Econometricians have an ambivalent attitude towards economic data. At one level, the "data" are the world that we want to explain, the basic facts that economists purport to elucidate. At the other level, they are the source of all our trouble. Their imperfection makes our job difficult and often impossible.