What If Ireland Defaults? (6 page)

Multiple Equilibria

The ‘second generation' literature explored one possible explanation for the sudden large change in the exchange rate (beyond what can be explained by changes in the shocks to the economy, including new information) and the failure of the exchange rate to equilibrate.

The peso crisis precipitated a massive loss of confidence in the currency and a full-on market panic. Though the importance of confidence is often mentioned, traditional macro models do not include independent variables that quantify confidence. Those that have tried to do so show that confidence can have significant explanatory powers, but few models incorporate confidence in a formal way.

One way of doing so is to assume that there are multiple equilibria.
⁸
Models of multiple equilibria (sunspots) that formally incorporated ‘confidence' suggest that a change in confidence can move the economy from one equilibrium to another. In the case of debt crises, Brazil (and perhaps Greece) provide examples: with low interest rates the country can easily service the debt, so it is rational that interest rates are low; but if interest rates become high, the country cannot service the debt, and it is rational that the interest rate is high to compensate for the risk of default (Greenwald and Stiglitz, 2003). The idea of a self-fulfilling market panic originated in the context of a run on a bank. In Diamond and Dybvig (1983), banks have relatively illiquid assets; in other words, if a bank has to sell assets on short notice it sacrifices a part of the asset value in the process. Every period some customers withdraw money from the bank to meet their spending needs. In a perfect world, all others keep their money in the bank. However, customers are aware that withdrawals will not be honoured if the bank runs out of money (no deposit insurance scheme exists). As they observe other customers' withdrawals, they could decide to take their own money out as well in anticipation of a bank run. Such self-fulfilling panics can leave everybody worse off.

Market Frictions

A third generation of models of crises focused on how various market frictions contribute to the onset of a financial crisis, amplifying (rather than dampening) shocks (for example, Kiyotaki and Moore, 1997; Caballero and Krishnamurthy, 2001; and Mendoza, 2010). Moreover, in these models what would normally be equilibrating adjustments to the shocks can be destabilising. While the importance attributed to specific financial frictions varies from model to model, a common theme in these theories of financial crises is the role of market imperfections in explaining both the fast pace of diffusion and the large extent of amplification of negative economic shocks, providing a recipe for a sudden crash.

Market frictions (information asymmetries, costly state verification, costs of contract enforcement, and bankruptcy (see Greenwald and Stiglitz, 1993a)) limit the extent to which firms can use equity or hedging contracts.
⁹
As a result, firms have to rely on debt, while remaining exposed to risk, and firms act in a risk-averse manner.
¹⁰
Optimal financial structures lead effectively to constraints on debt–equity ratios, so that a decrease in firm equity reduces its ability to borrow. The macroeconomic consequences of these micro imperfections are severe, with investment (including inventory accumulation), for instance, expanding in booms by a multiple of the change in equity (the financial accelerator), and the converse happening in downturns (for example, Greenwald and Stiglitz, 1993a; Bernanke, Gertler and Gilchrist, 1996). Not only are the effects of shocks amplified, but they can persist over time.

Other imperfections in financial markets can similarly trigger crises. Many borrowers face collateral constraints that limit borrowing capacity. Contract enforcement is complicated and lenders have only partial information. A collateral requirement can act both as a selection and incentive device (Stiglitz and Weiss, 1986) and can help manage default risk. For example, in Kiyotaki and Moore (1997) creditors cannot force repayment or seize the borrower's human capital, so borrowers can strategically default on the debt. Collateral-based borrowing constraints tied to the value of the firm's real assets become necessary.
¹¹
As a result, the maximum amount of debt the firm can take on, assuming collateral of a given value, is limited. Even a temporary shock to the value of collateral translates into reduced borrowing ability. Thus a shock sets in motion a feedback effect that decreases investment and the rate of growth for several years. Credit-constrained firms are forced to reduce investment, resulting in further declines in net worth, which in turn lead to tighter borrowing constraints and additional investment cuts.

