|
Activity Number:
|
144
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #301106 |
|
Title:
|
Use of Optimal Instrument Variables for Estimating Spatial Modeling Parameters with Application to Crime Survey Data
|
|
Author(s):
|
Krista Collins*+ and Avinash C. Singh
|
|
Companies:
|
Statistics Canada and Statistics Canada
|
|
Address:
|
100 Tunney's Pasture Drive, Ottawa, ON, K1A 0T6, Canada
|
|
Keywords:
|
spatial autocorrelation ; spatial lag model ; estimating functions ; instrumental variables ; crime data
|
|
Abstract:
|
The problem of using instrument variables in estimating regression and correlation parameters arises in the context of spatial modeling under a quasi-likelihood framework. The reason for this is that in the joint modeling of spatial lag dependence and regression, the dependent variables also appear as regressors in the model. Existing solutions to this problem do not use optimal instrumental variables. The main purpose of this paper is to show how optimal instruments can be defined using estimating functions (Singh and Rao 1997) and to examine what gains in efficiency are obtained with this approach. The methodology is illustrated using simulations of urban neighborhood crime rates in the Canadian city of Montréal.
|
