ABSTRACT
The chapter explains the basic econometric theory of how to identify the unknown latent factor. Empirical researchers have used the following two general identification strategies. First, earlier researchers are forced to settle for simply describing the factors using their shape, correlation to observed series, and factor loadings. Next, after excluding the variables in the first set, one may attempt to find the second set of explanatory variables that are highly correlated with the second principal component (PC) estimator. Again, from the common characteristics among the second set, one may be able to identify the nature of the second factor. More importantly, the false identification rate is possibly increasing as well. A panel data has asymptotically weak factors if the common components asymptotically vanish. PC-estimated factors are purely statistical factors, which are independent of each other. Crime rates in a few states can be responsible for the common factors to all of the states.
