In this issue we have: Forecasting Stock Market Volatilities Using MIDAS Regressions: An Application to the Emerging Markets ; Recognizing and Forecasting the Sign of Financial Local Trends using Hidden Markov Models ; Evaluating Volatility and Correlation Forecasts ; Experts' Stated Behavior.
|By:||Alper, C. Emre
|We explore the relative weekly stock market volatility forecasting performance of the linear univariate MIDAS regression model based on squared daily returns vis-a-vis the benchmark model of GARCH(1,1) for a set of four developed and ten emerging market economies. We first estimate the two models for the 2002-2007 period and compare their in-sample properties. Next we estimate the two models using the data on 2002-2005 period and then compare their out-of-sample forecasting performance for the 2006-2007 period, based on the corresponding mean squared prediction errors following the testing procedure suggested by West (2006). Our findings show that the MIDAS squared daily return regression model outperforms the GARCH model significantly in four of the emerging markets. Moreover, the GARCH model fails to outperform the MIDAS regression model in any of the emerging markets significantly. The results are slightly less! conclusive for the developed economies. These results may imply superior performance of MIDAS in relatively more volatile environments.|
|Keywords:||Mixed Data Sampling regression model; Conditional volatility forecasting; Emerging Markets.|
|JEL:||C53 C52 C22 G10|
|The problem of forecasting financial time series has received great attention in the past, from both Econometrics and Pattern Recognition researchers. In this context, most of the efforts were spent to represent and model the volatility of the financial indicators in long time series. In this paper a different problem is faced, the prediction of increases and decreases in short (local) financial trends. This problem, poorly considered by the researchers, needs specific models, able to capture the movement in the short time and the asymmetries between increase and decrease periods. The methodology presented in this paper explicitly considers both aspects, encoding the financial returns in binary values (representing the signs of the returns), which are subsequently modelled using two separate Hidden Markov models, one for increases and one for decreases, respectively. The approach has been tested with different exp! eriments with the Dow Jones index and other shares of the same market of different risk, with encouraging results.|
|Keywords:||Markov Models; Asymmetries; Binary data; Short-time forecasts|
|JEL:||C02 C63 G11|
|By:||Andrew J. Patton
Franses, Ph.H.B.F. (Erasmus Research Institute of Management (ERIM), RSM Erasmus University)
|We ask various experts, who produce sales forecasts that can differ from earlier received model-based forecasts, what they do and why they do so. A questionnaire with a range of questions was completed by no less than forty-two such experts who are located in twenty different countries. We correlate the answers to these questions with actual behavior of the experts. Our main findings are that experts have a tendency to double count and to react strongly to recent volatility in sales data. Also, experts who feel more confident give forecasts that differ most from model-based forecasts.|
|Keywords:||model forecasts;expert forecasts;decision making;stated behavior|
Taken from the NEP-FOR mailing list edited by Rob Hyndman.