This study undertakes a systematic analysis of several theoretic and statistical assumption s used in many empirical models of female labor supply. Using a singl e data set (PSID 1975 labor supply data) the author is able to replic ate most of the range of estimated income and substitution effects fo und in previous studies in this field. He undertakes extensive specif ication tests and finds that most of this range should be rejected du e to statistical and model misspecifications. The two most important assumptions appear to be the Tobit assumption used to control for sel f-selection into the labor force and exogeneity assumptions on the wi fe's wage rate and her labor market experience.

Empirical researchers sometimes misinterpret how additional regressors, heterogeneity corrections, and multilevel factors impact the interpretation of the estimated parameters in binary outcome models such as logit and probit. This can result in incorrect inferences about the importance of incorporating such features in these nonlinear statistical models. Some reports of biases in binary outcome models appear related to the arbitrary variance normalization required in binary outcome models. A focus on readily interpretable numerical quantities, rather than conveniently chosen “effects” as measured by arbitrarily scaled coefficients, would eliminate nearly all of the interpretation problems we highlight in this paper.

Most household production studies ignore unobserved inputs. Without additional assumptions, however, estimated impacts of the observed inputs cannot provide informative estimates of marginal products because of contaminating variations in unobserved inputs. Not even the signs of marginal impacts can be ascertained. One can establish the direction of these biases and significantly reduce them by including detailed information about nonproductive (pure consumption) goods in the analysis. Estimates from prior demand studies can also help determine the direction of the bias. We describe the stringent assumptions one must invoke in the absence of such additional information to be able to relate estimated effects to the true marginal products.