The main dispute here is the proportion of bad loans in the newest batch of lending between 02 and 04. The Chinese side says that fiscal discipline has improved and that the newest loans only went to financially qualified companies. E & Y assume about a 40% default rate on new loans based on a SRS (simple random sample) of the companies that the big 4 loaned to. However, the big difference here is that E & Y assumed historical trends among different loan size classes (basically that loans with principal between a and b, interest rate between x and y percent, and recipient market cap between m and n (I think those were the three principal explanatory variables in their model) would have an NPL rate between so and so percentage.) Summing the total NPL amounts, E&Y then calculated their figure of ~1 trillion dollars.

The NBS disagrees, obviously, by stating that it uses the historical NPL ratio of the regional bank branches and also adds in whether the company was public or private (of the new batch of loans, a big chunk went to private-sector construction companies, another big chunk went to public steel companies, but these "chunks" were distributed by different regional banks) to conduct a block survey. That's where the lower number comes in, b/c NE China branches of ICBC have historically had a low NPL rate (they decided against loaning like crazy to the failing SOEs, and because the assets of the SOEs were fixed and the ICBC had good guanxi they were able to recoup most of their assets) and the urban banks never really had a high NPL rate (because most of those Shanghai and Beijing banks started after the real estate crash of 1993). Outside observers state the data generated by this is flawed, which I agree, but I also disagree with E&Y's estimate because it's way too simplistic in its model, generating considerable probability problems which it then attempts to "rub away" with complex stochastic calculus (probably where the PBOC felt the report was "barely understandable") more appropriate for derivative volatility calculations than bank loans (because of pull-to-par).

I could give you my estimate, but then I'd get fired heheheh.

The People's Bank of China, the country's central bank, quickly attacked the research as “ridiculous and barely understandable�.

They actually admit that they don't understand economic research? :shock:

I'm afraid it is impossible to know what the actual numbers are, but you can be quite sure that the government numbers are not to be trusted. On the other hand, I have been waiting for a bank collapse for many years, and it still haven't happened.

Just want to get people's thoughts on this.