V Government Securities Markets in Industrial Countries
- International Monetary Fund
- Published Date:
- January 1994
Internationalization and Institutionalization
Large and recurring government fiscal deficits in the industrial countries have pushed up ratios of government debt to GDP. For the major industrial countries as the group, the ratio of general government debt to GDP was about 43 percent in 1980; the corresponding figure for 1994 is projected to be on the order of 68 percent. Moreover, the presence of large, unfunded public pension liabilities in many of these countries will complicate the already difficult task of reducing those debt ratios over the next decade.40
All of this has forced authorities to think hard about how they could minimize the cost of placing and servicing government debt. And the more they thought about that problem, the more convinced they seemingly became about three conclusions.
First, one could no longer rely almost exclusively on domestic investors. Given the size of the debt, crowding-out factors would push up domestic interest rates too high, and relaxation of capital controls cum the increasingly global competition for saving had rendered domestic investors less captive than before. No, if the debt was to be sold at low cost, governments would have to tap the international market. Moreover, in that international market, it would be the institutional investor that would be prime customer for these bonds. The share of public debt of the seven major industrial countries held by nonresidents now exceeds 20 percent, and this share is on the increase. For example, during 1993, on the order of one half of all domestic and foreign deutsche mark bonds were purchased by nonresidents, and they now hold over 30 percent of all deutsche mark bonds outstanding. Similarly, nonresidents now account for approximately 30 percent of the French Government’s negotiable debt and for roughly 50 percent of bond positions taken on MATIF. So too with the trend of institutional holdings, which have generally risen at the expense of the share held by households. Here, the U.K. figures are instructive. In 1980, households held 16 percent of gilts; by 1992, the household share had fallen to 9 percent.41
Second, if government debt was to be attractive to the international institutional investor, it would be necessary to institute a series of reforms in government bond markets. Those reforms, in turn, would be patterned on the standards of liquidity, transparency, issuing and trading efficiency, and tax treatment established in the world’s premier government securities market, namely, the market for U.S. Government securities.42
Third, if government debt management was to be more clearly formulated in terms of cost minimization and if these reforms in government securities markets were to be implemented effectively, government debt management would need to gain greater independence from the rest of government, and particularly from monetary and exchange rate policies. While much has been made in recent years of the trend toward increasing independence of central banks, this trend toward increasing independence of debt management has been just, if not more, in evidence. Where this has been done, the underlying assumption is that there are sufficient monetary policy instruments available to sterilize the impact of debt management operations on the monetary base. Under this approach, management of the maturity and currency composition of debt also cease to send signals about future monetary and exchange rate policy.
Probably the best examples of efforts to grant debt management formal autonomy from the rest of government and to orient it around explicit performance criteria are found in some of the smaller industrial countries. Since 1988, debt management in New Zealand has been placed in a Debt Management Office (DMO) that, although formally part of the Treasury, has a distinct mandate regarding debt management. It has been charged with maximizing the Crown’s net worth subject to some constraint on the risk it can assume. Ireland created the National Treasury Management Agency (NTMA) in 1990. Its debt management objective is to fund the Government’s debt at a lower cost than that of a benchmark portfolio. Sweden has been one of the most innovative sovereign issuers for some years now; the Swedish Debt Office has had responsibility for all key aspects of debt management since the late 1980s.
Reform of Government Securities Markets
While the sequencing and precise nature of reform in government securities markets inevitably differ across the industrial countries, there is enough of a consistent pattern of these reforms to talk about a consensus. Obviously, the features of the market are not all that counts. If a country has poor macroeconomic fundamentals, the debt management office is going to have more difficulty selling the bonds at a high price—no matter how they improve the market—than if the fundamentals are sound. That being said, the following practices and trends are widespread. The watchwords are liquidity and transparency. Liquidity in the secondary market is especially valued because nothing seems to frighten professional investors more than the prospect of “not being able to get out” when they want to without changing the price significantly. Professionals also do not like surprises. The more they can know in advance what the government intends to do in the bond market, the more comfortable they are about participating in that market.
