LOCLAB - an ANR-DFG funded project (2015-2019)
LOCal LABour markets: The causes and consequences of spatial differences in labour market outcomes across cities in France and Germany (contract ANR-15-FRAL-0007-01)
My work on the spatial mobility of workers is presented in a separate section: (spatial mobility).
Table of Contents
The end-of-project Conference
The end-of-project conference was held under the umbrella of the international conference cycle Spatial Dimensions of the Labour Market in Marseille on 11-12 July 2019. The invited keynote speakers were Frédéric Robert-Nicoud and Alan Manning.
Project Summary
LOCal LABour markets: The causes and consequences of spatial differences in labour market outcomes across cities in France and Germany
This research project has considered local labour markets in France and Germany, and has examined the causes and consequences of differences in local labour market outcomes. Why do workers’ earnings, firm-worker productivity, and housing costs exhibit such enormous differences between, say, Paris or Mannheim? To answer such a question we have examined the roles of each side of the local labour market: the behaviour of firms seeking to harness the gains in productivity stemming from the close proximity of firms in urban centres (« agglomeration economies »); the location decision of workers who trade-off wages, employment probabilities, and housing costs; as well as the interaction of these economic forces that lead to a spatial equilibrium.
We have addressed our research questions systematically and comprehensively in a theory-led empirical investigation using French and German administrative micro-datasets (DADs and LIAB). The research design has used a variety of empirical strategies, such as structural estimation based on economic modelling, and reduced form regressions.
Our discussion of the main research outcomes reflect the two main axes, agglomeration economies and worker mobility.
Summary 1: Productivity gains from agglomeration economies
We have estimated the static and dynamic agglomeration gains for the French economy. These estimates then made it possible to determine the productivity gains to be expected from the increased spatial concentration of economic activities in the Ile-de-France region linked to the installation of new transport infrastructures (the Grand Paris Express project). Our work shows that such gains are high, more than 10 billion euros per year in certain scenarios, but that a large fraction of these, up to 90%, could only come from transfers from other French regions. This occurs in particular in the case where the job creations due to the Grand Paris Express would mainly benefit former employees from French regions outside the Ile de France and do not correspond to net employment creation. A number of currently work-in-progress complement this first contribution. The first aims to quantify the static and dynamic agglomeration gains over a long period, since the 1950s. The second, carried out jointly on France and China, aims to compare the agglomeration gains with the cost of housing, in order to deduce real net earnings from urbanisation. Agglomeration costs appear to outweigh income gains for French cities, less for the Chinese economy. Finally, a third project studies the sensitivity of the estimates of agglomeration gains to the spatial nomenclature used, in particular with regard to a new strategy that we have developed to delimit urban areas on the basis of building density.
Summary 2: Labor market frictions and worker mobility
In a series of papers (spatial mobility) we have developed a general equilibrium life-cycle model of spatial job search across heterogeneous local labour markets in the presence of search frictions. Although local labour markets in the US and Europ exhibit substantial and persistent differences in terms of unemployment rates, nominal and net-of-housing-cost wages, as well as firm productivities, the observed spatial mobility of workers searching for jobs is low. This pattern has usually been framed as resulting solely from moving costs. However, to account for the observed geographic mobility, the implied moving costs should be extremely high. Stating the problem with a search-theoretic perspective, we establish a tractable model of location choice that accounts for the spatial dimension of search frictions. The model allows disentangling the different frictions that contribute to lowering geographic mobility, with a particular emphasis on the role of age. We estimated our model structurally using French and German administrative individual-level job transition data. Our results suggest first that job search frictions reduce internal migration much more than mobility costs. Second, mobility costs are more constraining for middle-aged workers than for young and senior workers.
Summary 3: Firm organisation and location
Firms are essential to the economic activity of cities. Yet, relatively little is known on how the firms’ vertical division of labor into layers of hierarchy varies across cities. Greg Spanos’ analysis bridges this gap. In “Firm Organization and Productivity across Locations” (Journal of Urban Economics, July 2019), he develops a theoretical model, based on the hierarchy framework of Garicano (2000) and Caliendo and Rossi-Hansberg (2012), which allows the size of local markets to affect the hierarchy of firms. The model implies that firms in larger markets organize into a greater number of layers, and, because they have more layers, firms in larger markets are more productive. Spanos then uses administrative data to confirm the model’s implications across French employment areas. He also observes that 8.8% to 22.4% of the log productivity gains from denser areas arise from differences in the organization of service firms. In the paper “Spatial Differences in Firm Wage Disparities: Firm Organization Matters”, he further shows that these productivity gains are mostly concentrated in the higher layers of firms. More generally, in the paper “Organization and Differences in Pay across Firms”, Spanos shows that differences in firm organization influence firms’ wage structure and account for an important share of the differences in wages associated with the size of local areas, and the size, productivity, and export status of firms.
