Active labour market policies and the efficiency of the European Social Fund in Spanish regions

ABSTRACT

Active labour market policies and the efficiency of the European Social Fund in Spanish regions. Regional Studies. This paper estimates the determinants of public expenditure on active labour market policies (ALMPs) at the regional level using panel data from a sample of the 17 Spanish regions (1989–2010). The estimations take into account endogeneity, dynamics and spatial dependence among regions. Estimations’ outcome is interpreted with the help of a simple theoretical model of intergovernmental grants. Results reveal that the European Social Fund is not effective in its aim of promoting ALMP expenditure and show that rent-seeking when regions compete to attract funds from a common pool provokes the opposite effect, which is exacerbated in the presence of imitation.

摘要

积极的劳动市场政策与欧洲社会基金在西班牙区域的效益。Regional Studies. 本文运用十七座西班牙城市（1989年至2010年）案例的面板数据，评估区域层级的积极劳动市场政策的公共支出之决定因素。该评估考量区域的内生性、动态与空间依赖。评估结果，则借助简易的跨政府基金理论模型进行诠释。研究结果显示，欧洲社会基金就其提倡ALMP支出的目标而言并不有效，并显示当区域彼此竞争以吸引共同来源的基金时的寻租行为，引发了反面效应，并在模彷中进一步恶化。

RÉSUMÉ

Politiques actives du marché du travail et efficacité du Fonds social européen dans les régions d’Espagne. Regional Studies. La présente communication effectue une estimation des déterminants des dépenses publiques sur les politiques actives du marché du travail à l’échelon régional, en utilisant des données de panel d’un échantillon de 17 régions d’Espagne (1989–2010). Les estimations prennent en considération l’endogénéité, la dynamique et la dépendance spatiale entre les régions, et les résultats des estimations sont interprétés à l’aide d’un simple modèle théorique de subventions intergouvernementales. Les résultats révèlent que le Fonds social européen n’est pas efficace dans son objectif de promotion des dépenses relatives aux politiques actives du marché du travail, et montrent que la «chasse aux rentes», lorsque les régions se font concurrence pour l’acquisition de fonds dans une réserve commune a un effet opposé, exacerbé par la présence de l’imitation.

ZUSAMMENFASSUNG

Aktive Arbeitsmarktpolitik und die Wirksamkeit des Europäischen Sozialfonds in spanischen Regionen. Regional Studies. In diesem Beitrag werden die Determinanten von staatlichen Ausgaben für aktive Arbeitsmarktpolitik auf regionaler Ebene mithilfe der Paneldaten einer Stichprobe von 17 spanischen Regionen (1989–2010) untersucht. Bei den Schätzungen werden die Endogenität, die Dynamik und die räumliche Abhängigkeit zwischen Regionen berücksichtigt. Die Ergebnisse der Schätzungen werden mithilfe eines einfachen theoretischen Modells von Finanzausgleichszahlungen interpretiert. Aus den Ergebnissen geht hervor, dass der Europäische Sozialfonds das Ziel der Förderung von Ausgaben für aktive Arbeitsmarktpolitik nicht wirksam erreicht und dass Rent-Seeking beim Wettbewerb zwischen Regionen zum Erhalt von Mitteln aus einem gemeinsamen Pool das Gegenteil bewirkt, was sich im Fall von Imitationen noch verschlechtert.

RESUMEN

Políticas activas del mercado laboral y la eficiencia del Fondo Social Europeo en las regiones españolas. Regional Studies. En este artículo se calculan los determinantes del gasto público en las políticas activas del mercado laboral (PAML) de ámbito regional mediante datos de panel de una muestra de 17 regiones españolas (1989–2010). La estimación tiene en cuenta la endogeneidad, la dinámica y la dependencia espacial entre regiones. Los resultados de las estimaciones se interpretan con ayuda de un sencillo modelo teórico de transferencias intergubernamentales. Los resultados indican que el Fondo Social Europeo no fomenta de modo eficaz el gasto de las PAML y muestran que la búsqueda de rentas cuando las regiones compiten por atraer capital desde un fondo común provoca el efecto contrario, lo que se agrava en presencia de imitación.

KEYWORDS:

• active labour market policies (ALMPs)
• European Social Fund (ESF)
• spatial panel data
• intergovernmental relations

关键词:

• 积极的劳动市场政策（ALMPs）
• 欧盟社会基金
• 空间面板数据
• 跨政府关系

MOTS-CLÉS:

• politiques actives du marché du travail (ALMP)
• fonds social européen
• données de panel spatiales
• relations intergouvernementales

SCHLÜSSELWÖRTER:

• aktive Arbeitsmarktpolitik
• Europäischer Sozialfonds
• räumliche Paneldaten
• Beziehungen zwischen Regierungen

PALABRAS CLAVES:

• políticas activas del mercado laboral (PAML)
• Fondo Social Europeo
• datos de panel espacial
• relaciones intergubernamentales

JEL:

• C23
• H53
• H77
• R19

ACKNOWLEDGEMENTS

This paper benefited from discussions had at the XXXth AIEL National Conference of Labour Economics, in Sardinia, Italy, and the 2015 Annual Meeting of the Association of Southern European Economic Theorists, in Granada, Spain. The valuable contribution of two anonymous referees is also gratefully acknowledged.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

SUPPLEMENTAL DATA

Supplemental data for this article can be accessed at https://dx.doi.org.10.1080/00343404.2017.1315396

ORCiD

Juan González-Alegre http://orcid.org/0000-0002-6637-8511

Notes

1. Hines and Thaler (1995) and Bailey and Connolly (1998) include interesting literature reviews.

2. Pallesen (2006) and González-Alegre (2012) are an exception.

3. Such as the budgetary objectives stated by the European Commission for every programming period or the limitations of the process of verification of additionality.

