Economic-Base Theory Chapter 3 - USP · Economic-Base Theory Chapter 3 WASchaffer 1 Draft 5/11/2010 REGIONAL MODELS OF INCOME DETERMINATION: SIMPLE ECONOMIC-BASE THEORY Economic-base

Economic-Base Theory Chapter 3

WASchaffer Draft 5/11/2010

CHAPTER 3

REGIONAL MODELS OF INCOME DETERMINATION: SIMPLEECONOMIC-BASE THEORY ................................................................... 1

ECONOMIC-BASE CONCEPTS ......................................................................... 1Antecedents .............................................................................................. 1Modern origins......................................................................................... 3

THE STRUCTURE OF MACROECONOMIC MODELS ........................................... 3THE "STRAWMAN" EXPORT-BASE MODEL...................................................... 5THE TYPICAL ECONOMIC-BASE MODEL ......................................................... 6TECHNIQUES FOR CALCULATING MULTIPLIER VALUES.................................. 8

Comparison of planner's relationship and the economist's model.......... 8The survey method ................................................................................... 8The ad hoc assumption approach ............................................................ 8Location quotients.................................................................................... 9Minimum requirements .......................................................................... 11"Differential" multipliers: a multiple-regression analysis ................... 11

CRITIQUE: ADVANTAGES, DISADVANTAGES, PRAISE, CRITICISM ................. 12

APPENDIX A REVIEW OF ECONOMIC-BASE LITERATURE ..... 13

APPENDIX B AN ECONOMIC-BASE MODEL OF ATLANTA....... 33

NOTE A. TECHNIQUES FOR DATA ANALYSIS .............................. 37

INTRODUCTION ........................................................................................... 37LOCATION QUOTIENTS ................................................................................ 37SHIFT-SHARE ANALYSIS .............................................................................. 38THOUGHTS ON WRITING AN AREA PROFILE ................................................. 39ELEMENTS TO INCLUDE IN A LOCATION QUOTIENT ANALYSIS ..................... 40


WASchaffer 1 Draft 5/11/2010

REGIONAL MODELS OF INCOMEDETERMINATION: SIMPLE ECONOMIC-

BASE THEORY

Economic-base concepts

Economic-base concepts originated withthe need to predict the effects of neweconomic activity on cities and regions.Say a new plant is located in our city. Itdirectly employs a certain number ofpeople. In a market economy theseemployees depend on others to providefood, housing, clothing, education,protection and other requirements of thegood life. The question which cityplanners and economists need to answer,then, is "what are the indirect effects ofthis new activity on employment andincome in the community?" With theseestimates in hand, we can work towardplanning the social infrastructure neededto support all of these people.

Economic-base models focus on thedemand side of the economy. Theyignore the supply side, or the productivenature of investment, and are thus short-run in approach. In their modern form,they are in the tradition of Keynesianmacroeconomics. In an introductoryeconomics course, we might start with asimple model of a closed economy,usually with some unemployment. Inregional economics we deal with an openeconomy with a highly elastic supply oflabor.

It is appropriate to start this chapterfirst with a look at the place of economic-base theory in the history of economicthought and proceed to a review thesimple Keynesian model and theelementary economic-base models. We

will then look at methods of estimatingthe values of multipliers.

Antecedents

We commonly divide economies intotwo often opposing parts. In action, it's usagainst them; in primitive life, it is huntersand gatherers; in analysis, it will beprimary and secondary, productive andnonproductive, basic and nonbasic, exportand support, fillers and builders,productive and sterile workers, necessaryand surplus labor, etc. The followingnotes trace obvious antecedents.1

Mercantilistic thought is a primeexample. During the period in which themercantilists were dominant, normallyconsidered to be from 1500 to 1776, thenation-states of Europe wereconsolidating their power and gainingstrength to resist or conquer others. Thewriters who documented the timesemphasized a philosophy not unlike thatof a modern merchant or chamber ofcommerce.

The mercantilists stressed accumulatinga supply of gold with which to pursue thenation's political and military objectives.The economic base of a nation includedthe sectors which created a favorablebalance of trade. Goods were producedfor export despite the needs of a poorpopulation, export of unprocessedmaterials was prohibited, shipping in local

1 This section is based on (Oser 1963) and (Kang andPalmer 1958). Oser's The Evolution of EconomicThought is one of the best short histories of economicthought in print.



bottoms was forced whenever possible,and colonies were exploited as a source ofraw materials.

Thomas Mun, a merchant in the Italianand Near Eastern trade and a director ofthe East India Company, was probably themost famous of these writers. Hisexposition of mercantilist doctrine inEngland's Treasure by Foreign Trade,written in 1630, explained how "… toenrich the kingdom and to encrease ourTreasure." He emphasized a surplus ofexports as the key:

Although a Kingdom may be enriched by giftsreceived, or by purchase taken from some otherNations, yet these are things uncertain and ofsmall consideration when they happen. Theordinary means therefore to encrease our wealthand treasure is by Forraign Trade, wherein weemust ever observe this rule; to sell more tostrangers yearly than wee consume of theirs invalue.(from Oser 1963 p.14)

The Physiocrats, led by FrançoisQuesnay and briefly prominent in Francein the second half of the 18th centuryprior to the French Revolution, respondedto the excesses of the mercantilists withseveral points important to later thought.They considered society subject to thelaws of nature and opposed governmentalinterference beyond protection of life,property, and freedom of contract. Theyopposed all feudal, mercantilist, andgovernment restrictions. "Laissez faire,laissez passer," the theme phrase for thefree enterprise system, is from thePhysiocrats. They opposed luxury goodsas interfering with the accumulation ofcapital.

But, for our purposes, they wereprecursors of economic-base thought intwo ways. First, they were important intheir treatment of the sources of value. Tothe Physiocrats, only agriculture wasproductive. The soil yielded all value;manufacturing, trade, and the professions

were sterile, simply passing value on toconsumers. This classification ofproductive and sterile activities is similarto the basic and service classification ineconomic-base discussions.

And second, the Physiocrats visualizedmoney flowing through the economicsystem in much the same way as bloodflows through the living body. Quesnay'stableau economique was a predecessor ofthe circular-flow diagrams popularized inKeynesian macroeconomics.

Adam Smith, writing in 1776, andheavily influenced by these Frenchauthors, took a less extreme butnevertheless strong position. Heemphasized production of material ortangible goods and considered service andgovernment as unproductive.

Karl Marx, in das Kapital, also dividedthe economy into two parts. To Marx,necessary labor was the source of wealthand was paid for with a wage barelysufficient to maintain its provider.Surplus labor was also provided byworkers but its value was appropriated bythe capitalists in the form of surplus value.Workers had to produce not only whatthey consumed but also a surplus for thecapitalist. Menial servants, landlords, theChurch, and commercial activities wereunproductive – they added nothing to totalvalue.

Others of the nineteenth century weremore generous. Jean Baptiste Say in hisTreatise on Political Economy (1803)popularized Adam Smith in France. Say'sfamous Law of Markets, paraphrased as"supply creates its own demand," requiredthat all work be productive, that allcompensated activity creates utility.

Nevertheless, we can see a strong lineof thought dividing economic activitiesinto two parts, and we can see economic-



base concepts as fitting into a centuries-old pattern.

Modern origins

Modern literature on the economic basehas been voluminous, but plaguedoccasionally by scholastic sloppiness inappropriate citations.

It seems that Werner Sombart, aGerman (historical) economist writing inthe early part of this century, shouldreceive major credit for modern concepts.2

Sombart was responsible for thedistinction between "town fillers" and"town builders," ("Städtegründer" and“Städtefüller") which appeared inFrederick Nussbaum's A History of theEconomic Institutions of Modern Europe(with full permission). But in a series ofarticles in the early 1950's, Richard B.Andrews quoted extensively fromNussbaum without mentioning the factthat Nussbaum had based his book onSombart's work. Andrew's work waswidely circulated and became the standardreference.

The structure of macroeconomicmodels

It is convenient to begin with a reviewof the basic elements of model building.We can start with the simplest of allmacroeconomic models, the Keynesianmodel of a closed economy. This modelis presented algebraically in IllustrationError! Reference source not found..1and follows the standard format we willuse in all of our models: we outlinedefinitions, behavioral or technical

2 I rely on Günter Krumme for this statement (Krumme1968). On his excellent web site Professor Krummepoints out that, according to Marc de Smidt, Sombarthimself traces the concept back to a 1659 manuscript bythe Dutch mercantilst Pieter de la Court. Seehttp://faculty.washington.edu/~krumme/papers/sombart.html

assumptions, equilibrium conditions, andfinally the solution. Since this is a processwe will follow with each new modelconsidered, it may be worthwhile toreview the nature of these modelelements.

A definition is a statement of fact. Bydefinition, it is always true. Inmathematics, the proper term is identity.One of the more important identities inmacroeconomics is the national incomeidentity: realized national income (actualexpenditures) is the sum of realizedconsumption and realized investment. Inthe simple national model, this has to be atrue statement—it is a tautology. Actualexpenditures have to equal their sum!

Another important identity in thesimple model is that income (which isanother term for 'output') is equal to thesum of consumption and savings. We, asrecipients of incomes, either spend ourincomes or we save (don't spend). Thisidentity can also be taken as a definitionof saving as the difference betweenincome and consumption.

Behavioral assumptions are equationsdescribing the behavior of certain groups,or actors, in the economy. In this case,the key behavioral relationship is theconsumption function, which postulatesconsumption as dependent on, or causedby, income:

C = f(Y)

which in its linear form may be expressedas:

C = a +cY

where a represents autonomousconsumption and c is the marginalpropensity to consume (dC/dY). Theparameters of the equation are a and c.Recall that if a>0, dC/dY<C/Y.



An incidental but important result ofthis assumption is that saving is also afunction of income:

S Y - C = -a + (1 - c)Y

The other important behavioralassumption in this simple model is that

investment, I, is determined outside thesystem. It is planned. In terms commonto model building, it is an exogenousvariable in contrast to consumption, whichis determined endogenously (that is,'within the system').

Illustration Error! Reference source not found..1 The simple Keynesianmodel

Definitions or identities:Planned Expenditures Consumption + Investment (planned sources of income)

(1) E C + IActual Income Consumption + Savings (actual disposition of income)

(2) Y C + SBehavioral or technical assumptions:

Consumption = A linear function of income (both planned and actual)(3) C = a + cY (c < 1 = the marginal propensity to consume)

Investment = Planned investment (an exogenously determined value)(4) I = I'Equilibrium condition:

Income = Expenditures, or actual income is equal planned expenditures(5) Y = E

or, with C + S = C + I, we can subtract C from both sides to form anequivalent equilibrium condition:

Drains = Additions(6) S = ISolution by substitution:

Y = C + I Substitute (1) into (5)Y = a + cY + I' Substitute (3) and (4)Y - cY = a + I' Gather the Y, or income, terms(1 - c)Y = a + I' Factor out YY = {1/[1 - c]}*(a + I') Isolate Y through division

The simple Keynesian investment multiplier is:dY/dI = 1/[1 - c]

(An example of a technicalassumption in economics is theproduction function. A productionfunction describes the relations between

inputs and outputs. A familiar exampleis Q=F(K,L), commonly used to describehow capital and labor are combined toproduce output.)



Equilibrium is a condition in whichthe expectations (plans) of decision-makers (actors) in the system are met.In this simple model, the equilibriumcondition is that income equals plannedexpenditures, or, what is the same thing,that saving (which sets the limits onactual investment) equals plannedinvestment.

