The evolution of China’s banking system: bank loan announcements 1996–2009
Abstract
This study investigates China’s evolving banking systems from 1996 to 2009 by testing the market response to bank loan announcements in the China. The results show a significant negative market response to bank loan announcements in the Chinese financial market for the sample period 1996–2004. However, after a series of reforms in the Chinese banking system, the significantly negative market response to bank loan announcements disappears for the sample period 2005–2009.
1. Introduction
Most studies examining the effects of bank loan announcements have focussed on the United States. For example, Billett et al. (1995) and James (1987) have found a significant, positive excess return following bank loan announcements. These studies empirically confirm Fama’s (1985) argument of the uniqueness of bank loans in the US financial market. Similarly, a number of studies have extended this topic to other markets, such as Canada (Aintablian and Roberts, 2000; Mathieu et al., 2006), the UK (Armitage, 1995b; Franks and Sussman, 2005), Australia (Fery et al., 2003), New Zealand (Koh, 2001), Japan (Kang and Liu, 2008), Hong Kong and South Korea (Boscaljon and Ho, 2005). The results of these revealed a common finding; banks operating in non-government-controlled systems grant loans based only on the borrowing firms’ performance rather than political or social objectives and are thus able to allocate funds efficiently. This implies that the decision to grant loans is based only on the borrowing firms’ performance. Thus, loan approval is considered a positive signal by the stock market. Banks have indeed played a role in screening and monitoring borrowing firms under the enriched notion of ‘inside debt’.
It is unclear whether the traditional predictions of bank performance for non-government-controlled banking systems also hold for government-controlled ones. In the government-controlled context, such as in China and Taiwan, results are ambiguous. For example, Chen and Tsai (2006) and Cui and Zhao (2004) report significantly positive abnormal returns, whereas Bailey et al. (2011) and Shen et al. (2007) find significantly negative returns. Bailey et al. (2011) explain their findings by suggesting that state-controlled banks may have to lend to weak firms for political reasons, such as avoiding widespread unemployment and social instability. If weak borrowers are prevalent, the market response to a bank loan announcement should be negative and vice versa.
This ambiguity may stem from different sampling periods or institutional settings across nations and industries. For example, Bailey et al. (2011) studied the period from 1999 to 2004, Cui and Zhao (2004) focussed on 1 January 2004–20 May 2004, whilst Shen et al. (2007) examined 1 January 2005–18 November 2005. However, three of the Big Four state-owned banks have become publicly listed and traded companies since 2005, with the government being the largest shareholder and retaining control. Listed banks are presumed to impose market discipline on directors and managers, improve accuracy and transparency of disclosure to international standards and subject bank performance to market appraisals of efficiency and profitability. Furthermore, under a World Trade Organization agreement in 2006, the banking sector in China was opened to foreign banks to increase competition.
This study investigates the market reaction to bank loan announcements in China’s financial market from 1996 to 2009, where the banking system is highly controlled by the government. To do so, we examine lending bank characteristics that may influence share price reactions to bank loan announcements in the Chinese financial market.
Zhang et al. (2012) studied the impact of bank loan announcements with regard to borrower characteristics. Their analysis demonstrates that Chinese market reforms regarding share splits had no effect on borrowers’ quality. The authors also found reforms relevant to borrowers did not change borrower behaviour. In contrast, this study considers the manner in which lender characteristics influence the response of borrower share prices to loan announcements over the same period examined by Zhang et al. (2012).
2. Literature review
Previous studies (e.g. Billett et al., 1995; Aintablian and Roberts, 2000; Boscaljon and Ho, 2005) argue that bank loan announcements yield positive abnormal returns for the borrowers’ stock when banks operate in non-government-controlled banking systems. Banks are thus able to allocate their funds efficiently. However, it is unclear how the market response to bank loan announcements differs when banks are owned or controlled by the state.
La Porta et al. (2002) argue that despite the wave of privatization in 1980s and 1990s, government ownership of banks is still prevalent, in particular, in countries with low levels of per capita income, backward financial systems, interventionist and inefficient governments, and poor protection of property rights. Altunbas et al. (2001), Barth et al. (2004) and Berger et al. (2004) argue that banks with a high degree of government ownership, especially banks owned by the state, may not be as efficient as privatized banks in lending because state-owned banks often pursue political objectives rather than profit and value maximization, which generally result in unfavourable economic consequences. Berger et al. (2005) and Iannotta et al. (2007) have found that government ownership of banks is associated with high risk-taking, negative net present value projects, weak monitoring and/or lack of aggressive collection procedures. These indicate that the market may unfavourably react to loans issued by banks operating with government interference, as this inhibits banks from allocating their assets according to market criteria.
Only six studies have investigated market response to bank loan announcements in government-controlled banking systems. Cui and Zhao (2004) used the event study method to investigate market response in China. Consistent with studies conducted in the United States, Canada, the UK, Australia, New Zealand, Japan, Hong Kong and South Korea, they found a significantly positive market reaction to such announcements. Limitations of that study include a relatively small sample (53 announcements) and a shorter period of observation than studies of other financial markets.
Chen and Tsai (2006) examined the information content of syndicated loan announcements in Taiwan and found they add significant positive value to borrowing firms. However, their sample was smaller than Cui and Zhao’s (2004), with only 40 announcements included.
Shen et al. (2007) investigated market response to bank loan announcements in the Chinese financial market and found a significant negative reaction to them. However, the authors do not provide a reasonable explanation for their findings. In addition, unlike prior studies using 2 or 3 days as the event window, Shen et al. (2007) employ a longer (30 days) period, suggesting that the Chinese financial market is not as efficient as others, implying that market response is relatively slow.
Bailey et al. (2011) reported that borrowers’ stock value reacts negatively to bank loan announcements in the Chinese financial market. According to the authors, the lending activities of state-owned or state-controlled banks are often driven by political considerations such as avoiding unemployment and social instability, rather than economic considerations. This implies that state-owned or state-controlled banks may have to lend to poorly performing firms for political reasons. Compared with the conventional notion that approval of a bank loan signals a high-quality, creditworthy borrower, approval of a bank loan in China may signal a poor-performing firm that cannot obtain sufficient funds from operations. Loans are then seen as a bailout from the government-controlled banking system. Thus, the direction of the market response to bank loan announcements is hypothesized to be negative in the Chinese financial market system.
