Investigating the Robustness of Macroeconomic Inference to Measurement Errors : The Case of the Consumer Decision Interval

That economists know next to nothing about measurement errors is not a good reason to assume them away. A more meaningful approach might be to investigate under which classes of measurement errors a given null hypothesis is rejected or not rejected. This paper illustrates this approach for the question of whether consumers make decisions at high or low frequency. It is shown that, with monthly data and assuming no measurement errors, a model of consumer behaviour that takes into account the durability of non-durable goods corroborates the low-frequency hypothesis. The paper then investigates the classes of measurement errors for which this conclusion does and does not hold