U.S. exorbitant privilege: Expected vs. realized

The expected returns on the U.S. net foreign asset position (NFA) feature prominently in theories of the U.S. exorbitant privilege. However, existing empirical studies of NFA dynamics focus on measuring realized outcomes. I use simulations of the NFA to show quantitatively that inferences based purely on realized NFA dynamics are misleading. I provide new estimates based on ex-ante expected returns by combining empirical asset pricing models with micro data on U.S. assets and liabilities. I investigate the most important determinants of expected NFA returns and any sources of asymmetry between expected returns on U.S. assets and liabilities. I document that expected returns on U.S. debt foreign assets are higher than the expected returns on U.S. debt foreign liabilities. I find that the most important determinant of NFA returns over the last 25 years has been changes to the composition of U.S. liabilities which along with the growth in the gross size of U.S. liabilities have led to the U.S. now having a net negative position in equity. These findings contrast with realized NFA returns, which over the same period are dominated by return differences between U.S. and foreign equities.