Reuters has done a nice wrap up of what we know -- and more importantly what we don’t know -- about how China manages its reserves. They correctly note that the TIC data misses so much that it, alone, fails to tell us much about the actual composition of China’s reserves.
[according to] Thomas Stolper, currency analyst at Goldman Sachs in London.
"The simple TIC-based calculation would almost certainly lead to an overestimation of FX reserves diversification."
A more realistic "guesstimate," many reckon, is that China holds about 25 percent of its hoard in non-dollar currencies.
But why is there such confusion?
... the TIC data is most flawed and even the Treasury is keen to emphasize the caveats.
Countries may hold some money in dollar deposit accounts outside the United States, which does not get caught in TIC data designed to monitor cross-border shifts. UBS analysts say $20-30 billion may be held this way.
Economists also reckon China is likely to be buying at least some U.S. securities via third party brokers and custodians based, possibly, in British or Caribbean financial centers. This means these holdings would not show up as China’s in TIC data, perhaps instead inflating the reported holdings ascribed to the countries of these third parties,
Another point that may tend to inflate estimates of diversification is difficulty in comparing data over time.
The TIC series is rooted in annual June surveys by the Treasury and Fed, the results of which are only released 10 months later. Monthly updates in between are compiled by adding net transactions -- numbers that often fail to distinguish factors such as maturing debt and repurchase agreements.
Taken together, it is just possible there has been no substantial shift at all.
"There’s no real complete way of knowing," said Daniel Katzive, senior currency strategist at UBS in Stamford, Connecticut. "The U.S. data is simply incomplete."
What do we know?
-- The Reuters story reports that China’s reserves now stand at $691 billion, up $32 billion in April and May.
-- The actual increase in April and May was far larger than $32 billion. China used $12.5 billion in reserves to recapitalize a state bank. And the value of China’s euro-denominated reserves fell in May. If 25% of China’s then $660 billion in (end March) reserves were in euros (a high estimate), it held $165 in euros. Do the math, those euros are now worth a lot less now than they were then. I suspect China’s underlying reserve increase in the past two months was closer to $50b than $30b.
-- If you add the reserves that have been transferred (in theory) to the balance sheet of the state banks back in, China’s total reserves stand at around $748.5 billion right now. China is rapidly gaining on Japan.
-- Only a small fraction of China’s reserve increase shows up in the TIC data. In 2004, adjusted for valuation, I estimate China’s reserves increased by $198b. China’s recorded holdings of US assets increased by only $67.4 billion (that includes what the TIC calls "foreign bonds" -- so only 34% of the increase showed up in TIC data. In the first quarter of 2005, only 27% of China’s $54 billion reserve increase (valuation adjusted) showed up in the TIC data. China’s recorded holdings of US assets increased by only $14.6 b in the first quarter.
-- the TIC data suggests that China holds a wide range of US assets, not just Treasuries. In 2004, the TIC data indicates that China bought $17.3 b of T-bills, $18.9b of Treasury notes, $16.4 b of Agencies, $12.1b of corporate bonds and $3b of foreign bonds. In q1 2005, China sold $7.2b of T-bills, and bought $7.8b of Treasuries, $7.1b of Agencies and $6.9b of corporate bonds. And that is only the subset of Chinese assets that were purchased through channels that registered in the TIC data.
The AP, in contrast, has a confusing article up about the potential impact of a change in the RMB. One graft in particular caught my attention, because it makes no sense. From the AP.
There are few authoritative calculations of the potential impact of a revaluation in China.
But an estimate by Zhai Shihong, a senior official at the National Bureau of Statistics, says a 5 percent rise in the yuan’s value would cut export growth to below 10 percent, from 35 percent last year.
So a tiny change in the RMB -- 5% -- would cut Chinese export growth from over 30% to under 10%? That is a bit hard to believe. A far bigger change in the dollar (over the past few years) against the euro certainly has not led to a comparable jump in US exports. So why does the author of this article take this argument at face value?
After all, many Chinese firms compete primarily against other Chinese firms. And others rely heavily on imported components -- components that would be a bit cheaper after a revaluation. Most analysts expect that a change in the RMB would, after a lag, slow the increase in China’s exports -- but they also expect the impact would be rather modest.
Goldman, for example, estimates that a 1% change in the real value of the RMB would slow export growth by about 1% -- which seems about right to me. A 5% revaluation might slow export growth by 5% from the current baseline.
I suspect that there is a relatively innocent explanation for this calculation. The National Bureau of Statistics was probably predicting a big slowdown in export growth this year, to say 15%. And then they forecast that a 5% revaluation would cut growth from 15% to 10%. No way would a 5% revaluation alone cut export growth by this much.
If that explanation is true, the National Bureau of Statistics massively underestimated China’s 2005 export growth. But in a broader sense they are right. At some point, and it is hard to know when, Chinese export growth will slow even absent a change the RMB. 30% y/y growth implies that Chinese exports double in three years. With $600b in exports at the end of 2004, continued 30% export growth consequently implies an absolutely enormous increase in China’s exports -- almost certainly an unsustainable increase even if the RMB stays constant. That will make evaluating the impact of any change in the RMB on export growth very difficult indeed.