Primary dealer systems. Authorities in many countries have designated a group of securities houses (or banks) as principal participants—primary dealers—in the government securities market. These firms obtain privileges in exchange for accepting certain obligations. The privileges variously include the right to submit noncompetitive bids at auctions, access to inter-dealer broker screens, designation as counterparties for the central banks’ open market operations, access to repo financing, and bond borrowing/lending facilities with the central bank. In addition, a certain cachet attaches to being a primary dealer. The obligations almost always involve requirements to place “reasonable bids” in auctions, to make a continuous secondary market in a range of issues, and to provide the central bank’s trading desk with market information. Although primary dealer systems restrict entry into the market, the beneficial effect on liquidity—particularly in bear markets and in countries with thin markets—appears to outweigh the drawbacks in most cases.
Issuance techniques. The methods of issuing government securities have undergone significant changes in recent years. The main change is the move away from issuing at fixed prices through syndicates, in favor of auctions. Auctions are now used to issue the bulk of domestic government debt; in contrast, when governments issue in the Euromarket or issue global bonds, underwriting by syndicates is the preferred option. Germany and Japan still rely on a syndicate to market a portion of their domestically sold debt.43 Of the seven major industrial countries, only Italy—a regular issuer on the Euromarkets—uses underwriting syndicates as a regular issuance technique. The gathering consensus seems to be that auctions, by opening up the bidding to a wider base of investors, produce a better price for the government—at least once the market gains a certain maturity; syndicate shares tend to remain fixed for too long, and often unduly favor domestic firms—both of which reduce competitive forces. Once the decision is made to opt for auctions, a choice has to be made between uniform-price and discriminatory methods. Under the former, the issue is sold to all participants at the same price as that of the lowest successful bidder, while in a discriminatory system, each bidder pays the price individually bid. Canada, France, and Germany use discriminatory auctions exclusively, while other countries use a mix of methods, generally with discriminatory auctions used for most of the issue volume. The United States has recently adopted on an experimental basis uniform-price auctions for sales of some Treasury notes (the lessons from that experiment are not yet in). There is no clear presumption—be it based on theory or practice—that one auction-pricing method is superior to the other. Although discriminatory auctions might be thought to increase revenue for traditional price discrimination reasons, critics have argued that this is offset by the tendency of this method to exacerbate the “winners’ curse” phenomenon (whereby successful bidders overestimate the value of the securities).44
Issue calendar. In response to calls from market participants, many countries have adopted a firm, preannounced issue calendar. Increased certainty about issue dates and about amounts of government securities to be issued is said to help market participants to place the issue. It allows institutional investors to structure the maturity of their investment portfolios in line with the issuing calendar. It is often maintained that greater predictability lowers the cost of issues. Some countries, such as France, have become strong believers in such a firm issue calendar and have shown great determination to stick to it, even during periods of extreme turbulence (e.g., during the ERM crisis). Some others (the United Kingdom, for one) are more skeptical; they emphasize the wisdom of having enough leeway in timing to take advantage of opportunities for obtaining the best price, as well as the need to prevent the market from changing the environment in their favor just before an auction.
Position financing. The key consideration here is the ease with which market participants can finance their securities positions. The most common financing vehicle is the repurchase agreement (repo); in a repo, an investor sells a security to another party, while at the same time agreeing to repurchase the security at a future date and at a prespecified price. Repos are in essence collateralized borrowing (which obviates the need for credit risk assessment): the trader that held the original security and executed the repo remains exposed to price movements on the securities used in the repo. The ability to execute repos is particularly important to foreign firms who do not have access to a domestic deposit base. The standardized, short-dated repo is becoming the main instrument for position funding in the major markets. It facilitates the taking of leveraged, long positions in securities. Where there are no repo markets, funding has to be in the form of uncollateralized lines of credit from the banking system.