The point is that planned investmentand saving do not have to be equal (eventhough, in the end, actual saving has toequal actual investment—this is afundamental principle of accounting).When they are equal, then all parties aresatisfied. When they are not, forces areat play which will take income to alower or higher level, bringing savinginto equality with planned investment.3

Good introductions to the art ofmodel-building can be found in severalreadily available books (e.g. Bowers andBaird 1971; Kogiku 1968; Neal andShone 1976). The simple Keynesianmodel is outlined in almost all texts onthe principles of economics. A goodreference is (Case and Fair 1994).

The "strawman" export-base model

It is common in economics toconstruct a "strawman" against which torail and argue. Nowhere is this practicemore common than in the regional

3 This paragraph brings “Say’s Law” into play. Statedby Jean Baptiste Say in the early 1800s as the “Law ofMarkets,” the idea that “supply creates its owndemand” was named in 1909 by Frederic Taylor.Keynes succinctly restated it as above and argued thatit did not apply. In Say’s time, since saving andinvestment were often done by the same landedpeople, it might have been more valid. But in moderntimes with complex banking systems, saving is doneby many people who do not buy capital goods andinvestment is done by people who borrow thosesavings. So the possibility of actual savings differingsignificantly from planned investment became real.

literature. The "export-base" model, inwhich the sole determinant of economicgrowth is exports, is often built torepresent the arguments of otherpractitioners. However, you can seldomfind an "export-base" theorist who is notalso an "economic-base" theorist readilyacknowledging many other determinantsof growth than exports alone.

Now let us construct this strawmanand see how a pure export-base stance isuntenable. We move into an openeconomy and make exports the soleexogenous factor. If any autonomousexpenditure is included (the easiest is forconsumption), then regional income canexist even when exports are zero (Ghali1977).

Presented in Illustration 3.2, themodel differs only slightly from thesimple Keynesian model. With Keynes,the key leakage was savings. Heexplained the underemployment of adepressed economy as resulting whenplanned investment fell below full-employment equilibrium levels due to alack of confidence in investmentmarkets. His endogenous variable wasconsumption, through which mostincome flows occurred—the flowsbecame disconnected in the saving-investment path.

In the export-base model, theendogenous flow remains consumption,redefined now as “domesticexpenditures.” We completely ignoresaving and hide investment expenditureswithin domestic expenditures (we areconcerned not about explainingdepression in the whole economy butabout explaining changes in regionalincome). The function of saving increating a leakage from the economy isnow assumed by imports, which isdefined as a function of income. The



function of investment is now assumedby exports, the driver of the export-based economy.

This model obviously stressesopenness and dependence of the regionon events beyond its reach.

Illustration 3.2 The pure export-base model

Definitions or identities:Total expenditures Domestic expenditures + Exports (inflows)

(1) E D + XIncome Domestic production +Imports

(2) Y D + M, or D Y - MBehavioral or technical assumptions:

Imports = a linear function of income(3) M = mY (m<1, the marginal propensity to import)

Exports = an exogenously (outside-region) determined value(4) X = X'Equilibrium condition:

Income = Total expenditures(5a) Y = E

orDrains = Additions

(5b) M = XSolution by substitution:

Y = Y - M + X Substitute (1) and (2) into (5a)Y = Y - mY + X' Substitute (3) and (4)Y - Y + mY = X' Gather the Y, or income, termsmY = X' ReduceY = (1/m)*X' Isolate Y through division

The export-base multiplier is:dY/dX = 1/m

The typical economic-base model

To make the model slightly morerealistic (or, rather, less simplistic!),saving and exogenously determinedinvestment can be added back into thesystem.

Illustration 3.3 includes these todevelop an almost typical economic-basemodel. Only minor interpretivecomments are required.

The missing element is autonomousconsumption (which appeared in thesimple Keynesian model). Whether ornot it is included seems to me to be amatter of personal preferences. On theone hand, it might be nice to becomplete and consistent with theKeynesian model. In addition, it servesto warn us that the consumption functionis probably curvilinear, originating at theorigin and rising at a decreasing ratewith respect to income. The marginal



propensity to consume at the range ofincomes over which we might work isless than the average propensity toconsume. A positive autonomousconsumption permits us to simulate thiscase.

On the other hand, we already haveone exogenously determined nonexportvariable, investment. The investment

multiplier is identical to that whichwould be calculated for autonomousconsumption—we have the resultswithout the bother. While this is a logicwhich might reduce a model to pulp ifpursued too rigorously, I have leftautonomous consumption out of thisillustration.

Illustration 3.3 The pure economic-base model

Definitions or identities:Total expenditures Domestic production + Exports + Investment

(1) E D + X + IIncome Consumption + Saving

(2) Y C + SConsumption Domestic expenditures + Imports

(3) C D + M, or D C - MBehavioral or technical assumptions:

Consumption = a linear function of income(4) C = cY (c = the marginal propensity to consume)

Imports = a linear function of income(5) M = mY (m = the marginal propensity to import)

Exports = an exogenously (outside-region) determined value(6) X = X'

Investment = an exogenously (outside-system) determined value(7) I = I'Equilibrium condition:

Income = Total expenditures(8a) Y = E

orDrains = Additions

(8b) M + S = X + ISolution by substitution:

Y = C - M + X + I Substitute (1) and (3) into (8a)Y = cY - mY + X' + I' Substitute (4), (5), (6) and (7)Y - cY + mY = X' + I' Gather the Y, or income, terms(1 - c + m)Y = X' + I' Factor out YY = {1/[1 - (c - m)]}*(X' + I') Isolate Y through division

The economic-base and investment multipliers are:dY/dX = 1/[1 - (c - m)], and dY/dI = 1/[1 - (c - m)]



Techniques for calculatingmultiplier values

Comparison of planner's relationshipand the economist's model

Concentrating purely on the practicalneed to develop an easy way to forecastcommunity change, early plannersdeveloped economic-base ratios (T/B forthe average ratio, and T/for themarginal ratio, where the lettersrepresent total (T) and basic (B) income(or employment) by pure observation asrules of thumb. By 1952, economists(Hildebrand and Mace 1950) haddeveloped export-base models in thesame analytic framework as theKeynesian macroeconomists, withmultipliers expressed as (1/(1-PCL),

where PCL represents either the averagepropensity to consume locally producedgoods (APCL) or the marginal

propensity (MPCL). Could these

approaches be equivalent? Yes. CharlesM. Tiebout showed us how (Tiebout1962). Tracing the metamorphosis foraverage propensities,

T/B = 1/(B/T) = 1/((T-NB)/T)) = 1/(1-NB/T) = 1/(1-APCL)

Here, the ratio of nonbasic activity tototal activity (NB/T) is the equivalent ofthe average propensity to consumelocally produced goods.

So, if we can obtain values of totaland basic variables over a period ofyears, we can estimate marginal export-base multipliers by regressing the totalon the basic values. With the regressionline formulated as T = a +bB, the slope bis the marginal multiplier (T/ forthe region.

The survey method

Of course, the most straight-forwardmethod is simply to ask businesses in thearea to specify how much of theirrevenues is basic and to use theirresponses to accurately divide localbusiness activities into basic and servicecomponents. In practice, this is seldomdone.

The neglect of the survey approach iseasy to explain. It is the most expensiveand time-consuming of approaches.Questionnaires on sensitive issues suchas revenues, employment, and marketsare seldom answered freely; to obtaineven a smattering of responses the studyteam must resort to personal interviews.And even then, the interviewers must beskilled and persuasive.

In addition, if the area is of any size,the survey would require carefulplanning. A canvass would beprohibitive and the sample must becarefully stratified and selected torepresent the broad spectrum ofactivities represented in moderncommunities. Such care and expensewould meet the test of rationality only ifdata collection were in the context of amuch larger study. The limit to thevalue of a simple export-base ratio isfairly low, in the hundreds of dollars.

A final argument against this simpleapproach is that the survey wouldprobably yield data for only one year,leading to calculation of an averagemultiplier when a marginal multiplier isthe most appropriate.

The ad hoc assumption approach

The easiest and least expensive ofmethods is simply to rely on arbitraryassignment of activities to basic ornonbasic categories. This could be done



by assignment of, say, employment orpayrolls for entire industries intocategories, or it could be accomplishedwith a little more finesse by estimatingproportions of employment involved inbasic activities.

Needless to say, the chance of errorsis large even for experienced analysts,and the multiplier will again be anaverage one with limited use inanalyzing the effect of change.

Location quotients

The location quotient is probablyresponsible for the long life andcontinuing popularity and use ofeconomic-base multipliers. Thesequotients provide a compelling andattractive method for estimating exportemployment (or income).

A location quotient is defined as theratio

LQi = (ei/e)/(Ei/E),

where ei is area employment in industry

i, e is total employment in the area, Ei is

employment in the benchmark economyin industry i, and E is total employmentin the benchmark economy. Normally,the "benchmark" economy is taken to bethe nation as the closest availableapproximation to a self-sufficienteconomy.

Assuming that the benchmarkeconomy is self-sufficient, then alocation quotient greater than one meansthat the area economy has more thanenough employment in industry i tosupply the region with its product. Anda quotient less than one suggests that thearea is deficient in industry i and mustimport its product if the area is tomaintain normal consumption patterns.

Surplus or export employment inindustry i can be computed by theformula

EXi = (1 - 1/LQi)*ei , LQi > 1,

which is easily shown to be thedifference between actual industryemployment in the area and the"necessary" employment in the area.

In fact, then, excess employment canbe computed without reference tolocation quotients through this reductionof the formula:

EXi = ei - (Ei/E)*e

It is convenient to retain the initialformula as a reminder of the logic, andto compute location quotients asreminders of the strengths of exportingindustries.

Now it is easy to estimate exportemployment for each industry in the areaand to sum these estimates to yield avalue for export employment for the areain some particular year. With thisnumber and total employment, anaverage multiplier for the area can becomputed. With a set of these valuesover 10-20 years, the more acceptablemarginal multiplier can be estimated bysimple regression.

While it is common to useemployment as the primary basis forthese calculations, other measures suchas wages and salaries are just asappropriate. Indeed, wage data is moreaccessible electronically, especially onCD-ROM. County Business Patterns, astandard source of employment andpayroll data, is available for years since1986, two years per disk. Inconsiderable detail, this is the best datafor recent years, but skill withmainframe computers, tapes, andprogramming is required to gain access



for earlier years. The RegionalEconomic Information System (REIS),updated on CD-ROM annually by theU.S. Department of Commerce with atwo-year lag, includes a relativelyaggregated 16-category employmentseries for the years 1969-2000 as well asa more detailed earnings series for everycounty in the nation (categories arebased on the old Standard IndustrialClassification (SIC) system). This datamakes earnings-based location quotientsa snap, especially if historic estimatesare desired. The REIS files releasedJune 2009, can be downloaded free ofcharge fromhttp://www.bea.gov/regional/docs/reis2007dvd.cfm .

From 2001 on, the industry categoriesare based on the new North AmericanIndustry Classification System (NAICS),with 23 categories of employment andeven more categories for earnings. Thisshift in industry definitions means thatcategorical data is not available in timeseries. Everything starts anew in 2001.

Location quotients have been in useby regional analysts for over 50 yearsnow, and have been commented on atlength. We should look at theassumptions involved in their use as wellas the advantages and disadvantages.

The literature records at least threespecific assumptions: (1) that local andbenchmark consumption patterns are thesame, (2) that labor productivity is aconstant across regions, and (3) that alllocal demands are met by localproduction whenever possible.

