Treasury Securities

In less than a decade, Treasury securities have grown from the third largest category, behind mortgages and corporate and foreign bonds, to the largest category, at $24.3 trillion. When the US government spends more than it collects in taxes and fees, which has been the case for most of the last 50 years, it needs to borrow money to fund its deficit spending. It does so through the array of instruments shown in Figure 0.5. Treasury bills or T‐bills mature in one year or less and are discount securities, which means that they sell for less than, or at a discount from, their promised payment at maturity. Treasury notes and bonds are coupon‐bearing securities; that is, they earn a fixed coupon or interest rate on their principal, face, or par amounts through maturity, and then repay that principal amount at maturity. Strictly speaking, and in the accounts of the government, notes are issued with 10 or fewer years to maturity, while bonds are issued with more than 10 years to maturity. The distinction had more meaning historically, when bonds were subject to a maximum, statutory rate of interest. In common parlance today, however, the words “notes” and “bonds” are used interchangeably. In any case, Chapter 1 describes the cash flows of Treasury notes and bonds in more detail.

Treasury Inflation Protected Securities (TIPS) protect investors against inflation with principal amounts that increase or decrease with changes in the consumer price index (CPI). Consider, for example, a TIPS with a principal amount of $100 and a coupon rate of 1% per year. If CPI has increased by 10%, the principal amount of the TIPS will have increased to $110, and the investor earns 1% on that higher principal amount. This investor is just as well off having $100 and earning $1 per year at the original price level as having $110 and earning 1% $110 or $1.10 per year at a price level that is 10% higher. Hence, TIPS earn a fixed real or inflation‐adjusted return, while coupon‐bearing Treasuries earn a fixed nominal or dollar return.² For this reason, by the way, in discussions that include both TIPS and Treasury bonds, the latter are often referred to as nominal bonds. In any case, while comprising only 5.6% of the total in Figure 0.5, TIPS have an outsized importance as measuring the market's perception of inflation and the price of inflation risk. As of January 2022, for example, the rate on five‐year nominal Treasury bonds was about 2.8% higher than the rate on five‐year TIPS. Roughly expressed, therefore, the market expects an average inflation rate of 2.8% over the subsequent five years. More precisely expressed, given inflation expectations and risk preferences, investors require a premium of 2.8% to buy five‐year nominal bonds and to assume inflation risk over that horizon.

Figure 0.5 next lists floating‐rate notes (FRNs). These are relatively new, having been first issued in January 2014. FRNs are sold with two years to maturity, and they pay a variable rate of interest equal to the going rate on 13‐week T‐bills plus a fixed spread. This spread entices some investors, who might otherwise roll investments of short‐term T‐bills, to sacrifice some liquidity and buy two‐year FRNs instead. From the perspective of the Treasury, FRNs lock in funding for two years, but at a cost only slightly above that of short‐term bills.³ In fact, over 2021, FRNs were sold at a spread of less than five basis points.⁴ In addition, somewhat cynically, FRNs seem to cost less than two‐year notes, because government accounting of interest rate cost does not penalize the risk of rates increasing in the future. In any case, the issuance of FRNs has remained limited, comprising about 2% of the total in Figure 0.5.

US Treasury issues are among the most actively traded securities in the world. This is due, in good part, to the significant role of the US dollar in international markets and to the perception of US government debt as among the best stores of value available. An additional explanation, however, is the careful management of Treasury debt issuance. More specifically, the US Treasury sets a regular auction schedule that lets investors know, in advance, which securities will be sold when and in what quantities. Furthermore, the Treasury has gradually modulated this schedule over many years to suit both the borrowing needs of the government and changing market conditions. To appreciate these points, Table 0.2 describes the auction schedule as of January 2022.

“Issue frequency” describes how often new securities of each type are issued. The Treasury has settled, for example, on issuing new notes with two, three, five, and seven years to maturity every month, while issuing new 10‐year notes, along with new 20‐ and 30‐year bonds, every quarter. Both the set of issues and their frequency have changed over time, however. For example, issuance of the 20‐year bond was eliminated in 1986 and brought back in May 2020, while issuance of 30‐year bonds was stopped after August 2001 and resumed in February 2006. And in 2000, the US Treasury introduced the concept of reopenings, which means auctioning or selling more of an existing issue. For example, Table 0.2 reports that 10‐year notes are issued quarterly and reopened monthly. In August 2021, the Treasury sold about $59 billion of a new 1.25% 10‐year note, that is, a note that had not been issued before, that pays interest on principal at an annual rate of 1.25%, and that matures on August 15, 2031. The following month, in September 2021, the Treasury sold another $42 billion of that same issue, that is, more notes with a coupon of 1.25% that mature on August 15, 2031. And again the next month, in October 2021, the Treasury sold another $41 billion of that same issue. In November 2021, however, a quarter after the first issuance of the 1.25% 10‐year notes, the Treasury sold $62 billion of a new 10‐year note, with a coupon of 1.365% and a maturity date of November 15, 2031.

One result of the auction schedule is that the most recently issued bonds of each type and maturity tend to be the ones most actively traded. In the previous example, as of November 15, 2021, the just‐issued 1.365% notes maturing on November 15, 2031, are called the 10‐year on‐the‐run notes and are likely to become the most actively traded of notes with approximately 10 years to maturity. The 1.25% notes maturing on August 15, 2031, which had been the 10‐year on‐the‐run notes, become the old notes, and over time, as even newer 10‐year notes are issued, become the double‐old notes, the triple‐old notes, etc.

