Sabri wrote:
It all depends on how my utility function differs from that of the average player, does it not? As a central bank I do not care if I lose a few billion dollars here and there, as long as I move in the direction of my optimal solution. You seem to be thinking in terms of "univariate risk aversion" whereas it is better to think in terms of "multivariate risk aversion".
I don't think it needs to be so complicated. If the central bank has many other assets in its portfolio, it can do same for each of them. What holds between 2 assets must hold among them all. At the margin, net benefits from switching between all assets must be zero. So, there's no loss of generality by thinking of just 2 assets: TT's Eurobonds are just like U.S. T-bonds, except for indenture details (for which you can adjust) *and default risk*. U.S. T's are thought of as the safer asset. Now, in solving your problem and choosing positions, you and the rest of the market determine their respective risk premia on TT's Eurobonds, each one based on its respective individual risk preferences. Other things equal, if, when the rest of the market is selling off TT's Eurobonds, you are buying them, what does that reveal about your (and the rest of the market's) risk preferences? I mean, you're shedding the safer asset and buying the riskier one! What kind of kinky utility function would you need to have to (1) be less risk averse than the rest of the market and, at the same time, (2) price TT's risk lower than the rest of the market so that (3) you're moved to buy TT and sell U.S. T's when the rest of the market is doing the opposite?
The question is: can I create such a market? That is, will Argentina, Venezuela, Brazil, Turkey and the like choose the cooperative equilibrium?
I don't understand. I thought you meant that each country's (e.g. Turkey's) central bank could hold its own country's Eurodebt. What you're saying here is that smaller economies could cooperate to bypass U.S. treasuries in their reserve portfolios. This latter proposition makes sense, to the extent the correlation is small.
