The Beveridge curve in the OECD before and after the great recession

Abstract

This paper analyses the Beveridge curve across twelve OECD countries from 1985 to 2013. Besides allowing for some customary labour-market institutions, we assess the role of technological progress and globalisation, as well as of the current recession, on the curve. Significant institutional variables include unemployment benefits, the tax wedge, minimum wages, and, to a lesser extent, employment protection legislation (the latter improving the unemployment-vacancies trade-off). Technological progress has a complex impact. R&D intensity worsens matching efficiency, producing evidence in support of a creative destruction effect, while an index of total factor productivity has the opposite effect. Unlike in previous work, we find a favourable impact of globalisation. The curve parameters are stable throughout the 2008–2013 period, once we allow for country-specific shifts in the curve coincidental with the Great Recession. On the other hand, there is no common shift of the curve that can be associated with the Great Recession.

Appendix

1.1 Table legends

The dependent variable is always the natural log of the unemployment rate. The vacancy rate, vr, is given by the ratio of vacancies over total employment. Among the Z variables, UB is the indicator for the generosity of unemployment benefits, TW the total tax wedge, ALMP the active labour market policy indicator (expenditure over GDP), EPL the employment protection legislation indicator, minW the minimum statutory wage (divided by the median wage), Coord_OECD an index of wage bargaining coordination, UD a measure of union density. Globalisation is measured by either ImpVa, the ratio of total manufacturing imports from non-OECD countries to manufacturing value added (both variables at current prices), or kaf, the KOF index of actual economic flows (allowing for external trade, capital flows and outsourcing), or Openness, a usual foreign trade indicator. Technical progress is measured both by r&d, a measure of business sector R&D intensity, and tfp, an index of total factor productivity. An initial l stands for a variable taken in natural logarithms, a _1 or a _2 termination indicates a first- or second-order lagged variable. Note that the index of R&D intensity turned out to be more significant if not logged: our reported estimates thus include its linear specification, unlike for the other indicators of technical progress and globalisation.

All estimates include country-idiosyncratic effects and year-specific effects, not shown in the interest of parsimony, in all specifications. Coefficient significances are denoted by stars: * means a p-value < 0.1; ** a p-value < 0.05; *** a p-value < 0.01. The r² is the coefficient of determination calculated as in Pesaran and Smith (1994). Diagnostics include the Arellano–Bond (AB (i)) test for first and second order serial correlation, and the Hansen J test of overidentifying restrictions. We provide p-values for all these tests.

Estimation is always carried out through a standard dynamic fixed-effects panel IV procedure. Instruments include, besides all lagged regressors, the following list of variables: lvr_2, EPL_1, EPL_2, minW_1, minW_2, lImpVa_1, lkaf_1, plus country-idiosyncratic linear trends, crisis dummies and linflow_1 when the estimated specification comprises trends, crisis dummies or the (log of the) inflow rate.

1.2 Data sources

Unemployment, labour force and employment data are taken from the OECD Statistic Portal. The unemployment rates are based on OECD standardised rates. Statistics for annual vacancies come from the OECD Registered Unemployed and Job Vacancies Dataset. Only for the US the vacancy rate before 2001 is obtained by extrapolating the OECD data through the composite Help Wanted Index (HWI) proposed by Barnichon (2010). The computation of the inflow rate was carried out by applying the procedure proposed by Elsby et al. (2013). For the estimation of this variable, we relied on the yearly and quarterly harmonised unemployment rates from the OECD Short-Term Labour Market Statistics Dataset, and, for the stock of unemployed workers by duration, on the OECD Unemployment by Duration Dataset.

For all the institutional variables, the primary data source was Nickell (2006), updated using information from the OECD Statistic Portal. More precisely, the generosity of unemployment benefits is taken from Allard (2005b), who uses OECD data to build an indicator combining the amount of the subsidy with its tax treatment, duration and the conditions that must be met in order to collect it. The total tax wedge is equal to the sum of the employment tax rate, the direct tax rate and the indirect tax rate. All three rates rely on information from the OECD National Accounts. The active labour market policy indicator (public expenditure as a percentage of GDP) relies on the series of Labour Market Programmes computed by the OECD. The overall indicator for employment protection legislation is taken from Allard (2005a), who basically follows the OECD methodology in computing it. Neither this indicator nor the generosity of unemployment benefit could be updated beyond 2013 because of the lack of basic OECD information. The minimum statutory wage is calculated, using OECD methodology and data, by converting statutory annual minimum wages into hourly wages. Then these wage rates are divided by median hourly wages.

Bargaining coordination is an index of bargaining coordination with range {1,5} taken from OECD (2004, Table 3.5). Its five categories range from Fragmented company/plant bargaining, little or no coordination by upper-level associates, to Coordination of industry-level bargaining by an encompassing union confederation or government imposition of a wage schedule/freeze, with a peace obligation.

Union density is union membership over employment, calculated using administrative and survey data from the OECD labour market statistics database.

In order to compute the globalisation measure ImpVa, total manufacturing imports from non-OECD countries are drawn from the OECD STAN Bilateral Trade Database and International Trade by Commodity Statistics, and value added from the OECD STAN Database for Industrial Analysis. We extrapolated through time-series techniques the 2012 and 2013 values for these indexes, which were not available from the original source. The KOF index for actual economic flows, kaf, comes from https://globalization.kof.ethz.ch. More information about them is provided in Dreher (2006). Openness, a traditional measure of trade openness, is obtained by taking the sum of exports and imports and dividing it by GDP (all data being from the OECD Statistics Portal).

R&D intensity, r&d, is measured by business enterprise expenditure on R&D as a percentage of GDP, taken from the Main Science and Technology Indicators, OECD. It represents the share over GDP of the intramural R&D expenditures within the business enterprise sector. Total factor productivity, tfp, is calculated as a Tornqvist index, using gross domestic output, employment, capital stock of the business sector and a smoothed share of labour. The source of the capital stock of the business sector (in fact, the private non-residential net capital stock) is the OECD Analytical Database, whereas gross domestic output and employment are drawn from the OECD Statistic Portal and the smoothed share of labour from the OECD Unit Labour Costs Dataset.