In this paper we suggest an alternative explanation of the high cross-
section association between shares of saving and investment in GDP whi
ch Feldstein and Horioka (1980) interpret as evidence of low capital m
obility. In OECD countries, saving and investment shares appear to be
I(1). We show that a solvency constraint implies that the current bala
nce is stationary and thus that saving and investment cointegrate with
a unit coefficient irrespective of the degree of capital mobility. It
is this long-run relation that the FH cross-section regression captur
es. Econometric results for 23 OECD countries over the 1960-92 period
are consistent with this explanation.