Zhang et al. (2012) studied the impact of bank loan announcements with regard to borrower characteristics. The authors found that negative market responses to bank loan announcements are particularly significant for borrowing firms with lower quality, including firms that are opaque, have a higher possibility of expropriation or tunnelling, have ineffective expropriation mechanisms, are controlled by the state and are in financial distress for the study period. The authors’ subperiod analyses (1996–2004 and 2005–2009) aimed to capture whether a split share reform (pertaining to corporate ownership) lessened the expropriation problem for borrowers. Results demonstrate that reforms did not affect the results on borrower’s quality for either sample period. The authors also found reforms relevant to borrowers were not successful in changing borrower behaviour.
Gao et al. (2006) suggested that the market may also react unfavourably to bank loan announcements when the bank’s location is in a province with a lower marketization level in credit allocation in China. Therefore, the negative effect of a bank loan announcement is particularly significant for loans from the Big Four state banks, state-owned or state-controlled banks, lower ranked banks and banks from such provinces.
Cui and Zhao (2004) found that both new loans and loan renewals can elicit a significantly positive reaction, but the positive effect of bank loan announcements is pronounced for loan renewals in the Chinese financial market. This result is similar to that of Aintablian and Roberts (2000) and Boscaljon and Ho (2005), but fails to confirm Fama’s (1985), Lummer and McConnell’s (1989) and Armitage’s (1995b) view that only renewal loans can generate a significantly positive response. In terms of loan maturity, Bailey et al. (2011) discover that in the Chinese market the negative effect of a bank loan announcement is particularly significant for loans with shorter maturity. This finding is inconsistent with Fama (1985), Flannery (1986) and Diamond (1991, 1993), who argue that the market reacts favourably to shorter-term loans. Bailey et al. (2011) explain that many Chinese listed firms use short-term loans to fund long-term assets. Therefore, the negative effect of a bank loan announcement is particularly significant for revised loans, loans for a greater amount, shorter term, with covenant/collateral, and less syndication.
With regard to loan purpose, Chen and Tsai (2006) argue there are significant abnormal returns for loans for capital expenditure, but found a non-significant response for loans for refinancing debt, purchasing facilities, stand-alone projects and other projects in the Taiwanese market. Bailey et al. (2011) found that the negative effect of a bank loan announcement is particularly significant for loans used to repay old loans in the Chinese market. These findings contrast with the Boscaljon and Ho (2005), Lee and Sharpe (2006), Lee and Sharpe (2009) and Slovin et al. (1992) finding that stock markets react favourably to loans used to refinance existing debt.
Past lending practices entirely based on political considerations have left Chinese banks with a large number of non-performing loans (NPLs) (Tian, 2004; Dobson and Kashyap, 2006). Comparing China with other major Asian economies in recent years, Allen et al. (2008) show that the rate of NPLs is the highest in China; profitability of China’s banking system is the lowest for the same group of countries. To resolve the NPLs, Chinese authorities restructured state-owned or state-controlled banks. In December 2003, the Bank of China (BOC) and the China Construction Bank (CCB) were selected as pilot banks for reform. The authorities announced the decision to recapitalize these two banks to strengthen their corporate governance structure and risk management, resolve NPLs, and specified that the use of reputable external auditors to assess the financial position of the banks as well as enhance external oversight of their’ operations.
Following financial restructuring, the CCB listed in Hong Kong in 2005, followed by the BOC in Hong Kong and Shanghai in 2006. In early 2005, the authorities approved the restructuring of the ICBC, the largest commercial bank in China following the same restructuring process of BOC and the CCB. The Industrial and Commercial Bank (ICBC) listed in Hong Kong and Shanghai in 2006 (Podpiera, 2006). Following the public listing, the CCB, the BOC and the ICBC incorporated as joint-stock companies, introduced new corporate governance structures, worked on changing risk management and internal organization and brought in strategic investors (Podpiera, 2006). These changes are especially noteworthy in that the ICBC, BOC and CCB are the top three banks in China in terms of their tier one capital and assets (PricewaterhouseCoopers, 2011).
Beyond state-owned banks, reforms have also been extended to state-controlled banks. For example, in 2005 and early 2006, foreign ownership was approved for a number of state-controlled banks such as Huaxia Bank, Bohai Bank, Guangdong Development Bank and Bank of Beijing (Podpiera, 2006). In addition, at the end of 2006, under a World Trade Organization agreement, the banking sector in China was opened to foreign banks. As a consequence, it is plausible that China’s banking system has improved since these reforms. The market response to bank loan announcements may thus differ from the results found in Bailey et al.’s (2011), Cui and Zhao’s (2004) and Shen et al.’s (2007) studies.
3. Details of data and sample selection and research method
3.1. Data and sample selection
This study sampled all bank loan announcements from companies listed on the Shanghai Stock Exchange (SHSE) and the Shenzhen Stock Exchange (SZSE) between 1996 and 2009.1 The sample period is divided into two subsamples, 1996–2004 and 2005–2009, owing to the series of reforms that began in 2005. These included allowing strategic investors, listing of banks’ shares and restricting the share of government ownership.
Bailey et al. (2011) argue that a credit agreement may be extended to all companies and therefore may be considered as a very weak loan ‘pre-qualification’, containing less information than an actual bank loan announcement.2 The authors report that credit agreement effects are much weaker than those of bank loan announcements. This study focuses only on bank loans that actually took place, that is, bank loan announcements with contracts. As a consequence, ‘credit agreement’ announcements were excluded from the study.