Returning to Figure 0.5, nonmarketable securities is another small category of Treasury issuance. Included in this category are about $140 billion of savings bonds, which are discount securities sold directly to individual investors, and about $120 billion of State and Local Government Series bonds, commonly known as SLUGs, which are mentioned later in the context of municipal bonds.

The final category in Figure 0.5 refers to Treasury bonds sold into government accounts. These bonds are also nonmarketable and represent debt that the US government owes itself. The social security trust funds, for example, at the end of 2020, held about $3 trillion in Treasury bonds. Most people argue that these bonds do not represent any additional Treasury indebtedness, or, in other words, that there is no difference between social security benefits being paid by the Treasury directly or being paid indirectly through the payment of interest and principal on bonds in the social security trust funds. By this logic, Treasury bonds held in government accounts are excluded from most descriptions of government indebtedness. More specifically, in terms of Figure 0.5, US government debt is usually equated to “Total Debt Held by the Public,” or $22.3 trillion, rather than to “Public Debt Securities,” which adds the $6.2 trillion held in government accounts, for a grand total of $28.5 trillion.

In discussing the magnitude of government debt, and in comparing government indebtedness across countries, debt held by the public is usually normalized by gross domestic product (GDP), which measures the value of the goods and services produced in a country in a single year. The idea here is that countries with greater GDPs can safely carry greater levels of debt. With US GDP at about $22 trillion as of June 2021, the ratio of debt to GDP in the United States is about 100%, which is extremely high by historical standards. The ratio was over 100% during World War II; subsequently declined to between 20% and 50%; climbed to about 80% in the aftermath of the financial crisis of 2007–2009; and then shot up to above 100% in the wake of the COVID pandemic and economic shutdowns. For comparison purposes, the ratio of debt to GDP in Japan is currently over 230%; in Greece about 175%; in France about 100%; in the United Kingdom about 85%; in Germany and China, less than 60%; and in Switzerland, less than 40%.

Despite the historically high ratio of US debt to GDP, foreign investors continue to find US Treasuries attractive and hold a large fraction of the amount outstanding. As of June 2021, investors outside the United States held $7.2 trillion Treasuries, or 33% of the $21.8 trillion marketable securities shown in Figure 0.5. These holdings include $1.28 trillion (5.9%) in Japan, $1.06 trillion in China (4.9%), and $0.53 trillion (2.4%) in the United Kingdom.

Municipal Securities

The $4 trillion municipal securities market includes more than 50,000 issuers and approximately one million individual bond issues. Municipal bonds, sometimes called municipals or simply munis, are issued by states and local governments to fund their expenditures. Unlike most fixed income markets, the muni market is dominated by retail investors: as of June 2021, over 70% of principal outstanding was held directly by individuals, or indirectly by individuals through mutual funds and other investment vehicles.

Interest payments from munis are exempt from federal tax so long as funds raised from selling those munis are used for public projects. Therefore, investors who pay federal taxes are willing to accept lower rates of interest from munis than they would from bonds whose interest is taxed at the federal level, like corporates and Treasuries, controlling, of course, for differences in credit quality. And municipal issuers can raise funds at rates below what they would otherwise have to pay. The story is somewhat more complicated, however, because capital gains on price appreciation from munis is not exempt from federal tax. To avoid this tax, however, the market has evolved to minimize the proportion of return in the form of (taxed) price appreciation rather than (untaxed) interest. More specifically, most munis – in the current low‐rate environment – are issued at a coupon rate of 5% and, consequently, at a premium to par, that is, at a price greater than face amount. In this way, these munis are unlikely to trade at a discount and thus subject subsequent buy‐and‐hold purchasers to taxes on capital gains. With respect to state taxes, the treatment of municipal interest varies by state. Most states tax bonds issued in other states and exempt their own bonds, while some states tax both their own and other states' bonds. Washington, D.C., does not tax any municipal bond interest, and Utah exempts municipals issued in states that exempt Utah's bonds! Utah's exemption is more significant than it may seem, because states with no income tax automatically qualify. The final piece of the tax story is that about $500 billion of municipals do not qualify for the federal tax exemption, because their proceeds are used for working capital, for funding private business development, or for refinancing existing debt through particular kinds of transactions.⁵ These taxable munis pay interest at rates comparable to those in the corporate and Treasury markets, again controlling for differences in credit quality.

