when t rade-offs gener-

al ly turn away from defense. He enjoyed a Republican major

i ty in Congress for only two yea rs c oin cid ing w ith demobili-

zation. In fairness to Russett , no one knew in 1982 tha t

Republican dominance of the Senate would l a s t s ix years.

But the large sh i f t in budgetary t rade-offs occurred in

con junct ion wi th a partisan sh i f t in the party control l ing

the Senate, which las ted for most of President Reagan's

term. This sh i f t in the control of the Senate is crucial .

While the Senate is typically not viewed as the centra l

actor in the budget process, the "correlat ion o f forces"

during the 1980s favored Republicans via control of two of

the three major ins t i tut ions involved in the budget process.

Had the domestic trade-off moved in the direct ion of

domest ic spending when this occurred, the whole enterprise

of using an part isan explanation would be in serious doubt.

Moreover, as predicted, the t rade-off moves back toward

168

original) .Russet t , "Defense Spending," 776, ( i ta l ics in



187

domest ic spending as the unity of Democrats in the House

increases. I t seems that House Democrats mounted a counter

offensive during the 1980s, with unity increasing as the

t rade-off returns to "normal" levels . This is entirely

consistent with theoret ical expectat ions.

Thus, the con junc tion o f these events lends more, not

less , credence to the th esis th at Republicans produce de

fense-favoring trade-offs and, more general ly, that part isan

ideology i s a factor in t rade-off outcomes.

To the extent that th is assessment is valid , we should

expect that i f the Republicans win control of a t leas t two

of the three ins t i tut ional ent i t ies involved in the budget

process during peacetime, we should see another sh i f t toward

defense spending and away from domestic spending.

After such a shi f t , the abi l i ty of the part ies to

produce int ra-party consensus wil l determines the direction

of the defense ver sus domestic t rade-off within a modal

pattern of t rade-offs . I f one party i s unable to generate

such a consensus, a united opposi t ion could win, with a

resul t ing increase in expenditures for i t s favored catego

r i es of spending relat ive spending on i t s opponent's favored

categories. This is a competitive model. This f inding runs

counter to the idea that budge ta ry choice is merely an i l lu -

sion, the resul t of incrementalism or economic determinism.

I f part ies make the kinds of changes I have found, then the



188

electorate has a ra ther clear choice between policy al terna

t ives , but only i f the part ies themselves are united.

Pol i t ica l hegemony by major coali t ions of par t ies , or by

bureaucracies in other countries might explain why t rade

offs in those count rie s a re relat ively stable.

Finally, one of the most important implications of the

fin dings p re sen ted here is that the defense versus domestic

spending t rade-off appears to be driven much more by domes

t ic pol i t ics than by the arms race. Changes in Soviet mili

tary spending do not add very much (around three to five

percent) to the explained varian ce o f the models I tes ted.

This model does not suggest that the collapse of the Soviet

Union wil l produce changes in the t rade-off as large as

those produced by changes party unity. Thus, the collapse

of the Soviet Union should not have a large direc t impact on

the direction of t rade-offs in f eder al out la ys .

In c lo sin g, return to th e que stio n I posed a t the

beg inning o f the study. Should we expect a "peace divi

dend?" Can I answer this question? For the United States,

I believe that I can.

I f we define a peace dividend as a t rade-off which

favors domest ic spending, I believe that there should be a

dividend ar ising from the loss of the Senate by the Republi-



189

cans .169 After Republicans lost control of the Senate,

t rade-offs immediately moved back toward domestic spending.

Meanwhile, Democrats have been increasing the i r unity in

voting. I f th is trend continues, there i s a good chance of

real iz ing a dividend. However, the Gulf War may have wiped

out a dividend for 1991. War does play an important role.

Peace dividends do exis t . But during peacetime the use of

the term seems anachronist ic. My model indicates that the

unity of pol i t i ca l par t ies i s a more powerful influence than

the dynamics of arms races in determining trade-offs, thus

the concept of a "peace dividend" i s less accura te t han what

might be cal led a "domestic pol i t i ca l dividend."

In For a discussion of the potent ia l spi l l -over effectsor consequences of a peace dividend for the economy, seeAppendix D.



APPENDIX A

Data Sources

Cross National Data:

Part isan placement are based on the opinions of the following

specia l is ts :

Joe Adams United StatesFrancis G. Castles Austral iaRobert Cox BelgiumConstantine Danopoulos (not used) GreeceDavid Wilsford FranceEric Einhorn NorwayEric Einhorn SwedenEric Einhorn DenmarkEric Einhorn FinlandM. Donald Hancock Federal Republic of GermanyJohn Logue SwedenJohn Logue NorwayJohn Logue .DenmarkJohn Logue FinlandRafael Banon Martinez (not used) SpainJohn Wil liams CanadaBarclay and Joan Ward ............•......... United KingdomBirol Yeshilada (not used) Turkey

Data on the distr ibut ion of seats held by each party

place on a five point , l e f t - r igh t continuum are and placements

for countries where specia l is ts were not ut i l ized were based

on th e o rd er and descriptions found in the following sources.

Comparisons between sources indicate tha t the convention in

presenting tables showing elect ion resul ts are arranged in

l e f t to r ight pat terns. These pat terns are consistent with

more detailed descriptions found in many of the sources below:

190



191

1. McHale, Vincent McHale. Pol i t ica l Part ies of Europe

2. Mackie, Thomas T., and Richard Rose. The Internat ionalAlmanac of Electoral History.

3. Banks, Arthur S. ed. Pol i t ica l Handbook of the World. New

York: McGraw-Hill Book Company, 1988.

3. Facts on File (various issues)

4 Keesing's Contemporary Archives (various issues)

Demographic and economic data were obtained from the following

sources:1. World Tables (various editions)

2. Alvarez, Michael R., Geoffrey Garrett , and Peter Lange."Government Partisanship, Labor Organization, andMacroeconomic Performance." American Pol i t ica l ScienceReview 85, no. 2 (June 1991): 539-56 .

3. U.N. Demographic Yearbook (various edit ions)

4. Summer, Robert. , and Alan Heston. "A New Set of Internat ional Comparison of Real Product and Price for 130Countries, 1950-1985." Income and Weal th 34 (1988): 1-25.

5. Internat ional Labour Organization, Year Book of LabourSta t is t ics (various years) .

Cross-national budget data are taken from:

1. OECD Social Expenditures 1960-1990: Problems of Growthand Control. Paris : Organization for Economic Cooperationand Development, 1985.

2. u. S. Arms Control and Disarmament Agency. World MilitaryExpenditures and Arms Trans fe rs (va riou s years) .

3. Arbetman, Marina. WESTUP (data se t based on IMP es t i -mates) .

4. Internat ional Monetary Fund, Government Finance Sta t i s -

t i c s (various years) .

u.S. Data:



192

Pol i t ica l data on the United States were obtained from:

1. Ornstein, Norman J . , Thomas E. Mann, and M ichael J .Malbin. Vita l Stat i s t ics on Congress, 1987-1988 .

2. Congressional Quarterly 's Guide to Congress.

3. Davidson, Roger H., and Walter J . Oleszek. Congress andI t s Members.

4. Shaffer , William R. Partv and Ideology in the UnitedStates Congress.

U.S. Budget data are taken from:

1 .Budget of the United States Government: Fisca l Year 1991.

2. Budget of the United States Government: Fiscal Year 1992.

Demographic and economic data are taken from:

1 . Bogue, Donald J . Populations of the United States .

2. Sta t i s t i ca l Abstracts of the United States (variousyears) .

Data on ba t t le deaths were taken from:

1. Sta t i s t i ca l Abstracts of the United States (various

years) .

2. "Army Batt le Casualt ies and Nonbattle Deaths in World WarI I : Final Report."

3. Meid, Pat and James M. Yingling . Operat ions in West Korea:U.S. Marine Operations in Korea 1950-1953.



APPENDIX B

Questionnaire Construction

In collaboration with Pro fe ssor s Marina Arbetman and M.

Donald Hancock, th e ques tio nn aires were send to country

specia l is t . The questionnaires l i s ted the par t ies and i n i t i a l

placements, corrections to these placements were requested.

