This paper assesses evidence, from the G-7 countries, of a link between a country's fiscal policy and its external balance, often referred to as the "twin deficits" phenomenon. It begins by reviewing the stylized facts, and then examines the theoretical case for such a link. A survey of the existing empirical literature reveals a rather mixed message: some authors find statistical evidence of a link, while others do not, and those who do find one report a wide range of estimates of the proportion by which the current account balance reacts to changes in the fiscal stance.

This paper focusses on two issues. First, it hypothesizes that important cross-country linkages may not have been incorporated in previous tests recorded in the literature. This is suggested because the standard theory behind the twin deficits link implies movements in the home country's exchange rate and external balance, and therefore by definition involves more than one country. In this regard, the paper investigates whether fiscal shocks should be modelled in a manner that takes account of fiscal stances in other countries when testing the strength of the twin deficits link. Second, the paper seeks to establish whether the data for the 1980s are substantially different from those for the 1970s, and whether studies focussing mainly on the former would give a fair impression of the strength of the link between the two balances.

These and a number of secondary issues are investigated in the context of six-variable vector autoregressions for each of the seven major industrial economies (United States, Japan, western Germany, France, Italy, the United Kingdom and Canada) over the 1972-90 period. It is found that the evidence in favour of the twin deficits link varies considerably across the seven countries. Also, while the evidence of such a link over the entire sample period is far from compelling, it is generally stronger when the 1980s data are examined in isolation. Although there is evidence of cross-country interdependencies that might influence single-country studies of the twin deficits link, tests based on relative measures of fiscal impulse yield mixed results. The paper concludes with some caveats put forward in the form of suggestions for further research.