STAR and ANN models: forecasting performance on the Spanish “Ibex-35” stock index

This paper studies whether it is possible to exploit the nonlinear behaviour of daily returns on the Spanish Ibex-35 stock index returns to improve forecasts over short and long horizons. In this sense, we examine the out-of-sample forecast performance of smooth transition autoregression (STAR) models and artificial neural networks (ANNs). We use one-step (obtained by using recursive and nonrecursive regressions) and multi-step-ahead forecasting methods. The forecasts are evaluated with statistical and economic criteria. In terms of statistical criteria, we compared the out-of-sample forecasts using goodness of forecast measures and various testing approaches. The results indicate that ANNs consistently surpass the random walk model and, although the evidence for this is weaker, provide better forecasts than the linear AR model and the STAR models for some forecast horizons and forecasting methods. In terms of the economic criteria, we assess the relative forecast performance in a simple trading strategy including the impact of transaction costs on trading strategy profits. The results indicate a better fit for ANN models, in terms of the mean net return and Sharpe risk-adjusted ratio, by using one-step-ahead forecasts. These results show there is a good chance of obtaining a more accurate fit and forecast of the daily stock index returns by using one-step-ahead predictors and nonlinear models, but that these are inherently complex and present a difficult economic interpretation.     

A multi-horizon comparison of density forecasts for the S&P 500 using index returns and option prices

We compare density forecasts of the S&P 500 index from 1991 to 2004, obtained from option prices and daily and 5-min index returns. Risk-neutral densities are given by using option prices to estimate diffusion and jump-diffusion processes which incorporate stochastic volatility. Three transformations are then used to obtain real-world densities. These densities are compared with historical densities defined by ARCH models. For horizons of two and four weeks the best forecasts are obtained from risk-transformations of the risk-neutral densities, while the historical forecasts are superior for the one-day horizon; our ranking criterion is the out-of-sample likelihood of observed index levels. Mixtures of the real-world and historical densities have higher likelihoods than both components for short forecast horizons.     

The jump component of S&P 500 volatility and the VIX index

Much research has investigated the differences between option implied volatilities and econometric model-based forecasts. Implied volatility is a market determined forecast, in contrast to model-based forecasts that employ some degree of smoothing of past volatility to generate forecasts. Implied volatility has the potential to reflect information that a model-based forecast could not. This paper considers two issues relating to the informational content of the S&P 500 VIX implied volatility index. First, whether it subsumes information on how historical jump activity contributed to the price volatility, followed by whether the VIX reflects any incremental information pertaining to future jump activity relative to model-based forecasts. It is found that the VIX index both subsumes information relating to past jump contributions to total volatility and reflects incremental information pertaining to future jump activity. This issue has not been examined previously and expands our understanding of how option markets form their volatility forecasts.     

证券分析师VS统计模型:证券分析师盈余预测的相对准确性及其决定因素

本文使用2005年35家券商对我国上市公司做出的每股盈余预测数据,考察了证券分析师盈余预测相对于统计模型的相对准确性及其决定因素。我们发现,我国证券分析师做出的盈余预测,同以年度历史数据为基础的统计模型得出的盈余预测相比,预测误差较小,证券分析师盈余预测具有一定的优势;但同某些以季度历史数据为基础的统计模型得出的盈余预测相比,预测误差较大,证券分析师盈余预测不具有优势。我们同时考察了决定证券分析师盈余预测相对准确性的决定因素。我们发现,公司每股盈余的波动性越大,公司上市越晚,跟踪公司的分析师越多,证券分析师的优势就越大。我们的研究对证券分析师以及投资者都有一定的启示作用。     

Combining forecasts: some results on exchange and interest rates

The aim of this work is to investigate whether the combination of forecasts plays an important role in the improvement of forecast accuracy Particular attention is paid to: (a) the methods of forecasting (the methods compared are neural networks, fuzzy logic, GARCH models, switching regime and chaotic dynamics); (b) combining the forecasts provided by the different methods. This work has also the aim of revising a short-term econometric forecast using a longer-term forecast. The revision process usually runs the opposite way (revision is made on a longer-term forecast using a short-term one to reflect the current available information, but it is not excluded that it is possible to proceed as described above. Daily data from the financial market is used. Some empirical applications on exchange and interest rates are given.     