The following example i s for Canada.

Belou ar e th e i deo log ica l pos lt lons selec ted (i n boldchar ac te rs ) f or each party during th e period 1955-1985. ~ l e a s ereview th is i n f o r ~ a t i o n fo r your c o u n ~ r y .KeY:RL Radical LeftHL Hoderate Lef:cor C e n ~ e rHR Hoderate RightRR Radical Right

~ l e a s e note any revisions in the existing data by circling th eposition uhich better approximates th e ideological character of theparty .To expedite data ent ry . r et u rn th e completed form on orbefore June 15. 1991.

CANADA

Neu Democratic 1955 RL HL CT HR RR( N D ~ ) 1960 RL ilL CT IIR RR

1965 RL ilL CT IIR RR1970 RL ilL CT IIR RR1975 RL ilL CT IIR RR1980 RL ilL CT IIR RR1985 RL ilL CT IIR RR1990 RL ilL CT IIR RR

Liberal ~ a r t y 1955 RL ilL CT IlR RR(LI?) 1960 RL ilL CT IIR RR

1965 RL ilL CT IlR RR1970 RL ilL CT IIR RR1975 RL ilL CT IIR RR1980 RL ilL CT IIR RR1985 RL ilL CT IIR RR1990 RL ilL CT IIR RR

~ r o g r e s s i v e Conservative 1955 RL ilL CT HR RR(I?CI?) 1960 RL ilL CT HR RR

1965 RL ilL CT IIR RR1970 RL ilL CT IlR RR1975 RL HL CT IlR RR1980 RL ilL CT IlR RR1985 RL ilL CT IlR RR1990 RL ilL CT IlR RR

Ralli:!ent de s 1965 RL HL CT Hit RRCredit is tes

Social Credit 1960 RL ilL CT IIR RR( S C ~ ) 1965 RL HL CT Hit RR

1970 RL ilL CT HIl RR1975 RL HL CT HR RR

193



APPENDIX C

Measurements: An Agenda for Research

The common perception that the Reagan's presidency was

somehow d if fe rent r ai se s questions about the nature of

president ia l partisanship and ideology. Reagan has been

associated with the r ight wing of the Republican party as

opposed to party moderates. This assessment of his posi t ion

suggests that gradations of l iberalism and conservatism

within each of the part ies might be re fle cte d in presiden-

t i a l leadership. Impressionistic comparisons of Eisenhower,

Nixon and Reagan suggest that the ideological posit ions of

presidents might be measured by the i r public speeches.

As a war hero, whose professional career in the mili-

tary was marked by the experience of pull ing together Allied

generals, Eisenhower was not renowned for his outspoken

opinions. While his vice president, Richard Nixon, bated

the Soviet leader Nikita Khrushchev, and Ronald Reagan aided

in the efforts of McCarthy-era demagogues in black-balling

members of the entertainment industry, Eisenhower might be

best remembered as the man who coined the expression "mili-

tary industr ial c0I:ClPlex," in his farewell address. Despite

the cornmon character izat ion of military men as bureaucrats

indoctrinated in the interests of the mili tary, Eisenhower

warned against mili tary expansion. Though surely a conser-

194



195

vat ive, Eisenhower cannot be remembered as an ideologue.

Moreover, he was not a professional pol i t ic ian , which might

otherwise have contributed to more pronounced ideological

convictions.

By comparison, Richard Nixon part icipated in more

over t ly ideological debate. Meeting with Khrushchev in

celebrated public appearances, Nixon made himself known for

his anti-Communist views. But as president, Nixon opened

relat ions with China, and began the period of detente with

the Soviet Union. He seemed w illin g to modify his views

according to circumstance.

Reagan, on the other hand, seemed to be consistently

more provocative in his rhetor ic and less willing to modify

his posi t ion according to circumstance. By comparison to

ei ther Nixon, or his successor, President Bush, Reagan

seemed less apt to accep t Sov ie t overtures. In many ways,

Reagan might be the quintessent ia l "Cold Warrior." Using

phrases l ike "evil empire," Reagan seemed intent upon keep

ing superpower r ivalry alive. One wonders what Reagan might

have said as p resid en t, in the wake of the Tianamen Square

incident in 1989. Would he have been as restrained as his

successor? Even Democrats in Congress complained tha t Bush

was too sof t on the Chinese. Would Reagan have been so

docile?

The problem of measuring pres ident ia l i nt en ti ons po in ts

to where th is kind of research should be aimed. Measurement

is the centra l problem. Perhaps a measure of the ideologi-



196

c al p os itio n of a president could be developed using content

analysis of the i r public speeches. Coding schemes similar

to those employed in the work of Ian Budge and others might

allow us to compare presidents in ideological terms .170

Factor analysis of such data might y ie ld i n te rva l- leve l

measures of a l ibera l conservative dimension in presidential

rhetoric .

Jus t as congressional part ies vary in the i r unity,

presidents may be more or less con sis te nt in thei r rhetoric.

American part ies are not monoliths, and presidents of the

same par ty may be as dissimilar as members of the House or

Senate in thei r orientat ions. I f such measures could be

developed, the predictive u t i l i ty of the models might be

improved.

Based on the work of congressional scholars and survey

researchers, I postulated tha t Democrats are "to the left"

of Republicans. Perhaps Reagan represented the l imi t of

what the "Right" means in the United States . In a sense, he

may be a kind of baseline, or bench mark, by which other

pol i t i ca l actors may be posit ioned. Further research on

th is subject needs to be done before the posi t ions of pres i -

dents can be compared longi tudinal ly.

170 See Budge, Robertson and Hearl, Ideology, Strategy andParty Change, 456-67; Ian Budge and Richard I . Hofferbert ."Mandates and Policy Outputs."



197

More innovative methods of measuring pol i t i ca l inten

t ions are desperately needed in th is area of research.

Counting parliamentary seats i s not sufficient . More dis-

criminating instruments are required to advance research in

th is area.

Measurement i s not the only problem encountered in

analyzing the in flu en ce o f part ies on budgets. Theories of

pol i t ical processes and ins t i tu t ions need to be conceptual

ized in such a way that systematic research of th is kind can

proceed in a cumulative manner. Unfortunately most ins t i tu -

t iona l studies do not deal with the problem of developing

measurements of structures or processes. Structural con

s t ra in t s are widely perceived, but rarely given operational

defini t ions. Of course, these things are di f f i cu l t to

operationalize, but many scholars have noted that there are

frequently sequential decisions that recur on a regular

basis . Formal theorist have developed a wide array of

analyt ical tools for solving sequential games, but fewer

deal with the empirical problem of operationalizing con

cepts.

Turning to the dependent variable, th is study has shown

the ut i l i ty of measuring trade-offs di rec t ly . I believe

that the measures used in th is study are the very best

available . The Berry-Lowery formula is a major contribution

to the f ie ld in providing a direct measure of a t rade-off .



198

Hypothesizing about the exis tence o f t rade-offs simply gets

nowhere. The potential uses for this measure have not been

exhausted here. The instrument i s equally applicable to the

o th er sid e of the budge ta ry equation: revenues.

Preliminary analyses of the sources of taxes promise to

be as in terest ing as th e a na ly si s of expenditures. Both

benefi ts and burdens are rendered more t ractable to analysis

using th is instrument. Not only does the Berry-Lowery

conceptualization provide a good foundation for developing

theories about the determinants of t rade-offs , but i t also

provides an excellent foundation for analyzing the effects

of trade-offs in a more coherent way (See Appendix D) .



APPENDIX D

Related Findings

In analyzing the models used in this study, I stumbled

upon a r at he r p ecul ia r problem. Confounding r es ults th re at

ened the leave me without any firm conclusions. Something

st range was going on, but I didn ' t know what. Solving th is

problem furnished a pleasant surpr ise.

War, Trade-offs and Unemployment

In examining the determinants of t rade-offs in the

federal budget I included, as a control variable , the ra te

of unemployment, in the b elie f th at the economic condit ion

of the country might impinge on the way tha t al locat ions are

made, especial ly with respect to cer ta in kinds of domestic

spending. Berry and Lowery include the percent of people

below the poverty l ine (as I do without resul ts ) . Seeing a

need to which a le f t - leaning pol i t ic ian might be sensi t ive ,

I tested models using unemployment as a control .

