Tests for a structural break in a time series typically involves partitioning the data into two sub-periods and compare the resulting mean rates of growth. A more formal approach involves estimating the regression parameters for each sub-period and testing theequality of the two sets of parameters. An important limitation of both approaches is that the breakdate must be known a priori. The researcher must either pick an arbitrary breakdate or pick a breakdate based on some known feature of the data. The test results can be "uninformative" because they can miss the true breakdate, or can be ‘misleading’ because the breakdate is endogenous and the test can indicate a break when none in fact exists. This paper tests for slower productivity growth in agriculture using techniques that allow for unknown structural breaks jointly with a possible unit root that can have either or both stochastic and deterministic components.