An examination of the effects of management earnings forecast form and explanations on financial analyst forecast revisions

Using a sample of 978 quarterly management earnings-per-share forecasts made during the period 1993 to 1999, we document that financial analyst revisions to management earnings forecasts are a function of management forecast form. More precise forecasts (measured three different ways) lead to greater revision of financial analyst consensus EPS forecasts for a given level of unexpected earnings as predicted by Kim and Verrecchia (1991) and Bayesian adjustment models. Also, consistent with our arguments, maximum forecasts are interpreted as bad news by analysts. Our results, while consistent with theory, are inconsistent with recent experimental studies which do not reject the null hypothesis of no effect of management earnings forecast form on the association between unexpected earnings and financial analyst forecast revisions. We also re-examine Baginski, Hassell, and Kimbrough's (2004) finding that attributions used to explain management forecasts affect the reaction to the forecast using analyst data. Consistent with their findings using stock prices, the attribution presence (especially external attributions) increases financial analyst revisions pursuant to management forecasts.     

The Taylor Rule and Forecast Intervals for Exchange Rates

In this paper, we examine the Meese–Rogoff puzzle from a different perspective: out‐of‐sample interval forecasting. While most studies in the literature focus on point forecasts, we apply semiparametric interval forecasting to a group of exchange rate models. Forecast intervals for 10 OECD exchange rates are generated and the performance of the empirical exchange rate models are compared with the random walk. Our contribution is twofold. First, we find that in general, exchange rate models generate tighter forecast intervals than the random walk, given that their intervals cover out‐of‐sample exchange rate realizations equally well. Our results suggest a connection between exchange rates and economic fundamentals: economic variables contain information useful in forecasting distributions of exchange rates. We also find that the benchmark Taylor rule model performs better than the monetary, PPP and forward premium models, and its advantages are more pronounced at longer horizons. Second, the bootstrap inference framework proposed in this paper for forecast interval evaluation can be applied in a broader context, such as inflation forecasting.     

Information criteria for GARCH model selection

In this paper, a set of appropriately modified information criteria for selection of models from the AR-GARCH class is derived. It is argued that unmodified or naively modified traditional information criteria cannot be used for order determination in the context of conditionally heteroscedastic models. The models selected using the modified criteria are then used to forecast both the conditional mean and the conditional variance of two high frequency exchange rate series. The analysis indicates that although the use of such model selection methods does lead to significantly improved forecasting accuracies for the conditional variance in some instances, these improvements are by no means universal. The use of these criteria to jointly select conditional mean and conditional variance model orders leads to performance degradation for the conditional mean forecasts compared to models which do not allow for the heteroscedasticity.     

Forecasting metal prices: Do forecasters herd?

We analyze more than 20,000 forecasts of nine metal prices at four different forecast horizons. We document that forecasts are heterogeneous and report that anti-herding appears to be a source of this heterogeneity. Forecaster anti-herding reflects strategic interactions among forecasters that foster incentives to scatter forecasts around a consensus forecast.     

Are more able managers good future tellers? Learning from Japan

Baik et al. (2011) find that high-ability managers in the U.S. are more likely to issue accurate management earnings forecasts. Focusing on Japan, where management earnings forecasts are effectively mandated, we extend the literature by exploring (1) whether the relationship between managerial ability and forecast accuracy is unique to the U.S. disclosure system, where management forecasts are voluntary, and (2) how high-ability managers increase their forecast accuracy. We find that managerial ability is negatively associated with forecast errors based on initial forecasts, suggesting that high-ability managers are more likely to issue accurate forecasts at the beginning of the fiscal year. We then show that high-ability managers are less likely to revise their initial earnings forecasts and less likely to use earnings management to improve the accuracy of their earnings forecasts. Our findings show that, while high-ability managers are more likely to issue accurate initial management forecasts, low-ability managers are more likely to revise their forecasts and conduct earnings management to reduce their forecast errors.     

