Changes in the organizational structures of enterprises, technical progress and shifts of preferences constitute the major sources of structural economic change. Nowadays, it is well recognized that labour-market dynamics, induced by the processes of structural change, can be understood only by considering the relevant flows of people and of jobs in the labour market. Research on the flow approach to labour markets is conducted along two routes: collecting data on gross flows, using panel data on individual firms or macroeconomic time-series data; and building models of labour-market flows. To bring together these two routes of research, and to investigate their policy implications, Gerard Van den Berg (Free University Amsterdam, Tinbergen Institute and CEPR) and Frank den Butter (Free University Amsterdam and Tinbergen Institute) organized a joint CEPR/TI workshop on ‘Structural Change, Labour Market Dynamics and its Policy Implications’. The workshop was held at the Tinbergen Institute in Amsterdam on 22/23 November 1996.

Tito Boeri (IGIER, Università Bocconi and CEPR) addressed two questions in his paper ‘Labour Market Flows in the Midst of Structural Change’. Why is there both a high pace of structural change and low unemployment turnover in the Eastern European transition economies? And why had there not been an increased rate of exits from unemployment as a result of reductions in the generosity of unemployment benefits (and the accompanying likely drop in reservation wages)?

The model Boeri developed to address these issues was based on the notion that, in the former socialist economies, it was very difficult to determine the individual labour product of workers. This was owing to, among other things, a highly compressed wage distribution, and narrowly based and rigid curricula in the vocational education system. Hence, unemployment persistence in Eastern European countries arises from informational asymmetries with respect to the actual productivity of former state workers. Because employers have no information on the productivity of an individual unemployed job-seeker, they are likely to rank employed above unemployed job-seekers. For Ronald Schettkat (University of Utrecht), another explanation might be a low job-continuation probability and, hence, a shorter vacancy chain in Eastern European countries.

Alison Booth (University of Essex) presented a paper, jointly written with Marco Francesconi (University of Essex) and Carlos Garcia Serrano (Universidad de Alcalá de Henares), entitled ‘Job Tenure: Does History Matter?’. The main questions posed were whether there was a difference in mobility patterns between people who entered the labour market at the beginning of the century and those who entered later, and whether their individual histories mattered. British men and women hold, on average, a total of five jobs during their careers; and half of all lifetime job changes occur in the first ten years. Job tenure increases with the number of jobs, but declines with the date of entry into the labour market. For later cohorts, the tenure patterns for males and females converge, but there remains a difference in the reasons for separation. Quits are important among men, whereas for women, family care seems to play a decisive role. Both factors decline with the number of jobs. The authors estimated non-monotonic job-separation functions, and showed that there is an increase in the hazard in the first months but a decline thereafter. Their conclusion was that both date of entry and personal history are important. For job tenure in later jobs, recent history is more important than the entire history of mobility.

Nathalie Greenan (INSEE) presented ‘Structural Change, Labour Market Dynamics and its Policy Implications’, jointly written with Dominique Guellec (OECD). The paper was about the relationship between job reallocation and product and process innovations. The main finding was that process innovation leads to more growth at the firm level than does product innovation. On the other hand, process innovation leads to employment losses at the sectoral level, while product innovation leads to sectoral employment growth. These findings are explained via a model in which process innovation is viewed as a positive supply shock, which decreases the price of the goods produced by the innovating firm, and product innovation is assumed to be a demand shock, because the variety of goods increases. Pieter Gautier (Free University Amsterdam) argued that new products often have non-rival designs and are only partly excludable, so that the entire sector can benefit from them.

Eric Bartelsman (Central Planning Bureau, The Netherlands) presented a paper jointly written with Martin Baily (Council of Economic Advisors, Washington) and John Haltiwanger (University of Maryland) and entitled ‘Labor Productivity: Structural Change and Cyclical Dynamics’. They explored how cyclical fluctuations influence average labour productivity. To date, most studies on labour productivity over the business cycle have used aggregated data; their study used a balanced panel on output, employment and hours for all large continuously operating plants in US manufacturing from 1972 to 1988. Their use of micro data made it possible to identify cyclical changes by using heterogeneity on the micro-level. They distinguished five possible sources of pro-cyclical labour productivity: labour hoarding, adjustment costs, external economies, increasing returns to scale and compositional changes. The authors showed that the pro-cyclicality of productivity is caused mainly by permanent downsizing. Plants that are upsizing in the long run exhibit little or no pro-cyclical productivity. This result favours an adjustment-cost model, although external economies may play a modest role. Labour hoarding and increasing returns to scale appear to play only a minor role.

Jan van Ours (Tinbergen Institute and CEPR) presented a paper jointly written with Jaap Abbring (University of Amsterdam) and Gerard van den Berg (Free University Amsterdam) on ‘Business Cycles and Compositional Variation in US Unemployment’. It is well known that, on average, the long-term unemployed have lower exit rates from unemployment than the short-term unemployed. The authors took a closer look at sorting and ranking as two possible reasons why the short-term unemployed leave unemployment first, and why the magnitude of the difference varies over the business cycle. Heterogeneity of the unemployed triggers a sorting process, e.g. those with high exit rates leave unemployment first. Alternatively, long unemployment duration may signal low skills, thus creating a ranking effect. In both cases, the short-term unemployed will be preferred. Ranking and sorting have different dynamic implications. An empirical duration model, which distinguishes between sorting and ranking, was used to study the interaction between cyclical fluctuations and exit rates. The data used to estimate the model were based on unpublished CPS data from the US Department of Labor. The analysis used time-series figures on monthly unemployment. Sorting was shown to be the dominant factor in explaining the interaction between the business cycle and duration-dependence during the first month of unemployment, while ranking was dominant during all longer durations.