Unemployment seemed to have a s tr ong i nf luence on the

values of other parameters in many of the models I tested,

but i t s effec t varied from one p erio d to another in a con

fusing pat tern .

Reconsidering the relat ionship between unemployment and

t rade-offs , I examined the pattern di rec t ly . The t rends are

shown below.

199



200

A oo.st1C Trade-off

10

9

8 .....

0..

7I

CQ)

U '0

Ii i coc-

o 6to-

r4S- O

u-I ....

II: 5Q)

:J CD

e0

4

Q

3

Returning to the l i terature on th e sub je ct of t rade

offs, I noted tha t Ron Smith had found a posi t ive correla

t ion with defense burdens and the unemployment level . l7 1

Smith's findings, however, did not hold in cross-national

s ta t i s t ica l tes ts .172 In fact, the United States, along

with the United Kingdom, is considered an out l ie r in being

one of the few countries where th is relat ionship holds.

As the domestic trade-off increases so does unemploy-

171 See Ron P. Smith, "Military Expenditures and Capitalism, "Cambridge Journal of Economics 1 (1977): 61-76.

172 See Chan, "Defense Spending and the Economy," 407.



201

ment, or so i t seemed during some periods. Looking closer ,

I noticed that the relat ionship rever se s -during wars.

Following th is observat ion I constructed a model, with

unemployment as a dependent instead of independent variable .

Two independent variables were included in an i n i t i a l model:

the domestic t rade-off and a dummy variable for wartime

( i .e . , any year with bat t le deaths as reported by the

Department of Defense). Results o f exp lo ra to ry analyses

were very good.

Several specif icat ions to the in i t i a l model were

tes ted, i nc ludi ng l ag s of one and two years. Results are

much weaker under any specification other than a simulta-

neous model .

Following the advice o f P ro fes so r Jacek Kugler, I

part i t ioned the variance between wartime t rade-offs and

peacetime t rade-offs , coding the counter circumstance as

zero ( i .e , wartime t rade-offs equal zero i f no bat t le

deaths, peacetime t rade-offs equal zero i f bat t le deaths are

greater than zero). This procedure s epar at es t he t rade-offs

occurring during wartime from those occurring dur ing peace -

time for the purpose of ascertaining whether both wartime

and peacetime tra deoff s a re s igni f icant .

Table 17 reports the results of th is specification:



202

TABLE 17

Explaining Unemployment

Model 1

1950-1989 1951-1989

Variable OLS COReRegression Regression

Constant 5.62*** 5.68***(0.1773) (0.2173)

Peacetime Trade-off 0.0118*** 0.0098***(0.0021) (0.0071)

Wartime Trade-off -0.0078*** -0.0076***(0.0018) (0.0019)


N = 40 39

Standard errors are reported in parentheses.Standardized coeff icients from OLS are .61 for peacetime

t rade-off and -.44 for wartime t rade-offs .* Signif icant a t the .05 level** Signif icant a t the .01 level*** Signif icant a t the .001 level

Given tha t the impact of the t rade-off between defense

and domestic spending on unemployment in one year may depend

on the level of unemployment during the previous yea r I

modified the model to in clude th e previous year ' s unemploy

ment. Adding the previous year 's unemployment rate to the

equation, the resu l t s are even be t te r . The results of th is

modification to the i n i t i a l model are reported in the

following Table:



203

Table 18

Explaining Unemployment

Model 2

1950-1989 1951-1989

Variable OLS CORC


Constant 3.11*** 5.68***(O.63) (26.16)

Peacetime Trade-off 0.0073*** 0.0098***(O.0020) (O.0021)

Wartime Trade-off -0.0057** -0.0056**(O.0016) (O.0022)

Unemployment It-l) 0.4519*** 0.4351***(O.109)

Adjusted R2 = .71 .66Durbin-Watson = n/a

N = 39 38

Standard e rro rs are reoorted in oarentheses.Standardized coeff ic ients from OLS are .38 for peacetimet rade-off and - .37 for wartime t rade-offs , and .45 forunemployment It-ll •

******

Signif icant a t the .05 levelSignif icant a t the .01 levelSignif icant a t the .001 level

The predict ion i s improved, with the t rade-off var i

ables continuing to be s igni f icant and in the direction

in i t i a l ly observed. Figure 23 shows the f i t of the model's

predict ion with actual ra tes of unemployment:



204

10 10

9 9

'C8 8

'DU U

>- >-0

7 70

... ...

0 . Q.

E Eu

6 6 ·c: :) :::»

4J

c5

cQ) 5 uu U

L L

U CD

Q. 4 4 Q .

3 3

2 2

4 8Predicted Percent UneIIployed

Model. 2 Predictions of unemploymentFIGURE 23

With these resu lts in hand, I turned to th e que stion of

whether tr ad e- of fs in other countries also resulted in

unemployment. No general relationship could be found.

Recalling the patterns shown a t the end of Chapter I I ,

I suspect that there is no cross-national pattern because

most advanced industr ial countries simply do not vary in the

way they d is tr ib u te t he ir resources. In this l ight , Smith's

findings make sense. Fluctuations in defense vis-a-vis

domestic spending are pract ical ly n i l in most of the other

countries I examined.

In the cross-national data I did not find a re la t ion-



205

ship between t rade-offs and unemployment, but then I do not

currently have data on comparable "discretionary domestic

spending." Gathering comparable data on th is var iable i s

the next step in th is l ine of research.

I f Iny" suspicions are correct, i t is not the "burden"

tha t "causes" unemployment. Instead, i t i s the change in

the distr ibut ion of resources.

At present , I can only speculate on the mechanisms

through which such re la t ionships might be produced. An

examination of the multipl ier effects of each kind of

expenditure might explain this f inding. Previous research

indicates that this may be the case .173

One thing worth noting i s that the relat ionship tha t I

have found is a re la t ionship between marginal changes in the

budget and the employment of people who are probably a t the

economic margins themselves. Rates of unemployment are

disproport ionate between people of different races and

levels of educat ional a ttainment. People who are a t lower

end of the socio-economic ladder are the ones most l ikely to

be affected by marginal changes in the econoIny".

Much more work needs to be done on th is matter. The

model i t s e l f might be refined by using more narrowly defined

t rade-offs . These findings are very encouraging for the

development of research in this area. This whole of area

173 See Chan, "Impact of Defense Spending."



206

research i s great ly enhanced by the use of a measure of the

dist r ibut ion, or sp l i t among budget items.

Moreover, the implications of this f inding are very

powerful. I f American part ies are responsible for generat

ing tra de -o ffs in budget allocations, then they are also

responsible for producing the rate of unemployment. Further

research on th is matter is warranted.

Based soley on the f i t of this model, i f the t rade-off

during 1991 and 1992 i s toward defense, as I expect i t wil l

be when data on that period i s available, it should explain

the high ra te of unemployment th at e xis ts at the time th is

study i s being writ ten. While combat deaths tota led more

than 200 for the Gulf War, there was no draft in effect as

there had been during World War I I , and the Korean and

Vietnam Wars. For this reason, I would expect t rade-offs

during the Gulf War to act in the same way as a peacetime

t rade-off .



BIBLIOGRAPHY

Abizadeh, Sohrab, and John Gray. "Wagner's Law: A PooledTime-Series, Cross-Sectional Comparison." National TaxJournal 38, no. 2 (June 1985): 209-16.

Abramson, Alan J . "Budgeting for Entitlement Programs in

Periods of Expansion and Consolidation." PaperPresented at the annual meeting of the AmericanPol i t ica l Science Association, Atlanta, AugustSeptember 1989.

Alvarez, Michael R., Geoffrey Garrett , and Peter Lange."Government Partisanship, Labor Organization, andMacroeconomic Performance." American Pol i t ica l ScienceReview 85 , no. 2 (June 1991) : 539-56.

Anderson, Marion. "The Soft Spot: How to Attack thePentagon." Journal of Sociology and Social Welfare 4(1977): 597-610.

