This article examines the effects of unemployment compensation finance in a labor market in which firms pay efficiency wages. Two self‐financing unemployment compensation systems are compared: one in which benefits are financed by a proportional payroll tax and another in which experience rating is introduced by taxing firms in proportion to their separations. We find that experience rating leads to less unemployment, less shirking, and higher output.

A straightforward generalization of the simple net present value rule that correctly predicts when to invest in two classes of projects that can be delayed is derived. The first class consists of projects for which the option to delay derives its value exclusively from uncertainty about interest rates. It is shown that the optimal rule for investing in such projects is to simply multiply the discount rate of the project by the ratio of the mortgage rate to the riskless rate and then use this new rate as the discount rate in a standard net present value analysis. The other class of investment opportunities that is considered is the firm's option to expand. It is shown that it is only optimal for the firm to expand when a particular call option on the firm's stock has no time value. The fact that mortgage bonds (in the form of GNMAs) and stock options are actively traded implies that these rules have potentially important practical and empirical value. Besides their simplicity, the rules have the added advantage that they do not depend on a maintained assumption on the dynamics of interest rates in the economy.

This paper considers conceptual and practical issues that arise in measuring labor's share of national income. Most importantly: How are workers defined? How is compensation defined? The current definition of labor compensation used the Bureau of Economic Analysis (BEA) includes the salary of business owners and payments to retired workers in labor compensation. An alternative series to the BEA's standard series is presented. In addition, a simple method for decomposing labor compensation into a component due to raw labor' and a component due to human capital is presented. Raw labor's share of national income is estimated using Census and CPS data. The share of national income attributable to raw labor increased from 9.6 percent to 13 percent between 1939 and 1959, remained at 12-13 percent between 1959 and 1979, and fell to 5 percent by 1996.

[This article uses bond market data to empirically test the asset pricing model of Kazemi (1992). According to this model the rate of return on a long-term, pure-discount, default-free bond will be perfectly correlated with changes in the marginal utility of the representative investor. The covariability between financial asset returns and returns on such a bond can therefore serve as a measure of the riskiness of assets. The aim of this study is to determine whether the model can explain cross-sectional differences in the monthly returns of bonds with different maturity dates. We estimate and test the restrictions imposed by the model on returns of default-free bonds, while allowing the conditional distribution of bond returns to be time varying. The model is rejected during the full sample period (1973-1995) and the subperiod (1973-1980) when the Federal Reserve's focus is on interest rates, while the model is not rejected during the subperiod (1981-1995) when the Federal Reserve's focus is on money supply.]

CORRIGENDA Volume 37, June 1999 In "The Numerical Reliability of Econometric Software", by B. D. McCullough and H. D. Vinod, there are misprints in some equations on page 647 in the upper part of the left-hand column. The passage should read as follows: The normal equations are 6?^ + 61.5? ^ = 7.261 61.5?^ + 630.55? ^ = 74.5053 which yields a regression line of Y^ = ?3.478 + 0.457X. Suppose that the normal equations had been computed only to four significant digits as 6?^ + 61.5? ^ = 7.261 61.5?^ + 630.6? ^ = 74.50.

Numerous examples show that some econometric software packages contain serious flaws, and that users cannot safely assume that their software is accurate. A brief survey of the fundamentals of computer arithmetic discusses the sources of numerical error and emphasizes that computer arithmetic is not at all like pencil-and-paper arithmetic. Both users and developers of econometrics software should first pay attention to accuracy, and only later consider user-friendliness. Details are provided for assessing the accuracy of basic estimation routines, statistical distributions, and random number generators. More accuracy benchmarks are needed, especially for specialized econometric procedures.

This paper examines the association between consolidation and changes in levels of bank branching as measured by changes in the number of bank branches per capita. Using a specially-constructed data set, we address this issue as well as how this relationship varies with the type of consolidation and initial regulatory, competitive, and market conditions. We find that merges where merging institutions have branch networks which overlap within a ZIP code (within-ZIP merger) are strongly associated with a reduction in offices per capita in that ZIP code. This result is robust across time and holds in both rural and urban areas. The findings also suggest that, contrary to popularly held views, consolidation is not unambiguously negatively associated with changes in the number of banking offices per capita. Neither within-market-but-not-within-ZIP mergers nor out-of-market mergers consistently show such a relationship. We also find that the relationships between within-ZIP and within-market-but-not-within-ZIP mergers and changes in the number of bank offices per capita are more negative in low-income neighborhoods than in other neighborhoods. However, because most states now have unrestricted branching and because savings associations are less prevalent and financially healthier than in the past, these findings may not be indicative of future branching patterns.

[The dynamics of the unobservable short rate are frequently estimated directly using a proxy. We examine the biases resulting from this practice (the "proxy problem"). Analytic results show that the proxy problem is not economically significant for single-factor affine models. In the two-factor affine model of Longstaff and Schwartz (1992), the proxy problem is only economically significant for pricing discount bonds with maturities of more than five years. We also describe two different numerical procedures for assessing the magnitude of the proxy problem in a general interest rate model. When applied to a nonlinear single-factor model, they suggest that the proxy problem can be economically significant.]

Auctions of government securities typically permit bidders to enter multiple price-quantity bids. Despite the widespread adoption of this institutional feature and its use by bidders, the motivations behind its use and its effects on auction outcomes are not well understood theoretically and have been little explored empirically. This paper proposes that bidders use multiple bids to adjust for winner's curse: By spreading her bids, a bidder aligns her outcome more closely to the aggregate outcome of the auction. This hypothesis is tested using bidding data from treasury bill auctions in Portugal. I find that, ceteris paribus, a bidder submits a greater number of bids and disperses prices on these bids more widely when there is a greater potential for winner's curse. In particular, both these measures of bid-spreading increase with the volatility of market interest rates and the expected number of participating well-informed bidders.