The Prison Boom and the Lack of Black Progress Since Smith and Welch

Transcription

1 The Prison Boom and the Lack of Black Progress Since Smith and Welch Derek Neal Univeristy of Chicago Department of Economics, COE, and NBER & Armin Rick University of Chicago Department of Economics November, 2013

2 Smith and Welch (1989) Smith and Welch (1989) used the 1940 through 1980 census files to document important relative black progress This progress did not continue, at least among men. Since 1980: prison populations have grown tremendously especially among black men employment rates for black men have fallen relative to those for white men Neal (2006) shows no black-white skill convergence since late 1980s

3 Our Main Results - Handout After 1970, differences between the ratios of observed median wages and our selection corrected ratios grow over time. Given < 15 years of potential experience, black-white ratios of median potential wage likely remained constant or fell after 1970 Given or years of potential experience, sin observed medians again imply substantial gains, but our selection corrected ratios do not. Black men suffered significant relative losses during the Great Recession, and this is particularly true of those with 6-15 years of experience.

4 What Caused Prison Growth i =(c i) s where c = the fraction of criminals in the population = the probability of arrest given engagement in crime = the probability of conviction given arrest = the probability of admission given conviction s = the expected time served given admission

5 Standard Approach Growth in incarceration is typically a consequence of growth in one or more of the sequence of stages leading to an increased prison population. Those stages begin with commission of crime,...some combination of increased effectiveness by prosecutors and punitiveness by judges in convicting arrested offenders and sending them to prison; increases in time served once sent to prison because of longer sentences (including mandatory minimum sentences), because the parole boards or other release policies are slower in offering release or because of more aggressive policies in recommitting parolees, either for a new offense or for a technical violation."

6 Problem Note discuss the last two terms, and s. mplicitly (wrong) assertion if there are changes in policy that lead to more long prison spells, e.g. mandatory minimum sentence provisions, restrictions on release to parole, etc., then researchers can detect the effects of these changes in policy by measuring changes in time-served among those admitted to prison.

7 Better Approach 0 s = k s 8s > 0 with k > = 1 S s=1 0 s. i =(c i) SX s s s=1 s = P S s=1 s 0 s P S s=1 0 s = P S s=1 s s P S s=1 s

8 New Parameters Further, we can create infinitely many new matrices of sentencing weights with elements, 0 sj = k sj sj, s.t. higher admission rates larger steady-state prison populations NO changes in the distribution of time-served among admitted prisoners. Our focus is not s but whether or not k sj > 1

9 Better Data The NCRP data provide detailed records of admissions and releases for many states in many years from 1983 to 2009 as well as stocks of prisoners in custody for years 2005 through Restricting our attention to states that filed NCRP reports on a fairly consistent basis with consistent release and admission data in the NCRP with consistent flows and age-specific stocks in the post-2005 NCRP files checks with stocks and flows from NPS

10 Better Data Eleven states where NCRP data contain only minor problems. We use data from 8 of the 11 states. CA, CO, M, ND, NJ, SC, WA, and W. NE, NY, and UT have release records with high missing rates concerning the type of admission or year of admission to prison in several years.

11 What are the k sj values? For all non-violent crimes, each k sj entry > 1, most are > 2 Drug Crimes, Simple Assault, Motor Vehicle Theft stand out The results for violent crime still imply a shift to harsher punishments Race specific k sj values tend to be larger for Blacks. But, some differences in sj values favored whites at baseline in 1985, i.e. drug crimes.

12 The nputs nto The Simulation 1. the probability that an offender arrested in 1985 enters prison as a new court commitment 2. the probability that a person who was paroled from prison in year (1985 p) enters prison in 1985 as the result of a parole revocation 3. the probability that a person exits prison to parole in year ( s) assuming that he entered prison in 1985 due to a new court commitment. 4. the probability that a person exits prison to parole in year ( s) assuming that he entered prison in 1985 due to a parole revocation. 5. the probability that a person exits prison without parole supervision in year ( s) assuming that he entered prison in 1985 due to a new court commitment. 6. the probability that a person exits prison without parole supervision in year ( s) assuming that he entered prison in 1985 due to a parole revocation.

13 Results

14 Results - Adjusted for ncapcitation

15 By Race

16 By Race