Publications
Articles in Peer-Reviewed Journals
Mixed-Frequency Machine Learning: Nowcasting and Backcasting Weekly Initial Claims with Daily Internet Search Volume Data
International Journal of Forecasting, forthcoming (with Daniel Borup and Eric Christian Montes Schütte)
Abstract: We propose an out-of-sample prediction approach that combines unrestricted mixed-data sampling with machine learning (mixed-frequency machine learning, MFML). We use the MFML approach to generate a sequence of nowcasts and backcasts of weekly unemployment insurance initial claims based on a rich trove of daily Google Trends search volume data for terms related to unemployment. The predictions are based on linear models estimated via the LASSO and elastic net, nonlinear models based on artificial neural networks, and ensembles of linear and nonlinear models. Nowcasts and backcasts of weekly initial claims based on models that incorporate the information in the daily Google Trends search volume data substantially outperform those based on models that ignore the information. Predictive accuracy increases as the nowcasts and backcasts include more recent daily Google Trends data. The relevance of daily Google Trends data for predicting weekly initial claims is strongly linked to the COVID-19 crisis.
[SSRN link]
Asset Pricing: Time-Series Predictability (open access)
Oxford Research Encyclopedia of Economics and Finance, June 20, 2022 (with Guofu Zhou)
Summary: Asset returns change with fundamentals and other factors such as technical information and sentiment over time. This review covers some of the major ideas, data, and methods used to model time-varying expected returns. The focus is on the out-of-sample predictability of the aggregate stock market return via extensions of the conventional predictive regression approach.
The extensions are designed to improve out-of-sample performance in realistic environments characterized by large information sets and noisy data. Large information sets are relevant because a plethora of plausible stock return predictors exists. The information sets include variables typically associated with a rational time-varying market risk premium, as well as variables more likely to reflect market inefficiencies resulting from behavioral influences and information frictions. Noisy data stem from the intrinsically large unpredictable component in stock returns. When forecasting with large information sets and noisy data, it is vital to employ methods that incorporate the relevant information in the large set of predictors in a manner that guards against overfitting the data.
Methods that improve out-of-sample market return prediction include forecast combination, principal component regression, partial least squares, the LASSO and elastic net from machine learning, and a newly developed C-ENet approach that relies on the elastic net to refine the simple combination forecast. Employing these methods, a number of studies provide statistically and economically significant evidence that the aggregate market return is predictable on an out-of-sample basis. Out-of-sample market return predictability based on a rich set of predictors thus appears to be a well-established empirical result in asset pricing.
[SSRN link]
Forecasting: Theory and Practice (open access)
International Journal of Forecasting, 2022, 38(3), 705–871 (with 79 co-authors)
Abstract: Forecasting has always been at the forefront of decision making and planning. The uncertainty that surrounds the future is both exciting and challenging, with individuals and organisations seeking to minimise risks and maximise utilities. The large number of forecasting applications calls for a diverse set of forecasting methods to tackle real-life challenges. This article provides a non-systematic review of the theory and the practice of forecasting. We provide an overview of a wide range of theoretical, state-of-the-art models, methods, principles, and approaches to prepare, produce, organise, and evaluate forecasts. We then demonstrate how such theoretical concepts are applied in a variety of real-life contexts.
We do not claim that this review is an exhaustive list of methods and applications. However, we wish that our encyclopedic presentation will offer a point of reference for the rich work that has been undertaken over the last decades, with some key insights for the future of forecasting theory and practice. Given its encyclopedic nature, the intended mode of reading is non-linear. We offer cross-references to allow the readers to navigate through the various topics. We complement the theoretical concepts and applications covered by large lists of free or open-source software implementations and publicly-available databases.
[My entries: Machine Learning with (Very) Noisy Data (Theory); Forecasting Stock Returns (Practice) | arXiv link]
Anomalies and the Expected Market Return
Journal of Finance, 2022, 77(1), 639—681 (with Xi Dong, Yan Li, and Guofu Zhou)
Abstract: We provide the first systematic evidence on the link between long-short anomaly portfolio returns—a cornerstone of the cross-sectional literature—and the time-series predictability of the aggregate market excess return. Using 100 representative anomalies from the literature, we employ a variety of shrinkage techniques (including machine learning, forecast combination, and dimension reduction) to efficiently extract predictive signals in a high-dimensional setting. We find that long-short anomaly portfolio returns evince statistically and economically significant out-of-sample predictive ability for the market excess return. The predictive ability of anomaly portfolio returns appears to stem from asymmetric limits of arbitrage and overpricing correction persistence.
[SSRN link | Internet Appendix | Data/R files]
Industry Return Predictability: A Machine Learning Approach
Journal of Financial Data Science, 2019, 1(3), 9–28 (with Jack K. Strauss, Jun Tu, and Guofu Zhou)
Abstract: In this article, the authors use machine learning tools to analyze industry return predictability based on the information in lagged industry returns. Controlling for post-selection inference and multiple testing, they find significant in-sample evidence of industry return predictability. Lagged returns for the financial sector and commodity- and material-producing industries exhibit widespread predictive ability, consistent with the gradual diffusion of information across economically linked industries. Out-of-sample industry return forecasts that incorporate the information in lagged industry returns are economically valuable: Controlling for systematic risk using leading multifactor models from the literature, an industry-rotation portfolio that goes long (short) industries with the highest (lowest) forecasted returns delivers an annualized alpha of over 8%. The industry-rotation portfolio also generates substantial gains during economic downturns, including the Great Recession.
