|
on Forecasting |
| By: | Alquist, Ron; Kilian, Lutz; Vigfusson, Robert J. |
| Abstract: | We address some of the key questions that arise in forecasting the price of crude oil. What do applied forecasters need to know about the choice of sample period and about the tradeoffs between alternative oil price series and model specifications? Are real or nominal oil prices predictable based on macroeconomic aggregates? Does this predictability translate into gains in out-of-sample forecast accuracy compared with conventional no-change forecasts? How useful are oil futures markets in forecasting the price of oil? How useful are survey forecasts? How does one evaluate the sensitivity of a baseline oil price forecast to alternative assumptions about future demand and supply conditions? How does one quantify risks associated with oil price forecasts? Can joint forecasts of the price of oil and of U.S. real GDP growth be improved upon by allowing for asymmetries? |
| Keywords: | Asymmetries; Demand and supply; Forecasting; Oil price; Predictability |
| JEL: | C53 Q43 |
| Date: | 2011–05 |
| URL: | http://d.repec.org/n?u=RePEc:cpr:ceprdp:8388&r=for |
| By: | Gary Koop (Department of Economics, University of Strathclyde); Dimitris Korobilis (Center for Operations Research & Econometrics (CORE), Universite Catholique de Louvain) |
| Abstract: | We forecast quarterly US inflation based on the generalized Phillips curve using econometric methods which incorporate dynamic model averaging. These methods not only allow for coe¢ cients to change over time, but also allow for the entire forecasting model to change over time. We find that dynamic model averaging leads to substantial forecasting improvements over simple benchmark regressions and more sophisticated approaches such as those using time varying coefficient models. We also provide evidence on which sets of predictors are relevant for forecasting in each period. |
| Keywords: | Bayesian, State space model, Phillips curve |
| JEL: | E31 E37 C11 C53 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1119&r=for |
| By: | Luc Bauwens (Université catholique de Louvain); Gary Koop (Department of Economics, University of Strathclyde); Dimitris Korobilis (Universite Catholique de Louvain); Jeroen Rombouts (HEC Montréal (École des Hautes Études Commerciales) (Business School) and Center for Operations Research and Econometrics (CORE) ECORE) |
| Abstract: | This paper compares the forecasting performance of different models which have been proposed for forecasting in the presence of structural breaks. These models differ in their treatment of the break process, the parameters defining the model which applies in each regime and the out-of-sample probability of a break occurring. In an extensive empirical evaluation involving many important macroeconomic time series, we demonstrate the presence of structural breaks and their importance for forecasting in the vast majority of cases. However, we find no single forecasting model consistently works best in the presence of structural breaks. In many cases, the formal modeling of the break process is important in achieving good forecast performance. However, there are also many cases where simple, rolling OLS forecasts perform well. |
| Keywords: | Forecasting, change-points, Markov switching, Bayesian inference. |
| JEL: | C11 C22 C53 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1113&r=for |
| By: | Gary Koop (Department of Economics, University of Strathclyde); Dimitris Korobilis (Center for Operations Research & Econometrics (CORE), Universite Catholique de Louvain) |
| Abstract: | Block factor methods offer an attractive approach to forecasting with many predictors. These extract the information in these predictors into factors reflecting different blocks of variables (e.g. a price block, a housing block, a financial block, etc.). However, a forecasting model which simply includes all blocks as predictors risks being over-parameterized. Thus, it is desirable to use a methodology which allows for different parsimonious forecasting models to hold at di¤erent points in time. In this paper, we use dynamic model averaging and dynamic model selection to achieve this goal. These methods automatically alter the weights attached to different forecasting model as evidence comes in about which has forecast well in the recent past. In an empirical study involving forecasting output and inflation using 139 UK monthly time series variables, we find that the set of predictors changes substantially over time. Furthermore, our results show that dynamic model averaging and model selection can greatly improve forecast performance relative to traditional forecasting methods. |
| Keywords: | Bayesian, state space model, factor model, dynamic model averaging |
| JEL: | E31 E37 C11 C53 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1118&r=for |
| By: | Gary Koop (Department of Economics, University of Strathclyde) |
| Abstract: | This paper is motivated by the recent interest in the use of Bayesian VARs for forecasting, even in cases where the number of dependent variables is large. In such cases, factor methods have been traditionally used but recent work using a particular prior suggests that Bayesian VAR methods can forecast better. In this paper, we consider a range of alternative priors which have been used with small VARs, discuss the issues which arise when they are used with medium and large VARs and examine their forecast performance using a US macroeconomic data set containing 168 variables. We ?nd that Bayesian VARs do tend to forecast better than factor methods and provide an extensive comparison of the strengths and weaknesses of various approaches. Our empirical results show the importance of using forecast metrics which use the entire predictive density, instead of using only point forecasts. |
