Sectoral transformation and labour-market flows

57© 2000 OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED

SECTORAL TRANSFORMATION ANDLABOUR-MARKET FLOWS

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 16, NO. 3

DAVID GREENAWAYCentre for Research on Globalisation and Labour Markets, University of NottinghamRICHARD UPWARDUniversity of Nottingham and CEPRPETER WRIGHTUniversity of Nottingham and CEPR1

This paper examines the pattern of sectoral transformation that has occurred in the United Kingdom in thepost-war period and documents the flows of workers that have occurred between industrial and servicessectors and the non-employment that has resulted. It then examines what consequences sectoral transforma-tion has had for wages and unemployment, both at the aggregate and the individual level. It concludes byexamining the policy implications of its findings.

I. INTRODUCTION

A number of features of labour-market adjustmenthave been linked both to the globalization of produc-tion and technological change. The most widelydiscussed in the literature is the growing wageinequality between skilled and unskilled workerswhich has been evident in the USA and UK.Growing trade with low-wage economies and skill-biased technical change are seen as alternativecandidates for driving this ‘wage gap’ (see, forexample, Wood, 1998; Slaughter, 1999; Atkinson,

2001). Since not all OECD economies have experi-enced growing wage inequality, a further dimensionhas been potential links between trade and employ-ment change in general, and the impact of trade withlow-wage economies in particular (see, for in-stance, Greenaway et al., 1999). A further area ofinterest has been the impact of cross-border invest-ment on wage levels, wage distributions, and em-ployment patterns.

It is often assumed that the speed with whicheconomies must adjust to change has accelerated

1 The authors acknowledge helpful comments from three referees on an earlier draft of this paper. Financial assistance from theLeverhulme Trust under Programme Grant F114/BF and from the ESRC under award number L138251007 is gratefullyacknowledged.

58


owing to globalization and technological change. Inparticular, they are seen as responsible for morefrequent reallocations of labour across sectors,occupations, and regions. If this is, indeed, the case,then globalization/technological change may be re-sponsible for less stability in the work place, dis-placement of growing numbers of (especially un-skilled) workers from declining sectors, longer peri-ods of unemployment, and declining long-term earn-ings. But what evidence is there that the speed oflabour reallocation has actually increased, or that jobsecurity has fallen? Is it true that workers displacedfrom declining industries face long periods of unem-ployment and lower wages when they eventuallyfind work? In this paper we draw together someempirical evidence on the flows of workers be-tween sectors to shed light on some of these issues,partly because limited evidence on sectoral adjust-ment is available, partly because observed sectoraladjustment can shed light on a range of trade-relatedissues, including the ‘smooth adjustment hypoth-esis’. This states that adjustment processes will besmoother (and adjustment costs lower) when theeconomy is faced with increases in intra-industrytrade as opposed to inter-industry trade.2

Section II begins by examining the pattern of sectoraltransformation in the United Kingdom over the last50 years, and the flows of workers that haveresulted. Section III rationalizes the observed move-ments and asks what this implies for previousstudies of worker flows. Section IV then examines

the consequences of sectoral transformation forwages and unemployment at both the aggregate andindividual levels. Finally, section V examines policyimplications.

II. SECTORAL TRANSFORMATIONAND LABOUR-MARKET FLOWS

What do we mean by sectoral transformation, andwhat does this imply for flows of workers? In Figure1 there are two aggregate sectors, one declining (D)and one expanding (E). In addition, workers whocannot find work are unemployed (U).

Crudely, we associate sectoral transformation withthe long-run ‘decline’ of one sector (D) and ‘expan-sion’ of another (E). This will occur if, as a result ofsome exogenous change such as a trade shock,there are changes in profitable opportunities be-tween sectors, which manifest themselves in differ-ential rates of job creation and destruction. As aresult, equilibrium unemployment (U) expands ordeclines in the long run. This adjustment is poten-tially important from a policy perspective because ithas an impact on the welfare of those who changeemployment status or industry. In the short run theremay be unemployment if the adjustment process issticky and workers are unable to move directly fromD to E. These individuals may also face changes inemployment conditions when re-employed—for in-stance lower wages. Even within such a simple

U

D E

Figure 1Aggregate Sectors

2 Intra-industry trade refers to the simultaneous import and export of products from a given industry; inter-industry trade ariseswhen imports and exports originate in different industries. It has long been presumed that the former results in factor reallocationwithin rather than between industries, thereby resulting in smoother adjustment. See Greenaway and Torstensson (1997) for areview of the literature.

59

D. Greenaway, R. Upward, and P. Wright

framework, two points are apparent. First, if directmovement between sectors becomes more difficultfor whatever reason, then unemployment attribut-able to sectoral transformation increases. Second, ifthe rate of sectoral transformation increases, thenthe unemployment attributable to people being dis-placed from declining sectors will increase.

(i) Employment Change in the UK, 1950–2000

The pattern of employment in the UK has changedmarkedly since the Second World War. The toppanel of Table 1 sets out labour reallocation acrossseven broad sectors between 1950 and 2000. Theproportion employed in distribution and services hasmore than doubled, and now accounts for 70 percent of the work-force. Manufacturing, by contrast,now provides only 16 per cent of employmentcompared to nearly 40 per cent in 1950. The bottompanel shows how manufacturing employment de-clined faster in the 1970s and 1980s than in other

decades, and that the speed of decline has halved inthe last decade. Similarly, the increase in the size ofthe service sector accelerated in every decade up tothe 1990s. Rowthorn (2000) describes this processas ‘de-industrialization’.3

The rate of restructuring through time can be sum-marized via a ‘turbulence index’,

which provides a measure of the rate of change ofindustry employment shares. These are plotted inFigure 2, which indicates that the 1970s and 1980ssaw greater sectoral employment reallocation thanany decade since the war. In the 1990s, the rate ofrestructuring decelerated rapidly and returned tolevels similar to those that prevailed in the 1960s.4

There are two important caveats in the calculationof such indices. First, since they measure absolute

Table 1Changing Employment Shares in the UK, 1950–2000

% of employees in employment 1950 1960 1970 1980 1990a 2000a,b

Agriculture, forestry, and fishing 5.60 4.10 1.74 1.57 1.37 1.27Mining, supply of electricity, gas, and water 5.16 4.73 3.68 3.19 1.74 0.86Manufacturing 38.02 37.66 38.69 30.28 20.52 16.52Transport, storage, and communication 8.00 6.97 6.94 6.52 6.07 6.09Construction 6.66 6.51 5.88 5.37 5.36 4.73Wholesale and retail distribution 12.74 13.88 12.08 14.61 15.79 17.04Services 23.82 26.16 30.98 38.47 49.15 53.50

Change in % over decade 1950s 1960s 1970s 1980s 1990s

Agriculture, forestry, and fishing –1.50 –2.35 –0.18 –0.20 –0.10Mining, supply of electricity, gas, and water –0.44 –1.04 –0.50 –1.44 –0.88Manufacturing –0.36 1.03 –8.41 –9.76 –4.00Transport, storage, and communication –1.03 –0.03 –0.42 –0.45 0.02Construction –0.15 –0.62 –0.51 –0.01 –0.64Wholesale and retail distribution 1.14 –1.80 2.53 1.18 1.25Services 2.34 4.82 7.48 10.68 4.35

Notes: a Figures for 1990 and 2000 refer to UK; earlier years refer to GB. b December 1999.Sources: Ministry of Labour Gazette, Department of Employment Gazette, Employment Gazette, andLabour Market Trends (various years).

