57A further problem with the MLP approach is that it seems implausible to assume that the typical consumer would do a sophisticated Muthian signal extraction problem (really, a special case of the Kalman filter) on a single observed aggregate variable (aggregate income), when they could do much better either by adding a few more variables that are equally easy to observe (interest rates, past consumption growth, etc). If they were really so intently focused on understanding where the aggregate economy is, they could do better yet, and much more easily, just by reading the available news stories reporting on professional forecasters’ forecasts. Discomfort with this somewhat schizophrenic set of assumptions is why in work after Lucas [1973], Lucas moved away from his assumption that microeconomic agents have imperfect information about aggregate data.