Measuring Transactions Costs from Observed Behavior: Market Choice in Peru

Farmers incur proportional and fixed transactions costs in selling their crops on markets. Using data for Peruvian potato farmers, we propose a method to measure these transactions costs. When opportunities exist to sell a crop on alternative markets, the observed choice of market can be used to infer a monetary measure of transactions costs in market participation. The market choice model is first estimated at the reduced form level with a conditional logit, as a function of variables that explain transactions costs. We then use these market choice equations to control for selection in predicting the idiosyncratic prices that would be received on all markets and the idiosyncratic proportional transactions costs that would be incurred to reach all markets. The net between the two gives us a measure of effective farm-level prices. This allows us to estimate a semi-structural conditional logit of the market choice model. In this model, the choice of market is a function of predicted effective farm-level prices, and of market information that accounts for fixed transactions costs. We can use the estimated coefficients to derive the price equivalence of the fixed cost due to information. We find that the information on market price that farmers receive from their neighbors reduces fixed transactions costs by the equivalent of doubling the price received, and is equal to four times the average transportation cost.