nep-for New Economics Papers
on Forecasting
Issue of 2009‒10‒31
nine papers chosen by
Rob J Hyndman
Monash University

  1. Nested forecast model comparisons: a new approach to testing equal accuracy By Todd E. Clark; Michael W. McCracken
  2. In-sample tests of predictive ability: a new approach By Todd E. Clark; Michael W. McCracken
  3. Macroeconomic forecasting with real-time data: an empirical comparison By Heij, C.; Dijk, D.J.C. van; Groenen, P.J.F.
  4. The Multivariate k-Nearest Neighbor Model for Dependent Variables : One-Sided Estimation and Forecasting By Dominique Guegan; Patrick Rakotomarolahy
  5. Inflation Expectations: Does the Market Beat Professional Forecasts? By Makram El-Shagi
  6. Evaluating German Business Cycle Forecasts Under an Asymmetric Loss Function By Jörg Döpke; Ulrich Fritsche; Boriss Siliverstovs
  7. Constructing Forecast Confidence Bands During the Financial Crisis By Ondra Kamenik; Marianne Johnson; Kevin Clinton; Huigang Chen; Douglas Laxton
  8. Exchange Rates and Stock Prices in the Long Run and Short Run By Morley, Bruce
  9. Modeling and prediction of surgical procedure times By Stepaniak, P. S.; Heij, C.; Vries, G. de