In addition, to minimize the effect of confounding events, ‘contaminated announcements’ (Lummer and McConnell, 1989; Bailey et al., 2011) that accompanied bank loan announcements with other corporate events, such as mergers and acquisitions, CEO turnovers, other types of concurrent financing arrangements and lawsuits within the event window of the loan announcements, were excluded from the study. Further, for the subsample period 2005–2009, we excluded announcements that accompanied share-split reform during the estimation window and event window to avoid the influence of share-split reform that might influence the results of analysis.
This study also considered only firms that were traded on the SHSE and the SZSE A-shares3 exchanges. This includes the A-share price when the borrowing firms issued both A-shares and B-shares4 but excludes firms that do not issue A-shares; otherwise, we would have to use the share price in foreign currencies, but such market values are not comparable (Mei et al., 2009; Chan et al., 2008). We also eliminated observations with missing borrower returns in the estimation of the event period, yielding a final sample of 501 bank loan announcements for the period 1996–2004 and 106 bank loan announcements for the period 2005–2009.
Because not all components of bank loan announcement are reported, sorting bank loan announcements by bank and loan characteristics can lead to a substantial decrease in the sample size. We did not divide the bank loan announcements by bank characteristics because bank characteristics are not unitary, that is, an announcement can include a number of loans from different banks. Similarly, if a bank loan announcement included a number of loans from the same bank, we did not divide the bank loan announcement by loan characteristics.
The daily individual stock returns and daily A-shares market returns were collected from the CSMAR® China Stock Market Trading Database. These returns were based on adjusted closing daily prices5 and were further adjusted for cash dividend reinvestment. The daily market returns were based on the total-value-weighted portfolio consisting of all the A-shares traded on the SHSE and the SZSE. The marketization index in credit allocation for China’s provinces was obtained from Fan and Wang (2001) and Fan et al. (2002, 2004, 2007). The index examines the magnitude of local government intervention in banks located in different provinces.
3.2. Research method
This study uses the event study to investigate borrowers’ stock price reaction to bank loan announcements. Abnormal bank loan announcement return was specified as the dependent variable in a multivariate cross-sectional regression analysis focussed on explaining market response to bank loan announcements.
We employ ‘Day 0’ as the date on which the announcement appears in the media. Following this, we allow the estimation window to end 30 days before the event date. The estimation period and the event period are chosen to avoid overlap so that the parameters of the model are not influenced by the event (Peterson, 1989; MacKinlay, 1997). We define 21 days [−10, 10] as the event window, following Armitage’s (1995a) method, with changes for application to Chinese capital markets. These include adjusting the period before the announcement date, because influential information may have been divulged before the firms formally released the information (Shen et al., 2007). Employing a slightly longer period after announcements allows for slower dissemination of information on less visible and infrequently traded stocks. The event windows in our study are wider than those in the US studies cited above. This is because the Chinese financial market is not as efficient as the US financial markets, which may result in a slower reaction to information in the Chinese capital market (Shen et al., 2007). We followed the standard market model approach and employ ordinary least squares (OLS) regression to estimate the model parameters.
The estimated abnormal returns are standardized before testing for statistical significance (James, 1987; Peterson, 1989) to reflect the statistical error in the determination of expected returns. Finally, average abnormal return is tested for significance using the sign test and the Wilcoxon signed rank test.
3.2.1. Multivariate cross-sectional models
()-
CAR(t1, t2) = cumulative abnormal return from time t1 to time t2;
-
BIG4_BANK = 1 if lender is a Big Four state-owned banks (0 otherwise);
-
BANK_OWNERSHP = 1 if lender is either a state-owned or state-controlled bank (0 otherwise6);
-
BANK_RANKING = 1 if loan is issued by a local branch below the provincial level (0 otherwise);
-
BANK_LOCATION = 1 if bank is in a lower marketization level province in credit allocation (0 otherwise);
-
LOAN_TYPE = 1 if the loan is identified as a subsequent loan (0 otherwise7);
-
LOAN_MATURITY = 1 end of the life of the loan8 (0 otherwise); and
-
COVENANT/COLLATERALS = 1 if the loan has covenants/collaterals (0 otherwise).
()Equation 3 that uses BANK_OWNERSHIP and LOAN_MATURITY as proxy for bank and loan characteristics to explain the market response to bank loan announcement for the combined sample period from 1996 to 2004.
()4. Empirical results
4.1. Summary statistics
The sample was split into two subperiods: 1996–2004 and 2005–2009. These were used to test for significant differences of the effects of bank efficiency on the market response to bank loan announcements following banking system reforms. Table 1 shows the descriptive statistics for banks, borrowers and loans by subsamples. Panel A presents the annual distribution of the 607 bank loan announcements in the second sample; there were fewer announcements from 1996 to 2000 than for 2001–2004. Chinese companies offering securities were not required to disclose to the public until 1998 (Trading Rules of Shanghai Stock Exchange, 1998; Trading Rules of Shenzhen Stock Exchange, 1998). Until 2000, companies offering securities were required to disclose important issues on one of the officially designated media by the CSRC (The First Revision of Trading Rules of Shanghai Stock Exchange, 2000; The First Revision of Trading Rules of Shenzhen Stock Exchange, 2000). Thus, records on bank loan announcements before 2000 are rare.
