Heckman correction
From Wikipedia, the free encyclopedia
The Heckman correction (the two-stage method, Heckman's lambda or the Heckit method) is any of a number of statistical methods developed by James Heckman in 1976 through 1979 which allow the researcher to correct for selection bias. Selection bias problems are endemic to applied microeconomic problems, which make Heckman’s original technique, and subsequent refinements by both himself and others, indispensable to applied econometricians. Heckman received 2000 Economics Nobel Prize for this achievement.
When a sample fails to represent reality, the statistical analyses based on those samples can lead to erroneous policy decisions. The Heckman correction, a two-step statistical approach, offers a means of correcting for sampling errors.
Heckman discussed bias from using nonrandom selected samples to estimate behavioral relationships as a specification error. He suggests a two stage estimation method to correct the bias. The correction is easy to implement and has a firm basis in statistical theory. Heckman’s correction has a Normality assumption, provides a test for sample selection bias and formula for bias corrected model.
Suppose that a researcher wants to estimate a wage relation using individual data, but only has access to wage observations for those who work. The Heckman correction takes place in two stages. First, the researcher formulates a model, based on economic theory, for the probability of working. Statistical estimation of the model yields results that can be used to predict this probability for each individual. In the second stage, the researcher corrects for self-selection by incorporating these predicted individual probabilities as an additional explanatory variable, along with education, age, etc. The wage relation can then be estimated in a statistically appropriate way.
Heckman's achievements have generated a large number of empirical applications in economics as well as in other social sciences. The original method has subsequently been generalized, by Heckman and by others.
[edit] Disadvantages
It has no "reject inference", it does not adjust for sample bias and portfolio quality estimates will be optimistic over the rejects.
Its reclassification is ad-hoc, implies Pf(bad | X) = 1 over a segment of the covariate space. It may also bias the scoring model over the accepts.
[edit] References
|
||||||||||||||||||||
|
||||||||||||||
