Nobel Lecture: Financial Intermediaries and Financial Crises
2023; University of Chicago Press; Volume: 131; Issue: 10 Linguagem: Inglês
10.1086/725793
ISSN1537-534X
Autores Tópico(s)Economic theories and models
ResumoPrevious articleNext article FreeNobel Lecture: Financial Intermediaries and Financial CrisesDouglas W. DiamondDouglas W. DiamondUniversity of Chicago and National Bureau of Economic ResearchPDFPDF PLUSFull Text Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinkedInRedditEmailPrint SectionsMoreI. IntroductionFinancial intermediaries are an important part of the financial system. Intermediaries today include banks, securitization vehicles, and more generally, shadow banks. They stand between investors who want to deploy their savings and borrowers who need funding. While financial markets channel some of this funding directly, even in financially developed economies there remains an important role for financial intermediation. My research on the theory of financial intermediation seeks to understand how financial intermediaries are best structured as well as the benefits and costs of the structure.This lecture has several goals. First, it describes some of my background that influences my interest and my approach to this area. It goes on to provide a nontechnical explanation of my first two models of intermediaries, as well as their implications. I share my current thinking on these models, in light of subsequent financial disruptions and crises. Next, I provide an overview of some of my subsequent research on financial intermediation (especially that with Raghuram Rajan) and compare my research to theories of financial constraints based on the value of collateral. Finally, I describe my research on short-term debt issued by nonfinancial firms.II. BackgroundMy research on financial intermediation began when I was a graduate student at the economics department of Yale University. My goal at that time was to understand why banks exist. The answer turned out to show how and why financial intermediaries can write contracts that allow better outcomes than financing directly. Because the results identify some problems that intermediaries solve and what their contracts do, they can be useful for understanding the effects of regulations on the form of these contracts and how intermediaries have macroeconomic effects.My first exposure to economics was in my third year of high school, in a seminar course on capitalism, where we read Milton Friedman's Capitalism and Freedom (Friedman 1962). In my last year of high school, I helped lead discussions of Paul Samuelson's classic textbook, Economics (Samuelson 1970). Economics made sense to me, but my plan was to become a molecular biologist. I applied to Brown University in part because I thought that it had good access to undergraduate research in science.At Brown, I found that I was more interested in (and better at) economics than in biological science. I took an amazing and transformational course in my senior year. In it, we read one of the best books in economics, A Monetary History of the United States, written by Milton Friedman and Anna Jacobson Schwartz (Friedman and Schwartz 1971). The book provides detailed descriptions of how monetary policy is made and the effects that it has. The discussion that had the biggest impact on me concerned the United States in the 1930s and the interaction between monetary policy, bank failures, and bank runs. Friedman and Schwartz describe the bank failures of the period and ascribe many of them to the Federal Reserve's unexpected unwillingness to serve as a lender of last resort. Bank failures and tight monetary policy led to a reduced supply of money, and this led to a falling price level. They argued that this was the way that bank failures damaged the economy. This fascinating argument made me curious about the financial system and monetary economics. It is interesting to note that this book also helped to form Ben Bernanke's research interests after Stanley Fischer suggested it to him in graduate school. Discussions with Professor Jerome Stein during and after the course helped me understand the process of economic research. Stein was a monetarist like Friedman, but he was a student of James Tobin, a Keynesian. He had the highest respect for both Friedman and Tobin. He encouraged me to go to Yale to work with Tobin.I began to study in the economics department at Yale in 1975. Some ideas from the emerging field of financial economics showed up in the first-year macroeconomics courses. James Tobin put financial markets and intermediaries at the center of his models of the effects of monetary and fiscal policy. I soon met Martin Shubik who was working on a theory of money and financial institutions. He pointed out many unsatisfactory predictions of the competitive market general equilibrium theory, as well as many questions that this theory could not address. Nothing could be more useful to someone looking for a topic to study. Martin's research used noncooperative game theory to model trade with money and price formation as a playable game. Banks were important in payments and the price formation mechanism. Although I never succeeded in adopting his methodology, I learned that one could use game theory to reexamine important applied issues in economics.I learned about progress in financial economics during three summers as a research assistant at the financial studies section of the Federal Reserve Board of Governors. I realized that I could not understand banks unless I had a better understanding of financial markets. After discovering that Yale had no course offerings on financial economics, I arranged to take John Lintner's course at Harvard. Between that course and a careful reading of the detailed lecture notes from Robert Merton's course at the Massachusetts Institute of Technology, I became familiar with a good part of the research in the area. I took my oral field exams in the fields of finance and money, where I first met Steve Ross, who was one of my examiners in finance. He joined the Yale faculty the following fall (1977). I managed to pass the exams. I took Steve's financial economics course that fall and got a much better understanding of financial