serafinelli.jpg
VoxEU Column Labour Markets Productivity and Innovation

Labour market-based knowledge spillovers: ‘Good firms’, worker flows, and local productivity

The productivity benefits of similar firms locating near one another are well accepted, but there is little agreement on how knowledge spillovers have local effects. This column presents evidence from Italy of how firm-to-firm labour mobility enhances the productivity of firms located near other, highly productive firms. The main finding is that the recruitment of workers with experience at good firms significantly increases the productivity of the firms hiring them.

A prominent feature of the economic landscape in advanced countries is the tendency for firms to locate near other firms producing similar products. Germany's Baden-Wuerttemberg and Britain’s ‘Motor Sport Valley’ are well-known examples of traditional manufacturing regions. The growth and diffusion of multinational corporations in the past two decades has led to the appearance of important new industrial clusters (Alfaro and Chen 2014). For instance, Slovakia is also known as the ‘Detroit of the East’ for the strong presence of the automotive sector.

Researchers have long speculated that firms in such industrial concentrations may benefit from agglomeration economies. Despite the difficulties involved in estimating agglomeration effects, a consensus has emerged that significant productivity advantages of agglomeration exist for many industries (Rosenthal and Strange 2003, Henderson 2003, Ellison et al. 2010, Combes et al. 2012, Baum-snow 2017). Disagreement remains, however, over the nature of the microeconomic mechanisms that can account for these advantages (Moretti 2011). This serves as a barrier to understanding differences in productivity across industry clusters and localities, and hinders the design of location-based policies (Kline 2010). Localised knowledge spillovers are one of the most commonly hypothesised sources of the productivity advantages of agglomeration, alongside the availability of specialised intermediate inputs, the sharing of a common labour pool, and better matching. Nevertheless, if information can easily flow out of firms, the question of why the effects of spillovers are localised must be clarified (a point well-made by Combes and Duranton 2006).

The importance of labour market-based knowledge spillovers

In a recent paper, I present direct evidence showing how firm-to-firm labour mobility enhances the productivity of firms located near highly productive firms (Serafinelli forthcoming). The paper lends support to the idea that the strong localised aspect of knowledge spillovers discussed in the agglomeration literature arises – at least in part – from the propensity of workers to change jobs within the same local labour market. Knowledge is partly embedded in workers and diffuses when workers move between firms. To fix ideas, I begin by presenting a simple conceptual framework in which some firms are more productive because they possess superior knowledge. The superior knowledge could include information about export markets, physical capital, new organisational forms, or intermediate inputs. Employees at these firms acquire relevant firm's internal knowledge – I refer to these employees as ‘knowledgeable’ workers. Other firms can then gain access to this superior knowledge by hiring such workers.

My central empirical goal is to measure the importance of labour market-based knowledge spillovers. I use a unique dataset combining Social Security earnings records and balance sheet information for Veneto, an administrative region in the Northeast of Italy (which contains Venice) with a population of around 5 million people (8% of the country's total). While the issues I analyse are of general interest, the case of Veneto is important because this region is part of a larger economic area of Italy where networks of specialised small and medium-sized firms, frequently organised in districts, have been effective in promoting and adapting to technological change during the past three decades. This so called ‘Third Italy’ region has received a good deal of attention by researchers in Europe and North America (Trigilia 1990, Whitford 2001, Piore 2009). The most famous industrial concentration in Veneto is the eyewear district in the province of Belluno, where Luxottica, the world's largest manufacturer of eyeglasses, has production plants. Manufacturing firms in Veneto also specialise in metal-engineering, gold-smithing, plastics, furniture, garments, textiles, leather, and shoes (Benetton, Sisley, Geox, Diesel, and Replay are Veneto brands).

Armed with these data, I show that firm-to-firm labour mobility happens more frequently within the same local labour market. I then identify a set of high-wage firms (‘good’ firms) and show that they are more productive than other firms. Using regression analysis, I then show that hiring a worker with experience at good firms increases the productivity of other firms. In order to understand the intuition behind my empirical strategy, it may be instructive to consider an illustrative example. Consider Firm 1 and 2, which are initially very similar in terms of observable characteristics. Firm 1 hires two workers from good firms in June 2000 (Worker A and Worker B). Firm 2 hires one such worker in June 2000 (Worker C). Workers A, B, and C are very similar in terms of observable characteristics. The questions are:

  • Is Firm 1 significantly more productive than firm 2 in December 2000 or in December 2001?
  • Is Firm 1 is significantly more productive in December 2000 (or December 2001) than it was in December 1999?

Given that workers do not move from firm to firm on a random basis, estimating the effect of recruiting a knowledgeable worker on the receiving firm’s productivity is difficult. Unobservable firm-level productivity shocks correlated with hiring may introduce biases in my econometric estimates (examples of such shocks are process innovations and new managerial techniques). I deal with these sources of bias using well-established control function methods drawn from the productivity literature.

