Bernasconi M., Kirchkamp O. (1999).

Why monetary policy matters – An experimental study of saving, inflation and monetary policies in an overlapping generations model. Sonderforschungsbereich 504 Publications 98-47, University of Mannheim.



We study experiments of an overlapping generations model where agents may transfer wealth from one period to the next (saving) and where government revenue is created through seigniorage. Inflation influences the amount of wealth an agent may wish to save in a given period. Inflation is determined by the monetary policy and by the amount of average saving within each period.

The framing of our experiment differs in several respects from the one that is used by Lim, Prescott, Sunder (1994), Marimon, Spear, Sunder (1993), Marimon, Sunder (1993,1994,1995). Some of the differences are:

  • participants are not forced to save optimally and risk-neutrally, given their expectations;
  • participants choose themselves whether they form expectations on inflation or on average savings;
  • participants describe their expectations graphically;
  • participants can test the implication of several different expectations before making a saving decision;
  • the market is presented to subjects as a market operating in the EMU;
  • monetary policies have labels and participants vote on monetary policies.
  • The above mentioned literature comes in particular to the conclusion that the choice of the monetary policy does not matter. Only the level of government deficit determines level and volatility of inflation.

Evidence from our experiments in Florence, Mannheim and Pavia challenges this point. We find that:

  • in contrast to the literature, agents do not form first order adaptive expectations;
  • while in Marimon and Sunder’s exeriments subjects are forced to behave risk-neutrally, subjects `oversave’ for precautionary reasons once they are allowed to.
  • This `oversaving’ or `precautionary saving’ implies (in theory and in the experiment) different levels of inflation and volatility for different monetary policies (which are supposed to be equivalent in the literature under optimal, risk-neutral behavior).
  • Therefore, in our experiment the so called Friedman conjecture holds, i.e. a money-growth style monetary rule turns out to be more stable
  • We relate our findings to the current research on the theories of coordination and equilibrium selection.