國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
- 24 -
(Sources: this paper data)
5. Empirical results
5.1 Univariate analysis
We compare the mean and medians of several key variables for the
international bank and domestic bank subsamples. Table 5.1.1 shows that the mean Z-score for international banks, which is 14.56, is lower than domestic banks’ value (24.55). This result supports the idea that international banks may take higher level of risk than purely domestic ones. Moreover, this conclusion holds even when we use different alternatives to measure risks, such as Sharpe Ratio.
On the other hand, according to Table 5.1.2, we try to reduce the concerns that our results are affected by specific bank size, so we make comparisons between the means and medians of Z-score for international banks and domestic banks by different bank sizes.Overall, our result indicates that international banks are riskier than purely domestic banks, no matter what sizes they possess.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
- 25 -
5.2 Regression analysis
To measure the effects of internationalization on the risk of Chinese banks, the following simple regression framework proposed by Berger (2013) are
implemented:
𝑅𝑖𝑠𝑘𝑖,𝑡−𝑘,𝑡= 𝛼 + 𝛽1∗ 𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−𝑘+ 𝛽2∗ 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑖,𝑡−𝑘+ 𝜀𝑖,𝑡−𝑘+1,𝑡
Risk is bank risk, which is measured by Z-score and other proxies that are stated in the Section4.1.1. Internationalization is a dummy variable signals whether a bank has international operations or not. Controls is the category that includes bank’s characteristics which are described in the Section 4.1.3.ε is the usual error term.
Considering the model may have lagged independent variables, so we measured our risk variables over k years from t-k+1 to t and independent variables are measured in the year t-k to attenuate the lagged effect. We use k=3 in our analysis.
The following Table 5.2 is our regression result. We regress Z-score on international dummy and bank’s control variables.
Table 5.2: Internationalization and Bank Risk: Main Regression Analysis This table reports regression estimates of the relation between the internationalization of China commercial banks and their risk using Z-score (3 years) as the dependent variable. The main internationalization measure is the International Dummy (Inter). We report in the table our main model, OLS with time fixed effects and clustering by bank (main model) for the full sample. Table 4.1 shows definitions for all variables. Robust t-statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
‧
We can clearly observe that international activities will lead banks in China with higher risks. The coefficient on international dummy is negative and statistically significant at the 5% level. From the “Overseas Strategy of Chinese Banking Industry” research conducted by the CBRC, there are several possible reasons for this result.First, more internationalization will lead banks to be exposed more of macroeconomic volatility. After global financial crisis, China has played an increasingly important role in funding other economies. However, European debt crisis lasted for a period of time and it was just solved a few years ago; furthermore, geopolitical concern continues in the region such as Russia and Middle East that many China corporations operate in and this will let Chinese banks have lower quality of offshore loans making them more risky. Second, diversified as well as strict supervision law in different markets make it harder for Chinese banks to adjust, costing more to adopt a new business strategy successfully. Third, during the merger and acquisition process, which Chinese banks usually adopted, it’s difficult for Chinese banks to evaluate merged target’s financial stability due to information asymmetry and overseas M&A prefer cash transactions, which may put Chinese
‧
environment than those who remain domestic operation.Moreover, our regression result also indicates that smaller size as well as lower management quality will leadbanks into a riskier status. Third, we observe that different structure of Chinese banks will result in different level of risk. From our regression, we learn that Bigfive banks, or state-owned banks, have the highest risk level while city commercial banks has the lowest risk. In Table 5.3, Bigfive banks have the characteristics such as high internationalization, large size and better quality of management. Comparing to Bigfive banks, city commercial banks possess lower level of internationalization, smaller size as well as worse quality of management. Although Bigfive banks equip larger size and better quality of management than joint stock banks and city commercial banks, Bigfive banks still suffer from higher level of risk due to higher internationalization comparing to others. To sum up, we confirm that internationalization is an important factor that drives Chinese banks into a riskier status even if they possess large bank size and better quality of management, which is not enough to mitigate the risks.
Table 5.3: Comparison between different bank structure (Bigfive, Joint Stock, City Commercial banks)
Table 5.3 reports differences in Z-score by bank structure. We report both difference in means and medians between the characteristics of Big5, joint stock and city commercial banks. Table 4.1 shows definitions for all variables.
Mean Median Mean Median Mean Median
Z-score 12.24 8 14.66 9.46 42.42 19.59
International Dummy 1 1 0.61 1 0.08 1
Size 6.32 6.38 5.36 5.64 4.29 4.2
Quality of Management 34.05 33.17 34.51 34.15 37.11 35.53 Observations
Big5 Joint Stock City Commercial
25 28 274
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
- 28 -
5.3 Region analysis
From our regression analysis above, we conclude that internationalization will increase the risk of Chinese banks. A further question is raised, will different expanding regions of international banks result in different level of risk? According to each bank’s official statements, we observe that each bank may have different focus market. For example, Industrial and Commercial Bank of China (ICBC), Bank of China and most joint stock commercial banks have international branches in Asia(ex. Hong Kong, Macau, Singapore…), while China Construction Bank has its key locationin the South Africa and Agricultural Bank of China puts emphasis on Europe (ex. London, Germany…) and Bank of Communication focuses on the North America (ex. U.S). Due to opacity of bank data on absolute statistics at each location in each year, we use the bank’s key operating location as a proxy representing the only overseas region of that bank and we take the average of Z-scores in each year of those banks whose key location remainsthe same. Results are demonstrated in Figure 5.1.
Figure 5.1: Different risk level in respective region between 2009-2013
Figure 5.1 shows the potential risk in each region, respectively. Given that Z-score is calculated using data over the previous 12 quarters, the sample period depicted ranges from 2009 to 2013.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
- 29 -
According to the figure above, region U.S has the highest Z-score, or the lowest risk over 2009 to 2013. Second, Europe region possess the highest risk if Chinese banks choose it as the overseas expansion location during 2009 to 2012. We consider this is reasonable since Europe has suffered from the Europe Debt Crisis from 2009； the overall economic situation as well as the individual country all went through a prolonged deteriorating market condition. However, Europe’s crisis seems to calm down recently and we can observe such a trend from the figure above that the Z-score is moving upward. Third, Asia seems to be the region possesses the highest risk level. Asia is an emerging market comparing with those developed countries, the region’s economic situation is influenced not only by developed countries but also by idiosyncratic events. Although Asia has maintained significant growth and attractsChinese banks to expand offshore business, it also create more risk for banks due to possible economic volatilities. To sum up, the U.S is the least risky region for Chinese banks to develop overseas business. Asia is the riskiest location and South Africa as well as Europe have medium risk for expanding.
Finally, regardless the pessimistic inference from the above discussion, Figure 5.1still indicates that through the experience of overseas expansion and a better
U.S.A
Europe
Asia
South Africa
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
- 30 -
ability to cope with local economic volatility, the regional risk is on a downward trend, which may be a good sign for Chinese banks to expand internationally in the future.