Search

News reports have become an imperative conduit of public information. Several recent studies have used news data from public media to investigate the impact of news on stock market returns. This study analyses the usefulness of news for predicting stock returns in the Taiwan stock market. We employ text mining of economic news, transform documents using a keyword matrix, and then convert the results into news variables. Subsequently, together with other quantitative variables, we construct a fixed effect model to investigate the behaviours of stock market returns in 20 subsectors from January 2008 to December 2014. Empirical analysis reveals that the news variables provide useful information for predicting Taiwan stock market returns, although the out-sample performance is only marginally improved. We also discover an asymmetric effect of economic news for predicting stock market returns. The prediction accuracy is higher when the stock market is booming than when it is glooming.

The aggregation of financial and economic time series occurs in a number of ways. Temporal aggregation or systematic sampling is the commonly used approach. In this paper, we investigate the time interval effect of multiple regression models in which the variables are additive or systematically sampled. The correlation coefficient changes with the selected time interval when one is additive and the other is systematically sampled. It is shown that the squared correlation coefficient decreases monotonically as the differencing interval increases, approaching zero in the limit. When two random variables are both added or systematically sampled, the correlation coefficient is invariant with time and equal to the one-period values. We find that the partial regression and correlation coefficients between two additive or systematically sampled variables approach one-period values as n increases. When one of the variables is systematically sampled, they will approach zero in the limit. The time interval for the association analyses between variables is not selected arbitrarily or the statistical results are likely affected.

In this paper we investigate the time interval effect of multiple regression models in which some of the variables are additive and some are multiplicative. The effect on the partial regression and correlation coefficients is influenced by the selected time interval. We find that the partial regression and correlation coefficients between two additive variables approach one-period values as n increases. When one of the variables is multiplicative, they will approach zero in the limit. We also show that the decreasing speed of the n-period correlation coefficients between both multiplicative variables is faster than others, except that a one-period correlation has a higher positive value. The results of this paper can be widely applied in various fields where regression or correlation analyses are employed.

Is money demand in Taiwan stable? Moreover, is money a luxury goods in Taiwan such that the income elasticity is greater than one? A casual application of Goldfeld type of money demand to the Taiwanese economy answers no to the first question and yes to the second one. This paper rigorously analyzes the money demand in Taiwan and attempts to provide more accurate answers to these questions. We employ both the ARMAX and cointegration models to study the money demand and use the rolling estimation approach to examine the stability of parameter estimates over time. Furthermore, we take into account the impact of stock market on money demand. Our empirical analysis concludes that the money demand in Taiwan is stable and that the income elasticity is less than one. Wrongly including a constant term within a dynamic model with lagged values of the dependent variable as regressors results in unstable estimates over time. In addition, the stock market is confirmed to have a significant impact on the demand of money.

This paper analyzes the Taiwan stock market and examines its price and volatility linkages with those of the United States. In particular, it tests the hypothesis that the short-term volatility and price changes spill over from the developed markets, mainly the United States, to the emerging Taiwan stock market. The model and the test are built upon Engle's ARCH (autoregressive conditional heteroskedasticity) and Engle and Kroner's M-GARCH (multivariate generalized ARCH) models. The paper differs from previous studies on the Taiwan stock market in three respects. First, instead of using daily closing prices, it uses close-to-open and open-to-close returns to avoid the problem of overlapping samples. It carefully models the day-of-the-week effect in daily data to avoid misspecification of the model. Second, to circumvent the generated regressor problem arising from the two-step estimation procedure, it also employs the M-GARCH model where all parameters are estimated simultaneously. Third, the misspecification test is carried out on various kinds of asymmetric ARCH factors. A substantial volatility spillover effect is found from the US stock market to the Taiwan stock market, especially for the model using close-to-open returns. There is also evidence supporting a spillover effect in price changes. The findings can be explained by the recent gradual opening of the Taiwan stock market to foreign investors. Taiwan's economy is closely linked to that of the United States through international trade and is likely to be affected by changes in the market fundamentals of the US economy. The increasing foreign participation in the Taiwan stock market has further strengthened the linkage. So, it is naturally easy to conjecture that price changes and volatility in the financial market also spill over from the US to Taiwan. In view of the different sizes of the markets, spillover effects in the other direction are deemed unlikely. Evidence of volatility spillover across international markets is reported in studies that use data from foreign exchange markets and stock markets. It is consistent with the view that information processing is the source of the volatility clustering. In a study by Engle et al. (1990), which looks at the Japanese yen=US dollar rate, it is shown that, except for the Tokyo market, each market's volatility is significantly affected by changes in volatility in the other markets, so that volatility is transmitted through time and different market locations just like a ''meteor shower''.

