Measuring bank competition under binding interest rate regulation: the case of China

ABSTRACT

Many empirical studies suggest that financial reform promoted bank competition in most mature and emerging economies. However, some earlier studies that adopted conventional approaches to measure competition have concluded that bank competition in China declined during the past decade, despite progressive reforms implemented since the 1980s. We show that this apparent contradiction is the result of flawed measurement. Conventional indicators such as the Lerner index and Panzar–Rosse H-statistic fail to measure competition in Chinese loan markets properly due to the system of interest rate regulation. By contrast, the profit elasticity (PE) approach does not suffer from these shortcomings. Using balance sheet information for a large sample of banks operating in China during 1996–2008, we show that competition actually increased in the past decade when the PE indicator is used.

KEYWORDS:

• Competition
• banking industry
• China
• lending markets
• regulation

JEL CLASSIFICATION:

• D4
• G21
• L1

Acknowledgements

The authors are grateful to Yiping Huang and Xun Wang for sharing the values of their FREP with us. We would like to thank our two referees for very useful comments. We would like to thank Claudio Borio, Chun-Yu Ho, Tom Minic, Christian Upper, and participants at a Bank for International Settlements (BIS) seminar, at the ‘VIII Annual Seminar on Risk, Financial Stability and Banking’, organized by Banco Central do Brasil (8–9 August 2013, São Paulo), and at the ‘Fifth Annual International Conference on the Chinese Economy’, organized by the Hong Kong Institute for Monetary Research (16–17 January 2014, Hong Kong), for helpful suggestions. Louisa Wagner provided excellent editorial assistance. All remaining errors are ours. The views expressed in this article are of the authors and do not necessarily reflect those of the Bank of Spain, BIS, or APG.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The analysis in this article concentrates on the shortcomings of the Lerner index in measuring bank competition in China. Those of the Panzar–Rosse H-statistic are assessed in the Bank for International Settlements (BIS) Working Paper (WP) version (Xu, Van Rixtel, and Van Leuvensteijn 2013) which is available at http://www.bis.org/publ/work422.pdf. See Bikker, Shaffer, and Spierdijk (2012) for further discussion on the shortcomings of Panzar–Rosse H-statistic in the empirical literature.

² See the BIS WP version of this article, Section 4.3.

³ According to the China Banking Regulatory Commission (CBRC), financial reform in China can be classified in three major stages (1978–1993, 1994–2002, and 2003–present); see Liu (2009). Clear overviews in English are presented in Allen, Qian, and Qian (2005) and Matthews and Zhang (2010).

⁴ The joint-stock commercial banks (JSCBs) initially offered banking services only regionally, but later they were allowed to operate freely nationwide, competing with the state-owned commercial banks (SOCBs) for large firms and with the city commercial banks (CCBs) for small and medium-sized enterprises.

⁵ CCBs offer commercial banking services to small and medium-sized enterprises and households in the main cities or in certain provinces, but have been expanding to larger companies that would normally do business with the SOCBs and JSCBs. The requirement for CCBs to operate only within the cities’ own administrative districts was lifted from 2007 onwards, allowing them to compete in larger geographical areas.

⁶ We present the data for the full sample of 1996–2008 and the two subsamples that we use (1996–2001 and 2002–2008), with 2001 being the year of China’s entry into the World Trade Organisation (WTO).

⁷ Moreover, as we shall discuss in more detail in Section III and especially in Section VI, the PBC reintroduced credit quotas for individual banks in 2007 in order to curb lending activities.

⁸ The People’s Bank of China (PBC) started to widen the floating band on banks’ interest rates from 1998 onwards, after it liberalized interbank interest rates. Hence, commercial banks got more discretion in setting loan rates (PBC, 2005; Feyzioğlu, Porter, and Takáts 2009).

⁹ For example, foreign banks were allowed to provide foreign currency services to Chinese residents and were permitted greater freedom in local currency operations as well. Furthermore, the participation of foreign investors in Chinese banks was promoted, with foreigners being allowed to take equity stakes of up to 25%.

¹⁰ The PBC provides only summary data on the percentage of loans issued around the reference rates from December 2004 onwards. Therefore, we lack sufficient data to formally test how binding the lending rate ceiling was prior to that date.

