[1] The author is chief economist and senior vice president of the World Bank, on leave from Stanford University. The views expressed are solely those of the author and do not necessarily represent those of any organization with which he is or has been affiliated. This paper is based on joint research with Bruce Greenwald.
[2] Thus, this theory differs from real business cycle theory, which views fluctuations as efficient responses to outside disturbances. Here, the hypothesis is that the economy responds to large shocks in ways which amplify their effects and make them persistent. Credit and equity rationing—lack of access to capital and of efficient risk distribution systems—play a central role in this theory.
[4] World Bank data and Consensus Forecasts. Note that the 16 percent figure is the current consensus forecast; measured relative to the pre-crisis consensus forecast for 1998, this represents a fall from trend of over 23 percent of GDP.
[5] World Bank (1998), p.105. Even in Thailand, where the crisis was far less severe, real wages fell by more than 10 percent between mid-1997 and mid-1998. Changes in product prices, e.g., in manufacturing relative to other prices, mean that real product wages may not have fallen by this magnitude, but the data did make clear that the hypothesis of price and wage rigidity is hard to maintain.
[6] In a sense, traditional economic theory has focused on one relative price, the price of goods relative to money, both as the cause of economic disturbances and as the mechanism for restoration towards full employment (the real balance effect). We argue that disturbances in other relative prices may play a larger role in inducing large economic fluctuations.
[7] There is a large literature on asset price bubbles, and the problems posed in particular by the absence of markets extending infinitely far into the future. For early examples, see Hahn (1966), Samuelson (1967), and Shell and Stiglitz (1967).
[8] World Development Indicators CD-ROM (World Bank 1998).
[10] Greenwald and Stiglitz (1990).
[11] See Greenwald and Stiglitz (1989, 1995). The failure to take these factors into account is one of the important limitations in the menu-cost literature (Mankiw 1985, Ball, Mankiw, and Romer 1988, Akerlof and Yellen 1985). If, for instance, a demand curve shifts, then either price or quantity has to be adjusted. While there may be some costs of adjusting prices, failure to adjust prices forces large adjustments in quantities. Traditional theories suggested that the latter were far larger than the former.
[12] This is, of course, the central insight of the efficiency wage literature. For surveys, see Akerlof and Yellen (1986) and Stiglitz (1987b). Signaling and selection effects may give rise to rigidities of product markets as well. For instance, lowering prices may convey a signal that the firm has no incentive to produce high quality products (to maintain its reputation); customers thus infer that the effective price—price per unit quality—may be lower, and hence the demand for the commodity may actually be reduced. (Stiglitz 1987b, 1989).
[13] See Edlin and Stiglitz (1995).
[14] The fact that deviations from norms convey information can result in the persistence of seemingly irrational behavior; for instance, it has long been noted that dividends represent an inefficient way of transferring money from the corporate to the household sector, resulting in unnecessarily high tax payments (the dividend paradox). The most convincing explanation of the dividend paradox is that deviations from the standard policy (to, say, a system with lower total tax payments) convey information that market participants do not know how to process appropriately, leading to adverse effects on market value.
[15] It is easy to show that there may be multiple equilibria: with the same "state" variables, the economy may exhibit two or more configurations.
[16] One of the explanations for the failure of exports to respond to the devaluations in the East Asia crisis is that customers became more anxious about firm bankruptcy, so that the East Asian firms became viewed as less reliable suppliers. Any action that conveys information about the likelihood of a bankruptcy thus can have an adverse effect on the demand curve facing a firm.
[17] Other explanations for wage rigidities have been provided; especially convincing is the insider-outsider model developed by Lindbeck and Snower (1988). In most of these models, there is considerable uncertainty about the consequences of wage changes, partly because their full impact depends on strategic interactions with the behavior of others.
[19] There are also option values associated with quantity adjustments. Fixed costs associated with hiring workers leads, in the presence of uncertainty, to the postponement of hiring decisions (an extension of the argument presented in Greenwald and Stiglitz 1995), and the fact that there are fixed costs associated with workers moving from one firm to another may lead to a slowness of workers in accepting lower wages. (Greenwald and Stiglitz 1987) The arguments of this and the previous section are related: some of the costs are associated with the publicly observable signal ("stigma") associated with accepting a low-wage job.
[20] Greenwald and Stiglitz (1993).
[21] As noted above, a substantial majority of Indonesia’s firms may be bankrupt. While a devaluation would also have led to an erosion of net worth of firms that had taken an exposed foreign exchange position, it is now clear that, at least for Thailand, the marginal death rate associated with devaluation would have been small (export firms would have gained in product sales what they lost in terms of increases in the Bhat value of liabilities, while the real estate firms were bust in any case, as a result of the collapse of the real estate boom). In any case, as anticipated by models focusing on the impact of bankruptcy and declines in economic activity on capital flows had predicted, the increases in interest rates did not arrest the decline in exchange rates. (See Furman and Stiglitz (forthcoming) and World Bank 1998.)
[22] In general, lower net worth interferes with economic efficiency. Risk-averse firms that are trying to avoid bankruptcy may take on too little risk, while firms (including banks) facing imminent bankruptcy may "gamble on resurrection," actually taking on excessive risk. See Kane (1987) and Stiglitz and Weiss (1992).
[23] Greenwald and Stiglitz (1990b).
[24] By the same token, policies that focus on the adverse consequences to those with foreign exchange exposure implicitly bail-out gamblers, and reinforce concerns about moral hazard.
[25] These hysteresis effects go well beyond those associated with the labor market that have previously been a subject of extensive discussion. Thus, a "temporary" increase in the interest rate, because it depletes firms’ net worth, continues to have effects long after the interest rate returns to normal levels.