| Year of announcement | Number of announcements |
|---|---|
| Panel A: Annual distribution of ‘clean’ bank loan announcements in the final sample (501 bank loan announcements) | |
| 1996 | 1 |
| 1997 | 0 |
| 1998 | 2 |
| 1999 | 8 |
| 2000 | 22 |
| 2001 | 75 |
| 2002 | 133 |
| 2003 | 135 |
| 2004 | 125 |
| 2005 | 40 |
| 2006 | 5 |
| 2007 | 21 |
| 2008 | 22 |
| 2009 | 18 |
| 1996–2004 | 2005–2009 | |
|---|---|---|
| Observations | Observations | |
| Panel B: Descriptive statistics of lending banks | ||
| Lender is one of the ‘Big Four’ banks | 259 | 37 |
| Lender is one of the state-owned or state-controlled banks | 473 | 85 |
| Lender is one of the privately owned banks | 14 | 9 |
| Lender is one of the local branch banks | 278 | 53 |
| Lender is one of the banks in the province with higher marketization level in credit allocation | 320 | 58 |
| 1996–2004 | 2005–2009 | |||
|---|---|---|---|---|
| Mean Observations | Mean Observations | |||
| Panel C: Descriptive statistics of loans | ||||
| Amount of loan (million yuan) | 101.72 | 488 | 457.92 | 82 |
| Maturity of loan (years) | 1.46 | 460 | 2.36 | 70 |
| Interest rate on loan (%) | 5.56 | 248 | 6.16 | 37 |
| Subsequent loans | 174 | 22 | ||
| Loans with covenants/collaterals | 294 | 68 | ||
| Non-syndicated loans | 496 | 97 | ||
- The sample consists of ‘clean’ bank loan announcements made by companies listed on the Shanghai Stock Exchange and the Shenzhen Stock Exchange from 1996 to 2004. ‘Clean’ loan announcements are not contaminated by any confounding corporate events and have returns based on actual transaction prices.
Panel A shows that bank loan announcements from 2005 to 2009 were much fewer than for 1996–2004. This is because we excluded announcements that accompanied share-split reform during the estimation window to avoid the influence of the share-split reform. Also, from 2004 to 2008, the China central bank raised the required reserve ratio for commercial banks 18 times, from 7.5 per cent to 17.5 per cent, with the lending rate increasing from 5.31 per cent to 7.47 per cent. These adjustments imply that the government wanted to discourage lending from 2005 to 2008. Listed firms thus preferred to raise funds externally, because it was cheaper.
Panel B provides the descriptive statistics of bank characteristics. From 1996 to 2004, over half of the bank loan announcements in the subsample (259 of 453) were made by one of the Big Four state-owned banks. Just over 97 per cent of the loans (473 of 487) are from state-owned or state-controlled banks, with only 13 from privately owned banks and one from a foreign bank. In 2005–2009, 45 per cent (37 of 83) of the loan announcements were made by the Big Four state-owned banks. This implies that the dominant position of the Big Four state-owned banks was challenged by the joint-stock banks and foreign banks.
Nearly, two-thirds of the loans were issued by local branches in both subsamples. In addition, almost two-thirds of the loans (320 of 485) were made by banks in provinces with higher marketization levels in credit allocation in 1996–2004. This percentage increases in the next period. Nearly three-quarters of loans, 72.5 per cent (58 of 80), were made by banks in provinces with higher marketization levels in credit allocation in 2005–2009.
Panel C presents descriptive statistics for loan characteristics. During 1996–2004, the average loan was 101.72 million yuan; loan maturity has mean of 1.46 years. The interest rate mean was 5.56 per cent. One-third of the loans (174 of 501) were identified as subsequent loans. Most loans (496 of 501) were made by single banks. Almost two-thirds of the loans required covenants or collateral.
During 2005–2009, the average loan was for 457.92 million yuan, significantly higher than during the first sample period. Mean loan maturity was 2.36 years, which is longer than first period. Similar to 1996–2004, most loans (97 of 106) were made by single banks, and almost two-thirds required covenants or collateral. However, only one-fifth were identified as subsequent loans, which was fewer than subsequent loans for the period 1996–2004. This implies that the ongoing monitoring process by banks was more efficient after the reforms. It was therefore difficult for troubled borrowers to get subsequent loans.
4.2. Abnormal returns around bank loan announcements
4.2.1. Summary statistics
Table 2 provides summary statistics for abnormal returns around bank loan announcements and parametric and nonparametric tests for the period 1996–2004. For each day in the event period, abnormal returns (
) were averaged across the 501 bank loan announcements. Based on the daily 
, there were significantly negative abnormal returns. For example, the 
on day 0 was −0.07 per cent, which is statistically significant at the 10 per cent level, and the 
on day 1 is −0.16 per cent, which is statistically significant at the 1 per cent level. For nearly all event windows, CARs are negative and significant (see Panel B in Table 3). The average [−1, 0] CAR is −0.17 per cent and the average [0, 1] CAR is −0.24 per cent, which are statistically significant at the 1 per cent level. Thus, the market response for the borrowing firm’s equity over several days following the bank loan announcement was typically negative. This result is consistent with Shen et al. (2007) and Bailey et al. (2011) studies, but inconsistent with studies in non-government-controlled banking systems that show bank loan announcements produce significant positive abnormal returns. These results imply that Chinese banks were not able to allocate their funds efficiently.
| Event day or window (0: announcement day) |  (%) |
t-test | Per cent of  > 0 (%) |
Sign test | Wilcoxon signed rank test |
|---|---|---|---|---|---|
Panel A: Average abnormal return ( ) |
|||||
| −10 | −0.03 | −0.30 | 45.31 | −2.10** | −1.42 |
| −9 | 0.16 | 1.98** | 49.90 | −0.04 | −1.08 |
| −8 | −0.02 | −0.30 | 47.11 | −1.30 | −1.29 |
| −7 | 0.06 | 0.51 | 47.40 | −1.03 | −0.06 |
| −6 | −0.05 | −0.48 | 46.11 | −1.74* | −1.60 |
| −5 | 0.01 | 0.26 | 46.11 | −1.74* | −1.04 |
| −4 | 0.02 | 0.25 | 49.10 | −0.40 | −0.40 |
| −3 | 0.12 | 1.18 | 46.71 | −1.47 | −0.28 |
| −2 | −0.04 | −0.68 | 45.11 | −2.19** | −1.53 |
| −1 | −0.10 | −1.02 | 44.71 | −2.37** | −1.92* |
| 0 | −0.07 | −1.73* | 43.91 | −2.73*** | −2.24** |
| 1 | −0.16 | −2.02** | 41.52 | −3.80*** | −2.91*** |
| 2 | −0.03 | −0.81 | 45.11 | −2.19** | −2.03** |
| 3 | −0.19 | −2.95*** | 41.52 | −3.80*** | −3.04*** |
| 4 | 0.05 | 1.15 | 48.50 | −0.67 | −0.16 |
| 5 | 0.06 | 0.95 | 50.90 | 0.40 | −0.24 |
| 6 | −0.06 | −0.89 | 47.50 | −1.12 | −1.23 |
| 7 | −0.15 | −1.76* | 47.11 | −1.30 | −2.06** |
| 8 | −0.04 | −0.37 | 46.31 | −1.65* | −1.43 |
| 9 | −0.11 | −1.22 | 47.31 | −1.21 | −1.76* |
| 10 | −0.11 | −1.48 | 42.12 | −3.53*** | −2.74*** |
| Event day or window (0: announcement day) |  (%) |
t-test | Per cent of  > 0 (%) |
Sign test | Wilcoxon signed rank test |
| Panel B: Cumulative average abnormal return (CAR) | |||||
| [−10, −1] | 0.14 | 1.18 | 50.10 | 0.04 | −0.18 |
| [−1, 0] | −0.17 | −2.74*** | 44.51 | −2.46** | −1.77* |
| [0, 1] | −0.24 | −3.74*** | 43.51 | −2.90*** | −2.68*** |
-
The average abnormal return
) and cumulative average abnormal return (CAR) were calculated using the market model and standard event study methodology. The estimation window for calculating the market model parameters was [−150, −31]. 