3 Rowthorn (2000) points out that the relative decline of manufactures is not peculiar to, but is particularly marked in, the UK.4 Abraham (1991) notes that since the indices are based on realized rather than desired changes in employment, they might fail

to capture the full extent of reallocation shocks. Information on vacancies may be used to assess the extent of desired demand.

( )∑ ∆ NNi21

60


Figure 2Turbulence Indices for the United Kingdom

changes in employment shares, an expansion of asector followed by an immediate contraction of thesame size will be picked up as two periods ofturbulence. Thus the index may pick up relativeexpansion and contraction of sectors over the busi-ness cycle. This can be ameliorated by extendingthe time period over which the difference is taken.For instance, by considering a decade turbulenceindex

any cyclical expansions and contractions whichoccur within the decade will not be recorded. Sec-ond, the level of industrial aggregation matters. Athigher levels of aggregation movements of employ-ment within a sector will not affect the index. Atlower levels these will be picked up as movementsbetween more disaggregated sectors.5

The influence of these elements can clearly be seenin Figure 2. The turbulence index based on 24sectors is always higher than that based on seven.The divergence of the two lines in the earlier periodsis therefore caused by the fact that in the 1950s and1960s most adjustment was between manufactur-ing sectors, which the 24-sector index is picking up

but the seven-sector index misses. In the 1970s and1980s on the other hand, embracing the period ofmost acute de-industrialization, most adjustment isdirectly from manufacturing to services, which bothindices pick up. A comparison of a decade-basedturbulence index with a 10-year moving average ofthe annual turbulence indices also indicates therelative importance of short- and long-run factors.From 1950 to 1980 these indices diverge periodi-cally, since temporary shocks are an importantcause of sectoral employment change. From 1980to 2000 the indices move more closely togethersince, over this period, employment is consistentlymoving from manufacturing to services and reverseflows are rare.

(ii) Gross Worker Flows and SectoralTransformation

Although the mechanism for sectoral transforma-tion is differential job creation and destruction, aconsequence is flows of workers from one sector toanother. Intuitively, one might imagine movementsof labour directly from declining to expanding sec-tors. That is, declining sectors have high outflow andexpanding sectors high inflow rates. Further, if thereis frictional unemployment, some flows might occur

( )∑ ∆ NNi1021

5 Abraham (1991) also notes that changes in the skill mix of workers required within an industry may occur without affectingthe industry’s share of employment and hence the index.

Year1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000

0

.005

.01

.015

24 sectors (10-year moving average)7 sectors (10-year moving average)7 sectors (10-year changes)

TurbulenceIndex

61


with an intervening spell of unemployment. Figure36 plots gross flows between declining and expand-ing sectors. It is apparent that these are pro-cyclicaland highest during booms, which suggests sector tosector flows are dominated by voluntary moves,with individuals moving when times are good ratherthan bad.7 Another interesting feature is that manyflows are not from declining to expanding sectors.While between 6 per cent and 11 per cent ofindividuals change firms each year, only 2–3 percent switch from the declining to the expandingsector or vice versa.8 There is, therefore, consider-able intra-sectoral movement of labour, which isconsistent with the observation on US firm-leveldata that much job creation and destruction occurswithin narrowly defined industries (Davis andHaltiwanger, 1992).

Some have argued that, for the USA, gross flowshave declined secularly over the 1970s and 1980s asworkers have become less mobile and that this hascoincided with an increase in unemployment

(Murphy and Topel, 1987; Jovanovic and Moffitt,1990). Indeed, Jacoby (1983) argues that job mobil-ity has been declining throughout the twentiethcentury. One explanation for this is that workerswith high levels of sector-specific human capital areunwilling to change sector, even at the expense oflonger periods of unemployment (Thomas, 1996).However, as is evident from Figure 3, the apparentdecline in the 1970s and early 1980s was stronglyreversed in the late 1980s as the UK economyrecovered. Moreover, the link between changes ingross flow rates and the level of unemployment isfar from direct, and more subtle than the abovecharacterization suggests. It is particularly impor-tant to realize, for instance, that flows other thanthose indicated on Figure 1 are important—intra-sectoral flows (which we have already mentioned)and reverse flows from unemployment and theexpanding to declining sector are also important.

Figure 4 shows job-to-job inflow and outflow ratesbetween ten sectors of economic activity in the UK

6 The data source used in this study is the UK Labour Force Survey (LFS) from 1975 to 1995. This is an annual (biennial from1975 to 1983; quarterly from 1992 onwards) survey of 60,000 households comprising about 120,000 adults. In every year of thesurvey, individuals are asked about their current labour-force status (working, unemployed, out of the labour force) and their currentindustry, if employed. Individuals are also asked about their status and industry 12 months previously.

7 It is well known from the literature of flows of workers between employment status that quits are pro-cyclical (Blanchard andDiamond, 1990; Burda and Wyplosz, 1994).

8 Clearly, the size of gross flows depends partly on the dimension of the gross-flow matrix, G: flows are greater if there are moreindustries. Sectors are classified as expanding or declining depending on whether they experienced increases or decreases inemployment shares over the period 1975–95.

Figure 3Gross Flows between Sectors, 1975–95

1-digit industries

and expanding sectorsGross flows between declining

Gross flows between 10

Gross flows between firms

Prop. gross flows

Year75 77 79 81 83 85 87 89 91 93 95

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

62


for each year from 1975 to 1995. The 45° lineequates inflow and outflow rates..Two features stand out. First, inflows and outflowsare of similar size. Second, they are positivelycorrelated.9 Thus, although a considerable numberwho leave declining sectors enter expanding sectorseach year, an almost equal number of people movein the opposite direction.

(iii) Net Flows of Workers

The fact that inflows and outflows are of a similarsize implies that gross flows between sectors arelarger than net flows. To see this, consider the grossflows of workers between declining and expandingsectors for 1993–4 (Table 2). The rows give thenumbers in each aggregate sector in 1993; thecolumns give the same numbers for the followingyear. Total gross flows are simply the number ofindividuals in a different sector in 1994 than 1993,and can be calculated as the sum of the off-diagonalelements of Table 2. Total net flows are the sum offlows not cancelled out by return flows. For exam-ple, 212,538 individuals move from the declining tothe expanding sector between 1993 and 1994. How-

ever, almost as many, 203,213, move from theexpanding to the declining sector over the sameperiod. Net flows between the declining and ex-panding sectors in this case are given by the differ-ence between these two flows.

