How Important is Geography Behind the Success of the New Tigers?

The Rationale Behind Looking at City Commercial Banks (CCBs)

We need to understand whether the New Tigers' better performance is caused by better corporate governance only or whether and to what extent the New Tigers are better simply because they do business in the most developed area of PRC, thus by their location avoiding the economic difficulties the SOBs have to deal with. To shed light on this issue, we draw on the results of a field survey commissioned by the Asian Development Bank Institute to check how far, within (a significant segment of) the New Tigers, bank performance differs depending on the economic development of the geographical area where banks do their business.⁸ This is exactly the rationale behind looking at City Commercial Banks (CCBs), one of the most vibrant components of the New Tigers, which includes banks located across most of the country.

Within our Bankscope sample, CCBs's market share (in terms of total assets) almost doubled, from 1.9 to 3.5% between 1997 and 2003, while their weight within the New Tigers fluctuated around 15% (Figure 5 [ PDF 94.5KB | 1 page ]). However, we should point out that Bankscope undersamples CCBs, including only 16 out of 112 of them. Considering all of the 112 CCBs, their total market share is about 5%.

CCBs came about after 1995 when the People's Bank of China put in order NPLendangered urban credit cooperatives. Urban credit cooperatives were salvaged with the injection of public funds, but at the same time, they were ordered to merge consolidating into the newly formed CCBs, established as joint-stock companies. CCBs inherited from urban credit cooperatives all NPLs formed during the ¡°nonstandard operating period¡± of 1985- 1995. (Girardin and Ping, 1997) At the end of 2003, 39 out of 112 CCBs had an NPL ratio above 20%, some of them even above 70%. CCBs' shareholders include urban enterprises, citizens and local governments (individuals are not allowed to become new shareholders). At present, city commercial banks are distributed in 112 central cities (district-level or above) of PRC ¨C one city, one city commercial bank without exception. Though they almost cover the whole of the country their distribution is uneven. Generally speaking, there are more CCBs in Eastern provinces (for example there are 11 CCBs in Jiangsu Province) than in Western provinces (in Gansu, Qinghai, Xinjiang and Ningxia, CCBs exist only in the capital cities). Since their foundations, the financial authority has required that all city commercial banks offer financial services only within the cities' own administrative districts.

By end June 2004, the 112 city commercial banks had 5,154 branches, 107,000 employees, and a 5-grade NPL ratio of 14.1%. Among the various categories of financial institutions, city commercial banks rank second in terms of business development, close to joint-stock commercial banks. CCBs focus on three main business lines; providing indirect financial services to SMEs; offering financial services for city residents; financing local government public works.

Field Survey on 20 CCBs from Three Provinces

By focusing on 20 CCBs located in three provinces of PRC featuring diverse levels of economic development, we keep corporate governance (relatively) constant and can thus ascribe any significant difference in performance across the provinces to their relative underlying prosperity.

We selected the three provinces to include; one with a level of prosperity just about the national average, this is Hubei province; one ranked amongst the most developed, this is Zhejiang province (with a GDP per capita about double the national average) average; one less affluent, this is Sichuan province (with a GDP per capita about two-thirds of the national average; Table 1 [ PDF 112KB | 1 page ]).

As shown in the Table 1 [ PDF 112KB | 1 page ]not only GDP per capita, but also growth is fastest in Zhejiang, while Sichuan, though less developed, is enjoying faster growth than Hubei. Thus, while Zhejiang stands out in both the level and the dynamics of GDP, Hubei is ranked before Sichuan if we look at GDP per capita, but the order is reversed if we take growth into account. The 20 interviewed CCBs are distributed as follows; 7 in Zhejiang, 5 Hubei, and 8 in Sichuan. For these CCBs the survey collected information on their asset-liability/profit-loss accounts over 2000-2003 as well as on their business features and several ownership and corporate governance aspects.