[Online Supplement | Data/R files | Alpha Architect blog post by Druce Vertes | ASX Prize for the Best Paper in Quantitative Finance and Derivatives, 28th Australasian Finance and Banking Conference]
Metro Business Cycles
Journal of Urban Economics, 2016, 94, 90–108 (with Maria A. Arias and Charles S. Gascon)
Abstract: We construct monthly economic activity indices for the 50 largest US metropolitan statistical areas (MSAs) beginning in 1990. Each index is derived from a dynamic factor model based on twelve underlying variables capturing various aspects of metro area economic activity. To accommodate mixed-frequency data and differences in data-publication lags, we estimate the dynamic factor model using a maximum-likelihood approach that allows for arbitrary patterns of missing data. Our indices highlight important similarities and differences in business cycles across MSAs. While a number of MSAs experience sizable recessions during the national recessions of the early 1990s and early 2000s, other MSAs escape recessions altogether during one or both of these periods. Nearly all MSAs suffer relatively deep recessions near the recent Great Recession, but we still find significant differences in the depth of recent metro recessions. We relate the severity of metro recessions to a variety of MSA characteristics and find that MSAs with less-educated populations and less elastic housing supplies experience significantly more severe recessions. After controlling for national economic activity, we also find significant evidence of dynamic spillover effects in economic activity across MSAs.
[Monthly MSA economic activity indices from the paper available on FRED | Forbes column by Adam Millsap | Citylab column (from The Atlantic) by Richard Florida | St. Louis Post-Dispatch story by Jim Gallagher]
Short Interest and Aggregate Stock Returns
Journal of Financial Economics, 2016, 121(1), 46–65 (with Matthew C. Ringgenberg and Guofu Zhou)
Abstract: We show that short interest is arguably the strongest known predictor of aggregate stock returns. It outperforms a host of popular return predictors both in and out of sample, with annual R-squared statistics of 12.89% and 13.24%, respectively. In addition, short interest can generate utility gains of over 300 basis points per annum for a mean-variance investor. A vector autoregression decomposition shows that the economic source of short interest’s predictive power stems predominantly from a cash flow channel. Overall, our evidence indicates that short sellers are informed traders who are able to anticipate future aggregate cash flows and associated market returns.
Updated short interest index (SII) data for 1973:01–2021:12
[Online Appendix | Data/MATLAB files | Wall Street Journal column by Mark Hulbert | CFA Digest article by Nitin Joshi | MarketWatch column by Mark Hulbert | Dallas Morning News column by Will Deener | Bloomberg Business article by Sam Mamudi and Saijel Kishan | American Association of Individual Investors Journal story | Seeking Alpha blog post by Fred Piard | Alpha Architect blog post by Wesley Gray]
Return Predictability and Dynamic Asset Allocation: How Often Should Investors Rebalance?
Journal of Portfolio Management, 2014, 40(4), 16–27 (with Himanshu Almadi and Anil Suri)
Abstract: To exploit return predictability via dynamic asset allocation, investors face the important practical issue of how often to rebalance their portfolios. More frequent rebalancing uses statistically and economically significant short-horizon return predictability to aggressively pursue the dynamic investment opportunities afforded by changes in expected returns. However, the degree of return predictability typically appears stronger at longer horizons, which, along with lower transaction costs, favors less frequent rebalancing. The authors analyze the performance effects of rebalancing frequency in the context of dynamic portfolios constructed from monthly, quarterly, semi-annual, and annual return forecasts for U.S. stocks, bonds, and bills, where the dynamic portfolios rebalance at the same frequency as the forecast horizon. Along the transaction-cost/rebalancing frontier, monthly (annual) rebalancing provides the greatest outperformance when unit transaction costs are below (above) approximately 50 basis points, and dynamic portfolios based on annual rebalancing typically outperform the benchmarks for unit transaction costs well in excess of 400 basis points.
[CFA Digest article by Gregory G. Gocek]
Forecasting the Equity Risk Premium: The Role of Technical Indicators
Management Science, 2014, 60(7), 1772–1791 (with Christopher J. Neely, Jun Tu, and Guofu Zhou)
Abstract: Academic research relies extensively on macroeconomic variables to forecast the U.S. equity risk premium, with relatively little attention paid to the technical indicators widely employed by practitioners. Our paper fills this gap by comparing the predictive ability of technical indicators with that of macroeconomic variables. Technical indicators display statistically and economically significant in-sample and out-of-sample predictive power, matching or exceeding that of macroeconomic variables. Furthermore, technical indicators and macroeconomic variables provide complementary information over the business cycle: technical indicators better detect the typical decline in the equity risk premium near business-cycle peaks, whereas macroeconomic variables more readily pick up the typical rise in the equity risk premium near cyclical troughs. Consistent with this behavior, we show that combining information from both technical indicators and macroeconomic variables significantly improves equity risk premium forecasts versus using either type of information alone. Overall, the substantial countercyclical fluctuations in the equity risk premium appear well captured by the combined information in technical indicators and macroeconomic variables.
[Online Appendix | Data/MATLAB files | Business Forecasting blog post by Clive Jones | CBS MoneyWatch column by Larry Swedroe]
An Intra-Week Efficiency Analysis of Bookie-Quoted NFL Betting Lines in NYC
Journal of Empirical Finance, 2013, 24(3), 10–23 (with Thomas W. Miller)
Abstract: We analyze the intra-week evolution of bookie-quoted National Football League betting lines in NewYork City and its implications for market efficiency. Our unique data set includes three sequential lines: (i) an outlaw line set by a single agent at the beginning of the week; (ii) Tuesday's opening line shaped by bets from a group of eight to ten agents; and (iii) a game-time closing line shaped by the wider public. While forecast encompassing tests show that information content increases during the betting week, consistent with a reasonably well-functioning market, we also uncover significant evidence of pricing inefficiencies relating to sentiment measures. In addition, actual bets made by a number of professional gamblers appear profitable, pointing to the existence of superior analysts.
