This paper reevaluates the theoretical and empirical debate surrounding the bivariate time series relationship between wholesale and consumer price inflation. Empirical work by Engle [3], Guthrie [7], and Silver and Wallace (S&W) [9] is based on a one-sided distributed lag model running from wholesale prices to consumer prices. Colclough and Lange (C&L) [1] and Granger, Robins and Engle (GR&E) [6] criticize this specification, arguing that it contradicts the standard theory of derived demand whereby changes in demand for final goods cause changes in demand for primary goods. S&W's finding that consumer prices fail to Granger cause wholesale price appears (to C&L and GR&E) to reject flexible price, demand led models in favor of cost-push, markup models of consumer price inflation.' In this paper we demonstrate that the patterns of feedback in U.S. data are consistent with standard, flexible price models of derived demand with strong aggregate demand elements. First, using Geweke's [4] linear feedback methodology, we confirm the general pattern of feedback found in previous studies. The incremental predictive content of wholesale prices for consumer prices is approximately twice that of consumer prices for wholesale prices. However, half of the feedback between the two series is purely contemporaneous. We find this pattern to be robust in a larger system that includes money as a proxy for aggregate demand. The large contemporaneous feedback implies that any structural inferences will rely heavily on maintained assumptions concerning the contemporaneous error structure. In section III we present a stylized model of price determination in an economy with two stages of production. Supply disturbances in the primary goods market affect wholesale prices and, with a lag, consumer prices. Thus the model predicts feedback from wholesale to consumer prices whenever supply disturbances are present. The key insight from the model is that the converse does not hold in the presence of demand disturbances. The demand for primary goods depends, not on current or past retail demand, but on expected future demand. Even with