| Keywords: | Bayesian, Minnesota prior, stochastic search variable selection, predictive likelihood |
| JEL: | C11 C32 C53 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1117&r=for |
| By: | Massimiliano Caporin; Michael McAleer (University of Canterbury) |
| Abstract: | In the last 15 years, several Multivariate GARCH (MGARCH) models have appeared in the literature. Recent research has begun to examine MGARCH specifications in terms of their out-of-sample forecasting performance. In this paper, we provide an empirical comparison of a set of models, namely BEKK, DCC, Corrected DCC (cDCC) of Aeilli (2008), CCC, Exponentially Weighted Moving Average, and covariance shrinking, using historical data of 89 US equities. Our methods follow part of the approach described in Patton and Sheppard (2009), and the paper contributes to the literature in several directions. First, we consider a wide range of models, including the recent cDCC model and covariance shrinking. Second, we use a range of tests and approaches for direct and indirect model comparison, including the Weighted Likelihood Ratio test of Amisano and Giacomini (2007). Third, we examine how the model rankings are influenced by the cross-sectional dimension of the problem. |
| Keywords: | Covariance forecasting; model confidence set; model ranking; MGARCH; model comparison |
| JEL: | C32 C53 C52 |
| Date: | 2011–05–01 |
| URL: | http://d.repec.org/n?u=RePEc:cbt:econwp:11/23&r=for |
| By: | Andrea Monticini (Universita Cattolica - Milano); Francesco Ravazzolo (Norges Bank (Central Bank of Norway)) |
| Abstract: | Market efficiency hypothesis suggests a zero level for the intraday interest rate. However, a liquidity crisis introduces frictions related to news, which can cause an upward jump of the intraday rate. This paper documents that these dynamics can be partially predicted during turbulent times. A long memory approach outperforms random walk and autoregressive benchmarks in terms of point and density forecasting. The gains are particular high when the full distribution is predicted and probabilistic assessments of future movements of the interest rate derived by the model can be used as a policy tool for central banks to plan supplementary market operations during turbulent times. Adding exogenous variables to proxy funding liquidity and counterparty risks does not improve forecast accuracy and the predictability seems to derive from the econometric properties of the series more than from news available to financial markets in realtime. |
| Keywords: | Interbank market, Intraday interest rate, Forecasting, Density forecasting, Policy tools. |
| JEL: | C22 C53 E4 E5 |
| Date: | 2011–06–06 |
| URL: | http://d.repec.org/n?u=RePEc:bno:worpap:2011_06&r=for |
| By: | Baumeister, Christiane; Kilian, Lutz |
| Abstract: | We construct a monthly real-time data set consisting of vintages for 1991.1-2010.12 that is suitable for generating forecasts of the real price of oil from a variety of models. We document that revisions of the data typically represent news, and we introduce backcasting and nowcasting techniques to fill gaps in the real-time data. We show that real-time forecasts of the real price of oil can be more accurate than the no-change forecast at horizons up to one year. In some cases real-time MSPE reductions may be as high as 25 percent one month ahead and 24 percent three months ahead. This result is in striking contrast to related results in the literature for asset prices. In particular, recursive vector autoregressive (VAR) forecasts based on global oil market variables tend to have lower MSPE at short horizons than forecasts based on oil futures prices, forecasts based on AR and ARMA models, and the no-change forecast. In addition, these VAR models have consistently higher directional accuracy. We demonstrate how with additional identifying assumptions such VAR models may be used not only to understand historical fluctuations in the real price of oil, but to construct conditional forecasts that reflect hypothetical scenarios about future demand and supply conditions in the market for crude oil. These tools are designed to allow forecasters to interpret their oil price forecast in light of economic models and to evaluate its sensitivity to alternative assumptions. |
| Keywords: | Forecast; Oil price; Real time; Scenario analysis |
| JEL: | C53 E32 Q43 |
| Date: | 2011–06 |
| URL: | http://d.repec.org/n?u=RePEc:cpr:ceprdp:8414&r=for |
| By: | Joshua Chan (Australian National University); Gary Koop (Department of Economics, University of Strathclyde); Roberto Leon-Gonzalez (National Graduate Institute for Policy Studies); Rodney Strachan (The Australian National University) |
| Abstract: | Time varying parameter (TVP) models have enjoyed an increasing popularity in empirical macroeconomics. However, TVP models are parameter-rich and risk over-fitting unless the dimension of the model is small. Motivated by this worry, this paper proposes several Time Varying dimension (TVD) models where the dimension of the model can change over time, allowing for the model to automatically choose a more parsimonious TVP representation, or to switch between different parsimonious representations. Our TVD models all fall in the category of dynamic mixture models. We discuss the properties of these models and present methods for Bayesian inference. An application involving US in?ation forecasting illustrates and compares the different TVD models. We find our TVD approaches exhibit better forecasting performance than several standard benchmarks and shrink towards parsimonious specifications. |
