Has China been diversifying away from the dollar?

The quick answer is I don’t know.

But the revisions to China’s holdings in the survey data – while large absolutely – are nonetheless a bit smaller than I expected. The latest survey data consequently suggest either:

China has reduced the dollar share of its reserves (counting those reserves that have been shifted to the state banks for management)

Changes in the way China manages its reserves – the growing use of state banks, more third party managers and the like – have reduced the share of China’s reserve assets that appear in the US survey data.

In June 2007, identified holdings on the US – a sum that will be smaller than China’s total dollar exposure because it leaves out funds managed by third parties, offshore dollar deposits and dollar-denominated bonds issued by other emerging economies and the World Bank – were $935. That is 70% of China’s reserves, but 62% of China’s “augmented reserves” (augmented reserves includes central bank funds shifted to the state banks through the recapitalization and swap contracts together with the CIC; my method for calculating this is explained here*). That is down from 76% of China’s reserves and 67% of China’s augmented reserves in the middle of 2006, and 77% of China’s reserves and 71%of China’s augmented reserves in the middle of 2005.

* It isn’t clear if my methodology captures the dollars the state banks are now holding to meet their dollar reserve requirement. This though isn’t an issue until the second half of 2007.

As important, from mid 2006 to mid 2007, the $218b rise in China’s identified claims on the US was only 58% of the $373b increase in China’s formal reserves and only 53% of the $411b combined increase in China’s reserves and the foreign assets of China’s state banks (both measures have been adjusted for valuation changes). The comparable numbers for mid 2005 to mid 2006 were 78% and 65%.

Data on China’s holdings come from the US Treasury survey of foreign portfolio investment in the US; my methodology for calculating China’s adjusted reserves is explained here.

More follows -- including more detailed graphs that include estimates of the rise in China’s total holdings after the last survey data point.

As a result, the gap between the increase in China’s reserves and its identified US holdings has increased.

Two points though are worth emphasizing.

First, there are lots of ways of accumulating dollar assets that would not appear in the US data. The data above represents a floor – not a ceiling – for China’s dollar exposure.

Second, starting in late 2005 and then with more force in 2006 China increased the fraction of its reserves managed by the state banks through the use of swap contracts.

Consequently, one plausible hypothesis is that changes in the way China’s manages reserves reduced the share of Chinese asset growth that shows up in the US data. We know from the Chinese balance of payments data that “private” Chinese investors bought $45b of debt in the first half of 2006 and another $64b in the second half of 2006. The $109b in private Chinese debt purchases over the course of 2006 doubled “private” Chinese holdings of foreign debt securities, so it was a big change. It seems likely that the state banks account for the majority of this increase.

This increase slowed in the first half of 2007, but there is still reason to think that the rise in the share of China’s foreign assets managed by the state banks contributed to the fall in the share of Chinese asset growth that appears in the US data. The state banks might buy more adventurous “structures” (like CDOs) that are legally offshore and thus don’t register in the US data, for example. Or they may buy through their Hong Kong branches. The best available data (from Goldman, based on the end-2006 annual reports of the state banks) indicates that the state banks hold the majority of their assets in dollars. We just don’t know if these dollar holdings register in the US data.

The smaller than expected upward revision in the survey data consequently isn’t conclusive .

The fall in China’s recorded holdings of corporate debt (a category in the US data that includes “private” mortgage backed securities) is particularly suspicious. I am reasonably confident – though the evidence is anecdotal – that Chinese state banks spent much of 2006 buying corporate bonds to get higher yields than available on Treasuries or Agencies. That changed in the second half of 2007. The recorded “portfolio investment” of the state banks fell sharply, and anecdotal evidence suggests that the Chinese banks have been unloading some of the risky debt they took on earlier.

Nonetheless, there is little doubt that something has changed.

The following plots – prepared with the help of Arpana Pandey of the Council on Foreign Relations – combine the monthly TIC data with the survey data. The large jumps every June show the impactof the survey revisions.

The latest survey revisions brought a plot of China’s identified US holdings back up to around 70% of China’s formal reserves.

But the revisions fell well short of bringing China’s identified US holdings up to 70% of my measure of China’s augmented reserves (augmented reserves is my name for the estimated total foreign financial assets of China’s state, including the CIC and the state banks )

It follows that the gap between China’s foreign asset growth and the increase in its recorded US holdings is rising -- and that the rise in Chinese holdings is now well below what is needed to maintain a 70% dollar share of China’s reserves (the orange line in the graph below represents what China would have needed to buy to keep a constant 70% dollar share in its broad reserves).

While I suspect some of the gap is explained by undercounting, it probably doesn’t explain the ensure gap. Macroman may have been right. An increase in Chinese demand could have been pushing the euro and pound (and Aussie and Canadian dollars?) up in the first part of 2007 …

A couple of other subplots are worth mentioning –

One – China apparently stopped buying US corporate debt in mid 2006. Its identified holdings fell by about $30b between mid-2006 and mid 2007. That is strange.

Two – China stepped up its equity purchases, with around $30b in purchases from mid 2006 to mid 2007. That was in advance of the formation of the CIC. The State Administration of Foreign Exchange now likely manages a small US equity portfolio.

Third – China has NOT stopped buying Treasuries or Agencies. Recently, it seems to have bought little else. Counting the rise in China’s short-term holdings, China’s purchases of “safe” US assets have been at around $200b since early 2005.

Four -- Even though China’s identified purchases of US debt are a bit on the low side relative to my expectations (and what would be required to keep the dollar share of China’s portfolio constant), they still dwarf China’s purchases of US goods.

From mid-2006 to mid-2007, China bought at least 3.5 times as many US financial assets as US goods. The total could be higher if the US data now understates Chinese purchases in much the same way that it understates Gulf purchases.

That highlights one of my pet peeves.

Most analysis of China’s impact on the US economy often focuses entirely on the goods market. But the US now exports far more debt than goods to China. China’s government is every bit as large a player in the Agency and Treasury market as it is in the commercial aircraft market. And if China really starts buying equities rather than bonds, its impact the equity market could be nearly as large.