Local responses to labor demand shocks: A Re-assessment of the case of Italy

Highlights

• Labor mobility is crucial for regional convergence in the employment rate.

• We focus on Italy, a country with lasting labor-market disparities across regions.

• Population mobility is hampered by wage rigidity and low housing supply elasticity.

• Acting on wages and housing supply elasticity is a first-order policy option.

• Place-based policies could be an alternative, but have shown limited effectiveness.

Abstract

The paper studies how local economies react to labor demand shocks. It focuses on the case of Italy, a country with lasting labor-market disparities in the context of rigid local wages and limited housing supply elasticity. Results document that allowing greater flexibility in local prices is a first-order option to reduce labor-market heterogeneity, while place-based policy seems to be a less viable alternative.

Introduction

Aggregate nation-wide trends in employment are often the result of heterogeneous local dynamics. For instance, Census data from the Italian Local Labor Markets (LLMs) show that the interquartile range in the local private employment to population ratio increased by almost 6 percentage points between 1971 and 2001. One of the main reasons is that areas characterized by a different economic structure tend to be hit by different labor demand shocks. Those specialized in sectors with worse performance will be more likely to show a decline in their relative employment with respect to the rest of the country, while those specialized in expanding sectors are likely to climb the ladder.

Nevertheless, the difference between the local and the aggregate employment and population dynamics depends on the reallocation of factors and the adjustment of local prices (see, for instance, Blanchard and Katz, 1992). Migration plays an important role, because the local unemployment toll may be alleviated if those who lose a job move elsewhere. With perfect mobility, individuals should move from declining to expanding LLMs in order to gain from better chances of getting a job. In other terms, population movement should be sufficient to avoid an increase in the geographical heterogeneity of the employment over population ratio. Mobility is also highly influenced by changes in relative wages and house prices. Take, for instance, an area where local labor demand is declining with respect to the overall nation-wide trend. This relative negative shock might be counterbalanced if wages can be reduced (or if their overtime increase can be contained with respect to other areas), thereby increasing the local comparative advantage. On the same line, a decrease in house prices may counterbalance the loss in the attractiveness of the area.

Starting from Bartik (1991) and Blanchard and Katz (1992) a large literature assessed how regional and local economies react to employment shocks. A non-exhaustive list includes Decressin and Fatás (1995), who compare European adjustments to the US ones, finding that migration plays a less relevant role in the former; Bentivogli and Pagano (1999) whose evidence suggests that the elasticity of labor mobility to both unemployment and income (and its variance) is stronger in the US than in Europe; Bound and Holzer (2000), who study the role of mobility across different groups of the population in the US during the 80s. A recent stream of literature, stimulated by the evidence that in more recent years the US exhibited declining migration rates (Molloy et al., 2011), has started to reassess the adjustment of regional labor markets. Autor et al. (2013) show that population mobility has not been sufficient to compensate the job-loss in areas specialized in sectors affected by the exposure to trade with China, although this result is questioned by Monras (2015), who argue that during downturns shocks are mostly absorbed by reduced in-migration rates, rather than out-migration. Dao et al. (2014), in extending the Blanchard and Katz's analysis to a longer period and to a comparison with European regions, suggest that the population response in the US has been decreasing since the nineties. Beyer and Smets (2015) provide evidence of a convergence in the adjustment process of regional LLM within Europe and the US, due to an increased reactivity of population in the former and a reduction in the latter. Détang-Dessendre et al. (2016), using a variation of Bartik shocks, suggest that in recent years the flexibility of French local labor markets is not far from the one in the US.¹ Notowidigdo (2011) highlights the non-linearity of the employment and population elasticity to the shocks. He shows that negative shocks are at least partially compensated by the strong decline in house prices and by the adjustment in nominal wages, while positive shocks tend to have a less strong impact on prices. As a consequence, the adjustment of employment and population is stronger for positive than for negative shocks. Dao et al. (2017) find similar results by looking at interstate mobility in the US. Amior and Manning (2018) try to explain the strong persistence of the local employment to population rate in the US commuting zones. The authors focus on decadal long adjustments in population using Bartik shocks as a source of exogenous variation and estimate a dynamic model where adjustments also depends on the initial level of the employment rate and not only on employment growth. Their findings suggest that the main driver over the long run is the high autocorrelation of the local labor demand shocks.