Francis Kramarz (CREST, INSEE) presented ‘The Entry and Exit of Workers and the Growth of Employment: An Analysis of French Establishments’. This was a joint paper with John Abowd (Cornell University) and Patrick Corbel (INSEE), and was based on a representative sample of all French establishments with 50+ employees in 1987-90. The authors found that, in growing establishments, the hiring rate was about three times the job-creation rate, while the firing/quit rate was twice the job-creation rate. Hence, one new job was associated with three hires and two separations. By contrast, job destruction involved one hire and two separations. Workers’ exit rates did not differ much between expanding and shrinking establishments; the distinction was caused by differences in hiring rates. Clerical and blue-collar workers had much higher entry and exit rates than engineers and managers. The number of short-term contracts might have played a role here (the bulk of hires was into short-term contracts). Simultaneous entry and exit in an establishment occurred in over 50% of the months in the sample, but this high ratio might have been owing to the average size of the sample establishments and the occurrence of voluntary quits. Furthermore, simultaneous entry and exit occurred more often in lower-skilled groups.

Stefan Profit (Humboldt-Universität zu Berlin) presented ‘Matching Across Space: Evidence on Mobility in the Czech Republic’, written with Michael Burda (Humboldt-Universität zu Berlin). The paper extended the matching literature by allowing for spatial spillovers between local entities. Unemployment outflow in their model is influenced also by labour-market conditions in neighbouring regions. The authors found a U-shaped pattern in the influence of unemployment in the neighbouring regions. At distances of less than 30 km and more than 120 km, the effect on local matches was positive. Jozef Konings (Katholieke Universiteit Leuven) remarked that the behaviour of firms was modelled rather passively, and that the mobility of vacancies is also important in explaining regional matches.

Stephen Millard (Bank of England) presented a paper entitled ‘The Effects of Increased Labour Market Flexibility in the United Kingdom: Theory and Practice’. For Millard, labour-market flexibility depends on how easy it is to hire and fire workers. Increased flexibility is associated with lower fixed job-creation and job- destruction costs.

Millard considered three different theoretical equilibrium business-cycle models, all calibrated for the United Kingdom. According to all of the models, increased output and consumption, and reduced unemployment, should be expected when labour-market flexibility increases. Empirical analysis for the United Kingdom provided some evidence for this. Additionally, the theory predicts a (slight) rise in consumption volatility, and a decline in employment variance (owing to reduced unemployment incidence and increased duration). No evidence was found for this, however, from Millard’s comparisons of the three business cycles in the 1979-96 period. In fact, employment variance increased over this period. Frank den Butter wondered whether calibration of all three models was based on the same data, and asked to what extent differences in their outcomes could be ascribed to differences in calibration.

Alan Manning (London School of Economics and CEPR) presented his paper entitled ‘Movin’ On Up: Interpreting the Earnings-Experience Profile’. It has been generally accepted that wages are positively correlated with job tenure. The question, however, is whether productivity increases, or whether experienced workers are older and have better jobs because they obtained more job offers. According to job-search theory, wages are related to labour-market transition rates.

Manning used a structural job-search model to distinguish between experience and labour-market transition rates. The data came from 20 years of the UK General Household Survey and Labour Force Survey. The earnings-experience profile of males is shown to be concave, with very rapid growth during the early years, while the profile of females has a bump around 10–15 years of experience. It seems that the on-the-job search model is able to explain age-earnings profiles for individuals aged over 30, while the accumulation of human capital is an important determinant of wage growth for youths. Over time, the gap between the female and male profiles has fallen. The model does well in interpreting the earnings-experience profiles of both males and females. This suggests that the theory of search capital satisfactorily explains the gender gap in the profile, although this is true also of human capital theory. In discussion, the difficulty of distinguishing between search capital and luck was mentioned, as was the possibility that ranking could be used to explain the wage differences.

In his paper on ‘Worker Flows and Job Flows in Danish Manufacturing, 1980–91’, Karsten Albaek (University of Copenhagen) used information about every individual manufacturing plant in Denmark. In contrast to a similar study by Davis and Haltiwanger, which found a counter-cyclical job-reallocation rate, Albaek found no large changes in the pace of job reallocation over the business cycle. Albaek also found an apparently large degree of heterogeneity among workers: when a plant is downsizing, only one out of every four workers whose job is destroyed takes up the job of an otherwise separating worker. Lastly, the differences between large and small plants was investigated. It turned out that, in an economic upswing, quits from small plants increased in contrast to quits from large plants. Lourens Broersma (Tinbergen Institute, Amsterdam) noted that Albaek’s reallocation rates were higher than suggested by results for other countries, when similar data sets are used.