The first assumption is not serious:not only can we not discern differencesin consumption patterns withoutextraordinary expense but we cansuspect that differences in productionpatterns are more important. Purchases

of intermediate goods by producersdiffer for regions depending on industrymix. (It turns out that we can accountfor industry mix with input-outputmodels, so this difference has beenaccounted for by the march of time.).

The constant-labor-productivityassumption is difficulty to avoid. Itsimpact can be ameliorated slightlythrough using earnings data, which canbe assumed to reflect regionalproductivity variation throughdifferences in wage rates. (Thisassumption could in turn be attacked ifwages vary more by area cost of livingthan by productivity.)

The assumption that local demandsare met first by local production is themore tenuous of the three. It isobviously not true, as any visit to agrocery or clothing store will attest. Butit is common, and a better alternative ishard to come by.

In addition to the disadvantagesaccruing from these assumptions,another major fault is that the method isdependent on the degree of aggregationof the data, making comparisons amongvarious studies of little value. Toillustrate the problem, consider the foodand kindred products industry in Atlanta.The location quotient computed for thisbroad industry should be less than one,and if excess employment werecomputed based on this classification,none would be credited to the foodindustry. But if the classification weremore detailed, the soft-drink industrywould show a large number of excessemployees, since the headquarters ofCoca-Cola is in the city.

The overpowering advantages ofusing location quotients are that themethod is inexpensive and the exerciseof computing excess employment may



give the analyst an opportunity to gaininsights of interest in themselves.

Minimum requirements

In the 1960's, when availablecomputing technology favored frequentuse of economic-base models, one of thealternatives to the use of locationquotients was the minimum-requirements approach (Ullman andDacey 1960). This variation involved aslight revision of the location-quotientformula to

EXi = ei - (Ei/E)min

*e ,

where (Ei/E)min

is the minimumemployment proportion for industry i incities of size similar to the subject city.You can readily see that we havesubstituted a varying benchmarkemployment proportion for a constantone:

LQi = (ei/e)/(Ei/E)min

.

While still appearing in various formsin the literature, the method suffers fromtwo major criticisms. One is that, ifenough cities are included in the selectedset, all regions will be exporting andnone may be importing. The other issimilar in that, if we use data defined ina fine level of detail (which should bean improvement, as it was in location-quotient estimates), we may reduce localneeds to near zero and make almost allproduction for export (Pratt 1968).

At any rate, the method is notcommonly used now. The location-quotient method remains the virtuallysole survivor as a simple means ofidentifying export industries.

"Differential" multipliers: a multiple-regression analysis

Another approach which has beenused in estimating economic-basemultipliers is to fit a multiple-regressionequation to regional data. The first ofthese studies arose in a study of theimpact of military bases on Portsmouth,New Hampshire in 1968 by Weiss andGooding (Weiss and Gooding 1968).

Simple economic-base models ignorethe possibility that different industriesmay have different impacts on theircommunities. The regression techniqueeliminates this simplifying assumption.Weiss and Gooding set up an equation

S = Q + b1 X1 + b2 X2 + b3 X3,

where S represents service employment,Q is a constant, and the X terms are, inorder, private export employment,civilian employment at the PortsmouthNaval Shipyard, and employment atPease Air Force Base.

With data fitted from 1955-64, theirresults were

S = -12905 + .78 X1 + .55 X2 + .35 X3

(.31) (.23) (.14)

The multipliers are 1+ bi for eachsector.

Weiss and Gooding used a mixture ofassumption and location quotientmethods in allocating exportemployment and assumed that the exportsectors were independent and thatworkers in the export sectors demandedsimilar services.

This variation on economic-basemodeling has not fallen into widespreaduse for several reasons: its flexibility (innumber of exogenous sectors) and thestatistical significance of coefficients arelimited by the number of observationsavailable; determining the export



content of industry employment remainsa demanding chore; and with the rise ofdesktop computing, input-output modelsare better sources of industry-specificmultipliers and are similar in cost.

Critique: advantages,disadvantages, praise, criticism

Economic-base models suffer fromold age: they have been built by somany analysts with varying levels ofquality and they have been criticized sooften that little remains except theconcept.

The indictment would include thefollowing phrases:

Short run

Nonspatial

Simple adaptations of national models

Data is normally available for administrativeunits (counties) which may be poorly definedas economic regions.

Ignores capacity constraints

Assumes perfect elasticity of supply for inputs

Pits the area against the rest of the world,showing no interdependence between regions

Multiplier varies with size of region. (As aregion grows it diversifies, importing less andso increasing local consumption and themultiplier (Sirkin 1959)). Also, larger regionstend to influence neighbors more and so toenjoy larger feedback effects (Richardson1972)

An employment multiplier is often used todiscuss income changes. (But this assumesthat employment and per capita income areperfectly correlated -- in a simple economywith perfectly elastic supplies of labor thismight be the case although, of course, theworld is not simple.)

Assumes that exports are the sole determinantof economic growth. (It is not reasonable forus to take the rap for this.) Any rationalperson can see that the determinants of growthare many -- the simple model just emphasizesone determinant. Perhaps the fault lies in earlyattempts to formulate multipliers and the easewith which the simple multipliers could beconstructed. (Ghali 1977; Sirkin 1959))

Direction of dependence may be questionable:which comes first, export growth or a strongservice sector, or interdependence? Shouldwe be concerned with preconditions for exportgrowth (setting up an attractive service sector)in this simple model? Are we planning growthor explaining the true basis?

Although castigated for decades, theeconomic-base model has survived as avery succinct expression of the power ofdemand in regional incomedetermination. The most current, andperhaps the clearest and most complete,statement of its status is found in arecent review by Andrew J. Krikelas(1992), reprinted below with permissionfrom the Atlanta Federal Reserve Bank..

Review of Economic-Base Literature Krikelas

Economic Review, FRB Atlanta 13 Jul/Aug 1992

APPENDIX A REVIEW OF ECONOMIC-BASELITERATURE

The following article appeared in theEconomic Review, Federal Reserve Bankof Atlanta, July/August 1992, pp. 16-29 ,and represents the latest in reviews andcritiques of economic-base literature.

Why Regions Grow:A Review of Research

On the Economic-Base Model

Andrew C. KrikelasThe author is an economist in theregional section of the Atlanta Fed'sresearch department.

Regional economic models are used in avariety of decision-making contexts.Government officials use them toprepare annual budgets. Businesses relyon them for producing short-run marketdemand forecasts and for analyzinglonger-term growth strategies. Urbanplanners and transportation officials usethem to develop long-range plans forurban and regional development. Finally,state and local policymakers turn to themto get new ideas for programs andpolicies to promote long-run regionalgrowth.Although it would be convenient if asingle model had been developed toserve all these purposes simultaneously,no such model is ever likely to exist.Instead, regional models tend to behighly specialized in terms of the issuesthat they are able to address and the timehorizons over which their analyticalresults are most reliable. For example, ashort-run forecasting model might serve

the needs of state or local governmentofficials engaged in the annualbudgeting process, but it wouldcontribute little information relevant tolong-run local economic developmentissues confronting planners andpolicymakers. Only rarely is a regionalmodel able to perform well in more thanone of these distinct decision-making

contexts.1

The rapid pace of urban growth duringthis century, along with the challenge ithas presented for planners trying toanticipate and influence this growth, hasensured a healthy demand for regionaleconomic models, particularly since1945. Unfortunately, models suppliedhave been inadequate.

At the beginning of the postwar period,the economic base model was probablythe only such instrument generallyavailable for regional economic analysis.This model focuses on regional exportactivity as the primary determinant oflocal-area growth; it is one of the oldestand most durable theories of regionalgrowth, with origins extending at least asfar back as the early 1900s. However,economic base theory received thegreatest amount of attention fromscholars in regional science between1950 and 1985. Despite the model'sacceptance over such a long period,when the noted regional scientist HarryW. Richardson, writing for a specialtwenty-fifth anniversary issue of theJournal of Regional Science, reflectedupon the more than forty years ofresearch conducted within this paradigm,



he concluded that "the findings oneconomic base models are conclusive.The spate of recent research has donenothing to increase confidence inthem.... The literature would need to bemuch more convincing than it has beenhitherto for a disinterested observer toresist the conclusion that economic basemodels should be buried, and withoutprospects for resurrection" (1985, 646).

Like Richardson, others over the yearshave expressed concern with the narrowfocus of economic base theory onexports—just one portion of the demandside of the regional growth equation—tothe exclusion of important supply-sidefactors and constraints. Many havesuggested that economic base theory, itsanalytical and methodologicaltechniques, and the public policies that itpromotes should be abandoned in favorof other, more comprehensive theories ofregional growth and development.

Nevertheless, economic base researchcontinues. Most notably, James P.Lesage and J. David Reed (1989) andLesage (1990) have provided empiricalevidence in support of the economicbase hypothesis as both a short-run andlong-run theory of regional growth.These authors suggest that their modelscould be used both for short-termforecasting of regional employment,income, and product and for longer-range regional economic planning andpolicy analysis. If these claims werevalid, then the economic base model,rather than being of little value, wouldbe one of the few regional models thatmight be useful in each of these verydifferent but crucially importantdecision-making contexts.

Because regional economic models playsuch an important role in planning andpolicy discussions, it is important tohave a clear understanding of theirstrengths and weaknesses. Limitations ofthe economic base model in particular,because it tends to be widely used,should be recognized. Recent researchhas provided evidence suggestingsubstantial improvement in traditionallystatic economic base modelspecifications through the adoption oftechniques routinely employed in themacroeconomics time-series literature.However, this author's research suggeststhat these studies may have overstatedthe usefulness of these new economicbase model specifications (Andrew C.Krikelas 1991).

The purpose of this article, therefore, istwofold. First, a concise analyticalhistory of the old and extensiveeconomic base literature generated by avariety of professional and academicdisciplines is provided in order to placerecent research in perspective. Thediscussion then turns to the centralquestion addressed in Krikelas (1991):Can techniques borrowed from statisticaltime-series literature successfullybreathe new life into the traditionaleconomic base model?

Definition of the Economic BaseConcept

As originally formulated, the economicbase model focused on regional exportactivity as the primary source of local-area growth. According to this theorytotal economic activity, ET is assumed tobe dichotomous, with a distinction beingmade between basic economic activity,EB (activities devoted to the productionof goods and services ultimately sold to



consumers outside the region), andnonbasic economic activity, ENB, whichincludes activities involved in producinggoods and services consumed locally:

ET = EB+ ENB (1)This division of regional economicactivity into these two distinct sectors isthe central concept of the model.2 Aserious empirical concern is immediatelyraised by this approach, however,because appropriate export data areavailable at any subnational level only athigh cost and with long lags. Variousalternative measures have been proposedand analyzed in the literature over theyears, but none has been found entirelyadequate. Data problems, therefore, havealways complicated economic baseresearch.

While the central concept of theeconomic base model is the duality ofregional economic activity, itsfundamental behavioral assumption isthat nonbasic economic activity dependson basic economic activity. In thisperspective, external demand for aregion's exportable goods and servicesinjects income into the regionaleconomy, in turn augmenting localdemand for nonexportable goods andservices. The model assumes that theincome injected into the regionaleconomy and the accompanyingpotential for developing locally oriented,nonbasic industries are in proportion tothe size of a region's export base. Staticand demand-oriented, the model ignoresfactors that affect the supply of aregion's output and other changes, suchas the introduction of new products, thataffect demands.

ENB =f(EB) = + * EB . (2)Equations (1) and (2) can then becombined into the reduced-formexpression in equation (3), which

indicates that total economic activity isprimarily a function of basic activity:

ET = + (1 + )* EB (3)

The expression (1 + ) is commonlyreferred to as the economic basemultiplier, and the parameter, , is calledthe economic base ratio.

