At the heart of Adam Smith’s admonition that society should strive to be just is the concept of debt not purely in a financial sense but also in a philosophical and moral sense. Every economic transaction involves an exchange of value between two parties. But it also involves an exchange of promises. Two parties undertake obligations to each other. One party offers something to another party, and this gives rise to an obligation by the other party to give something back. Smith was seeking to outline the types of institutional arrangements that would ensure that such exchanges are fair to both parties, or as fair as possible within the exigencies of human society. He was asking the question: What do human beings owe each other? What is a debt? What is an obligation? How do we balance the scales between people as evenly as possible?

Promises Aren’t What They Used to Be

Any discussion of the modern financial system, the disarray into which it was thrown in 2008, and the threats it faces after years of failed post-crisis policies must confront the overwhelming role that debt plays in virtually all financial arrangements. Among the most significant changes in the global economy in the three decades leading up to the financial crisis—particularly in the United States and other Western economies—was the incredible increase in indebtedness at all levels of society. Between 1980 and 2008, the share of household and consumer debt alone increased from 100 percent of the U.S. GDP to 173 percent, an increase of approximately $6 trillion.¹ The debt balloon expanded until it literally burst. The growth of debt financing and the increasing substitution of debt for equity in corporate capital structures and personal balance sheets became the gravamen of our age. This debt explosion accelerated after the crisis with total global debt reaching $199 trillion in 2014 according to the McKinsey Global Institute.² We cannot understand our economy or our society until we understand how central debt is both economically and culturally. In order to properly discuss debt, we need to understand the essential role that promissory arrangements play in every financial instrument or transaction known to man.

Every financial instrument involves some type of promise. Whether it is called a stock or a bond, equity or debt, an insurance policy or a pension trust, an option or a futures contract, every financial instrument involves a payment or promise of payment of money in exchange for a promise of future receipts. Sometimes that promise of future payment is contingent, as in the case of an equity security. In other cases, the promise is contractually certain, as in the case of a debt obligation like a bond or a bank loan, or in the unhappy case of a life insurance contract. The primary difference between debt and equity is the degree of certainty and the time horizon (fixed or uncertain) over which that the promise will be kept.

Any type of promise implies a belief in the future. Promises are by their nature a sign of optimism. Equity promises are more contingent than debt promises, so perhaps they denote the greatest degree of optimism of all financial instruments. In a leveraged capital structure, all of the debt is effectively a form of equity since there is a significant risk that it will not be repaid. Equity demands a higher return than debt because of this high degree of uncertainty regarding repayment. The forms that our financial promises assume say a great deal about our expectations about the future, not only economically, but philosophically, psychologically, and culturally.

In recent years, the traditional differences between debt and equity became blurred as leverage became an increasingly dominant component of the economy’s and corporations’ capital structures. Financial innovation created new forms of capital such as new types of preferred stock and convertible debt that combine the attributes of debt and equity. Many derivative instruments also combine debt and equity features. High yield bonds (corporate debt obligations rated less than investment grade by the major rating agencies) are another type of security that combine the attributes of debt and equity. The high yield bond market exploded in size since its founding in the 1980s. Today, these bonds constitute a significant component of the capital structures of many U.S. and European corporations (and post-crisis are becoming more common in Asia). While they are considered debt, however, high yield bonds are not really debt at all. Instead, they are in reality a hybrid debt/equity security that is best described as “equity with a coupon” that poses a high risk of default.

This conflation of debt and equity is one reason why traditional correlations between asset classes are breaking down. While traditional portfolio theory teaches that the returns on bonds and stocks should not correlate in certain types of economic environments, recent experience demonstrates that the returns on these different asset classes in fact correlate very strongly, particularly in extreme market conditions. This should not be surprising in view of the fact that these asset classes increasingly resemble each other in their constituent parts. The different debt and equity layers of a leveraged capital structure—which describes the vast majority of corporations in the United States and Europe—should logically trade in tandem as the financial condition of the underlying business improves or deteriorates. There is no rational reason why a leveraged company’s debt and equity should react any differently to positive or negative news about the business’s prospects. The fact that the stock of a bankrupt company can trade above zero while its debt trades well below par value remains one of the abiding anomalies of the financial markets, one that is best explained by the adage that markets don’t have to work in theory but have no choice but to work in practice.