4. As well as many empirical studies identifying the existence of spatial interactions of public expenditures at the state (Baicker, 2005) or local (Sole-Olle, 2006) level.

5. The standard spillovers model (Case et al., 1993) assumes that a citizen of a particular region benefits also from the provision of a public good in neighbouring regions; the existence of yardstick competition among regional governments (Besley & Case, 1995) implies that voters use neighbouring jurisdictions to evaluate the performance of elected officials; in the Tiebout competition (Wildasin, 1988), regions compete to attract residents from neighbouring jurisdictions with the purpose of enlarging the tax base.

6. As described by Brueckner (2003).

7. Based in Sole-Olle (2006), but introducing the weighting according to distance.

8. The regulation of the ESF’s distribution is extremely cumbersome, but details to confirm this statement can be found, for example, in the National Reference Strategic Framework for 2007–13 (European Union, 2007) and in the corresponding operational programmes.

9. This model, in contrast to theirs, does no internalize taxation and private consumption. Likewise, for the sake of simplicity, the dynamics considered by Park et al. (2005) are not introduced here either.

10. This assumption is in line with the findings in Milio (2007) for Italian regions, as the public administrations often share the same resources (basically, civil servants) for launching new programmes and search for funding.

11. In contrast to unconditional lump-sum grants.

12. This literature emerged in the 1960s, with Bradford and Oates (1971) being a milestone that includes the main results generally accepted by the subsequent literature.

13. In fact, this sort of verification could be officially implemented, as revealed in REGIO (2011).

14. In fact, as shown by Svensson (2000), the mere expectation of larger resources can incentive larger rent-seeking effort.

15. The category is labelled as ‘regional public expenditure on social promotion’; it captures public expenditure on active labour market policies, among other issues. Unfortunately, the level of breakdown available does not allow the isolation of ALMP from other policy areas. Nevertheless, the examination of recent databases on the public regional budget with better detailed accounting reveals that policies related to the labour market represent the vast majority of this type of expenditure.

16. This is a common strategy for most regressions using annual country-level fiscal variables. Alternatively, many studies make use of variables expressed in a frequency larger than one year to omit short-term dynamics (e.g., Bleaney, Gemmell, & Kneller, 2001).

17. The selection of the model was made on the basis of the outcome of the several spatial panel autocorrelation tests and preliminary estimations. In particular, the Lagrange multiplier (LM) spatial autocorrelation test (reported in the tables) rejects the null of no general spatial autocorrelation, and the absence of spatial autocorrelation is also rejected by Moran’s I-test, while the robust LM test cannot reject the null of no spatial lag in the dependent variable. Most tests outcomes are not reported for the sake of simplicity.

18. The use of physical distance instead of infrastructure-based measures ensures the exogeneity of the criteria. Each element (w_ij) of this matrix, whose diagonal consists of zeros, represents the inverse of the distance between two regions (i and j), normalized so that every row add to 1 (i.e., w_ij = [1/d_ij]/Σ_j [1/d_ij]).

19. The direct interpretation of coefficient estimates from the spatial Durbin model is incomplete due to the simultaneous impact of spatially lagged dependent variables in the dependent variable and its spatial lags. For that reason, LeSage and Pace (2009) define a partial derivative approach commonly followed in the literature. The description of the process is too long to be presented here. Nevertheless, the derivation of equation (9) illustrates intuitively the process.

20. Moral-Benito (2011) shows that the Arellano–Bond GMM estimator with time-persistent variables may underperform with respect a quasi-maximum likelihood estimator in small samples. In fact, Moral-Benito, Allison, and Williams (2017) develop a panel data estimator that considers predetermined explanatory variables and behaves better than the Arellano–Bond in that scenario. Unfortunately, they have not yet developed a version for spatial panels. For the sake of robustness, Table 3 has been re-estimated using the quasi-maximum likelihood estimator for spatial panels developed by Belotti, Hughes, and Mortari (2016). Results, not reported here but available from the author upon request, do not change dramatically except for the significance level of the spatially lagged regional expenditure. The parameter δ, which is estimated separately, is still strongly significant, however.

21. Two examples in which the dependent variable is, precisely, public expenditure in ALMP expressed as a share of GDP are Rueda (2005) and Vlandas (2011).

22. In fact, some of the studies referred estimate was a positive coefficient whose absolute value is close to 1.

23. Also using data from the 17 Spanish regions.

24. As an increase in the availability of funds E promotes the raise of µ_i in all regions, which turns into a lower provision of the public good a (AMLP) also because of the imitation effect.

25. Van Vliet and Koster (2011) find, instead, a significantly positive coefficient for the variable GDP per capita.

 

Additional information

Funding

This work was supported by the Cátedra Pasqual Maragall de Economia y Territorio under the Ayudas a la Investigación 2012 grants programme; the Generalitat de Catalunya AGAUR research grants programme [grant number 2014 SGR 1326]; and the Spanish Ministry of Economy and Competitiveness [grant number ECO2014-52999-R].