The largest repo market is, not surprisingly, located in the United States. In October 1993, primary dealers alone had open repo contracts in government securities of $852 billion, equivalent to 30 percent of U.S. marketable debt outstanding. Most repos tend to have a short maturity of a few days. The repo market in Japan (Gensaki) accounts for about 30 percent of transactions volume in government securities. Repo markets elsewhere tend to be substantially smaller. In the United Kingdom, only gilt-edged market makers buy repo gilts, and they must do so through the stock exchange money brokers. The Bank of England began in January 1994 to use gilts repos as a regular open market device. This has raised expectations among observers that the Bank may soon also liberalize the use of repos among private market participants.
In some countries, enthusiasm for the establishment and growth of private repo markets is restrained by the concern that their liquidity-enhancing attributes could well be offset—or even more than offset—by a higher potential for systemic instability, generated in turn by the high leverage ratios which repos facilitate. Indeed, repos are the preferred way for hedge funds to take large, highly leveraged positions in government securities.
Hedging instruments. Liquidity and efficient price discovery have also been underpinned by the development of organized futures markets (offering interest rate futures and options on government securities, for a few benchmark maturities). Dealers and other participants will be more inclined to hold trading portfolios if they can hedge interest rate risk. In addition, futures markets often have lower transactions costs than underlying cash markets; as such, trading volume tends to be far greater in some of the successful futures contracts than in the underlying cash markets. The when-issued market—a market in securities that have not yet been issued, with trades being settled on issue day—also allows the hedging of auction bids. Such markets now exist in most of the seven major industrial countries (Germany and Japan are the exceptions).
Withholding taxes. Withholding taxes, turnover taxes, and stamp duties are an anathema to institutional investors and traders alike. Such taxes tend to segment markets; they also reduce liquidity because they impede trading (Box 3). The trend is to eliminate these taxes altogether, or if this cannot be done for equity or political reasons, to reduce them and to rebate them to foreigners as quickly and as smoothly as possible. In New Zealand, it is reported that institutional demand for its securities increased so much in the few days after it abolished the withholding tax for nonresidents that it induced a fall in the yield large enough to more than offset lost tax revenue for the year.45
Box 3Withholding Taxes and Government Borrowing
Some countries levy a withholding tax on government securities. The revenue yielded by these taxes is often substantially offset by an increase in the interest rate paid by governments, while the taxes often create administrative burdens for foreign investors in government securities (including the rebating of such taxes). These burdens reduce the willingness of these investors to hold a country’s securities and can therefore increase the cost of debt service by more than the revenues raised by such a tax.1
Usually, such a tax is part of a tax on all interest income in the economy and is imposed on government securities partly as a way to treat government and private borrowing equitably. Both foreigners and residents are likely to be subject to the tax, although sometimes at different rates. Foreigners can get the tax rebated or reduced if their countries have tax treaties with the withholding country.
Among the seven major industrial countries, only Italy and Japan impose interest withholding taxes on nonresident holders of government securities. In Italy, the tax rate is 12.5 percent for foreigners, and in Japan the rate is 15 percent.2 The rate can be reduced for residents of some countries; German residents, for example, pay no tax in Italy, and German and U.S. residents pay a 10 percent rate in Japan. The United Kingdom and Germany impose withholding taxes on domestic residents, but exempt all foreign residents. A key issue with all such full and partial exemptions is the ease with which nonresidents can obtain a refund; market participants report that this process has in the past been cumbersome in Italy, although authorities have recently made efforts to improve the efficiency of the rebate process. Other industrial countries that impose withholding taxes include Australia and Switzerland.
The United States eliminated a 30 percent withholding tax in 1984; by then, it had come to be regarded as an ineffective means of taxing private capital flows. Loopholes and tax treaties meant that the effective U.S. tax rate on interest paid abroad was only 2 percent in 1983. One commonly used loophole permitted U.S. corporations to borrow on the Euromarket through subsidiaries in the Netherlands Antilles, which was tax exempt because of a tax treaty.3
Germany imposed a 10 percent withholding tax briefly in 1989. Borrowing by foreign residents in deutsche mark was exempt from the tax, and investors in deutsche mark shifted their assets toward these loans and away from debt issued by German residents. The result was an increase in the interest rates paid by domestic residents.4 The tax was undermined by large-scale evasion and was abolished after only six months. In 1993, Germany reimposed a 30 percent withholding tax on interest income; interest on government securities paid to residents is to be withheld by the bank transmitting the payments. Nonresident depositors and bond holders are exempt from this tax, but the new tax does not exempt borrowing by foreign residents. There have been ongoing efforts among countries in the European Union to harmonize withholding taxes across countries. Such a harmonization would, for example, significantly reduce the incentive for German capital markets business to flow through neighboring countries to avoid withholding tax.