[Alpha Architect blog post by Wesley Gray]
International Stock Return Predictability: What is the Role of the United States?
Journal of Finance, 2013, 68(4), 1633–1662 (with Jack K. Strauss and Guofu Zhou)
Abstract: We investigate lead-lag relationships among monthly country stock returns and identify a leading role for the United States: lagged US returns significantly predict returns in numerous non-US industrialized countries, while lagged non-US returns display limited predictive ability with respect to US returns. We estimate a news-diffusion model, and the results indicate that return shocks arising in the United States are only fully reflected in equity prices outside of the United States with a lag, consistent with a gradual information diffusion explanation of the predictive power of lagged US returns.
[Internet Appendix | Data/MATLAB files | CFA Digest article by Clifford S. Ang]
Forecasting US State-Level Employment Growth: An Amalgamation Approach
International Journal of Forecasting, 2012, 28(2), 315–327 (with Jack K. Strauss)
Abstract: We forecast US state-level employment growth using several distinct econometric approaches: combinations of individual autoregressive distributed lag models, general-to-specific modeling with bootstrap aggregation (GETS-bagging), and approximate factor (or ‘‘beta’’) models. Our results show that these forecasting approaches consistently deliver sizable reductions in mean squared forecast error (MSFE) relative to an autoregressive (AR) benchmark model across the 50 US states. On the basis of forecast encompassing test results, we also consider amalgamating these approaches and find that this strategy yields additional forecasting improvements. These improvements are particularly evident during national business-cycle recessions, where the amalgamation approach outperforms the AR benchmark for nearly all states and leads to a 40% reduction in MSFE on average across states relative to the AR benchmark.
[Complete reported/unreported results]
International Comovements in Inflation and Country Characteristics
Journal of International Money and Finance, 2011, 30(7), 1471–1490 (with Christopher J. Neely)
Abstract: Common shocks, similarities in central bank reaction functions, and international trade potentially produce common components in international inflation rates. This paper characterizes such links in international inflation rates with a dynamic latent factor model that decomposes 64 national inflation rates into world, regional, and idiosyncratic components. The world and regional components account for 35% and 16%, respectively, of annual inflation variability on average across countries, so that international influences together explain just over half of inflation variability. The importance of the world and regional components, however, differs substantially across countries. Economic policy choices and development measures strongly explain the cross-sectional variation in the relative importance of international influences. A subsample analysis reveals that the regional (world) factor increases in importance for a number of North American and European (Latin American and Asian) countries since 1980.
Predicting Market Components Out of Sample: Asset Allocation Implications
Journal of Portfolio Management, 2011, 37(4), 29–41 (with Aiguo Kong, Jack K. Strauss, and Guofu Zhou)
Abstract: The authors analyze out-of-sample return predictability for components of the aggregate market, focusing on the well-known Fama–French size/value-sorted portfolios. Employing a forecast combination approach based on a variety of economic variables and lagged component returns as predictors, they find significant evidence of out-of-sample return predictability for nearly all component portfolios. Moreover, return predictability is typically much stronger for small-cap/high book-to-market value stocks. The pattern of component return predictability is enhanced during business cycle recessions, linking component return predictability to the real economy. Considering various component-rotation investment strategies, the authors show that out-of-sample component return predictability can be exploited to substantially improve portfolio performance.
How Predictable is the Chinese Stock Market?
Journal of Financial Research (in Chinese), 2011, 9, 107–121 (with Fuwei Jiang, Jack K. Strauss, Jun Tu, and Guofu Zhou)
[Outstanding Paper Award (Third Prize), Journal of Financial Research (2011) | The Chinese Finance Association Best Paper Award in Investment (2010)]
Bagging or Combining (or Both)? An Analysis Based on Forecasting US Employment Growth
Econometric Reviews, 2010, 29(5-6), 511–533 (with Jack K. Strauss)
Abstract: Forecasting a macroeconomic variable is challenging in an environment with many potential predictors whose predictive ability can vary over time. We compare two approaches to forecasting US employment growth in this type of environment. The first approach applies bootstrap aggregating (bagging) to a general-to-specific procedure based on a general dynamic linear regression model with 30 potential predictors. The second approach considers several methods for combining forecasts from 30 individual autoregressive distributed lag (ARDL) models, where each individual ARDL model contains a potential predictor. We analyze bagging and combination forecasts at multiple horizons over four different out-of-sample periods using a mean square forecast error (MSFE) criterion and forecast encompassing tests. We find that bagging forecasts often deliver the lowest MSFE. Interestingly, we also find that incorporating information from both bagging and combination forecasts based on principal components often leads to further gains in forecast accuracy.
Out-of-Sample Equity Premium Prediction: Combination Forecasts and Links to the Real Economy
Review of Financial Studies, 2010, 23(2), 821–862 (with Jack K. Strauss and Guofu Zhou)
Abstract: Welch and Goyal (2008) find that numerous economic variables with in-sample predictive ability for the equity premium fail to deliver consistent out-of-sample forecasting gains relative to the historical average. Arguing that model uncertainty and instability seriously impair the forecasting ability of individual predictive regression models, we recommend combining individual forecasts. Combining delivers statistically and economically significant out-of-sample gains relative to the historical average consistently over time. We provide two empirical explanations for the benefits of forecast combination: (i) combining forecasts incorporates information from numerous economic variables while substantially reducing forecast volatility; (ii) combination forecasts are linked to the real economy.