research. Steve presented one of the first models of optimal contracts between agents and principals in Ross (1973). Steve later became my main advisor and mentor. He taught me how to do research and how to better develop my ideas into transparent models.Much of the theory in finance was real (nonmonetary) and based on perfectly competitive, informationally efficient markets. Corporate finance researchers were trying to overcome the Modigliani and Miller (1958) result on the irrelevance of capital structure. Incentives provision and agency theory had entered the field, along with studies of markets with private information. There was not much theory about financial intermediaries. Black (1975) argued that information revealed in equity market prices made obsolete many historic roles of banks. This motivated me to seek a possible role for banks in the presence of modern and informationally efficient financial markets.III. Why Do We Need a Theory of Financial Intermediation?I started by asking why one should add a layer between borrowers and investors. The answer in a competitive market must be that the extra layer has benefits that exceed its costs. Before the 1980s, the primary models of financial intermediaries assumed that intermediaries reduce transaction costs. A production function for producing financial assets and liabilities was assumed, and intermediaries with lower transaction costs than individuals were analyzed using neoclassical production theory and trade in financial markets (for surveys see Benston and Smith [1976] and Baltensperger [1980]). Some of the results described in this lecture do result in reduced costs, but the goal is to understand why financial intermediation reduces the costs.My approach to understanding the role of financial intermediaries has been to think about the mechanisms they design and the contracts they write, and about how these effect real economic outcomes. These mechanisms go beyond what one can do by trading in financial markets. What good things can intermediaries do, and what other effects do they have? I next explore one approach to answering these questions.IV. Financial Intermediation and Delegated MonitoringFinancial Intermediation and Delegated Monitoring, published as Diamond (1984) but originally from my doctoral dissertation, Diamond (1980), asks the question, If investments in or loans to business borrowers might need monitoring, what is the best way to set up financial contracts? I provide a very simple example of the model, without many details. A more detailed example is in Diamond (1996).Consider a borrower who needs to raise a large quantity of funding. Investors and borrowers are risk neutral, but borrowers have no funds of their own, and each investor's spare funds to invest are small relative to the amount needed to fund the borrower's project. For concreteness, suppose the borrower needs to raise 1, while each investor has 1/m units to invest, implying that a borrower needs to raise capital from at least m investors. I assume that m is very large, say m=10,000. Each investor requires an expected rate of return of 5%. Therefore, the borrower must offer investors as a whole an expected repayment of 1.05. There is uncertainty about how much the borrower can actually repay. The investors know that if they lend 1 to a borrower, the borrower's project will produce either 2 or 1, and the probability is 1/2 for each outcome. The expected value of the cash that the borrower will have is 1.5. Because this exceeds the expected return that investors require, all investors and the borrower agree that the borrower has a profitable, positive net present value project to fund. Because we will see that one needs to provide incentives for the borrower to repay investors and only the borrower will observe how profitable it turns out to be, it is challenging to design a financing arrangement where the investors will be repaid enough to induce them to invest (providing them an expected repayment of at least 1.05 in total).A. Providing Incentives to Repay via Financial ContractsThere are many conflicts of interest in finance, but the most important one involves getting borrowers to repay investors when borrowers may prefer other ways to spend the cash flow that the project produces. They could spend it themselves, on various investments and projects that may or may not be profitable, or they could do more nefarious things with it, such as investing in a shady deal with a brother-in-law. How do we provide incentives to overcome these preexisting conflicts of interest? One way, monitoring, requires acquiring detailed information about the borrower's activities and requires decision making by investors. I defer a discussion of that option for now.The other option to resolve the conflict is to write a contract that imposes a penalty on the borrower who does not repay, or more generally, a contract that sets penalties as a function of how much the borrower repays. This (complete) financial contract specifies in advance the penalties and consequences of each possible payment amount. It depends only on the amount paid and requires no extra information or decision making by investors. I will refer to the penalties as foreclosure, and focus on foreclosure as an example of a penalty, for the following reason. If you do not pay your home mortgage, the lender then has the right to take the house away from you. The lender may not get high recovery from selling the house, but that is only one effect of taking away the house. If you have children, a move might mean they have to go to another school district, which will cause family stress. There may also be stigma from failing to repay. This stress and stigma are penalties to the borrower. The borrower will generally value the loan collateral much more than the lender does, and would be willing to pay more than the lender's value of collateral to avoid foreclosure. To keep this example simple, I assume that the borrower will not repay at all if there will never be a penalty or foreclosure.The investors would like to impose a penalty for low payments to give incentives for higher payments. There are two possible penalties. The investor can liquidate the collateral if the borrower