Figure 5 plots net flows between sectors over timefor a number of aggregations.10 Several points arenoteworthy. First, gross flows are approximatelyten times greater than net flows. Only about one-fifth of 1 per cent of the labour force moves betweendeclining and expanding sectors in a way thatcontributes to sectoral adjustment. However, thechanges in sectoral employment during this period,shown in Table 1, would require net flows of greaterthan 1 per cent per year. It is clear, therefore, thatdirect job-to-job sectoral flows cannot account forthis adjustment. Second, the pattern of net flows isless obviously cyclical than gross flows. Indeed, thepeaks in net flows occur in recessions when lay-offsare prominent, which suggests that involuntary movesmay have a significant influence over net flow rates.

Figure 6 plots gross and net flows between employ-ment and non-employment.11 Note that net flowsare much larger than net flows between sectors, and

Figure 4Inflow and Outflow Rates by 1-digit Standard Industrial Classification, 1975–95

9 This relationship might occur if some sectors have high turnover while others have low turnover, and not because inflows andoutflows are positively correlated. Although some sectors do have higher turnover than others, OLS regressions of inflows onoutflows separately for each sector reveal a significant positive correlation within sectors for nine of the ten sectors.

10 As with gross flows, net job-to-job flows increase with the number of sectors.11 Non-employment is defined to include both unemployment and ‘not in the labour force’ (NILF). This grouping is necessary

because a proportion of individuals who classify themselves as NILF do in fact move into and out of employment.

Inflow rate

Outflow rate0.00 0.02 0.04 0.06 0.08 0.10

0.00

0.02

0.04

0.06

0.08

0.10

0

4

95

3

61

27

8

7

46

9

3

1

2

0

85

9

2

54

8

0

6

7

13

1

3

2

0

45

8

9

6

7

5

8

6

3

4

2

1

0

7

9

0

8

9

5 62

7

3

1

4 6

8

5

1

2

9

3

0

7 4

9

8

3

6

0

5

42

1

7

8

6

9 1

5

7

0

2

34

6

9

2

7

5

8

01

4

3

7

2

5

0

3

4

19

6

84

72

9

5

1

63

0

8

7

0

68

5

2

9

4

1

3

9

1 54

3

8 2

6

0

7

1

6

8

2 57

9

3

4

0

9

1

6

8

2

05

3

4

7

2

0

9

47

1

83

5

6

63


Table 2Gross Flow Matrix, UK, 1993–4

Aggregate sector at t

Declining Expanding Non-employment Total

Declining 7,665,379 212,538 571,264 8,449,181Row % 90.72 2.52 6.76 100.00Column % 90.94 1.52 6.54 27.15

Expanding 203,213 12,596,283 857,604 13,657,100Row % 1.49 92.23 6.28 100.00Column % 2.41 90.28 9.82 43.89

Non-employment 560,208 1,143,281 7,307,549 9,011,038Row % 6.22 12.69 81.1 100.00Column % 6.65 8.19 83.64 28.96

Total 8,428,800 13,952,102 8,736,417 31,117,319Row % 27.09 44.84 28.08 100.00Column % 100.00 100.00 100.00 100.00

Agg

rega

te s

ecto

r at

t–1

Figure 5Net Flow Rates Between Sectors

nearly half as big as gross flows. This is becauseflows from employment to non-employment arecounter-cyclical, while flows in the reverse direc-tion to employment are pro-cyclical. There is there-fore less of a tendency for them to cancel each otherout.

(iv) Implications for the Measurement ofSectoral Transformation

What does this imply about appropriate measuresfor assessing sectoral transformation, i.e. differen-tial job creation and destruction between sectors?

and declining industriesNet flows between expanding

Net flows between 10 1-digit industries

Prop. net flow

75 77 79 81 83 85 87 89 91 93 95

0.0000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

Year

64


Figure 6Gross and Net Flow Rates Between Employment and Non-employment

Consider a simple example. Suppose that the rate ofsectoral transformation between 1994 and 1995 hadbeen higher, with an additional 100,000 jobs de-stroyed in the declining sector and created in theexpanding sector. Such an increase in net job crea-tion would cause increases in net worker flows andin net employment changes. Net worker flows arenot identical to net employment changes becausethey are greater if adjustment comes about viaunemployment, U. However, net worker flows areuseful because they allow us to decompose adjust-ment into job-to-job and job-to-unemployment flows.What is surprising, however, is that gross workerflows might actually decrease in response to sectoraltransformation. A given rate of job creation anddestruction may be accommodated by widely vary-ing gross flow rates, because the balanced flowsmay increase or decrease with no impact on adjust-ment. Thus gross flows are not particularly informa-tive and the argument that decreasing engagementsand separations in manufacturing (turnover) imply adecreasing rate of sectoral transformation is falla-cious. Because workers may move between pre-existing jobs and in equal and opposite directions,there is no necessary relationship between grossflows and sectoral transformation.

Similarly, increased job instability does not neces-sarily indicate increased rates of sectoral transfor-mation but an increase in gross flow rates. So what,

then, are gross flows a useful measure of? They area measure of the average ease with which peoplecan move (or are forced to move) between jobs.Gross flow rates are therefore informative as ameasure of the ‘flexibility’ of the labour market interms of the ease with which employers can adjustlabour demand and employees can adjust laboursupply. Blanchflower and Freeman (1994), for ex-ample, use estimates of transition rates betweenemployment, unemployment, and out of the labourforce as a measure of the increased flexibility of theUK labour market over the 1980s.

Despite the widespread belief that labour-adjust-ment costs have fallen in the UK over this period,there is very little evidence of this in our transitionrates. The gross flows of workers between jobsappear pro-cyclical with little trend. This is sup-ported by the consensus view that job turnover hasnot declined significantly over the last 20 years. Inan accounting sense, the increase in unemploymentover the 1980s was due to decreased outflow ratesfrom unemployment rather than increased inflowrates or job turnover (e.g. Nickell, 1999).

(v) Changes in Occupational Structure

A major focus of the recent labour economicsliterature in the United Kingdom has been changingoccupational structure (e.g. Machin and Van Reenen,

Net flows between employment and non-employment

Gross flows between employment and non-employment

Prop. flow

Year75 77 79 81 83 85 87 89 91 93 95

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

65


1998). If expanding sectors are more skill intensivethan declining sectors, this may be driving skillupgrading. Evidence would seem to suggest that this isonly partially true. In general, skill upgrading appearsto be a more general phenomenon in the economy.Many argue that skill-biased technological change iscausing skill upgrading to occur even within nar-rowly defined sectors. Hence we would expect toobserve this process even within declining sectors.

To summarize, the rate of sectoral transformationaccelerated through the 1970s and 1980s and be-came increasingly dominated by movements ofemployment from manufacturing to services. Fur-ther, flows of workers between sectors do notcorrespond to a simple characterization that indi-viduals move from declining to expanding sectors.Intra-sectoral movements are common and manyindividuals move from expanding to declining sec-tors. Thus gross flows dominate net flows. Whilethere have been large shifts in employment betweensectors, only a small proportion of the adjustment ismade up of individuals moving directly from onesector to another, with flows into and out of non-employment playing a large part in the adjustmentprocess.