[Among RFS Top Cited Papers published in 2010]
Multi-Period Portfolio Choice and the Intertemporal Hedging Demands for Stocks and Bonds: International Evidence
Journal of International Money and Finance, 2009, 28(3), 427–453 (with Mark E. Wohar)
Abstract: We investigate the intertemporal hedging demands for stocks and bonds for investors in the US, Australia, Canada, France, Germany, Italy, and UK. Using the methodology of Campbell et al. [Campbell, J.Y., Chan, Y.L., Viceira, L.M., 2003a. A multivariate model of strategic asset allocation. Journal of Financial Economics 67(1), 41–81], we solve a multi-period portfolio choice problem for an investor in each country with an infinite horizon and Epstein–Zin–Weil utility, where the dynamics governing asset returns are described by a vector autoregressive process. We find sizable mean intertemporal hedging demands for domestic stocks in the US and UK and considerably smaller mean hedging demands for domestic stocks in the other countries. An investor in the US who has access to foreign stocks and bonds displays small mean intertemporal hedging demands for foreign stocks and bonds, while investors in Australia, Canada, France, Germany, Italy, and the UK who have access to US stocks and bonds all exhibit sizable mean hedging demands for US stocks.
[Data Appendix | Appendix with VAR Estimation Results | Data/GAUSS files]
Differences in Housing Price Forecastability Across US States
International Journal of Forecasting, 2009, 25(2), 351–372 (with Jack K. Strauss)
Abstract: Given the marked differences in housing price growth across US regions since the mid-1990s, we investigate forecasts of state-level real housing price growth for 1995–2006. We evaluate forecasts from an autoregressive benchmark model as well as models based on a host of state, regional, and national economic variables. Overall, our results highlight important differences in the forecastability of real housing price growth across US states, especially between interior and coastal states. More specifically, we find that autoregressive models, and especially models that incorporate information from numerous economic variables, often provide relatively accurate housing price forecasts for a number of interior states during the period 1995–2006; all forecasting models, however, tend to perform relatively poorly for a group of primarily coastal states that experienced especially strong housing price growth during this period, pointing to a “disconnect” between housing prices and economic fundamentals for these states.
States and the Business Cycle
Journal of Urban Economics, 2009, 65(2), 181–194 (with Michael T. Owyang and Howard J. Wall)
Abstract: We model the US business cycle using a dynamic factor model that identifies common factors underlying fluctuations in state-level income and employment growth. We find three such common factors, each of which is associated with a set of factor loadings that indicate the extent to which each state’s economy is related to the national business cycle. According to the factor loadings, there is a great deal of heterogeneity in the nature of the links between state and national economies. In addition to exhibiting geographic patterns, the closeness of state economies to the national business cycle is related not only to differences in industry mix but also to non-industry variables such as agglomeration and neighbor effects. Finally, we find that the common factors tend to explain large proportions of the total variability in state-level business cycles, although, again, there is a great deal of cross-state heterogeneity.
Structural Breaks and GARCH Models of Exchange Rate Volatility
Journal of Applied Econometrics, 2008, 23(1), 65–90 (with Jack K. Strauss)
Abstract: We investigate the empirical relevance of structural breaks for GARCH models of exchange rate volatility using both in-sample and out-of-sample tests. We find significant evidence of structural breaks in the unconditional variance of seven of eight US dollar exchange rate return series over the 1980–2005 period—implying unstable GARCH processes for these exchange rates—and GARCH(1,1) parameter estimates often vary substantially across the subsamples defined by the structural breaks. We also find that it almost always pays to allow for structural breaks when forecasting exchange rate return volatility in real time. Combining forecasts from different models that accommodate structural breaks in volatility in various ways appears to offer a reliable method for improving volatility forecast accuracy given the uncertainty surrounding the timing and size of the structural breaks.
[Appendix with Additional Results | Data/GAUSS files | Third most downloaded article from JAE in 2008]
Forecasting US Employment Growth Using Forecast Combining Methods
Journal of Forecasting, 2008, 27(1), 75–93 (with Jack K. Strauss)
Abstract: We examine different approaches to forecasting monthly US employment growth in the presence of many potentially relevant predictors. We first generate simulated out-of-sample forecasts of US employment growth at multiple horizons using individual autoregressive distributed lag (ARDL) models based on 30 potential predictors. We then consider different methods from the extant literature for combining the forecasts generated by the individual ARDL models. Using the mean square forecast error (MSFE) metric, we investigate the performance of the forecast combining methods over the last decade, as well as five periods centered on the last five US recessions. Overall, our results show that a number of combining methods outperform a benchmark autoregressive model. Combining methods based on principal components exhibit the best overall performance, while methods based on simple averaging, clusters, and discount MSFE also perform well. On a cautionary note, some combining methods, such as those based on ordinary least squares, often perform quite poorly.
[Appendix with Additional Results]
Forecasting the Recent Behavior of US Business Fixed Investment Spending: An Analysis of Competing Models
Journal of Forecasting, 2007, 26(1), 33–51 (with Mark E. Wohar)
Abstract: We evaluate forecasting models of US business fixed investment spending growth over the recent 1995:1–2004:2 out-of-sample period. The forecasting models are based on the conventional Accelerator, Neoclassical, Average Q, and Cash-Flow models of investment spending, as well as real stock prices and excess stock return predictors. The real stock price model typically generates the most accurate forecasts, and forecast-encompassing tests indicate that this model contains most of the information useful for forecasting investment spending growth relative to the other models at longer horizons. In a robustness check, we also evaluate the forecasting performance of the models over two alternative out-of-sample periods: 1975:1–1984:4 and 1985:1–1994:4. A number of different models produce the most accurate forecasts over these alternative out-of-sample periods, indicating that while the real stock price model appears particularly useful for forecasting the recent behavior of investment spending growth, it may not continue to perform well in future periods.