As the forms of debt and equity have changed, so have the guises in which financial promises travel. First, as noted above, debt rather than equity became the predominant form of capital in circulation. Second, the structures in which debt appears became increasingly complex and opaque and organized in a manner that ruptured the relationship between borrowers and lenders. These changes radically transformed the economy on both a local and global level and served as both cause and consequence of profound changes in our values and expectations.

The fact that debt came to play such a dominant role in our economy says a great deal about our values and beliefs as a society. By choosing to use debt as the dominant medium of exchange, we are telling each other that we trust each other. We are saying that we will keep our promises. We are confirming that our words have common meanings, that we speak a common language, and that we follow a common set of rules. More than anything, we are also affirming our belief in the future. We are telling each other that the world will continue to prosper and grow and that we will do everything reasonably necessary to ensure that happens. Most important, we are counting on the fact that future conditions will be sufficiently robust to allow this debt to be repaid.

But the world keeps minting new debt to an amount that it can never reasonably hope to repay. In view of the inconceivably large volume of global debt that exists in 2016, the underlying code of trust on which debt is based is likely to be called into question in the years ahead, posing serious threats not only to economic but to social and cultural stability. At the heart of every debt obligation lie two promises: a lender promises to lend money to a borrower, and a borrower promises to repay that money (with interest) to the lender. In today’s highly complex economy, modern debt obligations such as bonds or bank loans include an additional web of detailed legal promises known as “covenants” that govern the behavior of the borrower and the lender. These covenants require the borrower to conduct its business in a certain way as a condition for accepting the loan. For example, the borrower agrees not to incur additional debt, or not to pay dividends to its shareholders, or agrees to make regularly scheduled financial reports to the borrower. As long as the borrower keeps these promises, the lender promises to stand aside and allow the borrower to manage its affairs as it sees fit.

In 2008, Martin Wolf, the chief economic commentator for the Financial Times, wrote that “the central feature of the financial system... [is that] it is a pyramid of promises—often promises of long or even indefinite duration. This makes it remarkable that sophisticated financial systems exist.... Promises may not be kept.”³ Continuing, Wolf described the trillions of dollars of outstanding financial assets as “promises of future, of often contingent, receipts in return for current payment.... As the financial system grows more complex, it piles promises upon promises.”⁴ Promises are by their very nature uncertain in their fulfillment. Promises also contain a temporal element—they are executory in nature. As such, they are subject to the contingencies of time and human nature.

When we make a promise to another person, that person is depending on the fact that nothing will change (or change sufficiently) in the intervening time to alter our commitment and ability to keep our word. Yet the world is filled with changing circumstances that may affect our ability or willingness to keep our promises. As all of us know, the only certainty in life is change. So promises are, by their nature, highly contingent. In many respects, making a promise or accepting a promise is a great leap of faith. It denotes a commitment by the party making the promise, and trust by the party receiving the promise. In the global economy, trillions of promises big and small are made every day. While these promises are being made, circumstances both internal and external to the parties making them are changing, sometimes radically. Few of these changing circumstances make it easier for promises to be kept. Most of the time, reality is working to give people reasons or excuses to break, or amend, their promises. It is no small miracle that so many promises are ultimately kept. Large and complex financial transactions involve such a large number of complicated promises (both internal, such as the commitments of the parties, and external, such as the multifarious laws and regulations that must be satisfied) that it is miraculous that so many transactions get consummated at all.

The Digitalization of Promises

In recent years, the forms of debt, and therefore the character of our promises, have been drastically altered by the application of computer technology and advanced mathematics to traditional financial instruments such as bonds, mortgage loans, and corporate loans. The digitalization of financial information that was made possible by the computer chip ushered in a revolution in finance that profoundly altered the relationship between lenders and borrowers. The primary change it wrought was severing the personal connection that traditionally existed between the two parties to a loan. Instead of going down to the local bank and obtaining a mortgage from a banker with whom he or she has a personal relationship, today’s homeowner enters a transaction with a faceless corporation. This impersonal corporate entity finances the most important financial transaction the borrower will likely enter into in his lifetime and is granted enormous power over the borrower’s financial future. In effect, this regime creates a system of debts without promises, which in many respects is an oxymoron, or a system of impersonal promises, which is also a contradiction in terms. A recognition that many of the obligations created by modern finance are empty promises will help us to better understand the nature of the promises economic actors are now expected to keep, why they are harder to keep (or easier to break) than promises made in earlier times, and why this renders the financial sector increasingly unstable and vulnerable to periodic crises.

