In the Wake of the Crisis: Leading Economists Reassess Economic Policy

11. Optimal Financial Intermediation: Why More Isn’t Always Better

International Monetary Fund
Published Date:
March 2012
  • ShareShare
Show Summary Details
Adair Turner

How should the global economic crisis that began in 2008 affect our views about financial intermediation? This is a fundamental issue. We need to ask questions about the value of financial intermediation, about optimal levels of financial intermediation, and about reliance on free-market forces to select the optimal level and precise mix of financial intermediation.1

Over the last thirty years, financial intensity has grown remarkably, with increases in real-sector leverage but even more dramatic increases in financial-sector balance sheets as a percentage of gross domestic product, increases in trading volumes as a percentage of GDP, and the financial innovations of the derivatives market (figure 11.1). Before the crisis, the dominant conventional wisdom assumed and explicitly stated that this growth was beneficial because

Figure 11.1.Measures of increasing financial intensity.

  • it would increase allocative efficiency since increased financial intensity completed more markets and because increased market liquidity ensured more efficient price discovery, and

  • it would increase financial stability since risk would be dispersed more efficiently into the balance sheets of those best placed to manage it.

This conventional wisdom was strongly asserted, for instance, by the International Monetary Fund (2006) .

The second half of the conventional wisdom proved wrong, and we need to understand why. There are two broad schools of thought:

  • The first assumes that problems are essentially those of market imperfections, opacity, and perverse incentives. It seeks to identify the particular problems that prevented the system from reaching an efficient stable equilibrium and to put them right. This can be called the microstructuralist school.

  • The second believes that the drivers of instability are deeper than those amenable to increased transparency and the reform of incentives, and focuses on macroprudential oversight and policy response, including on a discretionary basis. This can be called the macro-Minsky school.

I comment on the relative merits of these two schools, express some preference for the latter, and support action to address incentives and structures as necessary but not sufficient.

The key problem that we are trying to solve is not the direct fiscal cost of the public rescue of otherwise failing banks. Although this is a key focus of popular outrage, the direct fiscal cost (as IMF figures show) is the small change of the macroeconomic harm produced by bank failure (table 11.1). In some countries, it could turn out to be negative. Public authorities in total may make a profit from the combination of equity injections, debt guarantees, and central-bank operations.

Table 11.1.International Monetary Fund estimates of public-support costs in the 2008 to 2009 financial crisis
Percentage of GDP
PledgedUtilizedRecoveryNet direct cost
Advanced economies6.
Emerging economies0.80.30.3
Source: International Monetary Fund, “A Fair and Substantial Contribution by the Financial Sector” (June 2010).
Source: International Monetary Fund, “A Fair and Substantial Contribution by the Financial Sector” (June 2010).

Instead, the essential problem is the supply of credit, which first was provided in excessive quantity and at too low a price in a self-reinforcing cycle with asset prices (particularly real estate) and then was constricted, driving destructive and deflationary processes of overrapid deleveraging (figure 11.2). A key measure of the success of public-policy responses to the crisis is therefore whether they will reduce the amplitude of that cycle.

Figure 11.2.Growth in lending and nominal income: United Kingdom, 2005 to 2010.

The structuralist school believes that if markets were made more efficient, and in particular if incentives were better aligned, then booms and busts would naturally be constrained. The core policy is therefore to fix the too-big-to-fail problem. The assumption is that bankers and traders did excessively risky things—in both the banking and shadow banking systems—because they knew that they enjoyed the put option of limited liability.

We must certainly ensure the resolvability of too-big-to-fail banks—enabling us to impose losses on debt holders. This means smoothly turning debt claims into equity claims when necessary (figure 11.3). This is a necessary part of the reform process, and in circumstances where we face the future idiosyncratic failure of a large bank (the future equivalent of a Continental Illinois failure), it would also be a sufficient response.

Figure 11.3.Mechanisms to increase loss absorption capacity and market discipline.

But turning debt claims into equity claims is a sufficient response to future problems of systemic instability (and of the possible simultaneous, interconnected, and self-reinforcing failure of large banks or multiple small banks) only if we can assume three things—that bank debt instruments will be held by unleveraged, nonmaturity-transforming investors, that these investors are capable of taking losses without producing knock-on systemic effects, and that they can take losses without collectively acting in a way that generates self-reinforcing fire sales and a downward asset-price cycle (table 11.2).

Table 11.2.Too big to fail: Conditions for success of debt-based solutions
Required to address the idiosyncratic failure of systemically important financial institutions:
Large enough percentage of balance sheet junior to depositors Subject to write-down or conversion to equity in smooth process Long enough in maturity to avoid precrisis runs
Additionally required to address systemic risks of multiple failures: Held outside the banking system and by investors able to take losses without
knock-on asset-first sale or debt-default consequences

These conditions would apply axiomatically if we can assume that investors always are foresightful and fully rational in their decisions—if they always consider the full range of future possible contingencies. But recent papers (such as Gennaioli, Shleifer, and Vishny 2010) challenge that assumption, arguing that many debt investors operate according to a model of local thinking in which during the good years they ignore the existence of the down tail of the distribution of possible results, essentially assuming that objectively risky instruments are close to risk free (figure 11.4).

Figure 11.4.Frequency distribution of bank bond payouts.