III. WHY DO INDIVIDUALS MOVESECTOR?

Intra-sectoral worker flows are consistent withmodels of job creation and destruction formulatedby Mortensen and Pissarides (1994), in which firmsexperience persistent idiosyncratic shocks and hencewhat is profitable for one firm may not be foranother. This leads to simultaneous job creation anddestruction even within narrowly defined industries,as evidenced by Davis and Haltiwanger (1992).Given this pattern, how do we rationalize observedpatterns of gross and net worker flows? Matchingmodels such as Jovanovic (1979) suggest the vastmajority of worker movements can be explained ina framework where firms and workers are search-ing for their most suitable job match.12 In models ofthis kind, the wage of worker i, wist, is given by:

wist = m.pst fst' (x, zst) = mwst.

That is, wages depend on: the marginal product oflabour in sector s at time t (measured in efficiencyunits, x); the output price in sector s at time t (pst);sector-specific factors (zst); and the quality of thematch between the worker and employer (m). It isassumed that the quality of the match is not knownprior to hiring. Following its realization, poorly suitedindividuals get a low wage and well-suited individu-als a high wage. Individual wages determine attitudeto the job, and poorly suited individuals will seekmore profitable alternatives. An individual may wellmove to a declining sector if their anticipated matchquality is high, since this may more than outweigh alower wage per efficiency unit. Hence we willobserve gross flows both into and out of the declin-ing sector, even in equilibrium. The total number ofgross flows depends on the cost of moving sectorand the spread of matching returns, because thelower the cost the greater the incentive to seek thehighest possible match, and the greater the spreadthe greater the likelihood of an individual receivinga poor match. Net flows between sectors occur aslong as there is a difference in expected net returns,perhaps because of differences in the wages paidper efficiency unit or, if unemployment is present, inthe expected probability of job offers.

What happens if there is a shock that impacts onsectors differentially? Suppose there is a decline inoutput price (pst) or an unfavourable movement in zstfor the declining sector. This causes a decrease inthe wage the firm is willing to pay per efficiency unitwhich can have several consequences. First, anumber of workers who were previously happy withtheir match in the declining sector will seek jobs inthe expanding sector. Likewise, those who wouldhave moved in the opposite direction now find thisunattractive. Second, if there is some institutionallimit on how far the wage can fall, firms lay offworkers. Hence there will be an increase in netflows from the declining to the expanding sectorsand possibly also into non-employment.

These models imply that authors such as Lilien(1982) may be misinterpreting the impact of demandshifts on worker flows, since many observed jobchanges may be occurring for matching reasons,rather than because the sector in which an individual

12 In addition, people may move within industries as part of the process of career development (Sicherman and Galor, 1990; Booth,1997).

66


is employed has been subject to a shock. Thisdifficulty also bedevils empirical work which fo-cuses on individual ‘displaced workers’ (e.g. Mincer,1986; Kletzer, 1996), where identifying those whosejob moves are enforced by sectoral shocks is simi-larly problematic.

IV. CONSEQUENCES OFADJUSTMENT

(i) Sectoral Adjustment and Wages

An important consequence of sectoral change is thelikely change in the structure of wages. Nickell(1996) argues that the movement of employmentfrom production to service sectors was not accom-panied by dramatic changes in the average realwage set in each sector. He argues, ‘if there havebeen demand shifts which have led to significantdislocation in the economy, these are related tooccupation rather than industry’ (p. 7). However,the simple model presented in section III suggeststhat examining average wages within sectors ispotentially misleading. Favourable shocks will leadto expanding sectors paying higher wages per effi-ciency unit than declining sectors. This will encour-age net movement to the expanding sector since thenumber of individuals who anticipate higher wagesfrom such a move will increase. The average wageof movers will therefore be higher than the wagethey received previously. However, those with goodquality matches in their existing job will be leastlikely to move. Therefore the average quality ofmatch in the declining sector will increase. Indeed,as Jovanovic and Moffitt (1990, p. 838) argue

a contracting sector . . . will therefore have higher wagesthan an expanding sector. The scenario above more orless describes US experience over the past 15 years or so:The manufacturing sector has shrunk while services haveexpanded, but manufacturing wages have tended toexceed those in the service sector.

Turning to individuals, a key impact of sectoraltransformation is that workers involuntarily dis-placed from their jobs are likely to suffer wagelosses, which will tend to be higher for more seniorworkers.13 But why should workers suffer wage

losses when displaced? A large body of evidencesuggests workers accumulate firm-specific humancapital. In empirical wage equations this manifestsitself in terms of returns to tenure within a firm,14 aswell as the wage falls that displaced workers expe-rience. General human capital that is not specific toa particular job, also accumulates, and this explainsthe positive relationship between wages and totallabour-market experience. It also seems likely thatsome element of human capital may be occupationand industry specific.

Clearly this has implications for workers forced tomove as a result of sectoral transformation, sinceworkers will lose returns to current industry andmaybe also occupational status. As well as being ofimportance to the individual, the extent to whichskills are industry specific is clearly of interest indetermining the cost of aggregate adjustment sincethis will determine the ease with which workersmove from one industry to another. Indeed, the viewthat these costs may be substantial underpins thesmooth-adjustment hypothesis.

For the USA, Neal (1995) finds that workers cantransfer skills acquired in one firm to another in thesame sector, suggesting that industry-level skills areimportant. Workers who change industry, on theother hand, suffer wage losses, as they are notrewarded for their (now) redundant skills. For theUK, this is investigated by Haynes et al. (1999),who show that crude returns to industry tenureappear to be large. However, once the correlationsbetween the measures of tenure and unobservedmatch-specific components of the wage are con-trolled for, returns to industry tenure are muchsmaller than returns to occupational tenure, whichimplies that workers moving between industriessuffer no greater wage losses than workers movingwithin industries, provided they remain in the sameoccupation. Of course, as Table 3 shows, workersmoving between industries are more likely to moveoccupation as well.

Further, Haynes et al. also find that returns to jobtenure are much smaller than returns to firm tenure,which is the usual measure in the literature. This isunsurprising, since a ‘job’ may be associated with a

13 Evidence for this comes from the large (mainly US) literature on displacement. Kletzer (1998) provides a summary.14 Returns to tenure are also consistent with a number of other theories of worker compensation, such as screening or signalling

theories (Weiss, 1995).

67


Table 3The Probability of Job, Industry, and Occupational Change

Annual probability

New job 0.176New industry 0.111New occupation 0.110

Same job, same industry, same occupation 0.775Same job, same industry, new occupation 0.017Same job, new industry, same occupation 0.029Same job, new industry, new occupation 0.004New job, same industry, same occupation 0.058New job, same industry, new occupation 0.039New job, new industry, same occupation 0.027New job, new industry, new occupation 0.051

particular nominal wage, and so longer tenure maylead to a declining real wage if individuals get ‘stuck’in a job. In a recent study of returns to tenure, Altonjiand Williams (1997) suggest the best estimate forreturns to 10 years firm tenure is about 0.11. Theresults of Haynes et al. (1999) suggest that it is notfirm tenure itself which causes this increase, butoccupational, and to a lesser extent, industry tenure.