Does co-integration help long-term forecasts? In this paper, we use simulation, real data sets, and multi-step-ahead post-sample forecasts to study this question. Based on the square root of the trace of forecasting error-covariance matrix, we found that for simulated data imposing the 'correct' unit-root constraints implied by co-integration does improve the accuracy of forecasts. For real data sets, the answer is mixed. Imposing unit-root constraints suggested by co-integration tests produces better forecasts for some cases, but fares poorly for others. We give some explanations for the poor performance of co-integration in long-term forecasting and discuss the practical implications of the study. Finally, an adaptive forecasting procedure is found to perform well in one- to ten-step-ahead forecasts. Copyright 1996 by John Wiley & Sons, Ltd.

The definition of causation, discussed in Granger (1980) and elsewhere, has been widely applied in economics and in other disciplines. For this definition, a series yt is said to cause xt+l if it contains information about the forecastability for xt+l contained nowhere else in some large information set, which includes xt−j, j ≥ 0. However, it would be convenient to think of causality being different in extent or direction at seasonal or low frequencies, say, than at other frequencies. The fact that a stationary series is effectively the (uncountably infinite) sum of uncorrelated components, each of which is associated with a single frequency, or a narrow frequency band, introduces the possibility that the full causal relationship can be decomposed by frequency. This is known as the Wiener decomposition or the spectral decomposition of the series, as discussed by Hannan (1970). For any series generated by , where xt, and are both stationary, with finite variances and a(B) is a backward filterwith B the backward operator, there is a simple, well-known relationship between the spectral decompositions of the two series.

Two alternative methods are considered for identifying what lags to use in a nonlinear model relating a pair of series. One is based on a mutual information function, the other is Kendall's r. They both have the property that if each variable is instantaneously transformed, such that ranks are preserved, then the functions are unchanged. Simulations find properties of the functions and allow application to generated nonlinear series. In simple cases, the methods appear to find frequently the correct lags.

If a simple non-linear autoregressive time-series model is suggested for a series, it is not straightforward to produce multi-step forecasts from it. Several alternative theoretical approaches are discussed and then compared with a simulation study only for the two-step case. It is suggested that fitting a new model for each forecast horizon may be a satisfactory strategy.

This paper investigates multistep prediction errors for non-stationary autoregressive processes with both model order and true parameters unknown. We give asymptotic expressions for the multistep mean squared prediction errors and accumulated prediction errors of two important methods, plug-in and direct prediction. These expressions not only characterize how the prediction errors are influenced by the model orders, prediction methods, values of parameters and unit roots, but also inspire us to construct some new predictor selection criteria that can ultimately choose the best combination of the model order and prediction method with probability 1. Finally, simulation analysis confirms the satisfactory finite sample performance of the newly proposed criteria., Comment: Published in at http://dx.doi.org/10.3150/08-BEJ165 the Bernoulli (http://isi.cbs.nl/bernoulli/) by the International Statistical Institute/Bernoulli Society (http://isi.cbs.nl/BS/bshome.htm)

