VoxEU Column EU institutions Financial Regulation and Banking

Centralised bank supervision and the composition of firm investment

The trade-off between stability and growth has long been a subject of policy debate and informs views on the extent to which the supervision of banks should be centralised. This column presents analysis of the ECB’s Single Supervisory Mechanism, using the announcement of the mechanism and its implementation as a quasi-natural experiment. It finds that centralised bank supervision is associated with a decline in lending to firms, which is accompanied by a shift away from intangible investment and towards more cash holdings and higher investment in easily collateralisable physical assets.

Does stability breed or hinder growth? Numerous papers have tackled various aspects of this question, providing opposing answers and conflicting policy implications. At the one extreme, Ramey and Ramey (1995) show that in countries where long-run GDP growth is less volatile, it is also higher on average. At the other extreme, Ranciere et al. (2008) show that countries in which long-run GDP growth is higher also tend to experience more frequent systemic crises. They argue that both development and crises are driven by the same economic force – risk taking – and that policymakers who clamp down on the sources of instability will pay a price in terms of lower growth.

In recent research (Ampudia et al. 2021), we contribute to this debate by studying the effect of one well-defined stability-relevant policy (the introduction of centralised bank supervision in the euro area) on one well-defined growth mechanism (corporate investment). After the Global Crisis and sovereign debt crisis (of the late 2000s and early 2010s, respectively), regulators introduced a number of reforms aimed at improving the resilience of European banks. The centrepiece of this drive was the introduction of the Single Supervisory Mechanism (SSM). Following an asset quality review as well as stress tests (together referred to as a ‘comprehensive assessment’), a number of significant euro area banks became supervised by the mechanism, while others remained under the supervision of their national authorities. While previous research has shown that the shift to centralised supervision resulted in stability-enhancing actions by the affected banks (Fiordelisi et al. 2017, Eber and Minoiu 2017, Altavilla et al. 2020), our research shows the impact of this change in supervisory architecture on the real economy. 

What do the theory and literature suggest?

On the one hand, Laffont and Tirole (1993) argue that local supervision results in better monitoring of banks. Colliard (2020) argues that local supervisors might be better able to extract information from banks than a centralised supervisor. Carletti et al. (2021) point to lower incentives for local supervisors to collect information if supervisory decisions are centralised. If local supervisors provide more rigorous supervision than centralised supervisors, we would expect firms whose lenders change to centralised supervision to increase their investment, including into less ‘collateralisable’ assets, such as intangible capital.

On the other hand, because bank supervision exhibits scale economies, centralised supervision might be more effective (Eisenbach et al. 2016). Centralised supervision might also be better able to reduce the risk of banks arbitraging differences in regulatory stringency across countries (Dell'Ariccia and Marquez 2006) and can increase supervisory independence (Rochet 2008). If centralised supervisors are more effective in holding in check banks' risk-taking, banks under their supervision might tighten lending standards and increase collateral requirements, with negative implications for investment, especially in intangible assets (which are less collateralisable).

The creation of the Single Supervisory Mechanism as quasi-natural experiment

On 29 June 2012, the heads of government of all euro area countries announced the creation of a Single Supervisory Mechanism, underpinned by the necessity to break the vicious circle between banks and sovereign (as first element of the so-called Banking Union). The ECB would exercise prudential supervision of all banks located in the euro area, whether directly by the ECB's own supervisory arm for the ‘significant institutions’, or indirectly by the national prudential supervisors (but under the general guidance of the ECB for the ‘less-significant institutions’). As of 2020, the ECB directly supervises 117 significant institutions – the actual supervisory activities are conducted by joint supervisory teams (JSTs), involving both ECB and national supervisory staff.

An important step in preparing the SSM to become fully operational was the comprehensive assessment between November 2013 and October 2014. This process included an asset quality review and stress test as financial health check of 130 banks in the euro area, covering approximately 82% of total bank assets. The results were published on 26 October 2014. Then, on 4 November 2014, the mechanism was born. 

