“It ain’t what you don’t know that gets you in trouble, it’s what you know for sure that just ain’t so.”

– Mark Twain

Last year, I debunked a popular measure of trade misinvoicing as the culprit for China’s capital outflows. Today, let’s scrutinize two other misconceptions bouncing around the China commentator echo chamber.

But first, a summary of the argument from last year: ‘Trade misinvoicing’ is when trading companies undercharge for exports or overpay for imports so that intermediaries can stash the difference in an offshore bank account. The consensus narrative claims that we can observe this directly via discrepancies in bilateral trade data. So, I showed how to generalize those discrepancies to a time series indicator; and, that this indicator offers zero explanatory power for China’s drawdown of FX reserves.

Today, let’s audit another proposed mechanism of capital flight involving artefacts of China’s trade data. I first saw this proposal in a blog post by Christopher Balding, economics professor at Peking University and owner of a delightful twitter handle. Since, many others have reproduced and repacked a similar analysis. Here is the basic idea:

The differences in key data surrounding trade data is illustrative… Whereas Chinese Customs reports $1.68 trillion and SAFE report $1.57 in goods imports into China, banks report paying $2.55 trillion for imports.  In other words, funds paid for imported goods and services was $870-980 billion or 52-62% higher than official Customs and SAFE trade data.  This level of discrepancy is extreme in both absolute and relative terms and cannot simply be called a rounding error but is nothing less than systemic fraud.

In other words, the difference between imports and bank payments data directly exposes the mechanism by which Chinese nationals ferret money out of the country.

Dr. Balding’s analysis is seductive for two reasons: One, we get to cry that China is imploding. Finally! Two, we’re supplied with further evidence implicating the statistical agencies in China Watchers v. Chinese Economic Data. Both get clicks.

However, a rigorous look at the data rejects this hypothesis. We can construct an indicator of this “systemic fraud” by subtracting the value of imports reported by Customs from the value of bank payments for imports (which is reported by SAFE). Let’s call it, “Bank Overpay.” Then, we compare bank overpay to changes in PBOC FX reserves. A final note, while Dr. Balding appears to be citing annual data from 2015, I use monthly data.

China began hemorrhaging foreign exchange reserves in late 2014. However, banks began ‘overpaying’ for imports nearly three years earlier. The eye-test alone debunks this mechanism as a direct monitor for capital flight. Furthermore, a battery of statistical tests show that a model of China’s FX reserves gains zero explanatory power by including ‘bank overpay’ as a predictor variable.

I can’t explain the discrepancy between Customs and SAFE import data, that could be topic for future digging – perhaps even a trip down to the docks. But, this particular statistical artefact and China’s capital flight are unrelated.

So, what does explain China’s recent bout of capital outflows? I still find this BIS report most persuasive. Robert McCauley and Chang Shu offer two drivers: corporates paying down FX debt and foreigners reducing their stock of RMB deposits. A simple model taking those two as independent variables explains around 70% of the variance in the PBOC’s FX reserves.

Now, let’s move on to China’s capital controls. The echo chamber commentators are tripping over each other to tell us that recent capital controls have staunched FX reserve depletion, and we should put worries of China’s capital flight behind us (read here ,here, and here).

What does the data say?

We can use the concept of covered interest parity to quantitatively observe the efficacy of capital controls. CIP assumes that in the absence of country risk, obscene transaction costs, or (most importantly, here) capital controls, an investor should not be able to capture a riskless profit by arbitraging the interest rate differential between two countries.

The methodology: we calculate an ‘FX implied rate’ for SHIBOR, which is what an investor should earn in China based on both USD funding costs and the expectation of future exchange rates. A divergence between the implied and real rates confirms the presence of capital controls limiting the free movement of capital. On the other hand, convergence would show a relative freedom of capital mobility.

A pile of studies (see here, and here) used this methodology to observe the effect of Japan’s capital controls last century finding that strong capital controls contributed to significant deviations from covered interest parity. Applying the same methodology in reverse, we see that China’s capital controls have recently actually become quite leaky.

The left-hand side plots the one year SHIBOR and FX implied rates, the right-hand side the difference between the two. The convergence of the two rates implies that China’s capital account has become more porous over the last 2 and half years, not less. And, recent capital controls certainly haven’t pinched complete capital mobility.

So what does this all mean?

Although PBOC FX reserves have stabilized for the moment, and the government has simultaneously implemented strict capital controls, I don’t believe the latter to be the cause of the former. Not only that, we still don’t have an accurate quantitative measure to observe in real time by which channel money is leaving the country.

Given China’s massive money supply, the RMB still faces tremendous downside risk should China get hit with another economic shock. Regarding capital flight, I believe that China is not yet out of the woods.