Today, all financial data is capable of being digitalized. When data is digitalized, it is reduced to 1s and 0s. This means that every financial instrument is reduced to the same basic constituent parts. The differences between various types of debt obligations, such as mortgages, automobile loans, or bank loans, are effectively erased by this process. The ramifications of this transformation of different financial instruments into their constituent elements are truly revolutionary. On a practical level, it became possible to analyze, manipulate, and stress-test voluminous amounts of financial data in relatively short periods of time. Lending decisions that used to be based, at least in part, on a personal relationship between a lender and a borrower instead came to rely on computer-based underwriting systems that substituted credit scores for human judgment. Among other things, this led to profound intellectual errors involving the use of flawed financial models that failed to consider whether the data was being tested against proper benchmarks.

The digitalization of information made possible the phenomenon that came to be known as “securitization.” This process involved the bundling of hundreds or thousands of individual mortgages into special purpose entities that could then be sold to institutional investors. Securitization dramatically increased the distance between individual borrowers and their lenders. Securitizations were effected through the formation of special purpose investment vehicles (i.e., corporations or limited liabilities companies) that were normally formed in a tax-favored jurisdiction like the Cayman Islands. The capital structures of these entities were divided into different pieces (Wall Street adopted the French word tranche to describe these different pieces) that were piled on top of each other and given descending credit ratings from AAA to BB with an unrated bottom tranche of equity.

Figure 4.1 compares the traditional model of credit (on the top of the figure) with the much more complex securitized model of credit (on the bottom of the figure). By the time the financial engineers were done, we sure weren’t in Kansas anymore.

Collateralized Mortgage Obligations

In the case of mortgages, these products were called collateralized mortgage obligations (CMOs). The theory behind CMOs was that a geographically diversified pool of mortgages would have a low risk of default because real estate is local in nature and heavily influenced by local economic conditions. Moreover, these entities were considered to be overcollateralized in the sense that there was believed to be more aggregate collateral (more mortgages) than necessary to repay each of the rated tranches and to produce an attractive return to the equity (bottom) tranche.

Both of these assumptions turned out to be woefully wrong. First, while real estate is local, the sale of the ultimate debt to financial institutions that are linked together in global markets eradicated the local nature of the underlying investment. Creators of CMOs should have been looking at the capital sources (i.e., the buyers of the debt issued by the CMOs), not the underlying borrowers, in seeking the protection ostensibly provided by diversification.⁵ Second, the underlying mortgages turned out—particularly with respect to subprime and Alt-A borrowers—to be worth far less than their face amount during the historic housing market collapse that began in 2006. Credit insurance contracts written on these types of products rendered insurance giant AIG insolvent and in need of a government bailout in 2008.

Compared to the traditional model of credit (a simple lender/ borrower relationship), the securitized model of credit introduced an enormous amount of complexity into the mix. To illustrate just how complex these concoctions became, we will first look at an example of a basic CMO and then jump to a more complicated real-world example that should make readers’ heads spin.

Figure 4.2 shows the structure of a basic CMO. Very few if any such structures exist in the real world today, although in the early years of the market deals generally had a limited number of tranches like the one in the figure. CMOs are generally designed so that payments on the underlying mortgages are applied sequentially from the top tranche to the bottom tranche through the life of the deal. The bottom tranche is unrated and considered equity because it assumes the first risk of loss when mortgages in the pool default.

One of Wall Street’s basic business principles is to introduce complexity into its products so it can obscure what it is really selling. CMOs, however, set new standards for complexity (topped only when credit default swaps came along and began to be used to provide insurance on the different tranches of CMOs, a topic discussed in Chapter 6). From the perspective of the manager of a CMO, these products often function like Rubik’s Cubes due to the complex web of covenants with which they must comply. Every time the manager wants to buy or sell a new mortgage, he must run that mortgage through a complex model to insure that all of the multiple covenants governing the CMO remain in compliance.