(ii) Sectoral Adjustment and Unemployment

As we have seen, an important consequence ofsectoral transformation is that a substantial propor-tion of those displaced are unable to move directlyfrom contracting to expanding sectors. Hencesectoral transformation and re-allocation may be animportant source of aggregate unemployment. In-deed, aggregate unemployment may increase if therate of change of sectoral transformation has risenor if individuals have become less mobile betweensectors for a given level of adjustment. For the USA,Lilien (1982) finds that inter-sectoral shocks werepositively correlated with US unemployment. Hismethodology has been questioned: Abraham andKatz (1986) point out that, if manufacturing employ-ment is more cyclical than services, then the disper-sion of employment growth rates may increaseanyway during slumps, even without any permanentreallocation of labour.15 Hence, a positive correla-

tion between the variance of employment growthand unemployment is not necessarily evidence forthe impact of restructuring. More recent studies—Loungani et al. (1990), Brainard and Cutler (1993),Mills et al. (1995)—have sought to remedy this andhave generally been supportive of the ‘sectoral shifthypothesis’ that inter-sectoral shocks are an impor-tant source of fluctuations in the unemploymentrate.

Unemployment spells also represent an importantcost of sectoral transformation to individuals. Anexamination of the movement of workers into andout of unemployment can also shed light on theproposition that labour-market adjustments to intra-industry trade are less costly in terms of dislocationthan adjustments to inter-industry trade. Previousliterature here is rare,16 though Murphy and Topel(1987) and Fallick (1993) provide evidence thatindividuals who change industry (‘movers’) tend tohave longer unemployment durations than thosewho return to the same industry (‘stayers’). Theysuggest that the greater wage losses of moversmean that individuals are prepared to stay unem-ployed for longer to return to their original sector andavoid losing returns to sector-specific skills. Thegreater the sector-specificity of skills the greater thepersistence of induced unemployment. This hasbeen tested on Canadian data by Thomas (1996),

15 As discussed earlier, this methodology picks up unemployment due to cyclical impacts as well as structural impacts becauseof its use of an annual turbulence index.

16 For a recent survey, see Matusz and Tarr (2001).

68


who finds that the link between increased aggregateunemployment and increased immobility of labour isrelatively weak.

Haynes et al. (2000) compare unemploymentdurations of those who find work in the sector inwhich they were originally employed, and those whofind work in a new sector. They also examine whatpersonal circumstances affect the probability ofindividual movement and the duration of unemploy-ment spells. Table 4 shows that individuals in theUSA experience a higher incidence of unemploy-ment than in the UK. Further a larger proportion ofspells in the USA end in a return to the same sector:46.5 per cent compared to 20.4 per cent. A corre-spondingly higher proportion of UK spells thereforeend in a movement to a new sector.17

One notable difference in the US data is that 13.6per cent of spells are coded as ‘temporarily laid off’and it would be expected that such individuals aremore likely to return to their previous employer, andtherefore remain in the same sector. This phenom-enon is rare in the UK, and, indeed, is not recognizedas an explicit category in the data.

Table 5 shows the probability of moving to particularstates from employment. The average duration ofspells in the USA is shorter with, for both countries,the duration being shortest for those spells ending ina return to the same industry.18

It is important to note that the use of raw data ispotentially misleading, however, since an individualwho is unemployed for a long time, but finds a job ina new sector, could have taken even longer to finda job in the same sector. That is, one outcome‘censors’ the other. To allow for this Haynes et al.(2000) adopt a competing risk model to allow for thepossibility of multiple exit states from unemploy-ment. They find that the longer mean durationobserved in the UK is not the result of a less sharplydeclining unemployment hazard (which means theunemployed find it increasingly hard to find a job), somuch as a lower overall hazard in the UK (whichmeans that they are always less likely to exitunemployment). They also find that the hazard tostaying in the same sector declines faster than thehazard to finding a job in a new sector in bothcountries. This suggests that individuals are morelikely to switch sector the longer they are unem-ployed in both countries. A plausible explanation forthis is that individuals initially attempt to find jobs thatcomplement their general and specific skills, butmove sector as this prospect diminishes. This isparticularly the case for workers who would beexpected to have higher levels of sector-specificskills (older workers, for example), which is consist-ent with the hypothesis that finding a job in theoriginal sector is less costly than finding a job in anew sector, at least for shorter unemploymentdurations. Indeed, even if it were the case thatrewards in the new sector were higher, their results

Table 4Characteristics of Unemployment in the UK and the USA

UK USA

Annual incidence of unemployment (spells/year) 0.264 0.348

Exit into job, of which 0.567 0.703(a) Exit into same industry 0.204 0.465(b) Exit into new industry 0.363 0.238

Temporarily laid off — 0.136

Source: Haynes et al. (2000).

17 Note also that the proportion of spells that are censored is higher in the UK (0.433 as opposed to 0.297). This occurs becausethe average duration of spells in the UK is longer.

18 It is longest for those which do not end before the end of the sample period.

69


Table 5Flows from Unemployment and Mean Durations in the UK and USAa

Status at t+1

Status at t–1 Employed, Employed, Unemployed/out Censoredb

same industry new industry of labour force

UK Employed 0.20(7.01) 0.36(8.34) 0.17(11.30) 0.26(28.21)

USA Employed 0.46(4.04) 0.24(4.07) 0.15(7.43) 0.15(14.92)

Notes: a Numbers in parentheses indicate mean duration in months. b Following status not known becauseof right censoring (occurs after end of sample period).Source: Haynes et al. (2000).

suggest that other costs of moving are sufficientlylarge to encourage search in the original sector.19

A further interesting result is that workers in bothcountries who enter unemployment from manufac-turing are more likely to change sectors. If, as isthought to be the case, the manufacturing sector hasexperienced long-term decline, this provides someevidence of a relationship between sectoral trans-formation and factor mobility, as might be expected.

While these results seem to be supportive of thesmooth-adjustment hypothesis, the real world ismore complicated than this characterization sug-gests. It should be noted that in the UK, the condi-tional probability of staying in the same sector isgenerally lower than that of moving. This may be fora number of reasons: the rate of turbulence in theUSA may be lower, and workers that are displacedcan return to the same industries; the costs ofmoving sector in the USA may be higher, discour-aging movement; institutional arrangements in theUSA may facilitate the return of a worker to thesame sector—for example, via temporary lay-offs.

(iii) Job Tenure

An alternative way to look at changing patterns ofemployment is via job stability. If job reallocation hasincreased in response to the greater requirements ofrestructuring, we might expect to find that average

job tenure has declined. There are a number ofreasons why this is not straightforward, however.As we have seen, increasing turbulence has am-biguous impacts on gross flows and therefore ontenure. Further, increased job instability may becaused by increased reallocation within rather thanbetween sectors. Finally, increased movement maybe voluntary—observing higher turnover tells younothing about its cause. Recent evidence for the UKfrom Gregg and Wadsworth (1995) and Burgessand Rees (1996) suggests that, contrary to popularperceptions, there is no evidence that the averagelength of jobs declined dramatically over the 1970sand 1980s. It does appear, however, that jobs forolder workers and less-skilled men have becomeless stable, which suggests that it is peripheralworkers who have found it increasingly difficult tomove and it is not a general phenomenon.