When applied to analyzing regionalgrowth, the economic base modelsuggests that the growth process will beled by industries that export goods andservices beyond regional boundaries. Iteven offers a prediction, captured in themultiplier, of the total regional impactlikely to result from a change in basiceconomic activity generated outside theregion. Understanding the future path ofa regional economy, the model implies,requires simply concentrating on theprospects for the base industries. Thesefew important industries are oftendubbed "engines of regional growth."

This simple model captures the essenceof economic base theory. Although themodel has been enhanced over the yearsto include additional variables as well asto capture more explicitly the dynamicnature of the regional growth process,most changes have been made within thescope of this simple demand-orientedspecification. In general, economic basemodels have not evolved toacknowledge the potential impact ofmany important variables that may affectregional growth—interregional capitalflows; labor migration patterns; changesin products, tastes, and productionprocesses; demographic shifts; andchanges in state and local tax laws, toname a few. Because these issues aregenerally too important to ignore, manyregional scientists have concluded thateconomic base theory lacks thecomplexity to provide a useful



framework for analyzing many regionaleconomic issues and policies. Thefollowing review of the developmentand testing of the model will summarizewhere the debate on this topic stands atthis point.

History of the Economic BaseLiteratureFive fairly distinct chronological periodscharacterize the history of the economicbase literature: (1) the origin of theconcept, 1916-21; (2) earlydevelopment, 1921-50; (3) the firstround of serious debate, 1950-60; (4) thesecond round of debate, 1960-85; and(5) a third and perhaps final round ofdebate begun in 1985 and continuingtoday. Decades of research within theeconomic base paradigm have created abody of conventional wisdomconcerning the uses and limitations ofthe model, both in theory and in practice.Nonetheless, as yet another round ofdiscussion has begun, it seems that fewlessons of the past have been learned andthat a brief summary of the history ofthis literature might be useful.

Origin of the Economic Base Concept.The essential duality of regionaleconomic activity that is central to thesimple model expressed in the equationsabove was first articulated in 1916 bythe German sociologist Werner Sombart,who wrote of "actual city founders,"identified as the "active, originative, orprimary city formers"—those whosepositions of authority, wealth, oroccupation allowed them to drawincome from outside the city—and the"passive or derived or secondary cityfounders," whose livelihood dependedon the city formers (Gunter Krumme1968, 114).3

In 1921 M. Arrousseau made a similarobservation in commenting on therelationship between what hedistinguished as a town's primary andsecondary occupations: "The primaryoccupations are those directly concernedwith the functions of the town. Thesecondary occupations are thoseconcerned with the maintenance of thewell-being of the people engaged inthose of primary nature" (John W.Alexander 1954, 246).4 Also in 1921,landscape architect Frederick LawOlmsted distinguished between what hecalled primary and ancillary economicactivity in an urban area (Alexander1954, 246.)5

Thus, although Sombart was apparentlythe first to observe formally the seemingduality of urban and regional economicactivity, the remarks of hiscontemporaries Arrousseau and Olmstedmake it abundantly clear that the conceptwas ripe for expression. By the early1920s, therefore, the economic baseconcept had generally surfaced as apotential theory for explaining theregional growth process.

Early Development of the Theory.Following establishment of the theory,the next logical step should have beenthe empirical testing of the validity ofthe model's central hypothesis. However,this step was almost universally ignoredand the model adopted as useful as therapid growth of cities early in thecentury pressured state and localofficials to improve the way in whichthey developed plans for urbanexpansion and the provision of publicinfrastructure and government services.The economic base model provided amuch-desired framework for developingsuch plans, and studies designed to



identify and measure basic industries—economic base studies—quickly becameprimary tools employed in acquiringinformation for long-range planning.

After identifying a region's export base,economic base studies calculate a local-area economic base ratio, . Oncecalculated, the economic base ratio canbe used with forecasts of the futuregrowth of the region's export baseindustries to predict the region's overallgrowth. The study's focus on the smallernumber of industries identified asregional export industries helpsstreamline the process of forecastingtotal regional economic activity. Inaddition, by identifying those industriesconsidered most important to theregional growth process, an economicbase study provides information thatadds insight to discussions of regionalindustrial policies and programs.

Sombart's analysis of the Berlineconomy, published in 1927, was thefirst economic base study conductedduring this period. Sombart, complainingthat "nobody makes the effort to sitdown with a pencil and figure out withthe help of occupational statistics howmuch there actually is of a city-formingindustry in a city such as Berlin,"developed an empirical approach fordividing an urban economy into its dualparts (Krumme 1968,116).6

Lacking detailed information on regionalexport activity, Sombart relied uponindustry employment data collected inBerlin in 1907 to estimate the basic andnonbasic sectors of the city's economy.Relying mainly upon his personaljudgment, Sombart estimated thatapproximately 262,000 of Berlin's totalwork force of 543,000 were employed in

export base industries (Krumme 1968,113). These calculations placed Berlin'snonbasic/basic ratio, , at 1.07, anapproximately one-to-one relationship.Although Sombart did not use thisinformation to forecast Berlin's growth,he could have done so. Making a morelimited forecast of the prospects forthose industries he had identified asbeing part of the city's export base andmultiplying that total by the city'seconomic base multiplier (1 + ) of 2.07(assuming that the city's base ratio hadremained relatively stable in theintervening twenty years since thecensus was conducted) would haveprovided a forecast of the change in totaleconomic activity expected in Berlin as aresult of some externally generatedchange in demand for its export product.

The reliance on secondary data sourcesfor Sombart's study of Berlin's economicbase is typical of most such research. Aspointed out earlier, even today theappropriate regional export data requiredto conduct an adequate economic basestudy are available only at relativelyhigh cost. The comprehensive economicanalysis of the city of Oskaloosa, Iowa,published in Fortune magazine in 1938illustrates this point ("Oskaloosa. . ."1938).Although published in a popularmagazine, this study represents animportant contribution to research on theeconomic base theory. The magazinestaff conducted a complete census of thetown's 3,000 families in order todetermine the origin and destination ofincome flows within the city. They alsoconducted a census of the town'sbusinesses, including an accounting ofthe destination of their output and thesource and value of the most important



inputs into the local-area productionprocess.

The results of the study indicated that in1937 Oskaloosa was a net exporter ofgoods and services to the rest of theworld and that manufactured goods andprofessional services were the town'sleading export industries. The study'sfindings are interesting because theywere based upon a census that provides arelatively accurate portrayal ofOskaloosa's export activity during theyear studied. Even by present standardsthis study represents one of the mostthorough economic analyses of a smallcommunity ever published.

The great effort required to collect thesedata, however, explains why a survey- orcensus-oriented approach to economicbase identification generally has beenabandoned for the nonsurveyidentification techniques made popularby Homer Hoyt in the late 1930s.Working with the Federal HousingAdministration during the mid-1930s,Hoyt developed and employed aneconomic base methodology forproducing forecasts of local housingmarket demand. His techniques becameknown to a wide audience with theoriginal publication of his textbook,Principles of Urban Real Estate(coauthored with Arthur M. Weimer in1939), which Richard B. Andrews calledthe first "complete statement of thetheory of the economic base." Incommenting on the impact of this work,Andrews continued, "This statementincluded much material that was newoutside of technical reports. Forexample, it introduced in formal fashionthe idea of a mathematical relationbetween basic employment and serviceemployment.... Hoyt considered the

economic base idea to be a tool thatmight be employed in analyzing theeconomic background of cities with theobjective of forecasting the future of theentire city" (1953a, 163).

In this text Weimer and Hoytdistinguished between "urban growth"and "urban service" industries,suggesting that a region's potential forgrowth depended primarily upon theprospects for the region's urban growthindustries. They provided a six-stepprocedure for identifying suchindustries. Using relatively accessibleincome and employment data, theauthors developed a methodology thatrepresented a combination of what hasbecome known as the assignmenttechnique and the location-quotienttechnique of economic baseidentification. The assignment techniqueis essentially identical to Sombart'smethodology, in which personaljudgment is used to assign industrieswithin a particular regional economy tobasic and nonbasic sectors. The location-quotient technique, on the other hand,relies upon regional economic data tomake such distinctions.

Location-quotient methodologycompares a region's concentration ofeconomic activity in a particular industrywith that of a benchmark economy,usually the entire country in which theregion is located. If the regionalconcentration, measured in terms of theindustry's share of total regionalemployment or income, exceeds thebenchmark economy's concentration inthat industry, the surplus level ofemployment or income is assumed tomeasure regional export activity. Forexample, if an industry accounts for 6percent of regional employment but only



2 percent of national employment, two-thirds of that industry's employmentwould be called basic. (If the regionalactivity in an industry is less than that atthe national level, the industry iscategorized as nonbasic.) Making thisidentification requires only industryemployment or income data for theregion and a similar set of data for anappropriate benchmark economy.

Although Weimer and Hoyt were not thefirst to propose using the locationquotient and assignment techniques asnonsurvey methodologies for dividingregional economic activity into its basicand nonbasic components, disseminationof the techniques through their textbookintroduced these shortcuts to a wideaudience. With these methodologiesavailable it became feasible for localdevelopment officials to adopt theeconomic base paradigm for purposes ofanalyzing specific urban and regionaleconomies. During the latter half of the1940s, once these techniques hadbecome more widely known, a muchlarger number of cities and states beganto use the economic base model in urbanand regional planning and economicanalysis.7

Theoretical Debate. By 1950 economicbase theory and its methodologicaltechniques had become established asthe primary tools of regional planning.The theory itself had been accepted,uncritically, as an explanation of local-area growth and economic development.Between 1950 and 1960, however,discussion at the theoretical andmethodological level turned directly tothe question of the validity of theeconomic base hypothesis itself.Unfortunately, only a handful of

empirical tests were reported during thisentire decade.

The earliest and most cogent critique ofeconomic base theory was presented byGeorge Hildebrand and Arthur Mace(1950) in their analysis of the LosAngeles metropolitan area. Thisimportant contribution identified thetheoretical model upon which theeconomic base paradigm was foundedand performed an empirical test thatprovided evidence supporting thevalidity of the economic basehypothesis, at least for short-runforecasting.

Hildebrand and Mace's most significantcontribution was their explicitformulation of economic base theory asa testable behavioral hypothesis. Theirresults, which demonstrated astatistically significant short-runrelationship between basic and nonbasicemployment in Los Angeles, representedthe first empirical confirmation of theeconomic base hypothesis. Furthermore,the authors formulated their tests withinthe context of an explicitly Keynesiannational income model and then outlinedthe inherent limitations of such a model.

Consider the familiar Keynesianrelationship:

Y = C + I + G + (X - M), (4)where total regional income, Y, isdivided into a number of distinct sectors,including consumption, C; investment, I;government expenditures, G; andexports minus imports, X - M. Thereduced-form expression of this modelwould include some smaller set ofexogenous variables, only one of whichwould be regional exports. (Otherexogenous variables would include theautonomous components of



consumption, investment, governmentexpenditures, and imports; marginalpropensities to consume locally, toinvest locally, and to import; and localand federal tax policies.) It is this set ofexogenous factors that would determine,theoretically, a region's total incomelevel, Y.