¹¹ See Section 4.3 of the BIS WP version (http://www.bis.org/publ/work422.pdf).

¹² In this section, we assume that cost functions for each bank type are similar, as only the constant term is allowed to vary across bank groups through bank type dummies. The alternative approach is to assume different cost functions for each bank type by allowing bank type dummies to interact with independent variables. We follow this approach in Appendix E (E.5) of the BIS WP version as an additional robustness test.

¹³ As a robustness test, we estimated in Appendix E (E.3) of the BIS WP version explicitly the conjectural variation parameter as a direct measure of competition.

¹⁴ There are other types of financial institutions in China, such as trust and investment corporations, rural commercial banks, savings banks, co-operative banks, investment banks, and policy banks. We exclude these institutions from our investigation for several reasons. First, in the 1990s, trust and investment corporations were important financial institutions that operated similarly to commercial banks (Hong and Yan 1997). However, in the late 1990s, they experienced significant problems and most of them were taken over by commercial banks. Since the primary focus of this article is to assess bank competition during 1996–2008, we believe it is safe to exclude trust and investment corporations from our analysis. Second, most of the other banks that are not included in our investigation capture only very small portions of Chinese lending markets and/or were established with different objectives from commercial banks. Third, there are significant data limitations for especially the large number of small banks that are excluded from the sample.

¹⁵ Banks with more than 50% foreign ownership are classified as foreign banks. We only include foreign banks that provide separate balance sheet data to the PBC. This means that several banks headquartered in Hong Kong (Special Administrative Region) and which are classified as foreign banks by the CBRC, but do not provide separate balance sheet data for their operations in mainland China, are excluded from our sample.

¹⁶ The results for the Panzar–Rosse H-statistic are reported in Appendix A of the BIS WP version (http://www.bis.org/publ/work422.pdf). The estimation results for cost Equation (2) and supply Equation (6) that form the basis for estimating the elasticity-adjusted Lerner index are reported in Appendix B of that WP.

¹⁷ These results are reported in Appendix E (E.5) and Appendix E (E.3) of the BIS WP version.

¹⁸ Estimating marginal costs from homogenous translog cost function (TCF) does not change our results. See our BIS WP version for robustness test.

¹⁹ The estimation of the TCF is reported in Appendix C (Table C.1) of the BIS WP version.

²⁰ In practice, a positive β_t is possible (Van Leuvensteijn et al. 2011), which could be the result of extreme collusion, market regulation or banks competing on quality (Tabak, Fazio, and Cajueiro 2012).

²¹ The Stock-Yogo weak ID test critical value at 10% (maximal Limited Information Maximum Likelihood (LIML) size) is 16.38 (Stock, Wright, and Yogo 2002).

²² Results with two-stage least squares (2SLS) are available upon request.

²³ Delis (2012) estimates worldwide bank competition with the profit elasticity (PE) indicator, including China for a sample covering 1988–2005 and using market shares as performance indicator. Our values are not that different from the values reported for China in that study, especially those reported for 2002–2004.

²⁴ Using year dummies instead of a time trend generates similar results for all estimations reported in this article. Results are available upon request.

²⁵ These robustness tests are presented in Appendix E of the BIS WP version, for an alternative definition of the PE indicator (E.4) and for an alternative calculation of marginal costs (E.5).

²⁶ See section 4.3 of the BIS WP version for theoretical proof.

²⁷ He and Wang (2012, 34): ‘… Using the regression results, we can then estimate the equilibrium interest rate by subtracting the effects of financial repression from the observed real interest rate: The equilibrium deposit rate in China was estimated at 4.7% in 2005. This estimated equilibrium deposit rate is significantly higher than the observed real deposit rate of 1.6% in 2005, which means that the deposit-rate ceiling must have been binding in China.’

²⁸ See section 4.3 of the BIS WP version of this article and Salvo (2010).

²⁹ The values of these indicators are shown in Appendix D of the BIS WP version, Table D.1.

³⁰ We also employed the current values of the financial reform indices to account for the possibility that banks may anticipate financial reform measures and adjust their competitive strategies accordingly. The results are similar to the ones we report here.

³¹ We find a similar result for the Panzar–Rosse H-statistic. This is presented in the BIS WP version, which is available at http://www.bis.org/publ/work422.pdf.