We use the announcement of the SSM in 2012 and its introduction in 2014 as a quasi-natural experiment to study the impact of a change in supervisory architecture on the performance of the real economy.


We combine firm- and bank-level data to gauge the effect of the change in supervisory architecture on firm investment patterns. Our firm-level data come from the Orbis data set provided by Bureau van Dijk (BvD), with financial data for 2010 to 2017. We follow the downloading methodology and cleaning procedure described in Kalemli-Ozcan et al. (2019) in order to ensure the database is nationally representative and contains minimal missing information. We also limit our sample to ten euro area countries with good coverage in Orbis (Austria, Estonia, France, Germany, Lithuania, Latvia, Luxembourg, Portugal, Slovenia, and Spain), and drop firms in agriculture and mining; sectors with high government ownership (such as public administration); and heavily regulated sectors (such as finance). 

In addition to financial statement information, the Orbis database provides (for each company) the name of the main bank the company conducts business with. This allows us to identify whether a company is related to a bank which became directly supervised by the mechanism when it was established, or whether it is related to a bank which is only indirectly supervised. We end up with a sample of 188,600 firms that have a relationship with a total of 294 individual banks. Of these, 179 are ‘significant institutions’, and 115 are ‘less-significant institutions’.

We also make use of monthly bank-level information from the ECB's Individual Balance Sheet Statistics (IBSI) dataset for 247 individual financial institutions in 18 European countries (comprising about 70% of the domestic banking sector). We focus on euro area countries with at least one significant and one less significant institution included in dataset, where the institutions in both groups are comparable in size. This leaves us with 186 banks in 11 euro area countries (Austria, Belgium, Finland, France, Germany, Greece, Ireland, Luxembourg, Malta, Netherlands, and Spain). We have information on the stock of total lending, as well as on the stock of lending to various classes of customers, such as governments, non-financial corporations (NFCs), and households. 

Main findings

We differentiate between the effect of the comprehensive assessment and the introduction of the Single Supervisory Mechanism in a difference-in-differences set-up, comparing firms whose lenders came under new supervision mechanism with firms whose banks stayed under the national supervision across three time periods: (i) before the announcement of the SSM  (2010-2012), (ii) during the period of the comprehensive assessment (2013-2014), and (iii) after the introduction of the SSM (2015-2017). 

  • Firms borrowing from banks supervised under the SSM experienced a significant reallocation across different types of investment, relative to firms borrowing from banks that remained under the supervision of national authorities. During the comprehensive assessment period (2013-2014), affected firms relatively increased their investment in current assets (i.e. cash) and reduced intangible assets. During the period after the mechanism took over (2015-2017), affected firms relatively increased their investment in tangible assets and cash, and reduced investment in intangible assets further. 
  • These results are robust across a number of sensitivity analyses, including a placebo test where we apply our empirical setting and estimation to European countries whose banks did not fall under the SSM from 2014 onward. We focus on Denmark, Hungary, and the UK – all of which have reasonable coverage in Orbis – and apply the SSM’s criterion to the banks which the firms have a credit relationship with. We do not find any significant differences in investment pattern across the two groups of firms (before or after the start of the euro area's SSM). We conclude that the main results in our paper are indeed driven by the transition to centralised supervision, rather than by a global trend in investment reallocation by firms borrowing from large banks.
  • We also test for parallel trends before the announcement of the mechanism and the comprehensive assessment. Figure 1 shows no evidence of pre-trends for any of the four types of assets. This means that firms borrowing from significant institutions were investing at the same rate as firms borrowing from less-significant institutions. After 2012, however, tangible assets and current assets increase and intangible assets decline for firms borrowing from significant institutions, compared to firms borrowing from less-significant institutions.
  • The decline in intangible investment is particularly pronounced in innovation-intensive sectors, especially during the early period of the mechanism. As such industries are instrumental in contributing to productivity-driven long-term growth in modern knowledge-based economies, this points to a trade-off between stability and growth.
  • We also find a significant decline in employment during the period of the comprehensive assessment, as well as a significant increase in employment during the SSM Mechanism period. This suggests a strong complementarity between physical capital and employment, but no such complementarity between employment and intangible capital. We also find a reduction in labour productivity for affected firms after their banks moved to SSM supervision. 
  • We also show that corporate lending by banks that came under supervision declined, both during the transition period and after the implementation of the mechanism (compared to corporate lending by banks not subject to SSM supervision). We record the same effect with firm- and bank-level data. In the latter case, we also find that the decline in lending was larger for banks with relatively low capital before the announcement of the mechanism.

