The Effect of E-Verify Laws on Crime

Abstract

E-Verify laws, which have been adopted by 23 states, require employers to verify whether new employees are eligible to legally work prior to employment. This study explores the impact of state E-Verify laws on crime. Using data from the 2004–2015 National Incident Based Reporting System, the authors find that the enactment of E-Verify is associated with a 7% reduction in property crime incidents involving Hispanic arrestees. This finding was strongest for universal E-Verify mandates that extend to private employers and its external validity bolstered by evidence from the Uniform Crime Reports. Supplemental analyses from the Current Population Survey suggest two mechanisms to explain this result: E-Verify-induced increases in the employment of low-skilled natives of Hispanic descent and out-migration of younger Hispanics. Findings show no evidence that arrests were displaced to nearby jurisdictions without E-Verify or that violent crime or arrests of African Americans were affected by E-Verify laws. The magnitudes of the estimates suggest that E-Verify laws averted $491 million in property crime costs to the United States.

Keywords

E-Verify immigration crime employment

Interior enforcement policies have grown substantially in the United States during the past two decades (Michaud 2010; Miles and Cox 2014; Treyger, Chalfin, and Loeffler 2014). One of the most widespread of these policies requires employers to verify the work eligibility of recent hires. Such statutes, known as E-Verify mandates, have been adopted by 23 states and the federal government.

Under an E-Verify mandate, employers must compare information from their new hires’ Employment Eligibility Verification (I-9) forms with electronic records from the Social Security Administration and Department of Homeland Security (DHS 2018). An employee’s name, Social Security number, date of birth, citizenship status, and (if applicable) non-citizen-related information is compared to electronic federal records to assess work eligibility. A mismatch prompts an alert to the employer that must be resolved within 10 federal workdays or the employee must be fired. Failure to comply with state E-Verify laws can result in substantial fines for employers as well as business license revocation. The majority of state E-Verify mandates (16/23) apply to public employers or private employers with public contracts, while eight (8) state mandates extend to private employers (NCSL 2015).¹

This study asks, Do E-Verify laws have important spillover effects on crime? The impact of state E-Verify mandates on crime is theoretically ambiguous. The policy’s net crime impact depends on the magnitudes of its effects on 1) labor market outcomes for low-skilled immigrants (unauthorized and authorized) and natives, 2) mobility of affected workers, and 3) the distribution of these effects across low-skilled populations with heterogeneous propensities for crime. On one hand, E-Verify may reduce unauthorized immigrants’ labor market prospects (Amuedo-Dorantes and Bansak 2014; Orrenius and Zavodny 2015), leading to an increase in property or drug crimes for income-generating purposes (Raphael and Winter-Ebmer 2001; Öster and Agell 2007). Furthermore, if unauthorized immigrants serve as complements to higher-skilled native workers (Lee, Peri, and Yasenov 2017; East, Luck, Mansour, and Velasquez 2018), unemployment among natives may increase crime. E-Verify laws may also shift the population composition away from likely unauthorized immigrants and toward low-skilled natives (Good 2013; Bohn, Lofstrom, and Raphael 2014; Orrenius and Zavodny 2016).

On the other hand, if low-skilled natives and immigrants are labor substitutes (Amuedo-Dorantes and Bansak 2014; Orrenius and Zavodny 2015), then improved labor market prospects for natives may reduce crime involving natives. E-Verify mandates may also reduce crime if a net out-migration of low-skilled Hispanics occurs.² Finally, E-Verify may reduce unauthorized immigrants’ willingness to report crime due to fear of detection.

Using data from the 2004–2015 National Incident-Based Reporting System (NIBRS) and the Uniform Crime Reports (UCR)—and a difference-in-differences empirical strategy—this study is the first to comprehensively examine the impact of state E-Verify mandates on crime, with particular attention to criminal incidents involving arrestees of Hispanic descent. Supplemental analysis of the 2004–2015 Current Population Survey (CPS) explores potential mechanisms through which E-Verify could affect crime: 1) employment opportunities for low-skilled immigrants and natives, and 2) the demographic (i.e., low-skilled immigrant) composition of states.

Background and Literature

History of Interior Immigration Reform

The Immigration Reform and Control Act (IRCA) of 1986 granted amnesty to previously unauthorized immigrants, while also attempting to limit the involvement of unauthorized immigrants in the US economy through mandated use of the I-9 system. However, IRCA did not require employers to verify the authenticity of workforce eligibility documentation provided by employees “if the document reasonably appear[ed] . . . to be genuine” (Kerwin and McCabe 2011). Employers complying in good faith with the I-9 system were entitled to an affirmative defense to federal sanctions (Castillo and Schulman 2011). During this period, false documentation was common, penalties were rarely administered (Baker 2015; Orrenius and Zavodny 2015), and attempts to increase enforcement were met by political opposition (Hanson 2006).

A decade later, the Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA) of 1996 addressed several problems left by IRCA. IIRIRA introduced the 287(g) program, authorizing Immigration and Customs Enforcement (ICE) to enter into agreements with state and local law enforcement agencies. The agreements allow these agencies to carry out some immigration enforcement actions, including, in some cases, interrogation and arrest of non-citizens.³ Later, the Secure Communities program was launched to target criminal unauthorized immigrants by comparing fingerprints collected at the time of arrest with the DHS Automated Biometric Identification System (Kubrin 2014). The system notifies ICE when a deportable criminal is arrested, though there is little evidence that Secure Communities had a discernable effect on crime (Miles and Cox 2014; Treyger et al. 2014).