Overall, the 20 interviewed CCBs have 13,400 employees over 1,160 branches, with 12 employees per branch on average. The largest (smallest) CCB is that of Wuhan in Hubei province (that of Zigong in Sichuan province) with almost 1,800 employees (with just 210 employees). CCBs business is largely concentrated in the city of establishment (on average 98.2% of the loans are granted there).

Different Patterns of Performance in More vs. Less Developed Provinces

Over the 4 years 2000-03, total assets of the CCBs expanded by 1.58 times in Hubei, by 2.15 times in Sichuan, and by 2.75 times in Zhejiang. Such ranking of the expansion of the banking business across the three provinces seems consistent with GDP growth.

Both size and performance of CCBs improve on average when we move from the less affluent Sichuan and Hubei to the most prosperous Zhejiang. The average size of Zhejiang CCBs is three times as large as that in the other two provinces (Figure 6 [ PDF 77.4KB | 1 page ]). Assets per employee, one of the basic indicators of productive efficiency, is twice as large in Zhejiang as in Hubei and Sichuan. The average ROA is close to 1.5%, which is five times as large as in the other two provinces. In addition, the NPL ratio is just below 5% in Zhejiang, which is 20% of that in Hubei and 25% of that in Sichuan.

Zhejinag CCBs stand out also in terms of ROE as well as in terms of their ability to generate profits out of net interest income. ROE is three times as large for them as in Sichuan, while Hubei CCBs are barely able to generate positive returns (Figure 7 [ PDF 79KB | 1 page ]). Profits/net interest income is close to 60% in Zhejiang, twice as large that in Hubei and three times that in Sichuan.

Zhejiang CCBs achieve higher profit efficiency in spite of larger loan-loss provisioning, where loan-loss provisions are very low in Sichuan and intermediate in Hubei. All in all, loan-loss reserves are probably insufficient; in 2003 the loan-loss reserves of the surveyed CCBs were merely 1.25% of total loans, well below their average 14% NPL ratio. In addition, CCBs have high loan concentration ratios; some CCBs overly pursue prime customers; insiders and big shareholders cause problems of loans to their related enterprises; and risk management systems are still not fully developed.

Before moving on to comment the additional information we gather from the survey, it is useful to check whether the different in performance across the three provinces are statistically significant. We test this running some basic econometric specifications where the dependent variable is, in turn, one of the standard performance measures: ROA, ROE, and the NPL ratio. In the light of the few observations we have, we only consider as regressors bank size (controlling for possible economies of scale) and two province dummies (identifying any specific effect for Hubei and Zhejiang, respectively the provinces with the intermediate and the highest level of development).

The results of these regressions are reported in Table 2 [ PDF 66.8KB | 2 pages ]. While larger bank size is generally associated with better performance, the hypothesis that performance varies among the provinces cannot be rejected. Specifically, both ROA and ROE are significantly lower for CCBs located in Hubei, the province which displays the lowest GDP growth. On the other hand, these performance measures do not appear to differ significantly between Sichuan and Zhejiang. Finally, the NPL ratio is significantly smaller in Zhejiang, whilst no significant differences emerge between Hubei and Sichuan.

Table 2. Regression on the impact of government shareholding on CCBs' performance [ PDF 66.8KB | 2 pages ]

Further Evidence from the Field Survey

CCBs are basically controlled by local government and their internal governance structures still need considerable improvement; on average, direct share-holding by the local government is 24.2%, but adding indirect shareholding brings to some two-thirds the share of total equity controlled by state capital; domestic private shareholders stand at 23.7%. The situation differs across the three provinces. While direct shareholding by local government is not very different, the picture changes drastically once we add shareholding by SOEs to that by local government. On average this sum reaches above 60% in Hubei, above 40% in Sichuan, whilst it is below 30% in Zhejiang (Figure 8 [ PDF 93.9KB | 1 page ]).

