Persistence in forecasting performance and conditional combination strategies

Aiolfi, Marco; Timmermann, Allan

doi:10.1016/j.jeconom.2005.07.015

Cited by 293 publications

(243 citation statements)

References 17 publications

Supporting

Mentioning

236

Contrasting

Unclassified

Order By: Relevance

“…As in Stock and Watson (2004), we use δ = 1.0 (no discounting of past forecasts) and δ = 0.9 (more weight on most recent forecasts) resulting in Stock and Watson's discounted mean-square forecast error (DMSFE) methods DMSFE(1) and DMSFE(0.9). Finally, we use cluster combinations to control for forecast persistence (see Aiolfi and Timmermann, 2006). Utilizing a hold-out period, the individual ARDL models are ranked by their mean square forecast error (MSFE) and clusters are created by consecutively adding the next lowest MSFE ARDL model to the cluster.…”

Section: Combination Predictor Modelsmentioning

confidence: 99%

Sticky prices or economically-linked economies: The case of forecasting the Chinese stock market

Jordan

Vivian

Wohar

2014

Journal of International Money and Finance

View full text Add to dashboard Cite

We explore whether economic links via trade affect aggregate Chinese stock market returns. We find that market return indices from countries that China net imports from can forecast the Chinese aggregate market return at the weekly time horizon. The stock returns of countries that China net exports to have no consistently significant OOS predictability.The economic intuition for our results follows from the fact that China has positioned itself as a low-cost provider competing on price. As a low-cost provider China has a more difficult time passing cost increases through to export customers because of sticky prices. However, import costs, e.g., raw materials, are subject to both consumption and speculative demand and thus vary. We can conclude that costs will drive short term economic gains for the overall Chinese economy. One interpretation of our results is that supply shocks are absorbed within 2 weeks. IntroductionCan aggregate Chinese stock returns be forecast? To date, there is mixed US evidence on out-of-sample (OOS) predictability using fundamentals and macro variables, the two work horses of the predictability literature. However, outside the US there is mounting evidence that markets are predictable by alternative (2012) and Lee and Rui (2000) for a review of the international fundamental predictability literature.2 Some papers examine the relation between China and other aggregate markets. However, these papers focus on the Greater China markets or a sample of Asian countries, sometimes including the US. Hsiao, Ching, and Wan (2012) explore the impact on 24 countries from the economic integration of Hong Kong with mainland China. The focus in this literature is on contagion or other market linkage characteristics. A significant difference between this line of literature and our paper is that this line of literature does not explore out-of-sample forecasting. 3 We explore the predictability of the aggregate Chinese market. However, there are two papers that explore predictability of cross-sectional differences. First, Wang and Xu (2004) apply a three-factor model to A-shares in the Chinese stock market using data from July 1996-June 2002. They find that size is able to explain the cross-sectional differences. Contrary to the findings using US data, the beta and book-to-market ratio did not account for return differences across individual stocks. Second, Chen, Kim, Yao, and Yu (2010) consider 18 firm-specific variables that have been show to predict cross-sectional stock returns in the US, and examine their relation to stock returns (at the annual horizon) in China over the period 1995-2007. They find that only 5 of the 18 variables predict Chinese stock returns. The explanation for this finding of weak predictability that they find support for is that, i) return predictors in China are less heterogeneously distributed than they are in the US and ii) stock prices in China are less informative in China than they are in the US, in the sense that there are persistent noisy valuations and persiste...

show abstract

Section: Combination Predictor Modelsmentioning

confidence: 99%

Sticky prices or economically-linked economies: The case of forecasting the Chinese stock market

Jordan

Vivian

Wohar

2014

Journal of International Money and Finance

View full text Add to dashboard Cite

show abstract

“…Garratt et al (2009 also propose to combine the nowcast from a large number of models in a two-step procedure. 10 Our approach is close to Aiolfi and Timmermann (2006) in the sense that we combine models in more than one stage. They find that forecasting performance can be improved by first sorting models into clusters based on their past performance, second by pooling forecasts within each cluster, and third by estimating optimal weights on these clusters (followed by shrinkage towards equal weights).…”

Section: Forecast Frameworkmentioning

confidence: 96%

Nowcasting GDP in Real-Time: A Density Combination Approach

Aastveit

Gerdrup²,

Jore³

et al. 2011

SSRN Journal

View full text Add to dashboard Cite

In this paper we use U.S. real-time vintage data and produce combined density nowcasts for quarterly GDP growth from a system of three commonly used model classes. The density nowcasts are combined in two steps. First, a wide selection of individual models within each model class are combined separately. Then, the nowcasts from the three model classes are combined into a single predictive density. We update the density nowcast for every new data release throughout the quarter, and highlight the importance of new information for the evaluation period 1990Q2-2010Q3. Our results show that the logarithmic score of the predictive densities for U.S. GDP increase almost monotonically as new information arrives during the quarter. While the best performing model class is changing during the quarter, the density nowcasts from our combination framework is always performing well both in terms of logarithmic scores and calibration tests. The density combination approach is superior to a simple model selection strategy and also performs better in terms of point forecast evaluation than standard point forecast combinations.JEL-codes: C32, C52, C53, E37, E52.

show abstract

“…Cluster combinations or combining, as developed by Aiolfi and Timmermann (2006), is a conditional combining approach which incorporates information about the forecast persistence and historical performance of individual models (Rapach & Strauss, 2010). In this study, we specifically employ the Previous Best conditional combination strategy.…”

Section: Cluster Combinationsmentioning

confidence: 99%

The Role of Asset Prices in Forecasting Inflation and Output in South Africa

Gupta

Hartley

2013

Journal of Emerging Market Finance

View full text Add to dashboard Cite

This paper assesses the predictive ability of asset prices relative to other variables in forecasting inflation and real GDP growth in South Africa. A total of 42 asset and non-asset predictor variables are considered. Forecasts of inflation and real GDP growth are computed using both individual predictor autoregressive distributed lag (ARDL) models, forecast combination approaches, as well as, large scale models. The large scale data models considered include Bayesian vector autoregressive models, and classical and Bayesian univariate and multivariate factor augmented vector autoregressive models. The models are estimated for an in-sample of 1980:Q2 to 1999:Q4, and then one-to eight step-ahead forecasts for inflation and real GDP growth are evaluated over the 2000:Q1 to 2010:Q2 out-of-sample period. Principle Component forecast combination models are found to produce the most accurate out-of-sample forecasts of inflation and real GDP growth relative to the other combination and more sophisticated models considered. Asset prices are found to contain particularly useful information for forecasting inflation and real GDP growth at certain horizons. Asset prices are however found to be stronger predictors of inflation, particularly in the long run.

show abstract

Persistence in forecasting performance and conditional combination strategies

Cited by 293 publications

References 17 publications

Sticky prices or economically-linked economies: The case of forecasting the Chinese stock market

Sticky prices or economically-linked economies: The case of forecasting the Chinese stock market

Nowcasting GDP in Real-Time: A Density Combination Approach

The Role of Asset Prices in Forecasting Inflation and Output in South Africa

Contact Info

Product

Resources

About