Sunday, March 23, 2014

International asset pricing - think of theory as well as data

There are some important empirical facts concerning international asset pricing:

1. There is the tendency for high interest rate currencies to appreciate - the forward premium anomaly or carry trade. This does not follow the logic of uncovered interest rate parity (UIP).

2. There is poor correlation between consumption growth differentials and exchange rate movements; growth different should be related to the currency changes. This should not occur given standard assumptions of preferences and complete markets.

Now the UIP relationship seems to hold well during the pre-10970's period, but has fallen over the last few decades which does not seem to make sense if we have free capital flows. The carry is the basis for much of international FX trading. The same change in empirical regulatory seems to apply for correlations of consumption which were well above .5 pre-1970's and now are closer to zero. During this time currency volatility increased substantially and there has been an increase in openness across countries as measured by the current count to GDP ratio. More openness, yet less correlation and a breakdown of basic relationships.

Ricardo Colacito and Mariano Croce in the December 2013 Journal of Finance provides a solution to this international asset anomaly through looking at risk-sharing mechanism and wealth transfer risk-return trade-off model. When there are, for example, long-term growth differences and capital mobility, preferences may cause capital flows which will change the sensitivity between currencies and interest differentials.

There is a lot going on with their general equilibrium model, but there are some simple take-aways for international investors. Do not accept recent empirical relationships as a given fact in all environments. When the structure of markets change, market frictions change, there will be an adjustment in empirical relationships. Place value on the data, but remember you need theory to explain the data. Making generalization with limited data can be a loser's game.


No comments: