The objective of this study, as stated in section 1.2, is to first confirm Gareth Cottam’s (2011) results then adjust the valuations of Bank of China, the Bank of Communications, and Minsheng Bank for inflation and potential non-performing loans. The structure of the following section is as follows. First, basic valuations of American banks during the 2000 dot com bubble, the 2007 subprime mortgage crisis, and the Chinese banks will be compared to confirm the validity of Cottam’s results. Next, the Chinese bank valuations will be adjusted to reflect the effects of inflation. The American bank valuations will not be adjusted as unexpected inflation during the 2000 dot com bubble and 2007 subprime mortgage crisis were negligible. The Chinese bank valuations will then be adjusted to reflect the effects of possible NPLs on their historical and forecasted valuations. Finally, the Chinese banks’ basic valuations will be adjusted to show how a potential NPL crisis will affect an investor’s RROR.
The Chinese and American stock markets behave very differently from one another. It has been long suspected that the Chinese bank stocks experience significant insider trading (Yuan, Ziwu
& Fe, 2012), speculation, and heavy share turnover when compared to the American stock market due to the Chinese market’s youth (Rawski&Brandt, 2008) and immaturity. Therefore, certain basic assumptions about the stock market and business environment are made by this study in modeling the effects of NPLs on Chinese banks. First, it is assumed that the share prices of the Chinese Banks on the Hong Kong stock exchange (SEHK) accurately represent all
investor sentiments towards the Chinese banks. Chinese bank shares behave differently on the Shanghai and Hong Kong stock exchange because there are two types of shares available to
16
investors. A-shares, which are listed on the Shanghai Stock Exchange, are available only to local investors and react slowly to information. H-shares, which are listed on the Hong Kong stock exchange, are available to international institutional investors and react to information more efficiently. With no chance for arbitrage, there is a price difference between A-shares and H-shares. Therefore, the SEHK will be assumed to be representative of the Chinese bank share price behavior. Secondly, it is assumed that NPLs from 2008-2011 are uniformly distributed as a result from the $700 billion stimulus in the form of loans to local governments (Shih, 2010). This is done to simplify the forecasting of Chinese banks. Finally, potential NPL percentages
forecasted by professionals pertaining to Chinese banks will be taken as a percentage of all loans rather than just loans made to local financial vehicles and local governments. The effects of non-performing loans on the Chinese banks will be modeled as a reduction in interest revenue with none of the impaired assets written off their balance sheets. This scenario assumes that the Chinese government will intervene should the state owned banks report a NPL crisis similar to that in the late 1990’s and early 2000’s when AMCs bought NPLs from the big 4 at face value with either cash or bonds (Ma&Fung, 2002 and Chanovec, 2009). This scenario is likely because the Chinese government has shown in the past that it is willing to support the banks in the event of any NPL crisis. However, the privately owned banks such as Minsheng bank may not be as fortunate. Minsheng bank’s valuation will also be forecasted to show what would happen if it writes down 100% of its non-performing loans.
Additionally, the simplified Roger Montgomery Value.Able method assumes that the companies in question payout all their earnings as dividends. This ignores the growth factor in valuation and is a limitation of this research.
17
3.2 Sampling
A total of 3 Chinese banks and 6 American banks were analyzed for this study. The three Chinese banks consists of one of China’s big four banks1, Bank of China, a state owned but partially privatized bank, Bank of Communications, and one private bank, Minsheng bank. This broad spectrum of banks is meant to give a cross section of the banking industry in China.
American banks were chosen from two time periods that have similarities to the situation currently faced by the Chinese banks. The time periods are the 2007 subprime crisis and the 2000 dot.com bubble. Citibank, JP Morgan & Chase, and Bank of America were chosen during the 2007 subprime crisis due to the parallels drawn between the American subprime mortgage problem and the potential Chinese NPL problem (Whitten, 2011). Morgan Stanley, Capital One, and Bank of America were chosen during the dot.com bubble because of their similar core businesses of consumer banking (Capital One and Bank of America) and investment banking (Morgan Stanley) to Chinese Banks.
Each bank’s financial data will be taken from their annual reports. For American banks, data will be taken from a 5 year span of time starting from the year prior to their respective crisis. Next, each bank’s annual reports from the year prior to each crisis will be examined to determine its exposure to their respective crises. The financial information will be used to model expected performance before the respective bubble bursts and the expected return afterwards. The Chinese banks’ annual reports will be taken from 2007 to 2011 to model their current valuations and be examined to determine their exposures to NPLs. Then the data will be extrapolated to reflect their value five years from now.
1 According to the financial statistics of the PRC, the largest banks in China are currently the Agricultural Bank of China, the Bank of China, China Construction Bank, and the Industrial and Commercial Bank of China.
18
The prices used in valuating America banks during the 2007 subprime crisis will be taken from the 60 day moving average on the day of their 2007 Annual report release and a 60 day moving average centered around the lowest price experienced by the company’s stock. This is done because the subprime crisis lasted for over two years and each individual bank’s shares crashed on separate occasions. For the 2000 dot.com bubble, the prices of the banks will be taken from the 60 day moving average before and after March 10th, 2000. This is done contrary to the method mentioned above because March 10th, 2000 is widely accepted as the date of the dot.com crash. Finally, share prices used in the valuation of the Chinese banks will be taken from the 60 day moving average before the release of their 2011 annual reports.
Inflation data will be taken from the year end YOY data as it best measures the actual inflation experienced by the economy during that year.
