Short Forecast Horizons Research Articles

Comparing the DSGE Model with the Factor Model: An Out-of-Sample Forecasting Experiment

In this paper, we put DSGE forecasts in competition with factor forecasts. We focus on these two models since they represent nicely the two opposing forecasting philosophies. The DSGE model on the one hand has a strong theoretical economic background; the factor model on the other hand is mainly data-driven. We show that by incooperating large information set using factor analysis can indeed improve the short horizon predictive ability, as claimed by manyresearchers. The micro founded DSGE model can provide reasonable forecasts for inflation, especially with growing forecast horizons. To a certain extent, our results are consistent with the prevailling view that simple time series models should be used in short-horizon forecasting and structural models should be used in long-horizon forecasting. Our paper compareds both state-of-the art data-driven and theory-based modelling in a rigorous manner.

Time Series Volatility Forecasts for Option Valuation and Risk Management

Option pricing models and longer-term value-at-risk models typically require volatility forecasts over horizons considerably longer than the data frequency. These are generally generated from short-horizon forecasts by successive forward substitution. We document deficiencies with the resulting long-horizon volatility predictions generated by GARCH type models, such as GARCH(1,1), EGARCH, and GJR. One, since volatility forecasts for forward periods are functions of forecast volatility for the next period, this recursive procedure keeps the relative weights of recent and older observations the same whether forecasting volatility in the near or distant future. In contrast, we find that older observations are relatively more important in forecasting at long horizons, e.g., more important in forecasting volatility next month than in forecasting volatility tomorrow. Two, forecasts of the return standard deviation - the most appropriate volatility measure for option valuation and value-at-risk models - are strongly positively biased. Three, GARCH(1,1) and GJR forecasts are especially biased following high volatility days. We find that the ARLS model of Ederington and Guan corrects these three deficiencies and generally forecasts better out-of-sample than GARCH, EGARCH, AGARCH and the GJR models across a wide variety of markets and forecast horizons.

Forecasting Interest Rates and Inflation: Blue Chip Clairvoyants or Econometrics?

This study examines the performance of the professional analysts in the Blue Chip Financial Forecasts vis-a-vis set of competing econometric benchmarks, including shrinkage versions that adjust for in-sample over-fit in improving out-of-sample performance. The individual participants perform the best overall for short horizon forecasts of short to medium term yields and inflation. Econometric models with shrinkage perform the best over longer horizons and maturities. Aggregating over a larger set of analysts improves inflation surveys while generally degrading interest rates surveys. We document predictability in the survey forecast errors, which exhibit substantial variability across di fferent economic episodes, and propose a new adjustment that can substantially improve the performance of the survey participants.

Technical Constraints on and Efficient Strategies for the Integration of Wind Energy

Based on simulations of the German and Austrian electricity systems, the impacts of wind generation on power plant operation are analysed. Accurate forecasts play the major role when integrating wind power because they determine additional reserve capacity as well as any additional imbalance. Additional balancing power can be limited to less than 10 % of the installed wind capacity if improved forecast tools or shorter forecast horizons are used. Additional balancing costs are below 5 € per MWh wind energy depending on the forecast accuracy and the available balancing capacity. Balancing markets need developments which improve the accessibility for demand-side options as simulations have shown potential cost reductions of 25%.

Management Earnings Forecasts and Simultaneous Release of Earnings News

This paper examines the hypothesis that when disappointing information regarding a firm's performance is released, the firm's management faces particularly strong incentives to counter the disappointing news with overly optimistic forward-looking statements. To address this issue, the paper investigates the properties of management earnings forecasts released simultaneously with earnings announcements. The paper predicts and finds that the more disappointing the earnings announcement news, the higher the optimistic bias in the simultaneously released management long-horizon forecasts of annual earnings. Since managers may use short-horizon forecasts of quarterly earnings to avoid disappointing earnings announcements, the paper also examines the interplay between management short-horizon forecasts of quarterly earnings and long-horizon forecasts of annual earnings that are released simultaneously. The paper documents a significantly negative association between the news contained in the short-horizon forecasts of quarterly earnings and the optimistic bias in the accompanying long-horizon forecast of annual earnings. The paper also investigates the link between management's decision to issue earnings guidance and the contemporaneous earnings announcement news, finding that, in general, more extreme earnings announcement news results in more earnings guidance. Finally, using market reaction tests, the paper documents that market participants are aware, at least to some extent, of management's tendency to counter disappointing earnings news with overly optimistic forecasts.

