 |
CURRENT ISSUES IN ECONOMICS AND FINANCE |
|
Who Buys Treasury Securities at Auction? |
|
| January 2007 Volume 13, Number 1 |
| ||
| JEL classification: H63, G18, G12 |
| ||
| |||
7 pages / 249 kb| Author: Michael J. Fleming The U.S. Treasury Department now releases fuller information about its auctions than in the past, including new information on investor class and bidder category. The investor class data shed light on the distribution of demand for government securities, and the bidder category data, released first, offer an early read on demand. Purchases by indirect bidders, in particular, are a fairly good proxy for foreign purchases of Treasury notes, but not Treasury bills. In recent years, the U.S. Treasury Department has expanded the information it routinely releases about purchases of its securities at auction. It now reports purchases by bidder category after each auction and discloses purchases by investor class with a shorter lag, and for a broader group of securities, than in the past. The additional, more timely data may be useful in understanding the performance of individual securities.1 A more direct application of the data, however, is in understanding the distribution of demand for government securities across investor groups. Foreign official accounts, in particular, held $1.3 trillion of Treasury securities at the end of 2005, or about 30 percent of all marketable Treasury securities outstanding.2 The amount purchased by “indirect” bidders is often assumed to be a proxy for purchases by foreign central banks, and is sometimes used to draw inferences about such banks' continued willingness to buy Treasury securities more generally.3 This edition of Current Issues describes the information available on the buyers of U.S. Treasury securities at auction and evaluates the extent to which the indirect bid is a good proxy for purchases by foreign investors as a group—that is, both foreign central banks and foreign private investors. The article begins with a description of auction data and an account of how disclosure has changed in recent years. Summaries of the bidder category and investor class data follow. The bidder category and investor class data are then compared in order to illuminate the bidding methods used by various investor classes and to assess how well the indirect bid explains purchases by foreign investors. The analysis reveals that the indirect bid is a fairly good proxy for foreign purchases of Treasury notes (securities with original maturities of more than one, but not more than ten years), but not Treasury bills (securities with original maturities of one year or less). The indirect bid is an imperfect proxy for foreign purchases of Treasury securities because other investors also bid indirectly. For bills, in particular, differences in the indirect bid across auctions are driven more by variation in investment fund purchases than by variation in foreign investor purchases. Publicly Available U.S. Treasury Security Auction Data The Treasury Department releases data on its auctions through a number of announcements and disclosures. In advance of an auction, a press release provides details of an offering, including the offering amount and the term and type of security being offered. The announcement also discloses the quantity of securities that the Federal Reserve has maturing on the issuance date—securities whose proceeds might be reinvested in the new security. In addition, the announcement reveals the amount of the new security that investors in TreasuryDirect are scheduled to purchase.4 Auction participants bid competitively or noncompetitively. Competitive bidders indicate the minimum yield they are willing to accept for a specified quantity of securities. Noncompetitive bidders specify the quantity of securities they are willing to buy at whatever price is paid by successful competitive bidders. All noncompetitive bids are accepted (subject to quantity limits). Competitive bids are then accepted in order of increasing yield until the offering amount is covered. All successful bidders pay the same price, computed from the highest accepted yield. Within minutes of the auction deadline, the Treasury releases the auction results, including the highest accepted yield, the associated price, and, for notes and bonds, the coupon rate of the new security. Also announced are the quantity of securities awarded to the Fed, the quantity awarded to foreign and international monetary authorities for noncompetitive bids made through the Federal Reserve Bank of New York, and the quantity awarded to other noncompetitive bidders. In addition, the Treasury discloses the quantity of securities awarded to investors in TreasuryDirect. Also within minutes of the auction close, the Treasury releases a post-auction addendum with information on competitive bids and awards by bidder category. This announcement discloses the quantity of securities that primary dealers, other direct bidders, and indirect bidders have bid for, as well as the quantity of securities awarded to each of these groups. Primary dealers are institutions that have a trading relationship with the Federal Reserve Bank of New York; other direct bidders are financial institutions that place their bids directly with the Treasury. Indirect bidders are bidders that place their bids through direct submitters; this group includes foreign and international monetary authorities that place their bids through the Federal Reserve Bank of New York. The next release of data occurs on the seventh business day of the month following each auction. At that time, the Treasury discloses information on the purchases of its securities by investor class. The reported classes are Federal Reserve Banks, depository institutions (including banks), individuals (including partnerships and personal trust accounts), dealers and brokers, private pension and retirement funds and insurance companies, investment funds, foreign and international investors, and “other” investors.Recent Changes in Data Release Practices Some of the information released by the Treasury is relatively new. The post-auction addendum was introduced by the Treasury in May 2003 “to improve the transparency and performance” of its auctions.5 The bidder category data included in this addendum are often mentioned in auction commentary because they are thought to provide an early indication of a security's buyers. In particular, the proportion of securities sold to indirect bidders is commonly cited as a proxy for purchases by foreign central banks.6 Additional changes were introduced in May 2005.7 Until then, the Treasury only released investor class allotment data for coupon securities and only on a quarterly basis in its Treasury Bulletin publication. In May 2005, the Treasury switched to a monthly release schedule, expanded the securities covered to