Littérature scientifique sur le sujet « Return-forecasting regressions »

AbstractI develop new results for long-horizon predictive regressions with overlapping observations. I show that rather than using autocorrelation robust standard errors, the standard t-statistic can simply be divided by the square root of the forecasting horizon to correct for the effects of the overlap in the data. Further, when the regressors are persistent and endogenous, the long-run ordinary least squares (OLS) estimator suffers from the same problems as the short-run OLS estimator, and it is shown how similar corrections and test procedures as those proposed for the short-run case can also be implemented in the long run. An empirical application to stock return predictability shows that, contrary to many popular beliefs, evidence of predictability does not typically become stronger at longer forecasting horizons.

We study time variation in expected excess bond returns. We run regressions of one-year excess returns on initial forward rates. We find that a single factor, a single tent-shaped linear combination of forward rates, predicts excess returns on one-to five-year maturity bonds with R2 up to 0.44. The return-forecasting factor is countercyclical and forecasts stock returns. An important component of the return-forecasting factor is unrelated to the level, slope, and curvature movements described by most term structure models. We document that measurement errors do not affect our central results.

AbstractI test for stock return predictability in the largest and most comprehensive data set analyzed so far, using four common forecasting variables: the dividend-price (DP) and earnings-price (EP) ratios, the short interest rate, and the term spread. The data contain over 20,000 monthly observations from 40 international markets, including 24 developed and 16 emerging economies. In addition, I develop new methods for predictive regressions with panel data. Inference based on the standard fixed effects estimator is shown to suffer from severe size distortions in the typical stock return regression, and an alternative robust estimator is proposed. The empirical results indicate that the short interest rate and the term spread are fairly robust predictors of stock returns in developed markets. In contrast, no strong or consistent evidence of predictability is found when considering the EP and DP ratios as predictors.

This paper examines the causality patterns between the lagged trading volume and returns of the Saudi stock market (TASI) for the period from2003:01 to April 2013:05, along with two consecutive sub-periods to account for pre- and post- market collapse of 2006. Using the quantile regression approach, the study finds that the return-volume relations are heterogeneous across quantiles with symmetric tendency across the mean for the full sample period. On the contrary, the study could not support the heterogeneous and symmetric effects for the first sub-sample period. The second sub-sample period is characterized by homogenous across quantiles with statistical evidence of symmetry. Thus, the study concludes that the dependence structure between the lagged volume and subsequent market returns seems to be randomly relying on the chosen period which makes volume unsuitable to be used as explanatory power for returns forecasting.

This paper analyzes the forecast performance of historical S&P500 and Dow Jones Industrial Average (DJIA) excess returns while using nonparametric functional data analysis (NP-FDA). The empirical results show that the NP-FDA forecasting strategy outperforms not only the the prevailing-mean model, but also the traditional univariate predictive regressions with standard predictors used in the literature and, most cases, also combination approaches that use all predictors jointly. In addition, our results clearly have important implications for investors, from an asset allocation perspective, a mean-variance investor realizes substantial economic gains. Indeed, our results show that NP-FDA is the only one individual model that can overcome the historical average forecasts for excess returns in statistically and economically significant manners for both S&P500 and DJIA during the entire period, NBER recession, and expansions periods.

AbstractPast research shows that prices move in response to World Agricultural Supply and Demand Estimates (WASDE) reports immediately prior to and after a report. This research develops trading models based on knowing the next WASDE report in advance. This should help traders evaluate investments to predict information contained within the report and in determining how best to use such forecasts. The price-forecasting models use regressions against the ratios of ending stocks to use. Results show a steady increasing return to trading over the report month. The highest returns are produced by trading during the growing and harvest seasons.

