Pischke (1995) uses both microeconomic and macroeconomic US data to test th
e idea that, within an otherwise standard PIH framework, ignorance by agent
s of aggregate labour income can account for the observed degree of excess
smoothness and sensitivity in consumption. His tests involve only the secon
d moments of aggregate consumption and labour income. In this paper our mai
n aim is to identify and test the restrictions his model implies for aggreg
ate consumption dynamics, using US quarterly data over the period 1959-1996
, but our framework allows us also to test an earlier, related model of Goo
dfriend (1992). We find that both models can be formally rejected: ignoranc
e of aggregate labour income cannot by itself account for aggregate consump
tion dynamics; some other relaxation of the assumptions of the standard PIH
is required. We give an example of one possible such relaxation and presen
t evidence indicating that Pischke's version of imperfect information may,
within that framework, have a significant role to play. Copyright (C) 2000
John Wiley & Sons, Ltd.