REGIONAL DIFFERENCES IN LABOUR PRODUCTIVITY IN THE NETHERLANDS

Tijdschrift voor Economische en Sociale Geografie – 2005, Vol. 96, No. 3, pp. 334 –343.© 2005 by the Royal Dutch Geographical Society KNAGPublished by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden MA 02148, USA

WINDOW ON THE NETHERLANDS

REGIONAL DIFFERENCES IN LABOUR PRODUCTIVITY IN THE NETHERLANDS

LOURENS BROERSMA* & JOUKE VAN DIJK**

*

Faculty of Economics and Faculty of Spatial Sciences, University of Groningen, PO Box 800, NL-9700 AV Groningen, the Netherlands. E-mail: [email protected]

**

Urban and Regional Studies Institute and Faculty of Spatial Sciences, University of Groningen, PO Box 800, NL-9700 AV Groningen, the Netherlands. E-mail: [email protected]

Received: January 2005

INTRODUCTION

One of the major goals of current Dutch econ-omic policy is to increase labour productivitygrowth. In order to achieve this goal a Dutchinnovation platform is established, chairedby the Dutch Prime Minister Balkenende, withthe aim to reinforce the innovativeness of theDutch economy. Innovation has a positive effecton productivity growth (Donselaar

et al.

2004).Besides this macro policy goal, the most recentmemorandum on regional policy ‘Pieken in deDelta’ (EZ 2004) makes it clear that enhancingproductivity growth is an important goal ofregional policy as well. However, until now therehas been hardly any information about thespatial variation in labour productivity. The aimof this ‘Window on the Netherlands’ is to givean overview of the regional disparities in labourproductivity in relation to regional differencesin welfare and to explore how differences inlabour productivity can be related to charac-teristics of the region.

Until now most studies of labour productivityare at the country level, whereas the analysis atthe regional level is largely neglected. This ispartly due to lack of suitable data. In an explora-tive study Broersma & Van Dijk (2003) showthat there are substantial differences in labourproductivity between Dutch regions. Further-more, the regional differences are larger than

in other European countries. From inter-national comparative studies it is well known thatthe Netherlands has one of the highest labourproductivity levels in the world when produc-tivity is measured as GDP per hour worked(McGuckin & Van Ark 2004). However, in termsof welfare, measured as GDP per capita, theposition of the Netherlands drops substantially.The difference is due to the fact that Dutchemployees work the lowest number of hourswithin the OECD. Dutch employees often workon a part-time basis and the official number ofhours worked in full-time employment is quitelow (Klomp & Roelandt 2004). During the1990s the productivity growth rate slowed downconsiderably. This is in line with many otherEuropean countries, but in marked contrastwith the United States. For the interpretationof the growth of labour productivity it is impor-tant to note that a high growth rate can be dueto a large increase in GDP, but also due to a lowgrowth rate of employment. In line with this, amoderate increase in GDP together with a highrate of employment growth may cause lowgrowth rates of labour productivity.

During the 1990s the labour market situationin the Netherlands underwent quite dramaticchanges and was dubbed the ‘Dutch miracle’.Unemployment fell from eight per cent in 1994to a mere three per cent in 2001. At the sametime employment increased by more than one


335

© 2005 by the Royal Dutch Geographical Society KNAG

million jobs between 1994 and 2001. In fact, thegrowth of employment in the Netherlands wasmore or less equal to the United States andmuch higher than in the rest of Europe.

1

In theUnited States this strong employment growthwas however accompanied by even strongerGDP growth rates, whereas this was hardlythe case for the Netherlands. A flexible labourmarket has enhanced these high employmentgrowth rates in the United States whereas in theNetherlands it was mainly attributed to a sus-tained policy of wage moderation. This led to afall in labour costs relative to competitive coun-tries. Besides low labour costs a high level of labourproductivity is an important factor of competi-tiveness both for countries and for regions.

Although in the 1990s all Dutch regions hadgrowing employment and falling unemploy-ment there remained substantial regional dif-ferences in unemployment (Atzema & Van Dijk2005). Therefore, regional differences in labourproductivity are possibly an important determi-nant for the explanation of regional economicdisparities. For the analysis of regional differ-ences in labour productivity in the Netherlandsa rich dataset is available at the regional levelfor the period 1990–2001 and in even moredetail for the period 1995–2001. Detailed infor-mation is also available about the amount ofcapital for regions, level of education, inno-vation and the regional economic structure.