Summary 4: Size distribution of cities.
Schluter and M. Trede (our German partner at the University of Munster) have published Size Distributions Reconsidered in Econometric Reviews. This paper looks at the general problem of the size distribution of objects, and demonstrates how Extreme Value Theory can be used to establish convincingly the tail behaviour of the size distribution. In an application to the size distribution of cities, they show that these are heavy tailed, i.e. Pareto- or Zipf-like (but not exactly so). Schluter has also published On Zipf’s law and the bias of Zipf regressions in Empirical Economics. He examines the properties of the tail exponent estimator obtained from ordinary least squares (OLS) rank size regressions (Zipf regressions for short), the most popular empirical strategy among urban economists. As city size distributions are not strictly Pareto, but upper tails are rather Pareto like (i.e. tails are regularly varying), the estimator is then biased towards Zipf’s law in the leading class of distributions. The Pareto quantile–quantile plot is shown to offer a simple diagnostic device to detect such distortions and should be used in conjunction with the regression residuals to select the anchor point of the OLS regression in a data-dependent manner. Applying these updated methods to some well-known data sets for the largest cities, Zipf’s law is now rejected in several cases.
Trionfetti revisits the city size distribution debate on « City size distribution: in search for an explanation ». In particular he compares the city size distribution of a sample of pre-agricultural, agricultural, and post industrial societies, finding that the city size distributions of these societies are statistically not distinguishable from one another. This is striking because millennia of economic history and industrial development separate these societies. Trionfetti then studies the organisational structure of three societies for which we have data: a post-industrial, an agricultural, and a pre-agricultural society. These three societies exhibit the same pattern of fractal organisational structure by which the same size-distribution repeats itself at different organisational layers. In related work, “Economic Aspects in the Oasis of Bukhara between Past and Present: An Archaeo-Economic Approach.” Federico Trionfetti and Rocco Rante use exceptionally good data on the urban system of the Bukhara region in the 9th century A.D. For this society they study the rank-size relationship of cities and the effect of the Silk Road on the structure of cities. They find that proximity to the Silk Road is associated with a larger “business district” of ancient cities. This is akin the the home market effect found in many modern economies. They also find that the rank-size relationship is concave like for modern economies but exhibits a Zipf coefficient only for a subset of cities (approximately 70% of the sample) in the middle of the distribution. This is somewhat different from the result found for modern economies that the Zipf law applies only to the largest cities.
Research Design
We use a variety of research designs. For instance, the dynamic spatial search models are estimated structurally: Using the model, we estimate the « deep primitives » (such as the technology and preference parameters), which then enables us to predict convincingly optimal location choices by workers, and then conduct policy experiments designed, for instance, to increase the spatial mobility of workers. By contrast, we investigate the static and dynamic agglomeration gains using empirical models exploiting panel data and specifications that include worker and firm fixed effects, and take into consideration the employment history of the worker. Instrumental variable technics allowed us to tackle possible endogeneity issues due to either missing variables or reverse causality arising when location choices are endogenous.
Contributions
We have attained these objectives in a coherent analytical framework which also considers the interaction of the discussed forces that lead to a spatial equilibrium of workers and firms. The ensuing empirical investigation has been based on this theory, and used high quality confidential administrative data for both France and Germany (the DADs and the LIAB).
Our empirical applications to the Franco-German context are new and contribute significantly to the established literature, specifically the estimation of empirical models using administrative employer-employee panels, as are the theoretical analyses of the spatial transmission and amplification mechanisms. For instance, our investigation of the static and dynamic agglomeration gains extend considerably the analysis of Puga and de la Roca (2012), as well as the results surveyed in Combes and Gobillon (2015). Our empirically estimated job search models introduce spatial search frictions to the seminal dynamic setting of Kennan and Walker (2012), and our modeling approach allows for individual level heterogeneity of workers, unlike Schmutz and Sidibé (2019).
The Team
This ANR-DFG project been coordinated by Christian Schluter (Aix-Marseille School of Economics, AMSE), with partners at AMSE (Federico Trionfetti, Greg Spanos (now at the University of Geneva), and Guillaume Wilemme (now at the University of Leicester)), and Pierre-Philippe Combes and partners at Lyon GATE LSE (Clément Gorin) and Paris School of Economics (Laurent Gobillon). The partners on the German side comprised the ZEW at Mannheim and the University of Munster.
