Neglected risks, financial innovation, and financial fragility

https://doi.org/10.1016/j.jfineco.2011.05.005Get rights and content

Abstract

We present a standard model of financial innovation, in which intermediaries engineer securities with cash flows that investors seek, but modify two assumptions. First, investors (and possibly intermediaries) neglect certain unlikely risks. Second, investors demand securities with safe cash flows. Financial intermediaries cater to these preferences and beliefs by engineering securities perceived to be safe but exposed to neglected risks. Because the risks are neglected, security issuance is excessive. As investors eventually recognize these risks, they fly back to the safety of traditional securities and markets become fragile, even without leverage, precisely because the volume of new claims is excessive.

Highlights

► We present a new model of shadow banking with several results. ► Growth of shadow banking is driven by growth in outside wealth. ► Banks hold AAA-rated assets as collateral for riskless debt finance. ► Bank assets and leverage are procyclical. ► Banks retain systematic risk as they diversify idiosyncratic risk. ► When investors neglect tail downside risk, banks are vulnerable to massive losses and liquidity dryups.

Introduction

Many recent episodes of financial innovation share a common narrative. It begins with a strong demand from investors for a particular, often safe, pattern of cash flows. Some traditional securities available in the market offer this pattern, but investors demand more (so prices are high). In response to demand, financial intermediaries create new securities offering the sought after pattern of cash flows, usually by carving them out of existing projects or other securities that are more risky. By virtue of diversification, tranching, insurance, and other forms of financial engineering, the new securities are believed by the investors, and often by the intermediaries themselves, to be good substitutes for the traditional ones and are consequently issued and bought in great volumes. At some point, news reveals that the new securities are vulnerable to some unattended risks and, in particular, are not good substitutes for the traditional securities. Both investors and intermediaries are surprised by the news, and investors sell these false substitutes, moving back to the traditional securities that have the cash flows they seek. As investors fly to safety, financial institutions are stuck holding the supply of the new securities (or, worse yet, having to dump them in a fire sale because they are leveraged). The prices of traditional securities rise while those of the new ones fall sharply.

A notorious recent example of this narrative is securitization of mortgages during the 2000s. Various macroeconomic events, including sharp reductions in government debt during the Clinton administration and massive demand for safe US assets by foreigners, created a shortage of safe fixed income securities. By pooling and tranching mortgages and other loans, financial institutions engineered AAA-rated mortgage backed securities (MBS) as substitutes for US government bonds. The perception that these securities were safe, apparently shared by both buyers and intermediaries who engineered them, was justified by historically low default rates on mortgages in the US and by more or less continuously growing home prices since World War II. Trillions of dollars of mortgage- and other asset-backed securities were created and sold to investors.

Both the holders of these securities and financial intermediaries appeared to be caught by surprise in the summer of 2007, when the news that AAA-rated securities were not safe hit the market. It is not that investors failed to understand that home prices could decline and mortgages could default. Yet two things came as rather substantial surprises. The first was how fast home prices declined and defaults grew. Gerardi, Lehnert, Sherlund, and Willen (2008) show that few, if any, Wall Street professionals expected the housing bubble to deflate so rapidly. The second surprise was the sensitivity of the prices of AAA-rated securities engineered from mortgages, especially collateralized debt obligations (CDOs), to home prices, a phenomenon largely overlooked by the models utilized by rating agencies (Jarrow et al., 2007, Coval et al., 2009). As these securities were downgraded, prices fell and new issuance stopped. The losses from MBS spread through the financial system, precipitating the market collapse in September 2008.