Muni bonds can be divided into three broad groups. General obligation (GO) bonds, which constitute about 25% of the market, are backed by the taxing power of the issuing municipality. Revenue bonds, which constitute about two thirds of the market, are backed by revenues from particular projects, like tolls from a bridge or highway. And within revenue bonds is a $600 billion subset of industrial revenue bonds, through which municipalities issue (tax‐exempt) debt to raise funds for private enterprises engaged in qualifying projects. The third group consists of prerefunded or defeased bonds. A muni in this category has been canceled as a municipal liability by the setting aside of sufficient cash and Treasury securities to fund all of its remaining payment obligations. These prerefunded or defeased bonds continue to exist and trade, but with their credit risk improved to that of Treasuries.⁶

Default rates on municipal securities have been very low. From 1970 to 2020, Moody's reports a five‐year cumulative default rate for the sector of 0.08%, which is much lower than the equivalent for corporate bonds, as discussed in Chapter 14. Nevertheless, credit risk is an important consideration for investors in the muni market, with underfunded pension obligations a particularly significant and perennial cause for concern. Historically, GO bonds, which are backed by the issuer's taxing power, were perceived as safer than revenue bonds, which are backed by particular sources of revenue that could diminish or disappear over time. This perception changed significantly, however, with the bankruptcy of Detroit in the summer of 2013. The settlement provisions that emerged in November 2014 were a combination of law, political forces, and negotiations across many interested parties. The results can be roughly summarized as follows. First, while a court ruled that state laws protecting pension benefits were trumped by federal bankruptcy law, Detroit pensions recovered over 95% of their value, although cost of living adjustments were reduced or eliminated. Health and life insurance benefits, however, recovered only 10% of value. Second, holders of water and sewer bonds, which had strong legal claims to the associated revenue streams, suffered no loss of principal. Third, GO bonds suffered significant losses, depending on their exact provisions. Unlimited tax bonds carry a pledge by issuers to raise taxes, if necessary, to pay bondholders. The unlimited tax bonds involved in the Detroit bankruptcy were, in addition, backed by specific, segregated, and voter‐approved tax receipts. Nevertheless, these bonds recovered only 74% of principal, and this negotiated, less‐than‐full recovery was attributed to the dwindling of their dedicated tax receipts due to deteriorating economic conditions in Detroit. Worse off, however, were Detroit's limited tax GO bonds, which had neither a pledge of unlimited tax increases nor dedicated tax receipts. They recovered only 34% of principal. The revelation that GO bonds can be treated like unsecured claims changed credit analysis and pricing in the muni market.

Before the financial crisis of 2007–2009, a majority of muni bonds were insured against default, for a fee, of course, by private insurers. In fact, most of Detroit's GO bonds were insured, which meant that insurers, rather than investors, suffered the losses described in the previous paragraph. In any case, muni insurers suffered massive losses during the financial crisis, not from their muni businesses, but from having insured mortgage‐related products. The ensuing damage to the industry ultimately resulted in less than 5% of new muni issues being insured. More recently, perhaps in part due to the Detroit bankruptcy, and perhaps in part due to the COVID pandemic and shutdowns, there has been a resurgence of muni insurance, rising to perhaps 10% of new issues.⁷

Other US Markets

Mortgages. The second largest sector in Figure 0.4 contains mortgages, at $17.3 trillion. Mortgage loans are used to purchase properties and are collateralized by those same properties. Mortgage balances finance the purchase of one‐ to four‐family residences (70%); commercial property (18%); multi‐family residences (10%); and farms (2%). A remarkable feature about the US mortgage market is that only about 45% of mortgage loan balances are held by the original lenders. The remaining 55% of balances are securitized, that is, sold by the original lenders; packaged into securities; and then sold to investors. Chapter 15 describes this market in much greater detail.⁸

Corporate and Foreign Bonds. The third largest sector in Figure 0.4 comprises corporate and foreign bonds, at $14.7 trillion, which includes bonds sold by US nonfinancial corporations (45%); by US financial corporations (31%); and by foreign corporations to US investors (24%), all to raise money to fund their operations and corporate transactions. Chapter 14 describes this market in detail.

Loans and Advances. Continuing counterclockwise in Figure 0.4, a little less than half of the $9.0 trillion of Loans and Advances are made by banks, and the rest by an assortment of nonfinancial and financial entities. An important feature of this sector is the securitization and trading of bank loans, which is described in Chapter 14. Consumer credit, at $4 trillion, is discussed in the next section, in the context of household balance sheets.

Agency/GSE Debt. The penultimate category of instruments in Figure 0.4 includes “agencies” or “agency bonds,” which are issued by government agencies and GSEs. These entities span a range of associations with the federal government, and their debt issues enjoy varying levels of support from the federal government. At one extreme are the Federal Housing Administration (FHA), the Small Business Administration (SBA), and the Government National Mortgage Association (GNMA). These agencies are part of the government, and their debt issues are backed by the “full faith and credit” of the United States. Moving from the full‐faith‐and‐credit framework, the Tennessee Valley Authority (TVA), which provides electricity for local power companies in Tennessee and surrounding states, is considered a federal agency. Formally, however, it is a corporation that is wholly owned by the US government, and its debt is backed by its own revenues, not – at least explicitly – by the federal government. Further away from full‐faith‐and‐credit are the Federal National Mortgage Association (FNMA) and the Federal Home Loan Mortgage Corporation (FHLMC). They are known as GSEs, but are owned by private shareholders and, having failed during the financial crisis of 2007–2009, are now under government conservatorship. Leading up to the crisis, their debt was not explicitly backed by the government, but the market expectations for full government support of their bonds was fully realized during and after the crisis. These two entities are discussed in great detail in Chapter 15.

Commercial Paper. The final category in Figure 0.4 is Commercial Paper (CP), through which the most highly rated corporations borrow short‐term funds from the public. CP is discussed further in Chapter 14.