and CAR were tested for significance using a two-tailed t-test. The sign test categorizes data into binary outcomes with the null hypothesis that the percentage of negative 
(CAR) equals the percentage of positive 
(CAR). The Wilcoxon signed rank test has a null hypothesis of no difference in magnitudes between the negative and positive 
(CAR). ‘***’, ‘**’ and ‘*’ indicate significance at the 1, 5 and 10 per cent levels, respectively. The number of observations is 501.
| Event day or window (0: announcement day) |  (%) |
t-test | Per cent of  > 0 (%) |
Sign test | Wilcoxon signed rank test |
|---|---|---|---|---|---|
| −10 | 0.16 | 0.63 | 49.06 | −0.19 | −0.26 |
| −9 | −0.05 | −0.15 | 47.17 | −0.58 | −0.12 |
| −8 | 0.09 | 0.46 | 44.34 | −1.17 | −0.37 |
| −7 | −0.11 | −0.03 | 45.28 | −0.97 | −1.13 |
| −6 | −0.23 | −0.72 | 48.11 | −0.39 | −1.25 |
| −5 | 0.31 | 1.59 | 50.00 | 0.00 | −0.35 |
| −4 | −0.15 | −0.27 | 44.34 | −1.17 | −0.98 |
| −3 | −0.25 | −1.04 | 41.51 | −1.75* | −1.65* |
| −2 | 0.12 | 0.44 | 53.77 | 0.77 | 0.31 |
| −1 | −0.18 | −1.11 | 41.51 | −1.75* | −1.06 |
| 0 | 0.03 | 0.59 | 46.23 | −0.77 | −1.04 |
| 1 | −0.23 | −1.27 | 41.51 | −1.75* | −2.12* |
| 2 | 0.39 | 1.63 | 55.66 | 1.17 | 0.35 |
| 3 | −0.10 | −0.26 | 50.00 | 0.00 | −0.65 |
| 4 | 0.07 | 0.40 | 47.17 | −0.58 | −0.53 |
| 5 | 0.35 | 1.22 | 50.94 | 0.19 | 0.37 |
| 6 | 0.14 | 0.83 | 44.34 | −1.17 | −0.91 |
| 7 | 0.33 | 1.13 | 52.83 | 0.58 | −1.22 |
| 8 | −0.40 | −1.20 | 40.57 | −1.94* | −2.01** |
| 9 | −0.14 | −0.47 | 47.17 | −0.58 | −0.44 |
| 10 | −0.47 | −1.54 | 38.68 | −2.33** | −1.46 |
-
The average abnormal return
) and cumulative average abnormal return (CAR) were calculated using the market model and standard event study methodology. The estimation window for calculating the market model parameters was [−150, −31]. 
and CAR were tested for significance using a two-tailed t-test. The sign test categorizes data into binary outcomes with the null hypothesis that the percentage of negative 
(CAR) equals the percentage of positive 
(CAR) The Wilcoxon signed rank test has a null hypothesis of no difference in magnitudes between the negative and positive 
(CAR). ‘***’, ‘**’ and ‘*’ indicate significance at the 1, 5 and 10 per cent levels, respectively. The number of observations is 106.
In addition, the average [−10, −1] CAR is positive and insignificant, implying that there was no systematic information leakage before the bank loan announcements. Thus, we focus on the CAR over a 2-day event window [0, 1].
Table 3 provides summary statistics for abnormal returns around bank loan announcements and parametric and nonparametric tests for the period 2005–2009. There were no significant abnormal returns under the parametric tests (two-tailed t-test) for the period 2005–2009. This result is inconsistent with findings for the period 1996–2004. This means the Chinese stock market no longer viewed bank loan announcements unfavourably following the series of reforms. From 2005, reforms included introducing strategic investors, listing of banks’ shares and restricting the share of government ownership, thereby reducing government intervention in Chinese banks. Banks may have more authority and freedom to grant loans for commercial reasons and take full advantage of their unique information to screen and monitor borrowers.
However, there does not appear to have been a significant positive market response to bank loan announcements in the Chinese financial market, although the banking system has improved following the reforms. Changes in the banking system are incomplete, and government interference in the system is still substantial in certain areas. For example, most Chinese banks are still controlled by the government, which makes it difficult for individuals or institutions to compete through ‘greenfield’ investment and direct participation in Chinese state-owned banks. Chinese banks cannot reject policy loans completely at this stage.