The economic base model focuses onone particular aspect of this relationship,regional export activity, X (EB inequation [1] above), and can beconsidered a special case of the moregeneral Keynesian model in equation(4). Given this interpretation, it becomesclear that for exports to be consideredthe only exogenous determinant ofregional growth, all other relevantfactors, related to both demand andsupply, must remain fairly constant or befunctions of export activity. Althoughthis might be a tenable assumption in theshort run, it probably is an extremelypoor one in the long run. Hildebrand andMace made this observation explicit andsuggested that the model was mostappropriate for anticipating regionaleconomic trends over a short timehorizon. In addition, they listed some ofthe other variables that they thoughtshould be taken into account indeveloping a more comprehensive modelof regional economic activity:population levels and interregionalmigration patterns, regional capitalinvestment levels and annual flows, stateand local tax policies, and changes in thecost of transportation to reach externalmarkets. Despite these reservations,Hildebrand and Mace offered a fairlyencouraging assessment of the prospectsfor this type of research, based on theavailability of additional census data andfurther empirical analysis across a ten-year span. 8

Unfortunately, the lessons contained inHildebrand and Mace's study were notwidely disseminated. Hildebrand andMace were among the first economists tocontribute to the economic baseliterature. Their article was published ina journal not normally read bygeographers and urban planners, who,before 1950, had played a dominant rolein the research conducted within theeconomic base paradigm. Therefore,rather than playing the role of a seminalarticle to a further body of empiricalresearch, the Hildebrand and Macearticle remained relatively unknown.The debate of the 1950s brought manyof their important insights to theattention of geographers and urbanplanners, but it took nearly a decade forall of these contributions to beuncovered.

Most of the 1950s' debate on economicbase theory was conducted in thegeography and planning literatures. Theorigin of this debate can be traced to aseries of nine articles published byAndrews between 1953 and 1956 (seereference list). These articles provided acareful exposition of economic basetheory and the methodologies that hadbeen developed to analyze urban andregional economic activity. The author'sstated purpose was to explore andevaluate the entire concept. "We haveoperated far too long on a set of ideaswhich appear valid but which, despitesubstantial conceptual omissions anddifficulties of application, seem to beaccepted all too blithely," he wrote,calling for "more fundamental thinkingon and questioning of the reality andutility of base theory as presentlyconceived" (1953a, 167).



While Andrews was somewhat critical inhis assessment of the economic baseparadigm, he clearly was a proponent ofits inherent validity and usefulness.Instead of suggesting the abandonmentof the model as a tool for urban andregional economic analysis, he identifiedways in which it could be improved toserve such purposes better. Hisrecommendation included better effortsat basic industry identification andmeasurement, improvements in thecollection of regional data, andmodifications in the way in whicheconomic base concepts were used.Given Andrews's criticism of the state ofthe economic base research prior to1950, it is surprising to note he did notaddress one of the most fundamentalshortcomings of this research: the lack ofempirical verification of the underlyinghypothesis. Krikelas (1991) identifiedonly five empirical tests of the economicbase hypothesis conducted during the1950s. Three of those studies, includingthat of Hildebrand and Mace, supportedthe validity of the economic basehypothesis, at least in the short run, andtwo provided evidence against it. Adecade of research, therefore, providedinsufficient empirical evidence fordetermining the validity of the model'scentral hypothesis.

When applied to analyzingregional growth, the economicbase model suggests that thegrowth process will be led byindustries that export goods andservices beyond regionalboundaries.

Instead, most of the debate of the 1950scentered on questions related to theory

and practice rather than testing. HansBlumenfeld (1955) was critical of theeconomic base model's narrow focus onexport activity as the primary source ofregional growth. While he agreed thatthis model might do well to explaineconomic growth in small or highlyspecialized economies, he argued that itwas inadequate to explain the growth ofcomplex urban economies. Blumenfeldwas also critical of the policyimplications of the model; these focusedalmost exclusively on supportingexisting export industries at the expenseof other reasonable alternatives, such asfostering the establishment anddevelopment of industries that wouldcompete with imported goods andservices.

Charles M. Tiebout (1956a, 1956b) andDouglass C. North (1955, 1956) engagedin a short but lively debate over theshort-run versus long-run applicability ofthe economic base model. Tiebout,explicitly recognizing the Keynesianroots of the economic base model,supported Hildebrand and Mace's (1950)contention that the economic base modelwas most appropriate for short-runeconomic analysis. He also argued thatthe economic base model minimized theimportant contribution that nonbasiceconomic activity made to local areagrowth and development. He wrote that,although export activity was important,"in terms of causation, the nature of theresidentiary industries will be a keyfactor in any possible development.Without the ability to developresidentiary activities, the cost ofdevelopment of export activities will beprohibitive" (1956a, 164).

North, however, objected to thecharacterization of the economic base



model as an adaptation of the demand-oriented Keynesian model. Instead, heargued that the most importantdeterminant of a region's long-rungrowth potential was its ability to attractcapital and labor into the region fromoutside. Such supply-enhancing flows inturn would respond quite favorably toprofit opportunities offered by regionsengaged in high levels of export activity.North observed that historically "it wasfrequently the opportunities inmanufacturing for the United Statesmarket which led to immigration oflabor and capital into a region. Theimportant point is that the pull ofeconomic opportunity as a result of acomparative advantage in producinggoods and services in demand in existingmarkets was the principal factor in thedifferential rates of growth of regions"(1956, 166).

Many regional scientists haveconcluded that economic basetheory lacks the complexity toprovide a useful framework foranalyzing many regional economicissues and policies.

The economic base model proposed byNorth explicitly recognized theimportant role of supply factors indetermining the nature and growthpotential of a region's export base. Inpractice, however, most economic basemodels of this and subsequent periodshave maintained a fairly strict demandorientation. This demand-oriented modelis also the one to which Tiebout raisedso many objections. As a result, althoughTiebout and North found themselves ondifferent sides concerning the validity ofthe model as a long-run theory of

regional growth, both ultimately agreedthat supply factors needed to be added tothe model in order to make it relevant forlong-run regional economic analysis.

One additional advance in the theoreticalliterature of this period that called intoquestion the adequacy of economic basemodeling techniques was thedevelopment of regional input-outputmodels. Before 1950 the economic basemodel represented the primary toolavailable to regional planners foranalyzing the impacts of anticipatedchanges in regional economic activity.During the first half of the 1950s,however, input-output modelingtechniques first developed by WassilyW. Leontief (1951) were adapted forpurposes of regional economic analysis.9

While a regional input-output modelcould distinguish between thedifferential regional impacts that mightbe associated with, for example, theconstruction of a specialty steelmanufacturer versus a mail-order catalogfacility—two very different kinds ofbasic economic activity—the simpletwo-sector economic base model couldnot make such a distinction. Given thislimitation, many urban planners began toadvocate input-output techniques asmore appropriate for forecastinganticipated changes in regionaleconomic activity.

The debate of the 1950s also focused onseveral important methodological issues.Papers by John M. Mattila and WilburR. Thompson (1955) and Charles L.Leven (1956) considered the adequacyof the location-quotient technique'sability to identify a region's economicbase industries. While suggesting certainimprovements to the traditionalformulation of the location quotient,



Mattila and Thompson concluded that "ifused with care, the index of surplusworkers in both its absolute and relativeform should prove to be a highly usefultool in regional economic base studies"(1955, 227).10 Leven, on the other hand,arrived at the opposite conclusion,stating that "the shortcomings of thistechnique render it useless as aquantitative measure of basic activity inan area" (1956, 256).

The issue of the appropriate measure tobe used for calculating location quotientswas also discussed. Becauseemployment data were more readilyavailable than wage or income data,most economic base studies of thisperiod used employment in identifyingregional export activity. This measure,however, has some serious drawbacks.In addition to placing equal weight uponpart-time and full-time employment andfailing to adjust adequately forproductivity and wage differencesbetween workers employed in differentindustries, employment data do notprovide any measure of the impact thattransfer payments and other sources ofunearned income, such as interestpayments, rents, and profits, have upon aregional economy.

Recognizing the serious weaknessesassociated with the use of employmentdata for purposes of identifying aregion's economic base, Andrews(1954a), Leven (1956) and Tiebout(1956c) all suggested the adoption ofalternative measures of regionaleconomic activity. Andrews and Tieboutadvocated the use of income received byresidents of the region, and Levenargued for a value-added measure.Income and value-added data, however,generally are not available for regional

economies, especially at the substatelevel, except with long lags.

By the beginning of the 1960sprofessionals engaged in urban andregional economic analysis had dividedinto three distinct camps concerning theconduct of research within the economicbase paradigm: those who stillconsidered the economic base model tobe a reasonable framework for urban andregional economic analysis; those whoquestioned its validity but sought moreempirical evidence before abandoningthe paradigm; and those who rejected thevalidity of the hypothesis, insteadturning to the investigation of othermethods of regional economic analysis,including regional input-output models.Whereas the debate of the 1950s wasconducted primarily at the theoreticallevel, the quarter-century between 1960and 1985 was filled with empiricalexaminations of a wide range oftheoretical and methodological questionsrelated to the economic base model.

Empirical Debate. Between 1960 and1985 a large number of articles andseveral books were published on theeconomic base model.11 Yet while thequestion of the empirical relevance ofthe economic base hypothesis wasarguably the most important issue facingthe profession on the heels of the debateof the 1950s, only a quarter of thesecontributions actually addressed it.To provide some perspective on theextensive literature of this period,Krikelas (1991) developed a taxonomy.The six categories listed representdistinct facets of the economic baseliterature of this period: (1) identificationof export base activity, (2) calibrationstudies, (3) extensions of the base model,(4) case studies, (5) theoretical works,



and (6) tests of the economic basehypothesis.

A thorough discussion of thecontributions that fall into each of thesecategories is beyond the scope of thisarticle. However, a summary of themajor developments in each categoryshould yield insights. It should be notedthat the majority of the researchpublished during this period—that is,categories (l)-(4)—assumed, at leastimplicitly, the validity of the economicbase hypothesis.

Identification of Export Base Activity.The most contentious issue facingresearchers using the economic basemodel is the identification of regionalexport activity. Much attention has beenpaid to the development of nonsurveytechniques, and during this periodseventeen studies were devoted tocreating new or improving oldmethodologies. Edward L. Ullman andMichael F. Dacey (1960) and Vijay K.Mathur and Harvey S. Rosen (1974)introduced two completely newnonsurvey methods for identifyingregional export activity, and severalother researchers suggested refinementsfor improving both the location-quotientand assignment methods of economicbase identification. Andrew M. Isserman(1980) offers an excellent survey of thedevelopments of this period, including acritique of each methodology.

Calibration Studies. Calibration studiesare research designed to test theadequacy of competing nonsurveyidentification techniques. Researcherseither compare nonsurvey estimates ofregional exports with benchmark surveyor census data on regional exports orsimply compare results of several

nonsurvey techniques. Anotherseventeen studies conducted between1960 and 1985 can be classified ascalibration studies, and Issermanprovides an excellent summary of suchresearch, concluding that althoughefforts to develop and refine thenonsurvey methods had been substantial,"the situation is lamentable" (1980, 179-79).

Extensions of the Base Model. Duringthis period at least two importantextensions were made to the simpleeconomic base model. In the first,additional variables other than basiceconomic activity were added to theoriginal specification in order toinvestigate their effects on the regionalgrowth process. Stanislaw Czamanski's(1965) study represents the first ofseveral in which a demographicvariable—population—was explicitlyincluded in the model specification. PaulE. Polzin (1977), on the other hand,developed a model designed to capturethe effects of local-area labor supplyconditions on regional economicactivity, and Ron E. Shaffer (1983) andShahin Shahidsaless, William Gillis, andShaffer (1983) included variablesdesigned to measure the contribution ofboth demographic and geographicfactors. Given the fact that these authorsgenerally found the additional variablesto be very important determinants ofregional growth, it is somewhatsurprising that relatively few studiesfocused on this issue.