Households

Table 0.3 gives the financial assets and liabilities of the household sector.⁹ The financial assets of the sector far exceed its financial liabilities, meaning that the sector, as a whole, has significant net worth. The sector invests significant amounts directly in equity and debt markets, although a greater fraction of assets is invested through intermediaries, like pensions, mutual funds, and the savings components of life insurance policies. A large fraction of financial assets is also held in near cash equivalents, that is, in deposits and shares of money market funds. On the liability side, households borrow mostly through mortgages, which are the subject of Chapter 15, or through consumer credit.

The primary components of consumer credit are home equity loans, auto loans, credit card loans, and student loans. The outstanding credit balances of these sectors over time are depicted in Figure 0.8. Home equity loan balances grew rapidly with the run‐up of housing prices preceding the financial crisis of 2007–2009 and have declined steadily since.¹⁰ Credit card loans also grew going into the crisis, though not as dramatically, and also declined after the crisis, but had then recovered, until declining again with the COVID pandemic and shutdowns. Auto loans, while also declining through the crisis, seemingly emerged as taking the place of declining home equity loans and relatively flat credit card loans. Student loans have increased independently of the economic cycle, which is a policy result: the federal government has been guaranteeing student loans for decades and, since 2010, directly owns all of its student loans. As of March 2021, government student loan balances comprised over 92% of the total.¹¹

Figure 0.9 reports balances that are 90 or more days delinquent, that is, balances on which the borrower has not made payments for 90 or more days. Not surprisingly, home equity delinquencies increased during and for several years after the financial crisis, as falling housing prices made it impossible for many homeowners to recover their outstanding mortgage and loan balances by selling their homes. Home equity delinquencies have since steadily declined, perhaps reflecting more careful underwriting in the aftermath of the crisis. Credit card delinquencies also increased after the crisis, perhaps reflecting weakened consumer balance sheets, but have since declined to pre‐crisis levels. Auto loan delinquencies increased after the crisis as well, then fell, and increased again, perhaps an expected consequence of the rapid growth in balances. Student loan delinquencies have been increasing for many years, which has raised a number of policy concerns, from the perspective of both students and the federal government. Note that the precipitous decline in student loan delinquencies in the second half of 2020 is an artifact of federal COVID forbearance programs.

Nonfinancial Business

Figure 0.10 shows the composition of liabilities for nonfinancial businesses, separated into corporate businesses, which are likely to be relatively large, and noncorporate businesses, which are likely to be relatively small. Corporate businesses, which are more likely to have track records of earnings and established creditworthiness, have broader access to markets. They can sell debt securities to raise 30% of their financial liabilities, while noncorporate businesses sell essentially none. The “Other” source for corporate businesses includes direct investment from abroad, which is not at all a part of small business liabilities. Noncorporate businesses, then, rely mostly on mortgages, for which they need only unencumbered property. The differences in the liability structures of these two groups reflect the life cycle of business borrowings, from “family and friends,” to bank loans, to investor groups, to private placements of debt, and finally – for the largest and most established companies – to public debt securities. Chapter 14 discusses private placements and public debt issues in more detail.

Commercial Banks

Commercial banks accept deposits from their customers, pay them interest, and offer them safety – in part through federal deposit insurance – and immediacy or liquidity, that is, the ability to withdraw funds whenever necessary. Deposits, which are not classified as debt securities or as loans, are not included in Figure 0.7, but constituted 93% of commercial bank liabilities. Other liabilities include investments by the bank's parent company, long‐term debt, commercial paper, assorted loans, and repurchase agreements, or repo, which are loans collateralized by debt securities, as discussed in Chapter 10.

While deposits are certainly the main source of funding for bank investments, they are actually a product or output of banking, just like business loans and mortgages are bank products. Depositors value the safety and immediacy of deposits, and they incorporate deposits into their management of cash and liquidity. Furthermore, banks actively manage the liquidity they offer to depositors, in part by having some fraction of liabilities in longer‐term debt, and in part by investing some fraction of assets in liquid products that can be sold quickly and easily to meet any unexpected withdrawals of deposits.

Turning to assets, Figure 0.11 shows the asset composition of large and small commercial banks. Commercial and industrial (C&I) loans, real estate loans, and consumer loans are all considered the main business of banking. In addition to these assets, however, along the lines of the previous paragraph, banks hold liquid assets. The most liquid, of course, are cash, reserves or deposits at the Federal Reserve, and other money market (MM) instruments. But these assets have the disadvantage of earning very low rates of return. Therefore, to earn higher rates of return while maintaining satisfactory liquidity profiles, banks also hold Treasuries, agency securities, and MBS, which can be sold relatively easily should the need arise.

Figure 0.11 also reveals some differences between large and small banks. First, commercial bank assets are highly concentrated. There are over 4,000 commercial banks in the United States, but $12.5 trillion of the sector's $19 trillion of assets, or 66%, are held by the largest 25 banks.¹² As an aside, the number of banks in the country has been declining gradually but swiftly: there were over 14,000 banks in 1984. This decline is likely an adjustment from historical restrictions on interstate banking and branching that prevented larger banks from satisfying market demand. In any case, a second difference between the largest and smaller banks is the difference in the fractions of their assets in real estate loans: 17.4% for the largest banks and 36.7% for the smaller banks. The concentration of a small bank's assets in real estate loans, which are often local, can challenge the bank's viability through regional economic downturns.

