Risk and Asian Exchange Rate Regimes

Abstract

A panel regression gives evidence that more flexibility in Asian exchange rates reduces risk associated with bank borrowing abroad, but deviations from mean exchange rates, and from the renminbi, increase risk. Since the exchange rate regime affects bank behavior and the incentives to hedge, the results broadly support the bank run over the moral hazard view of twin banking and currency crisis. The results suggest that flexibility in exchange rates is required for Asian EMEs, but the flexibility has to be limited, and it depends on more flexibility in the renminbi. This has implications for current global imbalances in reserves and feasible adjustment paths.

Keywords:

• exchange rate flexibility
• risk
• hedging
• moral hazard
• bank runs

Ashima Goyal thanks Fulbright, and Claremont Graduate University for providing a congenial working environment, Thomas Willett for comments, and Sven Arndt, Benjamin Cohen, and Randall Hennings for useful discussions on this topic and Saumik Paul for research assistance. Useful feedback was received after presenting some of the ideas at the Claremont Colleges.

Notes

1. Their model is based on Diamond and Dybvig's (1983) early modeling of rational bank runs.

2. During a period of steady depreciation of the Indian rupee over 1996–2002 only importers used to buy forward cover. Over 2003 as the rupee appreciated only exporters were hedging. Importers also rushed for cover when the rupee started depreciating in May 2004. There were news reports that major software firms such as Infosys and Wipro would not be able to gain from the depreciation because of the forward cover they had taken. The point is precisely that with hedging they can stop worrying about the impact of the rupee on their profits, and trying to make money from rupee movements. Instead they can focus on their key activity, producing better software.

3. In these regressions X3 and X4 for China reflect its link to the dollar. If the Chinese exchange rate is considered as fixed to its own mean, which makes X3 and X4 equal zero for China, the coefficients are still significant, and the signs remain unchanged, but their size falls.