Greenwald and Stiglitz (1993a) explain how with unindexed debt contracts a macroeconomic shock (for example, monetary policy tightening) that leads to lower than expected prices results in decreased equity, with real effects that are amplified by the financial accelerator. Non-convexities in the relationship between equity and investment also imply that a distributional shock (for example, an increase in the price of oil) has macroeconomic consequences, with the contraction in the losing sector exceeding the expansion in the benefiting sector.
¹²
These financial constraints cause one-time shocks to persist and result in widespread insolvencies.

The banking system itself can amplify especially large downturns. Banks can be viewed as highly leveraged firms (Greenwald and Stiglitz, 2003), so that, when their equity is diminished, they reduce their lending. Institutional features and regulatory design can increase the extent to which this is prevalent. Excessive reliance on capital adequacy requirements can result in a built-in destabiliser; countercyclical prudential regulations or appropriately designed policies of regulatory forbearance may be able to offset the effects (see Helmann, Murdoch and Stiglitz (2000) and the various essays in Griffith-Jones, Ocampo and Stiglitz (2010)). Regulation of maturity and currency mismatches in banks and the firms to which they lend can reduce the vulnerability of the banking system – and thereby the economy – to shocks.

During the Irish financial crisis, property developers facing declining real estate valuations were unable to refinance existing loans or obtain new loans. Asset write downs resulting from losses on property loans constrained the banks' ability to raise new financing, in turn limiting loan provision. Business and consumer credit reductions exerted downward pressure on the rate of new investment and consumption growth.

Other institutional rules and policies (in both home and foreign countries), such as the weakening of automatic stabilisers (for example, safety nets), can make countries more sensitive to shocks. Delegating authority of risk evaluation to rating agencies and imposing constraints on what pensions can invest in can contribute to volatility – a sharp downgrade by the rating agencies (as happened in Thailand in 1997) can precipitate a crisis (see Ferri, Liu and Stiglitz, 1999). In Ireland and other GIIPS (Greece, Ireland, Italy, Portugal and Spain) countries, downgrades of sovereign and bank credit ratings limited capital market access, causing a credit contraction and exacerbating recessionary pressures.

Systemic Crises

In the third generation models just described, financial constraints (operating through collateral requirements, debt–equity constraints or real balance effects), especially in the context of imperfectly indexed debt contracts, can lead to the amplification and persistence of shocks. While research on systemic shocks began well before the Great Recession, the recession has enhanced impetus for this work (see, for example, Haldane, 2009; and Haldane and May, 2011). Greenwald and Stiglitz (2003) and Allen and Gale (2000) describe bankruptcy cascades – how the bankruptcy of one firm can lead to that of others. The extent to which this occurs depends on financial interdependence. Pecuniary externalities arising in the presence of incomplete risk markets and imperfect information imply that the set of privately profitable contracts will not in general be socially optimal (Greenwald and Stiglitz, 1986). In fact, managerial contracts implicitly based on relative performance can lead to excessively correlated risk taking (Nalebuff and Stiglitz, 1983). Moreover, there are strong incentives, especially for large banks, to become excessively interdependent and correlated, so that in bad outcomes they will be bailed out (Stiglitz, 2010a; Acharya and Yorulmazer, 2008; etc.).

Market and Individual Irrationalities

In the original Diamond–Dybvig (1983) model, customers have no information about the bank's default risk. In real life, some depositors could have information about the bank's financial health. However, even when customers are able to assess the bank's financial condition, they sometimes end up ignoring their private knowledge and copying the actions of others, which is known as herding (see, for example, Banerjee, 1992; and Bikhchandani, Hirshleifer and Welch, 1992). As a result, bank runs or sudden market crashes can occur even when only a few investors or depositors possess negative information. Such herding may be rational.

