from Follow the Money

So, have central banks lost interest in Treasuries?

April 30, 2005

Blog Post
Blog posts represent the views of CFR fellows and staff and not those of CFR, which takes no institutional positions.

More on:

Monetary Policy

Two apologies in advance: First, this is one my wonkier/ data-intensive kind of posts. Second, this post responds to a subscription only Wall Street Journal C4 article by Agnes Crane that ran last Tuesday that draws on a Goldman Sachs Global Viewpoint paper (Asian Central Bank buying -- the Beginning of the End, 20 April, by Dominic Wilson) that also is not available to the public.

The conclusion of the Goldman Study:

"This year is likely to see significantly less buying of US Treasuries -- and US fixed income in general -- by Asian central banks"

I think the Goldman study is only 1/2 right.

There will be less central bank support of Treasuries. Every Asian central bank now adding to its reserves has said that they intend to hold a more diverse portfolio of dollar debt -- think Agencies, mortgage backed securities, corporate debt, perhaps even CDOs. You name it.

But if China holds on to its peg for the full year, or only makes cosmetic changes, the overall dropoff in demand for US assets from Asian central banks is likely to be a bit smaller than Goldman estimates. Moreover, focusing on Asian central banks misses a key part of this years’ story: very rapid reserve accumulation by oil exporters.

All in all, I don’t think global reserve accumulation will fall off much. When the financial history of 2005 is written, the big story will be same as in 2003 and 2004: unprecedented global reserve accumulation.

However, with constant reserve accumulation and a rising US external deficit, central banks will supply a smaller fraction of the net financing the US needs.

Dollar reserve accumulation may fall off just a bit. Central bank Treasury purchases will fall off even more. Above all, though, I expect a huge drop off in RECORDED central bank inflows. The countries now adding to there reserves tend to hide their tracks; the Japanese did not.

My previous post provides the gory details. Japan keeps almost all its reserves in dollars, invests almost all its reserves in the Treasury market and almost all Treasuries purchased by the Bank of Japan (for the MOF) seem to show up in US data. Other central banks don’t invest as a high share of their reserves in dollars, don’t invest as much in Treasuries, and above all, are more inclined to invest in ways that tend not to show up in the US data.

Central bank support for the Treasury market (assuming that central banks are not dumping any of their existing holdings) hinges on three things. 1) The pace of global reserve accumulation

2) The fraction of new reserves that are kept in dollars 3) The fraction of new dollar reserves that are invested in the Treasury market.

Let’s look at each of these three components in more detail.

The pace of global reserve accumulation.

Total global reserve accumulation in 2004 exceeded $710 billion, but that total includes valuation gains. New reserve purchases were probably closer to $625 billion, maybe a bit more.

The two biggest players by far were China ($210 b) and Japan ($178 b). But Korea ($44 b), Taiwan ($35b), Malaysia ($22b), Singapore ($14b), Thailand ($14 b), India ($29b) and Russia ($44b) all also added substantially to their reserves in 2004. Collectively, their reserve accumulation almost matched China’s.

What is likely to happen in 2005? The government of Japan, as of now, is out of the market. That seems likely to continue: a slumping Japan combined with higher US interest rates create an incentive for someone -- whether Japanese investors or US hedge funds -- to engage in the yen carry trade. China’s reserve accumulation looks set to accelerate on the back of a rapidly growing trade and current account surplus, strong FDI inflows, and continued "hot money" inflows. $275-300 b in Chinese reserve accumulation looks likely. Russia, India, Taiwan, Malaysia, Brazil and Korea all added substantially to their reserves in the first quarter -- the underlying pace of reserve accumulation in eleven major emerging economies was about $125 b in the first quarter. I expect that to continue and even accelerate on the back of China, though some countries may add more and some less over the course of the year.

Goldman recognizes this: they argue that the Asian reserve accumulation story is increasingly a BRICs (Brazil Russia India China) story.

