Incarceration length doesn’t appear to affect employment opportunities

In Incarceration Length, Employment, and Earnings, Jeffrey Kling looks to estimate the effects that incarceration length has on employment and earnings potential after release. In his analysis, which considers a variety of research designs, he finds no substantial evidence that there is a negative effect on employment or earnings level as the length of ones incarceration increases, and all together the effect is ambiguous.

Data is collected from numerous government agencies, including the administrative records of the state prison system in Florida and the federal judicial system in California that contains information about offender characteristics, incarceration experiences, and about ten years of earnings data reported through the Unemployment Insurance (UI) system. The Florida and California samples are both largely male, with the majority of the Florida sample being African-American and the California sample consisting of relatively more whites, Hispanics, and other races. An initial examination of the data revealed some interesting trends. The data from the Florida state system shows that prior to incarceration, employment rates were similar for those serving time. The data also shows that upon release the inmates’ employment rate immediately peaked for each group (a group consists of those who served the same amount of time) no matter how long they were incarcerated, but then slowly declines to levels consistent with those before incarceration. Kling suggests that this decline is due to the fact that though employed, the pay is small and offenders don’t stay at their first job for long.

Instead of comparing someone who served 1 year with someone who served 10 and then observing the difference, Kling’s goal is to isolate time served and find out what kind of difference in employment/earnings a prisoner serving one year would get if, instead, he served five years. Various different methods are utilized to examine if incarceration length influences prisoner employment and earnings. These methods include:

1. Controlling for observable differences in peoples characteristics that may be correlated with both incarceration length and employment/earnings.
2. Account for pre-existing differences in labor market prospects among individuals.
3. Examine variations in sentences that are not a result of individual characteristics but rather on the Judges strictness or leniency in ruling. For example, consider two offenders who committed a similar crime, but each was sentenced a different length of incarceration because they each had different Judges. (This is referred to as the instrumental variable analysis)

To examine the effects on medium-term labor market outcomes while controlling for individual characteristics, the employment rate amongst ex-cons is observed seven years after they were first imprisoned. The sample is limited to sentences served between .5 and 4.5 years so that data obtained in year 7 will consist of all individuals at least two years out, allowing any effects over the medium term to be examined. No significant correlation is found between incarceration length and labor market outcomes over the medium term when observable differences in people were controlled for. Also, when pre-existing differences are controlled for (different events in individuals lives before incarceration such as previous prison time) there was no statistically significant correlation in the medium term either.

Next, Kling attempts to control for Judges’ strictness or leniency. In other words, it is an attempt to examine how different prison lengths for similar offenders affects labor market outcomes. Observing two people who commit the same crime and have similar characteristics but receive different sentences, it is easier to isolate the affect of sentence length has on any differences in labor market outcomes upon the offenders’ release. For example someone got a four-year sentence for a certain crime and had he got another judge the sentence may have been 1 year. Again, however, there is no statistically significant correlation between incarceration length and labor market outcome over the medium term using this design method.

Lastly, the author examines short-term effects on labor market outcomes by focusing on inmates released at the same time, but with different incarceration lengths. Over the short-term, there is a relationship between longer incarceration time and positive labor market outcomes for inmates. However, as more factors are controlled for, the relationship is no longer statistically significant. The factor that influenced the data the most was the presence of work release programs. This explains why longer sentences were initially associated with positive labor market outcomes, because work release programs target those offenders with longer sentences and help place them in jobs. Those with shorter sentences aren’t imprisoned long enough to spend the necessary time in such programs.