More productive firms may choose to locate in industry clusters in order to benefit from agglomeration externalities. However, negative sorting and adverse selection effects are also possible. This column studies the location decisions of Japanese multi-plant firms in order to disentangle these effects. Agglomeration effects are present when firms compete for different, or dispersed foreign, markets. On the other hand, adverse selection occurs when firms face direct rivals in the product market. It suggests there is a need for policies that provide incentives to attract highly productive firms to industrial clusters.
Recent empirical studies show that firms located in industry clusters tend to be more productive (e.g. Andersson et al. 2019, Rosenthal and Strange 2020, Lavoratori and Castellani 2021). Two competing explanations have been put forward for this correlation. The first and predominant explanation is the notion of Marshallian agglomeration externalities, which contends that firms can enjoy positive externalities stemming from geographic industry clustering. Externalities can occur on the supply side in the form of the availability of specialised factors of production and on the demand side through reduced search costs for customers and heightened local industry demand. That these possible externalities motivate firms to choose locations where similar establishments are clustered has been supported by theoretical models (e.g. Krugman 1991) and empirical work (e.g. Belderbos et al. 2021).
A second explanation is the selection effect associated with the increased competition within clusters. Collocation leads to tougher competition, forcing the exit of weaker firms with lower productivity. Moreover, more productive firms may benefit more from agglomeration, leading to a positive sorting effect, with more productive firms self-selecting into high-density clusters (Baldwin and Okubo 2006). However, this conjecture has received less substantive support in empirical research. In fact, instead of positive selection effects, negative effects are also conceivable. When productive firms locate in industry agglomerations, they run the risk that their innovative technologies and organisational and process skills could be copied by collocated rival firms. Less productive firms, in contrast, have less to lose. Such asymmetry in knowledge spillovers due to productivity differences suggests a process of negative sorting (adverse selection) in which firms with relatively weaker (rather than stronger) productivity are more likely to locate within industry clusters.
In our paper (Belderbos et al. 2022), we aim to reconcile these two contrasting views by examining the location decisions of Japanese multi-plant firms for new manufacturing plants. We investigate the relationship between local industry agglomeration and location decisions for firms with different levels of productivity in the industry, distinguishing the influence of the local presence of rival firms within the same (domestic) market and the presence of firms not directly competing in the same market. We expect the adverse selection effect to occur if local agglomeration involves firms competing directly with the focal firm that decides on a plant location. We constructed establishment-level data for the period 2002–2008 drawing on Japan’s Census of Manufacture, conducted by Ministry of Economy, Trade and Industry. Plant-level total factor productivity (TFP) is measured using the index number method. Location choices are examined at the fine-grain level of towns, wards, and cities, using discrete choice conditional and mixed logit models.
Our analysis shows that the effect of industry agglomeration on firm location choice depends on their productivity level relative to their industry peers. Specifically, while there is a strong tendency for less productive firms to select locations with high levels of industry agglomeration, the most productive firms tend to avoid these. This pattern is observed for non-exporting firms that are facing a local industry cluster with firms that are likewise producing for the domestic market and directly competing with the focal firm, but not for exporting firms that target dispersed foreign markets. If existing establishments and high-productivity entrants compete in the same (domestic) product market, knowledge dissipation concerns are salient, as increases in the competitiveness of incumbent rivals directly affect the market share and profitability of the entrant. In contrast, if entrants and incumbents are less likely to compete in the same end market – i.e. if entrants target export markets and incumbents target the domestic market, or vice versa – industry agglomeration has a positive effect on location choice. Hence, adverse selection occurs, but only if investing firms face direct rivals in the product market.
The pattern for non-exporting firms is illustrated in Figure 1 below, which plots the elasticity of location choice with respect to industry agglomeration as a function of non-exporting firms’ TFP premium – that is, the ratio of firms’ TFP level to the industry average. In the figure, the solid line represents the elasticity, while the dotted line shows the 95% confidence interval. The figure indicates that while non-exporters with a low TFP premium are attracted to industry clusters, non-exporters with a high TFP premium are less likely to select a location with an industry cluster. Firms with a TPF premium of 1.4 or higher tend to avoid agglomerations for new investments, and at a level of 2.5 this disincentive effect becomes significant.
From a policy perspective, it seems desirable that highly productive firms collocate in industry clusters in order to maximise the positive externalities of such agglomerations. Our analysis suggests that in order to achieve industry clusters that are more desirable from the perspective of the economy overall, there is a need for policies that provide incentives to attract highly productive firms to industrial clusters. On the other hand, the fact that leading firms – i.e. firms with above-average TFP – have a greater incentive to avoid existing agglomerations may be good news for local governments in more peripheral areas that do not have industry clusters. Our findings suggest that if peripheral areas want to try to attract manufacturing investment, they are more likely to succeed if they target more productive rather than less productive firms. That said, the same rationale that would lead productive firms to locate to peripheral areas may also lead them to locate overseas in less developed regions without product market rivals. Peripheral areas in Japan may stand in direct competition with overseas economies in terms of attracting such firms.
Authors’ note: The main research on which this column is based (Belderbos et al. 2022) first appeared as a Discussion Paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.
This article first appeared on VoxEU on April 10, 2023. Reproduced with permission.