A Source of Long Memory in Volatility

The long memory characteristic of financial market volatility is well documented and has important implications for volatility forecasting and option pricing. When fitted to the same data, different volatility models calculate the unconditional variance differently and could have very different volatility persistent parameters. Hence, they produce very different volatility forecasts even when the projection is just beyond a few days. The popular GARCH and GJR models have short memory. This paper compares the out-of-sample forecasting performance of four long memory volatility models, viz. fractional integrated (FI), break, component and regime switching. Using S&P 500 returns, we find structural break model to produce the best in-sample fit and out-of-sample forecasts, if future volatility breaks are known. Without knowing the future breaks, GJR produced the best short horizon forecasts. For volatility forecasts of 10 days and beyond, FI dominates. The FI model projects the future unconditional variance from the exponentially weighted sum of an infinite number of past shocks. The persistence parameters then control how fast the forecasts converge to this unconditional variance. As the fractional differencing parameter gets closer to and exceeds 0.5, volatility is non-stationary. The success of the FI model in forecasting S&P 500 volatility suggests that the latter should be treated as nonstationary. Which volatility model is best for forecasting is an empirical issue. A best model for S&P 500 need not be the best for the other series, and may not always be the best, all the time, for forecasting S&P 500 volatility. Unusual events such as the 1987 crash, for example, call for unusual treatments to get better forecasting performance.

Historical Averages, Units Roots and Future Net Discount Rates: A Comprehensive Estimator

Abstract No abstract available.

Forecasting and Combining Competing Models of Exchange Rate Determination

This paper investigates the out-of-sample forecast performance of a set of competing models of exchange rate determination. We compare standard linear models with models that characterize the relationship between exchange rate and its underlying fundamentals by nonlinear dynamics. Linear models tend to outperform at short forecast horizons especially when deviations from long-term equilibrium are small. In contrast, nonlinear models with more elaborate mean-reverting components dominate at longer horizons especially when deviations from long-term equilibrium are large. The results also suggest that combining different forecasting procedures generally produces more accurate forecasts than can be attained from a single model.

How Good are Dynamic Factor Models at Forecasting Output and Inflation? A Meta-Analytic Approach

This paper surveys existing factor forecast applications for real economic activity and inflation by means of a meta-analysis and contributes to the current debate on the determinants of the forecast performance of large-scale dynamic factor models relative to other models. We find that, on average, factor forecasts are slightly better than other models' forecasts. In particular, factor models tend to outperform small-scale models, whereas they perform slightly worse than alternative methods which are also able to exploit large datasets. Our results further suggest that factor forecasts are better for US than for UK macroeconomic variables, and that they are better for US than for euro-area output; however, there are no significant differences between the relative factor forecast performance for US and euro-area inflation. There is also some evidence that factor models are better suited to predict output at shorter forecast horizons than at longer horizons. These findings all relate to the forecasting environment (which cannot be influenced by the forecasters). Among the variables capturing the forecasting design (which can, by contrast, be influenced by the forecasters), the size of the dataset from which factors are extracted seems to positively affect the relative factor forecast performance. There is some evidence that quarterly data lend themselves better to factor forecasts than monthly data. Rolling forecasts are preferable to recursive forecasts. The factor estimation technique seems to matter as well. Other potential determinants - namely whether forecasters rely on a balanced or an unbalanced panel, whether restrictions implied by the factor structure are imposed in the forecasting equation or not and whether an iterated or a direct multi-step forecast is made - are found to be rather irrelevant. Moreover, we find no evidence that pre-selecting the variables to be included in the panel from which factors are extracted helped to improve factor forecasts in the past.

Short‐term wind forecasting using off‐site observations

AbstractAccurate wind energy forecasts can be an essential component for economic viability of a wind project. Timely and accurate short‐term (hours) forecasts can increase the electric grid efficiency and minimize ancillary or other firming requirements, ultimately resulting in reduced costs. This article investigates the use of off‐site observations, at distances up to 200 km from the wind farm, as predictors in statistical forecast techniques. In combination with on‐site and off‐site observations, fine‐scale numerical weather predictions can also be used to further increase forecast accuracy at these short forecast horizons. An example from the Pacific Northwest of the USA is described. It is shown that significant forecast improvements are feasible when using off‐site observations and/or mesoscale numerical weather predictions in statistical forecast algorithms. Copyright © 2005 John Wiley & Sons, Ltd.

