The French competitiveness cluster policy (pôles de compétitivité, launched 2005) allocated R&D subsidies through a two-step procedure: the national government first selected 66 sector-location 'clusters' from 105 applications, and firms then freely joined the selected clusters as members eligible to apply for project subsidies. Fontagné, Koenig, Mayneris and Poncet (2013) showed that winners were (self-)selected at both steps — selected clusters contained ex ante larger, more productive, more export-oriented firms, and firm-level export premia were explained by individual characteristics except for a residual premium in 'National' clusters — but they deliberately left the driving mechanisms unexamined, stating that lobbying, policy-maker preferences, and policy-maker-cluster connections were unobserved, and left open whether the design of subsidies shapes firm self-selection. The open problem is to identify the causal mechanisms behind both stages of selection: why the particular sector-location applications were chosen by the government, and which frictions (application costs, information and networking asymmetries, expected subsidy value and its design) govern which firms self-select into membership.

Background

Cluster policies are place-based industrial policies that subsidize geographically concentrated groups of firms in a shared sector, intending to generate knowledge spillovers (benefits one firm's R&D confers on nearby firms) and agglomeration economies. France's competitiveness cluster policy (pôles de compétitivité) was launched in 2005: after a national call for applications, the government selected 66 of 105 sector-location applications, labeled 'worldwide', 'potentially worldwide', or 'national' clusters; firms then chose whether to become members of a selected cluster, membership being the gateway to applying for dedicated collaborative R&D subsidies. Because the government picks some places and sectors but not others, and firms decide for themselves whether to apply for membership, the policy combines public selection (government choice among applications) with self-selection (firms sorting themselves into treatment), a distinction emphasized in the theoretical literature on regional and industrial policy. That literature offers competing predictions about who ends up treated: models of asymmetric lobbying (Baldwin and Robert-Nicoud, 2007; Bombardini, 2008) predict that firms with the strongest lobbying capacity — typically larger firms, or firms in declining sectors where rents from protection are highest — capture support, implying selection of ex ante weaker or better-connected applicants; models of spatial sorting of heterogeneous firms (Baldwin and Okubo, 2006) predict that the most productive firms concentrate in agglomerated core locations, so better ex ante performance in targeted areas need not reflect the policy; and models of costly application for R&D subsidies (Takalo, Tanayama and Toivanen, 2013) predict that fixed application and administrative costs induce only larger or more productive firms to seek support. Fontagné, Koenig, Mayneris and Poncet (2013) audit the ex ante (pre-policy, 2004) performance of beneficiaries using French customs and annual business survey data. Their findings: selected clusters grouped firms that performed better within their sector at the national level; firms that joined were larger, more productive, and more export-oriented than sector-location peers (consistent with sunk application costs or better information and networking access); after controlling for firm size and productivity the export premium of cluster members vanished, except for a residual premium in 'National' clusters. Crucially, they analyze only the outcome of selection, not its mechanisms: they state explicitly that the exact mechanisms driving selection are unobserved and that they do not address lobbying, policy-makers' preferences, or connections between policy-makers and cluster managers; their conclusion flags whether subsidy type (R&D versus employment incentives, big versus small monetary subsidies) shapes firm self-selection as an open question. This matters because whether selection reflects political capture, policy-maker objectives, information asymmetries, or application frictions determines both the normative assessment of the policy and the correct econometric handling of selection bias when evaluating its causal effects.

Problem Statement

Identify the causal mechanisms that determined selection into the French competitiveness cluster policy at both of its stages. First, at the government stage: explain why, among the 105 sector-location applications received in the 2004-2005 call, precisely the observed 66 clusters were selected - in particular, discriminate empirically between political-economy mechanisms (lobbying and rent-seeking, political connections between applicants and policy-makers, policy-maker preferences and capture, territorial balancing) and merit-based or information-based selection. Second, at the firm stage: explain which frictions govern firm self-selection into membership of selected clusters - sunk application and administrative costs, information and networking asymmetries, expected subsidy value - and whether the design and type of subsidies available to members (R&D versus employment incentives, per-firm versus proportional, big versus small monetary subsidies) systematically shapes which firms self-select. An admissible answer must connect mechanisms to the documented ex ante patterns (larger, more productive, more export-oriented members; premium concentrated in 'National' clusters beyond size and productivity), and remain valid for the policy as implemented in 2005-2012 without restricting the mechanism set or the data used a priori.