Figure 1 SSM and firm assets over time: Significant vs less-significant institutions


Note: The figure uses annual data for the period 2010 to 2017. The graph plots period-by-period coefficients and 90% confidence intervals on yearly dummies interacted with a dummy equal to one for firms borrowing from SIs. The reference year is 2012.


Overall, our results suggest that centralised bank supervision is associated with a decline in lending to firms, which is accompanied by a shift away from intangible investment and towards more cash holdings and higher investment in easily collateralisable physical assets. This is an instructive result, in light of the fact that in the long run, capital investment has a negligible contribution to economic growth, while R&D investment accounts for the bulk of long-term growth (Fernald and Jones 2014). The combination of the two effects we document therefore raises the possibility that centralised bank supervision can slow down the shift from the ‘old’ (capital-based) to the ‘new’ (knowledge-based) economy. Returning to the debate mentioned at the start, these findings point to a trade-off between stability and growth in this specific incidence.


Altavilla, C, M Boucinha, J-L Peydro and F Smets (2020), “Banking supervision, monetary policy, and risk-taking: Big data evidence from 15 credit registers”, ECB Working Paper 2349.

Ampudia, M, T Beck and A Popov (2021), “Out with the New, In with the Old? Bank Supervision and the Composition of Firm Investment”, CEPR Discussion Paper 16225.

Carletti, E, G Dell'Ariccia and R Marquez (2021), “Supervisory Incentives in a Banking Union”, Management Science 67: 455-470.

Colliard, J-E (2020), “Optimal supervisory architecture and Financial integration in a Banking Union”, Review of Finance 24: 129-161.

Dell’Ariccia, G and R Marquez (2006), “Competition among regulators and credit market integration”, Journal of Financial Economics 79: 401-430.

Eber, M and C Minoiu (2016), “How do banks adjust to stricter supervision?”, working paper.

Eisenbach, T, D Lucca and R Townsend (2016), “The economics of bank supervision”, NBER Working Papers 22201.

Fernald, J and C Jones (2014), “The future of U.S. economic growth”, American Economic Review 104: 44-49.

Fiordelisi, F, O Ricci and F S S Lopes (2017), “The unintended consequences of the launch of the single supervisory mechanism in Europe”, Journal of Financial and Quantitative Analysis 52: 2809-2836.

Kalemli-Ozcan, S, B Sorensen, C Villegas-Sanchez, V Volosovych and S Yesiltas (2019), “How to construct nationally representative firm level data from the Orbis Global Database: New facts and aggregate implications”, NBER Working Paper 21558.

Laffont, J-J and J Tirole (1993), A Theory of Incentives in Procurement and Regulation, MIT Press. 

Ramey, G and V Ramey (1995), “Cross-country evidence on the link between volatility and growth”, American Economic Review 85: 1138-1151.

Ranciere, R A Tornell and F Westermann (2008), “Systemic crises and growth”, Quarterly Journal of Economics 123: 359-406.

Rochet, J-C (2008), Why Are There So Many Banking Crises? The Politics and Policy of Bank Regulation, Princeton University Press.

735 Reads