Nonlinear Dynamics and Out-of-sample Forecasts of Real Exchange Rates

The purpose of the paper is to contribute to the debate on the relevance of nonlinear modeling and forecasts in real exchange rates. Using monthly data from the posti-Bretton Woods periods for the developed seven countries, we found strong evidence of the presence of nonlinearities in the behavior of real exchange rate, which appeared to be modeled well by an exponential smooth transition autoregressive model. The presence of nonlinear meani-reverting adjustment for real exchange rates can be explained by the theoretical models that emphasize the role of transaction and monopolistic pricing. We then used this model to make outi-ofi-sample forecasting analysis. Overall, we find fairly limited evidence that favors the ESTAR model over a simple random walk benchmark. The nonlinear ESTAR forecasts tend to outperform a simple random walk at longer forecast horizons for UK and Switzerland, but at shorter forecast horizons for Japan and Canada, although the magnitude of the gain from using the ESTAR relative to a random walk is rather small in magnitude at entire forecast horizon.

The forecasting performance of various models for seasonality and nonlinearity for quarterly industrial production

Seasonality often accounts for the major part of short-run movements in quarterly or monthly macro economic time series. In addition, business cycle nonlinearity is a prominent feature of many such series. A forecaster can nowadays consider a wide variety of time series models that describe seasonal variation and nonlinear regime-switching behavior. In this paper we examine the forecasting performance of various models for seasonality and nonlinearity for quarterly industrial production series of 18 OECD countries. We find that the accuracy of point forecasts varies widely across series, across forecast horizons and across seasons. However, in general, linear models with fairly simple descriptions of seasonality outperform nonlinear at short forecast horizons, whereas nonlinear models with more elaborate seasonal components dominate at longer horizons. Finally, none of the models is found to render efficient forecasts and hence, forecast combination is worthwhile.

A comparison of tests of nonlinear cointegration with application to the predictability of US interest rates using the term structure

We test whether there are nonlinearities in the response of short- and long-term interest rates to the spread in interest rates, and assess the out-of-sample predictability of interest rates using linear and nonlinear models. We find strong evidence of nonlinearities in the response of interest rates to the spread. Nonlinearities are shown to result in more accurate short-horizon forecasts, especially of the spread.

A state space framework for automatic forecasting using exponential smoothing methods

We provide a new approach to automatic forecasting based on an extended range of exponential smoothing methods. Each method in our taxonomy of exponential smoothing methods provides forecasts that are equivalent to forecasts from a state space model. This equivalence allows: (1) easy calculation of the likelihood, the AIC and other model selection criteria; (2) computation of prediction intervals for each method; and (3) random simulation from the underlying state space model. We demonstrate the methods by applying them to the data from the M-competition and the M3-competition. The method provides forecast accuracy comparable to the best methods in the competitions; it is particularly good for short forecast horizons with seasonal data.

Evaluating Macro-Modelling Systems: An Application to the Area Wide Model

This paper considers approaches to considering the validity of the overall structure of macro-econometric models - specifically, the Area Wide Model of the European Central Bank. By structure, we refer to the dynamic (business-cycle) and steady-state features that the model purports to capture. This leads to two types of tests. The first, drawing on the DSGE literature, is concerned with whether the model matches business-cycle (high-frequency) data characteristics. This is implemented by a Cholesky bootstrap whereby the steady state of the model is stochastically simulated using historically consistent covariances. The generated data is analysed for stylised-facts fitting and, similarly, using the model's implied spectral characteristics, for congruence with the data in terms of persistence, periodicity and spectral fit. Moments matching, however, is only one aspect of overall model evaluation. Consequently, we move to tests that combine high-frequency aspects (short-horizon forecasts) with long-run features (such as the existence and identification of steady states and trends). Recursive forecasting tests form the second part. The forecasts attempt to measure the accuracy of model-based forecasts both simulated out-of-sample and in an in-sample exercise. The out-of-sample exercise analyses the 1- to 8-step-ahead forecasting ability of the model. For this, the model is re-estimated each time on a subset of the original sample, and the re-estimated model is used to generate a forecast over the remaining sample. The in-sample exercise omits the re-estimation step but performs a more thorough exercise, covering 1- to 12-steps-ahead forecasts over a larger part of the original sample. For this latter exercise, a thorough analysis of sources of forecast error is made. Both exercises incorporate alternative residual-projection methods, in order to assess the importance of unaccounted-for breaks in forecast accuracy. Conclusions reached are that, on the one hand, in-sample exercises should be preferred with systems of this size and typical samples; and, on the other, that mechanical residual adjustment or model re-estimation should be avoided except under strong evidence of mis-specification. The paper considers the testing procedure to be one applicable to the class of large, macro models and therefore of general interest.