"Army Batt le Casualt ies and Nonbattle Deaths in World WarII : Final Report," Prepared by Sta t is t ica l andAccounting Branch of the Adjunct General, (ProgramReview and Analysis Division, Office of the Comptroller

of the Army, O.C.S., June 1953, mimeo).

Au C laire, P hilip Arthur. "Public Atti tudes Toward SocialWelfare Expenditures." Social Work 29, no. 2 (MarchApril 1984) : 139-44.

Baer, Doug, Edward Grabb, and William A. Johns ton. "Clas s,Cris is and Pol i t ica l Ideology in Canada: RecentTrends." Canadian Review of Sociology and Anthropology24 (1987): 1-22.

Banks, Arthur S. ed. Pol i t ica l Handbook of the World. New

York: McGraw-Hill Book Company, 1988.

Baran, Paul, and Paul Sweezy. Monopoly Capital . New York:Monthly Review, 1966.

Bardach, Eugene. The Implementation Game. Cambridge: M.I.T.

Press, 1977.

Bayes, Jane H. Ideologies and In teres t Group Pol i t ics .

Novato: Chandler & Sharp Publishers, 1982.

207



208

Beam, David R., Timothy J . Conlan, and Margaret T.Wrightson. "Solving the Riddle of Tax Reform: PartyCompetition and the Poli t ics of Ideas." Pol i t ica l

Science Quarterly 105, no. 2 (Summer 1990) : 193 -217 .

Beck, Nathan. "Parties, Administrations, and AmericanMacroeconomic Outcomes." American Pol i t ica l ScienceReview 76 (March 1982): 92-3.

Beck, Morris. Government Spending: Trends and Issues. New

York: Praeger Publishers, 1981.

Beck, Morris. "Public Sector Growth: A Real Perspective."Public Finance 34 (1979): 313 -55.

Benoit, Emile. Defense and Economic Growth in DevelopingCountries. Lexington , Mass.: D.C. Heath, 1973.

Berman, Paul. "Designing Implementation to Match PolicySituat ion: A Contingency Analysis of Programmed andAdaptive Implementation." Paper Delivered a t the 1978Annual Meeting of the American Pol i t ica l ScienceAssociation.

Bernstein, Robert A. "Ideological Deviation and Support forRee lect ion: Par ty Differences in Senate Races." TheWestern Poli t ical Quarterly 44, no. 4 (December 1991) :987-1003.

Berry, William D. "Testing Budgetary Theories with BudgetaryData : Asses sing the Risks." American Journal ofPol i t ica l Science 30 (1986): 597-627 .

Berry, Wi11iam D., and David Lowery. "An Al ternativeApproach to Understanding Budgetary Trade-offs."American Journal of Pol i t ica l Science 34 (1990): 671705.

Bloom, Howard S., and H. Douglas Price. "Voter Response toShort-run Economic Conditions: the Asymmetric Effect ofProsperi ty and Recession." American Pol i t ica l ScienceReview 69, no. 4 (December 1975): 1240-54.

Bogue, Donald J . Populations of the United States . New York:The Free Press, 1985.



209

Brady, David, Joseph Cooper, and Patr icia Hurley. "TheDecline of Party in the U.S. House of Representat ives,

1887-1968." In Congress: Structure and Policy, eds.Matthew D. McCubbins and Terry Sullivan, 235-285.Cambridge: Cambridge University Press, 1987.

Browning, Robert X. "presidents, Congress, and PolicyOutcomes: U.S. Social Welfare Expenditures, 1949-77,"American Journal of Pol i t ica l Science. Vol. 29, no. 1,197-216, 1985.

Buchanan, James M., and Gordon Tullock. The Calculus ofConsent. Ann Arbor: University of Michigan Press, 1962.

Budge, Ian, and Richard I . Hofferbert .

Outputs: U.S. Party Platforms andExpenditures." American Poli t ical

no. l (March 1990): 111-31.

"Mandates and Policy

FederalScience Review 84

Budge, Ian, David Robertson, and Derek Hearl, eds. Ideology,Strategy and Party Chancre: Spatial Analyses of Post-WarElection Programmes in Nineteen Democracies. Cambridge:Cambridge University Press, 1987.

Budget of the United States Government: Fiscal Year 1987.Washington, D.C .: United S ta te s P ri nt ing Office, 1986.

Budget of the United States Government: Fiscal Year 1991.Washington, D.C .: United S ta te s P ri nt ing Office, 1990.

Budget of the United States Government: Fiscal Year 1992.Washington, D.C .: United States Printing Office, 1991.

Burke, Edmund. Reflections on the Revolution in France.Edited by Thomas H. D. Mahoney. New York: Bobbs-MerrillCompany, Inc. , 1955.

Cameron, David. "The Expansion of the Public Economy."American Pol i t ica l Science Review 72 (1978): 1243-61.

Cappenlen, Adne, Nils Petter Gleditsch, and Olav Bjerkol t ."Guns, Butter , and Growth: The Case of Norway," Paper

Presented at the annual convention of the InternationalStudies Association, Washington, D.C., April 1990.

Castles, Francis G. "The Impact of Parties on PublicExpenditures." In The Impact of Part ies , edi ted byFrancis G. Castles. Beverly Hil ls : Sage, 1982.



210

Castles, Francis G., and R. McKinlay. "Does Poli t ics Matter:An Analysis of the Public Welfare Commitment in

Advanced Democratic States ." European Journal ofPoli t ica l Research 7 (1979): 169-86.

Castles, Francis G., and Peter Mair. "Left-Right Pol i t ica l

Scales: Some 'Expert ' Judgements." European Journal ofPoli t ica l Research 12 (1984): 73-88.

Chan, Steve. "The Impact of Defense Spending on EconomicPerformance: A Survey of Evidence and Problems," Orbis(Summer 1985): 403-34.

Chan, Steve, "The Taiwanese Enigma." American Poli t ica l

Science Review 82, no. 3 (September 1988): 913-20.

Chase, Gordon. "Implementing a Human Service Program: How

Hard Will I t Be?" Public Policy. Vol. 27,no.4 (Fall1979) :387-435.

Child, John. "Organizational Structure, Environment andPerformance: The Role of Strategy Choice. " Soc io logy .Vol. 6 (1972) :2-22

Chubb, John E. "Federalism and the Bias for Centralization,"In The New Direction in American Pol i t ics , edited byJohn E. Chubb and Paul E. Peterson, 273-306.Washington, D. C.: The Brookings Ins t i tu te , 1985.

Clayton, James L. "A Comparison of Defense and WelfareSpending in the United States and the United Kingdom,1946-1976." Journal of Sociology and Social Welfare 4(1978): 401-18.

Congressional Quarterly's Guide to Congress (Washington, D.

C.: Congressional Quarterly Inc . , 1989-1990).

Coughlin, Richard M., and Phil ip K. Armour. "MethodologicalIssues in the Comparative Study of Social Securi ty:Quali ta t ive vs. Quanti tat ive Analysis and theAppropriate Use of Data." Internat ional Review ofModern Sociology 10 (July-December 1980): 25-48.

Cox, Robert. liDo Part ies Matter? Lessons from the Low

Countries," in eds. Joe Adams, Marina Arbetrnan, and M.

Donald Hancock, Pol i t ica l Part ies and the Ideology ofthe Welfare State, (Forthcoming),



211

Cull is , John G., and Philip R. Jones. "The Economic Theoryof Bureaucracy, X-Inefficiency and Wagner's Law: ANote." Pub lic Finance 39 (1984): 191-201.

Cusak, Thomas R. "On the Domestic Political-Economic Sourcesof American Military Spending." Paper Presented at theannual meeting of the American Poli t ical ScienceAssociation, Atlanta, August 1989.

Cutright , Ph il li ps . " Po li ti ca l Structure, EconomicDevelopment , and National Social Securi ty Programs."American Journal of Sociology 70 (March 1965): 537-50.

Daalder, Hans. "The Dutch Party System: From Segmentation toPolarization- And Then?" In Party Systems in Denmark,

Austria , Switzerland, the Netherlands and Belgium,edited by Hans Daalder. London: Frances Pinter , 1987.