A second innovation, which gained amuch broader acceptance in theliterature, was the disaggregation ofbasic activity into more than onesector—manufacturing, construction,services, and government, for example.



This work was stimulated by thechallenge posed by regional input-outputmodels and their clear demonstrationthat changes in regional activity indifferent export industries were likely tohave very different effects upon aregional economy. Steven J. Weiss andEdwin C. Gooding (1968) provide thefirst example of a multisectoraleconomic base model, and their workwas repeated and extended in manysubsequent studies. However, while theliterature of this period reported theresults of numerous multisectoraleconomic base models, the maximumnumber of sectors for which multiplierscan be estimated has always been limitedby the length of available data series,usually to ten sectors or fewer. As aresult, no economic base model has everbeen able to reproduce the level ofindustry disaggregation available in mostregional input-output models.

Case Studies. In most instances the mainpurpose of these base studies was thecalculation of multisectoral economicbase multipliers intended to demonstratethe significant impact of the sectorsunder consideration. Early studies hadfocused mainly on the role ofmanufacturing in the regional growthprocess. Many of these later works wereinstead devoted to showing theimportant contribution that the trade andservice sectors could also play inregional growth.12

Theoretical Works. Several contributionsduring this period were devotedexclusively to advancing the theoreticalfoundations of the economic baseparadigm. Edwin F. Terry (1965)explicitly derived the linkage betweenthe economic base model and theKeynesian model. John Mutti (1981), on

the other hand, demonstrated the closerelationship between economic base andinternational trade models. And finally,Wolfgang Mayer and Saul Pleeter(1975) and F.J.B. Stillwell and B.D.Boatwright (1971) developed economicbase theoretic models that demonstratedthat the location-quotient and minimum-requirements methods of export industryidentification could be derived from, andwere consistent with, economic basetheory. While these and othercontributions provided a formalstatement of the theoreticalunderpinnings of the economic basemodel and its methodologicaltechniques, they did not provideempirical evidence in support of thetheory's central hypothesis.

Tests of the Economic Base Hypothesis.In considering the empirical results ofstudies published during this period, it isimportant to distinguish betweendynamic and static tests of the economicbase hypothesis. Although the economicbase paradigm generally has been used,implicitly, to analyze dynamic regionaleconomic events, most specifications ofthe model, like that in equations (l)-(4),have been explicitly static in nature. Thispoint was made clear first by Charles E.Ferguson (1960). Subsequently, one ofthe major contributions of this periodwas the more explicit consideration ofthe dynamic properties of the economicbase model. Researchers began usingtime-series modeling and othereconometric techniques to analyze theshort-run versus long-run applicability ofthe economic base model as well as todevelop practical regional forecastingmodels.

The majority of these studies, however,were still predicated upon explicitly



static model specifications. Even someof the studies that ostensibly attemptedto capture the dynamic properties of theeconomic base model failed to do soadequately.13 Given that the utility of aneconomic base study depends upon itsuse for analyzing dynamic economicevents, it is unfortunate—andsurprising—that relatively few of theseempirical studies were specified in sucha way as to explore this issue.

In reviewing the literature of this period,Krikelas (1991) examined twenty-threestudies that reported the results of testsof the economic base hypothesis. Elevenwere static tests; twelve, dynamic. Ofthese, six static tests and seven dynamictests provided results consistent with theeconomic base hypothesis. Many of thedynamic tests of the hypothesis werefurther designed to explore the issue ofthe short-run versus long-run validity ofthe economic base hypothesis. Only fourstudies—Harold T. Moody and FrankW. Puffer (1970), Curtis Braschler(1972), Braschler and John A. Kuehn(1975), and James E. McNulty (I977)—provided any ostensible evidence insupport of economic base theory as along-run theory of regional growth.

As Shelby D. Gerking and Isserman(1981) have pointed out, however, themodel specifications adopted in three ofthese four studies actually tested only thecontemporaneous relationship betweenbasic and nonbasic economic activityrather than the long-run relationshippurportedly tested by the authors. Theyfurther concluded that Moody andPuffer's (1970) results, which werebased upon an appropriately specifieddynamic model, were more likely to beattributable to the authors' choice ofbifurcation methodology than to the

existence of a long-run economicrelationship between basic and nonbasicemployment. Thus, while a narrowmajority of the test results reportedduring this twenty-five-year periodprovided evidence in support of thevalidity of the economic basehypothesis, at least in the short run, verylittle empirical evidence suggested thatthe model could also perform well in thelong run.

By 1985 the most definite and positivecomment the literature could supportabout an economic base model was thatit would perform best in providingrelatively short-term forecasts of totalregional economic activity. More thanfifty years of research had failed toprovide any substantial evidence insupport of the model as a long-runtheory of regional growth—a seriouslimitation in light of the fact thatpolicymakers are generally moreinterested in long-run growth issues. Itshould be clear that the economic basemodel, because it fails to account forsome of the fundamental determinants ofthe regional growth process, should notbe adopted for long-range planning andpolicy analysis. These are the results thatled to Richardson's call (cited earlier) forburying economic base models "withoutprospects for resurrection" (1985, 646).

Third Period of Debate. DespiteRichardson's impassioned warning,research continues to be performedwithin the framework of the economicbase paradigm. Recently, a resurgence insuch research has been fueled by arecognition that some sophisticatedeconometric techniques used in analysisof macroeconomics time series may beapplied to the economic base model. Inparticular, it has been demonstrated that



the essential features of the economicbase model can be captured within thecontext of a bivariate vectorautoregression (VAR) linking basic andnonbasic economic activity.14 Oncespecified, such a VAR can be subjectedto the time-series econometric tests andanalytical procedures that have beendeveloped over the years. Grangercausality tests can be formulated in orderto test the validity of the economic basehypothesis. Impulse-response functions(the response of a variable to anunanticipated increase in other variables)can be derived and given a naturalinterpretation as dynamic basemultipliers. Forecasting competitionscan be held in order to assess how wellcompeting models improve the accuracyof a given forecast. Finally, co-integration tests can be performed inorder to assess whether there might be along-run relationship between basic andnonbasic economic activity.

Using such techniques, Lesage and Reed(1989) and Lesage (1990) foundempirical evidence in support of theeconomic base hypothesis. Lesage andReed reported Granger causality testresults that were generally consistentwith the economic base hypothesis, atleast in the short run. Proceeding further,the authors used their VAR modelspecifications to derive impulse-response functions describing thedynamic relationships between basic andnonbasic employment in eightmetropolitan statistical areas (MSAs) inOhio. The reasonable nature of themultipliers calculated from thisexperiment led the authors to concludethat this methodology offered promisefor regional economic forecasting andpolicy analysis purposes. When Lesage(1990) reported the results of co-

integration tests that demonstrated along-run economic relationship betweenbasic and nonbasic employment inseveral of these MSAs, the combinedresults of this research effort seemed toprovide evidence that such empiricalwork was both justified and could provefruitful.

A third period of debate on theeconomic base model centers onthe question of whether newtechniques borrowed frommacroeconomics time-seriesliterature can revive thetraditional economic base model.

The results of Lesage and Reed's (1989)and Lesage's (1990) studies are alreadybeing cited in the literature. David S.Kraybill and Jeffrey Dorfman (I992), forexample, used these authors'methodology to estimate a three-sectormodel for the state of Georgia. Theseand other recent contributions representexamples of what has become a thirdperiod of debate on the economic basemodel, centered on the question ofwhether new techniques borrowed frommacroeconomics time-series literaturecan revive the traditional economic basemodel.

Replicating and expanding this research,this author conducted extensive time-series econometric tests of the economicbase hypothesis on models specified forthe state of Wisconsin (Krikelas 1991).The results of this research, based upon alarge number of two-sector andmultisector model specifications, suggestthat these new techniques do not providethe convincing evidence to supportrevival of the economic base model for



purposes of long-term forecasting orplanning context.

First and foremost, the fundamentalproblems associated with derivingadequate estimates of regional exportactivity remain unresolved. AlthoughLesage and Reed (1989) claimed thattheir dynamic location-quotienttechnique "provides a more accuratedecomposition of local areaemployment" (1989, 616), this claimseems to be overstated. Krikelas (1991)confirms the results reported byIsserman (1980) and several others whohave found that the location-quotienttechnique tends to underestimate thelevel of regional export activity and,consequently, lend an upward bias toexport base multiplier estimates.

Second, in order to assess the stability ofmultiplier estimates derived from abivariate VAR, Krikelas (1991calculated impulse response functionsfor models that were based upon datagenerated from a variety of alternativesample separation techniques. Theresults of this experiment show thatsmall changes in the way in which agiven data set is divided into its basicand nonbasic components can lead tolarge changes in multiplier estimates.These results call into question theusefulness of the dynamic multipliersderived from a bivariate economic baseVAR for even short-run regional impactanalysis.

Finally, Krikelas (1991) explored thepossibility of deriving multipliers frommultisectoral VAR specifications andfound similar difficulties. As the numberof sectors included in a VAR isexpanded, establishing identifyingrestrictions required in order to derive

multiplier estimates becomes so arbitraryas to call into question the credibility ofthe multipliers derived from suchspecifications. As a result, any policyimplications that might be implicit in afinding of significant differencesbetween sectoral multiplier estimateswould also be questionable.

More fundamentally, however, Krikelasconcludes that the new techniquesemployed in Lesage and Reed andsimilar research do nothing to broadenthe economic base paradigm's focus onthe demand side of the regional growthequation. Past research has clearlyindicated that economic base models thatfail to account for important supply-sidefactors and constraints do not perform aswell as models that try to incorporatesuch relationships. Labor migrationpatterns, interregional capital flows, andstate and local tax policies all haveimportant effects upon regionaleconomic growth and development andneed to be incorporated into regionaleconomic model specifications for themodel to have value for anything otherthan short-term forecasting. Although itis possible to expand the bivariateeconomic base VAR to include some ofthese important supply-side variables,this author has concluded that suchresearch would be largely in vainbecause other problems would remain(see Krikelas 1991). The recent attemptto breathe new life into the economicbase model seems to have failed toresuscitate the patient.

ConclusionGiven the fact that several authors havebegun to report empirical results insupport of the validity of the economicbase hypothesis, a third round of debateon the model seems already under way



in the literature. An examination of someof the claims made by the proponents ofthese new dynamic economic basemodels, however, indicates that they areapparently unaware of the scope of theliterature preceding their efforts.15 Thisbrief analytical history should be

sufficient to convince users that theeconomic base model has severelimitations, especially for economicplanning and policy analysis, and to helpmake this next and perhaps final roundof debate a relatively short-lived one.

Notes1. Structural econometric models are often

used for purposes of both forecasting andpolicy analysis. However, the greatexpense required to specify and maintainsuch models has generally led economistseither to develop less complex models thatfocus narrowly on a small set of policyissues or to develop theoretical time-seriesmodels that perform well for purposes ofshort-run economic forecasting.

2. Besides the terms basic and nonbasic, anumber of others have been proposed todistinguish between the two types ofeconomic activity: town builders/townfillers, exchange production/ownproduction, primary/ancillary, export/local,as well as others. Andrews (1953b)directly addresses the issue of the profligateand confusing terminology of the economicbase paradigm.