| Keywords: | mixture model, model change, Bayesian |
| JEL: | C11 C24 C32 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1116&r=for |
| By: | Jan Willem Slingenberg; Jakob de Haan |
| Abstract: | This paper uses a Financial Stress Index (FSI) for 13 OECD countries to examine which variables can help predicting financial stress. A stress index measures the current state of stress in the financial system and summarizes it in a single statistic. We employ three criteria for indicators to be used in constructing a multi-country FSI (the index covers the entire financial system, indicators used are available at a high frequency for many countries for a long period, and are comparable) to come up with our FSI. Our results suggest that financial stress is hard to predict. Only credit growth has predictive power for most countries. Several other variables have predictive power for some countries, but not for others. |
| Keywords: | financial stress index; predicting financial stress |
| JEL: | E5 G10 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:dnb:dnbwpp:292&r=for |
| By: | Tsyplakov, Alexander |
| Abstract: | This is a comment on Mitchell and Wallis (2011) which in turn is a critical reaction to Gneiting et al. (2007). The comment discusses the notion of forecast calibration, the advantage of using scoring rules, the “sharpness” principle and a general approach to testing calibration. The aim is to show how a more general and explicitly stated framework for evaluation of probabilistic forecasts can provide further insights. |
| Keywords: | density forecasts |
| JEL: | C53 C52 |
| Date: | 2011–05–30 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:31184&r=for |
| By: | Michael P. Clements (University of Warwick); Ana Beatriz Galvão (Queen Mary, University of London) |
| Abstract: | Real-time estimates of output gaps and inflation trends differ from the values that are obtained using data available long after the event. Part of the problem is that the data on which the real-time estimates are based is subsequently revised. We show that vector-autoregressive models of data vintages provide forecasts of post-revision values of future observations and of already-released observations capable of improving real-time output gap and inflation trend estimates. Our findings indicate that annual revisions to output and inflation data are in part predictable based on their past vintages. |
| Keywords: | Revisions, Real-time forecasting, Output gap, Inflation trend |
| JEL: | C53 |
| Date: | 2011–06 |
| URL: | http://d.repec.org/n?u=RePEc:qmw:qmwecw:wp678&r=for |
| By: | Gary Koop (Department of Economics, University of Strathclyde); Lise Tole (Edinburgh University, Business School) |
| Abstract: | In an effort to meet its obligations under the Kyoto Protocol, in 2005 the European Union introduced a cap-and-trade scheme where mandated installations are allocated permits to emit CO2. Financial markets have developed that allow companies to trade these carbon permits. For the EU to achieve reductions in CO2 emissions at a minimum cost, it is necessary that companies make appropriate investments and policymakers design optimal policies. In an effort to clarify the workings of the carbon market, several recent papers have attempted to statistically model it. However, the European carbon market (EU ETS) has many institutional features that potentially impact on daily carbon prices (and associated ?nancial futures). As a consequence, the carbon market has properties that are quite different from conventional financial assets traded in mature markets. In this paper, we use dynamic modelaveraging (DMA) in order to forecast in this newly-developing market. DMA is a recently-developed statistical method which has three advantages over conventional approaches. First, it allows the coe¢ cients on the predictors in a forecasting model to change over time. Second, it allows for the entire forecasting model to change over time. Third, it surmounts statistical problems which arise from the large number of potential predictors that can explain carbon prices. Our empirical results indicate that there are both important policy and statistical benefits with our approach. Statistically, we present strong evidence that there is substantial turbulence and change in the EU ETS market, and that DMA can model these features and forecast accurately compared to conventional approaches. From a policy perspective, we discuss the relative and changing role of different price drivers in the EU ETS. Finally, we document the forecast performance of DMA and discuss how this relates to the efficiency and maturity of this market. |
| Keywords: | Bayesian, carbon permit trading, financial markets, state space model, model averaging |
| JEL: | C53 C24 |
| Date: | 2011–04 |
| URL: | http://d.repec.org/n?u=RePEc:str:wpaper:1110&r=for |
| By: | Adi Schnytzer (Bar-Ilan University); Janez Šušteršič (University of Primorska) |
