Economists have long emphasized the importance of expectations in determining macroeconomic outcomes. Yet there has been almost no recent e#ort to model actual empirical expectations data; instead,macroeconomists usually simply assume that expectations are `rational.' This paper shows that while empirical household expectations are not rational in the usual sense, expectational dynamics are well captured by a model in which households' views derive from news reports of the views of professional forecasters, which in turn may be rational. The model's estimates imply that people only occasionally pay attention to news reports; this inattention generates `stickyness' in aggregate expectations, with important macroeconomic consequences.

This paper argues that probability forecasts convey information on the uncertainties that surround macro-economic forecasts in a straightforward manner which is preferable to other alternatives, including the use of confidence intervals. Probability forecasts obtained using a small benchmark macroeconometric model as well as a number of other alternatives are presented and evaluated using recursive forecasts generated over the period 1999q1-2001q1. Out of sample probability forecasts of inflation and output growth are also provided over the period 2001q2-2003q1, and their implications discussed in relation to the Bank of England’s inflation target and the need to avoid recessions, both as separate events and jointly. The robustness of the results to parameter and model uncertainties is also investigated by a pragmatic implementation of the Bayesian model averaging approach.

Beginning in 1998, U.S. commercial banks may determine their regulatory capital requirements for financial market risk exposure using value-at-risk (VaR) models. Currently, regulators have available three hypothesis-testing methods for evaluating the accuracy of VaR models: the binomial, interval forecast and distribution forecast methods. Given the low power often exhibited by their corresponding hypothesis tests, these methods can often misclassify forecasts from inaccurate models as acceptably accurate. An alternative evaluation method using loss functions based on probability forecasts is proposed. Simulation results indicate that this method is only as capable of differentiating between forecasts from accurate and inaccurate models as the other methods. However, its ability to directly incorporate regulatory loss functions into model evaluations make it a useful complement to the current regulatory evaluation of VaR models.

We argue that the current framework for predictive ability testing (e.g.,West, 1996) is not necessarily useful for real-time forecast selection, i.e., for assessing which of two competing forecasting methods will perform better in the future. We propose an alternative framework for out-of-sample comparison of predictive ability which delivers more practically relevant conclusions. Our approach is based on inference about conditional expectations of forecasts and forecast errors rather than the unconditional expectations that are the focus of the existing literature. We capture important determinants of forecast performance that are neglected in the existing literature by evaluating what we call the forecasting method (the model and the parameter estimation procedure), rather than just the forecasting model. Compared to previous approaches, our tests are valid under more general data assumptions (heterogeneity rather than stationarity) and estimation methods, and they can handle comparison of both nested and non-nested models, which is not currently possible. To illustrate the usefulness of the proposed tests, we compare the forecast performance of three leading parameter-reduction methods for macroeconomic forecasting using a large number of predictors: a sequential model selection approach,

This paper addresses that issue

A rapidly growing literature has documented important improvements in volatility measurement and forecasting performance through the use of realized volatilities constructed from high-frequency returns coupled with relatively simple reduced form time series modeling procedures. Building on recent theoretical results from Barndorff-Nielsen and Shephard (2003c) for related bi-power variation measures involving the sum of high-frequency absolute returns, the present paper provides a practical framework for non-parametrically measuring the jump component in the realized volatility measurements. Exploiting these ideas for a decade of high-frequency five-minute returns for the DM/ $ exchange rate, the S&P500 aggregate market index, and the 30-year U.S. Treasury Bond, we find the jump components to be distinctly less persistent than the contribution to the overall return variability originating from the continuous sample path component of the price process. Explicitly including the jump measure as an additional explanatory variable in an easy-to-implement reduced form model for the realized volatilities results in highly significant jump coefficient estimates at the daily, weekly and quarterly forecasts horizons. As such, our results hold promise for improved financial asset allocation, risk management, and derivatives pricing, by separate modeling, forecasting and pricing of the continuous and jump components of the total return variability.

: We consider the forecasting of cointegrated variables, and we show that at long horizons nothing is lost by ignoring cointegration when forecasts are evaluated using standard multivariate forecast accuracy measures. In fact, simple univariate Box-Jenkins forecasts are just as accurate. Our results highlight a potentially important deficiency of standard forecast accuracy measures---they fail to value the maintenance of cointegrating relationships among variables--- and we suggest alternatives that explicitly do so. KEY WORDS: Prediction, Loss Function, Integration, Unit Root - 2 - 1. INTRODUCTION Cointegration implies restrictions on the low-frequency dynamic behavior of multivariate time series. Thus, imposition of cointegrating restrictions has immediate implications for the behavior of long-horizon forecasts, and it is widely believed that imposition of cointegrating restrictions, when they are in fact true, will produce superior long-horizon forecasts. Stock (1995, p. 1), for ...

Management, and anonymous referees for helpful comments. I am especially grateful for

England’s conference on “Credit Risk Modelling and the Regulatory Implications ” for their comments and suggestions. Evaluating Credit Risk Models Over the past decade, commercial banks have devoted many resources to developing internal models to better quantify their financial risks and assign economic capital. These efforts have been recognized and encouraged by bank regulators. Recently, banks have extended these efforts into the field of credit risk modeling. However, an important question for both banks and their regulators is evaluating the accuracy of a model’s forecasts of credit losses, especially given the small number of available forecasts due to their typically long planning horizons. Using a panel data approach, we propose evaluation methods for credit risk models based on crosssectional simulation. Specifically, models are evaluated not only on their forecasts over time, but also on their forecasts at a given point in time for simulated credit portfolios. Once the forecasts corresponding to these portfolios are generated, they can be evaluated using various statistical methods. I.

In situations where a regression model is subject to one or more breaks it is shown that it can be optimal to use pre-break data to estimate the parameters of the model used to compute out-of-sample forecasts. The issue of how best to exploit the trade-o that might exist between bias and forecast error variance is explored and illustrated for the multivariate regression model under the assumption of strictly exogenous regressors. In practice when this assumption cannot be maintained and both the time and size of the breaks are unknown the optimal choice of the observation window will be subject to further uncertainties that make exploiting the bias-variance tradeo di cult. To that end we propose a new set of cross-validation methods for selection of a single estimation window and weighting or pooling methods for combination of forecasts based on estimation windows of di erent lengths. Monte Carlo simulations are used to show when these procedures work well compared with methods that ignore the presence of breaks. JEL Classi cations: C22, C53.