include bills, and released a history of the bill data going back to August 2001. Limitations of the Auction Data Although the Treasury now releases more auction data than in the past, the information available about the ownership of specific securities remains limited. The bidder category and investor class data reveal only the allocation of securities at auction. However, the ownership structure of issues can and does change markedly over securities' life cycles. Primary dealers, in particular, buy large quantities of securities at auction and then sell them in the secondary market. Some of these securities are sold after a security is issued; others are sold before issuance in the so-called when-issued market. Despite these limitations, the auction data remain the sole source of information on the ownership of particular Treasury issues. In addition, although an issue's ownership can change over time, the allocation of an issue at auction sheds some light on an issue's future borrowing costs and presumably value (Jordan and Jordan 1997). Moreover, the fact that the bidder category data come out promptly—weeks before the investor class data—explains why they are monitored for ownership information. What the Bidder Category Data Show The bidder category data show that primary dealers alone account for 70.9 percent of Treasury securities sold to the public, on average (Table 1).8 Direct bidders other than primary dealers buy 2.4 percent, and indirect bidders buy 21.6 percent. The average purchase share of noncompetitive bidders is 5.1 percent. The large share of securities bought by primary dealers is notable given that there are only twenty-two of these institutions, whereas more than 800 financial institutions are set up to bid directly in Treasury auctions.9 Purchase shares by bidder category vary significantly across auctions. Primary dealers have bought as little as 33.6 percent of an issue in recent years and as much as 100 percent. Other direct bidders have bought as much as 31.6 percent of an issue competitively, and indirect bidders have bought as much as 64.8 percent. Security type alone can explain much of the variation in bidder category purchase shares (Table 2). Dealer shares are higher for bills (73.7 percent) than for both nominal and inflation-protected notes and bonds, and higher for nominal notes and bonds (60.2 percent) than for inflation-protected securities (53.5 percent). Relatively high dealer participation at bill auctions likely reflects other investors' lack of interest in very short-term bills. Conversely, relatively low dealer participation at auctions of inflation-protected securities may reflect the limited use of these securities as hedging and trading instruments (Sack and Elsasser 2004). Bidder category shares also vary across issues of particular maturities, most notably for bills (Chart 1). Dealer shares for bills decline with issue maturity from 93.1 percent for cash management bills, to 84.7 percent for four-week bills, 66.3 percent for thirteen-week bills, and 63.1 percent for twenty-six-week bills. The miniscule noncompetitive shares for cash management and four-week bills are at least partly explained by auction rules that prohibit TreasuryDirect investors from buying cash management bills and, until October 2005, four-week bills. The small indirect bidder shares for cash management and four-week bills may simply reflect certain investors' lack of interest in these securities because of their short maturities, high issue-size variation and, in the case of cash management bills, irregular issuance schedule. What the Investor Class Data Show The investor class data show that dealers and brokers alone account for 75.4 percent of Treasury securities sold to the public, on average (Table 3). Foreign and international investors account for the next largest share, 12.5 percent, followed by investment funds (6.5 percent), depository institutions (0.5 percent), individuals (0.5 percent), and private pension and retirement funds and insurance companies (0.0 percent). “Other” investors, not captured in any of the classes cited, accounted for 4.6 percent of the Treasury securities sold to the public.10 Again, there is significant variation in purchase shares across auctions, with depository institutions buying as much as 31.6 percent of an issue, investment funds buying 46.1 percent of an issue, and foreign and international investors buying 38.6 percent of an issue. Security type can explain much of the variation in investor class purchase shares across auctions—just as it explained the variation in bidder category purchase shares (Table 4). The pattern for dealers and brokers follows the pattern for primary dealers described earlier, as might be expected. Dealers and brokers' smaller share of nominal note and bond purchases relative to bill purchases is essentially offset by higher nominal note and bond shares for investment funds and foreign and international investors. In contrast, both dealers and brokers and foreign and international investors purchase smaller shares of inflation-protected securities than of nominal notes and bonds, with investment funds making up nearly all the difference. Investor class shares across issues are also revealing (Chart 2). On average, foreign and international investors account for roughly equal shares (20 to 24 percent) of issues with maturities between twenty-six weeks and five years, smaller shares (about 15 percent) of shorter and longer term securities (that is, thirteen-week bills and ten-year notes), and only tiny shares (less than 1 percent) of cash management and four-week bills. In contrast, investment fund shares tend to increase with maturity, with an average of 3.2 percent for cash management bills and 13.7 percent for ten-year notes. Bidding Methods Used by the Various Investor Classes A comparison of the bidder category and investor class data sheds light on the bidding methods used by the various investor classes. An understanding of different investors' bidding methods, in turn, provides useful background for efforts to estimate foreign purchases from the indirect bid. Regressions of bidder category purchase shares on investor class allotment shares suggest the share of purchases made by bid type for each investor class. Further analysis of the auction data reveals instances where investor classes must have made at least some purchases by means of a particular bidding method, because the amount purchased by the class or classes exceeds the amount sold through other methods.11 Results of Regression Analysis Regressions of direct bidder purchase shares on investor class allotment shares show that depository institutions and dealers and brokers (other than primary dealers) bid directly (Table 5).12 The coefficients reported in the table can be interpreted as shares, so that 93 percent of depository institution