Search Results
Title: A Simple Threshold Model of Theft
Resulting in 1 citation.
| 1. |
Williams, Geoffrey |
|
A Simple Threshold Model of Theft Working Paper, Social Science Research Network (SSRN), November 14, 2011. Also: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1978855 Cohort(s): NLSY97 Publisher: Social Science Electronic Publishing, Inc. Keyword(s): Adolescent Behavior; Crime; Delinquency/Gang Activity; Heterogeneity; Modeling Permission to reprint the abstract has not been received from the publisher. I propose a simple threshold strategy model of theft in which all individuals draw theft opportunities from the same random distribution, while individuals di ffer in terms of their actual or perceived costs of theft. I estimate the model using data from the NLSY 1997 Cohort for the years 1997-2003 with a number of specifi cations, including a bivariate structural model. Across all estimations covariates that measure or are closely correlated with time preferences and impatience are strong predictors of theft while measures such as opportunity cost show little or no explanatory power. The assumption of a common distribution of opportunities is not contradicted by the data. Structural and count estimations support the conclusion that unobserved heterogeneity across individuals plays a substantial role. I uncover a previously unnoticed pattern: theft is very spiky in that the median thief is active for only a brief period of less than a year in adolescence or early adulthood. Theft thus appears to be substantially a phenomenon of high impatience individuals entering a temporary period of intensi fied risk-taking in adolescence. Finally, and in contrast to the predictions of the literature, the two count data models favored in cases of unobserved heterogeneity perform very differently, suggesting that using count models in tandem with binary models o ffers more insight than using count models in isolation. |
|
|
Bibliography Citation
Williams, Geoffrey. "A Simple Threshold Model of Theft." Working Paper, Social Science Research Network (SSRN), November 14, 2011. |