pays too little, preventing the borrower from absconding with it, or the investor can impose a nonmonetary penalty on the borrower.1 Bankruptcy in the world today is some combination of these two actions. In ancient history, nonmonetary penalties, such as debtors' prison and physical penalties, were very common. Such sanctions are now illegal, but the loss of reputation of a borrower is similar to a sanction.Imposing penalties or taking collateral away from its best use is inefficient. Presumably, there was no great joy for an investor in sending anyone to debtors' prison, but the fact that you might face imprisonment served to increase your incentive to repay. The good news about foreclosure is that it makes the borrower repay what is owed, but the bad news is that imposing this penalty is tremendously inefficient. As an investor, you do not want to do it very often, both because it inefficiently hurts the borrower and because it may also reduce what an investor can recover.The optimal financial contract specifies a constant amount to repay, essentially saying, "If you pay investors this amount, there will not be foreclosure. Pay any smaller amount, and there will be." In the example considered here, the optimal way to impose it produces a simple debt contract without any other covenants or terms.2 This type of debt is the optimal financial contract. If the penalty is sufficient, the borrower will repay whenever possible. However, if the borrower owes more than 1 and only has 1, the investors will foreclose. This is bad for the borrower. To keep this simple, I assume that the penalty is sufficiently large that the prospect of imposing it will induce the borrower to repay. This may also be bad for the investors (destroying some of the cash of 1).Let me introduce some notation for readers desiring a bit more precision (feel free to skip this paragraph; I do not use any of this elsewhere). If the borrower has cash of V and pays the investors P, the borrower's payoff is V−P if there is no foreclosure specified for paying P. Any payment that leads to foreclosure gives the borrower a total payoff of B and the investors as a whole a total payoff of I. The borrower will prefer to pay P, if possible, instead of a payment that leads to foreclosure whenever V−P≥B. To keep the example simple, assume that B=0 (foreclosure drives borrower payoff to zero). Foreclosure is inefficient when the sum of investor and borrower payoffs from foreclosure is less than the value of cash available, or I+B 0. The cost of direct and duplicated monitoring of each small loan by each investor is incurred when making the loan. The total cost of monitoring is 10,000K and the expected net expected repayment after monitoring cost is no more than 1.5−10,000K. This must exceed 1.05 for investors to be willing to lend, which requires that K≤0.45/10,000. If monitoring costs more than this it would not be profitable for each investor to monitor their loan. Direct finance with monitoring would be too expensive, so the best possible direct lending contract would be an unmonitored debt contract. But, when a borrower faces this volatile cash flow, although the best direct lending contract for this situation is unmonitored debt, this contract works poorly. Neither type of direct finance allows the borrower to raise funding.Note that if the cost of monitoring were small enough, K≤0.45/10,000, it would allow monitored direct finance, but this would lead to a large duplication of effort. If possible, it would make more sense to delegate the monitoring and incur the cost only one time.E. Delegated MonitoringInstead of having all 10,000 investors monitor their loans, we delegate the monitoring to one person called a banker. As it will turn out, the best contract to provide these incentives is the contract that banks use. That is, the bank contract structure is an optimal way to provide incentives for delegated monitoring. This is the main insight of Diamond (1984).If monitoring of the firm and its cash flows is delegated to a banker, the borrower and the banker will know what was monitored. This monitoring will ensure that the borrower repays the banker. The question is, how do we get the bank to repay the investors? The banker can see, for example, whether the borrower and the brother-in-law are doing a deal, but the investors still cannot see that. Imagine that the banker, the brother-in-law, and the borrower get together and decide to say, "Let's just tell them we only have 1. We're not going to pay them more than that." The investors will only know how much the bank repays them, not the actual amount the borrower could pay to the bank. Without monitoring, we had a conflict of interest between the borrower and investors. With delegated monitoring, we inserted the banker to resolve the problem with the borrower, but it creates an added conflict of interest with the investors.Just as in the earlier example with direct finance, there is a way to resolve this added conflict of interest. In that case, the borrower had to repay the investors, and if the borrower paid too little, there was a foreclosure. In this case, the bank writes a debt contract that specifies that if it pays too little, the default forces foreclosure on the bank. This threat of bank failure turns out to be a great incentive for the bank to monitor borrowers and to repay investors.The optimal contract between the bank and investors (in this context, depositors) to provide incentives for delegated monitoring is a debt contract issued by the bank promising to pay the depositors. The bank will fail if it defaults on its contract and pays too little. The bank gets zero if it fails. This forces the bank to pay depositors. In addition, if the bank does not monitor, it will not have enough money to pay investors, and there will be a default. The bank will monitor as long as the value of its residual inside equity claim (a bonus pool from what is left after collecting loans and paying deposits) exceeds the cost of monitoring.For this contract to provide incentives and be cost effective regarding the delegated monitoring, the bank has to be large and diversified. To understand why diversification is beneficial, consider this extreme example of perfect