and seller concentration is not evidence of countervailing power but merely the result of spurious correlation induced by own industry purchases. When they test this proposition by regressing BCR on SCR, and including dummy variables to account for own industry purchases, the regression coefficient of SCR remains positive and statistically significant at the 0.05 level for 4 and 8 firm CR's. The positive relation becomes insignificant only when the definition of own industry purchases is expanded to include any sales in the same two-digit group. Their justification for a wider definition is that four-digit industries in the same two-digit group use similar technologies and therefore have similar degrees of economies of scale, so that the observed correlation is induced by natural forces rather than by countervailing power. The similarity of economies of scale is asserted, but no supporting evidence is offered. In fact, two-digit groups are rather broad and, for intermediate production, generally contain successive stages of production of a class of finished products. A look at the cases where the top four or eight purchasers are ifl the producing industry's own twodigit group will raise doubts about whether these represent similar, technologies. For BCR4, there were ten cases where the top ranked consuming industry was in the same two-digit group as the producing industry.4 They included the following pairs of four-digit industries: Yarn Mills and Broadwoven Fabrics, Processed Textile Waste and Upholstery, Knit Fabric Mills and Apparel Made from Purchased Materials, Logging Camps and Saw Mills, Pulp Mills and Paper Mills, Paper Mills and Sanitary Paper Products, Primary Aluminum and Aluminum Rolling and Drawing, Secondary NonFerrous Metals and Primary Aluminum, Industrial Controls and Switch Gear. It is doubtful that similar degrees of concentration in these pairs are primarily the result of similar economies of scale in production and not induced by attempts to attain countervailing power.

The Lucas-Sargent-Wallace (L-S-W) proposition states that expected policy changes have no influence on real economic variables even in the short run. (See Robert Lucas, 1972, and Thomas Sargent and Neil Wallace, 1975.) One form of empirical test of the L-S-W proposition (the neutrality proposition) began with Robert Barro's (1977; 1978) seminal articles supporting the neutrality proposition. However, more recent papers, including those by David Small (1979), John Makin (1982), and Frederic Mishkin (1982a, b) have reexamined the relationships between anticipated and unanticipated money vs. output, unemployment, and inflation, and have found evidence suggesting that anticipated monetary changes do have real effects. All of the above-cited studies use expectations procedures which estimate, for example, a money growth equation over the entire period under consideration.' However, this approach assumes that when expectations are being formed at the beginning of the period, economic agents have the relevant information for the entire period. In this note I reexamine the impact of expected vs. unexpected monetary changes focusing on the informational restrictions actually faced by economic agents when formulating expectations. The structure of the model used here is similiar to the models of Barro and Mishkin. A measure of real output is assumed to be a function of a policy variable and a vector of predetermined variables. The policy variable can be decomposed into anticipated and unanticipated components. Algebraically:

Michael T. Belongia, R. W. Hafer, Richard G. Sheehan, On the Temporal Stability of the Interest Rate-Weekly Money Relationship, The Review of Economics and Statistics, Vol. 70, No. 3 (Aug., 1988), pp. 516-520

Discount window borrowing by weekly reporting banks disaggregated by Federal Reserve district is used to estimate Marvin Goodfriend's (1983) model of borrowed reserves. Little evidence is found to support the argument that a bank's borrowing decision is determined by the spread between the funds rate and the discount rate and by prior bank borrowing. A weekly reporting bank has only a 2.7 percent chance of visiting the discount window during any given maintenance period. This result is consistent with the presence of considerable harassment costs imposed by the discount window officer. Copyright 1994 by MIT Press.