This recent episode is far from unique in recent US financial history. In the 1980s, investment banks began selling collateralized mortgage obligations (CMOs), securities created out of mortgage portfolios and intended to substitute for government bonds. To avoid a possible risk to the value of CMOs resulting from mortgage prepayments by homeowners (which would occur if interest rates fell and people refinanced their homes) and consequent prepayments on the high-yielding bonds, intermediaries engineered CMOs nearly invulnerable to prepayment risk if historical patterns continued. In the early 1990s, however, as the Federal Reserve sharply cut interest rates, prepayments skyrocketed to levels unprecedented by historical standards, so even the prices of CMOs most protected against prepayment risk declined sharply. The investors were caught by surprise and dumped the CMOs, turning back to government bonds (Carroll and Lappen, 1994). Financial intermediaries were evidently caught by surprise as well, and many (particularly those who sold prepayment insurance) suffered substantial losses. Like the collapse of home prices in 2007–2009, massive prepayments appear to have been unanticipated by the market.

A similar narrative describes what happened to money market funds in 2008. The industry was originally created to offer investors a substitute for bank deposits, with slightly higher returns, instant liquidity, and no risk. Because investment in money market funds was not protected by deposit insurance, however, these funds were engineered never to “break the buck”, that is, have their value per share drop below $1. To slightly raise returns, prime money market funds invested in generally safe nongovernment securities, such as commercial paper. The collapse of Lehman Brothers in September 2008 led to its default on commercial paper, which caused one large holder of that paper, the Reserve Fund, to break the buck (Kacperczyk and Schnabl, 2010). This event shocked investors and precipitated hundreds of billions of dollars in withdrawals not just from the Reserve Fund, but also from the whole prime money market fund sector, and a return to traditional bank deposits and government securities only funds (Pozsar, Adrian, Ashcraft, and Boesky, 2010). Only government guarantees of prime money market funds saved the industry.

In this paper, we present a model that captures some key elements of this narrative. The model shares with the traditional accounts of financial innovation, such as Ross (1976) and Allen and Gale (1994), the view that innovation is driven by investor demand for particular cash flow patterns. This demand allows intermediaries to profitably engineer these patterns out of other cash flows. We add two new assumptions to this standard story.

First, we assume that both investors and financial intermediaries do not attend to certain improbable risks when trading the new securities. This assumption captures what we take to be the central feature of the historical episodes we describe: the neglect of potentially huge defaults in the housing bubble and of the sensitivity of AAA-rated securities to these defaults, the neglect of the possibility of massive prepayments in the early 1990s, or the neglect of the possibility that a money market fund can break the buck. We model the neglect of certain states of the world using the idea of local thinking, introduced by Gennaioli and Shleifer (2010), which is a formalization of the notion that not all contingencies are represented in the decision maker's thought process. The neglect of some states of the world in models used to price CDOs is a good example.

Second, we make the preferred habitat assumption that investors have a very strong preference for safe cash flow patterns. We model this assumption through preferences, namely, infinite risk aversion, but it can reflect psychological or institutional characteristics of demand. An alternative way to model such demand might be to consider investors who have lexicographic preferences with respect to particular characteristics of investments (e.g., AAA ratings). Yet another approach might be to stress regulatory requirements imposed on investors such as banks and insurance companies that favor safe assets. This assumption on demand is not strictly necessary for our results but makes them much stronger.1 We have obtained similar results in a model with finite risk aversion and Epstein-Zin preferences.

We then examine a standard model modified by these two assumptions and obtain three main results. First, as in the standard model, there is room for financial innovation to offer investors cash flow streams that are not available from traditional securities in sufficient supply. However, when some risks are neglected, securities are over-issued relative to what would be possible under rational expectations. The reason is that neglected risks need not be laid off on intermediaries or other parties when manufacturing new securities. Investors thus end up bearing risk without recognizing that they are doing so.