Dabelko, David, and James D. McCormick. "Opportunity Costsof Defense: Some Cross-National Evidence." Journal ofPeace Research 2 (1977): 146-54.

Davidson, Roger H., and Walter J . Oleszek. Congres s and I ts

Members. 3d ed. Washington, D.C.: CongressionalQuar te rly Press , 1990.

Davis, Otto A., M . A ~ H . Dempster , and Aaron Wildavsky. "Toward a Predict ive Theory of Government Expenditures:U.S. Domestic Appropriations." Brit ish Journal o fPoli t ical Science 4 (October 1974): 419-52 .

Deger, Saadet. "Economic Development and DefenseExpenditures." Economic Development and Social Change35 (October 1986) : 178-96.

Deu ts ch , Kar l W. The Nerves of Government. New York: TheFree Press, 1962.

Devine, Joel A., and William Canak. "Redistribution in aBifurcated Welfare State: Quinti le Shares and the U.S.Case." Social Problems 33, no. 5 (June 1986): 391-405.

Devlin, Sir Patrick. "The Enforcement of Morals."Proceedings of the Brit ish Academy London: OxfordUniversity Press, 1959.



212

Domke, William K., Richard C. Eichenberg, and Catherine M.Kelleher. "The I l lus ion of Choice: Defense and Welfarein Advanced Indus t r ia l Democracies, 1948-1978." TheAmerican Poli t ica l Science Review 77 (March 1 9 8 3 ~ 1 9 35.

Doti, James L., and Esmael Adibi. Econometric Analysis withMicro TSP Student Software: An Applications Approach.Englewood Clif fs , NJ: Prentice-Hall Inc . , 1988.

Downs, Anthony. An Economic Theory of Democracy. New York:Harper and Row, 1957.

Downs, Anthony. "Why the Government Budget Is Too Small in aDemocracy." World Poli t ics 12 (1960): 541-63.

Durant, Robert F., e t . ale "From Complacence to Compliance:Toward a Theory of Intergovernmental Relations."Administration and Society. Vol. 17, no. 4 (February1986) :433-459.

Duverger, Maurice . Pol i t ica l Part ies : Thei r Organiza tion andActivity in the Modern State . Translated by Barbara andRobert North. Paris : Armond Colin, 1951. Reprint.London: Lowe & Brydone, 1967.

Dye, Thomas. "Party and Policy in th e S ta te s." Journal ofPol i t ics 46, no. 4 (November 1984): 1097-1116.

Eichenberg, Richa rd C., and Richard Stol l , "Straightjacket:Support for Defense Spending in Four WesternDemocracies ," Paper Presented a t the Annual Meeting ofthe American Poli t ica l Science Association, Atlanta,Georgia, August 31- September 3, 1989.

Erikson, Robert S., Gerald C. Wright, J r . , and John McIver."Pol i t ica l Part ies , Public Opinion, and State Policy in

the United States ." American Pol i t ica l Science Review83, no. 3 (September 1989) : 729-50 .

Facts on Fi le New York: Facts on File , Inc. , various issues .

Fair , Ray C. "The Effect of Economic Events on Votes fo rPresident ." The Review of Economics and Sta t is t ics 60(May 1978) : 159 -73.

Fenno, Richard F., J r . The Power of the Purse:Appropriations, Poli t ics and Congress. Boston : Lit t le ,Brown and Co., 1966.



213

Fiorina, Morris P. Congress: Keystone of the WashingtonEstablishment. New Haven: Yale Univer si ty P re ss , 1977.

Fiorina, Morris P. Retrospective Voting in American NationalElections. New Haven: Yale Univer si ty P re ss , 1981.

Fisher, Louis. Presidential Spending Power. Princeton:Princeton Univer si ty P re ss , 1975.

Fishkin, James S. Just ice, Equal Opportunity, and theFamily. New Haven: Yale University Press, 1983.

Flora, Peter and Arnold J . Heidenheimer. "The HistoricalCore and Changing Boundaries of the Welfare State ." InThe Development of Welfare States in Europe and North

America, edited by Pete r F lo ra and Arnold J .Heidenheimer, 17-34. New Brunswick, N.J. : TransactionBooks, 1981.

Freidland, Roger, and Jimy Sanders. "The Public Economy andEconomic Growth in Western Market Economies." AmericanSociological Review 50 (1985): 421 -37.

Frey, Bruno S. "Politico-Economic Models and Cycles."Journal of Public Economics 9 (1978): 203-20.

Friedman, Milton. There's No Such Thing as a Free Lunch.LaSalle, I l l inois : Open Court, 1974.

Friedman, Samuel R. "An Alternative to Szymanski." AmericanJournal of Sociology 79 (1974): 421-37.

Furniss, Norman. Futures of the Welfare State. Bloomington,I l l inois : Indiana Univer si ty P re ss , 1986.

Gould, Frank. "The Development of Publ ic Expendi tures in

Western, Industr ial ized Countries: A ComparativeAnalysis." Public Finance 38 (1983): 38-69.

Grier, Kevin B. "On the Existence of a Poli t ica l MonetaryCycle." American Journal of Pol i t ica l Science 33, no. 2(May 1989): 376-89.

Griff in, Larry J . , Joel Devine and Michael Wallace,"Monopoly Capital , Organized Labor, and MilitaryExpenditures in the United States, 1946-1976," AmericanJournal of Sociology 88 (1982): 113-53.



214

Griff in, Larry J . , Pamela Barnhouse Walters, Phil ipO'Connell, and Edward Moor. "Methodological Innovations

in the Analysis of Welfare-State Development: PoolingCross-Sections and Time Series." In Fu ture s of theWelfare State, edited by Norman Furniss, 101-38.Bloomington, I l l inois : Indiana University Press, 1986.

Griff in, Larry J . , and Kevin T. Leicht. "Poli t icizingWelfare Expenditures in the United States ." In Futuresof the Welfare State. edited by Nor.man Furniss, 320-56.Bloomington, I l l ino i s : Indiana Unive rs ity P re ss , 1986.

Griff in, Larry J , Joel Devine, and Michael Wallace. "OneMore Time: Mili tar iz ing the U.S. Budget: Rep ly toJencks." American Sociological Review 91 (1985): 384

90 .

Halperin, Morton H., Jacob A. Stockfish, and MurrayWeidenbaum. The Pol i t ica l Economy of the Military -Industr ial Complex, edited by Warren F. Ilchrnan and JoeS. Bain. Berkeley: University of California Press,1973.

Harrison, Anthony J . The Control of Public Expenditure 19791989. Reshaping the Pub li c Sec to r, Volume 5. SeriesEditors John Gretton and Anthony Harrison. Oxford:Transaction Books, 1989.

Hart , H.L.A. "Social Solidari ty and the Enforcement of

Morality." University of Chicago Law Review 35 (1967):1-13.

Hartley, Keith, and Pat McClean. "U.K. Defence Expenditure."Public Finance 36 (1981): 171-91.

Hartz, Louis. The Liberal Tradition in America. New York:Harcourt, Brace, 1955.

Hayek, Friedrick A. Law, Legislation and Liberty. Chicago:Unive rs ity o f Chicago Press, 1969.

Heclo, Hugh. Modern Social Poli t ics in Britain and Sweden:

From Relief to Income Maintenance. New Haven: YaleUniversity Press, 1974.

Henig, Stanley, ed. Pol i t ica l P arties in the EuropeanCommunity. London: George Allen & Unwin Limited, 1978.



215

Hewitt , Christopher. "The Effect of Pol i t ica l Democracy andSocial Democracy on Equality in Indust r ia l Societies: A

Cross-National Comparison." American SociologicalReview 42 (June 1977): 450-64.

Hibbs, Douglas A. "Pol i t ical Parties and MacroeconomicPolicy." American Poli t ical Science Review 71 (1977):1467-87.

Hibbs, Douglas A. Economic Interest and the Poli t ics ofMacroeconomic Policy. Centre for International Studies:MIT, 1975.

Hicks, Alexander, and Duane Swank. "On the Pol i t ica l Economyof Welfare Expansion." Comparative Pol i t ica l Studies 17

(1984): 81-119.