In the remaining text we will first relate regionallabour productivity, in terms of GDP per hour,to welfare, in terms of GDP per capita in 2001 atNUTS-2 levels. Next, we discuss factors that explainregional differences in labour productivity be-tween 1990 and 2001 at NUTS-3 levels. As a finalstep we briefly go into the recent policy measuresthat aim to stimulate productivity growth.

REGIONAL DIFFERENCES IN WELFARE AND THE ROLE OF LABOUR PRODUCTIVITY

The most commonly used indicator for labourproductivity in the scientific literature is GDPper hour worked. However, it is interestingto compare this indicator with the GDP peremployed worker or GDP per capita, becausethe latter are commonly used to compareregional disparities and serve, for instance,as indicators on which the entitlement forregional policy measures of the EuropeanUnion is based. The fifth column of Table 1shows that regional GDP per capita

as a percent-age of the national GDP per capita ranges from73 per cent in the province of Flevoland to126 per cent in Utrecht.

2

Regional disparitiesin GDP per capita can also be due to a low levelof labour participation or to a high share ofchildren and retired people in the total popu-lation. Therefore, we also calculated the GDP

Table 1. Regional differences in GDP and disposable income in 2001 (Netherlands = 100; mining excluded).

Province GDP per hour worked

(labour productivity)percentage of Netherlands

Effect of part-time work

and commuting percentage point

Effect of participation

and age compositionpercentage

point

GDP per capitapercentage of Netherlands

Disposable income

per capitapercentage of Netherlands

Groningen 101.6 −6.3 −5.4 89.9 90.9Friesland 92.3 −10.2 −4.6 77.5 90.0Drenthe 87.9 −9.2 −4.8 73.9 93.6Overijssel 89.1 −2.1 −2.5 84.5 91.8Flevoland 98.9 −27.3 1.2 72.7 91.8Gelderland 93.2 −6.4 −1.0 85.9 97.3Utrecht 105.7 12.1 8.2 126.0 107.3Noord-Holland 102.4 9.2 5.1 116.7 102.7Zuid-Holland 101.9 2.5 −0.7 103.7 107.3Zeeland 101.6 −8.3 −9.3 84.0 96.4Noord-Brabant 95.3 1.2 1.9 98.3 99.1Limburg 94.6 −5.5 −2.8 86.2 97.3

336

LOURENS BROERSMA & JOUKE VAN DIJK


per employed worker, taking into account dif-ferences in participation and age composition.Column four shows the difference in percent-age points with GDP for the total population.For the three northern provinces Groningen,Friesland and Drenthe about one third to halfof the difference in GDP per capita from thenational average can be attributed to the rela-tively low share of active participants on thelabour market in the total population and forthe island province of Zeeland it is even more.Lower participation rates in the age group 15–64 are much more important than differencesin the share of children and retired people. Theeffect of a greying population is most importantfor the Zeeland and Drenthe provinces with 2.6and 1.8 percentage points respectively, but evenin these provinces the lower participation ratesare much more important.

GDP per worker living in a region differsfrom GDP per hour worked in a region for tworeasons: working hours and commuting. Thetotal effect of these two factors is shown incolumn 3 of Table 1. It is clear that the productionof a worker will vary with the number of workinghours. Part-time work is very popular in theNetherlands: almost 45 per cent (75% forfemales and 22% for males) of the labour forceworks part time and this is almost three timesas much as the EU average (EU 2004, p. 29,Chart 18). Within the Netherlands there aresubstantial regional differences in part-timework. The most noticeable differences arefound for the provinces Groningen andZuid-Holland. The share of part-time workers isseven percentage points higher in Groningenand six percentage points lower in Zuid-Holland than the national average. Furthermore,relatively high shares of part-time workers arefound in Friesland, Drenthe and Zeeland,which have relatively low GDPs per capita.Another disturbance is caused by the fact thatpeople who live in a region can be productivein another region. Since GDP is measured atthe work location and the size of the populationand the labour force at the place of residence,this may cause a bias because there is substantialcross border commuting between provinces.The effect of commuting takes by far the largestpart of the 27.3 per cent calculated for the pro-vince of Flevoland, located on reclaimed landfrom the IJsselmeer, a lake that used to be part

of the former Zuiderzee. The high number ofcommuters from Flevoland to the provinces ofNoord-Holland and Utrecht accounts for alarge part the opposite effect of commuting inthese provinces.