(iv) Regional Adjustment

In the UK, much of the focus has been on regionalrather than sectoral mobility.20 Housing tenure hasattracted attention. Hughes and McCormick (1981)point to rigidities in the public rented sector, andOswald (1996) to the expansion of home ownershipat the expense of the private rented sector as the keyimpediment to mobility. Cameron and Muellbauer(1998) argue that fluctuations in house prices (par-ticularly negative equity) can discourage regionalmigration and increase inter-regional commuting.

19 This interpretation does not necessarily imply that potential wages in the original sector are greater than in any other sector.20 For example, Creedy (1974), Pissarides and Wadsworth (1989), Jackman and Savouri (1992), McCormick (1997). Pissarides

(1978) is an exception.

70


However, the links between regional and sectoralmobility have not been made explicit. If housing-market rigidities do cause workers to be less mobilebetween regions, does this have an effect on mobil-ity between sectors? If sectors are geographicallyevenly spread, one would expect the relationship tobe rather unimportant, since individuals will be ableto switch sectors without moving region. If, asseems more plausible, sectors are unevenly distrib-uted regionally, the relationship will be stronger.

Table 6 gives estimates of the average probability ofmoving between and within sectors and regions,split by employment status at t–1. These probabili-ties are another way of expressing gross flow ratesbetween sectors and regions. The second column inpanel (a) shows that the probability of moving to theexpanding from the declining sector is just 0.02 forindividuals who remain at the same address be-tween t–1 and t. This probability increases to 0.05for those who move addresses within region, whileindividuals who move region are by far the mostlikely to switch from the declining to the expandingsector, with a probability of 0.15. A similar patterncan be seen in the first column of panel (b): theprobability of moving sectors between two con-secutive years is highest for those who have also

moved address or region. Panel (c) shows there isalso a much higher probability of leaving non-employment for those who move region. The prob-ability of remaining in non-employment is only 0.59for those who move region, compared to 0.77 forthose who move address within region and 0.83 forthose who remain at the same address.

Thus, Table 6 shows that individuals who are geo-graphically mobile are more mobile between jobs,and more mobile between non-employment andemployment. However, there are some importantcaveats to this. The third column shows that indi-viduals who are geographically mobile are morelikely to enter as well as exit non-employment. Theprobability of entering non-employment from thedeclining sector is 0.18 for an individual who alsomoves region, compared to 0.06 for an individualwho remains at the same address. It should also benoted that the proportion of the sample who moveregion is extremely small. On average, the probabil-ity is less than 0.01. Thus, while there is clearly arelationship between ‘flexibility’ in terms of regionalmobility, sectoral mobility, and non-employment, it isnot necessarily important in terms of its overallcontribution to the sectoral transformation of theeconomy. Finally, note that it is not possible to make

Table 6Average Gross Flow Rates by Geographical Mobility, 1975–95

New firm, New firm, Not employeddeclining sector expanding sector

(a) Employed at t–1 in declining sectorSame address 0.05 0.02 0.06Same region (new address) 0.09 0.05 0.10New region 0.23 0.15 0.18Total 0.06 0.03 0.06

(b) Employed at t–1 in expanding sectorSame address 0.02 0.06 0.06Same region (new address) 0.04 0.13 0.11New region 0.06 0.32 0.20Total 0.02 0.07 0.06

(c) Not employed at t–1Same address 0.06 0.11 0.83Same region (new address) 0.07 0.16 0.77New region 0.12 0.30 0.59Total 0.06 0.01 0.82

71


any prediction about the extent to which geographi-cal mobility caused or is caused by job mobility,partly because the data do not reveal whether jobmoves precede or are preceded by geographicalmoves.

Figure 7 plots gross and net flows between regionsin the United Kingdom. Gross flows are the propor-tion of the sample who change address or regionbetween two consecutive years. Net flows aregiven by the sum of the (absolute) change in the sizeof each region, or alternatively the number of flowsout of region not cancelled out by flows into region.Although a large number of people move from thedeclining region (north) to the expanding region(south), an almost equal number move in the oppo-site direction. As with sectoral mobility of labour,gross flows greatly exceed net flows. Despite largedisparities in wage and employment rates betweenregions in the UK, net flows of labour have re-mained tiny over the whole period.

(v) The Effects of Sectoral Mobility on Skilledand Unskilled Workers

A central theme in the literature on the relationshipbetween globalization and labour markets has beenthe effects of increased trade on the distribution ofincome between high- and low-skill workers. Theanalysis of sectoral mobility provides an alternative,but neglected, view of the effects of internationaltrade (or technological change) on the distribution ofincome and employment between skilled and un-

skilled workers. As stressed by Haskel and Slaugh-ter (1999), in an economy with many sectors themovement of factors of production (e.g. labour)between sectors provides an alternative mechanismfor absorbing changes in the relative demand andsupply of those factors. However, it is not obviouswhat differences in sectoral mobility one mightobserve between skilled and unskilled workers. Thelatter might find it more difficult to move to newopportunities in expanding sectors, in which casethey would tend to have lower mobility rates thanskilled workers. Alternatively, if unskilled workersare more likely to be forced to move, they may havehigher mobility rates and, in particular, higher tran-sition rates into unemployment.

In Figure 8 we break down the patterns of grossflow rates by four simple occupational categories:skilled non-manual, skilled manual, unskilled non-manual, and unskilled manual.21 It is clear that less-skilled workers have higher gross flow rates bothbetween jobs and between jobs and non-employ-ment. What is particularly noticeable is the largedifferential in flow rates between employment andnon-employment. Panel (b) shows that in economicdownturns lay-off flows of unskilled manual work-ers into non-employment are three times greaterthan flows of skilled non-manual workers. Perhapsmore surprisingly, panel (a) shows that more-skilledworkers are also less mobile between sectors. Thismay reflect the fact that these workers have higherlevels of sector-specific human capital, as sug-gested earlier.

Figure 7Flows Between Regions

(a) Gross flows between regions (b) Net flows between regions

21 These groups are based on aggregations from the 1980 Standard Occupational Classification.

Gross flows between the "North" and the "South"

Gross flows between 10 UK standard regions

Gross flows between addresses

Prop. gross flow

Year75 77 79 81 83 85 87 89 91 93 95

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

Net flows between "North" and "South"

Net flows between 10 UK standard regions

Prop. net flow

Year75 77 79 81 83 85 87 89 91 93 95

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

72


The increased inequality between skilled and un-skilled workers, as observed in the UK and USAover that last 20 years, may therefore be related todifferential patterns of sectoral and employmentturnover between these groups. Unskilled workershave much higher exit rates into non-employment,as well as higher job-to-job turnover. As we have seen,both are likely to cause relative declines in wages.

V. CONCLUSIONS AND POLICYRESPONSES

Fundamental to most trade models of an economy’sresponse to external shocks is the reallocation offactors of production between sectors. These oftenmake extreme assumptions, either that factor real-location is frictionless (as in the Heckscher–Ohlinmodel) or very slow (as in the specific factorsmodel).22 In this paper we have drawn togetherevidence on two important questions. First, has therate of sectoral transformation in the UK increased,for whatever reason? Second, how difficult is it forlabour to move between sectors, and has it becomemore difficult?