IIRIRA also introduced an electronic employment verification system known as the Basic Pilot Program, which was made available to employers in all states for voluntary use in 2003 (National Immigration Law Center 2011), becoming the precursor to E-Verify. Colorado became the first state to mandate E-Verify for public employers and state contractors in 2006. Subsequently, 22 additional states have enacted E-Verify laws, with the majority of these mandates applying to public employers (or private employers with state contracts). Eight states mandate all employers use E-Verify. In addition, E-Verify is required for all federal workers and contractors and is annually reauthorized by Congress (Park and Friedman 2008).

Enforcement of E-Verify occurs at both the federal and state levels. The Monitoring and Compliance Branch of US Citizenship and Immigration Services (USCIS) enforces federal E-Verify policies with the goal of reducing abuse and fraud. Most state E-Verify laws include sanctions for non-compliance, such as fines (usually between $250 and $1,000 per violation, but sometimes reaching as high as $10,000 per offense), termination of business licenses, or temporary bans from state contracts. Some states grant firms immunity from liability for employing undocumented workers if E-Verify was used (Park and Friedman 2008). In 2015, 50% of new hires nationwide were verified through the system (Orrenius and Zavodny 2017), representing a 150% increase from 2011 (Rosenblum 2011).

Labor Market Effects of E-Verify Laws

The first wave of studies on E-Verify examined the labor market effects of these mandates. Using data from the 2004–2010 Current Population Survey (CPS), Amuedo-Dorantes and Bansak (2012) found that E-Verify mandates were associated with a 3 to 7% decline in employment for likely unauthorized immigrants. Similarly, Bohn and Lofstrom (2012) found that Arizona’s E-Verify law reduced employment of likely unauthorized men by 11 percentage points and shifted some toward informal work. Only limited evidence suggests that E-Verify laws increase job lock (Orrenius, Zavodny, and Gutierrez 2018).

E-Verify mandates also generate important spillover effects. Orrenius and Zavodny (2015) used data from the 2002–2012 CPS to explore the impact of E-Verify on employment and wages among workers who may compete with unauthorized immigrants. They found universal E-Verify mandates increased wage-and-salary employment among Mexican-born naturalized citizens by 8 percentage points and increased real earnings among native-born Hispanics by 9%. The authors found little evidence, however, that E-Verify affected employment among non-Hispanic whites. Of interest, these authors found evidence that E-Verify laws increased labor force participation among likely unauthorized females (often in the informal sector), and they attributed this result to female spouses responding to the decline in formal-sector employment for their husbands.

E-Verify may also affect labor market outcomes through selective migration. Bohn et al. (2014) used data from the 1998–2009 CPS to examine the impact of Arizona’s E-Verify mandate. Using a synthetic control approach, they found this law was associated with a 2 to 3% reduction in the share of the state population comprising non-native Hispanics. Extending this analysis to E-Verify mandates adopted nationwide, Orrenius and Zavodny (2016) found that E-Verify mandates were associated with a 50% reduction in the number of newly arriving, low-skilled, prime-age immigrants from Mexico and Central America.

Immigration, Labor Market Opportunities, and Crime

Becker’s (1968) theory of rational crime suggests that criminal behavior should be responsive to labor market conditions. Indeed, empirical studies that analyzed those on the margin of criminal behavior found that arrests are positively related to local unemployment rates (Levitt 2004; Lin 2008; Schnepel 2018) and business cycle contractions (Rosenfeld and Fornango 2007), and are negatively related to low-skilled wages (Gould, Weinberg, and Mustard 2002).

The relationship between labor market opportunities and crime appears to extend to immigrants. Baker (2015) found IRCA’s amnesty provision reduced property crime by 3 to 5% through enhanced levels of human capital and greater labor market opportunities. Subsequently, Freedman, Owens, and Bohn (2018) found that increased barriers to legal employment associated with the expiration of IRCA’s amnesty provisions led to a 59% increase in felonious prosecutions of Hispanic residents of Bexar County, Texas, for income-generating crime (i.e., theft, prostitution, fraud).⁴

E-Verify and Crime

Two papers of which we are aware have explored the relationship between E-Verify and arrests. Zhang, Palma, and Xu (2016) used 1998–2014 UCR data to examine the impact of Alabama’s HB 56 E-Verify statute on arrests. Using a synthetic control approach, the authors found that Alabama’s E-Verify mandate was associated with approximately 100 more violent crime arrests per year for every 100,000 adults, but it was statistically unrelated to property crime arrests. The authors attributed this result to diminished labor market opportunities for likely unauthorized immigrants but offered little compelling reason why violent crime rose while income-generating property crime was unaffected.⁵

Chalfin and Deza (forthcoming) explored the effect of Arizona’s E-Verify mandate on arrests using UCR data and two identification strategies: a state-level synthetic control approach and an agency-level difference-in-differences approach. They found that the mandate was associated with a 20% decrease in property crime arrests involving young men ages 15 to 24, but they found no statistically significant change in violent crime arrests. The authors posited that the property crime reduction was due to out-migration of foreign-born Mexicans.

The current study makes a number of contributions. It is the first to comprehensively examine the effect of state E-Verify laws on crime, exploiting policy changes in up to 23 US states. Second, we provide the first estimates of the effect of E-Verify on criminal incidents involving Hispanic arrestees. This contribution is important given that the E-Verify literature suggests heterogeneous labor market and migration effects for both Hispanic unauthorized immigrants and natives of Hispanic descent. Moreover, given that native Hispanic arrestees make up the majority of all Hispanic arrestees, the impact of E-Verify on immigrants alone will fail to capture the full policy impact.⁶ Third, we explore spillover effects of E-Verify on arrests involving African American and white adults. Finally, this study is the first to examine whether the crime effects of E-Verify laws differ by whether mandates require private as well as public employers to comply with E-Verify.