Second, markets in new securities are fragile. A small piece of news that brings to investors' minds the previously unattended risks catches them by surprise and causes them to drastically revise their valuations of new securities and to sell them. The problem occurs precisely because new securities have been over-issued: there are not enough cash flows in the neglected states of the world to make promised payments in full. When investors realize that the new securities are false substitutes for the traditional ones, they fly to safety, dumping these securities on the market and buying the truly safe ones.2

Third, in equilibrium financial intermediaries buy back many of the new securities. But their aggregate wealth might be much smaller than that of investors, which limits their ability to absorb the huge supply of the new securities. As a consequence, the prices of these false substitutes fall sharply, even without traditional fire sales due to leverage (Shleifer and Vishny, 1992). Prices of traditional securities rise as investors fly to safety.

The model thus delivers the basic patterns of financial innovation and financial fragility in a new way. The most important contribution is to connect financial innovation, the glut of new securities, surprise about risk, and corresponding financial fragility through a unified model of belief formation. We show that a model in the spirit of Allen and Gale (1994), even modified by a preferred habitat formulation of preferences but without neglect of certain risks, can deliver some aspects of the narrative, but not over-issuance and the fragility it entails. Without a deviation from rational expectations, one cannot get the basic idea of false substitutes: securities investors believe to be riskless turn out to be risky.

Our model of financial innovation is related to the behavioral finance idea of security issuance catering to investor demand as in Baker and Wurgler (2002) and Greenwood, Hanson, and Stein (2010). Also related is the idea that consumers ignore some attributes of products they buy (Gabaix and Laibson, 2006). Henderson and Pearson (2010) study equity derivative products called SPARQS, which they argue are introduced to capitalize on investor misunderstanding of equity payoff patterns. Shleifer and Vishny (2010) apply the idea of catering to the financial crisis, but they simply assume optimism as the stimulus for security issuance and pessimism as the shock precipitating a crisis. Here we present a unified model of belief formation that accounts for the whole story.

Our paper is also related to an important theme in the literature on financial fragility, namely that both banks and the shadow banking system create private money or liquidity that investors demand (Gorton and Metrick, forthcoming, Stein, 2010). Such creation of liquidity is usually seen as socially valuable, but entailing systemic risks due to leverage and resulting asset fire sales (Shleifer and Vishny, 2010, Stein, 2010). While we recognize the benefits of financial innovation, we take a more skeptical view about the social value of liquidity creation when investors neglect certain risks. In such a system, security issuance can be excessive and lead to fragility and welfare losses, even in the absence of leverage. In this respect, our paper is closer to Rajan's (2006) prescient analysis of the risks of financial innovation, although we emphasize neglect of unlikely events leading to over-issuance of securities instead of incentive problems as a source of instability.

In the next section, we present a benchmark rational expectations model of financial innovation in a pure exchange economy. Section 3 modifies this model to allow for local thinking and derives our main results on financial innovation and fragility. In Section 4, we study a production economy, in which innovation under local thinking can lead to investment distortions. In Section 5, we discuss welfare in both the exchange and the production economies. In the exchange economy, innovation under local thinking could benefit intermediaries and harm investors; in a production economy, because innovation distorts investment, it can leave everyone worse off. Section 6 examines the case of fully rational intermediaries dealing with locally thinking investors. Section 7 discusses some implications of our work for the recent financial crisis. All proofs are collected in the appendix.

Section snippets

The model

There are three dates t=0, 1, 2 and two assets, B and A, which pay off at t=2. The assets stand for cash flows from projects. Asset B pays R>1 for sure. Asset A pays yi with probability πi, where i=g (for growth), d (for downturn), r (for recession). We assume

A.1

yg>1>yd>yr and πg>πd>πr.

Growth is the most likely outcome, and a downturn is more likely than a recession.