Structural Breaks and Predictive Regression Models of US Stock Returns
Journal of Financial Econometrics, 2006, 4(2), 238–274 (with Mark E. Wohar)
Abstract: In this article we examine the structural stability of predictive regression models of US quarterly aggregate real stock returns over the postwar era. We consider predictive regressions models of S&P 500 and CRSP equal-weighted real stock returns based on eight financial variables that display predictive ability in the extant literature. We test for structural stability using the popular Andrews SupF statistic and the Bai subsample procedure in conjunction with the Hansen heteroskedastic fixed-regressor bootstrap. We also test for structural stability using the recently developed methodologies of Elliott and Müller, and Bai and Perron. We find strong evidence of structural breaks in five of eight bivariate predictive regression models of S&P 500 returns and some evidence of structural breaks in the three other models. There is less evidence of structural instability in bivariate predictive regression models of CRSP equal-weighted returns, with four of eight models displaying some evidence of structural breaks. We also obtain evidence of structural instability in a multivariate predictive regression model of S&P 500 returns. When we estimate the predictive regression models over the different regimes defined by structural breaks, we find that the predictive ability of financial variables can vary markedly over time.
The Out-of-Sample Forecasting Performance of Nonlinear Models of Real Exchange Rate Behavior
International Journal of Forecasting, 2006, 22(2), 341–361 (with Mark E. Wohar)
Abstract: We analyze the out-of-sample forecasting performance of nonlinear models of US dollar real exchange rate behavior from the extant empirical literature. Our analysis entails a comparison of point, interval, and density forecasts generated by nonlinear and linear autoregressive models. Using monthly data from the post-Bretton Woods period, there is little evidence to recommend either band-threshold or exponential smooth transition autoregressive models over simple linear autoregressive models in terms of out-of-sample forecasting performance at short horizons. Nonlinear models appear to offer more accurate point forecasts at long horizons for some countries. Overall, our results suggest that any nonlinearities in monthly real exchange rate data from the post-Bretton Woods period are quite “subtle” for band-threshold and exponential smooth transition autoregressive model specifications. Further evidence of this is provided by in-sample comparisons of the conditional densities implied by nonlinear and linear autoregressive models.
In-Sample vs Out-of-Sample Tests of Stock Return Predictability in the Context of Data Mining
Journal of Empirical Finance, 2006, 13(2), 231–247 ( with Mark E. Wohar)
Abstract: We undertake an extensive analysis of in-sample and out-of-sample tests of stock return predictability in an effort to better understand the nature of the empirical evidence on return predictability. We find that a number of financial variables appearing in the literature display both in-sample and out-of-sample predictive ability with respect to stock returns in annual data covering most of the twentieth century. In contrast to the extant literature, we demonstrate that there is little discrepancy between in-sample and out-of-sample test results once we employ out-of-sample tests with good power. While conventional wisdom holds that out-of-sample tests help guard against data mining, Inoue and Kilian [Inoue, A., Kilian, L., 2004. In-sample or out-of-sample tests of predictability: which one should we use? Econometric Reviews 23, 371–402.] recently argue that in-sample and out-of-sample tests are equally susceptible to data mining biases. Using a bootstrap procedure that explicitly accounts for data mining, we still find that certain financial variables display significant in-sample and out-of-sample predictive ability with respect to stock returns.
[Data Appendix | Appendix with Additional Results | Data/GAUSS files]
Regime Changes in International Real Interest Rates: Are They a Monetary Phenomenon?
Journal of Money, Credit, and Banking, 2005, 37(5), 887–906 (with Mark E. Wohar)
Abstract: In this paper, we use the Bai and Perron (1998, 2001, 2003) methodology to test for multiple structural breaks in the mean real interest rate for 13 industrialized countries. We find extensive evidence of structural breaks in the mean real interest rate for all 13 countries. In an attempt to explain the breaks in international real interest rates, we also test for multiple structural breaks in the mean inflation rate for the 13 countries. Once again, we find extensive evidence of structural breaks in the mean inflation rate for all of the countries. Interestingly, the breaks in inflation rates and real interest rates often coincide, with increases (decreases) in the mean inflation rate as we move from one regime to the next typically associated with decreases (increases) in the mean real interest rate.
Valuation Ratios and Long-Horizon Stock Price Predictability
Journal of Applied Econometrics, 2005, 20(3), 327–344 (with Mark E. Wohar)
Abstract: Using annual data for 1872–1997, this paper re-examines the predictability of real stock prices based on price-dividend and price-earnings ratios. In line with the extant literature, we find significant evidence of increased long-horizon predictability; that is, the hypothesis that the current value of a valuation ratio is uncorrelated with future stock price changes cannot be rejected at short horizons but can be rejected at longer horizons based on bootstrapped critical values constructed from linear representations of the data. While increased statistical power at long horizons in finite samples provides a possible explanation for the pattern of predictability in the data, we find via Monte Carlo simulations that the power to detect predictability in finite samples does not increase at long horizons in a linear framework. An alternative explanation for the pattern of predictability in the data is nonlinearities in the underlying data-generating process. We consider exponential smooth-transition autoregressive models of the price-dividend and price-earnings ratios and their ability to explain the pattern of stock price predictability in the data.
[Data/GAUSS files | Lead article]
Macro Variables and International Stock Return Predictability
International Journal of Forecasting, 2005, 21(1), 137–166 (with Mark E. Wohar and Jasper Rangvid)
Abstract: In this paper, we examine the predictability of stock returns using macroeconomic variables in 12 industrialized countries. We consider both in-sample and out-of-sample tests of predictive ability, with the out-of-sample forecast period covering the 1990s for each country. We employ recently developed out-of-sample tests that have increased power, namely, the McCracken [Asymptotics for out-of-sample tests of Granger Causality, Manuscript, University of Missouri-Columbia (2004)] variant of the Diebold and Mariano [Journal of Economics Business Statistics 13 (1995) 253] and West [Econometrica 64 (1996) 1067] test for equal predictive ability and the Clark and McCracken [Journal of Econometrics 105 (2001) 85] variant of the Harvey, Leybourne, and Newbold [Journal of Business and Economics Statistics 16 (1998) 254] test for forecast encompassing. In addition to analyzing the predictive ability of each macro variable in turn, we use a procedure that combines general-to-specific model selection with out-of-sample tests of forecasting ability in an effort to identify and test the ‘‘best’’ forecasting model of stock returns in each country. Among the macro variables we consider, interest rates are the most consistent and reliable predictors of stock returns across countries.