Starting with the GDP per capita in column5 of Table 1 we are now able to compare thiswith GDP per worked hour (column 2) takinginto account differences in active labour marketparticipation, working hours and commuting.GDP per capita is not a proper measure forlabour productivity in terms of competitivenessof workers’ performance for which GDP perhour worked is more suited. When we comparethose measures in column 2 and column 5 wemay conclude that regional disparities are nowmuch smaller and range from 87.9 per cent to105.7 per cent. However, with the exceptionof the new provinces of Flevoland and Zeeland,the rank order of the provinces is more or lessthe same, implying that on average the regionswith a high level of GDP per capita also show ahigh level of GDP per hour worked. This rank-ing of provinces in terms of economic perform-ance also corresponds quite close to the rankingwe get when we use the disposable income percapita as indicator of welfare. The regionalvariation in disposable income is of the samemagnitude as GDP per hour worked. This isremarkable because GDP per hour worked canbe seen as a measure of regional productive per-formance at the workplace, whereas disposableincome per capita is measured at the place ofresidence and includes the redistribution effectof taxes, subsidies and social security. The mostnotable exception is the province of Groningen,which is in terms of GDP per hour workedabove the national average but ranks in termsof disposable income together with Friesland atthe lowest level. Table 1 shows that this is causedby relatively low participation rates in combin-ation with a relatively low number of workinghours. These low participation rates are due tothe high number of young persons in universityand higher vocational education and the highnumber of non-participating females over 40years of age. The low number of working hours,resulting in a high share of part-time workers,is partly caused by the high share of governmentservices, like education and health care, wherepart-time work is very common. Lower partici-pation and less working hours are most likely


337


also due to a lower demand for labour due tothe remote location of Groningen. In addition,social security benefits, pensions and so on areincorporated in the disposable income, but notin regional GDP. These secondary income com-ponents mitigate regional differences in GDPper capita even further.

From column 2 in Table 1 we may concludethat the regional disparities in GDP per hourworked are smaller than in GDP per capita.However, we may also conclude that theregional differences in labour productivity aresubstantial in 2001. GDP per hour worked inthe centrally located province of Utrecht is 1.2times higher that in the province of Drenthe inthe northern part of the country. At the NUTS-3 level of 40 COROP-regions the differencebetween the highest and the lowest level is 1.5and thus substantially larger. Therefore, in thenext section we will analyse the regional differ-ences in labour productivity during the lastdecade in more detail at the spatial level of the40 COROP-regions (NUTS-3).

REGIONAL DIFFERENCES IN LABOUR PRODUCTIVITY

Figure 1 shows the regional variation in the levelof labour productivity in 2001 for COROP-regions in the Netherlands (see Figure 2 for thedemarcation of the COROP-regions).

3

GDP perlabour year ranges from

$

51,000 in Zuidoost-Drenthe to

$

78,000 in Zeeuwsch-Vlaanderen,whereas the average for the Netherlands is

$

64,000. Two types of regions show the highestlevel of productivity. The area consisting ofAmsterdam, Gooi en Vechtstreek and Utrechtis highly specialised in financial and businessservices, especially in ICT and creative indus-tries. The other regions with high productivitylevels are found in both the central part of thecountry around Rotterdam and IJmond and inthe peripheral areas of Zeeuws-Vlaanderen andDelfzijl. These regions have in common capital-intensive industries in basic metal and chemis-try that are very important. The regions withlow levels of productivity are located in theperiphery, especially in the east along theGerman border.

Figure 3 shows the average annual realgrowth rate of labour productivity over theperiod 1991–2001.

