Skip to content

Chinese Banks At Risk, Part 1

July 27, 2010

A couple of weeks ago, I published a column outlining my doubts about the health of China’s banking system in the lead-up to the Agricultural Bank of China (AgBank)’s IPO.  Given the lackluster performance of that share launch — despite the considerable political capital Beijing mobilized behind making it a success — it seems I was hardly alone in my concerns.

Now some concrete data is starting to emerge regarding the potential size of the problems that may be lurking on China’s bank balance sheets — in particular, the losses that may be incurred from risky stimulus loans made to development entities (known by some as LGFVs, or Local Government Financial Vehicles) sponsored and supposedly guaranteed by provincial and local governments.  Earlier this year, China’s central government nullified those guarantees, noting that local authorities often lacked the financial wherewithall to stand behind them.  The China Banking Regulatory Commission (CBRC), which had previously been touting such loans as safer-than-safe, launched an internal study to get some handle on just how big a problem they have on their hands.

According to analysts at Bank of America-Merrill Lynch (BoA-ML), as well as a report published today by Bloomberg, inside sources at the CBRC say the study has been concluded and some findings reached.  According to them, at the end of this June, LGFV loans amounted to RMB 7.7 trillion (US$1.1 trillion), or 16.2% of all loans in the banking system.  (Rough estimates I’ve heard had placed LGFV loans at 40% of all of new loans made last year and this year, which may still be possible, but there is nothing in the new data to confirm it). 

Of these LGFV loans, 27% were found to have funded projects with sufficient cash flow to repay the loans.  50% must rely on “alternative sources” for loan repayment (which I take to mean either seizing collateral or invoking the public guarantee).  23% are categorized as “facing high credit risks (ie invalid qualification of borrowers, invalid guarantee by local governments, or loans misappropriated).”  According to the BoA-Merrill analysts, if those 23% “high risk” loans (totaling RMB 1.7 trillion) were downgraded to NPLs and assigned 20% provisions, “it would lead to 74bp of extra credit cost for the sector, or some 30% cut to [bank] earnings.”  Or as Andrew Peaple of the Wall Street Journal points out, the roughly US$230 billion in new bad debt would amount to more than three times the US$67 billion in NPLs currently recognized by China’s banks.

Andrew argues that while “pretty sobering,” the figures aren’t quite as bad as they sound, because “even a fourfold increase in the amount of bad loans in the system would keep their level, as a proportion of overall lending, far off of highs seen a decade ago, when Chinese banks needed a major government-led bailout.”  As I mentioned to him yesterday, though, don’t find much comfort in that line of thinking, for two reasons. 

First, while it’s tempting to focus exclusively on the 23% “bad loan” category, these are just the worst of the bunch, the ones that are clearly rotten.  I find it remarkable that only 27% of LGFV loans can be repaid from project cash flows.  That suggests that the vast majority of such loans were not financially viable (that doesn’t mean the projects weren’t worth doing, in a broader sense, since some of the projects may have had positive social and economic externalities — but the point remains, from a commercial lending point of view, these were not good investments).  

When we talk about 50% of the loans being payable by “alternative” means, we’re either talking about a primary default, where the local government has to deploy its own revenues or assets to step up to its guarantee, the seizure of often-illiquid collateral such as undeveloped land owned by the same local governments (which they depend on for as much as 1/3 of their future revenues), or some kind of rescheduling or renegotiation of the debt (quite likely, since all parties are state entities, and a political deal would likely need to be struck to avoid any of them suffering too much embarassment).  None of these are “good” outcomes, and may involve losses for the lender, the guarantor, or both.  So the US$230 billion figure is by no means a cap on the LGFV losses that might eventually be sustained, one way or another.

Second, loans to LGFV aren’t the only risky or inappropriate loans that were made during last year’s lending boom.  They’re just one lending category that most alarmed the Chinese government because (a) it had been assumed to be “safe” and (b) the identity of the borrowers posed a political as well as financial dilemma.   There are plenty of other categories — loans to property developers, to industries with significant overcapacity, to overextended home buyers, to businesses that diverted the funds to stock and real estate speculation — that may pose equal or greater risk of loss.  In short, the LGFV figures coming out of CBRC could just be the tip of the iceberg — the first wave, if you will, crashing against the shore.  If the pattern persists, and subsequent revelations indicate 20-25% NPL rates for all or most of last year’s runaway lending, the resulting hole in the banks’ balance sheets will be a hard to ignore or easily fix.

It’s worth point out, by the way, that I started warning about all of this way back in May of last year, in an op-ed I wrote for the Wall Street Journal, which you can read here

Tomorrow, in Part 2 of this post, I’m going to highlight a recent article in Knowledge @Wharton that focuses on the profound structural challenges facing Chinese banks as they try to salvage the situation and get back on the track to reform.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: