Mr Knight addressed four issues in foreign exchange reserve management in an opening dinner speech delivered to the Swiss National Bank's conference on 'International reserve diversification and disclosure' on 8 September 2006. He first sketched the present investment of such reserves by currency and instrument, and, second, drew a parallel between the debates over the market impact of diversification and sterilised intervention. Third, he considered the relationship between the appropriate degree of disclosure and a country's exchange rate regime. And finally he questioned recent proposals for standardised or centralised management of foreign exchange reserves.
I appreciate the invitation of the Swiss National Bank and the Institute for International Economics for me to speak to you this evening. The subject of this conference, international reserve diversification and disclosure, is one that I regard as highly important. Indeed, I have worked on this question off and on, in various capacities, for almost three decades. Yet one theme of my remarks this evening may be controversial to some of you: it is that the case for encouraging individual central banks to disclose more information about their foreign reserve holdings is not cut and dried; it raises questions that deserve more debate in a conference like this one. But let me now lay the groundwork for this reflection.
Since the peak in the US dollar's effective exchange rate in early 2002, the global stock of foreign exchange reserves has continued to grow rapidly. Between 2001 and 2005, official international reserves doubled, compared with cumulative global trade growth of two thirds and global nominal GDP growth of two fifths. Some of this expansion presumably reflects central banks' conscious decision to accumulate precautionary reserves, particularly in the case of those countries that were affected by the Asian financial crisis of 1997-98. But there can be little doubt that most of the recent accumulation of official international reserves has resulted from purchases of dollars to limit the appreciation of the currencies of a number of countries. From this perspective, central bank reserve accumulation has been more an instrument than a goal of national monetary and financial policy.
As soon as we start thinking about the management of foreign reserves, two policy issues arise. The first is their allocation by currency; the second is their allocation across instruments.
To address these issues in the brief time I have this evening, I will pose four questions. First, what is the current composition of global foreign reserves by currency and instrument? Second, where do we stand in the debate on the effect of reserve diversification on asset prices and exchange rates? Third, what is the appropriate level, or degree, of disclosure by a central bank of the composition of its international reserves? Fourth, does a one-size-fits-all criterion make sense for the allocation of foreign exchange reserves by currency or instrument?
The currency allocation of global foreign exchange reserves has remained relatively stable over the past few decades, while the instrument allocation has changed significantly. Twenty-eight years ago, when H Robert Heller and I published a Princeton Essay containing the first comprehensive analysis of the currency allocation of central bank reserves, we found that the dollar share in foreign reserves was 80%. 1 Available data now put this share at about two thirds (see table 1). In fact, this ratio has shown intriguing stability over time. The empirical estimates in the paper by Bob Heller and me for a cross-section of 76 countries for the year 1975 found (page 19) that 'on average, the countries in our sample tend to hold 66% of their foreign-exchange reserves in dollars ... independently of their trading patterns and exchange arrangements'. In other words, the modest decline of dollar holdings from 80% of central bank reserves to 66% over 30 years may reflect the simple fact that fewer countries formally peg to the dollar than was the case in the mid-1970s, and that their trading pattern is less focused on dollar area countries. All in all, it seems that the two thirds share has held up remarkably despite a net decline over the past three decades of some 16% in the value of the dollar against a weighted basket of other major currencies and its secular declines of 24% and 44% against the Deutsche mark/euro and the yen, respectively.
Currency composition of official foreign exchange reserves,
|USD billions||Percentage of total||Percentage of allocated|
|Source: IMF, COFER. Table 1|
This remarkable persistence and resilience of the dollar as a reserve currency reflects several factors, including the liquidity of US financial markets and network externalities that lead to inertia in the use of vehicle currencies. Support has also come from the sustained share of global output produced by countries that we can think of as being part of a broadly defined dollar zone in the sense that their currencies tend to move in tandem with the dollar. Today, most of the currencies of the Middle East, Asia and the western hemisphere fluctuate less against the dollar than against the euro or yen.
In contrast, the allocation of foreign reserves by instrument has changed markedly. If we focus just on official foreign exchange reserves held in US dollars, existing data suggest that since the 1960s the proportion held in US Treasury securities has declined significantly. Today, probably less than half of central banks' US dollar reserves are invested in US Treasuries (see table 2). These holdings have declined in favour of investments in other US sovereign, semi-sovereign and agency paper, as well as US private (mortgage and corporate) bonds. US corporate equities also attract a surprisingly large amount of official investment, reflecting placements by central bank and other public long-term investment funds and central bank in-house pension funds. In addition, central bank reserve managers have extended the duration of their portfolios, and now hold most of their dollars in long-term instruments.
Instrument composition of identified US dollar reserves at end-June 2005
In billions of US dollars
|Repos and deposits in the United States||155|
Commercial paper and certificates of deposit
in the United States
|Total||900 (35.2%)||1,658 (64.8%)||2,558 (100%)|
Share of Treasury securities in assets of
the given maturity
|Total IMF-reported US dollar reserves at end-June 2005||1,789|
|Note: Figures for US Treasury, agency, and corporate bonds and equities are from US Treasury, Federal Reserve Bank of New York, Board of Governors of the Federal Reserve System, Report on foreign portfolio holdings of U.S. securities as of June 30, 2005, June 2006. Figures for deposits and money market paper in the United States are from BEA, International Transactions Table 4 (or the US Treasury Bulletin, Tables CM-I-2 and IFS-2). Figures for offshore US dollar deposits are from the BIS Quarterly Review , Table 5C, and the Japanese SDDS for June 2005. The US Treasury definition of foreign official institutions, including “national government-sponsored investment funds” (page 10), may be broader than those of the BIS and IMF. Table 2|
The second question is: where does the debate stand on how diversification of central bank portfolios across major reserve currencies affects global asset prices and exchange rates? Consider the case of a non-G3 central bank diversifying its reserves from dollars to euros.
One can assume that, before this shift, private investors were content with their holdings of dollar- and euro-denominated securities at prevailing interest rates and exchange rates. The central bank shift requires that private investors hold fewer euros and more dollars, and therefore implies that the dollar declines or that dollar rates of return rise.
The impact of such official reserve diversification is similar to that of sterilised intervention in which the Eurosystem or the Federal Reserve sells dollars, purchases euros, and takes measures to leave conditions in domestic money markets unchanged. Both operations imply more dollars to be absorbed by private portfolios. Intervention might signal a change in US or Eurosystem monetary policy, but diversification cannot directly modify domestic monetary conditions in the United States or in the euro area, though it might indirectly influence conditions through yield curve or exchange rate effects.
Let me now turn to the effect of official reserve diversification on exchange rates. How large and sustained can the effect be? Until recently, the prevailing view was that intervention in the markets for major currencies has only a transitory effect. This view followed both from a substantial body of academic literature and from the G10's Jorgensen report of the early 1980s which judged that private holders were, at the margin, practically indifferent to holding more dollar and fewer Deutsche mark securities. Today, in contrast, many market participants and economists consider that news of central bank foreign exchange reserve diversification can move major currencies in a big way. I do not wish to pre-empt any aspect of the discussion at this conference, but I would highlight the need for consistency in addressing this issue.
Can we reconcile these opposing views? One possible explanation focuses on scale. Potential reserve diversification today far exceeds formerly plausible levels of intervention, simply because of the unprecedented size of global current account imbalances and of foreign reserve holdings. Another possible factor is the growing exposure of private fund managers to movements in exchange rates: these managers may seek to 'front-run' official currency shifts.
Conventional wisdom and current conviction also differ with respect to the consequences of the instrument choice. In textbook treatments, whether reserves are invested in short-term or long-term securities makes no difference because bonds are valued on the basis of future short-term rates. However, financial market participants have become convinced that the investment strategies of official reserve managers can and do affect US bond yields. Again, scale may explain why these views are divergent.
My third question concerns the appropriate level of disclosure of the composition of each individual central bank's reserve holdings. Might the degree of transparency in this matter depend on the exchange rate regime that a country pursues? In general, disclosure of the level and composition of official reserve holdings has the virtue of improving public accountability regarding the management of public funds. But there is also a risk that opaque private portfolios might gain some, potentially lucrative, advantage over transparent public reserve portfolios.
To illustrate my point, consider various economies that disclose the currency composition of foreign reserves. In Australia, Canada, New Zealand, Sweden, Switzerland and the United Kingdom, the authorities do not intervene in exchange markets or do so only rarely. For these floating exchange rate regime countries, the benefits of disclosure for enhancing public accountability would seem to outweigh any untoward market effects. At the other extreme in terms of exchange rate arrangements is Hong Kong SAR's transparent currency board, where both the currency shares of reserves and individual interventions are disclosed.
Between these two ends of the exchange rate regime spectrum, however, active management of the exchange rate may tip the balance towards less disclosure. If market participants do not know precisely the currency shares of foreign reserves, unknown valuation effects arising from changes in the dollar's exchange rate against other reserve currencies will confound estimates of the scale of intervention. Authorities might, at least sometimes, welcome such uncertainty. Similarly, they also might not wish to announce the target currency composition, lest changes in these targets might provide unwanted signals of changes in their exchange rate management. In addition, countries with managed exchange rates can end up with very large reserves, a situation that is often associated with unwillingness to disclose currency shares. Indeed, Hong Kong is the only top 10 reserve holder to disclose its currency shares. One explanation might be that central banks which hold large amounts of international reserves may fear that this information could at times lead market participants to anticipate a rebalancing between major currencies.