There is some evidence that the rate of sectoraltransformation was greater in the UK in the 1970sand 1980s than in any other post-war decade.Although there has been a dramatic process ofsectoral transformation in the UK during the post-

war period, gross flows of workers are far greaterthan net flows. That is, changes in employmentshares across time disguise massive flows fromdeclining to expanding sectors and in the reversedirection and enormous flows within sectors. Al-though there is some evidence for the USA thatgross flows declined during the 1970s and 1980s, wefind that gross flows in the UK are basically pro-cyclical with no secular trend. We argue that grossflows are not in themselves indicative of the amountof sectoral reallocation occurring, because a sectoralshock can be accommodated by any amount ofgross flows. Rather they are useful as an indicatorof the costs of moving between sectors. In sectionIV a number of pieces of evidence were surveyedthat suggest that sectoral reallocation is costly,particularly in terms of aggregate unemploymentand unemployment duration. If the process of ad-justment between sectors is costly and unevenlydistributed, then there may be a case for interven-tion. Policies can broadly be described as having oneof two objectives: either to reduce the costs ofadjustment, or compensate losers.

The first is based on an efficiency argument. For agiven amount of reallocation, there will be a smallerloss of output if the transition process involves lessfrictional unemployment. ‘An obvious policy to dealwith this type of unemployment [i.e. mismatch] is tospeed up the process of adjustment by reducing theimpediments to intersectoral labour mobility’ (Nickell,

22 In contrast, the labour economic literature has primarily been concerned with the movement of labour between labour-marketstates rather than between sectors.

Figure 8Gross Flows Split by Skill Groups

(a) Job-to-job flows (b) Job-to-non-employment flows

Gross flows (%)

75 77 79 81 83 84 85 86 87 88 89 90 91 92 93 94 950.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Gross flows (%)

Skilled non-manual Skilled manual Unskilled non-manual Unskilled manual

Skilled non-manual Skilled manual Unskilled non-manual Unskilled manual

75 77 79 81 83 84 85 86 87 88 89 90 91 92 93 94 950.0

2.0

4.0

6.0

8.0

10.0

12.0

73


1991). However, this begs the question as to whythe market fails to reduce adjustment costs. Forexample, why do individuals fail to retrain to find jobsin the expanding sector or in more skilled occupa-tions? The second objective is based directly on anequity argument and indirectly on an efficiencyargument. ‘Government programmes are often jus-tified on the grounds that society should compensatethe losers for structural changes that benefit us inthe aggregate’ (Fallick, 1996). It is, of course, thisperspective which underpins broadly based socialadjustment regimes, such as, for example, unem-ployment insurance. As a result, one might questionthe rationale for specific trade-related adjustmentprogrammes. On this front there may be an effi-ciency dimension if potential losers have an incen-tive to lobby against adjustment. It seems plausible,for example, that trade adjustment assistance pro-grammes in the USA are implemented to compen-sate organized labour with the political power tolobby against increased adjustment.

In the USA, programmes targeted specifically atdisplaced workers have offered income replace-ment, re-employment, and retraining services forsome limited period. More recently, support has alsoincluded the requirement of advance notification forplant closures or mass lay-offs. However, empiricalevidence on the efficacy of these programmes is, atbest, mixed. As income losses following displace-ment are often long-lived (e.g. Jacobson et al.,1993), it seems unlikely that temporary incomereplacement will fully compensate for this loss. Theimpact of retraining programmes has been subjectto widespread empirical study (see Heckman et al.,1999, for a recent survey). The consensus is thatexpenditure often outweighs benefits in terms ofincreased employment probabilities or earnings.Evidence is also emerging that the benefits ofreceiving advance notice are ‘modest at best’(Kletzer, 1998).

Trade-related adjustment policies targeted specifi-cally at displaced workers are less common in

Europe than the USA. This is, in part, a reflection ofthe presence of more general social safety nets inthe form of unemployment assistance and insur-ance. Sapir (2000) suggests that the adjustmentcosts of globalization have tended not to fall on themedian voter in Europe, and that organized labourhas therefore voiced less opposition to the adjust-ment and reallocation of labour. In the UK, attentionhas focused most on regional mobility. Compensa-tion has often taken the form of regional assistanceprogrammes rather than payments to displacedindividuals. Some have argued that rigidities in thehousing market contribute to immobility, althoughfor different reasons. Hughes and McCormick (1981)argued that social housing prevents individuals mov-ing, while Oswald (1996) argues that high levels ofhome ownership have contributed to aggregateunemployment because of the high costs of moving.An attempt to revive the private rented sector maytherefore suggest itself as a remedy. Many havelooked to the education system and human-capitalformation as an explanation for the inability ofworkers to adjust to changing patterns of demand.For example, Nickell and Bell (1996) suggest that‘The very high level of education and trainingembodied in the vast bulk of the German labourforce enables them to respond in a flexible mannerto demand shifts.’

Some have dismissed sectoral reallocation as asuspect in the search for the causes of high levels ofunemployment and increasing skilled–unskilled wagedifferentials observed in some OECD countriesover the last 30 years. We feel that this may bepremature, for three reasons. First, industrial turbu-lence in the UK peaked in the 1970s and 1980s, andhas subsequently returned to post-war levels. Sec-ond, net flows of labour between sectors have beenlargely facilitated by movements in and out of thelabour force rather than directly from job to job.Third, micro-econometric evidence suggests thatmovements between sectors are associated withlonger unemployment spells than movements withinsectors.

74


REFERENCES

Abraham, K. (1986), ‘Mismatch and Labour Mobility: Some Final Remarks’, in F. Padoa Schioppa (ed.), Mismatch andLabour Mobility, Cambridge, Cambridge University Press, 453–81.

— Katz, L. (1986), ‘Cyclical Unemployment: Sectoral Shifts or Aggregate Disturbances?’, Journal of PoliticalEconomy, 94, 507–22.

Altonji, J., and Williams, N. (1997), ‘Do Wages Rise with Job Seniority? A Reassessment’, NBER Working Paper 6010.Atkinson, A. (2001), ‘Is Rising Inequality Inevitable? A Critique of the Transatlantic Consensus’, The World Economy,

21, forthcoming.Blanchard, O. J., and Diamond, P. (1990), ‘The Cyclical Behaviour of the Gross Flows of US Workers’, Brookings Papers

on Economic Activity, 2, 85–153.Blanchflower, D., and Freeman, R. (1994), ‘Did the Thatcher Reforms Change British Labour Market Performance?’,

in R. Barrell (ed.), The UK Labour Market, Cambridge, Cambridge University Press, 51–92.Booth, A. (1997), ‘Career Mobility in Britain’, ESRC Research Centre on Micro-Social Change Working Paper 97-21.Brainard, S., and Cutler, D. (1993), ‘Sectoral Shifts and Cyclical Unemployment Reconsidered’, Quarterly Journal of

Economics, 108, 219–43.Burda, M., and Wyplosz, C. (1994), ‘Gross Worker and Job Flows in Europe’, European Economic Review, 38, 1287–

315.Burgess, S., and Rees, H. (1996), ‘Job Tenure in Britain 1975–92’, The Economic Journal, 106, 334–44.Cameron, G., and Muellbauer, J. (1998), ‘The Housing Market and Regional Commuting and Migration Choices’,