Purpose This paper aims to ask whether a range of stock market factors contain information that is useful to investors by generating a trading rule based on one-step-ahead forecasts from rolling and recursive regressions. Design/methodology/approach Using USA data across 3,256 firms, the authors estimate stock returns on a range of factors using both fixed-effects panel and individual regressions. The authors use rolling and recursive approaches to generate time-varying coefficients. Subsequently, the authors generate one-step-ahead forecasts for expected returns, simulate a trading strategy and compare its performance with realised returns. Findings Results from the panel and individual firm regressions show that an extended Fama-French five-factor model that includes momentum, reversal and quality factors outperform other models. Moreover, rolling based regressions outperform recursive ones in forecasting returns. Research limitations/implications The results support notable time-variation in the coefficients on each factor, whilst suggesting that more distant observations, inherent in recursive regressions, do not improve predictive power over more recent observations. Results support the ability of market factors to improve forecast performance over a buy-and-hold strategy. Practical implications The results presented here will be of interest to both academics in understanding the dynamics of expected stock returns and investors who seek to improve portfolio performance through highlighting which factors determine stock return movement. Originality/value The authors investigate the ability of risk factors to provide accurate forecasts and thus have economic value to investors. The authors conducted a series of moving and expanding window regressions to trace the dynamic movements of the stock returns average response to explanatory factors. The authors use the time-varying parameters to generate one-step-ahead forecasts of expected returns and simulate a trading strategy.

There has been substantial research effort aimed to forecast futures price return volatilities of financial assets. A significant part of the literature shows that volatility forecast accuracy is not easy to estimate regardless of the forecasting model applied. This paper examines the volatility accuracy of several volatility forecast models for the case of the Mexican peso-USD exchange rate futures returns. The models applied here are a univariate GARCH, a multivariate ARCH (the BEKK model), two option implied volatility models and a composite forecast model. The composite model includes time-series (historical) and option implied volatility forecasts. Different to other works in the literature, in this paper there is a more rigorous analysis of the option implied volatilities calculations. The results show that the option implied models are superior to the historical models in terms of accuracy and that the composite forecast model was the most accurate one (compared to the alternative models) having the lowest mean-squared-errors. However, the results should be taken with caution given that the coefficient of determination in the regressions was relatively low. According to these findings it is recommended to use a composite forecast model if both types of data are available i.e. the time-series (historical) and the option implied.

Purpose: The aim of our paper is twofold. First, we examine the predictive ability of log book-market, dividend-price, earnings-price and dividend-earnings ratios on the most recent data set of the strongest securities in the UK economy; unlike the majority of the studies in this data set, our analysis is not limited on returns but further investigates dividend and earnings growth predictability under the presence of the most recent global financial recession. Second, we exploit the long-run equilibrium relationship in two systems, [p_t,d_t,e_t] and [p_t,b_t,e_t] and examine the predictive ability of our newly formed variables, namely 〖pde〗_t and 〖pbe〗_t. Design/methodology/approach: In this study, we examine the most recent data set of Financial Times Stock Exchange 100 (FTSE 100) and analyze it based on the formation of size portfolios. The main focus is placed on the index’s returns, dividend and earnings growth rates and the predictive ability of the four financial ratios we have selected following their reputation as strong predictors. We also formulate two extra ratios based on their long-run equilibrium relationship. Finding: Our study’s main findings can be summarized as following. First, we retrieve evidence that in-sample return predictability is evident in the medium and large-sized portfolios and is better captured by 〖pde〗_t at 35% and 47% equivalently. Second, forecasts on dividend growth are even more linked to the size criterion we employ. Third, in-sample regressions of continuously compounded earnings growth rate show that most predictive benefits are obtained by 〖dp〗_t in the medium portfolio with an R^2 of 45%. Research limitations/implications: A first constraint is the forecasters we employ; we have used the most indicative ones due to their popularity in similar data sets but there are other macroeconomic variables such as spreads and interest rates that could be tested in future research. Also, we could examine the sensitivity of our results on whether we use nominal, excess or real returns and then, attempt to alter our data’s frequency so as to address the seasonality effect observed mainly in dividends and earnings. Originality/value: We believe that our paper contributes to the ongoing debate of the traits that make return predictable and the information included in either dividends or earnings to explain that predictability. Finally, the novelty of this paper lies in the links it tries to retrieve among market capitalization value and predictability in a market whose predictive components have not been entirely explored. Our paper may prove informative to investors focused on short-term forecasting and interested in the effects of size in portfolio formation.