4

In all regions real labour

productivity growth is positive, ranging from0.2 per year in Alkmaar and environs to 1.7 percent per year in Delft and Westland. The averageannual growth is 1.1 per cent. Of the regions withthe highest level of productivity in Figure 1 onlyDelfzijl is also in the highest growth category inFigure 3. In general the regions with growthrates above the national average are also theregions with a level of productivity above thenational average. However, the relation is notvery strong as becomes clear from the ratherlow value of 0.26 for the correlation coefficientbetween level and growth. The regions with astrong service sector and located in the centreof the country, on the border or just outside theRandstad show the fastest growth in labourproductivity. In the regions in the Randstad andthe province of Brabant, the main cause of thisgrowth is an increase in GDP. In the regionswith high productivity growth in Gelderlandand Overijssel and in the traditional trouble spotsof Oost-Groningen and Zuid-Limburg laggingproductivity growth is mainly a consequence ofthe slow growth of employment. Instead of anincrease in economic activity, the high labourproductivity growth for these regions may indi-cate the removal of slack capacity. This implies

Figure 1. Level of labour productivity as value added (in1000 $) per labour year worked in 2001.

338



that regional disparities in productivity levelsincrease slightly over time. Especially the peri-pheral regions with low levels of productivityare lagging behind more and more. The caseof Delfzijl clearly shows that regions with a highlevel of productivity and a high growth rate arenot always very prosperous regions: during thewhole period 1991–2001 the peripheral regionof Delfzijl shows the highest unemploymentrate of all Dutch regions.

As a next step we will use Figures 4–7 to shedsome light on the explanation of the observed

regional differences in labour productivity. Aswill be clear from the descriptions of Figures 1and 3, the sector structure plays a role. Regionswith an over-representation of capital-intensiveindustries or specialised services do better interms of productivity. There is also an obviousrelation between the level of education andlabour productivity reflected in the variation inwages by educational level. From the viewpointof competitiveness of a region a lower level ofpayment can compensate a lower level of labourproductivity, because a low level of productivity,

Figure 2. Regional demarcation of the COROP-regions.


339


combined with a low wage, might lead to thesame unit labour cost that a high level ofproductivity combined with a high wage wouldyield. Therefore, Figures 5 and 6 provide mapswith the share of the higher educated in thelabour force and the cost of labour per labouryear.

5

Finally, the map in Figure 7 reflects thedensity of the number of jobs per square kilo-metre, which can be seen as an indicator for thepresence of agglomeration effects. Figures 4–7are based on average values for the period1991–2001.

In the discussion about the regional differ-ences shown in the maps the sectoral compo-sition in a region is several times mentioned as apossible explanatory factor. From recent studiesby Broersma & Van Dijk (2003) and Broersma& Oosterhaven (2004) who analyse Dutch datafor 1990–2000, it becomes clear that regionaldeviations from the national sectoral compo-sition account for about 25 per cent of theregional variation in the levels of productivity.Figure 4 gives an impression of the regionalvariation of the effect of the sectoral composi-tion on the level of productivity from a shift-share-analysis based on 20 sectors (mining isexcluded). The map reflects that the high pro-

ductivity level in the area Amsterdam, Gooi- enVechtstreek, Utrecht and The Hague is partlycaused by an over-representation of servicesectors with high levels of productivity. Theregions with high productivity levels due tothe presence of capital-intensive industries inIJmond and Rijnmond and the peripheralareas Delfzijl and Zeeuws-Vlaanderen alsoshow positive effects of the sectoral compo-sition. The negative sectoral component formost of the peripheral regions with low produc-tivity levels indicates that low productivity sec-tors are over-represented.

Figure 5 shows the regional variation in theaverage share of higher educated (academicsand higher vocational graduates) over theperiod 1990–2001. The share ranges from 36per cent in Amsterdam to 12 per cent in Oost-Groningen, where the share of the highereducated in the national labour force is 25 percent. The regions with a high share of highereducated outside the Randstad are mainly theregions where institutes for higher educationare located. When Figure 5 is compared withFigure 1, it is clear that regions with low

Figure 3. Real average annual growth of labourproductivity (in per cent), 1991–2001.

Figure 4. Industry mix or sectoral component of the levelof labour productivity based on shift share analysis of 20sectors (mining excluded), 1990–2000 (in percentagedifference from national sector component).

