The Economic Costs of Bank Closures: Evidence from a Natural Experiment

Kerstins Rum Session 2: Central banking – a science or an art? organized by Lars Fredrik Øksendal and Anders Ögren


Seán Kenny, Jason Lennard, Emma Horgan


A fundamental feature of major banking crises in economic history is the closure of banks (Reinhart and Rogoff, 2009; Grossman, 2010; Jalil, 2015). A classic example is the Great Depression, when more than 9,000 banks suspended in the United States between 1930 and 1933 (Friedman and Schwartz, 1963, p. 35). According to Bernanke (1983), bank closures result in lost expertise, information, and relationships, which increases the cost of credit intermediation and decreases the volume of credit and economic activity. In many historical episodes, however, there are confounding factors that obscure a clean estimate of the macroeconomic effects of bank closures, such as weak fundamentals, panics, lending of last resort, and bail outs.

An ideal experiment would observe two identical islands. In the first, banks would be closed; in the second, banks would be open. The difference in economic outcomes between the first and second island is equal to the causal effect of bank closures. While this randomised control trial is impossible, there was a natural experiment in history: Ireland in the 1960s and 1970s. All banks were closed between 7 May and 30 July 1966, 1 May and 17 November 1970, and 28 June and 6 September 1976, due to industrial disputes (Murphy, 1978). While there has been general interest in this episode (Financial Times, 2015; Independent, 2015), there has been little academic research.

To estimate the unobserved counterfactual Ireland with open banks, we use synthetic control methods (Abadie and Gardeazabal, 2003; Abadie et al., 2010, 2015; Born et al., 2019). To do so, we construct a doppelganger that is a weighted average of countries in the donor pool, where the weights are determined by minimising the distance in outcomes between actual and doppelganger Ireland before treatment. The causal effect is the difference in outcomes after treatment.

Our hypothesis is that the complete closure of banks involved a significant economic contraction. However, it was not a macroeconomic disaster, as agents substituted away from the scarce resource – bank credit – to other forms, such as cheques and micro credit, which ameliorated the impact. This is consistent with recent research showing that substitution after sudden stops is an important macroeconomic shock absorber (Bachmann et al., 2022).


No PDF available.