To give readers a flavor of how complex these instruments became, Figure 4.3 shows the structure of an actual CMO issued in 1994 that issued 17 different tranches of debt.

In view of the fact that Figure 4.3 was typical of the actual types of instruments that regulators were trying to decipher in the midst of the financial crisis, it is a miracle that the system survived at all. This complexity makes it extremely doubtful that regulators can even begin to understand these types of investments before being able to determine whether they might cause systemic threats. It required several years to unwind these structures after the crisis (and many of them were still being unwound as of early 2016).

One issue with respect to all of these increasingly complex derivative concoctions is the degree of separation between the underlying borrower and the ultimate lender. As the forms of these contracts became increasingly convoluted, the relationship between borrower and lender became increasingly attenuated. Promisor and promisee (borrower and lender) basically had no relationship with each other but were instead connected through a chain of contracts that removed any meaningful promissory connection from their relationship. In really exotic structures, like some of the ones that brought insurance giant AIG to its knees, credit default swaps were tied to individual tranches of collateralized debt obligations (CDOs) that themselves consisted of pools of underlying obligations in other CDOs (these are known as CDO-squareds). In such structures, the distance between the lender and ultimate borrower was so attenuated that they might as well have been circulating in different galaxies.

The distance between the individual mortgage borrower and the ultimate investor in a CMO is illustrated in Figure 4.4.

Figure 4.4 is intended to show the structure of a typical transaction. The process begins when a group of mortgages are packaged together, which occurs on the far left hand side of the diagram. Individual borrowers are aggregated into a pool of borrowers by the originator, who then sells this portfolio to a special purpose vehicle, which is the CMO itself. The CMO then sells rated tranches of debt to investors in order to fund the purchase of these assets. In the real world, these two steps basically occur simultaneously. The original borrower is found all the way on the far left-hand side of the diagram, and the ultimate lender is found all the way on the right side. We are a long way from going to the local bank to get a loan from your friendly neighborhood banker. Figure 4.4 illustrates why CMOs and other types of CDOs are the ultimate fetish instrument (borrowing from Marx’s terminology). The actual CDO tranches that are sold to investors are completely untethered from the underlying human and economic obligations on which they are based, and create an enormous distance between the ultimate lenders and the actual borrowers who must repay the loans.

This raises important questions about how to determine the value of these instruments, because one of the key points about fetish instruments is that they obscure rather than reveal the relationships that support them. As discussed in Chapter 2, money is already one step removed from the commodities whose value it represents. As money assumes increasingly complex forms, its relationship with these commodities becomes increasingly complicated and obscured. The value of a stock or bond, which are just two of the virtually unlimited forms of capital in circulation today, is determined by a complex group of factors because these securities are representations of complex underlying economic relationships. As the forms of money grow increasingly complex, it becomes more and more difficult to determine their value. By the time we reach complex derivative instruments, which are sophisticated legal contracts with many moving parts, the determination of value is almost forbiddingly difficult and requires as much art as mathematics. The liquidity function, which is the ultimate arbiter of value in market economies, is strongly influenced by the form that money assumes. For this reason, complex derivatives are necessarily far less liquid than more simple forms of money like stocks and bonds.

CMOs were constructed using several assumptions that were expected to work in theory but failed miserably in practice. The first assumption was that diversification of the underlying portfolio by property type and location would minimize losses based on the belief that these varying property types would not behave the same way (in terms of price) in an economic downturn. In point of fact, virtually all real estate prices correlated downward—sharply downward—when the real estate bubble burst beginning in 2006. Why did this happen? This is where things got interesting. As noted earlier, the very fact that so many properties became securitized rendered obsolete the age-old real estate adage that “all real estate is local.” Securitization rendered all real estate, regardless of its physical location, local, because its ultimate ownership became vested in CMO debt tranches owned by institutional investors. These investors were spread throughout the global financial system rather than located inside local banks whose financial fortunes were closely tied to the local communities near the properties. The money that made it possible to bid up real estate prices in local markets in states like Florida, Arizona, and California no longer originated in those states but instead came from financial centers in Europe and Asia (and even, remarkably, Iceland) where the rated tranches of CMOs were sold. The globalization of the real estate market was the logical outcome of the sundering of the promissory relationship between lender and borrower, but the ramifications were far more profound than a borrower’s inability to maintain a personal relationship with his or her lender. Rather than spreading risk, securitization concentrated it among a group of electronically linked investors subject to herd-like behavior.

