Time-varying foreign-exchange risk and central bank intervention: estimating profits from intervention and speculation

Abstract

Failure to risk-adjust estimates of profits, from central-bank foreign exchange intervention or from private speculation, can have large effects on the estimated profits, including changing signs. Many choices arise in deciding how to adjust profits for risk. The time period over which a market model is fit has mixed effects on calendar-year profits; variations in profits across calendar years is much more important than the period over which the market model is fit. In some cases, but not in all, results are sensitive to whether a US stock market index is used or a world market index. For non-US central banks or private speculators, the relevant market index might be denominated in USD, but alternatively might be denominated in a foreign currency. For the Swedish central bank, estimated profits decline importantly if an index measured in USD is used instead of an index measured in SEK. In estimating market models where beta is conditioned on some measure of intervention, likely candidates are intervention or cumulative intervention; the first has an effect for one or a few days, the second has long-term effects. Estimates show that the choice can make an important difference, though the effects are not all one way.