Data and Methods

National Incident-Based Reporting System

We use agency-by-month data from the 2004–2015 National Incident-Based Reporting System (NIBRS) to estimate the impact of state E-Verify mandates on criminal incidents involving Hispanic arrestees. The NIBRS is collected by the Federal Bureau of Investigation (FBI) and includes incident-level crime data collected from local law enforcement agencies. The reporting agency is usually a city or township, with NIBRS agencies often serving relatively small populations.⁷ Approximately 93 million Americans, or 29% of the US population, are covered by the NIBRS, accounting for 27% of all US crime (FBI National Press Office 2015).

The NIBRS data are especially useful for this study because they contain information on the arrestee’s ethnicity; however, their coverage is much sparser than the UCR. Thus, in contrast to studies using UCR data to generate county or state arrest rates (Gould et al. 2002; Cáceres-Delpiano and Giolito 2012; Anderson 2014), best practices using NIBRS data examine agency-level observations in panel-based analyses with agency fixed effects (Card and Dahl 2011; Heaton 2012; Lindo, Siminski, and Swensen 2018).⁸

We measure criminal incidents at the agency-by-month level. Our main sample consists of a balanced panel of agency-months, which is expected to reduce measurement error generated by clustered or inconsistent crime reports. However, we experimented with alternate samples, including agencies that reported in at least half the years covering the sample period, or agencies serving counties of at least 10,000 population. These strategies produced a qualitatively similar pattern of results.

Our main outcome from the NIBRS, Hispanic property crime, is an agency-by-month count of property crime incidents involving a Hispanic arrestee age 16 to 64. Property crime is defined as burglary, larceny theft, motor vehicle theft, and arson. We generate Hispanic violent crime analogously. These crimes include murder, rape, robbery, and aggravated assault.⁹ In Online Appendix Table A.1A (hereafter, numbering for Online Appendix material is prefaced with an “A.”), we report the mean number of property and violent criminal incidents involving Hispanics ages 16 to 64, and populations of non-Hispanic whites and African Americans, respectively, are also explored below.

Uniform Crime Reports

While the NIBRS data allow us to identify criminal incidents involving Hispanic arrestees, these data lack national coverage. By contrast, the UCR data are representative of the entire United States, covering 98% of the population. As shown in Table 1, all 23 states that enacted E-Verify legislation between 2004 and 2015 contribute to identification in the UCR. Unfortunately, the UCR do not report arrestee ethnicity. We are able to measure criminal arrests of adults by race only, and race is coded as white or African American. According to 2004 data from the Surveillance Epidemiology and End Results (SEER) program, 15.7% of all whites in the United States were of Hispanic descent and 4.6% of all African Americans were of Hispanic descent.¹⁰ While we cannot measure ethnicity in the UCR, we do explore heterogeneity in the effects of E-Verify laws on arrests of whites by jurisdictions that had relatively higher shares of Hispanic white residents in 2004 prior to the adoption of E-Verify laws.

Table 1.

Effective Dates of State E-Verify Laws, 2004–2015

			Identifying variation?
State	Effective date	Coverage	NIBRS	UCR	CPS
Alabama	April 1, 2012	Universal	No	Yes	Yes
Arizona	December 31, 2007	Universal	No	Yes	Yes
Colorado	August 7, 2006	Public	Yes	Yes	Yes
Florida	January 4, 2011	Public	No	Yes	Yes
Georgia	July 1, 2007	Public	No	Yes	Yes
	January 1, 2012	Universal	No	Yes	Yes
Idaho	July 1, 2009	Public	Yes	Yes	Yes
Indiana	July 1, 2011	Public	No	Yes	Yes
Louisiana	August 18, 2011	Partial	Yes	Yes	Yes
Michigan	March 1, 2013	Public	Yes	Yes	Yes
Minnesota	January 1, 2008	Public	No	Yes	Yes
Mississippi	July 1, 2008	Universal	No	Yes	Yes
Missouri	January 1, 2009	Public	No	Yes	Yes
Nebraska	October 1, 2009	Public	Yes	Yes	Yes
North Carolina	January 1, 2007	Public	No	Yes	Yes
	October 1, 2012	Universal	No	Yes	Yes
Oklahoma	February 2, 2010	Public	No	Yes	Yes
Pennsylvania	January 1, 2013	Public	No	Yes	Yes
Rhode Island	October 17, 2008^a	Public	Yes	Yes	Yes
South Carolina	January 1, 2009	Public	Yes	Yes	Yes
	January 1, 2012	Universal	Yes	Yes	Yes
Tennessee	October 1, 2011	Partial	Yes	Yes	Yes
Texas	September 1, 2015	Public	Yes	Yes	Yes
Utah	July 1, 2009	Public	Yes	Yes	Yes
	July 1, 2010	Universal	Yes	Yes	Yes
Virginia	December 1, 2012	Public	Yes	Yes	Yes
West Virginia	June 24, 2012	Public	Yes	Yes	Yes

Sources:National Conference of State Legislatures (2015) and Urban Institute (2017).

Notes: CPS, Current Population Survey; NIBRS, National Incident-Based Reporting System; UCR, Uniform Crime Reports.

The E-Verify mandate enacted in Rhode Island in 2008 was repealed on January 5, 2011.

In contrast to the NIBRS, which is measured at the incident level, the UCR data are measured at the arrest level. We measure agency-by-month criminal arrest counts for adults ages 18 and older. We restrict our sample to agency-month observations that report in at least 90% of our sample period. Our key outcome variables in the UCR include African American property arrests, African American violent arrests, White property arrests, and White violent arrests. As noted above, to examine E-Verify law effects on Hispanics, we interact indicators for quartiles of the distribution of white Hispanics in the state in 2004 with the E-Verify policy. Table A.1B shows descriptive statistics for arrests from the UCR.