4.3. Size of the market response to bank loan announcements
To further examine the effects of bank and loan characterizes on the size of the market response to bank loan announcements, our study groups bank loan announcements into subsamples according to bank and loan characteristics then conducts univariate tests and utilizes multivariate cross-sectional regression analyses to test whether the abnormal returns are statistically different between the two groups in the subsamples. The effects of bank and loan characteristics on the size of the market response to bank loan announcements can then be assessed for the sample period 1996–2004. The results of univariate tests on subsamples are reported in Table 4.
| Category | No. of observations | CAR [0, 1] (%) | t-test of CAR = 0 | Per cent of CAR positive | Sign test | Wilcoxon signed rank test |
|---|---|---|---|---|---|---|
| Panel A: Bank characteristics | ||||||
| Bank type | ||||||
| ‘Big Four’ state banks | 259 | −0.25 | −3.18*** | 43.51 | −2.55** | −2.75*** |
| Other banks | 194 | −0.24 | −1.76* | 45.88 | −0.85 | −0.75 |
| Bank ownership | ||||||
| State-owned/state-controlled banks | 473 | −0.30 | −4.44*** | 42.49 | −3.26*** | −3.15*** |
| Private banks | 14 | 0.83 | 2.12** | 71.43 | 1.60 | −1.54 |
| Bank ranking | ||||||
| Loans issued by local branches | 278 | −0.22 | −3.08*** | 42.09 | −2.64*** | −2.28** |
| Loans issued by headquarters or main provincial branches | 173 | −0.24 | −1.59 | 48.55 | −0.38 | −0.48 |
| Bank location | ||||||
| Bank in the province with the higher marketization level in credit allocation | 320 | −0.15 | −2.57** | 45.00 | −1.79* | −1.97** |
| Bank in the province with the lower marketization level in credit allocation | 165 | −0.42 | −2.82*** | 40.00 | −2.60*** | −2.00** |
| Panel B: Sorted by loan characteristics | ||||||
| Loan type | ||||||
| First loans | 327 | −0.13 | −2.13** | 45.26 | −1.71* | −1.60 |
| Subsequent loans | 174 | −0.43 | −3.43*** | 40.22 | −2.58*** | −2.39** |
| Size | ||||||
| Above median | 244 | −0.23 | −2.76*** | 42.62 | −2.30** | −1.93* |
| Below median | 244 | −0.20 | −1.94* | 45.49 | −1.41 | −1.33 |
| Maturity | ||||||
| 1 year or shorter | 376 | −0.36 | −4.30*** | 42.67 | −2.84*** | −2.88*** |
| Longer than 1 year | 84 | 0.34 | 0.81 | 51.19 | 0.22 | −0.40 |
| Covenants/collateral | ||||||
| With covenants/collateral | 294 | −0.28 | −3.01*** | 44.22 | −1.98** | −2.07** |
| Without covenants/collateral | 196 | −0.15 | −2.09** | 42.35 | −2.14** | −1.63 |
| Syndication | ||||||
| Non-syndicated loans | 496 | −0.26 | −3.90*** | 43.54 | −2.87*** | −2.75*** |
| Syndicated loans | 5 | 2.19 | 1.36 | 40.00 | −0.44 | −0.27 |
-
Table 4 reports the [0, 1] cumulative average abnormal returns (CAR) around bank loan announcements by subsamples with parametric and nonparametric significance tests. The sign test has a null hypothesis that the percentage of negative
(CAR) equals the percentage of positive 
(CAR). The Wilcoxon signed rank test has a null hypothesis of no difference in magnitude between the negative and positive 
(CAR). ‘***’, ‘**’ and ‘*’ denote significance at 1, 5 and 10 per cent levels, respectively. The lack of data leaves the total number of observations below 501.
4.3.1. Effects of bank characteristics
Table 4 reports the summary statistics for the [0, 1] cumulative average abnormal returns divided by bank (Panel A) and loan (Panel B) characteristics for the sample period 1996–2004.
Panel A in Table 4 presents the univariate statistics on bank loan announcements by bank characteristics. The negative stock return effect is significantly stronger for loans from Big Four state banks than for loans from other banks. In addition, there is a significant positive response to loans by private banks, although the sample is quite small (2.8 per cent; 14 of 501) with only 13 loans from a privately owned bank (China Mingsheng Bank) and one from a foreign bank (Bank of East Asia). The 2-day CAR for loans issued by provincial level branches and headquarters are not significantly different from zero. However, the negative CAR is significant for loans made by a bank’s local branch, such as municipal or township branch bank below the provincial level. This reveals that local branches are more inclined to suffer intervention by the local government and suffer higher moral hazard.
Analysis also shows that loans issued by banks in provinces with the lower marketization levels in credit allocation have more negative CAR than loans made by banks in provinces with higher marketization levels. Banks in provinces with lower marketization levels in credit allocation are routinely pressured by the government to supply policy loans. Our results are consistent with the Gao et al. (2006) and Bailey et al. (2011) findings, which show that negative market responses to bank loan announcements are prevalent if lending banks are operating under strong political interference and suffer from greater pressure to issue loans for non-value-maximizing purposes.
Loans from banks with higher government ownership and stronger political interference transmit a negative signal. The Big Four state-owned banks in China are subject to more political interference and assume more policy loans than other domestic banks. Chinese banks with lower rankings (local branches) may be influenced by local governments and pressured to issue loans to pursue the political interests of those local governments (Bailey et al., 2011). Banks in provinces with lower marketization levels in credit allocation suffer from stronger interference by local governments because marketization levels in credit allocation varies in different provinces in China (Fan and Wang, 2001; Fan et al., 2002, 2004, 2007, 2009; Gao et al., 2006). Therefore, the negative market response to bank loan announcements is significantly larger if the lending bank is a state-owned or state-controlled bank, especially the Big Four state-owned banks, banks with lower rankings and banks in provinces with lower marketization levels in credit allocation.
4.3.2. Effects of loan characteristics
This study also divides the bank loan announcement data by loan characteristics. Results of this analysis are reported in Panel B of Table 4. Here, we find that both first loans and subsequent loans can elicit a significant negative reaction, but the negative effect of bank loan announcements is pronounced for subsequent loans in the Chinese financial market. This is consistent with Aintablian and Roberts (2000), Cui and Zhao (2004) and Boscaljon and Ho (2005) who have argued that there is greater information content in bank loan announcements for subsequent loans than for first loans. This finding implies that government-controlled banking systems such as in China may not favour subsequent loans, because such loans may be used to prop up troubled firms. Therefore, a subsequent loan is perceived by the Chinese stock market as a negative signal, indicating that the borrowing firm is experiencing financial distress.