3. Krumme was translating Werner Sombart'sDer Moderne Kapitalismus, Erster Band:Die Vorkapitalistische Wirtschaft, 2nd rev.ed. (Munich: Duncker and Humblot, 1916).Sombart identified the city formers as "aking who collects taxes; a landlord whoreceives rent payments; a merchant whoprofits from trade with outsiders; acraftsman, a manufacturer, who sellsindustrial products to the outside; anauthor, whose writings are being boughtoutside the gates; a physician, who hasclients in the countryside; a student, who issupported by his parents in another place,etc. These are the people who live and letlive."

4. Alexander was citing M. Arrousseau, "TheDistribution of Population: A ConstructiveProblem," Geographical Review 11 (1921).

5. Alexander cites a letter dated February 21,1921, to John M. Glenn, a member of theNew York Regional Planning Committee inwhich Olmsted wrote, "The multiplicity of

their productive occupations may beroughly divided into those which can beconsidered primary, such as carrying on themarine shipping business of the port andmanufacturing goods for general use (i.e.,not confined to use within the communityitself), and those occupations which may becalled ancillary, such as are devoteddirectly or indirectly to the service andconvenience of the people engaged in theprimary occupations."

6. According to Krumme's translation,Sombart wrote, "It is necessary to find outfor each trade how much of it is engaged inwork for local consumption and how muchin work for exports out of the city. Thisfigure then is the city-forming ratio for theindividual trade. Naturally, the ratio can befound accurately only with the assistance ofan extensive enquete (survey). However,one could gain at least an approximateimpression of the shares of the exportindustries in the total gainful employmentby a careful investigation of the results ofthe occupational census" (1968, 116). Theempirical study cited by Krumme waspublished for the first time in the secondrevised edition of Sombart's Der ModerneKapitalismus, Drifter Band: DasWirtschaftsleben im Zeitalterr desHochkapitalismus, in 1927. Krumme,however, was quoting from the thirdprinting of this edition, published in Berlinin 1955.

7. The following list identifies a few of thecommunities that performed economic basestudies during the 1940s, the individuals orinstitutions that performed these analyses,and the base ratios (,B) calculated,respectively: New York, The Regional PlanAssociation Inc., 2.1; Detroit, Detroit CityPlan Commission, 1.1; Cincinnati, VictorRoterus and the staff of Cincinnati CityPlanning Commission, 1.7; Washington,



D.C., National Capitol Park and Planning,1.1; Brockton, Massachusetts, Homer Hoyt,0.8; the state of New Jersey, Homer Hoyt,1. 1; and Albuquerque, New Mexico,Federal Reserve Bank of Kansas City, 0.9.This information was originally compiledby Edward Ullman and published in thethird edition of Weimer and Hoyt's text in1954 and was reprinted in Pfouts (1960,30).

8. Hildebrand and Mace wrote, "Theforthcoming Census of 1950 will permitfurther advances in this research.Recalculation of location quotients andcomparisons with 1940 will indicatechanges in external markets and locationalconcentrations during the war decade,particularly in communities undergoinglarge gains or losses in population. Withmonthly statistics of insured employment, acurrent record of employment in non-localized industries can be maintained.Improved multiplier analysis, with currentlocal labor force statistics, should thenpermit more precise depiction of localunemployment problems, and attainment ofmore adequate policies at the o r-all andcommunity levels" (1950, 249).

9. Perhaps the most often-cited contribution tothe early regional input-output literaturewas an article coauthored by Isard andKuenne (I 95 3).

10. The index of surplus workers is simply ameasure of the number of workers inexcess of that which would be required ifthe region's employment profile matchedthe national average.

11. Krikelas (1991) identified eighty-fourcontributions to the literature during thisperiod.

12. Some of the sectoral multiplier studiesconducted and the region or project for

which they were calculated, include thefollowing, respectively: retail trademultipliers calculated by Friedly (1965) forRedondo Beach, California; trade andservice sector multipliers calculated byTerry (1965) for St. Louis, Missouri;defense industry multipliers calculated byBillings (1970) for the state of Arizona andby Erickson (1977) for the BadgerAmmunition Plant, near Baraboo,Wisconsin; rural area multipliers calculatedby Garrison (1972) for fivenonmetropolitan counties in Kansas; anduniversity sector multipliers calculated byWilson (1977) for Tulsa, Oklahoma.

13. See Gerking and Isserman's (1981)discussion of the results of Braschler(1972), Braschler and Kuehn (1975), andMcNulty (1977).

14. A VAR model consists of an equation foreach variable in which the equations areestimated by regressing each of thevariables against lagged values of all thevariables. By not imposing any particulartheoretical connection among the variables,the VAR will capture any correlations thatexist in the data. In this sense, VARs aredistinct from traditional structural models,which typically include a large number ofvariables that are theoretically linked.

15. Lesage, for example, reported on one ofthe few empirical tests recorded in thehistory of the literature that supports theeconomic base hypothesis as a long-runtheory of regional growth and wrote that"this finding would not be particularlysurprising to most regional economists"(1990, 309). His is one of severalcomments published recently that havepointed toward the need for presentation ofa comprehensive history of the extensivebody of literature that exists.

References

Alexander, John W.,. "The Basic-Nonbasic Concept of Urban Economic Functions." Economic Geography30 (1954): 246-6 1.

Andrews, Richard B., "Mechanics of the Urban Economic Base: Historical Development of the BaseConcept." Land Economics 29 (1953a): 161-67.

______. "The Problem of Terminology." Land Economics 29 (1953b): 263-68.

.______. "A Classification of Base Types." Land Economics 29 (1953c):343-49.

______. "The Problem of Base Measurement." Land Economics 30 (1954a): 52-60.



______. "General Problems of Base Identification." Land Economics 30 (1954b): 164-72.

______. "Special Problems of Base Identification." Land Economics 30 (1954c): 260-69.

______. "The Problem of Base Area Delimitation." Land Economics 30 (1954d): 309-19.

______. "The Concept of The Base Ratios." Land Economics 31 (1955): 47-53.

______. "The Base Concept and the Planning Process." Land Economics 32 (1956): 69-84.

Billings, R. Bruce, "Regional Defense Impact-A Case Study Comparison of Measurement Techniques."Journal of Regional Science 10 (1970): 199-216.

Blumenfeld, Hans, "The Economic Base of the Metropolis." Journal of the American Institute of Planners21 (1955): 114-32.

Braschler, Curtis,. "A Comparison of Least-Squares Estimates of Regional Employment Multipliers withOther Methods." Journal of Regional Science 12 (1972): 457-68.

______., and John A. Kuehn. "Industry Sectors and the Export Base Determination of NonmetropolitanEmployment Change in Four Midwestern States." Review of Regional Studies 3 (1975): 82-89.

Czamanski, Stanislaw, "A Method of Forecasting Metropolitan Growth by Means of Distributed LagsAnalysis." Journal of Regional Science 6 (1965): 35-49.

Erickson, Rodney A., "Sub-Regional Impact Multipliers: Income Spread Effects from a Major DefenseInstallation." Economic Geography 53 (1977): 283-94.

Ferguson, Charles E., "Statics, Dynamics, and the Economic Base." In Techniques of Urban EconomicAnalysis, edited by Ralph W. Pfouts, 325-39. West Trenton, N.J.: Chandler and Davis PublishingCompany, 1960.

Friedly, Philip., "A Note on the Retail Trade Multiplier and Residential Mobility." Journal of RegionalScience 6 (1965): 57-63.

Garrison, Charles B., "The Impact of New Industry: An Application of the Economic Base Multiplier toSmall Rural Areas." Land Economics 48 (1972): 329-37.

Gerking, Shelby D., and Andrew M. Isserman, "Bifurcation and the Time Pattern of Impacts in theEconomic Base Model." Journal of Regional Science 21 (1981): 451-67.

Hildebrand, George, and Arthur Mace, "The Employment Multiplier in an Expanding Industrial Market:Los Angeles County, 1940-47." Review of Economics and Statistics 32 (1950): 241-49.

Isard, Walter, and Robert Kuenne, "The Impact of Steel upon the Greater New York-Philadelphia IndustrialRegion." Review of Economics and Statistics 35 (1953): 289-30 1.

Isserman, Andrew M., "Estimating Export Activity in a Regional Economy: A Theoretical and EmpiricalAnalysis of Alternative Methods." International Regional Science Review 5 (1980): 155-84.

Kraybill, David S., and Jeffrey Dorfman,. "A Dynamic Intersectoral Model of Regional EconomicGrowth." Journal of Regional Science 32 (1992): 1-17.

Krikelas, Andrew C., "Industry Structure and Regional Growth: A Vector Autoregression ForecastingModel of the Wisconsin Regional Economy." Ph.D. diss., University of Wisconsin-Madison, 1991.

Krumme, Gunter,. "Werner Sombart and the Economic Base Concept." Land Economics 48 (1968): 112-16.

Leontief, Wassily W. The Structure, of the American Economy. New York: Oxford University Press, 195 1.

Lesage, James P. "Forecasting Metropolitan Employment Using an Export-Base Error-Correction Model."Journal of Regional Science 30 (1990): 307-23.

______., and J. David Reed. "The Dynamic Relationship between Export, Local, and Total AreaEmployment." Regional Science and Urban Economics 19 (1989): 615-36.



Leven, Charles L. "Measuring the Economic Base." Papers and Proceedings of the Regional ScienceAssociation 2 (1956): 250-58.

Mathur, Vijay K., and Harvey S. Rosen. "Regional Employment Multiplier: A New Approach." LandEconomics 50 (1974): 93-96.

Mattila, John M., and Wilbur R. Thompson. "The Measurement of the Economic Base of the MetropolitanArea." Land Economics 31 (1955): 215-28.

Mayer, Wolfgang, and Saul Pleeter. "A Theoretical Justification for the Use of Location Quotients."Regional Science and Urban Economics 5 (1975): 343-55.

McNulty, James E. "A Test of the Time Dimension in Economic Base Analysis." Land Economics 53(1977): 359-68.

Moody, Harold T., and Frank W. Puffer. "The Empirical Verification of the Urban Base Multiplier:Traditional and Adjustment Process Models." Land Economics 46 (1970): 91-98.

Mutti, John. "Regional Analysis from the Standpoint of International Trade: Is It a Useful Perspective?"International Regional Science Review 6 (1981): 95-120.

North, Douglass C. "Location Theory and Regional Economic Growth." Journal of Political Economy 63(1955): 243-58.

______. "A Reply." Journal of Political Economy 64 (1956): 165-68.

"Oskaloosa vs. The United States." Fortune, April 1938, 58ff. Pfouts, Ralph W., ed. Techniques of UrbanEconomic Analysis. West Trenton, N.J.: Chandler and Davis Publishing Company, 1960.

Polzin, Paul E. "Urban Labor Markets: A Two-Sector Approach." Growth and Change 8 (1977): 11-15..

Richardson, Harry W. "Input-Output and Economic Base Multipliers: Looking Backward and Forward."Journal of Regional Science 25, no. 4 (1985): 607-61.

Shaffer, Ron E. "A Test of the Differences in Export Base Multipliers in Expanding and ContractingEconomies." Regional Science Perspectives 13 (1983): 61-74.

Shahidsaless, Shahin, William Gillis, and Ron Shaffer. "Community Characteristics and EmploymentMultipliers in Nonmetropolitan Counties, 1950-70." Land Economics 59 (1983): 84-93.

Stillwell, F.J.B., and B.D. Boatwright. "A Method of Estimating Interregional Trade Flows." Regional andUrban Economics 1 (1971): 77-87.

Terry, Edwin F. "Linear Estimators of the Export Employment Multiplier." Journal of Regional Science 6(1965): 17-34.

Tiebout, Charles M. "Exports and Regional Economic Growth." Journal of Political Economy 64 (1956a):160-64.