Life Insurance Companies

Life insurance products often pay death benefits, of course, but they are also often savings vehicles, through which policy holders invest funds with the advantages of tax deferral. Life insurance companies are, therefore, financial intermediaries that collect and invest premiums so as to meet their obligations under policies sold and to earn additional returns for their shareholders. Furthermore, long‐term fixed income assets are natural hedges to the long‐term nature of their policy liabilities. Reflecting these considerations, the asset portfolios of life insurance companies contain large fractions of corporate bonds and equities, 36.4% and 8.5%, respectively, with an additional 18.8% in mutual fund shares that are some mix of bonds and equities. In fact, their corporate bond investments make life insurers very significant players in that market: their direct holdings of $3.5 trillion of corporate bonds comprise about 24% of the total $14.7 trillion outstanding. While Treasuries are theoretically useful as a match for long‐term liabilities, they do not earn enough to meet insurer return hurdles. As a result, Treasuries comprise only 2.4% of life insurance company assets. Finally, life insurance companies also use derivatives to achieve their return and hedging objectives.

Pension Funds

Historically, the majority of pensions were defined benefit (DB) plans, in which the sponsor promises to pay retirees according to a formula that depends on the number of years worked, contributions to the plan, salary history, etc. Sponsors collect employee and their own contributions into a pension fund, and then invest the assets of that fund so as to be able to honor promised obligations. A low‐risk strategy combines relatively high contributions to the pension fund with low‐risk investments, while a high‐risk strategy combines relatively low contributions with aggressive investments. In any case, investment risk in DB plans resides with the sponsor, which, in the end, is responsible for paying the promised benefits. For decades now, however, defined contribution (DC) plans have become more important. In these plans, employees and employers contribute to individual employee accounts, and each employee can typically choose among a few investment options. Upon retirement, employee benefits are determined completely by the funds accumulated in their respective accounts. Hence, in DC plans, investment risk resides with employees. Employers can, of course, offer both types of plans or a hybrid of the two types.

Government employees, at the federal, state, and local levels, nearly always have DB plans or an option to participate in a DB plan. In the private sector, however, the trend has been for corporations and other employers to avoid the risks and costs of managing pension funds, that is, to migrate from DB to DC plans. In 1975, there were about 33.0 million participants in private DB plans and 11.5 million in private DC plans. In 2019, the numbers were 32.8 million and 109.1 million, respectively.¹³ Furthermore, corporations are actively shedding DB pension fund risk through pension risk transfer transactions, in which they pay insurance companies to assume their pension liabilities. In any case, for any portfolio manager of a DB plan, long‐term debt instruments naturally hedge the present value of fixed liabilities. On the other hand, allocations to equities hold out the prospect of having to make smaller contributions to the fund. A study of the largest 100 DB plans offered by US public companies in 2020 found that 50% of assets were invested in fixed income, 32% in equity, and 18% in other categories (e.g., real estate, private equity, hedge funds).¹⁴

Money Market Funds

As mentioned earlier, in the context of deposits, there exists significant demand for assets that provide both safety and immediacy. Money market funds, created in the 1970s, were designed to offer safety and immediacy, while paying higher rates than banks at the time were allowed to pay on deposits. Money market funds are divided into three broad categories: government funds, which purchase only short‐term, government‐backed debt; prime funds, which invest predominantly in short‐term, high‐quality corporate debt, like commercial paper; and tax‐exempt funds, which invest in short‐term, high‐quality, tax‐exempt municipal debt. Before changes implemented after the financial crisis of 2007–2009, investors could buy money market fund shares for $1 per share; their money was invested in relatively safe and liquid assets; and, except in extraordinary circumstances, they could sell their shares at any time for $1 per share. More specifically, a fund that i) complied with Securities and Exchange Commission (SEC) rules governing the safety and liquidity of fund investments, known as 2a‐7 rules; and ii) had a portfolio or net asset value (NAV) corresponding to a value per share of between 99.5 cents and $1.005, could offer and redeem shares at a “fixed NAV” or “stable NAV” of $1 per share. But if the value of the fund fell such that the value per share fell below 99.5 cents, the fund would “break the buck” and shares would no longer be redeemed at $1, but rather at a value corresponding to the fund's NAV. Hence, money market fund shares were very similar to bank deposits, but did not have the benefit of an explicit government guarantee, like federal deposit insurance. Instead, money market shareholders had to rely on their fund sponsors or management companies to make up for any NAV shortfalls. Only one money market fund had ever broken the buck, in 1994, but it was to happen a second time during the financial crisis of 2007–2009.

In September 2008, a few days after the failure of the mortgage government‐sponsored enterprises, FNMA and FHLMC, and a few days before the failures of the investment bank, Lehman Brothers, and the insurer, AIG, money market fund investors embarked on a flight‐to‐safety, through which they withdrew huge volumes of cash from prime funds and deposited huge volumes into government funds. Relatively suddenly, investors came to believe that financial entities might not be able to pay off their maturing commercial paper, which comprised a significant part of prime money market fund portfolios. And, in fact, the day after the bankruptcy of Lehman Brothers, the Reserve Primary Fund became the second fund in history to break the buck: the value of its sizable holdings of Lehman Brothers commercial paper had fallen so precipitously that the fund's NAV fell to 97 cents per share.¹⁵ Fearing that investor flight from prime money market funds would exacerbate already stressed conditions in money markets, including the ability of financial entities to continue borrowing in CP markets, the Treasury instituted a program through which it guaranteed money market fund shares for one year, and the Federal Reserve created a facility that extended nonrecourse loans to banks collateralized by asset‐backed commercial paper bought from money market funds.

