This study examines the voluntary asset sales that took place in Japan from 2000 to 2007. Asset sales increased rapidly after the introduction of a new stock swap and transfer scheme under the Commercial Code Revision in the late 1990s. The unwinding of cross shareholding also forced managers to sell underperforming divisions to maximize shareholder wealth. The study investigates the effect of asset sales announcements and analyzes the motivation for asset sales. In particular, it verifies that information asymmetry influences asset sales. Based on asymmetric information, management can observe the performance of each division, but markets can only observe the total performance of a diversified firm. Hence, markets are likely to underestimate a firm's value. When firms are undervalued, they not only resort to selling overvalued assets for funds, but they also receive a proper evaluation by providing clearer financial information on undervalued assets. This suggests that the announcement of asset sales is positively related to the stockholder's wealth and mitigates the information gap. The results are as follows. This study uses 262 voluntary asset sales and a market model to investigate the announcement effects. We find that the announcement of asset sales is significantly positive for abnormal returns of 1.09% per day (-, +1). This suggests that asset sales convey information on corporate fundamental values if the management has access to complete information and attempts to evaluate the wealth of existing shareholders. In addition, we analyze the relationship between the cumulative abnormal return (CAR) and motivation of asset sales. Hite et al. (1987) suggest that management only operates assets for which they have a comparative advantage and sells inefficient assets. This is the so-called efficient hypothesis of asset sales. John and Ofek (1995) report that returns are higher, and the firm experiences increased operating performance, when asset sales lead to a more focused firm. This is called the focus hypothesis. However, Lang et al. (1995) argue that asset sales are used to raise capital and show that announcement returns depend on the use of the proceeds, and positive returns only occur when proceeds are paid out to creditors or shareholders. They call this the financing hypothesis. The main empirical results are not consistent with the above three hypotheses. First, to test the financing hypothesis, the entire sample is classified into two subsamples: the payout sample (54 asset sales) and there investment sample (146 asset sales). The abnormal returns of payout firms are not significantly different than those of reinvestment firms. This result is not consistent with the financing hypothesis, which asserts that an agency problem, through management's discretion, affects asset sales return. Second, we find that the sample has 118 focused asset sales and 82 non-focused asset sales. The CAR (-1, 1) of focused firms is 0.17%, which is 1.35% lower than that of the other firms, but the difference is not significant. This result also does not support the focus hypothesis. Third, according to the efficient hypothesis, management who inefficiently operate their assets will transfer it to others who may operate it more efficiently. We use Tobin's q as a proxy variable representing firm efficiency. Tobin's q for sellers is not lower than that of buyers. In addition, the median CAR of the buyer is significantly negative. These results are not consistent with the hypothesis that the positive effect of asset sales is due to improved efficiency. Fourth, we find that the degree of information asymmetry of asset sales firms is significantly higher than that of matching firms. We identify matching firms using the methodology developed by Barber and Lyon (1996). The degree of information asymmetry significantly decreases after asset sales. This suggests that information asymmetry is related to asset sales. Furthermore, in cross-sectional regressions of CAR (-3, +3), the coefficient of information asymmetry is significantly positive, whereas the coefficients of the debt-pay dummy and focus dummy are non-significantly negative. This also supports the notion that information asymmetry influences asset sales. The results of the other control variables are as follows. Firm size is not related to the announcement effect. Tobin's q is significantly positive and market return is significantly negative, suggesting that positive market response is due to the availability of information on which firms are undervalued. The coefficients of ROA and leverage are statistically insignificant, which does not support the findings of Hite et al. (1987) and Lang et al. (1995). Finally, we analyze whether asset sales are associated with investments and find that the sensitivity of investments and the proceeds of asset sales are positive and statistically significant. This is consistent with the view that firms improve their market value using asset sales that reduce information asymmetry.