The purpose of this thesis is to examine the predictability of stock indices with regression models based on intermarket factors. The underlying idea is that there is some correlation between past price changes and future price changes, and that models attempting to capture this could be improved by including information derived from correlated assets to make predictions of future price changes. The models are tested using the daily returns from Swedish stock indices and evaluated from a portfolio perspective and their statistical signicance. Prediction of the direction of the price is also tested by Support vector machine classication on the OMXS30 index. The results indicate that there is some predictability in the market, in disagreement with the random walk hypothesis.
Syftet med denna uppsats är att undersöka förutsägbara tendenser hos aktieindex med regressionsmodeller baserade på intermarket-faktorer. The bakomliggande idén är att det existerar en viss korrelation mellan föregående prisrörelser och framtida prisrörelser, och att modeller som försöker fånga det kan förbättras genom att inkludera information från korrelerade tillgångar för att förutspå framtida prisförändringar. Modellerna testas med dagliga data på svenska aktieindex och utvärderas från ett portföljperspektiv och deras statistiska signifikans. Förutsägelser av riktningen hos priset testas också genom klassifikation med en Stödvektormaskin på OMXS30-index. Resultaten indikerar att det finns vissa förutsägbara tendenser i motsats till hypotesen om slumpmässiga aktiepriser.

Defence date: 4 April 2016
Examining Board: Prof. Evi Pappa, EUI, Supervisor; Prof. Agustín Bénétrix, Trinity College Dublin; Prof. Christian Brownlees, Universitat Pompeu Fabra; Prof. Peter Hansen, EUI.
The thesis consists of three essays in the fields of international finance and applied econometrics. The first chapter analyzes the co-movement of market premia for rare adverse events, addressing the important issue of contagion. The second chapter studies the impact of rare adverse events on the estimates of the risk-aversion coefficient and on household's portfolio composition. This chapter shows that the threat of a rare disaster justifies household's positive bond holdings. Finally, the last chapter studies if the information not contained in the domestic yield curve, but contained in the foreign yield curve helps to predict future dynamics of domestic yields. The first chapter proposes a novel approach to assessing volatility contagion across equity markets. More specifically I decompose the variance risk premia of three major stock indices into: crash and non-crash risk components and analyse their cross-market correlations. I find that crash-risk premia exhibit higher correlations than non-crash risk premia, implying the existence of volatility contagion. This suggests that investors believe that equity returns will be more highly correlated across countries during market crashes than during more normal times. The main result of the analysis holds when I apply other measures of co-movement as well as when I allow correlation to be time varying. Moreover I document that crash-premia constitute a large portion of the overall variance risk premia, highlighting the importance of crash-risks. Unlike the existing literature, my approach to testing the existence of volatility contagion does not rely on short periods of financial distress, but allows for crash-risk premia to be computed in tranquil times. The second chapter assesses the impact of the Peso problem on the econometric estimates of the risk aversion coefficient. Rietz (1988) and subsequently Barro (2006) showed that the introduction of the crash risk allows the canonical general equilibrium framework to generate data consistent equity premia even under low risk aversion of the representative agents. They argue that the original data used to calibrate these models suffer from a Peso problem (i.e. does not encounter a crash state). To the best of my knowledge the impact of their Peso problem on the estimation of the risk aversion coefficient has not to date been evaluated. This chapter seeks to remedy this. I find that crash states that are internalized by economic agents, but are not realized in the sample, generate only a small bias in the estimates of the risk aversion coefficient. I also show that the introduction of the crash state has a strong bearing on the household's portfolio composition. In fact, under the internalized crash state scenario, households exhibit positive bond holdings even in a frictionless environment. In the third chapter, co-authored with Andrew Meldrum and Peter Spencer, we show, using data on government bonds in Germany and the US, that 'overseas unspanned factors' - constructed from the components of overseas yields that are uncorrelated with domestic yields - have significant explanatory power for subsequent domestic bond returns. This result is remarkably robust, holding for different sample periods, as well as out of sample. By adding our overseas unspanned factors to simple dynamic term structure models, we show that shocks to those factors have large and persistent effects on domestic yield curves. Dynamic term structure models that omit information about foreign bond yields are therefore likely to be mis-specified.

