On February 28th, the US Treasury released the preliminary data for the annual TIC survey. While this data is subject to minor revision when the final data is released in April, this data provides a more detailed and accurate depiction of foreign ownership of US assets than the monthly surveys – or so it is commonly believed. This is particularly of interest when attempting to estimate China’s allocation of its foreign reserve purchases, and its requisite effects.
As was outlined in a previous post, in 2012 the US Treasury introduced a new monthly survey of the holdings of US assets by foreigners, in addition to the existing survey of transactions involving US assets. The new survey is aimed at providing an estimation of foreign holdings of US assets that is more timely than annual survey which has a near 12-month lag, and more accurate than the level estimated from the transaction survey, which is subject to fairly large revisions. However, just how accurate the new monthly positions survey (or SLT survey) was was not entirely known because it couldn’t be compared to a full year of data from the other two surveys. With this most recent data release, that benchmark exists.
The SLT survey did indeed give a drastically better estimation of the levels of foreign holdings as of June 2013 than the transactions survey (or S survey) did. Exhibit 1 below shows that the SLT survey gave a more accurate estimate of China’s holdings for each asset class – indeed within $1 billion of the annual survey results for each asset category, whereas the transaction data could differ by nearly $100 billion at times. The drawback of the SLT versus the S survey is that the SLT takes one month longer for the complete set of data to be released. However, so-called major holders of US Treasuries are reported at the same time as the S-survey. The US Treasury is instituting a number of measures to close the reporting gap however, including mandatory electronic filing and expedited data verification systems, some of which will take effect in May of this year.
Taking China as an example, the differences between the surveys is not uniform across asset classes. The transaction data seems to exhibit a downward bias with respect to US Treasuries and equities, but a sizable upward bias in corporate and agency bonds. This has been a trend over the last several years as well. However, the SLC survey exhibits very little bias in any direction. This should come as little surprise however, as SLC survey was designed to largely match the annual survey.
Exhibit 1: Comparison of US Treasury surveys of China’s holdings of US assets as June 2013
Source: US Treasury, author’s calculations
So what is all of this new and accurate data telling us? Firstly, that China was a net seller (in the broad sense of the word) of US Treasuries in December. This was reflected in both the transaction data and the holdings data. This would be broadly welcome as it would indicate that China slowed its intervention in its currency market in December. But not so fast – that’s not what the monthly FX-adjusted headline reserve data from the PBoC shows (exhibit 2). It shows that headline reserves continued to increase in December, albeit at a much slower rate than in October and November. There are two possibilities that we see here: 1) China continued to intervene through December, but is buying less US Treasuries in doing so; or 2) China continued to intervene through December and bought the corresponding amount of US Treasuries, but did so via offshore banking centers. The former is fairly straight-forward, and certainly possible. As we outlined in a previous post, China’s allocation to US assets has clearly been declining over the past 3-5 years. As to the latter, we would look to the known offshore banking centers for any unusual changes in holdings, changes of a magnitude that could be commensurate with actions by a massive market participant such as the PBoC. The marked uptick in holdings by Belgium raises such suspicions: a 25% increase in holdings in one month for a small country whose monetary policy is outsourced to Frankfurt seems unusual. It should be added that transactions entirely between two foreign residents will not show up in any TIC survey. Given that foreigners were net sellers of Treasury bonds in December, it seems plausible that China was indeed a net buyer of Treasuries, but was able to do so offshore in the secondary market.
Exhibit 2: China’s holdings of US Treasuries versus headline FX reserves
Source: US Treasury, Bloomberg
Exhibit 3: Holdings of US Treasuries by select countries
Source: US Treasury