diversification. Suppose the bank makes many loans, and assume that exactly half of the borrowers' projects will produce a return of 2 and the other half will produce 1. There is no uncertainty about the amount the average borrower is going to pay, and the amount of cash that the bank could collect from the loans will always be up to 1.5. What investors do not know is which borrowers will have 2 and which borrowers will have 1, but it turns out that they do not need to know that.Investors require a 5% expected rate of return, and this implies that the bank must promise depositors collectively at least 1.05 per borrower. The bank can always collect up to 1.5 per borrower, so any promised payment to depositors above 1.05 and below 1.5 per borrower can be made with certainty and will attract deposits. The promised payment from the borrower to the bank must be set a bit above 1.05 to leave a sufficient bonus payment to the banker to provide incentives to monitor and to avoid defaulting on the deposits. The bank will never fail. In this best case, bank diversification means foreclosure will never be needed, but will only be a threat. The bank issues perfectly safe deposits.This is perfect diversification, but in the real world and Diamond (1984), banks have imperfect diversification. Banks are going to fail sometimes—maybe because they are small or exposed to a common source of risk such as the prospects of single industry, or maybe because they are very large but still extremely exposed to macroeconomic aggregate risks.It is important that banks diversify and limit their exposure to aggregate risks. Banks can hedge some aggregate risks in markets. Banks might not have a sufficiently strong incentive to stay diversified and hedged if no one could see the choice. Because banks do not (or cannot) commit contractually to stay diversified and hedge aggregate risks; bank supervisors and regulators need to make sure they do both in order to make banks safe and efficient. Even if banks could contractually commit to diversification and hedging, there is an added role for supervision if there are negative external effects of their failure on other parts of the financial system. There has been a focus on diversification and hedging in supervision and regulation, but since the 2008–2009 financial crisis, stress tests have explicitly looked at the aggregate risks on bank balance sheets. They measure how exposed a financial institution is to large aggregate macroeconomic shocks that might bring down a diversified bank or even the financial system. In addition to measuring and limiting these risks, enforcing substantial diversification within a large bank is essential for banking to work well.This contract structure provides optimal incentives for delegated monitoring using diversification of assets and issuing debt claims to investors. A modern name for it is pooling and tranching. Pooling is just another name for diversification: take a large number of imperfectly correlated loans and put their returns into a pool from which to pay claims to investors. Tranching refers in this context to giving investors senior claims, and having the banker retain some junior (inside) claims on the diversified pool to incentivize monitoring and payment. This structure also resolves many other financial conflicts of interest, such as the costly screening out of bad loans at the origination stage. DeMarzo (2005) presents a more general model of this. Most securitizations of financial assets use this contract structure, replicating this structure first used by banks.As mentioned before, diversification does not work perfectly to eliminate risk, particularly when asset returns become highly correlated. We saw this in the housing crisis of 2008–2009: when house prices went down worldwide and in the United States in particular, the correlation between mortgage defaults increased dramatically. The diversification effect from pooling almost disappeared, and senior mortgage-backed securities became, unexpectedly, much riskier.Nonetheless, in most situations, implementing delegated monitoring with pooling and tranching is an important and widely used financial technology. It provides incentives for bankers to take actions to increase the repayments of loans they make. These loans represent the intermediary's assets. Section V describes a contracting technology that improves the characteristics of an intermediary's liabilities.Bernanke (1983) presents empirical evidence that bank failures in the 1930s led to adverse effects on subsequent access to credit. His interpretation of this finding is that the failures removed institutions with the ability to distinguish good and bad borrowers. A related interpretation is that a given bank's failure would destroy information. This would make it difficult for the bank's existing borrowers to get monitored finance, forcing some of them to raise costly direct finance or lose all access to finance.F. The Choice between Bank Loans and Directly Placed DebtA borrower with a sufficiently profitable project with low volatility of cash can use the threat of foreclosure to commit to repaying debt without frequently incurring the costs of foreclosure. Such a borrower would prefer direct finance to bank loans. There is a related point based on lending dynamics. New borrowers without good reputations will borrow from banks, whereas those that survive and get better credit ratings will issue directly placed debt.Diamond (1991b) considers bank monitoring that can resolve an added conflict of interest where borrowers with limited liability may prefer risky projects (which is sometimes referred to as risk shifting; see Fama and Miller [1972] and Jensen and Meckling [1976]). Young borrowers without a good reputation have poor incentives; they would prefer to choose excessively risky projects if not monitored, and must borrow from banks. Some of these borrowers always have risky projects, even when monitored. A borrower who repays monitored bank loans for a long time without default acquires a reputation consistent with not always choosing risky projects and this reputation would be lost on default. Once this reputation is acquired, it deters the borrower from choosing r
Referência(s)