Perhaps a card game metaphor can illustrate these different policy perspectives. Authorities that actively manage a currency are often playing a strategic game with private market participants, so to them disclosure may seem tantamount to showing their hand and inviting adverse information asymmetries. By contrast, those authorities that rarely intervene may believe that mirrors behind them make for a 'better game', in the sense that they allow for a more efficient determination of exchange rates.
There might be a stronger case for disclosing the composition of reserve holdings by instrument. The advantages of greater accountability may be greater and the attendant disadvantages fewer. This is all the more true for those central banks that choose to accept higher risk in pursuit of higher returns. Losses are easier to explain if the investment strategy of public reserve managers is clearly communicated.
My fourth, and last, question addresses recent proposals for a standardised approach to the investment of foreign reserves. For currencies, Ted Truman has spoken in favour of 'an international standard' to prevent destabilising shifts, although he no longer advocates a single currency allocation for all holders. For instruments, Larry Summers has argued for reserves to be invested in a centrally managed pool of riskier assets to achieve higher returns over long investment horizons, while providing political cover during inevitable down years. Indeed, if authorities want to reallocate reserves towards riskier assets in order to increase long-run returns, several factors would support the idea of doing so collectively. First is the expense that each central bank incurs in acquiring, or developing, the expertise to invest in such additional asset classes as mortgage-backed securities, or in the complex systems needed to extend duration through interest rate swaps. Second is the risk that newcomers can face in investing in opaque markets, such as that for gold. Finally, there is the saving in external managers' fees that can be realised by the pooling of national reserves.
In my view, however, these proposals for a standardised approach rely on the implicit assumption that reserve management should be similar across economies. This assumption is, at the very least, debatable. Indeed, one can say that economic differences across countries argue against different central banks holding the same mix of currencies or financial instruments. On the currency side, many countries align the composition of foreign reserves with their trade shares. For instance, UK reserve holdings put almost twice the weight on the euro as on the dollar, consistent with UK trading patterns. Because sterling trades in relation to the euro and the dollar in such a manner as to keep the effective exchange rate relatively stable, the value of the UK reserve portfolio in sterling terms is also fairly stable.
Turning to instruments, I frankly doubt the appropriateness of investing reserves in a single, common pool across countries. Some official reserve holdings reflect past accumulations of quasi-fiscal surpluses that amount to national wealth, for instance in countries transforming non-renewable commodities into external financial assets. In many countries, however, reserve holdings have been accumulated through sterilised exchange rate interventions. These are in effect financed by interest bearing domestic currency liabilities of the public sector. Is a single global pooled investment model appropriate to both 'real money' reserves and such 'borrowed funds'?
From this perspective, a country-specific approach may be preferred to a centralised or standardised one. One important advantage is that each central bank remains accountable for its overall risk/return profile. Another is that a very large global forex reserve fund could lead to unwelcome market disruption in response to portfolio shifts.
I may not be an entirely objective observer, but I believe that the BIS, as a cooperative international financial institution, has been – and continues to be– instrumental in assisting in central bank international reserve management, as well as in the recent foreign reserves diversification drive of its central bank clients. At the BIS we currently manage about 7% of global reserves on our balance sheet, and we invest the counterpart funds in high-quality fixed income assets. Thus, the BIS already aggregates a large pool of official liquidity and deploys it in a diversified way in global money and capital markets. The BIS also offers a broad range of asset management services, which help to raise the efficiency of global capital allocation. And, by helping to boost the return of public official assets, the welfare of the respective countries is increased.
Let me conclude. Certainly, powerful arguments can be made for the disclosure of the composition of official foreign exchange reserves. Nevertheless, any norms to be considered need to respect differences in exchange rate regimes and other factors in a thoughtful manner. Similarly, comprehensive programmes designed to address any perceived reserve overhang raise the question of whether they can respect national differences in the nature of reserve holdings and economic circumstances. I know these views are not shared by everyone in the room this evening. But, after all, a key objective of this conference is to discuss these issues from diverse perspectives. That is exactly what you will be doing tomorrow. I wish you a very stimulating and successful conference!
1 H Robert Heller and Malcolm D Knight, “Reserve currency preferences of central banks”, Princeton Essays in International Finance, no 131, December 1978.