Current Population Survey

To explore the mechanisms through which E-Verify may affect crime, we use the 2004–2015 Current Population Survey (CPS) Basic Monthly Data (BMS). The surveys are administered by the US Bureau of Labor Statistics and are representative of the US population when weighted using appropriate sample weights. The CPS data include information on labor market outcomes, citizenship status, and other demographic characteristics.

In analyzing the potential mechanisms, we restrict our sample to lower-skilled individuals ages 16 to 64 with at most a high school diploma (Orrenius and Zavodny 2015). We measure whether the respondent is employed (Any employment) and then isolate work for pay (Employed, salary and wages). We focus on the labor market effects of E-Verify separately for 1) likely unauthorized immigrants defined as less-educated non-citizens of Hispanic descent (Amuedo-Dorantes and Bansak 2014; Orrenius and Zavodny 2015), and 2) US-born Hispanics (citizens). In supplemental analysis, we also examine naturalized citizens of Hispanic descent. Summary statistics of each of the above variables are available in Table A.3.

Methods

In the NIBRS, our unit of analysis is the agency-by-month from 2004–2015 and our dependent variable is a count of criminal incidents involving working-age individuals of Hispanic descent. Following Card and Dahl (2011) and Lindo et al. (2018), we first estimate the following Poisson regression model:

H i s p a n i c C r i m e_{a t} = κ_{a t} Exp (β_{0} + β_{1} E V e r i f y_{s t} + β_{2} l o g (p o p_{a t}) + β_{3}' X_{c t} + β_{4}' Z_{s t} + α_{a} + τ_{t} + ε_{a t})

(1)

where our primary outcome of interest, $Hispanic Crim e_{at}$ , measures the number of criminal incidents involving a 16- to 64-year-old Hispanic arrestee reported by agency a in month-by-year t. Each agency is linked to its primary county and state to code the covariates in Equation (1). Our key policy variable, EVerify_st, is an indicator for whether an E-Verify mandate had been enacted in state s at time t (NCSL 2015; Urban Institute 2017). In alternate specifications, we explore heterogeneous treatment effects by whether the law applies solely to public employers (Public E-Verify), is mandatory and extends to all private employers (Mandatory universal), or extends to private employers, but allows employers an alternate means of compliance (Partial E-Verify).

Our controls include the following: log(pop_at) denotes the natural log of the agency-level population;¹¹ the vector X_ct includes county-level controls including demographic characteristics (the age distribution of the county population and the shares of the county population that are male and African American); county-level immigration policies (an indicator for the presence of a 287(g) program, Secure Communities, and an omnibus immigration law); and an indicator for the presence of a ban-the-box law (Doleac and Hansen 2017, 2020; Sabia, Mackay, Nguyen, and Dave 2021).¹² The vector Z_st includes state-level controls for economic conditions (the natural log of per capita income, the natural log of the state unemployment rate, and the share of population ages 25 and older with a bachelor’s degree); the political climate (an indicator for whether the governor is a Democrat); crime policy controls (the natural logs of lagged police expenditure per capita and lagged police employment per capita); gun policy controls (shall issue concealed carry permit laws [see Donohue et al. 2019], stand-your-ground laws, and the lagged number of background checks); and social policy controls (the natural log of the state minimum wage, the refundable Earned Income Tax Credit [EITC] refundable credit rate, Affordable Care Act-related [ACA] Medicaid expansions, and Supplemental Nutrition Assistance Program [SNAP] asset test vehicle exemptions). (See Table A.4 for the sources for these variables.) Finally, θ_a is a time-invariant agency fixed effect and τ_t is an agency-invariant month-by-year fixed effect.

Identification of β₁ comes from temporal variation in E-Verify laws across states. Figure A.1 depicts the states adopting an E-Verify mandate between 2004 and 2015, and Table 1 lists the effective dates of the laws and comprehensiveness of the state statute. For our NIBRS-based analysis, 12 states contribute to identifying variation. Figure A.2 shows trends in crime for E-Verify and non-E-Verify states.

A causal interpretation of β₁ requires that crime would have evolved similarly in the treatment and control states in absence of the policy change. While this is fundamentally untestable, we examine whether crime among Hispanics evolved similarly prior to the implementation of an E-Verify mandate in treatment and control states:

\begin{array}{l} Hispanic {Crime}_{a t j (s, t)} = κ_{a t} Exp (β_{0} + \sum_{j \neq - 1} β_{1}^{j} 1 {E V e r i f y}_{s t} * 1 {E v e n t Y e a r}_{j (s, t)} \\ + β_{2} l o g (p o p_{a t}) + β_{3}' X_{c t} + β_{4}' Z_{s t} + α_{a} + τ_{t} + ε_{a t}) \end{array}

(2)

The event study in Equation (2) differs from our primary specification in how it expresses policy variation. The subscript j denotes the number of years before and after a state enacts an E-Verify Law (“event time”). Each β^j₁ describes the change in criminal incidents involving Hispanic arrestees in states that enacted E-Verify compared to those that did not. Specifically, it involves a differential change from year j relative to the event period j(s,t) = −1, one year prior to enactment. This approach is designed to descriptively explore whether pre-existing trends in Hispanic arrests drive the adoption of E-Verify laws.

Additionally, we experiment with including state-specific linear time trends, county-level linear time trends, and county-level quadratic time trends to account for time-varying spatial heterogeneity correlated with E-Verify and arrests. Note that the inclusion of geographic-specific trends can bias estimated treatment effects (Goodman-Bacon 2021).