The Persistence in International Real Interest Rates
International Journal of Finance and Economics, 2004, 9(4), 339–346 (with Mark E. Wohar)
Abstract: In this paper, we investigate the degree of persistence in quarterly postwar tax-adjust ex post real interest rates for 13 industrialized countries using two recently developed econometric procedures. Our results show that international tax-adjusted real interest rates are typically very persistent, with the lower bound of the 95% confidence interval for the sum of the autoregressive coefficients very close to 0.90 for nearly every country. A highly persistent real interest rate has important theoretical implications.
Financial Variables and the Simulated Out-of-Sample Forecastability of US Output Growth Since 1985: An Encompassing Approach
Economic Inquiry, 2004, 42(4), 717–738 (with Christian E. Weber)
Abstract: We reconsider the out-of-sample forecasting ability of a large number of financial variables with respect to real output growth over the 1985:1–1999:4 period. We show that models including financial variables display almost no forecasting ability relative to an autoregressive benchmark model over this period according to a mean squared forecast error metric. However, tests based on forecast encompassing indicate that many financial variables do, in fact, contain information that is useful for forecasting real output growth over the 1985:1–1999:4 out-of-sample period. Our results suggest that the extant literature exaggerates the demise of the forecasting power of financial variables with respect to real activity since the mid-1980s.
Testing the Monetary Model of Exchange Rate Determination: A Closer Look at Panels
Journal of International Money and Finance, 2004, 23(6), 841–865 (with Mark E. Wohar)
Abstract: In this paper, we undertake an extensive evaluation of panel tests of the long-run monetary model of exchange rate determination. We first show how poorly the monetary model performs on a country-by-country basis for US dollar exchange rates over the post-Bretton Woods period for a large number of industrialized countries. In sharp contrast, we find considerable support for the monetary model using panel procedures, as in Groen (2000) and Mark and Sul (2001). Given the disparity in the country-by-country and panel approaches, we carefully analyze the homogeneity restrictions inherent in the panel procedures. The evidence on the appropriateness of the homogeneity restrictions is mixed. In the end, whether the monetary model conforms to post-Bretton Woods data largely depends on one’s prior beliefs.
Are Real Interest Rates Really Nonstationary? New Evidence from Tests with Good Size and Power
Journal of Macroeconomics, 2004, 26(3), 409–430 (with Christian E. Weber)
Abstract: In this paper, we re-examine the stationarity of international real interest rates, an issue first investigated by Rose [Journal of Finance 43(5) (1988) 1095], using a new set of unit root tests developed by Ng and Perron [Econometrica 69(6) (2001) 1519] with good size and power. Using conventional unit root tests, Rose finds that the nominal interest rate is I(1), while the inflation rate is I(0), for each of the many countries he considers, indicating a nonstationary real interest rate for each country. Using an extended sample period and the Ng and Perron unit root tests, we find that the nominal interest rate is I(1) and the inflation rate is I(0) for only three of the 16 countries we examine. For a number of countries, the Ng and Perron tests indicate that the nominal interest rate and inflation rate are both I(1), so that we need to test for cointegration in order to decipher the integration properties of the real interest rate. Using either the Ng and Perron unit root tests in conjunction with a pre-specified cointegrating vector or the Perron and Rodriguez [Residual based tests for cointegration with GLS detrended data (2001)] cointegration tests for an unspecified cointegrating vector, there is little robust evidence of cointegration. While our results are mixed, they usually provide support for the Rose finding that international real interest rates are nonstationary, albeit often for different reasons than Rose.
International Evidence of the Long-Run Impact of Inflation
Journal of Money, Credit, and Banking, 2003, 35(1), 23–48
Abstract: In this paper. I use a structural vector autoregression framework to analyze the effects of a permanent change in inflation on the long-run real interest rate and real output level in 14 industrialized countries. Long-run monetary superneutrality iu rejected for all 14 countries using annual data: the results indicate that a permanent increase in inflation lowers the long-run real interest rate in each country: a permanent increase in inflation also increases the long-run real output level in a number of countries. Long-run monetary superneutrality is also rejected for four out of the five countries examined uslng quarterly data.
Testing the Monetary Model of Exchange Rate Determination: New Evidence from a Century of Data
Journal of International Economics, 2002, 58(2), 359–385 (with Mark E. Wohar)
Abstract: We test the long-run monetary model of exchange rate determination for a collection of 14 industrialized countries using data spanning the late nineteenth or early twentieth century to the late twentieth century. Interestingly, we find support for a simple form of the long-run monetary model in over half of the countries we consider. For these countries, we estimate vector error-correction models to investigate the adjustment process to the long-run monetary equilibrium. In the spirit of Meese and Rogoff [Journal of International Economics 14 (1983) 3–24], we also compare nominal exchange rate forecasts based on monetary fundamentals to those based on a naïıve random walk model.
The Long-Run Relationship Between Inflation and Real Stock Prices
Journal of Macroeconomics, 2002, 24(3), 331–351
Abstract: In this paper, we use recent developments in the testing of long-run neutrality propositions to measure the long-run response of real stock prices to a permanent inflation shock for 16 individual industrialized countries. The estimation results provide considerable support for long-run inflation neutrality with respect to real stock prices. Ranges of plausible identifying parameter values are also consistent with a positive long-run real stock price response to a permanent inflation shock. There is little plausible evidence for a negative long-run real stock price response to a permanent inflation shock. Overall, our results indicate that inflation does not erode the long-run real value of stocks.