We also learn from the survey that the local government has the final influence in selecting the chairman and president of CCBs, who are in charge of key decisions. It appears that related party loans to finance local government initiatives (such as for infrastructure) may be a problem (also this could hide quasi-fiscal liabilities): 4 of the 20 CCBs declare that a share between 10 and 20% of their loans is in response to the intervention of the local government (the level escalates to between 30 and 40% for one of the CCBs, while 3 CCBs do not answer this question), reportedly augmenting NPLs.

The survey offers novel information on the ownership structure and on some key corporate governance aspects. On average, the largest shareholding belongs to the local government plus SOEs (43.7%), followed by other Chinese institutional investors (29.7%), by private non-financial firms plus households (24.8%), while only in one CCB are there collective shares and none of our CCBs has participation by foreign capital (Figure 9A [ PDF 90.9KB | 1 page ]). The situation however varies across the CCBs and among the provinces. As expected based on its relative affluence, in Zhejiang on average the shareholding by the local government plus SOEs is lowest (26.7%) and that of private non-financial firms plus households is largest (33.0%), while the situation is reversed in Sichuan and, even more so, in Hubei.

The influence of the various parties on bank conduct may, however, differ with respect to their nominal weight as shareholders. This is revealed by examining who appoints the bank directors. As reported in Figure 9B [ PDF 90.1KB | 1 page ], the local government plus SOEs appoint 56.5% of the directors, well above their shareholding, while private non-financial firms plus households appoint only 11.9% of the directors, well below their shareholding. Across the three provinces, private shareholders' ability to appoint directors relative to their shareholding is minimal in Hubei (no director vis-a-vis 9.7% of private shareholding), intermediate in Sichuan (10.3% of the directors vis-a-vis 21.3% of private shareholding), and the maximum in Zhejiang (20.7% of the directors vis-¨¤-vis 33.0% of private shareholding). This suggests that the ability of private shareholders to affect bank conduct may be less than indicated by their nominal shareholding. We will return to this later.

A further point worth stressing hinges on the potentially dangerous link between large shareholding by the local government and SOEs, whereby CCBs might be captured by political influence. The survey offers some evidence on this. First from individual bank data, there is a strong positive correlation between the weight of the local government together with SOEs as shareholders and the share of CCB loans going to the local government plus SOEs (Figure 10A [ PDF 89.5KB | 1 page ]).

This is consistent with the hypothesis that CCBs may be captured the local government plus SOEs: When these are large shareholders, it may be very difficult for the banks to deny them credit. Second from individual bank data there is also a positive correlation, , between the share of loans to the local government plus SOEs and NPL ratios(see Figure 10B [ PDF 80.2KB | 1 page ]) although the correlation is very much weaker than in the first relation.

To take this analysis a little further, we run some econometric specifications on this issue. Given the limited number of observations, this analysis can only be exploratory. The results are reported in Table 3 [ PDF 68.3KB | 2 pages ]. Clearly, bank size is an important determinant of performance if we consider ROA (Panel 3A) and ROE (Panel 3B), possibly stemming from the presence of economies of scale. Neither the province of operation nor the presence of private shareholders significantly affect performance (all panels). The weight of the local government together with SOEs as shareholders impacts neither ROA (Panel 3A) nor ROE (Panel 3B) directly.

However, we can identify two channels through which government ownership negatively affects performance indirectly. First, the NPL ratio proves a significant determinant of both ROA (Panel 3A) and ROE (Panel 3B). In turn, as the weight of the local government together with SOEs as shareholders of the CCB increases, the NPL ratio increases significantly also (Panel 3C). This happens even though the share of loans going to the local government and SOEs ¨Cwhich is positively related to the weight of the local government plus SOEs as shareholders¨C does not significantly affect the NPL ratio (Panel 3C). The second channel through which government ownership worsens performance is via the allocation of loans at the banks. We detect that, as expected, a larger share of loans going to manufacturing improves CCB performance, at least in terms of ROE (Panel 3B). In this respect, we identify a significant negative link between the share of loans going to manufacturing and that going to the local government plus SOEs (Panel 3D), and, at the same time, the latter share increases with the weight of government ownership (Panel 3D). Accordingly, it seems that by shifting the distribution of loans towards the government itself and SOEs, government ownership crowds out manufacturing loans and, through this, weakens CCB performance.