Next, we turn to the UCR, for which we estimate Equation (1), but replace the left-side variable with White arrests or African American arrests, measured separately for property and violent crime arrests. An important advantage of the UCR over the NIBRS is that all 23 E-Verify states contribute to identification.¹³ A limitation, however, is that we cannot explicitly separate arrests of Hispanic whites versus non-Hispanic whites. Therefore, we proxy for arrests that are more likely to be of Hispanic whites using the share of the state population that comprises Hispanics:

\begin{matrix} White Arrest s_{at} = κ_{at} Exp (γ_{0} + γ_{1} EVerif y_{st} + γ_{2} EVerif y_{st} * PctHIS P_{s 2004} \\ + γ_{3} \log (po p_{at}) + ϒ_{4}' X_{ct} + ϒ_{5}' Z_{st} + α_{a} + τ_{t} + μ_{at}) \end{matrix}

(3)

where PctHISP_s2004 indicates the pre-E-Verify (year = 2004) distribution of state share of the white population in the state who are of Hispanic descent (focusing on quartiles of the ranked state-specific distribution).

Finally, our CPS-based analysis explores the mechanisms through which E-Verify may affect crime, including employment and state population demographics. We estimate ordinary least squares (OLS) regressions for low-skilled, working-age Hispanics with a high school degree or less. We then stratify that low-skilled Hispanic sample by citizenship status, focusing on non-citizen immigrants and US-born citizens of Hispanic descent.¹⁴ Our primary outcomes of interest measure employment and demographic composition of the respondent. First, we generate a variable indicating whether an individual is employed; then we generate an employment indicator for those employed for wage-and-salary pay (that is, not self-employed). Second, we examine the impact of E-Verify on the demographic composition of the state, with attention to Hispanic immigrants and natives.

Results

Our main findings appear in Tables 2 through 9. Our NIBRS-based tables focus on the estimate of $β_{1}$ . Coefficient estimates on control variables from our main specifications are shown in Table A.5. Standard errors are clustered at the state level (Bertrand, Duflo, and Mullainathan 2004).

Table 2.

Estimated Effect of E-Verify on Crime Involving Hispanic Arrestees, NIBRS 2004–2015

	(1)	(2)	(3)	(4)	(5)	(6)
Property crime	−0.118**(0.051)	−0.112**(0.048)	−0.123***(0.047)	−0.103***(0.039)	−0.080**(0.032)	−0.075**(0.036)
N	255,744	255,744	255,744	255,744	255,744	255,744
Violent crime	−0.043(0.063)	−0.023(0.053)	−0.022(0.053)	0.015(0.046)	0.021(0.042)	−0.007(0.049)
N	255,744	255,744	255,744	255,744	255,744	255,744
Agency fixed effects?	Yes	Yes	Yes	Yes	Yes	Yes
Year-by-Month fixed effects?	Yes	Yes	Yes	Yes	Yes	Yes
Demographic controls?	No	Yes	Yes	Yes	Yes	Yes
Political & economic controls?	No	No	Yes	Yes	Yes	Yes
Crime policy controls?	No	No	No	Yes	Yes	Yes
Immigration policy controls?	No	No	No	No	Yes	Yes
Social policy controls?	No	No	No	No	No	Yes

Notes: Poisson estimates are generated using agency-level data drawn from the 2004–2015 National Incident-Based Reporting System (NIBRS). Each regression has controls for agency fixed effects and year-by-month fixed effects. Demographic controls include the share of population ages 25 and older with a bachelor’s degree, the share of county population ages 25–54 and ages 55 and older, the share of population that are male, and the share of population that are African American. Political and economic controls include the natural log of per capita income, the natural log of unemployment rates, and an indicator if the state governor is a Democrat. Crime policy controls include the natural logs of police expenditure per capita and police employment per capita and indicators for shall issue laws, stand-your-ground laws, and background check laws. Immigration policy controls include indicators for 287(g) programs, Secure Communities, and omnibus immigration bills. Social policy controls include the natural logs of minimum wages, refundable Earned Income Tax Credit (EITC) rates, indicators for ban-the-box laws, Supplemental Nutrition Assistance Program (SNAP) vehicle exemptions, and Affordable Care Act (ACA) Medicaid expansion. Standard errors are clustered at the state level.

***

Significant at 1% level; ** at 5% level; * at 10% level.

E-Verify Laws and Criminal Incidents Involving Hispanics

In Table 2, row (1), we present estimates of $β_{1}$ from Equation (1) for property crime incidents involving Hispanic arrestees. Our most parsimonious specification, which includes agency fixed effects and month-by-year fixed effects (column (1)), shows that the enactment of an E-Verify mandate is associated with an 11.1% [1-exp (–0.118)] reduction in property crime incidents involving Hispanic arrestees. The inclusion of county-level demographic controls (column (2)), economic and political controls (column (3)), and crime policy controls (column (4)) has little effect on this estimate. In column (5), we add controls for other state immigration policies, and in column (6), we include a wide set of social welfare policy controls. The result from our most saturated specification (column (6)) suggests that E-Verify mandates are associated with a 7.2% reduction in property crime involving Hispanics. The stability of our estimated policy effects across specifications lends credence to the hypothesis that E-Verify mandates are implemented exogenously to other crime determinants.

In contrast to our property crime estimates, we find no evidence that E-Verify reduces violent criminal incidents among Hispanics (row (2), Table 2). The estimated effects are smaller in magnitude (0.021 to −0.043) and statistically insignificant.¹⁵

The results in Table 2 suggest that E-Verify mandates reduce economically motivated crimes, which is consistent with E-Verify-induced increases in employment among low-skilled US citizens of Hispanic descent (Bohn, Lofstrom, and Raphael 2015; Orrenius and Zavodny 2015; Orrenius et al. 2018). How plausible is a 7.2% decline in property crime? Orrenius and Zavodny (2015) found that E-Verify was associated with an 8 percentage point increase in wage-and-salary employment among Mexican-born naturalized citizens. Lin (2008) found that a 1 percentage point increase in unemployment was associated with a 2 to 4% increase in property crime. Thus, even after accounting for negative employment effects for unauthorized immigrants (Amuedo-Dorantes and Bansak 2014), a net crime decline of 7.2% among Hispanics is certainly plausible. Given the absence of violent crime declines, our results are unlikely to be entirely driven by out-migration of Hispanics.