After the crisis, the SEC changed several rules governing money market funds.¹⁶ First, 2a‐7 rules were tightened to increase the safety and improve the liquidity profiles of money market fund portfolios. Second, institutional prime and tax‐exempt money market funds, as opposed to funds with only retail investors, must allow share price to float with the fund's NAV. Proponents of this change argue that floating NAVs raise awareness that fund values can fluctuate and may discourage withdrawals timed to precede a fund's breaking the buck. Opponents argue that money market fund investors, particularly institutional investors, are well aware of the risks; that floating NAVs have little to no bearing on flights‐to‐safety away from prime funds; and that floating NAVs significantly increase the accounting, operational, tax, and legal complexities of using money market funds for cash management. The third post‐crisis change was that prime and tax‐exempt money market funds had to have the power, under various stress conditions, to impose redemption fees of up to 2% and gates that prevent withdrawals for up to 10 business days in any 90‐day period. Redemption fees, which are paid into the fund, are intended both to discourage investor withdrawals in a crisis and to recover losses from liquidating assets in a crisis to meet those withdrawals. Gates are intended to give funds a grace period in which to manage through stressed market conditions. If, however, a fund cannot restore stability by the end of its grace period, the fund is liquidated. Government funds, by the way, may choose to include the power to impose fees and gates, but very few have done so. While redemption fees and gates are intended to increase fund stability, they might actually encourage earlier, preemptive redemptions. In any case, both the floating NAV rule on institutional prime funds and the inclusion of redemption fees and gates on all prime funds went into effect in October 2016.

Figure 0.12 shows the balances of money market funds over time, by sector. The timing of the steep drops in the balances of both prime and tax‐exempt funds corresponds closely to the October 2016 effective date just mentioned. The inconvenience of floating NAVs and the potential loss of immediacy from redemption fees and gates clearly reduced the attractiveness of prime funds. In 2020, as part of a broad flight‐to‐quality brought on by the COVID pandemic and economic shutdowns, balances in government money market funds increased markedly. Over the whole of 2020, institutional prime fund balances stayed relatively constant at $600 billion, while retail prime fund balances fell gradually from about $500 billion to $200 billion. Prime balances did fall dramatically, however, in March 2020, and institutional balances fell more than retail balances, but, due to swift action by the Treasury and Federal Reserve in support of financial markets in general and money market funds in particular, balances recovered relatively quickly. In any case, it seems that the possibility of redemption fees and gates in March 2020 did encourage preemptive withdrawals by prime investors and sales of assets by prime fund managers, who raised liquidity in order to avoid triggering fees and gates. Consequently, at the time of this writing, the SEC is revisiting fees and gates and considering other changes to the regulation of institutional prime funds.¹⁷

Electronification

Equity and futures markets, which comprise relatively few distinct securities, migrated to electronic trading much more quickly than did bond markets, which comprise a vastly greater number of distinct issues. Traditionally, bonds were traded by voice. Customers traded with dealers, and dealers traded with each other, either directly or anonymously through interdealer brokers (IDBs). While a lot of bond trading is still by voice, electronic trading has been growing steadily. Dealer‐to‐customer (D2C) trades can take place on single‐dealer or multi‐dealer platforms, mostly through request‐for‐quotes (RFQ). In this process, a customer requests dealers to provide price quotes to trade a particular size of a particular security or portfolio of securities. The customer then chooses with which dealer to trade.³⁶ A customer can also receive direct streams from dealers, through which dealers continuously send bid and offer prices for particular securities. The customer can then decide if and when to trade at the streamed prices. Dealer‐to‐dealer (D2D) trades can take place on interdealer electronic platforms, usually through a CLOB. Through a CLOB, traders submit limit orders to buy or sell a security, as described earlier, or they buy or sell that security immediately by lifting the offers or hitting the bids of other dealers who have submitted limit orders. PTFs have become more important in recent years and are discussed presently. PTFs often participate directly on interdealer platforms and also set up direct streams from individual dealers.

In the Treasury market, from February 2019 to May 2020, 65% of all trades were electronic. This percentage breaks up into 33% D2D trades, which nearly all trade electronically, and 67% DTC trades, of which 48% trade electronically. The electronic trades over this time period were 54% through CLOBs, 33% through RFQ, and 13% through streaming. In the corporate bond market, electronic trading of investment‐grade bonds increased from 10% in 2011, to 19% in 2017, and 31% in 2020, and reached 38% in December 2020. Lower but growing percentages of high‐yield bonds trade electronically, from 11% in 2017, to 21% in 2020, and to more than 25% in December 2020. Electronic trading of corporates is further ahead in Europe, growing from 39% of the market in 2017 to 47% in 2020.³⁷