Previous analysis on forecasting theory either assume knowing the true parameters or assume the stationarity of the series. Not much are known on the forecasting theory for nonstationary process with estimated parameters. This paper investigates the recursive least square forecast for stationary and nonstationary processes with unit roots. We first prove that the accumulated forecast mean square error can be decomposed into two components, one of which arises from estimation uncertainty and the other from the disturbance term. The former, of the order of $\log(T)$, is of second order importance to the latter term, of the order T. However, since the latter is common for all predictors, it is the former that determines the property of each predictor. Our theorem implies that the improvement of forecasting precision is of the order of $\log(T)$ when existence of unit root is properly detected and taken into account. Also, our theorem leads to a new proof of strong consistency of predictive least squares in model selection and a new test of unit root where no regression is needed. The simulation results confirm our theoretical findings. In addition, we find that while mis-specification of AR order and under-specification of the number of unit root have marginal impact on forecasting precision, over-specification of the number of unit root strongly deteriorates the quality of long term forecast. As for the empirical study using Taiwanese data, the results are mixed. Adaptive forecast and imposing unit root improve forecast precision for some cases but deteriorate forecasting precision for other cases., Published at http://dx.doi.org/10.1214/074921706000000969 in the IMS Lecture Notes Monograph Series (http://www.imstat.org/publications/lecnotes.htm) by the Institute of Mathematical Statistics (http://www.imstat.org)

The three most important Chinese holidays, Chinese New Year, the Dragon- boat Festival, and Mid-Autumn Holiday have dates determined by a lunar calendar and move between two solar months. Consumption, production, and other economic behavior in countries with large Chinese population including Taiwan are strongly affected by these holidays. For example, production accelerates before lunar new year, almost completely stops during the holidays and gradually rises to an average level after the holidays. This moving holiday often creates difficulty for empirical modeling using monthly data and this paper employs an approach that uses regressors for each holiday to distinguish effects before, during and after holiday. Assuming that the holiday effect is the same for each day of the interval over which the regressor is nonzero in a given year, the value of the regressor in a given month is the proportion of this interval that falls in the month. Bell and Hillmer (1983) proposed such a regressor for Easter which is now extensively used in the U.S. and Europe. We apply the Bell and Hillmer's method to analyze ten important series in Taiwan, which might be affected by moving holidays. AICC and out-of-sample forecast performance were used for selecting number of holiday regressors and their interval lengths. The results are further checked by various diagnostic checking statistics including outlier detection and sliding spans analysis. The empirical results support this approach. Adding holiday regressors can effectively control the impact of moving holidays and improves the seasonal decomposition. AICC and accumulated forecast error are useful in regressor selection. We find that unemployment rates in Taiwan have holiday effects and seasonal factors cannot be consistently estimated unless the holiday factor is included. Furthermore, as the unemployment is rising, the magnitude of holiday and seasonal factor are decreasing. Finally, we find that holiday factors are generally smaller than seasonal factors but should not be ignored.

The Asian financial crisis broke out in Thailand in July 1997, and rapidly spread throughout the neighboring countries. An important question then arises? Is it possible to predict next financial crisis? If yes, then what are the predictors? The answer lies in combined usage of economic theory and econometric methods. By using the economic theory, one can locate possible potential crisis predictors whereas appropriate econometric models can pinpoint effective ones. In this paper we suggest using the Stochastic Search Variable Selection (SSVS) developed by George and McCulloch (1993) to identify the crisis predictors. As is suggested by the name, SSVS stochastically searches for practically significant variables. Each variable coefficient is assumed to come from a mixture of two normal variates with respectively large and small variances. For the former case, this variable is considered as insignificant and should be excluded from the model whereas for the latter, this variable is significant and should be included in the model. SSVS is not affected by the ordering of the candidate variables and is particularly effective when the sample size is much smaller than the number of all possible models. By employing SSVS method, we conclude that annual growth rate of money supply, $M_2$, and the ratio of government debt to GDP are promising predictors for financial crisis. It is worth mentioning that the frequently mentioned factors, such as ratio of total foreign reserve to GDP and the ratio of current deficit to GDP are not selected by our analysis. Our empirical analysis implies that monetary and fiscal policy play a crucial role in exploring the Asian financial crisis.