Active and passive labour-market policies: the outlook from the Beveridge curve

ABSTRACT

Following a panel ARDL approach, we appraise the impact of various indicators of active and passive labour-market policies within the framework of the Beveridge curve across fourteen OECD countries from 1985 to 2013, controlling for other factors, both institutional (tax wedge) and structural (technological progress, globalization). We embed the role of these variables within the specification of the Beveridge curve, finding that the generosity of unemployment benefits has a detrimental impact on labour-market matching, with the duration of benefits and the strictness of the rules pertaining to the deployment of benefits taking a key role in driving this result. Among active labour-market policies, employment incentives and especially training have a favourable effect on matching. There is evidence of a virtuous interaction between active and passive policies. A significantly detrimental role emerges for the tax wedge. These results are consistent across various specifications, and structural relationships are stable throughout the 2008–2013 period.

KEYWORDS:

• Unemployment
• vacancies
• training
• unemployment benefits

JEL CLASSIFICATION:

• E24
• J20
• F60
• O40

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Two recent and perceptive accounts of this literature are provided in Goulas and Zervoyianni (2018) and in Pignatti and Van Belle (2018).

² In fact, Bova, Tovar Jalles, and Kolerus (2018) relied heavily on temporal disaggregation procedures to convert data that are available only on an annual basis into quarterly series. As is well known (see, for instance, Bisio and Moauro 2017), the resulting series may be subject to measurement error, linked to the appropriateness of the benchmark quarterly variables utilized and to other features of the disaggregation procedure.

³ These points are treated in detail in Dumont et al. (2005) and Schmidt (2011).

⁴ These are basic measures of dispersion for age-, area-, skill-, or sector-specific employment; in the past, unemployment and vacancies have also been considered, besides employment.

⁵ These effects are considered in some detail for the US Beveridge curve in Dickens and Triest (2012). Dickens and Triest pointed out that shifts in the Beveridge curve during a recession may be only temporary, a possibility that was also highlighted in the historical analysis conducted by Diamond and Şahin (2014).

⁶ As emphasized by Nickell et al. (2003) and Bouvet (2012), endogeneity of the labour-market institutions and of the vacancy rate is likely to characterize the estimation of the Beveridge curve.

⁷ Bova, Tovar Jalles, and Kolerus (2018) used the GDP, whereas Arpaia, Kiss, and Turrini (2014) took a function of the number of the unemployed. The number of the unemployed and the labour force were taken in related contexts by Arranz, Garcia Serrano, and Hernanz (2013) and Goulas and Zervoyianni (2018), respectively.

⁸ Job search assistance, as well as services and sanctions, also emerged as significant factors in these papers. This is not vindicated in our analysis. It could be argued that this happens because our data for public employment services and administration include administrative expenses that have little correlation with job search assistance or with services and sanctions. Furthermore, we find a significant role for training, which is highly correlated with the expenditures on public employment services and administration.

⁹ To compute these values, we use the actual (not rounded) parameter values from Table 3 and focus on the long-run parameters, which, for NRW1, INCENTIVES_U, TRAINING_U, and TW, are semi-elasticities. We multiply these semi-elasticities by the mean values of the unemployment rate to obtain long-run derivatives, with which we calculate absolute changes (in percentage points) from the mean value of the unemployment rate. Dividing these absolute deviations by the mean value of the unemployment rate yields our measure of relative (percentage) changes.

Additional information

Funding

The work was supported by the Funda??o para a Ci?ncia e a Tecnologia [UIDB/00315/2020].