Recruiting knowledgeable workers increases productivity

I find that, on average, recruiting a knowledgeable worker increases the productivity of a non-high-wage firm by between 1.8% and 3%. In interpreting my estimates, it is important to highlight that non-high-wage firms are quite small – their median number of employees is 33. Further, as many as 78% of non-high-wage firms in a given year do not employ any knowledgeable workers. Hiring one knowledgeable worker therefore implies a significant change in terms of workforce for most firms in my data.

Additional evidence supports the main finding that the recruitment of workers with experience at good firms significantly increases the productivity of the non-high-wage firms hiring them. I observe greater productivity gains in firms hiring workers in higher-skilled occupations. Moreover, the productivity effect of knowledgeable workers is not associated with recently hired workers in general – there is no similar productivity effect for recently hired workers with experience at firms that have lower productivity than the receiving firm. Finally, the productivity effect of knowledgeable workers does not appear to be driven by unobserved worker quality (I obtain a proxy for worker ability using estimates of worker fixed effects from wage equations).

It is also possible that knowledgeable workers are attracted to join firms that are ‘on the rise’ (i.e. firms that offer better prospects than the good firms at which these workers are employed), rather than knowledgeable workers moving to firms and causing the increase in productivity. To address this concern, I exploit variation in the number of knowledgeable workers employed by a non-high-wage firm arising from the number of good firms locally in the same industry that downsized in the previous year. Following a downsizing event at a good firm, it is more likely that a knowledgeable worker applies for a job at local non-high-wage firms because she is unemployed and does not want to relocate far away. Put differently, in the scenario captured by this approach, the strategic mobility explanation is less likely to play a role. While the timing of these moves is arguably unrelated to unobservable trends at non-high-wage firms, knowledgeable workers may still decide which new employer to join among the set of non-high-wage firms after being displaced by good firms. However, in small labour markets and specialised industries, workers are likely to have a limited set of alternatives. Applying this approach, the regressions return estimates consistent with knowledge transfer through labour mobility.

I then turn to evaluate the extent to which labour mobility can explain the productivity advantages of firms located near other highly productive firms. Here I relate my findings on the effect of firm-to-firm labour mobility to the existing evidence on the productivity advantages of agglomeration, focusing on a study by Greenstone et al. (2010). They find that following the opening of a large manufacturing plant, the productivity of incumbent plants in the US counties that were able to attract these large plants increased significantly relative to the productivity of incumbent plants in counties that survived a long selection process but narrowly lost the competition. The observed effect on productivity in Greenstone et al. (2010) is larger if incumbent plants are ‘economically’ close to the large plant. Further, this productivity effect increases over time. These facts are consistent with the presence of intellectual externalities that are embodied in workers who move from firm-to-firm. I am able to evaluate the extent to which worker flows explain the productivity advantages of agglomeration, by predicting the change in local productivity following an event analogous to that studied by Greenstone et al. (2010) within the worker mobility framework described above. I find that the predicted change in productivity is around 10% of the overall local productivity change observed after the event.

References

Alfaro, L, and M X Chen (2014), “The global agglomeration of multinational firms”, Journal of International Economics, 94, 263-276.

Baum-Snow, N (2017), “Urban transport expansions, employment decentralization, and the spatial scope of agglomeration economies”, Working Paper.

Combes, P, and G Duranton (2006), “Labour pooling, labour poaching and spatial clustering”, Regional Science and Urban Economics, 36(1), 1-28.

Combes, P, G Duranton, L Gobillon, D Puga, and S Roux (2012), “The productivity advantages of large cities: Distinguishing agglomeration from firm selection”, Econometrica, 80 (6), 2543-2594.

Ellison, G, E L Glaeser, and W R Kerr (2010), “What causes industry agglomeration? Evidence from coagglomeration patterns”, American Economic Review, 100 (June 2010), 1195-1213.

Greenstone, M, R Hornbek, and E Moretti (2010), “Identifying Agglomeration spillovers: evidence from winners and losers of large plant openings”, Journal of Political Economy, 118 (3), 536-598.

Henderson, V (2003), “The urbanization process and economics growth: the so-what question”, Journal of Economic Growth, 8 (1), 47-71.

Kline, P (2010), “Place based policies, heterogeneity, and agglomeration”, American Economic Review: Papers and Proceedings, 100 (May 2010), 383-387.

Moretti, E (2011), “Local Labor Markets”, Handbook of Labor Economics, 4, 1237–1313.

Piore, M J (2009), “Conceptualizing the dynamics of industrial districts” In The Handbook of Industrial Districts, Edward Elgar.

Rosenthal, S, and W Strange (2003), “Geography, industrial organization, and agglomeration”, The Review of Economics and Statistics, 85 (2), 377-393.

Serafinelli, M (forthcoming), “Good Firms, Worker Flows and ProductivityJournal of Labor Economics.

Trigilia, C (1990), “Work and politics in the third Italy’s industrial districts”, In F Pyke, G Becattini and W Sengenberger (eds.), Industrial Districts and Inter-Firm Co-operation in Italy (160-184), Geneva: International Institute for Labor Studies.

Whitford, J (2001), “The decline of a model? Challenge and response in the Italian industrial districts”, Economy and Society, 30 (1), 38-65.

630 Reads