The following figure shows the relationship between an index of productivity and farm size in Brazil in three different years. The relationship is specified as follows: Yield, ?, is the measure of land productivity, which is measured as

agricultural output per unit of land. It is postulated that ?=??(?), where ? is the area of the farm, indicating that yield may be a function of farm size. If farm size and yield are unrelated, we have ???(?)/??=0.

a. Can this result be taken as evidence for decreasing return to scale?(See graph below)

b. Many argue that this evidence is the result of imperfect labour markets. How do imperfect labour markets explain this relationship?

c. Many argue that this result is caused by omitted variable bias. What are the omitted variables in this case and what effect might they have on the relationship?

d. Benjamin (1995) uses instrumental variable technique to test the omitted variable hypothesis in Java. Provide a discussion for the appropriateness and the strength of his instruments? What evidence does he find regarding the farm size-productivity relationship?

Farm Size and Land Productivity, Brazil (logs) Far Sue (logs] 1085 --- 1900 Empirical work has regularly found dyw(A)/0A