Thursday, December 11, 2008

Cost of default on sovereign debt

Now that various governments seem committed to increasing their debt drastically, it may become appropriate to investigate whether it could be optimal for them to default on this debt. What would this entail? Eduardo Borensztein and Ugo Panizza discuss this and identify four costs to default. This is based on an empirical exercise using 200 years of default data.

They find that reputation costs, as measured by credit ratings and rate spreads are significant, but short-lived. Indeed, consider how many times Argentina has defaulted yet managed to get decent credit conditions. International trade also is negatively affected, and this beyond what can be explained by a reduction in trade credit. The impact on GDP is also significant, especially when the cause for default was not very compelling. The impact on the economic activity is, however, short-lived. Finally, defaults may cause banking crises, but not vice-versa. And absent a banking crisis, the financial industry is not more affected than the rest of the economy.

Reading this, it does not seem that defaulting is that bad. In fact, it seems better than a personal bankruptcy, even US style, where establishing decent credit conditions is a long and hard battle. So, who is going to be the first to default? Not so quick, because it seems that some people suffer greatly from sovereign debt default: governments, and especially their finance ministers.

1 comment:

Ugo said...

Thanks for your interest in our paper. I think that there are two small imprecisions in the post. The first is that in our analysis, we don't look at 200 years of defaults, but to a much shorter sample (basically, we look at the last 20-30 years).

Moreover, our paper finds that default have not been very costly so far because most defaults have been "justified" by difficult economic conditions. In our paper we make the point that "strategic" defaults (i.e. defaults which are not justified by difficult economic conditions) could be extremely costly and this exactly why we normally don't observe these types of defaults. So, I would not interpret the finding of our paper as, please go ahead and default.