Baxter and Crucini [Am. Economic Rev. 83 (1993) 416] propose a country-size argument to explain the Feldstein-Horioka puzzle. This paper augments the empirical literature by examining the threshold effect of country-size, measured by relative GNP share, on the magnitude of saving-retention coefficient. Evidence from a panel of 23 OECD Countries indicates that the saving-retention coefficient increase as the relative GNP share becomes larger, which substantially Supports the country-size argument. (C) 2003 Elsevier Inc. All rights reserved.