The results also show that the negative effect of bank loan announcements is more pronounced, if the loan size is larger. This implies that the market may not favour larger sized loans, because they may be used to support troubled firms in a banking environment where non-commercial motivations are common. This is consistent with the Easterwood and Kadapakkam (1991), Krishnaswami et al. (1999) and Kang and Liu (2008) argument that there is greater information content in bank loan announcements for loans of larger than smaller sizes. This finding implies that government-controlled banking systems may not favour larger loans because the loans are used to help troubled firms. In the Chinese government-controlled banking system, approval of a loan may be perceived by the Chinese stock market as a negative signal that the borrowing firm is in trouble and requires the loan to keep afloat (Bailey et al., 2011). Thus, a larger loan is perceived as a signal that the firm is experiencing financial distress.
In addition, the negative CAR is significant for loans with shorter maturity (1 year or less) but insignificant for loans with longer maturity. This is consistent with Bailey et al.’s (2011) argument that in China, market considers short-maturity loans as bad news. This result implies that non-government-controlled banking systems favour shorter-maturity bank loans, whereas government-controlled banking systems such as that in China do not favour shorter-maturity bank loans. This is because borrowing firms are required to roll over shorter-maturity bank loans frequently. In non-government-controlled banking systems, repeated refinancing transactions with borrowers make banks continuously re-evaluate the borrowing firms, thereby strengthening the banks’ unique ability to assess inside information and monitor the loans effectively. However, in government-controlled banking systems such as in China, the complementary monitoring functions of shorter-maturity bank loans cannot be implemented, because the bank monitoring function is not efficient. Furthermore, many Chinese listed firms use short-term loans to fund long-term assets (Bailey et al., 2011).
Analysis also revealed significant negative reactions for both loans with covenants/collateral and for loans without. Loans with covenants/collateral have a slightly more negative effect than those without covenants/collateral. Indeed, very few Chinese listed firms are liquidated because of the support from the government. Thus, equity investors in China may not need to favour loans with covenants/collateral. This finding is consistent with Rajan and Winton’s (1995) and James and Smith’s (2000) argument that there is greater information content in bank loan announcements for loans with covenants/collateral. In non-government-controlled banking systems, the presence of debt with covenants/collateral increases the banks’ incentive to monitor by increasing the sensitivity of a lender’s return to information or by decreasing the banks’ payoff if monitoring is absent. However, in government-controlled banking systems, the complementary monitoring functions of covenants/collateral cannot be implemented because the bank monitoring function is not efficient.
The negative effect of a bank loan announcement is significant for non-syndicated loans, signifying that the Chinese stock market does not favour non-syndicated loans. This finding contradicts Preece and Mullineaux (1996), Aintablian and Roberts (2000) and Fery et al. (2003). For example, Preece and Mullineaux (1996) argue that the capacity to renegotiate a bank loan is relatively inexpensive in corporate restructuring and should complement monitoring as a source of value to borrowers. Consequently, as the number of lenders increases in a syndicate, contracting costs increase and the capacity to renegotiate declines. However, in the Chinese government–controlled banking system, banks operate in a relatively non-competitive environment because of the dominant (albeit declining) position of state-owned banks in total lending and the continuing political intervention by the government. In addition, under soft budget constraints, it is not costly for borrowing firms to switch banks. Thus, non-syndicated loans in China are not favoured by the stock market because Chinese listed firms do not use non-syndicated loans to enhance contractual flexibility and limit contracting costs. Furthermore, our results suggest that there is no significant financial market response to syndicated bank loan announcements in China. This result contradicts the Rajan (1992) and Houston and James (1996) argument that multiple bank lending or syndicated loans could enhance contractual flexibility and limit hold-up problems, which would result in a statistically positive relationship between the borrowers’ abnormal returns and syndicate size. This is because the sample of syndicated loans is quite small, just 1 per cent (five of 501 loans), in our study.
4.4. Multivariate cross-sectional analysis
There was no significant market response to bank loan announcements for the sample period 2005–2009. We therefore also investigated the effects of bank and loan characterizes on the size of the market response to bank loan announcements for the sample period 1996–2004.
Table 5 shows the state-owned or state-controlled banks coefficient (BANK_OWNERSHIP) is significantly negative. This result is consistent with the evidence in Panel A from Table 4, which implies that the government may have to issue a large number of policy loans with low operational efficiency to avert unemployment and potential instability in the country. The result is also consistent with Bailey et al.’s (2011) finding that negative market response to bank loan announcements is much stronger if the lending bank operates under strong political interference and suffers from greater pressure to issue loans for non-value-maximizing purposes.
| Equation | 1 | 2 | 3 |
|---|---|---|---|
| INTERCEPT | 0.008 (0.978) | 0.015 (1.157) | 0.368 (0.793) |
| BIG4_BANK | 0.000 (−0.118) | ||
| BANK_OWNERSHIP | −0.012 (−1.449) | −0.011 (−1.648)* | |
| BANK_RANKING | 0.002 (0.804) | ||
| BANK_LOCATION | −0.004 (−1.146) | ||
| LOAN_TYPE | −0.002 (−0.493) | −0.002 (−0.807) | |
| LOAN_SIZE | 0.000 (−0.059) | −0.001 (−0.111) | |
| LOAN _MATURITY | 0.002 (2.075)** | 0.002 (1.703)* | 0.002 (2.279)** |
| SYDICATION | −0.019 (−0.573) | ||
| COVENANTS/COLLATERAL | −0.003 (−0.852) | −0.015 (−1.415) | |
| Observations | 393 | 447 | 447 |
| Adjusted R2 | 0.002 | 0.006 | 0.012 |
- ** and * denote significance at 5 and 10 per cent levels.