Probability Forecasts of Macroaggregates in the Turkish Economy

We provide probability forecasts of key Turkish macroeconomic variables such as inflation and output growth. The probability forecasts are derived from a core vector error correction model of the Turkish economy and its several variants. We use model and window averaging to address uncertainties arising from estimated models and possible structural breaks. The performances of the different models and their combinations are evaluated using relevant forecast accuracy tests in different pseudo out-of-sample settings. The results indicate that successful directional forecasts can be obtained for output growth and inflation. Averaging over both the models and the estimation windows improves the level of accuracy of the forecasts.     

A New Model of Inflation,Trend Inflation,and Long‐Run Inflation Expectations

This paper develops a bivariate model of inflation and a survey‐based long‐run forecast of inflation that allows for the estimation of the link between trend inflation and the long‐run forecast. Thus, our model allows for the possibilities that long‐run forecasts taken from surveys can be equated with trend inflation, that the two are completely unrelated, or anything in between. Using a variety of inflation measures and survey‐based forecasts for several countries, we find that long‐run forecasts can provide substantial help in refining estimates and fitting and forecasting inflation. It is less helpful to simply equate trend inflation with the long‐run forecasts.     

Determinants of analysts’ revenue forecast accuracy

This study investigates financial analysts’ revenue forecasts and identifies determinants of the forecasts’ accuracy. We find that revenue forecast accuracy is determined by forecast and analyst characteristics similar to those of earnings forecast accuracy—namely, forecast horizon, days elapsed since the last forecast, analysts’ forecasting experience, forecast frequency, forecast portfolio, reputation, earnings forecast issuance, forecast boldness, and analysts’ prior performance in forecasting revenues and earnings. We develop a model that predicts the usefulness of revenue forecasts. Thereby, our study helps to ex ante identify more accurate revenue forecasts. Furthermore, we find that analysts concern themselves with their revenue forecasting performance. Analysts with poor revenue forecasting performance are more likely to stop forecasting revenues than analysts with better performance. Their decision is reasonable because revenue forecast accuracy affects analysts’ career prospects in terms of being promoted or terminated. Our study helps investors and academic researchers to understand determinants of revenue forecasts. This understanding is also beneficial for evaluating earnings forecasts because revenue forecasts reveal whether changes in earnings forecasts are due to anticipated changes in revenues or expenses.     

预测中国宏观经济变量:专家与模型的组合预测

本文使用组合预测方法,探究以“朗润预测”为代表的专家预测以及计量模型对于中国宏观经济变量的预测效果,并研究对不同预测进行组合预测是否有助于改进预测效果。本文发现,对我国CPI和GDP的增长率,专家预测效果总体上优于模型预测。从原因看,一方面,专家在预测时已经考虑了计量模型的预测信息;另一方面,在经济出现“拐点”的时期,专家通过对实际经济环境和政策的把握,得出更准确的经济预测。组合预测有助于提升预测精度,对专家预测进行组合得到的预测效果优于大多数的专家预测,“模型—专家”组合预测的效果也优于所有的模型和大部分专家预测。     

Pitfalls in backtesting Historical Simulation VaR models

Historical Simulation (HS) and its variant, the Filtered Historical Simulation (FHS), are the most popular Value-at-Risk forecast methods at commercial banks. These forecast methods are traditionally evaluated by means of the unconditional backtest. This paper formally shows that the unconditional backtest is always inconsistent for backtesting HS and FHS models, with a power function that can be even smaller than the nominal level in large samples. Our findings have fundamental implications in the determination of market risk capital requirements, and also explain Monte Carlo and empirical findings in previous studies. We also propose a data-driven weighted backtest with good power properties to evaluate HS and FHS forecasts. A Monte Carlo study and an empirical application with three US stocks confirm our theoretical findings. The empirical application shows that multiplication factors computed under the current regulatory framework are downward biased, as they inherit the inconsistency of the unconditional backtest.     