In addition to the rational reasons for herding, many have argued that irrationality plays a crucial role in both the onset and the creation of the conditions for and the spread of financial crises (for example, Stiglitz, 1999b, 2004; and Hirshleifer and Teoh, 2009). For instance, as Kindleberger, Aliber and Solow (2005) note, changes in the sentiment of borrowers and creditors over time can explain the well-known cyclical nature of bank lending. (Such changes in sentiment also play an important role in Minsky cycles and credit crises.) Increases in loan supply can be attributed to optimism in good times, while decreases in credit can be linked to pessimism in bad times. Irrational investor pessimism causes rapid declines in lending, asset prices and exchange rates, typically seen during crises. Investor irrationality can stem from the inability to correctly process available data, compounded by behavioural biases that cause investors to make suboptimal decisions based on the beliefs they have formed (see Barberis and Thaler (2003) for a detailed survey).
¹³
The resulting overreaction to economic news can cause small negative shocks to trigger large-scale market panics that spread across national borders.

When bubbles break (or when panics lead to irrationally depressed prices), there are large real balance effects and the other effects delineated above arising from the financial accelerator, and these can give rise to a macroeconomic crisis. The devastating consequences of a burst housing market bubble have been seen in the recent US and Irish financial crises.

In open economies with firms that have substantial foreign currency debt (with mismatches in the currency and maturity structure of assets and liabilities), large changes in exchange rates similarly can have dramatic effects on equity values or lead to large increases in collateral requirements, precipitating a crisis, for instance, as firms make large cutbacks in investment. During the 1997–1998 Asian financial crisis, firms with foreign currency liabilities and home currency assets were vulnerable to depreciation of the home currency (Stiglitz, 2001; Cespedes, Chang and Velasco, 2004).

In standard dynamic stochastic general equilibrium models the sources of crises are exogenous shocks, but the most important crises involve the breaking of bubbles, most of which can be attributed to internal market dynamics. Housing prices, for instance, rise to the point where further increases are not sustainable given the constraints imposed by the institutional and regulatory system (even with mild forbearance). When home prices can no longer increase at the rate that has been anticipated, demand for housing decreases suddenly with the follow-on effects described above. This pattern, repeated historically, presents a challenge to rational expectations models. There are two possibilities. One is that with short-sighted market participants the economy can evolve in a manner that is consistent with inter-temporal arbitrage equations for a very long time, before a (say, non-negativity) constraint becomes binding (for example, Shell and Stiglitz, 1967). The other is that there is uncertainty about the date of unravelling of the process, and a bubble can then be consistent with rational expectations for an extended period of time (Abreu and Brunnermeier, 2003).

We suspect though that it is challenging to fully reconcile bubbles with perfect rationality. In the US, Irish and most other bubbles (Gurdgiev, Lucey, Mac an Bhaird and Roche-Kelly (2011) discuss the Irish property bubble), large numbers of investors recognised that there was a very high probability of a bubble (and took short bets), even if others believed it was not the case. The question is, why couldn't those who knew better correct the market irrationality? Note that the analysis of such situations requires the construction of models in which individuals have different beliefs, and even as they extract information from the market, they do not converge to the same beliefs. Recent models focusing on the consequences of short sale restrictions for asset bubbles have provided insights, since those who are more optimistic are given more weight during booms than during recessions (for example, Scheinkman and Xiong, 2003). This gives rise to higher market volatility, with real consequences of the kind that we noted earlier in this essay.

In practice, delineating rational and irrational causes of crises can be hard not only because investors face imperfect markets, but also because rationality and irrationality interact: there are rational actors willing to exploit the irrationality of others (and imperfections in the regulatory framework). While standard models assume that such rational exploitation of market irrationality stabilises the economy, in fact that often does not seem to be the case. The crisis of 2008 serves as an example. The lending during the housing bubble illustrates a high level of irrationality on the part of market participants. Incentive distortions led to excessive risk taking in mortgage provision. In the end, it was rational for major institutions to make contracts with each other which amplified risks and made them less transparent, because it ensured (under the assumption of too-big-to fail) large and sustained government subsidies.