A second reason might be a specification bias
generated by the strong assumption that we made regarding the transport costs specification (equation 16).
First, going beyond the traditional two-variable relationship, a VAR model is built in the production function context to avoid a possible specification bias
. Second, Riezman et al.
But even a sophisticated model that assumes varying transition rates and absorbs their deviations from the average transition rates across all districts into the error term (in a varying coefficient model) will fail, because these deviations are usually correlated with the independent variables to produce a classic case of specification bias
in which the error term is correlated with the included independent variables.
The answer to this puzzle, we believe, lies in the presence of specification bias
in the risk of death coefficient.
Feldstein (1970) demonstrated that studies using the nominal instead of the real interest rate were subject to a serious specification bias
. Carlino (1982) employing four different interest rates estimated contradictory signs for the interest rate coefficients and rejected the hypothesis that variations in the real interest rate altered consumption decisions.
This paper handles the problem of possible specification bias
in the money demand equation through use of extreme bound analysis [Leamer 1983; 1985] on the estimated coefficients.
This is an important question because shelter costs form the largest budget item for most consumers and, from an econometric viewpoint, failing to account for structural attributes of dwellings found in the general housing stock may result in specification bias
when estimating tenure model parameters.
Moreover, total reliance on continuous measures may create the possibility of significant specification bias
It has been shown (Heckman, 1979; Lee, 1978, 1979) that this problem of missing data on the dependent variable can be transformed into one of specification bias
due to the omission from (1) of an explanatory variable.
They analyze a significant geographic dispersion phenomenon which can cause either negative or positive bias in travel cost model (TCM) estimates of site value when travel costs to alternative sites are omitted.(1) However, the derivation of specification bias
in the TCM which they provide is incomplete and could lead to generalizations which are erroneous.