References

  1. Fontagné, Lionel; Koenig, Pamina; Mayneris, Florian; Poncet, Sandra (2013). Cluster Policies and Firm Selection: Evidence from France. Journal of Regional Science 53(5), 897-922. DOI 10.1111/jors.12050. https://doi.org/10.1111/jors.12050 (open-access full text: https://hal.science/hal-00975554)
  2. Baldwin, Richard E.; Robert-Nicoud, Frédéric (2007). Entry and Asymmetric Lobbying: Why Governments Pick Losers. Journal of the European Economic Association 5(5), 1064-1093. DOI 10.1162/jeea.2007.5.5.1064. https://doi.org/10.1162/jeea.2007.5.5.1064
  3. Baldwin, Richard E.; Okubo, Toshihiro (2006). Heterogeneous Firms, Agglomeration and Economic Geography: Spatial Selection and Sorting. Journal of Economic Geography 6(3), 323-346. DOI 10.1093/jeg/lbi020. https://doi.org/10.1093/jeg/lbi020
  4. Bombardini, Matilde (2008). Firm heterogeneity and lobby participation. Journal of International Economics 75(2), 329-348. DOI 10.1016/j.jinteco.2008.03.003. https://doi.org/10.1016/j.jinteco.2008.03.003
  5. Takalo, Tuomas; Tanayama, Tanja; Toivanen, Otto (2013). Market failures and the additionality effects of public support to private R&D: Theory and empirical implications. International Journal of Industrial Organization 31(5), 634-642. DOI 10.1016/j.ijindorg.2013.02.002. https://doi.org/10.1016/j.ijindorg.2013.02.002
  6. Mar, Modou; Massard, Nadine (2021). Animate the cluster or subsidize collaborative R&D? A multiple overlapping treatments approach to assess the impacts of the French cluster policy. Industrial and Corporate Change 30(4), 845-867. DOI 10.1093/icc/dtab002. https://doi.org/10.1093/icc/dtab002
  7. Hassine, Haithem Ben; Mathieu, Claude (2020). R&D crowding out or R&D leverage effects: An evaluation of the french cluster-oriented technology policy. Technological Forecasting and Social Change 155, 120025. DOI 10.1016/j.techfore.2020.120025. https://doi.org/10.1016/j.techfore.2020.120025
  8. Chaudey, Magali; Dessertine, Marion (2018). Impact sur l'emploi de la participation aux projets de R&D des pôles de compétitivité. Méthode et résultats. Revue d'économie industrielle 162, 73-98. DOI 10.4000/rei.7061. https://doi.org/10.4000/rei.7061
  9. N'Ghauran, Konan Alain; Autant-Bernard, Corinne (2022). Effects of cluster policies on regional inventor networks: evidence from France. Regional Studies 56(11), 1903-1918. DOI 10.1080/00343404.2021.2015534. https://doi.org/10.1080/00343404.2021.2015534
  10. Quignon, Aurélien (2025). Cluster and local science-industry collaborations: evidence from a place-based innovation policy. Journal of Technology Transfer. DOI 10.1007/s10961-025-10284-w. https://doi.org/10.1007/s10961-025-10284-w
  11. Okamuro, Hiroyuki; Nishimura, Junichi (2018). Whose business is your project? A comparative study of different subsidy policy schemes for collaborative R&D. Technological Forecasting and Social Change 127, 85-96. DOI 10.1016/j.techfore.2017.07.017. https://doi.org/10.1016/j.techfore.2017.07.017
  12. Okubo, Toshihiro; Okazaki, Tetsuji; Tomiura, Eiichi (2022). Industrial cluster policy and transaction networks: Evidence from firm-level data in Japan. Canadian Journal of Economics 55(4), 1990-2035. DOI 10.1111/caje.12629. https://doi.org/10.1111/caje.12629
  13. Lehmann, Erik E.; Menter, Matthias (2018). Public cluster policy and neighboring regions: beggar-thy-neighbor? Economics of Innovation and New Technology 27(5-6), 420-437. DOI 10.1080/10438599.2017.1374039. https://doi.org/10.1080/10438599.2017.1374039
  14. Howell, Anthony (2020). Picking 'winners' in space: Impact of spatial targeting on firm performance in China. Journal of Regional Science 60(5), 1025-1046. DOI 10.1111/jors.12497. https://doi.org/10.1111/jors.12497
  15. Grashof, Nils (2021). Putting the watering can away - Towards a targeted (problem-oriented) cluster policy framework. Research Policy 50(9), 104335. DOI 10.1016/j.respol.2021.104335. https://doi.org/10.1016/j.respol.2021.104335
  16. Nathan, Max (2022). Does light touch cluster policy work? Evaluating the tech city programme. Research Policy 51(9), 104138. DOI 10.1016/j.respol.2020.104138. https://doi.org/10.1016/j.respol.2020.104138