Are distance effects really a puzzle?

Given the distance proxies for trade costs, the onset of globalization implies that geographical distance would matter less for trade. However, year-on-year regressions of a log-linearized gravity model estimated by the ordinary least squares (OLS) method usually suggest that the negative impact of distance on trade is rising since the 1950s during the late 20th century. These seemingly counter-intuitive results may occur due to the omission of the extensive margin as well as the neglect of the Jensen's inequality. This paper investigates these two potential solutions but that only the second seems to work. After considering Jensen's inequality, the distance effects declined over the period 1950–1999. In addition, this paper proposes a simple theoretical model to identify trade costs. The empirical results also show a declining trend of trade costs over the same time period.