The second erroneous assumption used to promote CMOs was the belief that a man’s house is his castle, that he will do anything to defend it (in modern times this has come to mean he will do anything to prevent being kicked out of it by his lender), and it is therefore a sound credit risk. In fact, just the opposite was true. As the economist Robert Shiller points out, “[a] home represents a highly leveraged exposure to a single, stationary plot of real estate—about the riskiest asset one can imagine.”⁶ Moreover, the riskiness of this asset was increased when lending standards were tossed out the window by subprime lenders who engaged in some of the most reckless lending practices one can imagine. For example, NINJA loans were extended to borrowers with no jobs, no income, and no assets; other loans were extended in amounts that exceeded the value of the underlying properties. The thought that combining hundreds or thousands of highly risky assets into one big pile of risky assets would somehow reduce overall risk actually should have been highly counterintuitive. Yet this was precisely the conclusion that was reached by teams of highly educated mathematicians who apparently were so caught up in their theories and equations that they forgot to apply any common sense to their work. One can concede, for the sake of argument, that it is a rare individual who possesses both the advanced mathematical talents required to design and analyze complex financial instruments and deep knowledge of the financial markets in which those instruments will be traded. But the firms that sold hundreds of billions of dollars of these products certainly possessed inside their walls the combination of talents that should have been brought to bear on the validity of the basic assumptions underlying the CMO financial models. But instead of acting as a series of checks and balances, the different parts of these firms appeared to reinforce reckless behavior rather than rein it in.

A third flawed assumption involved the wholesale dependence on FICO scores to measure the creditworthiness of borrowers. Moody’s Investors Service and Standard & Poor’s, the two agencies whose ratings were required for CMOs to sell debt to investors, used FICO scores to evaluate the creditworthiness of the underlying borrowers in these deals. FICO scores are credit rating scores that are generated by Fair Isaac Corp. of Minneapolis, Minnesota, that evaluate a person’s creditworthiness. The rating agencies (as well as underwriters and investors) failed to take into account the fact that FICO scores had not been in existence during previous recessions. As a result, these scores were incapable of providing accurate predictions of consumer behavior during a sharp housing downturn. Rating agencies interpreted data without any reference to historical context. By accepting past data at its face value and failing to adjust it for changes in economic conditions, the rating agencies ended up issuing wildly flawed ratings on hundreds of billions of dollars of CMOs. The rest, as they say, is history, but history as nightmare.

HSBC Drinks the Mortgage Kool-Aid

The folly of this type of thinking was illustrated by what happened when London-based HSBC Holdings PLC, one of the world’s largest banks with operations in 76 countries and territories, joined the subprime party with its 2003 acquisition of Household International, Inc. Household was a large subprime lender based in Prospect Heights, Illinois, the heart of the United States. Less than four years later, in February 2007, the 142-year-old British banking giant announced that it was adding $1.7 billion to its loan reserves to account for losses in its subprime mortgage portfolio. But that was only the beginning. Before the subprime crisis was over, HSBC would suffer many more billions of dollars of losses from the work of its clever finance doctorates.

The business that HSBC acquired from Household focused on second lien loans, sometimes known as “piggyback loans.” These loans allowed a buyer to combine a bigger mortgage from a first mortgagor with a second lien from a second lender into an amount that often exceeded 100 percent of the purchase price of a home. In the event of default, the second lien holder would only be paid off after the first mortgage was satisfied. For this reason, second liens paid higher interest rates than first liens and were ostensibly more attractive to some lenders who believed they were capable of evaluating the risk. HSBC believed it was one such lender. It was wrong.

Shortly after its purchase of Household, HSBC’s then Chief Executive William Aldinger (in a comment he surely came to regret) bragged that the bank employed 150 PhDs skilled at modeling credit risk. He didn’t define what he meant by “risk,” and clearly neither did his hard-working PhDs, because it turned out that they clearly lacked the expertise required to properly analyze subprime credit and default probabilities. By the end of the first quarter of 2009, HSBC’s subprime losses reached $8.3 billion and were still running. Management admitted that the write-downs were not over either. At that point, it did not require a doctorate in mathematics to calculate the loss at an astonishing $55.3 million per PhD (assuming any of the original 150 were still around and not been fired).