Management disclosure bias and audit services

This paper considers the level of bias observed in management disclosures of earnings forecasts and historic earnings data in Australian prospectuses. Management forecasts and naïve forecasts derived from managements’ normalised historic data are analysed. A key focus is upon the possible association between such forecast bias and differential audit services performed upon the data. Audit firm size and level of engagement are modelled against bias. The full sample revealed no overestimation bias for any of the forecast models, but underestimation was observed for elements of the management and random walk naïve forecasts. Cross-sectionally, a significant association was observed between forecast bias and audit firm size across all three forecast models. Specifically, the audit firm size variable (Non Big-5/Big-5) was inversely associated with the extent to which forecasted and normalised historic earnings data were upwardly biased. On the other hand, the level of engagement was not a significant discriminator for forecast bias. These outcomes are contrasted against others reported elsewhere in the literature and suggest a risk in generalising across contexts. The findings imply a level of ‘disclosure management’ regarding company IPO forecasts and normalised historic accounting data, with forecast overestimation and error size more extreme when the monitoring expertise and/or reputation of auditors is lower (JEL D80, G14, M41, N27).     

Forecasting government bond yields with large Bayesian vector autoregressions

We propose a new approach to forecasting the term structure of interest rates, which allows to efficiently extract the information contained in a large panel of yields. In particular, we use a large Bayesian Vector Autoregression (BVAR) with an optimal amount of shrinkage towards univariate AR models. The optimal shrinkage is chosen by maximizing the Marginal Likelihood of the model. Focusing on the US, we provide an extensive study on the forecasting performance of the proposed model relative to most of the existing alternative specifications. While most of the existing evidence focuses on statistical measures of forecast accuracy, we also consider alternative measures based on trading schemes and portfolio allocation. We extensively check the robustness of our results, using different datasets and Monte Carlo simulations. We find that the proposed BVAR approach produces competitive forecasts, systematically more accurate than random walk forecasts, even though the gains are small.     

Forecasting Changes in Copper Futures Volatility with GARCH Models Using an Iterated Algorithm

There is a gap in the literature regarding the out-of-sample forecasting ability of GARCH-type models applied to derivatives. A practitioner-oriented method (iterated cumulative sum of squares) is applied to detecting breakpoints in the variance of two copper futures series. Short-, intermediate-, and long-term out-of-sample forecasts of copper future series are compared to forecasts from a benchmark random walk model for each series. Not only do the GARCH-type models dominate the random walk model, but the relative improvement is fairly consistent across series, forecast horizon, and GARCH-type model. The evidence makes clear that, with few exceptions, the forecast improvement of the GARCH-type models over the RW model lies somewhere between 20–30%. It is particularly true that for the long-term close to close forecasts, there is great coherence among the forecasts. These all fall within a fairly narrow range.     

Refining the Degree of Earnings Surprise: A Comparison of Statistical and Analysts' Forecasts

This paper compares the relative predictive ability of several statistical models with analysts' forecasts. It is one of the first attempts to forecast quarterly earnings using an autoregressive conditional heteroskedasticity (ARCH) model. ARCH and autoregressive integrated moving average models are found to be superior statistical forecasting alternatives. The most accurate forecasts overall are provided by analysts. Analysts have both a contemporaneous and timing advantage over statistical models. When the sample is screened on those firms that have the largest structural change in the earnings process, the forecast accuracy of the best statistical models is similar to analysts' predictions.     

Market implied future earnings and analysts’ forecasts

This study provides evidence on market implied future earnings based on the residual income valuation (RIV) framework and compares these earnings with analyst earnings forecasts for accuracy (absolute forecast error) and bias (signed forecast error). Prior research shows that current stock price reflects future earnings and that analyst forecasts are biased. Thus, how price-based imputed forecasts compare with analyst forecasts is interesting. Using different cost of capital estimates, we use the price-earnings relation and impute firms’ future annual earnings from three residual income (RI) models for up to 5 years. Relative to I/B/E/S analyst forecasts, imputed forecasts from the RI models are less or no more biased when cost of capital is low (equal to a risk-free rate or slightly higher). Analysts slightly outperform these RI models in terms of accuracy for immediate future (1 or 2) years in the forecast horizon but the opposite is true for more distant future years when cost of capital is low. A regression analysis shows that, in explaining future earnings changes, analyst forecasts relative to imputed forecasts do not impound a significant amount of earnings information embedded in current price. In additional tests, we impute future long-term earnings growth rates and find that they are more accurate and less biased than I/B/E/S analyst long-term earnings growth forecasts. Together, the results suggest that the RIV framework can be used to impute a firm’s future earnings that are high in accuracy and low in bias, especially for distant future years.