  1. By: Todd E. Clark; Michael W. McCracken
    Abstract: This paper develops bootstrap methods for testing whether, in a finite sample, competing out-of-sample forecasts from nested models are equally accurate. Most prior work on forecast tests for nested models has focused on a null hypothesis of equal accuracy in population - basically, whether coefficients on the extra variables in the larger, nesting model are zero. We instead use an asymptotic approximation that treats the coefficients as non-zero but small, such that, in a finite sample, forecasts from the small model are expected to be as accurate as forecasts from the large model. Under that approximation, we derive the limiting distributions of pairwise tests of equal mean square error, and develop bootstrap methods for estimating critical values. Monte Carlo experiments show that our proposed procedures have good size and power properties for the null of equal finite-sample forecast accuracy. We illustrate the use of the procedures with applications to forecasting stock returns and inflation.
    Keywords: Economic forecasting
    Date: 2009
  2. By: Todd E. Clark; Michael W. McCracken
    Abstract: This paper presents analytical, Monte Carlo, and empirical evidence linking in-sample tests of predictive content and out-of-sample forecast accuracy. Our approach focuses on the negative effect that finite-sample estimation error has on forecast accuracy despite the presence of significant population-level predictive content. Specifically, we derive simple-to-use in-sample tests that test not only whether a particular variable has predictive content but also whether this content is estimated precisely enough to improve forecast accuracy. Our tests are asymptotically non-central chi-square or non-central normal. We provide a convenient bootstrap method for computing the relevant critical values. In the Monte Carlo and empirical analysis, we compare the effectiveness of our testing procedure with more common testing procedures.
    Keywords: Economic forecasting
    Date: 2009
  3. By: Heij, C.; Dijk, D.J.C. van; Groenen, P.J.F. (Erasmus Econometric Institute)
    Abstract: Macroeconomic forecasting is not an easy task, in particular if future growth rates are forecasted in real time. This paper compares various methods to predict the growth rate of US Industrial Production (IP) and of the Composite Coincident Index (CCI) of the Conference Board, over the coming month, quarter, and half year. It turns out that future IP growth rates can be forecasted in real time from ten leading indicators, by means of the Composite Leading Index (CLI) or, even somewhat better, by principal components regression. This amends earlier negative findings for IP by Diebold and Rudebusch. For CCI, on the other hand, simple autoregressive models do not provide significantly less accurate forecasts than single-equation and bivariate vector autoregressive models with the CLI. This amends some of the more positive results for CCI recently reported by the Conference Board. Not surprisingly, all forecast methods improve considerably if `ex post' data are used, after possible data updates and revisions.
    Keywords: vintage date;leading indicators;forecast evaluation;recessions;industrial production;composite coincident index
    Date: 2009–10–19
  4. By: Dominique Guegan (CES - Centre d'économie de la Sorbonne - CNRS : UMR8174 - Université Panthéon-Sorbonne - Paris I, EEP-PSE - Ecole d'Économie de Paris - Paris School of Economics - Ecole d'Économie de Paris); Patrick Rakotomarolahy (CES - Centre d'économie de la Sorbonne - CNRS : UMR8174 - Université Panthéon-Sorbonne - Paris I)
    Abstract: Forecasting current quarter GDP is a permanent task inside the central banks. Many models are known and proposed to solve this problem. Thanks to new results on the asymptotic normality of the multivariate k-nearest neighbor regression estimate, we propose an interesting and new approach to solve in particular the forecasting of economic indicators, included GDP modelling. Considering dependent mixing data sets, we prove the asymptotic normality of multivariate k-nearest neighbor regression estimate under weak conditions, providing confidence intervals for point forecasts. We introduce an application for economic indicators of euro area, and compare our method with other classical ARMA-GARCH modelling.
    Keywords: Multivariate k-nearest neighbor, asymptotic normality of the regression, mixing time series, confidence intervals, forecasts, economic indicators, euro area.
    Date: 2009–07
  5. By: Makram El-Shagi
    Abstract: The present paper compares expected inflation to (econometric) inflation forecasts based on a number of forecasting techniques from the literature using a panel of ten industrialized countries during the period of 1988 to 2007. To capture expected inflation we develop a recursive filtering algorithm which extracts unexpected inflation from real interest rate data, even in the presence of diverse risks and a potential Mundell-Tobin-effect. The extracted unexpected inflation is compared to the forecasting errors of ten econometric forecasts. Beside the standard AR(p) and ARMA(1,1) models, which are known to perform best on average, we also employ several Phillips curve based approaches, VAR, dynamic factor models and two simple model avering approaches.
    Keywords: Inflation Expectations,Rational Expectations,Inflation Forecasting
    JEL: E31 E37
    Date: 2009–10
  6. By: Jörg Döpke (University of Applied Sciences Merseburg, Merseburg/Germany); Ulrich Fritsche (Hamburg University, Faculty Economics and Social Sciences and German Institute of Economic Research, Hamburg/Germany); Boriss Siliverstovs (KOF Swiss Economic Institute, ETH Zürich)
    Abstract: Based on annual data for growth and inflation forecasts for Germany covering the time span from 1970 to 2007 and up to 17 different forecasts per year, we test for a possible asymmetry of the forecasters' loss function and estimate the degree of asymmetry for each forecasting institution using the approach of Elliot et al. (2005). Furthermore, we test for the rationality of the forecasts under the assumption of a possibly asymmetric loss function and for the features of an optimal forecast under the assumption of a generalized loss function. We find evidence for the existence of an asymmetric loss function of German forecasters only in case of pooled data and a quad-quad loss function. We cannot reject the hypothesis of rationality of the growth forecasts based on data for single institutions, but based on a pooled data set. The rationality of inflation forecasts frequently is rejected in case of single institutions and also for pooled data.
    Keywords: Business cycle forecast evaluation, asymmetric loss function, and rational expectations
    JEL: C53 E42
    Date: 2009–08
  7. By: Ondra Kamenik; Marianne Johnson; Kevin Clinton; Huigang Chen; Douglas Laxton
    Abstract: We derive forecast confidence bands using a Global Projection Model covering the United States, the euro area, and Japan. In the model, the price of oil is a stochastic process, interest rates have a zero floor, and bank lending tightening affects the United States. To calculate confidence intervals that respect the zero interest rate floor, we employ Latin hypercube sampling. Derived confidence bands suggest non-negligible risks that U.S. interest rates might stay near zero for an extended period, and that severe credit conditions might persist.
    Keywords: Bank credit , Credit restraint , Economic forecasting , Economic models , European Union , Inflation targeting , Interest rates , Japan , Monetary policy , Oil prices , United States ,
    Date: 2009–09–30
  8. By: Morley, Bruce
    Abstract: Using the ARDL bounds testing approach to cointegration this paper provides evidence of a stable long run relationship between the exchange rate and stock prices for the UK, Japan and Swiss currencies with respect to the US dollar. The resultant error correction models suggest a positive relationship between stock prices and the exchange rate, which in an out-of-sample forecast outperforms the random walk. We compare these results with a similar model incorporating interest rates, suggested by Solnik (1987), however this does not in general improve the results.
    Keywords: Exchange Rates; Stock Prices; Forecast; Cointegration
    Date: 2009
  9. By: Stepaniak, P. S.; Heij, C.; Vries, G. de (Erasmus Econometric Institute)
    Abstract: Accurate prediction of medical operation times is of crucial importance for cost efficient operation room planning in hospitals. This paper investigates the possible dependence of procedure times on surgeon factors like age, experience, gender, and team composition. The effect of these factors is estimated for over 30 different types of medical operations in two hospitals, by means of ANOVA models for logarithmic case durations. The estimation data set contains about 30,000 observations from 2005 till 2008. The relevance of surgeon factors depends on the type of operation. The factors found most often to be significant are team composition, experience, and daytime. Contrary to widespread opinions among surgeons, gender has nearly never a significant effect. By incorporating surgeon factors, the accuracy of out-of-sample prediction of case durations of about 1,250 surgical operations in 2009 is improved by up to more than 15 percent as compared to current planning procedures.
    Keywords: operation room;surgeon factors;lognormal distribution;ANOVA model;planning;European hospital;health care management;current procedure terminology (CPT)
    Date: 2009–10–19

This nep-for issue is ©2009 by Rob J Hyndman. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.