Menzie Chinn on the relationship between the current account and budget deficits:
On the current account deficit, "We have met the enemy, and he is us".
[T]he issue of global imbalances ... is never far away from policy-makers minds. Of course, there are two major camps in the debate. First, there is the Bernanke "global saving glut" view, and the closely related but intellectually distinct "Bretton Woods Revived" force. The U.S. current account deficit, according to adherents of this view is made abroad, or at least made anywhere but in America. Second, there is the "Twin Deficits View". This perspective has taken a lot of lumps in recent years, partly because the current account deficit kept on rising even as the budget balance moved toward surplus over the latter part of the 1990's. Of course, since budget surplus was achieved in 2000, the Administration's policies have re-established the positive correlation between budget deficits and current account deficits by driving both to new nominal heights (and percent of GDP records, for the current account).
But what of the 1990's experience? Doesn't the divergent trends in the two deficits necessarily invalidate the twin deficits hypothesis? Certainly, this talking point has been taken as a way to absolve fiscal policy of any responsibility in the current account imbalance. Well, my response to this bivariate view of the world is that questions like this led man to create "multiple regression analysis". Taking this perspective, Hiro Ito and I have updated the results [PDF] discussed in this post.
Specifically, we have "rigorously interrogated" the data... We find the following (for the industrial countries, as defined by the IMF), over the 1975-2004 period:
- The pooled OLS estimate of the response of the current account balance to the budget balance, using 5-year-averaged data and time fixed effects, is 0.15 to 0.16. This means a one percentage point (ppt) change in the budget balance to GDP ratio induces a 0.l5 to 0.15 ppt improvement in the current account to GDP ratio (statistically significantly different from zero, using robust standard errors, and the 10% marginal significance level). ...
- The two stage least squares estimate (i.e., accounting for endogeneity) of the response is 0.325, statistically significantly different from zero at the 10% level.
- The estimate using Hodrick-Prescott (HP) trends (to identify the medium term values) is 0.095, statistically significantly different from zero at the 10% level.
- The OLS estimate using HP trend data and fixed effects is 0.485, statistically significantly different from zero at the 10% level. This result means that a 1 ppt. increase in the budget balance to GDP ratio purged of business cycle frequency effects induces an approximately 0.5 ppt. improvement in the the current account balance purged of business cycle frequency movements, after conditioning for country specific effects.
- In general, the point estimates range from a minimum of 0.095 (HP trends and OLS) to a maximum of 0.485 (HP trends, fixed effects), with a median of the estimates reported in the paper of 0.325.
In other words, in an exhaustive study that has addressed issues of specification (pooled OLS versus fixed effects), extraction of medium term variation (five year averages versus Hodrick-Prescott filtering), endogeneity (OLS versus 2SLS), the importance of institutions (w/ and w/o institutional variables), we find that fiscal policy has important economic and statistically significant effects on the current account imbalance. (By the way, as pointed out by Calculated Risk in this post, don't hold your breath for actual progress on the "on-budget" budget deficit).
This shouldn't deny the fact that international factors may have an impact on our current account imbalance. But those who hold blameless U.S. fiscal policy should reconsider the empirical evidence.