Figure 1 plots our event study coefficients. In panel (a), we find no evidence that Hispanic property crime arrests were trending downward in states that later adopted E-Verify compared to states that did not. A reduction occurs approximately one year following enactment and grows over time. By contrast, we continue to find no evidence that E-Verify mandates affect violent crime incidents involving Hispanic arrestees (panel (b)).¹⁶

Figure 1.

Event Study Analysis of E-Verify Mandates and Criminal Incidents Involving Hispanic Arrestees, NIBRS 2004–2015

In Table 3, we explore the sensitivity of our estimates to including state- or county-specific time trends. Column (1) reproduces the estimates from column (6) of Table 2. The inclusion of state-specific linear time trends (column (2)), county-specific linear time trends (column (3)), and county-specific quadratic time trends (column (4)) do not change our main finding. Across these specifications, we find that E-Verify laws are associated with a 6.2 to 8.3% reduction in property crime incidents involving Hispanic arrestees.

Table 3.

Robustness of Hispanic Arrest Effects of E-Verify to Controls for State- and County-Level Time Trends

	(1)	(2)	(3)	(4)
Property crime	−0.075**(0.036)	−0.087**(0.041)	−0.061*(0.032)	−0.064**(0.032)
N	255,744	255,744	255,744	255,744
Violent crime	−0.007(0.049)	−0.031(0.034)	0.046(0.032)	0.021(0.034)
N	255,744	255,744	255,744	255,744
Agency fixed effects	Yes	Yes	Yes	Yes
Year-by-Month fixed effects	Yes	Yes	Yes	Yes
State-specific linear time trends	No	Yes	No	No
County-specific linear time trends	No	No	Yes	Yes
County-specific quadratic time trends	No	No	No	Yes

Notes: Poisson estimates are generated using agency-level data drawn from the 2004–2015 National Incident-Based Reporting System (NIBRS). Each regression has controls for agency fixed effects, year-by-month fixed effects, linear time trends, and controls listed in Table A.2. Standard errors are clustered at the state level.

***

Significant at 1% level; ** at 5% level; * at 10% level.

Heterogeneity in Effects of E-Verify Mandates on Criminal Incidents Involving Hispanics

Tables 4 and 5 explore heterogeneity in the effects of E-Verify mandates by age, gender, breadth of mandate, and offense type. With regard to age, we find that our estimated effects are concentrated among Hispanics ages 20 to 44, a demographic group for whom E-Verify mandates have been shown to have relatively larger labor market effects (Amuedo-Dorantes and Bansak 2014). By contrast, we find no evidence that E-Verify affected property crime incidents involving younger (16–19) or older (45–64) Hispanic arrestees (panel I, column (1)). We find that the property crime-reducing effects are concentrated among males (panel II), a population with substantially higher crime rates.

Table 4.

Heterogeneity in Hispanic Arrest Effects of E-Verify, NIBRS 2004–2015

	(1)	(2)
	Property crime	Violent crime
	Panel I: Age
Ages 16–19	0.034(0.035)	0.078(0.052)
Ages 20–24	−0.090**(0.038)	−0.038(0.058)
Ages 25–34	−0.166***(0.054)	−0.031(0.054)
Ages 35–44	−0.109**(0.055)	−0.054(0.050)
Ages 45–64	0.008(0.039)	0.061(0.090)
N	255,744	255,744
	Panel II: Gender
Men	−0.091**(0.039)	−0.010(0.048)
N	255,744	255,744
Women	−0.047(0.039)	0.045(0.062)
N	255,744	255,744
	Panel III: Type of E-Verify mandate
Public or partial E-Verify	−0.068**(0.034)	−0.004(0.048)
Universal E-Verify	−0.265*	−0.116
	[p value = 0.071]^a	[p value = 0.286]^a
N	255,744	255,744
	Panel IV: Controls for state-by-year share of population that are low-skilled immigrants and natives
E-Verify	−0.081**(0.036)	−0.008(0.052)
N	255,744	255,744

Because only two states identify Universal E-Verify, we generate our permutation-based p values from placebo tests on non-E-Verify states (Buchmueller et al. 2011; Cunningham and Shah 2018).

***

Significant at 1% level; ** at 5% level; * at 10% level.

Table 5.

Examination of Detailed Criminal Incidents Involving Hispanic Arrestees, NIBRS 2004–2015

	Panel I: Property crime
	Larceny	Burglary	Motor vehicle theft	Arson
E-Verify	−0.078**(0.037)	−0.017(0.044)	−0.166*(0.101)	−0.082(0.136)
N	255,744	255,744	255,744	255,744
	Panel II: Violent crime
	Aggravated assault	Murder	Rape	Robbery
E-Verify	0.012(0.050)	−0.090(0.122)	−0.097(0.073)	−0.077(0.074)
N	255,744	255,744	255,744	255,744
	Panel III: Other crime
	Drug	Stolen property	Weapon law violation	Sex offenses
E-Verify	0.048(0.041)	−0.288***(0.060)	0.008(0.055)	−0.020(0.046)
N	255,744	255,744	255,744	255,744

***

Significant at 1% level; ** at 5% level; * at 10% level.