Dealer Regulation and Risk Taking

Dealers traditionally take risk in order to make markets. Customers can buy securities in significant size from dealers, because dealers are willing to bear the risks of buying and holding significant inventories in those securities. Customers can sell securities in significant size to dealers, because dealers are willing to bear the risks of holding and subsequently selling those securities. There are really two distinct risks here: market risk and execution risk. With respect to market risk, a dealer holding an inventory of corporate bonds can pretty easily hedge the risk of general interest rate changes, and even the risk of general changes to credit spreads, but hedging the unique credit risk of individual bonds can be difficult and expensive. With respect to execution risk, a dealer must typically “work out” of a large position. For example, after buying a large block of securities from a customer, the dealer cannot usually sell that whole block at once: the loss from the resulting price impact would likely wipe out any spread or fee that the dealer had charged the customer. Therefore, the dealer usually sells the larger position gradually, in smaller pieces, running the risk that other market participants happen to jump in to sell the same security at the same time, and also extending any idiosyncratic market risks of the position.

Dealers in banking entities were subjected to more stringent regulation after the financial crisis of 2007–2009. In particular, capital requirements were increased, both adjusted for risk and as a raw percentage of assets; liquidity requirements were formally added; and trading purely for the account of the bank, rather than for facilitating customer transactions, was – with the exception of Treasury bond trading – severely restricted by the Volcker rule. Because of this new regulatory regime, and perhaps also because of their own choices to reduce risk, banking entities are less willing to hold securities inventory and less willing to bear the risks of making markets. Figures 0.17 and 0.18 illustrate this point in the US Treasury and corporate bond markets, respectively.

Figure 0.17 shows that the amount of 10‐year Treasury notes sold at auction has been increasing steadily, and particularly rapidly in recent years, while, at the same time, dealers – who are mostly within banking entities – have been purchasing lower and lower percentages of those sales. Both the issue sizes and allotments to dealers, given in the figures by the very light gray lines, vary significantly from month to month, mostly because of the auction cycle of new and reopening sales described previously. To show the trend more clearly, therefore, the solid lines are centered seven‐month averages of the underlying series. In any case, in the last few month of the observation window, the allotment to dealers is well under 15%. Figure 0.18 shows that primary dealers are holding much smaller net positions in corporate bonds than before the crisis, while the total amount of corporate bonds outstanding has increased dramatically.³⁸

Growth of PTFs

In the Treasury market, the reduction of dealer market making has been made up, at least in part, by PTFs. These firms typically trade to make money on their accounts and try to make relatively small amounts of money on each of very many trades. They are often high‐frequency trading firms (HFTs) as well, but do not have to be. In any case, the PTF business model or trading style relies heavily on electronification, and the advance of electronification is in good part due to demand from PTFs. Market participants were shocked in 2015 by an inadvertent leak of the top 10 firms by trading volume on the largest platform for the electronic trading of Treasuries: only two were banks. Figure 0.19 uses more recent data to show how PTFs have grown in prominence in the Treasury market. Over the sample period, PTFs and dealers each comprised 48% of trading in the interdealer broker market (with buy‐side firms, like asset managers, comprising the remainder). Furthermore, true to their usual nature, about 93% of PTF trades were electronic, compared with 58% of dealer trades. Note, by contrast, that only 18% of buy‐side trades were electronic.

Growth of ETFs

Shares of an ETF represent fractional holdings of the fund's underlying portfolio and trade on an exchange. Authorized Participants (APs) of an ETF can create ETF shares by buying underlying assets, adding them to the fund's portfolio, and selling ETF shares against those additional assets. APs can also reduce the number of outstanding ETF shares by taking assets from the underlying portfolio, selling them, and purchasing existing ETF shares corresponding to the sold assets. An ETF has a NAV, usually expressed as a value per share, which is an estimate of the value of the underlying portfolio. However, investors buy or sell shares of an ETF on an exchange at a market‐determined price, which may be greater than, equal to, or less than the NAV. Unlike APs, investors cannot exchange their shares for securities in the underlying portfolio, nor can they deliver securities in exchange for new ETF shares.

ETFs in general, including fixed income ETFs, have experienced phenomenal growth. The first fixed income ETF was listed in 2002, and, by the end of 2020, assets under management in global bond ETFs exceeded $1.5 trillion.³⁹ An important motivation for bond ETFs is the creation of liquid trading vehicles from relatively illiquid assets. As discussed earlier, in part because there are so many distinct corporate bond issues, individual corporate bonds are relatively illiquid, particularly as they age. By contrast, a small number of large and diversified portfolios, traded on exchanges and traded electronically in the form of ETFs, can be very liquid. Furthermore, growing liquidity and electronic trading in ETFs drive more electronic trading of corporate bonds, which further improves ETF liquidity, and so forth. In fact, a market has developed for trading portfolios comprising bonds that are all or mostly all held by ETFs.⁴⁰ In any case, Figure 0.20 illustrates the liquidity advantages of ETFs. For each asset class shown, ETF assets under management (AUM) are a small fraction of total amounts outstanding. At the same time, however, ETF ADV is a large fraction of total ADV. For example, ETFs hold only 4.8% of high‐yield corporate bonds outstanding, but the ADVs of these ETFs comprise 24.3% of the ADV of all such bonds. In other words, high‐yield corporate bond ETFs are significantly more liquid than high‐yield corporate bonds.