I would say it is a BRICs + NICs + Malaysia + the oil exporters story. The IMF estimated that the oil exporters in the middle east will add about $80 b to their official reserves. The ruling families in many oil sheikdoms will also add to their private reserves.

Add up the additional reserve accumulation in the oil exporters and China, and I suspect global reserve accumulation will stay around $600 b. No big drop off.

But the US current account deficit is gonna be a lot bigger than it was in 2004 -- so as a share of the US current account deficit, global reserve accumulation will fall.

What share of these reserves will be invested in dollars?

Less than in 2003, when about 90% of all new reserves were placed in dollars (global holdings of euro reserves grew, in dollar terms largely because existing holdings of euro went up in value). Less than in 2004, when I would guess a bit over 75% of all new reserves were invested in dollars. Remember, almost all of Japan’s reserves are invested in dollars, so even if some other countries were only putting 67% or even 50% of their reserve growth in dollars, the overall percentage will be higher. Maybe more like 67%.

That is just a guess though. The oil exporters in particular seem far less inclined to invest in dollars than the Japanese, or even the Chinese.

Remember, 1/3 of $600 billlion is $200 billion, which is a lot of money. It might even have something to do with low Eurozone yields right now. Australia needs some inflow to finance its own current account deficit, as does the UK -- but the Eurozone doesn’t. Some official inflows to Europe can be offset by private outflows from Europe, but at the end of the day, the world’s current account surplus mostly has to be used to finance the US current account deficit.

So my baseline estimate for dollar reserve accumulation in 2005 is about $400b -- somewhat less than in 2003 and 2004.

That means private investors are playing a larger role financing the US this year than in previous years. Given the math, the "private" funds that are flowing to the US are ultimately coming mostly from Japan and the world’s oil exporters. They are the countries with current account surpluses that are not captured in reserves, and thus are available for private investment -- but these savings may be coming to the US via a rather roundabout route. Some of these private flows are not really that private: think of the private accounts of the oil sheiks. Others reflect, I would argue, expectations that central bank will intervene to limit dollar appreciation. Think of the yen-dollar and the yen carry trade.

Finally, what fraction of the reserve inflow will go into the Treasury market?

Again, a lower fraction than in 2004. Why? Japan. Japan not only bought lots of Treasuries, but it did so in ways that showed up in the US data.

The trend in reserve management is "diversification across dollar assets" -- that means more agencies, more mortgage backed securities, more corporate bonds and fewer Treasuries. Central banks already have all the super-safe short-term Treasuries they need; increasingly they are looking for a bit more yield. If only 50% of the expected $400 billion increase in central bank dollar reserves is invested in the Treasury market, there would be a substantial drop off in central bank support for Treasuries -- and probably an even sharper drop off in observed central bank support for Treasuries.

In 2004, recorded central bank purchases of Treasury bonds and notes in 2004 totaled $203 billion (data from here; the number implied by the change in the TIC stock data is ever so slightly lower). Add in bills, and the central banks bought $236 billion of US Treasury debt. Add in bank deposits, central bank purchases of agencies, and a few other things and you get total RECORDED central bank inflows of $355 billion - I cannot get the BEA number for Treasury purchases ($261 b) to match the TIC data ($233 b) though. Even $260 billion though probably understates real central bank purchases of Treasuries, if you include the treasuries central banks likely purchased through indirect channels. Some of the $153 billion of Treasuries purchased by "private" investors abroad in 2004 ($153 b = TIC data/ $108 b = BEA data) were bought on behalf of the world’s central banks.

All told, I suspect central banks bought well over $300 b in Treasuries in 2004. Suppose that falls to $200b in 2005 (with recorded inflows falling by even more). $200 b is not $300b, but $200 b is still a decent sum of money.

That is a long-winded way of saying Goldman is right -- there will be less central bank support for the Treasury market. However, overall reserve accumulation though is not likely to fall as much as Goldman suggests (don’t forget about the oil exporters) and there could well be more central bank support for other US fixed income markets.

More on:

Monetary Policy