Bayesian Estimation of ARMA-GARCH Model of Weekly Foreign Exchange Rates

Three Bayesian methods (Markov chain Monte Carlo, Laplace approximation and quadrature formula) are developed to estimate the parameters of the ARMA-GARCH model. The ARMA-GARCH model is applied to weekly foreign exchange rate data of five major currencies, and their stochastic volatilities are judged by the posterior probabilities of stationarity and other conditions. This revised version was published online in August 2006 with corrections to the Cover Date.     

A Bayesian test of a parameter shift and an application

The paper formulates a Bayesian test of a parameter shift in two regressions whose error terms have multivariate student-t distributions with zero location vectors. The test is derived first with diffuse and then with natural conjugate prior probability density functions. The Bayesian test is then applied to examine whether or not a parameter shift in expenditure on vitamins and other nutritional supplements can be observed since 1969 in Japan. The empirical test implies that there is a sudden parameter shift in 1971 due to a consumer protection movement.     

Some tests for the constancy of regressions under heteroscedasticity

For testing the equality of coefficients of a linear regression model under heteroscedasticity, we suggest an F criterion conditioned on the posterior mean of the ratio of standard deviations of error terms in two subsamples. For pairable subsamples, and exact F test is derived. Sampling experiments show that the Chow test differs substantially from the nominal significance level when the two subsample sizes are unequal, and that the F test conditioned on the posterior mean is superior to other tests when sample sizes are small.     

Japanʼs intangible capital and valuation of corporations in a neoclassical framework

This paper estimates the economic value in the 1980s and 1990s of corporate assets in Japan, including both tangible and intangible assets, based on the neoclassical framework of McGrattan and Prescott (2005). Our estimates use a new micro-data set that comprises the accounting statements of all listed, non-financial companies in Japan. We find that in 1980–1986, a period that immediately preceded Japan?s so-called “bubble economy”, our assessed value of corporate productive assets, net of the value of corporate debt, is approximately equal to the actual stock market value of Japanese corporate equity. The finding differs from previous results based on studies of aggregate data sets or based on studies of micro-data sets that neglected intangible capital. We also show that the Japanese ratio of the amount of intangible capital stock to the amount of tangible capital stock is comparable to the analogous ratios for the U.S. and U.K.     

Personal Identification Through Pedestrians’ Behavior

This article focuses on a new approach for personal identification by exploring the features of pedestrian behavior. The recent progress of a motion capture sensor system enables personal identification using human behavioral data observed from the sensor. Kinect is a motion sensing input device developed by Microsoft for Xbox 360 and Xbox One. Personal identification using the Microsoft Kinect sensor (hereafter referred to as Kinect) is presented in this study. Kinect is used to estimate body sizes and the walking behaviors of pedestrians. Body sizes such as height and width, and walking behavior such as joint angles and stride lengths, for example, are used as explanatory variables for personal identification. An algorithm for the personal identification of pedestrians is defined by a traditional neural network and by a support vector machine. In the numerical experiments, pictures of body sizes and the walking behaviors are captured from fifteen examinees through Kinect. The walking direction of pedestrians was specified as 0°, 90°, 180°, and 225°, and then the accuracies were compared. The results indicate that identification accuracy was best when the walking direction was 180°. In addition, the accuracy of the vector machine was better than that of the neural network.     

Estimating unknown join points: Determination of the yen-dollar exchange rate

We propose a Bayesian procedure to estimate a switching regression in which the number of switching points (i.e. join points) is not known. We apply the Bayesian procedure to a regression model for the yen-dollar exchange rate using monthly data from January 1973 to June 1992. We identify three join points in January 1978, September 1988, and March 1990. We compare the post-sample forecast performances of our switching regression model to those of other regression models. The post-sample forecasts show that the Bayesian switching model performs better than the other models.     

Encoding of high-frequency order information and prediction of short-term stock price by deep learning

Predicting the price trends of stocks based on deep learning and high-frequency data has been studied intensively in recent years. Especially, the limit order book which describes supply-demand balance of a market is used as the feature of a neural network; however these methods do not utilize the properties of market orders. On the other hand, the order-encoding method of our prior work can take advantage of these properties. In this paper, we apply some types of convolutional neural network architectures to order-based features to predict the direction of mid-price trends. The results show that smoothing filters which we propose to employ rather than embedding features of orders improve accuracy. Furthermore, inspection of the embedding layer and investment simulation are conducted to demonstrate the practicality and effectiveness of our model.