As a result, the financial system—particularly if it is intense, complex, and innovative—is capable of generating an excessive quantity of debt instruments and a quantity of apparently risk-free instruments greater than can be objectively risk free, given the fundamental risk and indeed Knightian uncertainty inherently present in the real economy. When initial problems emerge, investors and depositors bring the down tail of the distribution into their consciousness and decision-making processes, generating self-reinforcing downward cycles of confidence, liquidity, asset prices, and credit supply.

Shleifer’s analysis fits well with what actually occurred in 2000 to 2007 to 2009 (figure 11.5). For instance, the credit default swap spreads of major banks fell to historically low levels in June 2007, immediately ahead of the crisis; the discovered market price first provided us with no useful forewarning of impending problems before the crisis but then swung to excessive overreaction.

Figure 11.5.Financial firms’ credit default swaps (CDSs) and share prices. Source: Moody‘ KMV; Financial Services Authority calculations.

Shleifer’s analysis means that both incentives and myopia can be problems. If that is the case, then fixing incentives is highly desirable but insufficient to ensure stability. The implication of this in relation to the too-big-to-fail debate is that we should strongly prefer solutions that increase the equity ratios of large systemically important financial institutions (SIFIs) because only with equity instruments can we be reasonably certain that the instruments will be held by investors that are able to take losses without knock-on systemic consequences. More generally, this illustrates that the core issues of financial stability are (1) the balance within the economy between debt and equity contracts and (2) the aggregate maturity transformation that the financial system is in total performing.

An economy has equity and debt contracts. Debt contracts, like fixed-wage contracts, respond to people’s desire for apparent certainty in income flow. Not all contracts express equity partner shares in the underlying value added that economic projects produce (Einaudi 2006). But debt instruments introduce into the economy important potential rigidities, irreversibilities, and procyclical tendencies (table 11.3). These arise from a combination of the institutions of bankruptcy, the possibility of fire sales, the need for debt to be rolled over continually, the existence of multiple equilibria depending on an endogenously determined creditrisk spread, and self-reinforcing credit and asset-price cycles.

Table 11.3.Rigidities and vulnerabilities of debt contracts
• Bankruptcy costs: nonsmooth adjustment
• Fire-sale costs
• Need for continual rollover
• Multiple equilibria, depending on interest rate
• Credit and asset-price cycles

Together these mean that cycles of irrational exuberance in debt markets are inherently more dangerous than equivalent cycles in equity markets. The Internet equity-price boom and bust of 1995 to 2000 to 2002 produced large individual wealth gains and losses but little macroeconomic harm. Debt cycles, as both IMF analysis and Reinhart and Rogoff (2009) have shown, produce far greater harm. Therefore, the total extent of leverage in the real economy and in the financial system is an important macrovariable.

But also important is the aggregate degree of maturity transformation—the extent to which the financial system in total, whether through bank balance sheets or via liquid markets, enables the nonfinancial sector to hold financial assets of shorter-term maturity than liabilities. This vital and socially value-creative transformation function produces a yield structure of interest rates more favorable to long-term investment than would otherwise exist. But it is also inherently risky.

Measuring aggregate financial system maturity transformation is therefore vital, but its difficulty increases as the financial intermediation system becomes more complex, interconnected, and multistage. For several decades before the crisis, aggregate maturity transformation increased. Households, for instance, accumulated far more long-term liabilities (mortgages), but the household and corporate assets that funded these were predominantly short term—a development that, fatally, we failed to understand.

In the future, we need to monitor aggregate leverage and aggregate maturity transformation, and we cannot assume that the financial system, left to itself (even with better incentives in place), will select the optimal levels of leverage and maturity transformation. We need policy instruments that are designed to influence those key macro stability parameters.

Levels are important, with high levels of either leverage or maturity transformation creating vulnerability, but changes in levels (that is, cycles) are also important. It is unclear whether any set of constant rules can be relied on to limit occasional excess cyclicality. Also necessary, therefore, is a degree of countercyclical discretion for leaning against credit and asset-price cycles.

So I am arguing that the second half of the precrisis conventional wisdom—that increased financial intensity, complexity, and innovation would ensure stability—was wrong. Empirically, it was proved so, and the theoretical reasons why it was wrong can be identified.

But where does this leave the first question, about whether increased financial intensity and complexity deliver allocative efficiency benefits? I cannot address this important question in this chapter, but I would like to finish with two points.

First, we cannot assume axiomatically that increasing financial intensity produces valuable allocative efficiency benefits. A wealth of theory suggests that financial intensity can be rent extracting rather than value creative and that any beneficial effect of increasing financial intensity in terms of allocative efficiency must be subject to declining marginal returns.

Second, the answer to this question has implications for financial stability policy. Many of the measures that could be taken to increase stability—such as higher capital requirements against trading activities or against intrafinancial system claims—might reduce the scale of trading activity and the liquidity of some markets. If these activities and related liquidity are value creative (at the social level), then we may need to make a tradeoff between stability and allocative efficiency. If they are zero sum or rent extracting, then there is no such tradeoff. The less certain we are that increased financial activity delivers improved allocative efficiency, the more radical we can be in the pursuit of stability-oriented reforms.


These comments present a summary of conclusions that were set out in more detail in a lecture that was delivered at Clare College, Cambridge: “Reforming Finance: Are We Being Radical Enough?,” available at

    Other Resources Citing This Publication