340



productivity levels (often located in the periph-ery) usually have a lower than average share ofhigher educated in the labour force and viceversa. This positive relation is confirmed withthe value of 0.42 for the correlation coefficient.As is clear from a comparison of Figure 5 withFigure 3 there is no clear relation between theshares of higher educated and the growth rateof labour productivity and this is confirmedby the correlation of 0.03 between these twovariables.

From the map in Figure 6 reflecting theregional differences in labour costs, measuredin thousand

$

per labour year, and a correlationcoefficient of 0.69 with regional difference inlabour productivity, it is clear that lower levelsof productivity are partly compensated by loweraverage wages. The highest labour costs arefound in Amsterdam (

$

33,400 per labour year)and the lowest in Flevoland (

$

28,500 perlabour year), whereas the national averageequals

$

30,200. The correlation coefficient of0.56 suggests that a lower average wage in aregion is related to the lower share of highereducated in that region. Most probably thiseffect is even stronger because wage cost is in

this comparison measured at the work locationand education at the place of residence andthus a bias may occur due to commuting effects.The positive relation between wage costs andthe level of productivity indicates that low pro-ductivity does not necessarily lead to higher unitlabour costs and thus might not have a negativeeffect on the regions competitiveness. In linewith the result for education, comparison ofFigure 6 with Figure 3 does not indicate thatthere is a relation between wage cost level andthe growth rate of labour productivity and thisis confirmed by the correlation of

−

0.03 betweenthese two variables. Instead we expect wage

growth

to correlate with productivity growth, butregional wage growth rates are largely similardue to the fact that in the Netherlands collectivewage bargaining results on wage growth ratesapply to all regions. Therefore differences inregional labour productivity growth andregional wage growth are not related.

In Figure 7 the job density per square kilo-metre is shown. Job density can be seen as anindicator for the presence of agglomerationand cluster effects that may have a positive influ-ence on labour productivity (Ciccone 2002). By

Figure 5. Average share of the higher educated (academicand higher vocational level) in the employed labour forceover the period (in per cent) 1990–2001.

Figure 6. Average total labour cost (in $1000) per labouryear, 1990–2001.


341


far the highest spatial concentration of jobs isfound in the government centre at The Hague,with 1,483 jobs per km

2

, where in Oost-Groningenand Zuidwest-Friesland there are only 52 jobsper km

2

.The areas with a high concentration of jobs

are all located in the Randstad and correspondlargely to high productivity areas. In the peri-pheral regions job density is substantially lowerthan the national average of 186 per km

2

. Thepositive relation between job density and thelevel of productivity is also confirmed by thevalue of the correlation coefficient of 0.70. Jobdensity is also positively correlated with theshare of higher educated in Figure 5 (r = 0.70)and labour cost in Figure 6 (r = 0.56). WhenFigure 7 is compared with the growth of pro-ductivity in Figure 3 the similarities are less clearand the correlation coefficient is even negative(r =

−

0.13). It seems likely that the growth ofproductivity is hampered when the spatial con-centration of jobs is extremely high and causescongestion, as is the case in the Randstad area.Regression results obtained by Broersma &Oosterhaven (2004) confirm this hypothesis.They find that higher spatial concentrations ofjobs are significantly positive related to the level

of productivity, but are significantly negativerelated to productivity growth.

CONCLUSIONS AND POLICY IMPLICATIONS

At the NUTS-2 level of twelve provinces,regional differences in economic performancemeasured in GDP per capita are substantial inthe Netherlands. GDP per capita in the richestprovince of Utrecht is 1.7 times higher than inthe ‘poorest’ province of Flevoland. We haveshown that a substantial part of this inequalitycan be attributed to regional differences in par-ticipation rates, commuting and working hours(Table 1). When we control for these variableswe end up with GDP per hour as an appropriatemeasure of labour productivity. Labour pro-ductivity in Utrecht and Drenthe are the mostextreme cases, but the regional difference goesdown from 1.7 to 1.2. At the NUTS-3 level of40 COROP regions the regional differences aresubstantially larger with a factor of 1.5. Thehighest level of productivity is found in theRandstad regions that are highly specialised infinancial and business services and in a fewregions with capital-intensive industries in basemetal and chemical industries. The regionswith low levels of productivity are located inthe periphery, especially in the east along theGerman border. Regions with high levels oflabour productivity can be characterised by ahigh share of higher educated in the labourforce and a high concentration of jobs. Fromthe positive correlation between labour pro-ductivity and labour cost we can conclude thatthe advantage of higher productivity is partlyoffset by higher cost. This implies that in termsof competitiveness the regional differences aresubstantially smaller than the figures aboutlabour productivity suggest. It is clear, however,that regions with lower levels of labour pro-ductivity also show lower levels of GDP per capitaand this often also goes together with lower par-ticipation rates and a higher share of part-timeworkers in these regions. Labour productivityand participation rates are both positively cor-related with education. In accordance with thework of Barro (1991) this may suggest that apolicy aiming to increase the level of educationin a region should be advocated. The latter isonly a good recipe if the higher educated can