Researchers have known for some time that nonlinearity exists in the financial markets and that neural networks can be used to forecast market returns. Unfortunately, many of these studies fail to consider alternative forecasting techniques, or the relevance of the input variables. The following research utilizes an information-gain technique from machine learning to evaluate the predictive relationships of numerous financial and economic input variables. Neural network models for level estimation and classification are then examined for their ability to provide an effective forecast of future values. A cross-validation technique is also employed to improve the generalization ability of the models. The results show that the classification models generate higher accuracy in forecasting ability than the buy-and-hold strategy, as well as those guided by the level-estimation-based forecasts of the neural network and benchmark linear regression models.

Researchers have known for some time that nonlinearity exists in the financial markets and that neural networks can be used to forecast market returns. Unfortunately, many of these studies fail to consider alternative forecasting techniques, or the relevance of the input variables. The following research utilizes an information-gain technique from machine learning to evaluate the predictive relationships of numerous financial and economic input variables. Neural network models for level estimation and classification are then examined for their ability to provide an effective forecast of future values. A cross-validation technique is also employed to improve the generalization ability of the models. The results show that the classification models generate higher accuracy in forecasting ability than the buy-and-hold strategy, as well as those guided by the level-estimation-based forecasts of the neural network and benchmark linear regression models.

This paper contains an analysis of withdrawal data for North West Indiana to compute consumptive-use coefficients and to describe monthly variability of withdrawals and consumptive use. Concurrent data were available for most water-use categories from 1990 through 2008. Average monthly water withdrawals are discussed for a variety of water-use categories, and average water use per month is depicted graphically. Water quality analysis is presented and historic water quality data of Northwest Indiana, (Lake, Porter and LaPort Counties) were downloaded from USEPA website and they were examined for the trends in different water quality constituents. Individual station based analysis and regional analysis were conducted using MK Test. Water quality data indicated an improvement trend. Water withdrawals data were analyzed using regression and Artificial Neural Network (ANN) models. The ANN model performed a better forecasting while compared to a linear regression model. For most water-use categories, the summer months were those of highest withdrawal and highest consumptive use. For public supply, average monthly withdrawals ranged from 2,193 million gallons per day (Mgal/d) (February) to 3,092 Mgal/d (July). North West Indiana energy production had large increases in average monthly withdrawals in the summer months (17,551 Mgal/d in February to 26,236 Mgal/d in July, possibly because of increased electricity production in the summer, a need for additional cooling-water withdrawals when intake-water temperature is high, or use of different types of cooling methods during different times of the year. Average industrial withdrawals ranged from 31,553 Mgal/d (February) to 36,934 Mgal/d (August). The North West Indiana irrigation data showed that most withdrawals were in May through October for golf courses, nurseries, and crop irrigation. Miscellaneous water withdrawals ranged from 12.2 Mgal/d (January) to 416.3 Mgal/d (October), commercial facilities that have high water demand in Indiana are medical facilities, schools, amusement facilities, wildlife facilities, large stores, colleges, correctional institutions, and national security facilities. Consumptive use and consumptive-use coefficients were computed by two principal methods in this study: the return-flow and withdrawal method and the winter-base-rate method (WBR). The WBR method was not suitable for the industrial and miscellaneous water-use categories. The RW method was not used for public-supply facilities. The public-supply annual average consumptive-use coefficient derived by use of the WBR methods is 8 percent from 1990 to 2008 for North West Indiana; the summer average consumptive-use coefficient was considerably higher with the amount of 20 percent. The energy production annual consumptive-use coefficient was 13 percent by the WBR method, which increased to 28 percent for summer. In terms of maximum accuracy and minimal uncertainty, use of available withdrawal, return-flow, and consumptive-use data reported by facilities and data estimated from similar facilities are preferable over estimates based on data for a particular water-use category or groups of water-use categories. If monthly withdrawal, return flow, and consumptive use data are few and limited, monthly patterns described in this report may be used as a basis of estimation, but the level of uncertainty may be a greater than for the other estimation methods.