This note illustrates the connections between the Hessians of numerical optimization problems, variance-covariance matrices for parameter vectors, and the influence that data mismeasurement may have on parameter estimates. Condition numbers provide a central guide to the sensitivity of common numerical problems to data mismeasurement. Examples are provided that clarify their importance. Two simple prescriptions arise from this analysis. First, data must be of an ‘appropriate’ scale. In some cases this means that the data need similar means and similar variances. Second, in numerical algorithms it is desirable to ascertain the condition number of the Hessian implied by the initial parameter values used for numerical optimisation algorithms. Condition numbers are easy to compute and indicate whether the updates from an initial starting value are likely to be poor.
Download Info
To download:
If you experience problems downloading a file, check if you have the
proper application to
view it first. Information about this may be contained
in the File-Format links below. In case of further problems read
the IDEAS help
file. Note that these files are not on the IDEAS
site. Please be patient as the files may be large.