Are Real GDP Levels Nonstationary? Evidence from Panel Data Tests
Southern Economic Journal, 2002, 68(3), 473–495
Abstract: Ever since the seminal paper of Nelson and Plosser (1982), researchers have focused on the potential nonstationarity of important macroeconomic variables, and unit root tests are now a standard procedure in empirical analyses. While there are many findings of unit roots in macroeconomic variables using the popular augmented Dickey and Fuller (1979) test, this test has low power against near-unit-root alternatives. Recently, panel data procedures have been proposed as an avenue to increased power. This paper applies panel unit root tests to international real GDP and real GDP per capita data. The results overwhelmingly indicate that international real GDP and real GDP per capita levels are nonstationary.
[Lead article]
Monetary Shocks and Real Exchange Rate Hysteresis: Evidence from the G-7 Countries
Review of International Economics, 2001, 9(2), 356–371
Abstract: Long-run monetary neutrality specifies that nominal disturbances do not affect long-run real exchange rates. However, the “over-depreciation” of the US dollar in the late 1980s, after its strong appreciation earlier in the decade, suggested to a number of observers that nominal disturbances alter long-run real exchange rates; that is, money supply shocks entail real exchange rate hysteresis. Using data from the G-7 countries and the post-1973 float, the paper measures the long-run effects of relative money supply disturbances on real US dollar exchange rates. Little evidence of hysteretic money policy effects is found.
Macro Shocks and Real Stock Prices
Journal of Economics and Business, 2001, 53(1), 5–26
Abstract: In this paper, I examine the effects of money supply, aggregate spending, and aggregate supply shocks on real US stock prices in a structural vector autoregression framework. Overall, the empirical results indicate that each macro shock has important effects on real stock prices. The real stock price impulse responses to the various macro shocks conform to the standard present-value equity valuation model, and they shed considerable light on the well-known negative correlation between real stock returns and inflation. An historical decomposition indicates that the late 1990s surge in real stock prices is due to a series of favorable structural shocks emanating from different sectors of the US economy.
[Lead article]
Macro Shocks and Fluctuations
Journal of Economics and Business, 1998, 50(1), 23–38
Abstract: This paper assesses the importance of money supply, money demand, real spending, and supply shocks in explaining short-run real output fluctuations. The framework for the analysis is a structural vector autoregression model, in which long-run restrictions identify the above shocks. Results indicate that spending and supply shocks are mainly responsible for fluctuations, while monetary shocks play a very limited role. These findings are compared to other recent studies of fluctuations.
Monetary Shocks and Relative Farm Prices: A Re-examination
American Journal of Agricultural Economics, 1997, 79(4), 1332–1339 (with Alan G. Isaac)
Abstract: The effect of monetary policy on the farm sector remains controversial. Studies of the effects of monetary disturbances on relative farm prices report conflicting results: some find that positive monetary shocks increase relative farm prices in the short run, and others detect no such effect. We offer a resolution of these conflicting findings by reestimating existing models on a common data set. When sample periods corresponding to the original studies are used, the conflicting results are confirmed. In contrast, when samples are updated through 1993, all models supply the same result: monetary shocks do not affect relative farm prices.
Articles in Federal Reserve Publications
Common Fluctuations in OECD Budget Balances
Federal Reserve Bank of St. Louis Review, 2015, 97(2), 109–132 (with Christopher J. Neely)
Abstract: The authors use a dynamic latent factor model to analyze comovements in OECD surpluses. The world factor underlying common fluctuations in budget surpluses across countries explains an average of 28 to 44 percent of the variation in individual country surpluses. The world factor, which can be interpreted as a global budget surplus index, declines substantially in the 1980s, rises throughout much of the 1990s, peaks in 2000, and declines again after the financial crisis of 2008. The authors then estimate similar world factors in national output gaps, dividend-to-price ratios, and military spending that significantly explain the variation in the world budget surplus factor. Idiosyncratic components of national budget surpluses correlate with well-known “unusual” country circumstances, such as the Swedish banking crisis of the early 1990s.
Real Interest Rate Persistence: Evidence and Implications
Federal Reserve Bank of St. Louis Review, 2008, 90(6), 609–643 (with Christopher J. Neely)
Abstract: The real interest rate plays a central role in many important financial and macroeconomic models, including the consumption-based asset pricing model, neoclassical growth model, and models of the monetary transmission mechanism. The authors selectively survey the empirical literature that examines the time-series properties of real interest rates.
Forecasting Real Housing Price Growth in the Eighth District States
Federal Reserve Bank of St. Louis Regional Economic Development, 2007, 3(2), 33–42 (with Jack K. Strauss)
Abstract: The authors consider forecasting real housing price growth for the individual states of the Federal Reserve's Eighth District. They first analyze the forecasting ability of a large number of potential predictors of state real housing price growth using an autoregressive distributed lag (ARDL) model framework. A number of variables, including the state housing price-to-income ratio, state unemployment rate, and national inflation rate, appear to provide information that is useful for forecasting real housing price growth in many Eighth District states. Given that it is typically difficult to determine a priori the particular variable or small set of variables that are the most relevant for forecasting real housing price growth for a given state and time period, the authors also consider various methods for combining the individual ARDL model forecasts. They find that combination forecasts are quite helpful in generating accurate forecasts of real housing price growth in the individual Eighth District states.