In panel III of Table 4, we explore heterogeneity in the impacts of E-Verify by mandate type: those that apply only to public employers or allow employers an alternate means of compliance with E-Verify (Public or partial E-Verify), and those that are mandatory and apply to both private and public employers (Universal E-Verify). Our findings suggest substantially larger declines in property crime incidents involving Hispanic arrestees following the passage of mandatory, universal E-Verify mandates (23.3%). Because this estimated treatment effect is identified from only two NIBRS states, South Carolina and Utah, we conduct placebo tests on untreated states to generate a permutation-based p value (Buchmueller, DiNardo, and Valletta 2011; Cunningham and Shah 2018) of 0.07. An event study analysis of universal, mandatory E-Verify laws (panel (c) of Figure 1) suggests that the Hispanic property crime decline takes approximately two years to unfold. We fail to detect evidence that universal mandates affect violent crime involving Hispanic arrestees (panel (d) of Figure 1).

In panel IV, we explore whether the estimated crime effects we obtain can be explained by endogenous demographic composition changes of low-skilled immigrants and natives of Hispanic descent. The estimated effect of E-Verify on crime is largely unchanged with the inclusion of controls for the share of the state population that comprises working-age (ages 16–64), less-educated (attained a high school degree or less) Hispanic non-citizen immigrants and working-age, less-educated Hispanic natives. This result suggests that demographic composition changes to E-Verify mandates cannot fully explain their property crime-reducing effects.

In Table 5, we explore the property crime offenses that drove this decline (panel I), as well as examine whether particular violent crimes (panel II) or other criminal incidents (panel III) were affected by E-Verify. Our results show that the decline in property crime is largely driven by larcenies, which account for more than 80% of all property crimes. A substantial negative effect of E-Verify on motor vehicle theft is also seen, though the relationship is less precisely estimated.

We generally fail to detect statistically significant relationships between E-Verify mandates and violent offenses or other non-violent offenses. The only exception is for stolen property, for which we find that E-Verify is associated with a substantial decline, a finding consistent with economically motivated criminal behavior.

Finally, in Table 6, we examine whether E-Verify displaces Hispanic crime to other jurisdictions in close geographic proximity to an E-Verify mandate. We generate two measures: Border-state E-Verify, which turns on when a border state adopts an E-Verify mandate, and Census-division E-Verify, which turns on when a state within the state’s own census division enacts E-Verify. In odd-numbered columns of Table 6, we restrict the sample to jurisdictions that have never implemented E-Verify and estimate Equation (1), replacing E-Verify with Border-state E-Verify (panel I) and Census-division E-Verify (panel II). In even-numbered columns, we pool all available jurisdictions and add Border-state E-Verify (panel I) or Census-division E-Verify (panel II) to the right side of Equation (1). In no case do we uncover evidence that property or violent crime was displaced to neighboring NIBRS jurisdictions.

Table 6.

Exploring Hispanic Crime Displacement in Jurisdictions Neighboring E-Verify States, NIBRS 2004–2015

	(1)	(2)	(3)	(4)
	Property crime		Violent crime
	Panel I: Spillover to border state
Border-state E-Verify	−0.023(0.083)	−0.035(0.036)	0.015(0.076)	−0.016(0.038)
E-Verify		−0.078**(0.036)		−0.007(0.048)
N	105,840	255,744	105,840	255,744
	Panel II: Spillover within Census division
Census-division E-Verify	0.019(0.078)	0.008(0.041)	−0.107(0.109)	−0.032(0.054)
E-Verify		−0.076**(0.035)		−0.004(0.047)
N	105,840	255,744	105,840	255,744
Sample	Non-E-Verify	Pooled	Non-E-Verify	Pooled

***

Significant at 1% level; ** at 5% level; * at 10% level.

Mechanisms to Explain Decline in Hispanic Property Crime

Table 7 explores the mechanisms through which E-Verify affects crime involving Hispanic arrestees using data from the CPS-BMS. We present results for all working-age individuals and then individuals ages 20 to 44, the age group for which we find the strongest evidence of crime reductions.

Table 7.

Exploring Employment and Demographic Composition Mechanisms, CPS-BMS 2004–2015

	(1)	(2)	(3)	(4)
	Hispanic immigrants		US-born Hispanics
Ages	Men	Women	Men	Women
	Panel I: Any employment
16–64	−0.007(0.006)	0.003(0.009)	0.010(0.009)	−0.002(0.012)
N	249,080	214,237	225,720	223,643
20–44	−0.004(0.008)	−0.003(0.012)	−0.004(0.014)	−0.007(0.010)
N	177,758	145,939	115,217	109,479
	Panel II: Wage-and-salary employment
16–64	−0.010(0.010)	−0.006(0.011)	0.022*(0.013)	0.001(0.013)
N	249,080	214,237	225,720	223,643
20–44	−0.005(0.011)	−0.009(0.014)	0.012(0.020)	−0.004(0.012)
N	177,758	145,939	115,217	109,479
	Panel III: Demographic composition
16–64	−0.0010(0.0008)	−0.0011**(0.0004)	−0.0019**(0.0009)	−0.0015*(0.0008)
N	12,507,441	12,507,441	12,507,441	12,507,441
20–44	−0.0009(0.0005)	−0.0008**(0.0003)	−0.0012***(0.0004)	−0.0007*(0.0004)
N	12,507,441	12,507,441	12,507,441	12,507,441

Notes: Weighted ordinary least squares (OLS) estimates are generated using individual-level data drawn from the 2004–2015 Current Population Survey Basic Monthly Survey(CPS-BMS). Each regression has controls for agency fixed effects, time (year and month) fixed effects, state-specific linear time trends, and controls listed in Table A.2. Standard errors are clustered at the state level.

***

Significant at 1% level; ** at 5% level; * at 10% level.