H icks , A lexander , and Alex Mintz. "Theoretical Insights andOversights in 'Methodological Problems'," AmericanSociological Review 91 (1985): 379-83.

Hobday, Charles. Communist and Marxist Part ies of the World.Santa Barba ra , California: ABC-Clio, 1986.

Hobson, J.A. Imperialism: A Study. London: Allen & Unwin,1938.

Hofstadter, Richard . Liberal ism in America. New York:Vintage Books, 1958.

Jackman, Robert W. "Social is t Parties and Income Inequali tyin Western Indus t r ia l Societies." Journal of Poli t ics

42 (February 1980) : 135 -49.

Jacobs, Francis, Richard Corbett, and Michael Shackleton.The European Parliament. Boulder: westview Press, 1990.

Jacobson, Gary C. The Poli t ics of Congressional Elections.2d ed. Glenview, I l l inois : Scott Foresman and Company,1987.

Jenks, Christopher. "Methodological Problems in Studying

'Mil i tary Keynesianism' ." American Journal of Sociology91 (1985): 373- 79.

Kaldor, N. "Technical Change in the Defence Industry." InTechnical Innovation and Brit ish Economic Performance,edited by K. Pavit , 100-24. London: Macmillan, 1980.



216

Keech, Wil li am R., and Kyougsan Pak. "Electoral Cycles andBudgetary Growth in Veteran's Benefi ts ." American

Journal of Pol i t ica l Science 33, no.4 (November 1989):901-11.

Keefe, William J . Part ies , Poli t ics and Public Policy inAmerica. Fif th Edition. Washington, D.C.:CongressionalQuar te rl y I nc , 1988.

Keeney, Bradford P. Aesthetics of Change. New York: TheGuil fo rd P re ss , 1983.

Keesing 's Contemporary Archives London: Keesing's Limited,various issues.

Keman, Hans, and Die tmar Braun. "The Limits of Poli t ica lControl: A Cross-National Comparison of Economic PolicyResponses in Eighteen Capital is t Democracies." EuropeanJou rn al of Pol i t ica l Research 12 (1984): 101-8.

Kendall, Willmoore. "The 'Open Society ' and I t s Fal lac ies ."American Pol i t ica l Science Review 54 (December, 1960):972-79.

Key, V. 0 . , J r . Pol i t ics Part ies and Pressure Groups. FourthEdition, New York: Thomas Y. Crowell Co., 1958.

Kiewiet, D. Roderick. Macro-Economics &Micro-Poli t ics: TheElectoral Effects of Economic Issues. Chicago: TheUniversity of Chicago Press, 1983.

Kiewiet, D. Roderick. Macroeconomics &Micropoli t ics.Chicago: University of Chicago Press, 1983.

Kingdon, John W. "Models of Legislative Voting." Journal ofPoli t ics 39, no. 3 (August 1977): 563-95.

Kluegel, James R. "Macro-economic Problems, Beliefs aboutthe Poor and Atti tudes Toward Welfare Spending." SocialProblems 34, no. 1 (February 1987): 82-99.

Kmenta, Jan. Elements of Econometrics. 2d ed. New York:

Macmillan Publishing Company, 1985.

Kohl, Jurgen. "Trends and Problems in Postwar PublicExpenditure Development in Western Europe and NorthAmerica." In The Development of Welfare S tates inEurope and North America, edi ted by Pet er F lo ra andArnold J . Heidenheimer. New Brunswick, N.J . :Transaction Books, 1981.



217

Korford, Kenneth. "Dimensions in Congressional Voting."American Poli t ical Science Review 83 (1989): 949-62.

Korpi, Walter. "Social Policy and Dis tr ibu tiona l Confl ic t inthe Capi ta l is t Democracies: A Preliminary Framework."West European Pol i t ics 4 (1980): 296-316.

Koven, Steven G. Ideological Budgeting: The I nf lu ence o fPol i t ica l Philosophy on Public Policy._ New York:Praeger, 1988.

Koven, Steven G. Ideological Budgeting: The Influence ofPoli t ica l Philosophy on Public POlicy._ New York:Praeger, 1988.

Kraus, Franz. "The Histor ical Development of IncomeInequali ty in Western Europe and the United States ." InThe Development of Welfare States in Europe and NorthAmerica, edited by Peter Flora and Arnold J .

Heidenheimer, 187-223. New Brunswick, N.J. : TransactionBooks, 1981.

Krehb ie l, Kei th . "A Technique for Estimating Congressmen'sIdeal Points." Journal of Pol i t ics 48, no. 1 (February1986): 97-115.

Kurdel, Rober t T., and Theodore R. Marmor. "The Developmentof Welfare States in North America." In The Developmentof Welfare States in Europe and North America, edited

by Peter Flora and Arnold J . Heidenheimer, 81-121. NewBrunswick, N.J. : Transaction Books, 1981.

Ladd, Everett Carll , Jr . and Charles D. Hadley. Poli t ica l

Part ies and Pol i t ica l Issues: Patterns andDifferentiat ion Since the New Deal. Edited by RandallB. Ripley. American Poli t ics Series, Vol. 1. BeverlyHills: Sage Publications, 1973 ..

Lane, Robert E. Poli t ical Ideology: Why the American CommonMan Believes What He Does. New York: The Free Press,1962.

Leiserson, Avery. "Coordination of Federal Budge ting andAppropriations Procedures under the LegislativeReorganization Act of 1946." National Tax Journal 1(1948): 118-25.

Leiserson, Avery. Parties and Poli t ics: An Ins t i tu t ional andBehavioral Approach. New York: Alfred A. Knopf Inc . ,

1958.



218

Light, Paul C. The ~ r e s i d e n t ' s Agenda: Domestic PolicyChoices from Kennedy to C arter. Baltimore: The Johns

Hopkins University Press, 1983.

Lindblom, Charles. "The Science of Muddling Through." PublicAdministration Review 19 (1959): 79-88.

Lipset, SeYmour M. "The Economy, Elections, and PublicOpinion." Tocgueville Review 5 (1983): 431-70.

Lowi, Theodore J . The End o f L iberali sm : The Second Republicof the United States . 2d ed. New York: W.W. Norton &Company, 1979.

Lowi, Theodore J . "Four Sys tems of Polic y, Pol i t ics , andChoice." Publ ic Adminis tra tion Review. 72 (July/August1970) :298-310.

MacDonald, Elaine, Ola Listhaug, and George Rabinowitz."Issues and Party Support in Multiparty Systems."American Pol i t ica l Science Review 85, no. 4 ( D e c ~ e r1991): 1107-31.

MacIntyre, Alsdair. Whose Justice? Which Rationali tv? NotreDame: University of Notre Dame Press, 1988.

MacIntyre, Ala sd air. A fte r V irtu e. 2d ed. Notre Dame:University of Notre Dame Press, 1984.

MacIntyre, Alasdair . Against the Self-Images of the Acre.Notre Dame: Univer si ty o f Notre Dame Press, 1978.

Mackie, Thomas T., and Richard Rose. The Internat ionalAlmanac of Electoral History. 2d ed. New York: Facts onFile , Inc ., 1982.

Madsen, Henrik J . "Electoral Outcomes and Macro-EconomicPolic ies: The Scandinavian Cases. II In Models ofPol i t ica l Economy, edited by P. Whitley, 15-46. Londonand Beverly Hil ls : Sage Publications, 1980.

Magat, Wesley A., Alan J . Krupnick, and Winston Harrington.

Rules in the Making: A Sta t i s t i ca l Analysis ofRegulatory Agency Behavior. Washington, D.C.: Resourcesfor the Future, 1986.

Mannheim, Karl. Ideology and Utopia. Translated by LouisWirth and Edward Shils . New York: Harcourt BraceJovanovich, Publishers, 1985.



219

Marks, Ga:ry W. "State-Economic Linkages in AdvancedIndustr ial ized Societ ies ." In Pol i t ica l Economy inWestern Democracies, edited by Norman J . Vig and StevenE. Schier . New York: Holmes &Meier, 1985.