In addition, loan maturity carries a significant positive coefficient. It was expected that short-term loans generate a greater positive bank loan announcement effect than long-term loans in non-government-controlled banking systems, because banks’ monitoring powers increase with more frequent renewals. However, in China, the monitoring function of banks is weakened because of intervention by the government. Thus, the positive slope on LOAN_MATURITY implies that equity investors recognize that short-term loans may not be employed as one of the complementary monitoring mechanisms by banks. In fact, some Chinese listed companies use short-term loans to fund long-term assets (Bailey et al., 2011). This behaviour leads to a stronger negative bank loan announcement effect following loans with shorter maturities than those with longer maturities. This result is consistent with the evidence presented in Panel B in Table 4.
5. Conclusions
This study’s empirical results show that the negative effect of bank loan announcements according to bank characteristics is particularly significant when loans are granted by state-owned or state-controlled banks. This is especially so for the Big Four state-owned banks, banks with lower rankings and banks in provinces with lower marketization levels in credit allocation for the sample period 1996–2004. Results are consistent with Bailey et al.’s (2011) and Gao et al.’s (2006) findings, which show that a negative market response to bank loan announcements is prevalent if the lending bank operates with strong political interference and suffers from greater pressure to issue loans for non-value-maximizing purposes.
Contrary to the previous studies that show a larger positive response to bank loan announcements for loans with larger size, shorter maturity, with covenants/collateral and with less syndication in non-government-controlled banking systems, our study results show a larger negative response to bank loan announcements for loans with larger size, shorter maturity, with covenants/collateral and with less syndication in the Chinese financial market for the sample period 1996–2004. These results imply that the market recognizes that complementary monitoring functions of traditional loan characteristics employed by banks in loan contracts are not efficient in China. Thus, the stock market in China does not favour bank loans with a complementary monitoring function, such as larger size, shorter maturity, with covenants/collateral and with less syndication.
In sum, we found that the Chinese stock market did view bank loan announcements favourably following reforms during the subsample periods. Results document an increase in efficiency in the Chinese banking environment as a result of evolving bank policies and regulation and that reforms were successful in changing lender behaviour in the context of bank loan announcements.
6. Implications
Given the finding that reforms help to improve banking efficiency to a limited extent, it might be prudent to continue reforming the Chinese banking system to further increase efficiency. The government’s huge percentage of bank ownership makes it difficult for individuals or institutional investors to compete through greenfield investment and direct participation in Chinese state-owned banks. Thus, financial liberalization via a reduction in government intervention may provide useful increases in banking efficiency.
Over the past few years, China has made substantial progress in financial liberalization. These efforts have included gradually introducing market practices into the banking system, liberalizing interest rates and opening up to foreign competition (Garcia-Herrero et al., 2006). However, government interference in the banking system is still substantial in certain areas. For example, most Chinese commercial banks and financial institutions have limited autonomy in setting their deposit and lending rates. This is because the central bank not only sets the official lending deposit rates in retail banking, but also restricts the ranges of these rates (Feyzioglu et al., 2009) by setting both ceilings and floors for lending and deposits. Commercial institutions are allowed to vary their rates, but only within the officially specified price ranges. Such regulation thereby places tight limits on pricing autonomy (The People’s Bank of China, 2005). In 2004, the central bank removed the ceiling on the lending rate; however, the unchanged floor on the rate-restricted competition for loans. They also removed the floor on the deposit rate, but the unchanged ceiling limit on deposits has restricted competition for deposits. These restrictions mean that few banks can improve their lending rate or lower their deposit rate, because they would lose clients if they did so.
In addition, the current floor of the reference lending rate and the ceiling on reference deposit offer a safe margin for commercial banks to maintain a relatively high net interest margin of more than 7 per cent. Therefore, so long as banks make loans, they can make profits. There is no need for most commercial banks to change their lending operations significantly. This safe and stable interest margin set by the People's Bank of China (PBOC) provides little incentive for banks to improve their efficiency in credit allocation and monitoring of loan performance.
Furthermore, interest rate controls limit competition among banks. Non-state banks, particularly foreign banks, cannot break the market monopoly of state banks, resulting in an inefficient use of credit and serious structural imbalances in the Chinese banking system. The role of foreign competition is limited, even with WTO commitments.
Interest rate liberalization is thus a potentially useful building block for China to consider for enhancing the role of market forces in credit allocation. The PBOC could remove the ceiling on the deposit rate and the floor on the lending rate and set one prime rate on which commercial banks can borrow for short-term liquidity. However, if interest rate constraints are removed, competition may encourage banks to assume risky practices and support excessive borrowing in the economy if there is no appropriate regulation and supervision (Feyzioglu et al., 2009). To combat an increase in such activities, the China Banking Regulatory Commission could implement vigilant regulatory and supervisory frameworks to keep up with the changing financial landscape and guard against a banking crisis from aggressive competition between banks. The PBOC could impose a proactive monetary policy to contain excessive lending and strengthen indirect monetary policy instruments. Such changes to the banking system might then ensure risks following interest rate deregulation are manageable.
Reforms in the Chinese banking system cannot be independent of other reforms outside banking, such as reforms in state-owned enterprises (SOEs) and the social welfare system. This is because if banks have the autonomy and incentives to make sound decisions, they would not subsidize poor-performing SOEs. This would thus suggest that SOEs must also become more efficient, perhaps by increasing their independence. Reforms in the banking system and SOEs would thus necessarily be accompanied by a social welfare system that provides unemployed persons with benefits, pensions and health insurance. If the social welfare system can mitigate possible social unrest resulting from reforms in the banking system and SOEs, the Chinese government would likely not need to intervene in the banking sector. Banks therefore would have the autonomy necessary to make decisions based on fundamental commercial principles. Chinese banks could take advantage of the inside information they possess to assess firms’ perspectives, issue loans to borrowing firms that have bright prospects and monitor borrowers effectively. In turn, the Chinese stock market would likely develop increased confidence in the efficiency of bank loans and respond positively to bank loan announcements.