Figure 7. Average job density (in jobs per km2 landsurface) 1990–2001.

342



indeed find employment within the region.Otherwise the only effect will be an increase inout-migration of the higher educated, becauserecent empirical evidence suggests that causalitymay run mainly from employment growth toeducation and not vice versa (see Bils & Klenow2000; Van Dijk & Bosch 2003). In this case apolicy aiming to diminish the share of workerswith the lowest level of education via formaleducation or via on-the-job-training might bemore successful in reducing regional differ-ences in welfare than a policy focusing on theincrease of the share of the highly educated,who might leave the region after finishing theireducation. Instead of a policy focusing on edu-cation, a policy aimed at creating new jobs forboth low and high skilled might be a betteralternative to solve the problem.

Even though the level of labour productivityin the Netherlands is very high, the rate ofgrowth of Dutch labour productivity hasdecreased in the 1990s compared to its com-petitors in Europe (EU-15) and to the UnitedStates.

6

This causes the competitive advantageof the Netherlands over other countries toerode. When labour productivity remains at thisdeclining growth path additional labour inputwill yield increasingly smaller additional output.However, a further increase in labour supply,especially for women and the elderly, is pre-cisely what the Dutch Government is aiming atby means of tax policy, increasing the statutoryworking week and with revisions of socialsecurity and disability arrangements. This callfor additional labour supply does not stop thedownward trend in productivity growth. It istherefore necessary that productivity growth isstimulated in another way to make this addi-tional labour also more productive labour. Animportant policy handle in this respect isstimulation of innovative behaviour of bothcompanies and government. We focus here onregional issues that may help to enhance pro-ductivity growth, since productivity growth isnot only the goal of macroeconomic policy inthe Netherlands, but also of regional policy (seeEZ 2004).

In the new regional policy plans of the DutchGovernment focus is on regions with a nationalinterest and the aim is to remove barriers inthose regions that hamper productivity growth.Because of the national interest of these regions

this helps to stimulate national productivitygrowth. The government suggests that theseregions are basically the Randstad and thesoutheastern parts of the Netherlands. Themain reason for focusing on these regions isthe assumption that they have agglomerationadvantages due economies of scale, spilloversand are near to other economic activities. Allo-cating regional policy measures to these regionsis supposed to give the highest return on invest-ment and to lead to an extra boost of pro-ductivity growth. A crucial question in this debateis the relation between agglomeration effectsand productivity. There is indeed a positiverelation between job density, as approximationof agglomeration effects, and the

level

of labourproductivity. However, we found a negativerelation between job density and the

growth

oflabour productivity. This suggests that investingin already highly dense regions will lead to morecongestion and less space, which in the endleads to even more slowing down of productivitygrowth. Instead, investing in less dense regions(the light areas in Figure 7) seems to be a muchmore promising route to get the productivitygrowth rate back on track. These results severelydoubt the assumption of the government thatthe allocation of the major part of the budgetfor regional policy to the congested regions inthe Randstad and southeast Netherlands willlead to higher returns on investment thaninvesting in more peripheral regions. Besideshigher returns at the national level, investmentsin peripheral regions may also help to reduceregional disparities in welfare, although this isno longer an official goal of regional policy forthe present government.

Acknowledgements

The authors thank Bertus Talsma for his assistancein processing the data and the maps and VeroniqueSchutjens for comments on an earlier version.

Notes

1. Employment growth in this context measured inpersons employed. In the Netherlands a highshare of new entrants in the labour market workpart time and, therefore, the growth in employ-ment measured in full-time equivalents is substan-tially lower.