Mayer, Lawrence, and John H. Burnett . Poli t ics in Indus t r ia lSocieties: A Comparative Perspective. New York: JohnWiley & Sons, 1977.

Mayhew, David R. Placing Part ies in American Pol i t ics . New

Haven: Yale Univer si ty P re ss , 1986.

Mayhew, David R. Divided We Govern: Partv Control, Law

Making and Investigations 1946-1990. New Haven: YaleUniversity Press, 1991.

Mayhew, David R. The Electoral Connection. New Haven: YaleUniver si ty P re ss , 1974.

Mazmanian, Daniel A ., and Paul A. Sabat ier . Implementationand Public Policy. Glenview, I l l inoi s : Scott , Foresman,and Company, 1983.

McCubbins, Mathew D., and Ter:ry Sull ivan. Congress:Structure and Policy. Cambridge: Cambridge Universi tyPress, 1987.

McDowall, David, Richard McClea:ry, Errol E. Meidinger, andRichard A. Hay, Jr . Interrupted Time Series Analysis.Quanti tat ive Applications in the Social SciencesSeries. Beverly Hills : Sage Publications, 1980.

McDowall, David, Richard McClea:ry, Errol E. Meidinger, andRichard A. Hay, J r . Interrupted Time Series Analysis.Quanti tat ive Applications in the Social SciencesSeries. Beverly Hills : Sage Publications, 1980.

McHale, Vincent E. McHale, ed. Pol i t ica l Part ies of Europe.Westport, Conn.: Greenwood Press, 1983.

McKay, David. Domestic Policy and Ideology: Presidents andthe American State 1964-1987. Cambridge: Cambridge

University Press, 1989.

McKay, David. "Pres iden ti al S tr at egy and the Veto Power: AReappraisal," Pol i t ica l Science Quarterly 104, no. 3(Fall 1989): 447-61.



220

Meid, Pat and James M. Yingl ing. Operat ions in West Korea:U.S. Marine Operations in Korea 1950-1953, Vol. V,

(Washington, D.C.: Historical Division Headquarters,U.S. Marine Corps, 1972.

Melman, Seymour. Our Depleted Society. New York: DellPublishing Company, 1965.

Melman, Seymour. The Permanent War Economy; AmericanCapitalism in Decline. New York: Simon and Schuster ,1974.

Michels, Robert. Pol i t ica l Part ies , t ranslated by Eden andCedar Paul. Glencoe, I l l inois : The Free Press, 1915,reprinted 1949.

Miller, Leonard S., and Marleen Clark. liThe Welfare Effortof The United States: Know Thyself. II Journal ofSociology and Social Welfare 4 (1977): 382-400 .

Minarik, Joseph J . Making America's Budget policy. New York:M.E. Sharpe Inc. , 1990.

Mintz, Alex. "Guns Versus Butter: A Disaggregated Analysis. II

American Pol i t ica l Science Review 83 (1989): 1285-93.

Mintz, Alex, and Alexander Hicks. "Mili tary Keynesianism inthe United States, 1949-1976: Disaggregating MilitaryExpenditures and the i r Determination." American Journalof Sociology 90 (1984): 411-17.

Montero, Darrel. "The Guns and Bette r Is su e: Trends inAmerican Public Opinion. II Journal of Sociology andSocial Welfare 5 (1978): 135 -49.

Moon, Bruce E., and William J . Dixon. "Poli t ics , the State,and Basic Human Needs: A Cross-Nat iona l S tudy ."American Jou rn al of Poli t ica l Science 29 , no. 4(November 1985): 661-94.

Morgan, Michael John. liThe Modelling of GovernmentalCoalit ion Formation: A Policy Based Approach with

Interval Measurement.II Ph.D. diss . , Universi ty of

Michigan, 1976.

Nathan, Richard P., Fred C. Dooli t t le , and Associates.Reagan and the States . Princeton, NJ: PrincetonUniversity Press, 1987.



221

Nie, Norman H., Sidney Verba, and John R. Petrocik. TheChanging American Voter. Cambridge: Harvard University

Press, 1979.

Nigro, Lloyd and Kenneth Meir. "Representative Bureaucracyand Policy Prefe rence: A Study of the Atti tudes of Federa l Executives." Public Administration Review 36(July/August 1976): 456-58.

Niskanen, William. Bureaucracy and RepresentativeGovernment. Chicago: Aldine and Atherton, 1971.

Nordhaus, William D. "The Pol i t ica l Business Cycle." Reviewof Economic Studies 42 (April 1975): 169 -90.

Nozick, Robert. Anarchy, State and Utopia. New York: BasicBooks, 1974.

OECD Social Expenditures 1960-1990: Problems of Growth andControl. Paris: Organization for Economic Cooperationand Development, 1985.

Olson, Mancur. The Rise and Decline o f Natio ns. New Haven:Yale University Press, 1982.

Organski, A.F.K., Jacek Kugler, J . Timothy Johnson, andYoussef Cohen. Births, Deaths, and Taxes . Chicago :universi ty of Chicago Press, 1984.

Orloff, Ann Shola, and Theda Skocpol. "Why not equalProtection? Explaining the Poli t ics of Public SocialSpending in Britain, 1900-1911, and the United States ,1890s-1920." American Sociological Review 49 (December1984): 726-50.

Ornstein, Norman J . , Thomas E. Mann, and Michael J . Malbin.Vital Sta t is t ics on Congress, 1987-1988. Washington,D.C.: Congressional Quar te rly Inc. ,1987.

Ostrom, Charles W. J r . , and Robin F. Marra. "U.S. DefenseSpending and the Soviet Estimate." American Pol i t ica l

Science Review 80, no. 3 (September 1986): 820-41.

Palmer, G lenn. "Alliance Poli t ics and Issue Areas:Determinants of Defense Spending." American Jo urn al ofPol i t ica l Science 34 (1990): 190-211 .

Pa lmer, G lenn. "Alliance Poli t ics and Issue Areas:Determinants of Defense Spending." American Journal ofPol i t ica l Science. 34 (February 1990) : 190-211 .



222

Pampel, Fred C ., and John B. Williamson. "Welfare Spendingin Advanced Industr ial Democracies 1950-1980." American

Journal of Sociology 93 (1988): 1424-56.

Patterson, Samuel C., and Gregory A. Caldeira, "Party Votingin the United States Congress." Brit ish Journal ofPoli t ica l Science 18 (January 1988):111-31.

Patterson, Samuel C. and Gregory A. Caldeira. "Party Votingin the United States Congress." Brit ish Journal ofPoli t ica l Science 18 ( January 1988) : 111-31.

Peacock, Alan, and Jack Wiseman. The Growth of PublicExpenditure in the United States . Princeton : P rincetonUniversi ty Press, 1961.

Peterson, Paul E., and John E. Chubb, eds. The New Directionin American Pol i t ics . Washington, D.C.: The BrookingsIns t i tu t ion , 1985.

Peterson, Paul E. "The Rise and Fall of Special In teres tPoli t ics ," Pol i t ica l Science Ouarterly 105 (Winter1990-91): 539-56.

Phil l ips, Steve. "Budget Negotiations and the Sequester,"World of Poli t ics (October 1990): 6-7.

Poole, Keith, and R. Steven Daniels. "Ideology, Party andVoting in the U.S. Congress, 1959-1980." American

Poli t ica l Science Review 79 (March 1985): 373-99.

Prothro, James W., and Charles M. Grigg. "FundamentalPrinciples of Democracy: Bases of Agreement andDisagreement." Journal of Pol i t ics . 22 no.2 (1960) :276294.

Ram, Rati . "Government Size and Economic Growth: A NewFramework and Some Evidence form Cross-Section andTime-Series Data ." American Economic Review 76 (1986):191-203.

Rapoport, Ronald B., Alan I . Abramowitz, and John McGlennon,

eds. The Life of the Pa rt ie s: Ac ti v is ts i n P re si dent ia lPoli t ics . Lexington, KY: The University Press ofKentucky, 1986.

Rawls, John. A Theory of Just ice. Cambridge: HarvardUniversity Press, 1971.



223

Rice, Tom W. "The Determinants of Western EuropeanGovernment Growth." Comparative Pol i t ica l Studies 19 ,

no 2 (July 1986): 233-57.