______. "Rejoinder." Journal of Political Economy 64 (1956b): 169.

______. "The Urban Economic Base Reconsidered." Land Economics 32 (1956c): 95-99.

Ullman, Edward L., and Michael F. Dacey. "The Minimum Requirements Approach to the UrbanEconomic Base." Papers and Proceedings of the Regional Science Association 6 (1960): 175-94.

Weimer, Arthur M., and Homer Hoyt. Principles of Real Estate. Rev. ed. New York: Roland PressCompany, 1948.

Weiss, Steven J., and Edwin C. Gooding. "Estimation of Differential Employment Multipliers in a SmallRegional Economy." Land Economics 44 (1968): 235-44.

Wilson, J. Holton. "Impact Analysis and Multiplier Specification." Growth and Change 8 (1977): 42-4.

Techniques for Analysis Note A

© WASchaffer 33 Draft 5/11/2010

APPENDIX B AN ECONOMIC-BASE MODELOF ATLANTA

The following article is an example of a simple economic-base multiplier, probably thelast computed for Atlanta. With modern data sources and computing equipment, suchmodels are seldom seen; but just a few decades ago they were quite common. Theirvirtue is the ease with which they are understood.



NOTE A. TECHNIQUES FOR DATAANALYSIS

Introduction

Now, with a few explanatory modelsunder our belts, let us take a look at acouple of techniques for examining data.The term “analysis” comes from theGreek term for “a breaking up.” I like tothink of it as a “laying out” of essentialfeatures for the better understanding ofsome phenomenon or thing. The processof analysis can involve a simple re-ordering of elements or it can involve acomplex statistical tool or maybe even amodel.

Sometimes, the simple approach is thebest and can yield insights with greatbenefits at small costs. Here, I wouldlike to re-visit our old friend the locationquotient and consider a new technique,shift-share analysis. Both of these toolshave their critics in academic circles, butboth can be used to suggest strengthsand weaknesses in an economy and bothcan point toward actions for developinga community.

Location quotients

A location quotient is commonlydefined as the ratio

LQi = (ei/e)/(Ei/E), a-1

where ei is area employment in industry

i, e is total employment in the area, Ei is

employment in the benchmark economyin industry i, and E is total employmentin the benchmark economy. Normally,the "benchmark" economy is taken to bethe nation as the closest available

approximation to a self-sufficienteconomy.

As pointed out in our discussion ofeconomic-base models, it is easilyconverted for use in estimating exportemployment. We simply assume that alocation quotient greater than one meansthat an industry produces more thanexpected in a self-sufficient economyand thus would be an export industry.On this assumption and locationquotients for a local economy, we builtan estimate of the economic basemultiplier for an economy.

But all of these computations fordetailed industries can lead to questionsas well as answers. The analyst couldproceed to ask why the local economyvaries from the benchmark economy.What resources are present or absent?What opportunities for importsubstitution are unexploited? What isthe apparent comparative advantagesignaled by a large location quotient?Could it be an impediment or anenhancement for future growth?

Obviously, this means that locationquotients are suggestive in nature -- theypoint toward further analysis, and theyforce us to proceed next to understandproduction processes and to explorecomparative advantage and ways tochange the local economy and promotegrowth.

The popularity of location quotientshas lead to a large number of variations.As Avrom Bendavid-Val points out inhis excellent practitioner’s book, the listof derived measures includes “...



coefficient of localization, coefficient ofspecialization, index of diversification,coefficient of redistribution, coefficientof geographic association, coefficient ofparticipation, index of occupationaldiscrimination, coefficient of deviation,friction ratio, and more. All of theseamount to little more than imaginativeapplications of the basic locationquotient technique, computing a ratio ofratios, in response to particular analyticneeds.” (Bendavid-Val 1983)

Advantages and disadvantages ofusing location quotients are discussed inChapter 3 along with sources of data.

Shift-share analysis

Sometimes, it may be helpful toincrease your general knowledge ofchange in the area in which you areconducting a regional impact analysis.For this task, “shift-share analysis” maybe appropriate. It is the most commontechnique for breaking economic changein areas into components is called “shift-share analysis.”1

Although originating in the 1940’s,the technique was introduced to frequentuse in 1960 by a team of economistsundertaking a massive study of regionsand economic growth (Perloff et al.1960). Since then, it has been extended,used, criticized and revived numeroustimes. As a projection technique, it hasbeen abandoned by all (including theU.S. Department of Commerce, astaunch supporter in the 1960’s) but themost faithful. As an expositorytechnique, it has enjoyed continued life.Economists have criticized it as merely a

1 A clear statement of this technique is contained in anessay by Charles F. Floyd included as an appendix to(Schaffer 1976). My comments rely on ProfessorFloyd’s work as well as appendix B in (Jackson et al.1981)

definitional manipulation of datacontaining no explanation of phenomena(that is, it is simply a way to organizedata) -- the explanation of change has tocome from further investigation.

But its major fault is that itsinterpretation relies heavily on the levelof aggregation of the data used.Nevertheless, it is a great andinexpensive way to start a review of theindustrial structure of an area.

Now, to see how this tool works, letus try various ways to estimate regionalemployment in year 2 for industry i(R2i), given knowledge of growth innational employment in industry i (N2i/N1i ), growth in total nationalemployment (N2./N1.), and evenknowledge of actual employment inindustry i in the region (R2i/R1i). Howcould we proceed?

There are four alternative estimatingtechniques consistent with shift-shareanalysis. First, we could assume thatemployment in industry i is the same inperiod 2 as it is in period 1:

R2i = R1i a-2

Here, we simply have neither knowledgenor hope for growth.

Second, we could assume that thelocal employment in industry i grows atthe same rate as does the nationaleconomy:

R2i = R1i*N2./N1. a-3

This, of course, is also a naîveassumption. In effect, we have assumedthat both the industry and the economyhave grown at the same rate, that localelements have each retained a constantshare of national growth (hence theterm).

Third, we could assume that the localemployment in industry i grows at the



same rate as does the employment inindustry i in the national economy:

R2i = R1i*N2i/N1i a-4

This is a little better. We can reasonablyassume that local industries are subjectto the same demand pressures felt bytheir competitors at the national level.

Fourth, we could assume that the localemployment in industry i grows at thesame rate as actually occurs (if wesomehow could know this!):

R2i = R1i*R2i/ R1i a-5

Now, what if we try subtraction tomake equations a-3 through a-5 intoexpressions of net change so that theycan be added? We can do this bysubtracting the right side of thepreceding equation from each andnaming the results.

The change in employment due tonational growth alone, assuming theregion gets its share, we call the nationalgrowth effect (NGi):

NGi = R1i*N2./N1 - R1i

NGi = R1i(N2./N1 - 1) a-6

The other two changes represent shiftsaway from the national trend. The firstis called the industry-mix effect (IMi)and shows the additional change due tothe growth characteristics of industry i:

IMi = R1i*N2i/N1i - R1i*N2./N1.

IMi = R1i(N2i/N1i - N2./N1.) a-7

The last change is called the regional-shift effect (RSi) and shows theadditional change due to the specificcharacteristics of the region itself:

RSi = R1i*R2i/ R1i - R1i*N2i/N1i

RSi = R1i(R2i/ R1i - N2i/N1i) a-8

Total change for each industry can besummarized as:

TCi = NGi + IMi + RSi a-9

These elements can then be summed toyield:

TC = TCi (R2 ii

i

R1i) a-10

Demonstration that all of these equationsfit together to form this identity is left asan exercise.

We should note that the terminologyused here, while it seems the mostcommon, is not unique. The techniquehas been used and reinterpretedfrequently, leading to renaming ofeffects. “Effects” are sometimes“components.” The “industry-mixeffect” has been the “structuralcomponent,” the “proportional shift,”and the “industry-composition effect;”and the “regional-share effect” has beencalled the “differential shift” and the“competitive effect.” (Floyd 1976)

Thoughts on writing an area profile

An area profile might be defined as aneconomic description which may lead toaction or insight. Profiles may havemany purposes. They may be intendedto inform potential investors, to attractvisitors, to inform citizens about thestructure of their community, etc. Thechallenge for an analyst is to break freeof the exclamatory hype associated withpromotional literature. You shouldorganize data to proclaim advantagesand accomplishments, to identifyproblems and opportunities, or tosuggest future policy, as appropriate.The following words and phrases arerandom points derived from a generalreading of published and unpublishedprofiles; I suggest a scanning ofcollections of brochures, of Federal



Reserve Economic Reviews, of theWorld Wide Web, etc. for furtherguidance:

What is the region? Describe itabsolutely and relative to otherplaces.

How is the region defined? If it is apolitical region, what is its economiccomponent. If it is a statistical region,are there any unusual features?

Geography and significant landscapefeatures, advantages, and limitations.

Demography: population, age andethnic composition if significant,education, etc.

Employment -- Have recent changesbeen due to dramatic changes forparticular industries, or have theyoccurred across the board?

Governments -- fragmented,consolidated, etc.

Subregions -- definitions andboundaries, uneven growth orpopulation distribution?

Economic base, before, now,expected. Location quotient analysis.How similar to the nation or state?Should it change? Where is themarket for its exports? Where shouldmajor imports originate? Are theseother markets stable or volatile (thatis, is the region subject to aninterregional business cycle)?

Will the economic base help or hinderin the future?

What are the prospects for improvingthe industry mix to get higherpersonal incomes, less pollution,whatever?

Which industries lead the region’sperformance? Do they contribute tocycles or to stability?

What is the reason for the region’sexistence? Why do people live there?

What will happen in the next two,three years -- expected newindustries or activities, departures,significant events?

Development policies stated byauthorities and in practice.

Problems

Trends in various indicators andperformance relative to nation orother regions.

Why should the region be optimisticor pessimistic about its future?(Remember that you are an objectiveanalyst, not a promoter.)

Elements to include in a locationquotient analysis

A location-quotient analysis should becarefully planned to take advantage ofmodern spreadsheets such as Microsoft’sExcel. The challenge is to lay out thesystem so that you can sort and resort onthe basis of the values of locationquotients and on calculated surplusemployment and back into industryorder as needed.

You should pay careful attention topresentation of data in tabular format:

Does the table title clearly identify theregion under study, the year, and thedata?

Are the columns clearly identified inthe caption?

Does the stub identify all rows andare sections specified and set apart?

Are columns wide enough for data?

Are decimals set consistently and atthe appropriate level for discussion?



In other words, does the presentationmeet the standards of goodscholarship and sound workmanship?

Have you presented only data that isrelevant, excluding items that areinconsequential?

The written part of the analysis is alsoimportant. Here are some points toconsider:

At a very basic level, do you define“location quotients” and how they areused?

Do you show awareness of your datasource, the level at which you work,and problems which may beassociated with your data?

Did you experiment with severalsources (e.g. both the RegionalEconomic Information System and

County Business Patterns) to see ifthey yield similar results?

Did you use both quotients andestimated excess employment?

Did you set up the tables and yourstatements such that the reader couldeasily check your results and evenmake their own conclusions?

Did you speculate on apparentlyquestionable results?

Did you explore changes over time?(Since REIS data is since 1969 andthe CBP data is available in yearlychunks on CD-ROM since 1986, wemight expect an alert and energeticanalyst to look at variation over timeeither to show trends or to identifydata problems.)

Documents

Economic-Base Theory Chapter 3 - USP · Economic-Base Theory Chapter 3 WASchaffer 1 Draft 5/11/2010 REGIONAL MODELS OF INCOME DETERMINATION: SIMPLE ECONOMIC-BASE THEORY Economic-base