343


2. In some publications the province of Groningenis listed with a very high GDP per capita due tothe large amount of natural gas originating fromthis region. Because natural gas is easy to trans-port, the profits are used nationwide and thebenefits for the province of Groningen are ratherlimited. To avoid this possible statistical bias thatmay also occur in some other regions in a minorway, the sector mining is excluded from all ana-lysis in this paper.

3. For the analysis by COROP-regions we use GDPper labour year as an indicator for labour pro-ductivity. This is because employment in terms ofnumber of hours worked is unavailable by regionbefore 1995. Instead we use the labour volumeas measure of employment, expressed as thenumber of labour years. This means that (part-time) jobs are converted to their full-time equi-valent, i.e. two part-time jobs of 20 hours a weekeach equal one full-time job of 40 hours a week.GDP per labour year is very close to GDP per hour,but may differ slightly because the definition offull time is not exactly the same for each sectordue to differences in collective agreements(CAOs) with regard to the length of the standardworking week, days off, etc. Because the sectoraldistribution differs by region, labour productivityin terms of output per hour worked may slightlydiffer from output per labour year divided by theaverage annual full-time working hours.

4. We measure the growth of labour productivity inreal terms because the growth in nominal termsincludes changes in prices in GDP and this givesbiased information about the performance oflabour as a production factor. In the period 1991–2001 prices in services increased substantiallywhereas the price levels in the industrial sectorremained more or less constant. If we had usedthe nominal figures instead of the real figures,the increase in labour productivity in regionswith a high share of services would have beenoverestimated.

5. Like the labour input in our productivity measure,this implies that all jobs are converted to full-timeequivalent jobs.

6. See Groningen Growth and Development Centre,Total Economy Data Base, at <www.ggdc.net>.

REFERENCES

A

tzema

, O.A.L.C. & J.

van

D

ijk

(2005),

The Persistenceof Regional Unemployment Disparities in the Nether-lands

. Heidelberg: Springer.B

arro

, R.J. (1991), Economic Growth in a Cross-country Section of Countries.

Quarterly Journal ofEconomics

106, pp. 407–443.B

ils

, M. & P.J. K

lenow

(2000), Does SchoolingCause Growth?

American Economic Review

90,pp. 1160–1183.

B

roersma

, L. & J. O

osterhaven

(2004),

RegionaleArbeidsproductiviteit: Niveau, Groei en Verklaring.

Groningen: University of Groningen, Eindrapportaan het Ministerie van Economische Zaken.

B

roersma

, L. & J.

van

D

ijk

(2003),

Arbeidsproductiviteitin Fryslân.

Groningen: University of Groningen,URSI-rapport 305.

C

iccone

, A. (2002), Agglomeration Effects in Europe.

European Economic Review

46, pp. 213–227.D

onselaar, P., H. Erken & L. Klomp (2004), R&Dand Innovation: Drivers of Productivity Growth. In:G. Gelauff, L. Klomp, S. Raes & T. Roelandt,eds., Fostering Productivity: Patterns, Determinants andPolicy Implications, pp. 75–91. Amsterdam: ElsevierScience.

EU (2004), Employment in Europe 2004. Luxembourg:European Commission.

EZ (2004), Pieken in de Delta. The Hague: Ministry ofEconomic Affairs.

Klomp, L. & T. Roelandt (2004), Innovation Per-formance and Innovation Policy: The Case of theNetherlands. De Economist 152, pp. 365–374.

McGuckin, R.H. & B. van Ark (2004), Performance2002: Productivity, Employment and Income in theWorld’s Economies. The Conference Board, ReportR-1328-03-RR, New York.

Van Dijk, J. & S. Bosch (2003), Firm Related Train-ing in a Tight Regional Labour Market. In: E. Schamp& V. Lo, eds., Knowledge, Learning and Regional Devel-opment pp. 83–105. Münster/Hamburg/London:Lit Verlag.

REGIONAL DIFFERENCES IN LABOUR PRODUCTIVITY IN THE NETHERLANDS

Documents

Transcript of REGIONAL DIFFERENCES IN LABOUR PRODUCTIVITY IN THE NETHERLANDS