The Hispanic-American Incarceration Rate Has Fallen

But it remains relatively high even after adjusting for age and sex

By Jason Richwine on September 7, 2021

Because so many immigrants to the U.S. in the past 50 years have come from Latin America, the socioeconomic status of Hispanic Americans is of interest to immigration researchers. Stanford Professor Keith Humphreys recently pointed out the encouraging news contained in a Bureau of Justice Statistics (BJS) report: The Hispanic incarceration rate has been moving downward toward the non-Hispanic white rate, especially over the last 10 years. Humphreys suggested the decline is all the more notable because Hispanics are younger on average than non-Hispanic whites, and young adults are of course the most crime-prone age group.

So what do the age-adjusted numbers look like? The BJS report cannot tell us, but we can use the American Community Survey (ACS) to investigate. The ACS is an annual mini-census that records whether respondents are “institutionalized”. Institutionalization is not the same as incarceration, but among the non-elderly it’s a decent approximation. (See “Methodological Notes” below for more on this topic.)

Each cell in the following table displays a ratio — the Hispanic institutionalization rate divided by the non-Hispanic white rate — and each ratio is statistically adjusted to account for differences in age, sex, and region of residence. The top portion uses 2009 ACS data the bottom portion uses 2019 data. For example, U.S.-born Hispanics with Mexican ancestry had an institutionalization rate that was 2.3 times as high as non-Hispanic whites of the same, age, sex, and region in 2009. By 2019, that ratio had declined to 1.9.

Hispanic Institutionalization Rate
Divided by the Non-Hispanic White Rate


Ancestry Foreign-Born U.S.-Born Overall
Mexican 1.1 2.3 1.7
Puerto Rican* 3.2 3.1 3.2
Cuban 1.4 1.8 1.5
Dominican 1.4 1.7 1.5
Central American 0.9 1.1 0.9
Overall Hispanic 1.2 2.4 1.7


Ancestry Foreign-Born U.S.-Born Overall
Mexican 0.9 1.9 1.4
Puerto Rican* 2.4 2.1 2.2
Cuban 0.9 1.1 1.0
Dominican 1.1 1.8 1.3
Central American 1.0 1.2 1.0
Overall Hispanic 1.0 2.0 1.5

Source: American Community Survey, 2009 and 2019.

Figures are adjusted for age, sex, and region of residence.

* Puerto Ricans are U.S. citizens but are considered "foreign-born"
in this analysis if they were born outside of the 50 states.

The table shows a mixed picture. Institutionalization rates have certainly declined among Hispanics, to the point where foreign-born Hispanics have reached parity with non-Hispanic whites. At the same time, rates remain high in the U.S.-born Hispanic category. In fact, U.S.-born Hispanics are still twice as likely to be institutionalized as non-Hispanic whites, even after adjusting for differences in age, sex, and region. Whether the overall decline can continue as the Hispanic population shifts increasingly into the U.S.-born category remains to be seen.

Furthermore, although the recent decline would ideally be the product of cultural assimilation, it may have more prosaic explanations. For example, it was widely reported during the early years of the Trump administration that Hispanics had become less willing to report crime, including intra-ethnic crime, although more research is needed on that issue. In addition, perhaps the lower background rate of crime in the U.S. tends to compress group differences, only for those differences to “decompress” when the background rate turns higher. We’ll continue to watch these trends with interest.

Methodological Notes

I first limited the sample to people between the ages of 16 and 64, and then used regression analysis to derive risk ratios that are adjusted for differences in age, sex, and region of residence. The age controls are eight age groupings covering about six years each. The region controls are the nine “divisions” identified by the Census Bureau.

In the ACS, institutionalization covers not just correctional facilities, but also mental hospitals and nursing homes. Setting the maximum age at 64 excludes most of the nursing home residents, and lower age caps did not produce substantively different results. As for distinguishing between foreign- and native-born respondents in institutions, the Census Bureau has improved its ability to do so since 2000, but some lingering problems may remain.

I employed a similar analysis for the American Enterprise Institute in 2010.