purchases and 48 percent of dealer and broker purchases are estimated to come from direct bids.13 No other investor class coefficients are significantly different from zero. Regressions of indirect bidder purchase shares on investor class allotment shares show that dealers and brokers, private pension and retirement funds and insurance companies (hereafter, pension funds), investment funds, and foreign and international investors bid indirectly. The coefficients suggest that half of dealer and broker purchases and nearly all purchases of pension funds, investment funds, and foreign and international investors come from indirect bids. (At most 100 percent of an investor class's purchases can come through a particular bidding method. The 1.86 coefficient for pension funds is insignificantly different from one.) Lastly, regressions of noncompetitive purchase shares on investor class allotment shares show that individuals, foreign and international investors, and “other” investors bid noncompetitively. The coefficients suggest that about half of individual purchases, a small fraction of foreign and international purchases, and most purchases by “other” investors come from noncompetitive bids. The results are generally similar when estimated separately for bills and coupon securities. For bills, however, the coefficient for pension funds is insignificant for bidding indirectly and the coefficients for individuals and foreign and international investors are insignificant for bidding noncompetitively. For notes and bonds, the coefficient for dealers and brokers is insignificant for bidding directly (and close to one for bidding indirectly) and the coefficient for “other” investors is significant for indirect but not noncompetitive bids.14 Additional Results from Analyzing Auction Data Consistent with the regression results, further analysis of the auction data reveals several instances when depository institutions must have bought securities directly and numerous instances when dealers and brokers (other than primary dealers) must have bought securities directly. The analysis does not reveal any instances when any other particular investor class must have bought securities directly, but it does reveal many instances when some other investor class or classes must have bought securities directly. In a few of these instances, the direct purchasers must have included investment funds and/or foreign and international investors.15 In contrast to the direct bidder results, there are multiple instances when each of the investor classes must have bought securities indirectly. This includes those investor classes for which the indirect bidder coefficient is insignificant. That is, depository institutions, individuals, and “other” investors bought securities indirectly in at least some auctions, even though the regression results fail to indicate significant indirect purchases by these classes. There are no instances when any particular investor class must have bought securities noncompetitively (excluding the reported noncompetitive purchases of foreign and international monetary authorities). How Well Does the Indirect Bid Explain the Foreign and International Allotment? Another way to analyze the data is to explore the extent to which differences in bidder category purchases across auctions can explain differences in investor class allotments. Such an approach would seem to offer a means of testing the common view that purchases by indirect bidders are good proxies for purchases by foreign central banks. However, the data do not allow a test of this particular relationship, because information is lacking on foreign central bank purchases of Treasury securities at auction (aside from the noncompetitive purchases made through the Fed). Nevertheless, the data do allow a test of how well indirect bids explain purchases by foreign and international investors as a whole. While all investor classes bid indirectly to a certain extent, the earlier results can be used to show that foreign and international investors account for the largest share of purchases by indirect bidders (averaging 46 percent), followed by investment funds (32 percent), dealers and brokers other than primary dealers (18 percent), and remaining investor classes combined (4 percent).16Average allocations vary substantially across issues, however, so that foreign and international investors account for only about 2 percent of cash management bills purchased indirectly, but 74 percent of twenty-six-week bills purchased indirectly (Chart 3). Regressions of foreign and international allotment shares on indirect bidder purchase shares show that the indirect bid does indeed proxy for foreign and international purchases and in the expected way (Table 6). For the two-year note, for example, the average share allotted to foreign and international investors is estimated as 8 percent of the issue plus 50 percent of the indirect bid. The R2—a measure of the variance explained by the statistical analysis—suggests that for the two-year note, 48 percent of the variation in foreign and international purchases is explained by variation in the indirect bid. The regressions also reveal significant differences across issues in the extent to which indirect bidder purchases can explain foreign and international allotments. In particular, the R2 measures are much higher for notes than for bills. Variation in the indirect bid across auctions thus explains 75 percent of the variation in ten-year-note purchases, but only 16 percent of the variation in thirteen-week-bill purchases. The low R2s for the thirteen- and twenty-six-week bills in particular are somewhat surprising as foreign and international purchases of these securities are quite large. Nonetheless, while the average level of foreign and international purchases of these bills is high, the variation in such purchases is low when compared with investment funds' purchases (the funds also buy nearly all of their securities indirectly).17 As a result, variation in the indirect bid for bills is determined to a greater extent by investment fund purchases. Put another way, the indirect bid for bills is a better proxy for investment fund purchases than it is for purchases by foreign and international investors. Conclusion Steps to increase the transparency of U.S. Treasury auctions in recent years have provided useful information to market observers. Investor class allotment information—shown to be related to the performance of individual securities—is now available on a more timely basis and for more securities than in the past. In addition, bidder category data are now available, and their prompt release offers observers an early read on the allocation of securities. The indirect bid, in particular, seems to be a fairly good proxy for foreign purchases of Treasury notes, although not Treasury bills. |
