Business cycles and bank regulations – what happens to bank provisioning? A more comprehensive look at 49 countries

Abstract

Numerous researchers provide evidence that many banks intend to increase their Loan Loss Provisioning (LLP) when the economy is in a downward trend. However, the answer whether banks provide sufficient provisions when the economy is in an upturn trend remains unsolved. Furthermore, provisioning must be influenced not only by business cycles and bank earnings, but also by the regulatory system. Nevertheless, empirical research on this issue has been scarce. Thus, this study intends to answer two questions shown above. The evidence shows that with steady growth in both the economy and bank earnings, the bank management will tend to increase LLP, whereas with a buoyant economy but negative growth in bank earnings, the management will exhibit an inclination to reduce LLP. As regards the influence of bank regulation on provisions, the evidence shows that, under certain circumstances, banks make more provision based on regulatory considerations. This explains why bank regulations regarding LLP across countries do have an impact on banks’ provisioning behaviour.

Acknowledgements

The authors are grateful for the financial support provided by the Taiwan National Science Council (NSC) ⟨94-2416-H-025-010-⟩.

Notes

¹ Rather than just serving as a window dressing for a financial report, Kim and Santomero (1993) argued that a positive correlation between earnings and provisions may well be the result of optimal statistical forecasting with respect to loan losses and hence is not necessarily due to misleading provisioning behaviour as is supposed in income-smoothing theory. Defond and Chul (1997) used discretionary accruals as the basis for predicting next period earnings. The evidence suggests that in considering job security when current earnings are ‘poor’ and expected future earnings are ‘good’, managers ‘borrow’ future earnings for use in the current period. Conversely, when current earnings are ‘good’ and expected future earnings are ‘poor’, managers ‘save’ current earnings for possible use in the future.

² In addition to the research studies mentioned here that empirically verified the pro-cyclicality of banks’ provisioning behaviour, Borio et al. (2001) used a simulation approach to investigate the important links between financial development and business cycles; loan loss provisioning is another one of the factors. Lowe (2002) also investigated the issue during periods when the economy is in good shape, especially under circumstances marked by a dramatic growth in loans and a rapid increase in the credit risk of bank loans.

³ The definition of pro-cyclical effect given by Jackson (1999), Bikker and Metzemakers (2005) is stated as below: during a cyclical downturn, the quality of banks’ assets is likely to deteriorate, which would increase risk exposure and, hence, capital requirements, exactly at a time when new capital becomes more expensive or, for weaker institutions, simply unobtainable.

⁴ Countries allowing general provisions to be counted as part of Tier II capital include some G-10 countries – France, Germany, Italy, Japan, the UK and the USA – and some non-G-10 countries, such as Argentina, Australia, Chile, China, the Czech Republic, Hong Kong, India, Mexico, Saudi Arabia, Singapore, South Africa, South Korea, the Russian Federation and Taiwan.

⁵ In Taiwan, according to Article 5 of the ‘Regulations Governing the Procedures for Banking Institutions to Evaluate Assets and Deal with Non-performing/Non-accrual Loans’, amended on 6 January 2004, the minimum standard for LLP shall be the sum of 2% of the balance of category 2 credit assets, 10% of the balance of category 3 credit assets, 50% of the balance of category 4 credit assets and the full balance of category 5 credit assets.

⁶ The differences between LLP and LLR, where the former is a flow concept and is considered in income statement and the latter is the stock concept and is in balance sheet. Along with Bikker and Hu (2002), Cavallo and Majnoni (2002) and Bikker and Metzemakers (2005), this study also adopts LLP as the independent variable.

⁷ Our positive EBPT or negative EBPT are actual values rather than a dummy variable used in Laeven and Majnoni (2003). Since a dummy variable is created according to that banks earnings are below or above the zero threshold, it does not tell us the strength of EBPT's different values. Using positive EBPT as an example, the values of 1 and 1000 are assumed to have the same effect in Laeven and Majnoini (2003), but they are different in our case. This may wrongly treat a profitable bank and a normal bank to be the same. Thus, we modified the research methodology proposed by Laeven and Majnoni (2003).

⁸ Under the consideration of data quality, our sample follows the work of La Porta et al. (1998), covering 49 countries from Europe, North and South America, Africa, Asia and Australia. There are no socialists of ‘transition’ economies in the sample.

⁹ Readers are referred to Borio et al. (2001) and Beattie et al. (1995) in the case where the financial system vis-à-vis provisioning is not forward looking.