The SAS Analytics 2015 Conference is coming soon. It is my first time attending, so when I discovered that the conference is in Las Vegas, I must admit I became more than a little excited to partake in some casual gambling. My thing is sports betting, specifically college football, and you can be sure I’ll be wagering a few dollars on the various games the weekend preceding the conference. Note to readers: As of this writing, the N.C. State Wolfpack are 4-0 against the spread. But how does my interest in sports and betting relate to panel data?

When I’m not busy analyzing point spreads and over/unders, I work on the econometrics team developing software for panel data analysis at SAS, mainly PROC PANEL in SAS/ETS. My colleague Ken Sanford already linked panel analysis to sports in a post last year on using panel data to assess the economic impact of the Super Bowl. So the upcoming conference reminded me of a paper I read about a panel data analysis that measured whether crime rates were affected by the presence of nearby casinos.

The paper is by Falls and Thompson (2014), who performed a panel-data analysis of Michigan’s 83 counties over the years 1994-2010. Following the passage of the Indian Gaming Regulatory Act of 1988, nineteen Native American (and three additional non-tribal’’) casinos were established in Michigan, so the goal of the paper was to analyze how this change affected crime rates. The authors fit random-effects (RE) models of the form

$r_{it}=\alpha+C_{it} \beta+X_{it} \gamma+e_{i}+ v_{it}$

where i  is the county and t is the year. The response variable $r_{it}$ is the log crime rate for one of four crime categories: burglary, larceny, robbery, and motor vehicle theft. The $C_{it}$ variables are the casino-related variables of interest, such as the presence of a casino in the county, presence of a nearby casino (within 50 miles), casino age and casino scale of operations. The $X_{it}$ variables are control variables that are known to be associated with crime, such as population density, demographics, and police presence. The $e_{i}$ are county-level effects and the $v_{it}$ are the overall residuals.

I found the results somewhat surprising. The analysis revealed very few statistically significant effects due to the casino-related variables:

1. The rate of robberies decreases slightly as casinos age, once the “newness” wears off, I presume.
2. The rate of motor vehicle thefts decreases by about 16% when there is a nearby casino, although the reason is not clear.
3. The rate of motor vehicle thefts increases slightly the larger the nearby casino.

In short, the impact of legalized gambling on the crimes considered was minimal. That certainly bodes well for my trip to Las Vegas!

The authors fit random-effects models throughout, but they had some concern about police presence being correlated with county effects. To deal with this endogenous regressor, they employed an instrumental variables approach with bootstrap standard errors. You could reproduce their analysis in SAS with some custom bootstrap code, but it would be interesting to compare the authors’ approach with some of the more readily-available estimators that PROC PANEL supports for dealing with endogenous regressors, for example, Hausman-Taylor estimation and the generalized method of moments (GMM).

If you are curious to know more about panel data and the problem of endogeneity, you can catch my talk on Working with Panel Data at Analytics 2015. If we cross paths, I’ll be sure to let you know if the Wolfpack are good to cover against Wake Forest that Saturday.

References:

Falls, G. A. and P. B. Thompson. 2014. Casinos, casino size, and crime: A panel data analysis of Michigan counties. The Quarterly Review of Economics and Finance, 54: 123-132.

SAS Institute, The PANEL Procedure, SAS/ETS(R) 14.1. documentation

Image credit: photo by www.david_baxendale.com // attribution by creative commons

Share