We study nonparametric and regression estimators of treatment and spillover effects when interference is mediated by a network. Inference is nonstandard due to dependence induced by treatment spillovers and network-correlated effects. We derive restrictions on the network degree distribution under which the estimators are consistent and asymptotically normal and show they can be verified under a strategic model of network formation. We also construct consistent variance estimators robust to heteroskedasticity and network dependence. Our results allow for the estimation of spillover effects using data from only a single, possibly sampled, network.

Empirical studies of forecasts often fail to reconcile the rational expectations hypothesis with a minimum mean square error objective function. Recent studies, however, have argued that observed bias may be rational in certain advising games, or for objective functions that include publicity or forecasting reputation as additional arguments. This paper analyzes multistep forecasting behavior for individuals forecasting bond yields in the Blue Chip Financial Survey over the 1987–1996 period and uncovers statistically significant evidence supportive of Ehrbeck and Waldmann's rational stubbornness. I find that forecasters rationally place too much weight on their previous forecasts in an attempt to mimic the behavior of more able forecasters (perhaps attempting to fool their clients). Jointly, I also find that this pattern of under-revision is positively correlated with mean square forecasting errors. Rational stubbornness is sensitive to the forecasting horizon as well as bond maturity.

Current sales of most consumer durable goods are accounted for by replacements. However, only in recent years has the economic literature provided a more rigorous analysis of replacement purchases by incorporating elements of dynamic programming and of the theory of stochastic processes. This paper is an empirical study of household replacement decisions modeled as an optimal stopping rule. Using data from the Residential Energy Consumption Survey (RECS) of the U.S. Department of Energy, we conclude that demographic variables, operation and replacement costs, and equipment characteristics may affect ownership spells of appliances such as electric heaters and central air conditioners.

This study examines the Treasury's experiment with single-price bond auctions in the mid-1970s and finds that, controlling for factors unrelated to auction technique, markups of auction average rates over when-issued rates shortly after auctions were a statistically significant seven to eight basis points higher at single-price auctions than at discriminating-price auctions. These results suggest that single-price auctions raised Treasury borrowing costs by roughly 3/4 percent of the issuing price of auctioned securities.

The results of forecasting experiments based on an error correction mechanism (ECM) model and various types of vector autoregressive (VAR) and Bayesian vector autoregressive (BVAR) models are presented. A Bayesian error correction mechanism (BECM) model is also tested. This model represents a hybrid of the BVAR and ECM models. The results from experiments using fifty industries and monthly Ohio labor market data demonstrate that the ECM model produces forecasts with much lower errors than any of the alternative VAR or BVAR models when the variables used in the model pass the statistical tests for cointegration. The findings confirm many of the beliefs expressed by Granger (1986) and Engle and Yoo (1987) based on theoretical consideration of the ECM model versus the VAR model. A result contradictory to the contentions of Engle and Yoo is that the BECM model performs well at the longer forecast horizons for both cointegrated and non-cointegrated industries.

The economic consequences of debilitating illness are defined and then estimated for one such illness--multiple sclerosis. Economic losses are defined as the consumption losses to the affected household because of illness. These losses are measured as the change in earnings of all family members plus the increase in gross (not out-of-pocket) medical costs. Earnings models are specified and estimated, and gross medical costs calculated for multiple sclerosis. The average annual loss to the multiple sclerosis household is $5, 336 per year in 1976 dollars; the estimated aggregate decline in consumption for the year 1976 was $0.656 billion. Lifetime costs (discounting by 0.06) total $207, 200 per multiple sclerosis household and $25.50 billion for society as a whole for the current (1976) pool of multiple sclerosis patients, and $30.45 billion for all future multiple sclerosis patients.

T HE Heckscher-Ohlin (H-O) theory suggests international trade is determined by relative resource supplies among countries. Prior empirical research, such as work on the determinants of U.S. comparative advantage in a single year, has concentrated on the static predictions of H-O theory. However, H-O theory also suggests changes over time in resource supplies will alter trade structure. The present paper investigates this aspect of H-O theory. Particular emphasis is given to the role of world resource changes as an explanation of changes in U.S. trade and of the increased international competition in manufactured goods faced by the United States. Some recent studies have examined the relationship between changes in resource endowments and trade. Heller (1976) examined changes in the factor content of Japan's trade between 1958 and 1968 and found the observed patterns-consistent with changes in Japan's physical and human capital endowments. Stern and Maskus (1981) investigated changing factor input determinants of U.S. trade by estimating annual cross-section regressions over 1958-77. They suggested U.S. net exports made less direct use of unskilled labor over time. Further, analysis of the factor content of U.S. trade suggested increased U.S. abundance in physical capital relative to human capital between 1958 and 1971. Balassa (1979), using a 1970 crosssection of countries, concluded that physical and human capital accumulation largely explained changing patterns of comparative advantage in manufactures. Although these studies made important contributions, a number of issues remain concerning the relationship between resources and trade. Stern and Maskus, in listing directions for further research, cite first an examination of How and why endowments of physical capital, human capital and labor have changed within the U.S. and our major trading partners. This paper reports a substantial data effort which addresses this topic. Another important consideration is that the direct effect of resource endowment variation on trade has yet to be determined. Previously, inferences about the effect of resource endowments on trade have been based primarily on results from industry cross-section regressions. Authors of such work indirectly infer the effect of resource variation on trade by assuming the coefficients from such regressions reflect resource abundance. However, Leamer and Bowen (1981) recently demonstrated that signs of coefficients from such regressions need not reflect a country's true resource abundance. Thus, the usual negative coefficient for the capital-labor ratio in an analysis of U.S. trade cannot be used to infer the scarcity of capital and thus cannot be used to infer the effect of an increase in capital endowment on U.S. trade. Similarly, Balassa's procedure of first regressing, for each of a sample of countries, industry trade on input intensity and then using the estimated coefficients as the dependent variable in a cross-country regression on resources is an inappropriate method for inferring the effect of resources on trade. This paper advances consideration of these issues by investigating aspects of the relationship between resources and trade. Section II examines changing patterns of resource supply among thirty-four countries over 1963-75. Section III investigates whether these resource changes are associated with altered comparative advantage in manufactured goods. Section IV uses cross-country regressions to estimate the resource endowment, as opposed to factor input, determinants of U.S. manufacturing trade and thereby the direct effect of resource variation on U.S. trade. The Received for publication April 23, 1981. Revision accepted for publication November 30, 1982. * New York University. This paper is an outgrowth of research in Bowen (1980a) and of further work conducted at UCLA under a Ford Foundation grant directed by Edward E. Leamer. An earlier version was presented at the 1980 Southern Economic Association meetings in Washington, D.C. Comments by C. Michael Aho, Robert Baldwin, Edward Leamer, Joseph Pelzman, Leo Sveikauskas and an anonymous referee are gratefully acknowledged. The author remains responsible for errors.