A withholding tax has different effects on private and government borrowing. For a private borrower, the tax raises the interest the borrower must pay to give lenders the same net return. On the other hand, a withholding tax does not necessarily raise the cost of the government borrowing, since the government also collects the tax revenues. If lenders continued to demand the same net return, the government would have the same debt costs as before, once its new tax revenue is taken into account. The administrative costs imposed by the tax can, however, increase the net return demanded by foreign residents and potentially increase the government’s net debt-service costs.
In a country with a withholding tax, holders eligible for a lower tax rate can apply for a rebate. As noted above, however, this process can be cumbersome. The administrative cost and uncertainty in the rebate process are real economic burdens caused by withholding taxes. To reduce this burden, rebates should be automatic and prompt.
That such costs can be substantial is illustrated by experience in New Zealand, where the authorities credit the effective elimination of withholding tax on interest payments to nonresident holders of government bonds with a significant reduction in borrowing costs. Since 1991, nonresident holders of government bonds have been able to exempt themselves from the interest withholding tax—currently between 10 percent and 15 percent—by paying an Approved Issuer Levy (AIL) of 2 percent of the interest payment. In July 1993, the Debt Management Office (DMO) began paying the AIL on behalf of nonresident holders out of its own resources. On the night the change in policy was announced, yields on ten-year bonds declined by 30 basis points—resulting in an interest cost savings that more than outweighed the AIL payments the DMO had undertaken to make. Although other factors certainly played a role, the DMO’s decision contributed to a decline of 70 basis points in the spread of ten-year government bonds versus U.S. Treasury bonds by the end of October 1993.
Despite this evidence, there are two counteracting effects through which a withholding tax may potentially lower the government’s debt costs. Because taxpayers in many countries receive a tax credit for taxes paid to foreign governments, they do not bear the full burden of the withholding tax.5 In such a case, these investors will settle for a lower interest rate from the withholding country, net of the withholding tax, because their tax credits mean that they do not pay the full withholding tax. The second way a withholding tax can lower debt costs is through the tax it imposes on all borrowing (private plus public); this reduces the amount of competing borrowing in the capital market and may therefore allow the government to pay a lower interest rate on its borrowing. Although both of these effects may reduce government borrowing costs, they come at the expense either of foreign governments (which must pay the tax credits) or of private borrowers (who incur higher borrowing costs).1 Similar effects can occur with taxes on transactions of securities. Such taxes often lead investors to modify their behavior to avoid the tax, either by trading related but untaxed instruments, or by trading offshore; in other cases, investors simply reduce their trading volume, directly shrinking the tax base. The experience of Sweden and the United Kingdom with such taxes is described by Campbell and Froot (1993). In Sweden, a tax on domestic equity transactions led to increased offshore trading, while the tax on fixed income transactions, including government bonds, led to domestic trading in related—but untaxed—instruments. Total turnover of Swedish fixed-income instruments also fell by two thirds with the imposition of the tax. Although the effects of the British stamp tax on transactions in U.K. equities are less clear, it appears to reduce the number of transactions and to cause a switching into untaxed derivatives and American Depository Receipts (ADRs).2 See Huizinga (1994).3 See Goulder (1990).4 See Goulder (1990).5 See Huizinga (1994).
