The motivation for this research derives from the pressing need to assess the reliability of econometric models for cross-section data. The potentially large amount of information now available in cross-section surveys has led to a recent explosion in applied economic studies using these data sources. However, although many policy prescriptions are based on models estimated using such data, for example in taxation and labour supply, the properties of estimators for these models rely heavily on strong and usually untested stochastic assumptions. This importance of the stochastic specification plays a critical role in cross-section models in particular wherever the data under study suffer from censoring, truncation or grouping. In such common situations, the properties of the estimators depend crucially, for example, on the whole shape of the assumed distribution of unobservables. Despite the importance of these assumptions in econometric models for cross-section data, it is only recently that simple tests for their reliability have been developed. Furthermore, there has been little systematic application of these tests to cross-section models or data sets commonly used in economics and no controlled experiments or finite sample analyses to assess their performance on such data. It is this gap that the research intends to fill.



Testing the normality assumption multivariate simultaneous limited dependent variable model

Author: R J Smith Date: 11 September 2000 Journal article

Distributional specification tests against semiparametric alternatives

Author: Simon A Peters Date: 11 September 2000 Conference paper/presentation

Estimation in a class of simultaneous equation limited dependent variable models

Author: Richard Blundell Date: 11 September 2000 Conference paper/presentation

Distributional specification tests against semiparametric alternatives

Author: Simon A Peters Date: 01 January 1991 Journal article

Finite sample behaviour of heteroskedasticity robust wald tests

Author: Andrew Chesher Date: 01 January 1991 Journal article

Specification error detetction in the multinominal logit model

Author: Andrew Chesher Date: 01 January 1991 Book

Asymptotive expansions of the information matrix test statistic

Author: Andrew Chesher Date: 01 January 1991 Article (submitted)

Hajek inequalities, measures of leverage and the size of heteroskedasticity robust wald test

Author: Andrew Chesher Date: 01 January 1989 Journal article

On the use of distributional unspecification checks in limited dependent variable models

Author: R J Smith Date: 01 January 1989 Journal article

Approximations to the distributions of heterogeneity tests in the censored linear model

Author: Andrew Chesher Date: 01 January 1989 Book

Symmetry, regression design and sampling distributions

Author: Andrew Chesher Date: 01 January 1988 Book

The bias of a heteroskedasticity consistent covariance matrix estimator

Author: Andrew Chesher Date: 01 January 1987 Journal article

Start date
30 September 1986
End date
30 March 1989
Co-applicants
Professor Richard Blundell
Grant amount
£31,950.00
Grant reference
RB00232150
Grant type