330 likes | 428 Vues
JOB SUBSIDIES AND CUTS IN EMPLOYERS’ SOCIAL SECURITY CONTRIBUTIONS: THE VERDICT OF EMPIRICAL EVALUATION STUDIES. Ive Marx Centre for Social Policy University of Antwerp, Belgium. Spending on targeted employment subsidies, 2001. Source: OECD (2003). Our purpose.
E N D
JOB SUBSIDIES AND CUTS IN EMPLOYERS’ SOCIAL SECURITY CONTRIBUTIONS: THE VERDICT OF EMPIRICAL EVALUATION STUDIES Ive Marx Centre for Social Policy University of Antwerp, Belgium
Spending on targeted employment subsidies, 2001 Source: OECD (2003)
Our purpose • To bring together empirical findings: • What is known about the impact on employment ? • What is known about subsequent mobility trajectories ?
Types of employer subsidies • Roughly two types: • permanent but typically modest subsidies/ employer SS reductions applying to low-paid workers • temporary but typically more substantial subsidies/SS reductions for employers hiring people deemed at high risk of structural labour market exclusion
Gross impact: take up • There appears to be a connection between the scope, the generosity and the duration of initiatives. • Relatively generous measures with a relative broad scope involving a substantial subsidy or reduction generally generate a greater response. • In the case of more strongly targeted measures, take up rates are often considerably lower, often well below 50 per cent
Reasons for non take up • Ignorance or inadequate information • particularly a factor in small and medium-sized companies • multitude, temporary nature and complexity also factors • (Perceived) bureaucratic fuss and cost • Voucher schemes more effective for targeting very specific groups • downside: stigmatisation
Gross to net impact: intervening factors • Deadweight losses • Substitution effects • Displacement effects
Three evaluation methods • Employers surveys • Time series analysis • Experiments
Net impact Employer survey based evidence
Deadweight losses • Losses which occur when subsidised individuals would have found employment had the subsidy not been in place • Estimated losses are found to be consistently and considerably higher than what is assumed in theoretical models and simulations • Snower (1994) in ‘Converting unemployment benefits into employment subsidies’ assumes a deadweight cost of 25 per cent
Substitution effects • Targeted groups need to be clearly demarcated and this inevitably leads to distortions at the margins, i.e. subsidized workers being recruited at the expense of non-subsidised employees • Empirical studies generally find substitution to be significant
Displacement effects • job losses which occur through distortion of competition, i.e. job losses caused by the fact that enterprises that do not receive subsidies lose market share. • difficult to estimate using employer surveys; respondents are usually unable to attribute an increase or decrease in market share to this single factor.
Employer surveys: methodological limitations • Response rates tend to be low • Estimates rely on statements by an interested party – the employer
Net impact Time series based evidence
Time series analysis • The approach taken in these studies is to ascertain whether the introduction of a particular measure coincided with additional job growth (or slower job destruction) that could not be attributed to any other measurable factor. • The value of these estimations depends on the thoroughness with which one tests for other potential explanatory factors, such as cyclical movements of the economy.
Time series analysis: additional problems • Usually there are multiple policy changes at the same time the effects of which are difficult to disentangle • e.g. subsidies are sometimes coupled with training or additional job seeking support • Never a ‘perfect’ counterfactual / reference group available
Net impact Experimental evidence
A classic: Dayton experiment • A controlled experiment conducted in 1980-1981 by the US Department of Labor among a group of welfare recipients • One group was given a tax credit voucher, which entitled the employer hiring the vouchered job applicant to a tax credit (of the magnitude of the earlier discussed Targeted Tax Job Credit) • A second group was given a voucher that entitled the employed to a direct cash subsidy • A third control group was not given a voucher, even though they qualified in principle. • The individuals were assigned to one or the other group randomly, so that the groups were comparable in terms of composition. • All three groups got two weeks of job search training • Efforts were undertaken to make sure that the three groups did not receive different treatments from administrators or trainers
Findings from experimental research: Dayton Voucher Experiment Source: Burtless (1985)
Experimental research: issues • In theory the superior method since only treatment and not (unobserved) composition effects explain differential outcomes • But…. • a degree of contamination is difficult to prevent (in reality it is difficult to deal seperately but equally with treatment and control groups) • experimental studies are expensive and time-intensive • ethical issues: is it ethical to deny potentially positive ‘treatment’ to equally entitled people for the sake of science ?
From subsidised to regular work Evidence on mobility
Evidence on mobility • Number of studies available up until recently • Data requirements are substantial • Collecting adequate longitudinal data is costly and methodologically complex. • Increasingly studies use administrative data (for example from social security databases) • These allow longer term analysis (say over several years) at reasonable cost. • The problem is that such data bases often do not contain information on all relevant variables
Why limited mobility ? • beneficiaries get locked into their subsidised jobs • stigmatisation • job experience acquired inadequate to escape from the ‘productivity trap’
Conclusion: prior remark • very significant differences between the measures reviewed across a very wide range of dimensions • same applies to evaluation methodologies
Conclusion • The measured net employment effects of targeted subsidies and employer SS reductions tend to be consistently and considerably smaller than what ex-ante evaluations (i.e. theoretical models and simulations) invariably suggest • The empirical evaluation literature remains ambiguous as to the longer term impact of spells in subsidized employment – exact design features seem to matter a lot
Policy implications • More scepticism vis-a-vis ex-ante policy evaluation warranted • Need for systematic empirical evaluation • Evidence based policy adjustment