Richardson, Bradley. "European Par ty Loyal ti es Revisi ted."American Pol i t ica l Science Review 85, no. 3(September): 751-75.

Riker, William H. and Peter C. Ordeshook, An Introduction toPositive Pol i t ica l Theory. Englewood Clif fs : PrenticeHall, Inc, 1973.

Rokeach, Milton. The Open and Closed Mind. New York: BasicBooks, 1960.

Rose, Richard. Do Parties Make a Difference? 2d ed. , rev.and enl . Chatham, New Jersey: Chatham House Publishers,1984.

Rose, Richard. "How Exceptional is the American Pol i t ica lEconomy?" Pol i t ica l Science Quarterly 104 (Spring1989): 91-115.

Russet, Bruce. "Doves, Hawks and u.S. Public Opinion."Pol i t ica l Science Quarterly 105, no 4 (Winter 19901991): 515-38.

Russett , Bruce. "Defense Expenditures and National Wellbeing." American Pol i t ica l Science Review 76 (September1982): 767-77.

Russett , Bruce. What Price Vigilance? The Burdens ofNational Defense. New Haven: Yale Unive rs ity P re ss ,1970.

Sahni, Balb ir S ., and Balvir Singh "On the Causal DirectionsBetween Income and Government Expenditure in Canada."Public Finance 39 (1984): 359-93.

Sandel, Michael J . "Morality and the Liberal Ideal ." The NewRepublic 17 May 1984, 15-7.

Sani, Giacomo. "The Electorate: An Ambiguous Verdict , "1987.in I ta ly a t the P olls . 1983: A Study o f Nati onalElections. Edited by Howard Penniman. Duke UniversityPress: Durham, North Carolina .



224

Sani, Giacomo, and Giovanni Sartori . "Polarization,Fragmentation and Competition in Western Democracies."

Western European Party Systems. Edited by Hans Daalderand Peter Mair. Beve rly Hills : Sage, 1983.

Sartori , Giovanni. "Concept Misformation in ComparativePoli t ics ." The American Poli t ical Science Review 64,no. 4 (December 1970): 1033-53.

Sartori , Giovanni. Parties and Party Systems: A Frameworkfor Analysis. Cambridge: Cambridge University Press,1976.

Scarborough, Elinor. Pol i t ica l Ideology and Voting: An

Exploratory Study. Oxford: Clarendon Press, 1984.

Schneider, Jerrold E. Ideological Coali t ions in Congress.Westport, Conn.: Greenwood Press, 1979.

Schneider, William. "Public Opinion." In The Making ofAmerica's Sov ie t Pol icy, edited by Joseph S. Nye, 16-7.New Haven: Yale University Press, 1984.

Seldes, George. The Great Thoughts. With a Foreword by HenrySteele Commager. New York: Ballantine Books, 1985.

Shaffer , William R. Party and Ideology in the United StatesCongress. Washington, D.C.: University Press ofAmerica, 1980.

Shapiro, Robert Y., and John T. Young. "Public Opinion andthe Welfare State: The United States in ComparativePerspective." Poli t ical Science Quarterly 104, no. 1(Spring 1989): 59-89.

Sharkanski, I ra . Spending in the American States . Chicago:Rand McNally & Company, 1968.

Shepsle, Kenneth. "Representation and Governance: The GreatLegislat ive Trade-off." Pol i t ica l Science Quarterly103, no. 3 (Fall 1988): 461-84.

Shibata, Hirofumi, and Yoko Kimura. "Government Debt andGrowth in Public Spending." Public Finance 42 (1987):291-96.

Sloan, John, and Kent L. Tedin. "The Consequences of RegimeType for Pub li c Pol icy Outputs." Comparative Pol t ica lStudies 20 , no. 1 (April 1987): 98-124.



225

Smith, Steven, and Christopher J . Deering. Committees inCongress. Washington, D.C.: Congressional Quarterly

Inc. , 1984.

Smith, R. P. "Mili tary Expenditures and Capitalism."Cambridge Journal of Economics 1 (1977): 61 -76.

Souraf, Frank J . Party Poli t ics in America. Fif th Edition.Boston: Lit t le , Brown and Co., 1984.

Sowell, Thomas. A Conflict of Visions. New York: WilliamMorrow and Company, Inc, 1987.

Stata Reference Manual. Fif th Edition. Version 3.0. SantaMonica, California: Computing Resource Center, 1992.

Sta t i s t i ca l Abstract of the United States . United StatesPrintng Of fi ce , (various years ) .

Stevenson, Paul. "A Defense of Baran and Sweezy." AmericanJournal of Sociology 79 (1974): 1456-8.

Surmner, Robert. , and Alan Heston. "A New Set ofInternational Comparison of Real Product and Price for130 Countries, 1950-1985." Income and Wealth 34 (1988):1-25.

Swank, Duane H. "The Pol i t ica l Economy of GovernmentDomestic Expenditures in the Affluent Democracies,

1960-80." American Journal of Pol i t ica l Science 32(1988): 1121-50 .

Szymanski, Albert . "Military Spending and EconomicStagnation." American Journal of Sociology 79 (1973):1-15.

Szymanski, Albert . "A Reply to Friedman, Stevenson, andZeit l in ." American Journal of Sociology 79 (1974):1462-77.

Thurow, L. "How to Wreck the Economy," New York Review ofBooks. May 14,1981.

Tocquevil le , Alexis, de. Democracy in America. edited byPhil l ips Bradley. Vol. I , p t . 2, New York: Knopf, 1945.

Truman, David B. The Governmental Process. New York: AlfredA. Knopf, Inc. , 1951.



226

Tucker, Rober t C. The Marx-Engels Reader. 2nd ed. New York:W.W. Norton & Company, 1972.

Tufte, Edward R. Poli t ical Control of the Economy.Princeton : P rinceton University Press, 1978.

U.s. Arms Control and Disarmament Agency. (Various Years)World Military Expenditures and Arms Transfers.Washington, D.C.: U.S. Printing Office.

Uslaner, Eric M. "Partisanship and Coali t ion Formation inCongress." Pol i t ica l Methodology 2, no. 4 (1975): 381414.

Uusitalo, Hannu. "Comparative Research on the Determinants

of the Welfare State: the State of the A rt." EuropeanJournal of Pol i t ica l Research 12 ( 1984 ): 403-22.

Vig, Norman J . , and Steven E. Sch ier, ed s. Pol i t ica l Economvin Western Democracies. New York: Holmes &MeierPublishers Inc. , 1985.

Voegelin, Eric . Science, Poli t ics and Gnosticism. Chicago:Regnery, 1968.

Wander, W. Thomas, F. Ted Hebert, and Gary W. Copeland, eds.Congressional Budgeting: Pol i t ics , Process and Power.Baltimore, MD: Johns Hopkins University Press, 1984.

Weber, Max. "Poli t ics as a Vocation." In From Max Weber,t rans la ted by H.H. Gerth and C Wright Mills , 77-128.New York: Oxford University Press, 1958.

Whiteley, Paul. Pol i t ica l Control of the Macroeconomy: ThePol i t ica l Economy of Public Policy Making. London andBeverly Hills: Sage Publications, 1986.

Wildavsky , Aaron . The Poli t ics of the Budgetary Process.Boston: Lit t le Brown, 1964.

wildavsky, Aaron. "President Reagan as a Pol i t ica lStra tegis t . " Paper presented a t the Thomas P. O'Neal,

J r . Symposium on Elections in America, Boston, October1985.

Wilensky, Harold L. The Welfare State and Equalitv:Structural and Ideological Roots of PublicExpenditures. Berkeley: Univer si ty o f California Press,1975.



227

Wilensky, Harold L. "Leftism, Catholicism, and DemocraticCorporatism: The Role of Poli t ica l Parties in Recent

Welfare State Development." In The Development ofWelfare States in Europe and North America, edi ted byPeter Flora and Arnold J . Heidenheimer, 345-82. New

Brunswick, N.J. : Transaction Books, 1981.

Documents

Parties, Ideologies, and Budgets: A Study of Budgetary Trade-offs