Are bank mergers procyclical or countercyclical? Theory and evidence from Taiwan

Abstract

This work develops a theory of countercyclical bank mergers and finds supported empirical evidence in Taiwan. Procyclical (countercyclical) mergers tend to involve higher (lower) measures of merger activities during economic booms than downturns. Several previous studies suggest that bank mergers are procyclical in developed countries, possibly driven by the higher capital liquidity that accompanies an economic expansion. Alternatively, this work emphasizes the role of economic situation and the government in bank mergers. Our results suggest that the depressed industrial situation, which leads to severe competitive market and impoverished revenue basis, drives banks to increase their market share for reducing competition through merger with the merger cost as low as possible by utilizing government's offering and the beneficial condition of stock market. Merger activities may be explained by the motivation of generating higher profit and able to survive, whereas the banks with insufficient competitiveness are more likely to suffer serious loss and thus forced to be merged.

Keywords:

• countercyclicality
• merger
• industrial recession
• banking industry

JEL Classification::

• G21
• G34

Acknowledgements

We wish to express our gratitude to an anonymous referee for the very valuable suggestions provided. All remaining errors are ours.

Notes

¹ The number of commercial banks increases from 25 to 53 until 2002, including 16 new banks and the other transformed ones.

² One purpose of the policy is to stabilize the banking system. The government may control competition by restricting opening new branches, bail out failing banks, or require healthy bank to rescue them (Hoshi, 2002).

³ Berger and Humphrey (1992), Pilloff and Santomero (1998) support the opinion using the bank data in the US, whereas Vander Vennet (1997), Focarelli et al. (2002) are European studies.

⁴ ‘too-big-to-fail’ policy means government explicit or implicit guarantee debt holders or shareholders of these large organizations. This may evoke incentives to increase size through consolidation to benefit from government's financial safety net, lower the cost of funding, or even increase the value of shares (Berger et al., 1999).

⁵ In Japan, a variety of banks have been consolidated since the 1990s when most banks suffered from a huge amount of nonperforming loans.

⁶ The point supported not only by the US bank that approximately half of bank mergers are in-market but also by European bank mergers that are of this type as well (Vander Vennet, 1997).

⁷ We regard firm 1 as active banks and firm 2 as passive banks in the following empirical explanations.

⁸ The assumption that firm 1 owns higher productivity than firm 2 will be confirmed in empirical data subsequently.

⁹ Commercial banks, whose main business is granting loans and offering money deposits, are in the majority in Taiwan banks. Without loss of generality, we assume the spread, loan interest rate minus interest on deposit, as the price of financial services that banks provide.

¹⁰ Precisely, is required. If the default risk of loan is relatively small, both firms will not suffer default risks. To emphasize the role of bankruptcy costs on merger activities, the case without bankruptcy costs is ignored.

¹¹ Okada (2005) provides an alternative motivation of mergers, to take advantage of the government's too-big-to-fail policy.

¹² The assumption that only one period profit and merger cost is clearly special. That can be justified in a situation in which the future profit streams and merger costs can be discounted into one period. It is remarked that the assumption of one-period merger costs is common even in a dynamic model in the literature, such as Lambrecht (2004). Here we simplify the problem by assuming that only one period of profits and merger cost. In a more general framework with multiple periods such a real option model, the timing of merger activities might be discussed.

¹³ One might wonder about the assumption that duopoly model can picture the fierce competition in the financial sector. This is analogous to the representative individual assumption in macroeconomics that makes it possible to connect our empirical samples, regarding firm 1 as a group of active banks and firm 2 as the other group of passive banks. The model just simplifies numerous banks to two types, generating no effects to our empirical implications.

¹⁴ Specifically, the merging dummy equals 1 if there will be active banks joining in merger activities; the merged dummy is set to 1 when the passive banks will be merged, and the merger dummy will be 1 when either kind of banks participates in merger next year.

¹⁵ Literature such as those of Vander Vennet (1997) and Focarelli et al. (2002) apply the logit model which only has different assumption of error-term distribution from the probit model. Note that we obtain similar results for logit models.

¹⁶ The title may be somewhat different among literature, such as nonperforming loan ratio in Peristiani (1993) or bad loan ratio in Hosono et al. (2007), but the formula is all the same.

¹⁷ Some literature may use ROA or return on total assets, and some may work with ROE, whereas others apply both (Focarelli et al., 2002). We just choose for either one, having no influence on our results.

¹⁸ The laws or policies denote Financial Institutions Merger Act, Financial Holding Company Act and Financial Reformations, the central actions facilitating merger activity.

¹⁹ We exclude investment banks and some small trust companies, focusing on only commercial banks, the major bank type in Taiwan.

²⁰ For a comparison, Cerasi et al. (2002) report that average number of branches per bank in Belgium, Denmark, France, Germany, Greece, Italy, the Netherlands, Portugal and Spain in the year 1996 are 199, 49, 79, 45, 116, 151, 210, 186 and 432, respectively. Actually, our sample does not include unlisted local noncommercial financial institutions such as local credit cooperation. Most of these financial institutions have small number of branches.