The HSBC saga illustrated the risk of undue reliance on financial models. Among the problems involved in analyzing HSBC’s portfolio, according to The Wall Street Journal, was an absence of data on loans to subprime borrowers making small or nonexistent down payments (i.e., borrowers with no equity in their homes who found it relatively easy to walk away from their loans). It turned out that HSBC was relying heavily on FICO scores. Prior to this, FICO scores had never been tested against a downturn in the housing market or against second lien loans, rendering them of limited predictive utility. Douglas Flint, HSBC’s finance director, told investors that “what is clear now is that FICO scores are less effective or ineffective” when lenders are granting loans in an unusually low interest rate environment. It turns out that using FICO scores in such an environment is akin to using peak earnings to calculate and then project a corporation’s future earnings, or applying the decade’s lowest default rates to project future corporate bond default rates (all errors that were made by major financial industry players at various times and were being committed again by many with respect to loans to the energy industry after the financial crisis).

Actually, if you think about it, it shouldn’t have required a single doctorate, not to mention 150 of them, to figure out that different credit tools were needed to build and monitor subprime loan portfolios. As Peter Bernstein noted in his book on risk, “Likeness to truth is not the same as truth. Without any theoretical structure to explain why patterns seem to repeat themselves across time or across systems, these innovations provide little assurance that today’s signals will trigger tomorrow’s events. We are left with only the subtle sequences of data that the enormous power of the computer can reveal. Thus, forecasting tools based on nonlinear models or on computer gymnastics are subject to many of the same hurdles that stand in the way of conventional probability theory: the raw material of the model is the data of the past.”⁷ The markets would see this error repeated many times over the next couple of years or, to be more precise, they would see the consequences of an error that was being committed incessantly in the financial world throughout the 1990s and 2000s materialize in frightening dimensions in the 2007–2008 time frame. And then, once the smoke cleared after that crisis, markets would see similar errors committed all over again as an epic, central bank-sponsored bull market began in March 2009 and created a massive credit bubble in energy and commodities that began to collapse in mid-2014.

A Fetish Is Not a Promise

Securitization eradicated the individual identity of borrowers and substituted a broad-based credit rating system on which investors came to rely in determining whether to purchase CDOs. It no longer mattered whether a borrower was an individual purchasing a home or a car, a corporation using the money to build a new plant or to finance the acquisition of a competitor, or a private equity firm paying too much to buy a company in order to generate transaction fees for its general partners. The individual borrower came to have absolutely no identity or meaning to the ultimate lender. There was absolutely no relationship between borrower and lender.

The securitization of financial assets such as mortgages or bank loans represented a modern manifestation of what Karl Marx termed the “fetish character of commodities,” the commodity in this case being the mortgage or bank loan itself. Marx’s description of the lack of relationship between the underlying economic commodity and the form in which it is traded in the marketplace (which was discussed in more detail in Chapter 2) perfectly describes a collateralized debt obligation. Any relationship between lender and borrower in such a product is completely eradicated. A collateralized debt obligation is the ultimate fetish instrument. The distance between the individual mortgage holder, the economic actor generating cash flows in the form of mortgage payments, and the lender, is about as attenuated as possible. This effectively obliterates the promissory character of the original mortgage and leaves room for all kinds of mischief that can interrupt the flow of funds and prevent repayment of the underlying debt. In more blatant situations, courts in the United States have prevented loans from being foreclosed on due to the inability of the collateralized loan obligation to prove ownership of the underlying loan. In those cases, the entity lost track of the deed to the home!⁸ Instruments that were designed to reduce risk by slicing and dicing it and reallocating it among parties who could pick and choose their favorite flavor instead ended up choking on a poisonous buffet.

Parties that have no relationship are in no position to make promises to each other. Financial instruments, even the debts packaged into large pools and then resold to investors in securitized form, still involve formal but depersonalized promises between the parties that created them. Adam Smith’s view that the regard for the opinion of others leads them to act morally and responsibly can play no role in a system in which personal relationships have been sundered. Changing the form of these instruments devalued the promises that stand behind them. It also left the global economy holding sacks of empty promises when the debts came due.