Skeptics have argued that day‐to‐day ETF liquidity lulls investors into a false sense of security, in the sense that, under stressed market conditions, ETF liquidity will revert to the liquidity of the underlying bonds and essentially disappear. This theory was tested in March 2020, when markets roiled from the COVID pandemic and economic shutdowns. Underlying bond market liquidity did evaporate, with even primary issues of both investment‐grade and high‐yield bonds coming to a halt. Over the same time, however, ETF trading volumes increased dramatically. In other words, investors could not trade individual bonds, but could trade ETFs. The pricing of ETFs over the period, however, surprised some observers. As bond prices plummeted, ETFs traded at significant discounts from NAV; that is, the prices of the ETFs were very much below the estimated value of their underlying portfolios. Similarly, when bond prices rapidly recovered, ETFs traded at significant premiums to NAV, that is, at prices well above estimated portfolio values. Critics took this pricing behavior as evidence that ETFs were not working well, in the sense that investors could not trade shares at their true values. But by the definition of illiquidity during stress events, “true values” of individual bonds cannot be accurately determined at those times. It is most likely, therefore, that observed discounts from NAV resulted from stale bond prices: as prices swiftly fell with nearly no liquidity, bond prices recorded for the purposes of estimating NAV were stale, higher prices. Similarly, as prices rebounded quickly with little liquidity, NAV was estimated with stale, lower prices, making it seem that ETFs traded at a premium. Supporting this narrative was the fact that when prices were falling and ETFs were selling at a discount to NAV, investors were nevertheless giving bonds to APs to create new ETFs. Despite the apparent loss in value, investors could sell their bonds most efficiently by first exchanging them into ETFs.⁴¹

Growing Concentration of Asset Managers

Concentration in the asset management industry has been growing for some time. Assets under management at the top 20 global managers were 29% of total assets in 1995, 38% in 2000, and 43% in 2020. Furthermore, in 2020, the top 10 global managers managed 31% of total assets, and the top five, 21%. With respect to flows, a 2018 study found that 42% of asset management trading in investment‐grade corporate bonds came from the top five managers.⁴² Increasing concentration in asset management implies a growing demand in the market for larger trades or, equivalently, a growing amount of execution risk from having to break up large orders into smaller transactions.

Summary of Trends

The trends just presented can be combined into the following narrative. Dealers are allocating less risk to market making, while asset managers need to execute larger and larger trades. Therefore, dealers focus on their largest and best clients, pushing others to electronic platforms. The growth of PTFs makes up for some of the liquidity withdrawn by dealers, but by providing frequent liquidity on small trades rather than by assuming the execution risk of larger orders. The net results of these intersecting trends, therefore, are as follows:

• Liquidity has become plentiful for small trades in relatively liquid, electronic markets.
• Liquidity can still be challenging for larger trades, even in relatively liquid markets, and, of course, for trades in less liquid markets.
• More customers must bear the execution risk of their own trades, instead of paying dealers to assume this risk.

Liquidity and Fragility

The changing nature of liquidity in bond markets has raised the issue of market resilience or, its reverse, market fragility. In particular, some believe that liquidity provision by PTFs is less stable than by dealers. The argument is that PTFs will shut down in stress conditions, so as to avoid any losses on their own accounts, while dealers will continue to make markets for their clients, with whom they have ongoing and valuable business relationships. The “flash rally” in the Treasury market on October 15, 2014, provided a case study in which to examine this argument.

On that day, there was a release of retail sales data at 8:30am, but it is generally agreed that the news was not particularly surprising. Nevertheless, Treasury yields dropped relatively steeply over the next hour or so. Then, over the 12 minutes from 9:33am to 9:45am, the yield on the 10‐year Treasury fell by 16 basis points and then rose by 16 basis points (i.e., prices rose by a lot and then fell by a lot). This was an extremely large move for such a short time period – the daily standard deviation of the 10‐year yield was perhaps four or five basis points per day. Was any particular group of market participants to blame for this flash rally? Does the event forewarn of worse occurrences to come?

Subsequent analysis showed that over the 12 minutes in question, trading volume increased dramatically and market depth dropped precipitously. In other words, trading continued throughout the interval, and in large quantities, but through very many small orders and a consistent replenishment of limit orders. Market depth provided by both PTFs and dealers fell during the window. While market depth supplied by PTFs fell by a much larger percentage, PTFs supplied much more of the total throughout. Furthermore, PTFs did not change their bid–ask spreads by much, while dealers did at times. A joint study of regulators concluded that: “In very broad terms…PTFs, as a group, reacted…primarily by reducing limit order quantities, while the bank‐dealers reacted by widening bid–ask spreads and, for brief periods of time, removing their offers to sell securities.”⁴³ Evidence from the flash rally, therefore, is consistent with the different nature of liquidity provision by PTFs and dealers; is not consistent with PTFs closing shop during a stress event; and is consistent with a liquidity regime that is shifting execution risk to the buy side.

There was a similar flash event on February 25, 2021, though this time Treasury prices fell significantly and then recovered in a short amount of time. The event was again characterized by sharply reduced market depth and sharply elevated trading volumes, that is, by rapid replenishment of limit orders.⁴⁴ The extent to which markets are susceptible to similar and perhaps worse occurrences remains a topic of concern.