The behavior of money demand in the Chinese hyperinflation

In this paper I study the behavior of money demand during the episode of hyperinflation that occurred in China after World War II. I consider two popular and competing money demand specifications – the log–log and the semi-log – and show that the log–log performs better than the semi-log in its ability to track the behavior of the money demand. The choice between the two specifications is of great importance, as it implies that welfare cost estimates are very different for hyperinflation. The findings also contribute to the understanding of Cagan’s paradox and the failure of Cagan inflationary finance models. The paradox might be attributable to the popular semi-log schedule for money demand, and the log–log schedule might be an appropriate form for the analysis of hyperinflation.