Estimating the Implied Distribution of the Future Short Term Interest Rate Using the Longstaff-Schwartz Model

This paper proposes the use of the two-factor term-structure model of Longstaff and Schwartz (1992a, LS) to estimate the risk-neutral density (RND) of the future short-term interest rate. The resulting RND can be interpreted as the market's estimate of the density of the future short-term interest rate. As such, it provides a useful financial indicator of the perceived uncertainty of future developments in the short-term interest rate. The LS approach used in this paper provides an alternative to option-based estimation procedures, which may be useful in situations here options markets are not sufficiently developed to allow estimation of the implied distribution from observed option prices. A simulation-based comparison of these two approaches reveals that the differences in the results are relatively small in magnitude, at least for short forecast horizons. Furthermore, the LS model is quite successful in capturing the asymmetries of the true distribution. It is therefore concluded that the LS model can be useful for estimating the distribution of future interest rates, when the purpose is to provide a general measure of the market's perceived uncertainty, for example as an indictor for monetary policy purposes.

The role of dividends in valuation models used by analysts and fund managers

The value of a share is given by the dividend discount model as a simple function of future dividends; but the actual determination of the share price is rarely based upon the direct estimation of these future dividends. A ranking of the valuation models used by analysts and fund managers shows a preference for ‘unsophisticated’ valuation using, for example, the dividend yield rather than the dividend discount model. This finding is shown to depend upon the practical difficulty of using currently-available information to forecast future cash flows. This difficulty limits the quantitative basis of valuations to short forecast horizons, while the subjective, qualitative estimation of terminal value assumes great importance. Crucially, both analysts and fund managers use their own assessment of management quality to underpin the estimation of terminal value, on the basis that superior quality causes outperformance and that, whereas management quality can be assessed now, future performance itself is unobservable. Linked with this and with information asymmetry, valuation is a dynamic, company-specific process, focused on personal communication with management and embodying ongoing signalling and implicit contracting, using both dividends and other variables. This method of valuation causes formal valuation models such as the dividend yield to play only a limited role. They offer a benchmark of relative price differences, which serves as a basis from which to conduct subjective, company-specific analysis and to make investment decisions; but valuation models are not used exclusively, in themselves, to value shares.

Assessing the forecasters: an analysis of the forecasting records of the Treasury, the London Business School and the National Institute

This paper analyses the historical inflation and output growth forecasting records of the Treasury, the London Business School and the National Institute. An analytical framework is adopted that is based on a recent assessment by Granger of the reliability and quality of economic forecasts. It is found that forecasts are not of much use at horizons greater than 18 months (that is, 6 months before the year being forecast). Forecasters have a tendency to be incorrect in the same direction and to underestimate actual changes. Major economic events, in particular, tend to be poorly forecasted, with both serious bouts of inflation and deep recessions being underpredicted. However, short-horizon forecasts tend to be both unbiased and efficient and usually get the change correctly signed.

Seasonal unit roots and forecasts of two-digit European industrial production

Monthly industrial production in important sectors of the German, French and UK economies are shown to exhibit very strong seasonality, such that typically 80% or more of the variation in monthly growth can be attributed to seasonal effects. Seasonal unit root test results imply that most of these series should be modelled using conventional first differences with the inclusion of monthly dummy variables, rather than as specifications involving other levels of differencing. However, when the post-sample forecast accuracy is compared for various models, annual difference specifications often produce the most accurate forecasts at horizons of up to a year. First difference models appear to be most accurate at short forecast horizons for series where seasonality is particularly marked. The study also examines the impact of updating coefficient estimates and model respecification within the forecast period.

Predictive Evaluation of Econometric Forecasting Models in Commodity Futures Markets

The predictive accuracy of various econometric models, including random walks, vector-autoregressive and vector-error-correction models, are investigated using daily futures prices of four commodities (the S&P 500 index, treasury bonds, gold, and crude oil). All models are estimated using a rolling-window approach, and evaluated by both in-sample and out-of-sample performance measures. The criteria considered include system criteria, where we evaluate multiequation forecasting models, and univariate forecast-accuracy criteria. The five univariate criteria are root mean square error (RMSE), mean absolute deviation (MAD), mean absolute percentage error (MAPE), confusion matrix (CM), and confusion rate (CR). The five system criteria used include the trace of second-moment matrix of the forecast-errors matrix (TMSE), the trace of second-moment matrix of percentage-forecast errors (TMAPE), the generalized forecast-error second-moment matrix (GFESM), and a trading-rule profit criterion (TPC) based on a maximum-spread trading strategy. An in-sample criterion, the mean Schwarz information criteria (MSIC), is also computed. Our results suggest that error-correction models perform better in shorter forecast horizons, when models are compared based on quadratic loss measures and confusion matrices. However, the error-correction models which we consider perform better at all forecast horizons (one to five steps ahead) when models are compared based on a profit-maximization loss function. Further, our error-correction model, where the error-correction term is constructed according to a cost-of-carry equilibrium condition, outperforms our alternative error-correction model, which uses the price spreads as the error-correction term.