Scottish Journal of Political Economy, 45, 420–46.Creedy, J. (1974), ‘Inter-regional Mobility: A Cross-section Analysis’, Scottish Journal of Political Economy, 21, 41–

53.Davis, S., and Haltiwanger, J. (1992), ‘Gross Job Creation, Gross Job Destruction and Employment Reallocation’,

Quarterly Journal of Economics, 107, 819–63.Fallick, B. C. (1993), ‘The Industrial Mobility of Displaced Workers’, Journal of Labour Economics, 11, 302–23. — (1996), ‘A Review of the Recent Empirical Literature on Displaced Workers’, Industrial and Labor Relations

Review, 50, 5–16.Greenaway, D., and Torstensson, J. (1997), ‘Back to the Future: Taking Stock of Intra-industry Trade’, Weltwirtschaftliches

Archiv, 133, 249–69. — Hine, R. C., and Wright, P. W. (1999), ‘Modelling the Impact of Trade on Employment in the United Kingdom’,

European Journal of Political Economy, 15, 485–500.Gregg, P., and Wadsworth, J. (1995), ‘A Short History of Labour Turnover, Job Tenure, and Job Security, 1975–93’,

Oxford Review of Economic Policy, 11(1), 73–90.Haskel, J., and Slaughter, M. (1999), ‘Trade, Technology and UK Wage Inequality’, Centre for Research on

Globalisation and Labour Markets Research Paper 99/2, University of Nottingham.Haynes, M., Upward, R., and Wright, P. (1999), ‘Estimating the Wage Costs of Inter- and Intra-sectoral Adjustment’,

Centre for Research on Globalisation and Labour Markets, Research Paper 99/15, University of Nottingham,mimeo.

— — — (2000), ‘Smooth and Sticky Adjustment: A Comparative Study of the US and UK’, Review ofInternational Economics, 8, 517–32.

Heckman, J., LaLonde, R., and Smith, J. (1999), ‘The Economics and Econometrics of Active Labor Market Programs’,in O. Ashenfelter and D. Card (eds), Handbook of Labor Economics, Amsterdam, North-Holland, 1865–2097.

Hughes, G., and McCormick, B., (1981), ‘Do Council House Policies Reduce Migration between Regions?’, TheEconomic Journal, 91, 919–37.

Jackman, R., and Savouri, S. (1992), ‘Regional Migration in Britain: An Analysis of Gross Flows Using NHS CentralRegister Data’, The Economic Journal, 102, 1433–50.

Jacobson, L., LaLonde, R., and Sullivan, D. (1993), ‘Earnings Losses of Displaced Workers’, American EconomicReview, 83, 685–709.

Jacoby, S. (1983), ‘Industrial Labor Mobility in Historical Perspective’, Industrial Relations, 22, 261–82.Jovanovic, B. (1979), ‘Job Matching and the Theory of Turnover’, Journal of Political Economy, 87, 972–90. — Moffitt, R. (1990), ‘An Estimate of a Sectoral Model of Labor Mobility’, Journal of Political Economy, 98, 827–

52.Kletzer, L. G. (1998), ‘Job Displacement’, Journal of Economic Perspectives, 12, 115–36.Lilien, D. (1982), ‘Sectoral Shifts and Cyclical Unemployment’, Journal of Political Economy, 90, 777–93.

75


Loungani, P., Rush, M., and Tave, W. (1990), ‘Stock Market Dispersion and Unemployment’, Journal of MonetaryEconomics, 25, 367–88.

McCormick, B. (1997), ‘Regional Unemployment and Labour Mobility in the UK’, European Economic Review,41, 581–9.

Machin, S., and Van Reenen, J. (1998), ‘Technology and Changes in Skill Structure: Evidence from Seven OECDCountries’, Quarterly Journal of Economics, 113, 1215–44.

Matusz, S., and Tarr, D. (2001), ‘Adjusting to Trade Policy Reform’, Michigan State University, mimeo.Mills, T. C., Pelloni, G., and Zervoyianni A. (1995), ‘Unemployment Fluctuations in the United States: Further Tests

of the Structural Shifts Hypothesis’, Review of Economics and Statistics, 77, 294–304.Mincer, J. (1986), ‘Wage Changes in Job Changes’, Research in Labor Economics, 8, 171–97.Mortensen, D., and Pissarides, C. (1994), ‘Job Creation and Job Destruction in the Theory of Unemployment’, Review

of Economic Studies, 61, 397–415.Murphy, K., and Topel, R. (1987), ‘The Evolution of Unemployment in the United States 1968–1985’, in S. Fischer (ed.),

NBER Macroeconomics Annual, Cambridge, MA, MIT Press.Neal, D. (1995), ‘Industry-specific Human Capital: Evidence from Displaced Workers’, Journal of Labour Economics,

13, 653–77.Nickell, S. (1991), ‘Mismatch and Labour Mobility: Some Final Remarks’, in F. Padoa Schioppa (ed.), Mismatch and

Labour Mobility, Cambridge, Cambridge University Press, 481–4. — (1996), ‘Sectoral Structural Change and the State of the Labour Market in Great Britain’, CEP Discussion Paper

Series, The Labour Market Consequences of Technical and Structural Change, No. 2. — (1999), ‘Unemployment in Britain’, in P. Gregg and J. Wadsworth (eds), The State of Working Britain

Manchester, Manchester University Press. — Bell, B. (1996), ‘Changes in the Distribution of Wages and Unemployment in OECD Countries’, American

Economic Review Papers and Proceedings, 86, 303–8.Oswald, A. (1996), ‘A Conjecture on the Explanation for High Unemployment in the Industrialised Nations: Part I’,

Department of Economics, University of Warwick, mimeo.Pissarides, C. (1978), ‘The Role of Relative Wages and Excess Demand in the Sectoral Flow of Labour’, Review of

Economic Studies, 45, 453–67. — Wadsworth, J. (1989), ‘Unemployment and the Inter-regional Mobility of Labour’, The Economic Journal, 99,

739–55.Rowthorn, R. (2000), ‘The Political Economy of Full Employment in Modern Britain’, Oxford Bulletin of Economics

and Statistics, 62, 139–73.Sapir, A. (2000), ‘Who is Afraid of Globalisation’, paper presented at IEA conference on Globalisation and Labour

Markets, University of Nottingham, July 2000.Sicherman, N., and Galor, O. (1990), ‘A Theory of Career Mobility’, Journal of Political Economy, 98, 169–92.Slaughter, M. (1999), ‘Globalisation and Wages: A Tale of Two Perspectives’, The World Economy, 22, 609–30.Thomas, J. M. (1996), ‘An Empirical Model of Sectoral Movements by Unemployed Workers’, Journal of Labour

Economics, 14, 126–53.Weiss, A. (1995), ‘Human Capital vs Signalling Explanations of Wages’, Journal of Economic Perspectives, 9, 133–

54.Wood, A. (1998), ‘Globalization and the Rise in Labour Market Inequalities’, The Economic Journal, 108, 1463–82.

Documents

Sectoral transformation and labour-market flows