This post is a bit geeky, but I've been mulling over some comments Nouriel Roubini offered a few days back in an email to RGE Monitor subscribers:

Today we look at how Asian Financial Crisis changed Asia and the world...

Policy makers throughout Asia clearly are determined not to return to the IMF.  In Stephen Jen's words the Asian Financial Crisis taught Asian central banks to never be caught without enough foreign exchange reserves...

Asia's crisis generated an international consensus that emerging economies need to hold sufficient reserves to cover their short-term debt.  No comparable consensus emerged, though, over the right exchange rate regime.   The G-7 and the IMF argued that Asia's crisis showed the risks of currency pegs, at least those currency pegs not backed up by an institutional commitment like a currency board.   But it isn't clear that Asian policy makers drew the same lesson many seem to have concluded the real problem was not pegs, but an over-valued currency.

In a simple cut at the issue, there is a distinction between pegging an exchange rate below its freely-floating value -- the case of an under-valued currency -- and pegging the exchange rate above its freely-floating value -- the case of an over-valued currency.  In the case of an over-valued currency, the central bank has to trade from its holdings of foreign currency reserves to "soak up" the excess supply of its own currency.  The ability to do this is obviously limited by its reserves, which can become seriously depleted in the event of a persistent misalignment between the pegged value and where the market would take the exchange rate in the absence of the peg.  And if market participants get a sniff of the possibility that the government lacks sufficient reserves to support the peg, a speculative run is all but guaranteed.

The situation is a bit different when the currency is under-valued.  In this case a central bank would react to upward pressure on the exchange rate by expanding its own money supply to buy foreign currencies, thus eliminating the excess supply of those currencies.  Since there are no inherent resource restrictions in creating fiat money, there is no obvious limit to the government's ability to play the game.  Money supply expansion may eventually be inflationary, but that would itself tend to drive down the freely-floating value of the currency.  But therein, according to Nouriel, lies the problem:

... reserves are not an unmitigated blessing. Sterilizing the region's huge reserve growth poses growing difficulties.  In some countries - notably China - strong money and credit growth has fueled inflationary pressures, stock market bubbles and housing bubbles.  Nouriel Roubini argues that current Asian exchange rate and financial policies have left Asia vulnerable to new kinds of crises. See "Have Asia's Sterilization and Reserve Accumulation Policies Shielded It From Another Crisis Or Have Led To Increased Vulnerabilities?"

I confess that I have been used to thinking like the Asian policy makers in assuming that the impact of pegging an exchange at too low a level is more benign than pegging at too a high level.  Because inflation reduces the value of a currency relative to others, there is a sense in which the actions of a central bank attempting to damp currency appreciation are consistent with moving the exchange rate closer to the unfettered equilibrium value. 

The same argument, however, ought to hold for a currency that is over-valued.  When a central bank buys back its own currency with foreign reserves it is contracting the domestic money supply.  That in turn should reduce the domestic rate of inflation and result in a nominal appreciation, moving the exchange rate's fundamental value toward the peg.  If there are problems, then, they must arise because the necessary price adjustments are too slow when the pressure on the currency is persistent.

But if prices are slow to adjust, why not on the upside as well as the downside?  And though depleting reserves are no problem in the case of an under-valued exchange rate, the misallocation of resources associated with excessive monetary creation could be, which I believe is exactly Nouriel's point.  I'm still not sure that such misallocations have as sharp a destabilizing potential as running out of reserves and losing the capacity to intervene all together, but I'm convinced Nouriel's argument is worth thinking about.