3.3 Basic Valuation
The primary valuation metric used in this study is return on equity divided by a required rate of return. Required rates of return can be calculated through the weighted average cost of capital, capital asset pricing model, or as a simple premium added to an index or risk free asset
(Montgomery, 2010) and differs with each individual and rises with a security’s risk and inflation. The required rate of return used in this study is 10% for simplicity. Share prices are taken as a 60 day moving average at the end of the year. Cross-section book values per share are taken from quarterly reports relative to time periods in question. Return on equity divided by the required rate of return results in a book value multiplier that is combined with the book value per share of the firm to obtain the fair price per share for the firm.
19
Equation 3.3.1
The share price of the firm is then divided by the fair price per share to obtain the premium ratio.
Equation 3.3.2
The premium ratio sheds light on whether a stock is overpriced and by how much when compared to its peers. A premium ratio greater than one indicates the shares are overvalued while a ratio of less than one indicates the shares are trading at a discount. When comparing two premium ratios, a higher premium ratio indicates the shares are trading at a premium relative to the shares in question and vice versa. In addition, a disproportionate decrease in ROE will result in a higher premium ratio due to ROE being in the numerator of the fair price per share of firm equation. Conversely, a disproportionate decrease in the price per share of firm will result in a decrease in the premium ratio due to it being in the numerator of the premium ratio equation.
Results in line with H1 will result in higher premium ratios before a bubble burst than after. If the bank was previously in a bubble, the price will decline proportionally more than the fair price due to overselling, resulting in a lower premium ratio. The resulting fair price ratios of Citibank, JP-Morgan and Chase, and Bank of America before and after their respective crises will be compared to the fair price ratio of the Chinese banks in question. This metric is useful in
20
determining the existence and the scale of a bubble since a bubble results in inflated prices compared to the actual value leading to an increase in the fair price ratio during a bubble and decrease in the following collapse.
3.4 Adjusting for Inflation
The value of unexpected inflation will be determined by dividing the actual inflation rate by the forecasted inflation rate. As discussed in earlier chapters, expected inflation doesn’t affect an investor’s return as they have accounted for it in their required rate of return. However, if the inflation rate is higher than expected, then the actual rate of return of the investor is in fact lower.
Valuations of Chinese banks will be adjusted for inflation by dividing a bank’s ROE with the inflation rate.
Equation 3.4.1
Equation 3.4.2
Inflation will also erode the earnings of banks and the future value of its reserves. However, this study will focus on inflation’s effects on return on equity as ROE is the basis for Roger
Montgomery’s valuation methods.
21
Inflation’s effect on American Bank valuation will be ignored as unexpected inflation in the United States during the two time periods tested was negligible.
3.5 Modeling for Non-Performing Loans
3.5.1 NPL Adjustment
The modeling of the effect of NPLs on the financials of the Chinese banks was done with forecasted NPL levels of 5%, 10%, 15%, and 25%. Since it is assumed that forecasted NPL percentages is taken as a portion of a bank’s entire loan portfolio rather than a percentage of a bank’s local government owned debt, the effects on the income statement of the banks were estimated by reducing net income by the net interest income affected by NPLs. For example, if bank A has $100 net income from $150 net interest income, a level of 10% NPL means net income would be reduced by $15 ($150 x 10%). NPLs were modeled to cover two extremes. The first assumption was that the NPLs would only affect revenue. As the Central Bank of China (CBC) had previously bailed the Chinese financial system out of a previous NPL crisis in the early 2000’s by selling toxic assets to government owned asset management companies, it is feasible that they may do so should another NPL crisis occur. The second model will examine Minsheng bank’s valuations if its NPLs were written off as bad assets. This would occur if the Central Bank of China (CBC) refuses to bail out the private banking sector of a NPL crisis.
In order to model the effects of NPL on return on equity, the following formula was used.
22
Equation 3.5.1.1
Multiplying the level of NPL with each bank’s exposed revenue gives an approximation of what effects different levels of NPL will have on each bank’s revenues. Since NPLs still incur normal expenses, the loss in revenue translates directly to a loss in net income. This new net income is then divided by equity to find the NPL adjusted return on equity.
If the CBC declines to bail out the private banking sector, then Minsheng Bank would have to write-off its impaired assets due to non-performing loans. According to IFRS, impaired assets are off set on the balance sheet by the allowance for impaired assets account. If the actual level of impaired assets exceeds this allowance, then the allowance for impaired assets account is debited and net income is decreased (Cromwell, 2012). To determine the ROE for year effected by a 100% write-off of all non-performing assets, the following formula is used.
Equation 3.5.1.2
The above formula is the same as equation 3.4.1 but with the amount of written-off assets subtracted from the net income. As mentioned in chapter 2, loans given in 2009 aren’t expected to become non-performing loans for three years (Shih, 2010) so for the purpose of this study, it is assumed that all NPLs will occur three years after the date given.
23
3.5.2 Forecasting
Forecasting for 2012 was done by taking the yoy growth of interest income, net income, and book value reported in the 2012 interim reports of the Bank of China, Bank of Communications, and Minsheng bank and extrapolating for year-end results in 2012. The values were then used to calculate the effects of potential levels of non-performing loans on the valuations of the Chinese banks.
3.6 NPL Adjusted Required Rate of Return
The purpose of this section is to determine new required rates of return for basic valuation that would result in identical premium ratios for the various levels of anticipated non-performing loans. Each banks’ return on equity when faced with different levels of non-performing loans must be examined in order to determine what RROR would give the basic valuations similar premium ratios to NPL-adjusted valuations. When adjusting valuations for NPLs, each banks’
ROE was reduced to reflect the loss of revenue each banks will face. The percent change faced by each return on equity must be first determined by the following equation.
Equation 3.6.1
Since return on equity is in the numerator and the required rate of return in the denominator of the premium ratio equation (equations 3.3.1.1&2), the inverse of the percent change of ROE must be the percent change in RROR. The following equation shows this relationship.
24
Equation 3.6.2
25