RDP 2007-10: Trade Costs and Some Puzzles in International Macroeconomics 4. Capital Controls

Another potential explanation for the three puzzles could be capital-market restrictions. Lewis (1996) argued that such restrictions could explain the lack of consumption risk sharing and Engel (2000) suggests that many of the puzzles may be explained by allowing for financial market imperfections. The following regression explores whether capital-market restrictions could play a role together with trade costs in explaining the Feldstein-Horioka puzzle:

where K is a dummy variable equal to 1 when a country has capital controls and zero otherwise (based on IMF data). If the puzzle is due to an interaction of the role of trade costs and capital controls, then the relevant interaction coefficient, ϖ₃, will be positive. As can be seen from Table 1, many countries removed capital controls during the sample period, with around 80 per cent of observations having capital controls early in the sample but only about 10 per cent having it towards the end.

Table 8 presents the results of investment, the real exchange rate and consumption equations using a variety of specifications. The middle column reports the coefficient ϖ₃ in Equation (5) or its analogue, while the last column reports the coefficient γ₁ or its analogue. The coefficient φ on the main independent variable (saving, the lag of the real exchange rate or output) appears in the left-hand column and tends to be positive and significant, suggesting that a puzzle still exists even after controlling for trade costs and capital controls. The coefficient γ₁ is still positive, indicating that even in the absence of capital controls, trade costs still appear to play a role in explaining the puzzle.