The Long-Run Relationship Between Consumption and Housing Wealth in the Eighth District States
Federal Reserve Bank of St. Louis Regional Economic Development, 2006 2(2), 140–147 (with Jack K. Strauss)
Abstract: The authors examine the long-run relationship between consumption and housing wealth for the seven individual states in the Federal Reserve System’s Eighth District. Given that state-level consumption data are not available, the authors develop a novel proxy for state-level consumption based on state-level data for personal income and savings income. They use this consumption proxy to estimate a cointegrating relationship between consumption spending and housing wealth, stock market wealth, and income in each of the Eighth District states. Their results indicate that increases in housing wealth produce sizable increases in consumption for most of the states in the Eighth District. Interestingly, the authors also find that consumption typically responds much more strongly to changes in housing wealth than to changes in stock market wealth. Their results imply that the strong increases in housing prices and home construction over the past decade have helped to buoy consumption and decrease saving in most Eighth District states.
Forecasting Employment Growth in Missouri with Many Potentially Relevant Predictors
Federal Reserve Bank of St. Louis Regional Economic Development, 2005, 1(1), 97–102 (with Jack K. Strauss)
Abstract: In this paper, the authors examine different approaches to forecasting monthly Missouri employment growth in the presence of many potentially relevant predictors, including both regional and national economic variables. Following Stock and Watson (2003, 2004), they first generate simulated out-of-sample forecasts of Missouri employment growth at horizons of 3, 6, 12, and 24 months using individual autoregressive distributed lag (ARDL) models based on 22 potential predictors. They then consider 20 different methods from the extant literature for combining the forecasts generated by the individual ARDL models. At longer horizons of 12 and 24 months, combining methods based on Bayesian shrinkage techniques produce out-of-sample forecasts that are substantially more accurate than forecasts from an autoregressive (AR) benchmark model. Combining methods based on Bayesian shrinkage techniques also outperform simple combining methods (such as those that use the mean or median of the individual forecasts) at longer horizons. Nevertheless, simple combining methods consistently outperform the AR benchmark model at all horizons and appear to offer a low-cost way of generating reliable combination forecasts.
Chapters in Edited Volumes
Time-Series and Cross-Sectional Stock Return Forecasting: New Machine Learning Methods
Machine Learning for Asset Management: New Developments and Financial Applications, 2020, Emmanuel Jurczenko (Ed.), Hoboken, NJ, Wiley, pp. 1–34 (with Guofu Zhou)
Abstract: This chapter extends the machine learning methods developed in Han et al. (2019) for forecasting cross-sectional stock returns to a time-series context. The methods use the elastic net to refine the simple combination return forecast from Rapach et al. (2010). In a time-series application focused on forecasting the US market excess return using a large number of potential predictors, we find that the elastic net refinement substantively improves the simple combination forecast, thereby providing one of the best market excess return forecasts to date. We also discuss the cross-sectional return forecasts developed in Han et al. (2019), highlighting how machine learning methods can be used to improve combination forecasts in both the time-series and cross-sectional dimensions. Overall, because many important questions in finance are related to time-series or cross-sectional return forecasts, the machine learning methods discussed in this chapter should provide valuable tools to researchers and practitioners alike.
Forecasting Stock Returns
Handbook of Economic Forecasting, 2013, Vol. 2A, Graham Elliott and Allan Timmermann (Eds.), Amsterdam, Elsevier, pp. 328–383 (with Guofu Zhou)
Abstract: We survey the literature on stock return forecasting, highlighting the challenges faced by forecasters as well as strategies for improving return forecasts. We focus on US equity premium forecastability and illustrate key issues via an empirical application based on updated data. Some studies argue that, despite extensive in-sample evidence of equity premium predictability, popular predictors from the literature fail to outperform the simple historical average benchmark forecast in out-of-sample tests. Recent studies, however, provide improved forecasting strategies that deliver statistically and economically significant out-of-sample gains relative to the historical average benchmark. These strategies—including economically motivated model restrictions, forecast combination, diffusion indices, and regime shifts—improve forecasting performance by addressing the substantial model uncertainty and parameter instability surrounding the data-generating process for stock returns. In addition to the US equity premium, we succinctly survey out-of-sample evidence supporting US cross-sectional and international stock return forecastability. The significant evidence of stock return forecastability worldwide has important implications for the development of both asset pricing models and investment management strategies.
[Data/MATLAB files | Business Forecasting blog post by Clive Jones]
Forecasting Regional and Industry-Level Variables
Advances in Economic Forecasting, 2011, Matthew L. Higgins (Ed.), Kalamazoo, Michigan, W.E. Upjohn Institute for Employment Research, pp. 51–64
Forecasting Stock Return Volatility in the Presence of Structural Breaks
Forecasting in the Presence of Model Uncertainty and Structural Breaks, 2008, Vol. 3 of Frontiers of Economics and Globalization, David E. Rapach and Mark E. Wohar (Eds.), Bingley, UK, Emerald, pp. 381–416 (with Jack K. Strauss and Mark E. Wohar)
Abstract: We examine the role of structural breaks in forecasting stock return volatility. We begin by testing for structural breaks in the unconditional variance of daily returns for the S&P 500 market index and ten sectoral stock indices for 9/12/1989–1/19/2006 using an iterative cumulative sum of squares procedure. We find evidence of multiple variance breaks in almost all of the return series, indicating that structural breaks are an empirically relevant feature of return volatility. We then undertake an out-of-sample forecasting exercise to analyze how instabilities in unconditional variance affect the forecasting performance of asymmetric volatility models, focusing on procedures that employ a variety of estimation window sizes designed to accommodate potential structural breaks. The exercise demonstrates that structural breaks present important challenges to forecasting stock return volatility. We find that averaging across volatility forecasts generated by individual forecasting models estimated using different window sizes performs well in many cases and appears to offer a useful approach to forecasting stock return volatility in the presence of structural breaks.