Investor base. Different investors have different needs. Germany, for example, just recently issued a 30-year Bund to take advantage of the needs of pension funds to have an income profile that matches the profile of expected disbursements. For much the same reason, “stripped” government securities (i.e., securities that break the principal and interest components of a bond apart and sell them as separate securities) have become very popular in the United States because different investors have different requirements for the timing of interest and principal revenues. Insurance companies, mutual funds, dealers, and banks may, in turn, each have their own preferences for maturity profile, coupon, tax status, and so on. The investor base also carries implications for liquidity. For example, because retail investors tend to be less active in trading than institutional investors, issuers in some smaller markets may be reluctant to reserve too high a share for them in view of the adverse effect on liquidity. In other circumstances, when the desire may be to increase long-term holding, retail investors could be the target group.
Large issue sizes in a few benchmark maturities. Generally, the larger the issue, the more liquid it is. By concentrating on a few benchmark or “on-the run” issues, liquidity can be increased substantially. Transactions volume tends to be concentrated in benchmark issues, and other fixed-income securities are typically priced with reference to these benchmarks. Italy, Belgium, and Sweden have refinanced smaller issues with larger benchmark issues, yielding saving of anywhere from 5 to 15 basis points. Likewise, trading in the Japanese Government bond markets is sufficiently concentrated in the benchmark issues that the yield differential between benchmark issues and nearby issues is also often on the order of 5–15 basis points. Issue size is particularly relevant for smaller government bond markets where there may not be enough investor demand to spread across too wide a spectrum of issues. Some countries have increased the size of issues by reopening the issue of an existing security instead of issuing a different security.
A smoothly functioning clearance and settlement system. This is yet another liquid-enhancing mechanism. Where this is absent, trading will be held back by delivery problems, and systemic risk will increase as the outstanding volume of unsettled trades increases. A book-entry system for securities that is closely tied to the wholesale payments system is almost indispensable in this regard. A trade netting system can also contribute to efficiency by reducing the number of payments that brokers and dealers need to make. Links between domestic clearing systems and international clearing systems, especially Cedel and Euroclear, are likewise regarded as facilitating international transactions in a country’s debt.
Global bonds. The legal separation of Euro-issues and domestic issues has long operated to segment domestic and Euromarkets. Recently, an instrument known as a “global bond” has been developed in an attempt to overcome this segmentation. Global bonds are issued in a number of major financial centers simultaneously (usually, Europe, the United States, and Japan). They require bridging arrangements among different national clearing systems; when these arrangements can be put in place, global bonds can flow easily across jurisdictions, thereby enhancing liquidity by allowing buyers in different locations to react to price fluctuations. The World Bank and a variety of public issuers—Italy, Sweden, China, and Argentina, among them—have successfully used the global structure for some of their debt.
It is far from straightforward to put together a robust ranking of liquidity in the major government bond markets. For one thing, some national markets may have very high liquidity in a few benchmark issues but relatively low liquidity in all the others, whereas other markets may have somewhat lower liquidity in the benchmark issues but higher average liquidity across the whole spectrum of outstanding securities. For another, there is no single measure of liquidity that is preferred to all others and for which there is a published, long time series available for the major markets. When asked for their own ordinal rankings, a large sample of private firms indicated that they would place the U.S. Government securities market in first place by a considerable margin. The next tier would contain the benchmark issues of Japanese Government bonds, the benchmark deutsche mark bonds, and French OAT issues. The major gilt issues would be at the top of the next tier, followed in no special order by Canadian federal bonds, Italian Government bonds, Belgian linear bonds, and Spanish Government bonds.
See International Monetary Fund (1993), Table 10, p. 60.
In sharp contrast to many other countries, households reportedly hold the dominant share—about two thirds—of government debt in Italy. Only about 3 percent of Italian debt is issued on international markets, although more may be held by foreigners. Portugal is another country with little foreign-held debt.
Discussions with market participants and authorities revealed that institutional holders turn over their security holdings more frequently than retail holders, but that turnover does not depend particularly on the holder’s country of residence. The impact of turnover on market volatility is also unclear.
The market for U.K. Government securities—for gilts—has also served as a model for some of the reforms.
Syndicated bonds in Japan are issued at the weighted-average price from auctions for the other tranche of the issue.
For a review of different auction formats, see Feldman and Mehra (1993).
Ongoing efforts in the European Union are aimed at harmonizing withholding taxes across member countries.