As the development of methodology in new empirical industrial organization (NEIO), there have been increasing number of studies that estimate market power of sellers and/or buyers in various kinds of agricultural supply chains. Many of them, however, do not capture time variation of market power during the sample periods, even though they use long time series data. If the degree of market power actually changes along the time, such an estimation that uses only whole sample of the data may provide a misleading conclusion. The main objective of this study is to establish a way of estimating time variation of market power and making a time series index of it. Such an index enables us to make a comparison with indices of market structure, such as market share, which has been one of the main research topics of traditional industrial organization studies. Empirical analysis is conducted using the case of the U.S. soybean exports. The methodology employed in this study to estimate market power of the U.S. exporters is residual demand model, which enables us to derive the degree of market power and has widely been used in the context of international markets. To capture time variation of the market power, rolling window regression method is applied to the model, which is a methodology that repeats regressions using subsamples of total data by shifting the start and end points with a fixed window. Using the parameter of residual demand elasticity in each rolling estimation, a time series index of market power is calculated. The estimation results of rolling regressions of residual demand model using GMM-nonIV and the window size of 30 show that the U.S. had market power over importers to some extent until 1995, but had less and little power from the late 1990s to 2010. It also showed that the U.S. had almost no market power over China from 1996 to 2010. On the other hand, especially from the late 1990s, the U.S. had increasing market power over Mexico and Japan, although the extent was larger to Japan than to Mexico. Using the estimated index of market power and published index of market shares, then the relation between market structure and performance was analyzed. The analysis indicates that the changes in market power of the U.S. soybean exporters to importers average correspond to the decrease of the U.S. market share in the world soybean exports and to the increase of importers concentration due to the increase of China’s imports. It is also pointed out that indices of the U.S. market power over Mexico and Japan who depend soybean imports heavily on the U.S. may correspond to the changes in the market structure of the U.S. grain exporting industry.