It seems that the extent of local government ownership has a direct impact on the NPL ratio and on the share of manufacturing loans, and through these it worsens performance indirectly. It is also possible that, beside these two channels, government ownership induces general inefficiencies in the CCB business conduct, but we have no direct evidence on this.

Finally, we consider a non-standard measure of performance, namely the share of loans targeted to small enterprises (defined in PRC as firms with fewer than 100 employees). A high value of this share seems advantageous in view of the desirability to promote the transition by supporting new business formation. We regress this share against the usual explanatory variables (Panel 3.E). As expected ¨Cin light of the positive correlation between bank and firm size found in the literature¨C we show that larger CCBs are less prone to support small business. In addition, we find that the extent of ownership by the local government plus SOEs has a negative effect on such a share. Thus, even within the CCBs we detect indications consistent with the findings of Wei and Wang (1997) that government ownership tends to disfavor (smaller) private business.

Table 3. Regression on the impact of government shareholding on CCBs' performance [ PDF 68.3KB | 2 pages ]

As dependent variable we consider alternatively ROA, ROE, NPL, LOAGOV (the share of CCB loans going to the local government + SOEs), and LOASE (the share of CCB loans targeted to small enterprises). Among the explanatory variables we consider (not all at the same time for reasons of few observations): the logarithm of the shareholding of the CCB by the local government + SOEs (LGOVTSH); the total assets of the CCB (SIZE); a variable controlling for the province where the CCB operates (PROV; to be more parsimonious we used here a graduated dummy taking a value of 1 for Sichuan, 2 for Hubei and 3 for Zhejiang), a dummy variable identifying CCBs with private shareholding above 50% and/or the fact that some of the directors are appointed by private shareholders (PRIVATE) and, where applicable, LOAGOV and the share of CCB loans going to manufacturing (LOAMAN). Reported t-statistics are obtained via OLS and are Huber-White heteroskedastic consistent. The superscripts ***, ** and * indicate that the coefficient is different from zero respectively at the 1%, 5% and 10% confidence level.

This evidence is consistent with the results of the above mentioned PBOC survey on the causes of NPLs at SOBs (Zhou, 2004a).

The hypothesis that captive CCBs are potentially endangered by political interference may also explain why, performance is relatively poor in Hubei province with respect to what one might expect from its level of development indicated by a GDP per capita on par with the national average.. As reported in Figure 10A [ PDF 89.5KB | 1 page ], Hubei CCBs stand out at the top in terms of both shareholding by local government plus SOEs and share of loans allocated to them. Unsurprisingly, Hubei CCBs are at the top also in terms of the NPL ratio.

Thus, it seems that, keeping corporate governance (relatively) constant, geography is a strong determinant of performance, although it should be noted that in our regression analysis provincial dummies are sometimes not significant. Accordingly, the view that the New Tigers are the solution to bring better banking to PRC seems too simplistic. To be sure, we have remarked that corporate governance differs even within CCBs, where those CCBs more exposed to political influence are worse performers. It is possible that higher development induces better governance by decreasing the role of local government and SOEs. However this implies that corporate governance is to some extent endogenous and, in any event, what works in PRC's affluent Eastern Belt may not work in less developed areas. This casts doubt on the possibility that the New Tigers may offer an effective national solution to deal with the country's banking problem and, as a result, forces policy makers to turn again to consider improvements to the situation of the NPL-endangered SOBs.

Download this Discussion Paper [ PDF 226.3KB| 25 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.