My beef? China's reported reserve increase combines two things. The amount of foreign exchange the central bank bought in the market. And changes in the dollar value of China's reserve assets stemming from changes in the dollar/ euro, dollar/ yen, dollar/ pound and so on. The central bank's actual purchase of foreign exchange -- the key number -- can only be inferred from the reported reserve increase.
Valuation changes are potentially quite significant. So significant that a wrong call on the euro/ dollar ruined the career of a senior Chinese official back when China only had $200 billion in total reserves. Now China has $820 billion in reserves, and at least $160 billion (20%) and perhaps as much as $240 billion (30%) in reserve assets denominated in euros, yen and other currencies not named the US dollar. So a 10% annual move against a composite of the euro/ yen/ pound would reduce (or increase) China's reported reserves by somewhere between $16 and $24 billion.
Rather than just complain, I decided to do something. RGE monitor is now producing a "China reserve watch." Alas, I have to reserve this kind of analysis for RGE subscribers. But I do want to highlight a couple of key points for the broader audience that participates in this blog.
Before going further, though, I want to acknowledge that the analysis that Casson Rosenblatt and i conducted has been heavily influenced by Eswar Prasad and Shang-Jin Wei's paper "The Chinese Approach to Capital Inflows: Patterns and Possible Explanations." In broad terms, our reserve watch just applies their methodology to the quarterly data rather than the annual data.
Adjusting for valuation - along with reserves transferred to state banks/ and currency swaps that reduce the central bank's reported reserves -- is clearly important.
If you just look at the reported increase in China's 2004 and 2005 reserves, it looks like China's reserves increased at a similar pace in 2004 and 2005.
They didn't. China's 2005 reserve increase was certainly far stronger than its 2004 reserve increase. Moves in the dollar/ euro subtracted from China's reserve growth in 2005, and added to China's 2004 reserve growth. If China ha 20% of its reserves in euro and 10% in yen, its 2005 reserve growth (including $15 billion shifted to one state bank and at least $6 billion in currency swaps) was around $260 billion, v. roughly $190 billion in 2004. The euro/ yen split doesn't matter much - a 70%/ 30% dollar/euro reserve basket produces a similar estimate for 04 gains and 05 losses. An 80/20 split implies slightly smaller valuation losses, and thus reserve growth of around $250 billion rather than $260 billion.
Adjusting the quarterly reserve data for valuation changes also is important -- and yields new insights about the pattern of hot money flows into China.
To calculate hot money flows, Casson Rosenblatt and I estimated quarterly valuation changes to estimate the "underlying" increase in the central banks reserves and compared that total to China's reported quarterly current account surplus and FDI inflows. Hot money is portion of the valuation adjusted reserve increase not explained by the current account surplus and (net) FDI inflows. It corresponds with the non-FDI capital account and the errors and omissions line of the balance of payments.
One caveat: China hasn't reported its current account surplus for q3 or q4, so we estimated the surplus based on the reported merchandise trade data. That is an obvious source of potential error.
What specifically did we learn?
Hot money inflows into China peaked in the second quarter of 2005, not the fourth quarter of 2004. The transfer of $15 billion to ICBC and big valuation losses masked a lot of the q2 increase. Chatter about hot money flows, however, peaked in q4 - as analysts worked off measures of China's reserve increase that counted valuation gains as hot money inflows.
Hot money inflows fell sharply in q3 and q4. That is not what I expected in July. At the time, I thought China's small exchange rate move would lead speculators to bet on further moves, and thus do little to slow the pace of capital inflows into China. Lardy and Goldstein made a similar argument.
We all were wrong. Inflows fell after the revaluation. In part because China tightened controls on inflows. In part because the fed kept hiking US rates. In part because - it now seems - China's central bank flooded the Chinese banking system with liquidity to drive down the interbank rate and widen the interest rate differential between China and the US (see this HSBC publication). In part because the Shanghai real estate market ain't so hot. And in part because China convinced folks that it would allow the RMB to appreciate at a very measured pace.
Finally, Jonathan Anderson overstates his case. The pace of inflows slowed in the fourth quarter, but they did not completely stop. Anderson is certainly not alone in making this argument, but his argument is more sophisticated than most.
What explains the difference between Anderson's calculations and my calculations? I don't think it is valuation: Anderson - unlike many, including Bear Stearns - does adjust for currency moves. Rather it reflects differences in how to account for the November currency swap (I think it reduced China's reserves) and Anderson's decision to use seasonally adjusted reserve data. I think valuation adjustment is essential, but don't like seasonal adjustments in this context.
I wonder what the January reserve data will show. Currency moves increased the value of China's euros and yen - the dollar fell by 2.3% against the euro in January and a bit over 3% against the yen in January, more or less reversing the valuation losses from the fourth quarter. Capital flows into much of Asia picked up, leading other Asian central banks jump back into the market. That suggests a pick-up in hot money flows into China.
Korea's reserves are up $6.5 in January, both from valuation gains and renewed intervention. Stay tuned.