We contribute to this literature in two respects. First, following the recent stream of studies, we re-assess the local responses to employment shocks in Italy, which can be considered as a case-study for countries with limited nominal wages adjustments. Using the traditional Blanchard and Katz (1992) framework, but discussing also issues raised in more recent papers (Notowidigdo, 2011; Amior and Manning, 2018), we show that the rigidities in housing supply and wage bargaining lead to a low population mobility in comparison with other countries. Furthermore, we find that local employment is more sensitive to negative rather than positive shocks. This result is the opposite of what Notowidigdo (2011) finds for the US and is at odds with the prediction from a spatial equilibrium model with concave housing supply, but it can be explained by the presence of local wage rigidities.

Secondly, we use the estimated local responses to discuss the relative merits of place-based policies (PBP) versus interventions intended to allow greater responses of local prices to shocks. With respect to the purpose of reducing geographical inequality in the employment rate, the two kinds of policy could be seen as substitutes. In fact, by means of simulations based on our estimates, we show that both an increase in population reactivity to the US level and a sequence of PBP increasing employment in the lagging-behind areas would have had a similar counteracting effect on LLM heterogeneity. In the first scenario the increased mobility raises the employment rate in declining LLMs and limits its growth in expanding ones. Differently, in the second scenario the PBP-driven boost in employment ends up raising the local employment rate, because in a context of limited mobility the increase in the number of jobs goes mostly to the benefit of local residents (Kline and Moretti, 2014; Neumark and Simpson, 2015). Although we are unable to provide a full assessment of the different alternatives, which would require a structural model calibrated on the Italian regional economies, we nevertheless illustrate that in the Italian context there seems to be only a limited rationale for using PBP to reduce labor-market geographical heterogeneity. The main reason is that the empirical literature on Italian PBPs highlights a very limited effectiveness in raising local employment. This implies that reaching the same effect on spatial inequality of increasing mobility to the US level, as in the simulations, would entail a very high fiscal cost. Furthermore, while Austin et al. (2018) suggest that there might be a rationale in intervening in lagging-behind areas if they are more reactive to PBP, we show that in Italy the elasticity of local employment to shocks is very similar across places, irrespective of their initial level of development. More in general, our estimates highlight that the rigidities in wage and house prices adjustments play a crucial role in impeding the convergence of different areas. Policies aimed at reducing the strength of housing regulation and the bite of collective bargaining might therefore be preferable to PBPs in order to achieve convergence in the employment rate.

Overall, our findings confirm results from a large literature that highlights the limited mobility of population in Italy. Several papers analyze the responsiveness of population and migration to regional changes in employment (or unemployment), wages and social transfers. Decressin and Fatás (1995) use a framework similar to Blanchard and Katz (1992) and find that the regional migration response to employment changes is very limited in Italy compared to other countries, although they do not instrument employment using a proxy of demand shocks. Faini et al. (1997) discussed the puzzling evidence of increasing unemployment North-South differentials and decreasing migrations. Brunello et al. (2001) showed that migration from the South to the rest of the country shrunk in the 80s and 90s because of reducing wage differentials and increasing transfers, although only part of the divide could have been absorbed if migration had been as it was during the 70s. We contribute to this literature focused on Italy in several respects. First of all, we are the first to analyze the local responses to labor demand shocks at the LLM level over a long period of time (1971–2011). We believe that the LLM is a more interesting unit of spatial analysis than regions, because it is designed to represent an approximately self-contained area in terms of commuting. Furthermore, once we account for the strong differences between the South and the Centre-North, around 2/3 of the remaining variance of the time-demeaned private employment to population rate at the LLM level is within regions. Secondly, even if we use a simplified empirical specification that does not account for dynamics, we directly focus on a the reactivity of employment, population, house prices and wages to a plausibly exogenous proxy of labor demand shocks, while previous papers generally looked at the relation between the different outcomes using their lagged values as instruments. We also study non-linearities of the LLM adjustments to demand shocks, which have not been discussed in any of the previous papers for Italy. Finally, we make use of the estimated reactions to shocks to simulate and compare two hypothetical policy options.

The paper is structured as follows. Section 2 illustrates the empirical framework and data. Section 3 re-assesses the local reactivity in Italy, jointly looking at employment, population, and local prices (wages and house prices). Section 4 proposes some policy-motivated simulations and discusses the rationale behind PBP aimed at reducing the strong geographical heterogeneity. Section 5 concludes, highlighting the links between our estimates and the previous literature.