There is a representative intermediary, who is patient and risk neutral. At t=0, the intermediary owns both assets and sells claims

Financial innovation under local thinking

We consider departures from rational expectations due to agents' limited ability to represent uncertainty. To do so, we follow the Gennaioli and Shleifer (2010) model of local thinking. That model, which provides a unified explanation of several anomalies in judgments but admits Bayesian rationality as a special case, builds on the notion that agents' inferences are made on the basis of a selected subset of possible events, not on the entire state space. Intuitively, not all states of the world

Innovation and local thinking in a production economy

Suppose that instead of owning assets, the intermediaries have exclusive access to production technologies (or activities) B and A. Activity B yields R at t=2 for any unit invested at t=0. The return of activity A is stochastic, equal to yi with probability πi, where as before i=g, d, r. The riskless activity is in limited supply, in the sense that investment IB in activity B cannot exceed one. Investment IA in activity A is in principle unbounded.

The intermediary has initial wealth wint<1 but

Welfare analysis

Section 2 showed that under rational expectations financial innovation is socially beneficial. It boosts intermediaries' profits while leaving investors' welfare unchanged.8

Rational intermediaries

We have assumed so far that intermediaries and investors share the same incorrect beliefs. We now show that the false substitutes effect holds even if the intermediaries hold rational expectations. Rationality of the intermediaries introduces two changes into our previous setting. First, the intermediaries evaluate returns correctly, which influences their investment and issuance decisions. Second, intermediaries could try to profit from the possible drop in prices of the new securities by

Discussion

We propose a new approach to modeling financial markets, one that emphasizes the central role of the neglect of low probability risks in accounting for the nature of financial innovation and financial fragility. We motivate the model using several examples in which the neglect of risks appears important, including the recent financial crisis. In conclusion, it might be useful to return to the crisis and to explain some of the ways in which our model offers a novel perspective on the events.

We

References (31)

  • N. Barberis et al.

    A model of investor sentiment

    Journal of Financial Economics

    (1998)
  • E. Dekel et al.

    Recent developments in modeling unforeseen contingencies

    European Economic Review

    (1998)
  • A. Shleifer et al.

    Unstable banking

    Journal of Financial Economics

    (2010)
  • V. Acharya et al.

    Manufacturing tail risk: a perspective on the financial crisis 2007–2009

    Foundations and Trends in Finance

    (2010)
  • F. Allen et al.

    Financial Innovation and Risk Sharing

    (1994)
  • M. Baker et al.

    Market timing and capital structure

    Journal of Finance

    (2002)
  • R. Caballero et al.

    Collective risk management in a flight to quality episode

    Journal of Finance

    (2008)
  • M. Carroll et al.

    Mortgage-backed mayhem

    Institutional Investor

    (1994)
  • Joshua Coval et al.

    The economics of structured finance

    Journal of Economic Perspectives

    (2009)
  • P. Demarzo et al.

    A liquidity-based model of security design

    Econometrica

    (1999)
  • Diamond, D., Rajan, R. Fear of fire sales and the credit freeze. Quarterly Journal of Economics 126 (2),...
  • K. French et al.

    The Squam Lake Report: Fixing the Financial System

    (2010)
  • X. Gabaix et al.

    Shrouded attributes, consumer myopia, and information suppression in competitive markets

    Quarterly Journal of Economics

    (2006)
  • N. Gennaioli et al.

    What comes to mind

    Quarterly Journal of Economics

    (2010)
  • K. Gerardi et al.

    Making sense of the subprime crisis

    Brookings Papers on Economic Activity

    (2008)
  • Cited by (437)

    • The climate-sovereign debt doom loop: what does the literature suggest?

      2024, Current Opinion in Environmental Sustainability
    • Banking supervision with loopholes

      2024, European Economic Review
    View all citing articles on Scopus

    We are grateful to Pedro Bordalo, Robin Greenwood, Sam Hanson, Anil Kashyap, Brock Mendel, Vladimir Mukharlyamov, Adriano Rampini, Michael Rashes, Joshua Schwartzstein, Jeremy Stein, seminar participants at the Harvard Business School, National Bureau of Economic Research, Stern School, and Chicago Booth, as well as the editor and two referees for helpful comments. Gennaioli thanks the Barcelona GSE Research Network and the European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013)/ERC Grant agreement no. 241114 for financial support.

    View full text