| Abstract: | Standard methods to assess the statistical quality of econometric models implicitly assume there is only one person in the world, namely the forecaster with her model(s), and that there exists an objective and independent reality to which the model predictions may be compared. However, on many occasions, the reality with which we compare our predictions and in which we take our actions is co-determined and changed constantly by actions taken by other actors based on their own models. We propose a new method, called a regression tournament, to assess the utility of forecasting models and taking these interactions into account. We present an empirical case of betting on Australian Rules Football matches where the most accurate predictive model does not yield the highest betting return, or, in our terms, does not win a regression tournament. |
| Date: | 2011–03 |
| URL: | http://d.repec.org/n?u=RePEc:biu:wpaper:2011-10&r=for |
| By: | Zafer Dilaver (Surrey Energy Economics Centre (SEEC), Department of Economics, University of Surrey); Lester C Hunt (Surrey Energy Economics Centre (SEEC), Department of Economics, University of Surrey) |
| Abstract: | This paper investigates the relationship between Turkish aggregate electricity consumption, GDP and electricity prices in order to forecast future Turkish aggregate electricity demand. To achieve this, an aggregate electricity demand function for Turkey is estimated by applying the structural time series technique to annual data over the period 1960 to 2008. The results suggest that GDP, electricity prices and an underlying energy demand trend (UEDT) are all important drivers of Turkish electricity demand. The estimated income and price elasticities are found to be 0.17 and -0.11 respectively with the estimated UEDT found to be generally upward sloping (electricity using) but at a generally decreasing rate. Based on the estimated equation, and different forecast assumptions, it is predicted that Turkish aggregate electricity demand will be somewhere between 259 TWh and 368 TWh in 2020. |
| Keywords: | Turkish Turkish Aggregate Electricity Demand; Structural Time Series Model (STSM); Energy Demand Modelling and Future Scenarios. |
| JEL: | C22 Q41 Q48 |
| Date: | 2011–05 |
| URL: | http://d.repec.org/n?u=RePEc:sur:seedps:132&r=for |
| By: | Adi Schnytzer (Department of Economics, Bar Ilan University); Martien Lamers (Ghent University; Department of Financial Economics, Ghent University); Vasiliki Makropoulou (Utrecht School of Economics, Utrecht University) |
| Abstract: | This paper uses a new variable based on estimates of insider trading to forecast the outcome of horse races. We base our analysis on Schnytzer, Lamers and Makropoulou (2008) who showed that inside trading in the 1997-1998 Australian racetrack betting market represents somewhere between 20 and 30 percent of all trading in this market. They show that the presence of insiders leads opening prices to deviate from true winning probabilities. Under these circumstances, forecasting of race outcomes should take into account an estimate of the extent of insider trading per horse. We show that the added value of this new variable for profitable betting is sufficient to reduce the losses when only prices are taken into account. Since the only variables taken into account in either Schnytzer, Lamers and Makropoulou (2008) or this paper are price data, this is tantamount to a demonstration that the market is, in practice, weak-form efficient. |
| Date: | 2011–03 |
| URL: | http://d.repec.org/n?u=RePEc:biu:wpaper:2011-14&r=for |
| By: | Adi Schnytzer (Department of Economics, Bar Ilan University) |
| Abstract: | The purpose of this paper is to make a novel contribution to the literature on the prediction market for the Australian Football League, the major sports league in which Australian Rules Football is played. Taking advantage of a novel micro-level data set which includes detailed per-game player statistics, predictions are presented and tested out-of-sample for the simplest kind of bet: fixed odds win betting. It is shown that player-level statistics may be used to yield very modest profits net of transaction costs over a number of seasons, provided some more global variables are added to the model. A comparison of different specifications of the linear probability model (LPM) versus conditional logit (CLOGIT) regressions reveals that the LPM usually outperforms CLOGIT in terms of profitability. It is further shown that adding significant variables to a regression specification which is clearly superior in econometric terms may reduce the efficacy of the prediction and thus profits. |
| Date: | 2011–03 |
| URL: | http://d.repec.org/n?u=RePEc:biu:wpaper:2011-15&r=for |
| By: | Charlotte Christiansen (Aarhus University, Business and Social Sciences and CREATES) |
| Abstract: | Severe simultaneous recessions are de?ned to occur when at least half of the countries under investigation (Australia, Canada, Germany, Japan, United Kingdom, and United States) are in recession simultaneously. I pose two new research questions that extend upon stylized facts for US recessions. One, are the occurrences of simultaneous recessions predictable? Two, does the yield spread predict future occurrences of simultaneous recessions? I use the indicator for severe simultaneous recessions as the explained variable in probit models. The lagged yield spread is an important explanatory variable, where decreasing yield spreads are a leading indicator for severe simultaneous recessions. |
| Keywords: | Business cycle, Recessions, Yield spread, Probit model |
| JEL: | C25 E32 E43 G15 |
| Date: | 2011–05–31 |
| URL: | http://d.repec.org/n?u=RePEc:aah:create:2011-20&r=for |