Meese-Rogoff Redux: Micro-Based Exchange-Rate Forecasting By MARTIN D . D . EVANS AND RICHARD K . LYONS* This paper compares the true, ex ante fore- casting performance of a micro-based model against both a standard macro model and a random walk. In contrast to existing literature, which is focused on longer-horizon forecasting, we examine forecasting over horizons from one day to one month (the one-month horizon being where micro and macro analysis begin to over- lap). Over our three-year forecasting sample, we find that the micro-based model consistently outperforms both the random walk and the macro model. Micro-based forecasts account for almost 16 percent of the sample variance in monthly spot rate changes. These results pro- vide a level of empirical validation as yet unat- tained by other models. The forecasting experiment proposed by Richard Meese and Kenneth Rogoff (1983) re- mains a benchmark against which exchange- rate models are judged. Their result that structural macro models cannot outperform a naive random walk has proved robust over the decades. Yet, the Meese-Rogoff paper was never about forecasting in the true sense (i.e., using tirrie-f information to forecast exchange rates at t + 1). B y using concurrent, realized values of the forcing variables, their regressions were more about concurrent explanation than about ex ante forecasting. Their only forecast- ing element is in their reliance on ex ante data to estimate equation parameters, which appro- priately penalized models whose estimated pa- rameters were unstable. tion for the random-walk nature of exchange rates is that there exists some unobserved fun- damental that itself follows a random walk (un- observed being important because fundamentals proposed in most macro models do not follow random walks). They offer a different explana- tion, one rooted in the asset approach to ex- change rates and the present-value relation that follows from it. Specifically, they show that i f fundamentals are 1(1), but not necessarily ran- dom walks, then as the discount factor in the present-value relation approaches 1, the ex- change rate will follow a process arbitrarily close to a random walk. Intuitively, given that an 1(1) process can be split into random-walk and stationary components, a discount factor near 1 means that most all of the weight is placed on fundamentals far into the future, ex- pectations of which are dominated by the random- walk component. Charles Engel and Kenneth West (2004, 2005) provide a valuable perspective on the forecastability of exchange rates. One explana- This paper takes the analysis of Engel and West as an important reorientation of thinking and brings it to the natural next step. Specifi- cally, if there is little room for forecasting based on stationary components of fundamentals, then one needs to focus on where all the action is, namely, exchange-rate dynamics that come from expectational surprises. Though the sur- prise part is, by definition, orthogonal to public information, our micro-based model shows that there should exist types of nonpublic infor- mation that are useful for forecasting this part, and where to look for these types of informa- tion. We then locate data on these types of nonpublic information and test whether they have true, ex ante forecasting power. We should * Evans: Department of Economics, Georgetown Uni- versity, Washington, DC 20057, and NBER; Lyons: Haas School of Business, University of California, Berkeley, C A 94720-1900, and NBER. We are grateful to Ken West for his comments and to the National Science Foundation for financial support. We refrain from using the word news because it has too strong an association with macro information that is public, which, even in the most careful of event studies, explains less than 5 percent of exchange-rate variation in total (see Torben Andersen et al., 2003).

We analyze a general equilibrium model in which there is both adverse selection of, and moral hazard by, banks. The regulator can screen banks prior to giving them a licence, audit them ex post to learn the success probability of their projects, and impose capital adequacy requirements. Capital requirements combat moral hazard when the regulator has a strong screening reputation, and they otherwise substitute for screening ability. Crises of confidence can occur only in the latter case, and contrary to conventional wisdom, the appropriate policy response may be to tighten capital requirements to improve the quality of surviving banks.

From a macroeconomic perspective, the shortterm interest rate is a policy instrument under the direct control of the central bank, which adjusts the rate to achieve its economic stabilization goals. From a finance perspective, the short rate is a fundamental building block for yields of other maturities, which are just riskadjusted averages of expected future short rates. Thus, as illustrated by much recent research, a joint macro-finance modeling strategy will provide the most comprehensive understanding of the term structure of interest rates. In this paper, we discuss some salient questions that arise in this research, and we also present a new examination of the relationship between two prominent dynamic, latent factor models in this literature: the Nelson-Siegel and affine no-arbitrage term-structure models. I. Questions about Modeling Yields 1. Why Use Factor Models for Bond Yields?—The first problem faced in term-structure modeling is how to summarize the price information at any point in time for the large number of nominal bonds that are traded. In fact, since only a small number of sources of systematic risk appear to underlie the pricing of the myriad of tradable financial assets, nearly all bond price information can be summarized with just a few constructed variables or factors. Therefore, yield-curve models almost invariably employ a structure that consists of a small set of factors and the associated factor loadings