First, generally consistent with prior work (Amuedo-Dorantes and Bansak 2014), we find that the implementation of an E-Verify mandate is associated with a (statistically insignificant) decline in any employment among likely unauthorized male immigrants, on the order of 0.5 to 1.0 percentage points (column (1), panel I).¹⁷ However, we also find that E-Verify is associated with a 2.2 percentage point increase in wage-and-salary employment among low-skilled, US-born Hispanics (column (3), panel II).¹⁸ Given that most prosecuted Hispanic defendants are US citizens (Lott 2018; Landgrave and Nowrasteh 2019), this positive employment effect—in conjunction with the fact that immigrants (particularly recent immigrants) are less likely to be criminally prone than are natives (Butcher and Piehl 2008a; Chalfin 2013)—is likely an important channel to explain the net decline in Hispanic property crime.

In panel III of Table 7, we examine a sample of all adults and define the left-side variable as an indicator set equal to 1 if the respondent is a member of the demographic group listed in the column heading, and 0 otherwise. Thus, the estimated policy impact can be interpreted as the effect of E-Verify on the demographic composition of the state. We find that the enactment of an E-Verify law is associated with a 5.3 (–0.0009/0.017) to 5.7% (–0.0008/0.014) decline in the state population share of 20- to 44-year-olds who are less-educated Hispanic immigrants. We also find that E-Verify is associated with a 6.4 (–0.0007/0.011) to 10.9% (–0.0012/0.011) decline in the share of the 20- to 44-year-old state population who were US-born Hispanics.¹⁹ This finding could suggest that mixed-status Hispanic families composed of both likely unauthorized immigrants and natives out-migrate in response to E-Verify enactment. Out-migration is unlikely to be the only mechanism explaining the property crime reductions given that 1) we do not observe a similar decline in violent crime, 2) our property crime results persist after controlling for state shares of low-skilled Hispanic immigrants and natives, and 3) we fail to detect crime spillovers to neighboring jurisdictions without E-Verify laws. However, it is worth noting that we cannot measure international out-migration. If E-Verify mandates induce unauthorized immigrants to return to their native countries, it is possible that we do not detect crime spillovers because they occur in neighboring nations.

Figure 2.

Event Study Analysis of E-Verify Mandates and Hispanic Demographic Composition for 20- to 44-Year-Olds, CPS-BMS 2004–2015

In Table 8, we examine the effect of Universal E-Verify mandates and Public or partial E-Verify mandates on labor market outcomes. While our findings are mixed, the results in panel II do suggest that the magnitude of the effect of E-Verify mandates on wage-and-salary employment for low-skilled male citizens of Hispanic descent is greater for E-Verify mandates that extend to private employers relative to only public employers. Event-study estimates in Figure 3 indicate positive employment effects of Universal E-Verify for low-skilled Hispanic natives. This finding is consistent with larger property crime effects for more expansive E-Verify laws.²⁰

Table 8.

Exploring Heterogeneity in Employment and Demographic Composition Effects, by Breadth of E-Verify Mandate, CPS-BMS 2004–2015

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Hispanic immigrants				US-born Hispanics
	Men		Women		Men		Women
	16–64	20–44	16–64	20–44	16–64	20–44	16–64	20–44
	Panel I: Any employment
Public E-Verify	−0.010*(0.006)	−0.007(0.007)	−0.003(0.009)	−0.006(0.012)	0.008(0.009)	−0.004(0.014)	−0.008(0.014)	−0.012(0.010)
Universal E-Verify	0.006(0.012)	0.022(0.012)	0.000(0.016)	−0.025(0.020)	0.011(0.022)	−0.002(0.028)	0.036**(0.018)	0.037(0.026)
N	249,080	177,758	214,237	145,939	225,720	115,217	223,643	109,479
	Panel II: Wage-and-salary employment
Public E-Verify	−0.013(0.009)	−0.007(0.010)	−0.015(0.011)	−0.016(0.014)	0.018(0.013)	0.008(0.021)	−0.003(0.014)	−0.008(0.011)
Universal E-Verify	0.004(0.015)	0.011(0.016)	−0.003(0.022)	−0.030(0.027)	0.030(0.023)	0.037(0.034)	0.035(0.021)	0.041(0.031)
N	249,080	177,758	214,237	145,939	225,720	115,217	223,643	109,479
	Panel III: Demographic composition
Public E-Verify	−0.0014(0.0009)	−0.0011*(0.0006)	−0.0009**(0.0004)	−0.0007**(0.0003)	−0.0022**(0.0009)	−0.0013***(0.0004)	−0.0017**(0.0008)	−0.0008*(0.0004)
Universal E-Verify	−0.0010(0.0025)	−0.0005(0.0016)	−0.0002(0.0013)	−0.0011(0.0008)	−0.0014(0.0009)	−0.0006(0.0007)	−0.0012(0.0007)	0.0002(0.0004)
N	12,507,441	12,507,441	12,507,441	12,507,441	12,507,441	12,507,441	12,507,441	12,507,441

Notes: Weighted ordinary least squares (OLS) estimates are generated using individual-level data drawn from the 2004–2015 Current Population Survey Basic Monthly Survey (CPS-BMS). Each regression has controls for agency fixed effects, time (year and month) fixed effects, state-specific linear time trends, and controls listed in Table A.2. Standard errors are clustered at the state level.

***

Significant at 1% level; ** at 5% level; * at 10% level.

Figure 3.

Event Study Analysis of Universal E-Verify Mandates and Wage-and-Salary Employment for Low-Skilled US-Born Hispanics

External Validity

Finally, given that our NIBRS results may not be nationally representative and thus not generalizable, we turn to the UCR. Our results in Table 9 show no evidence E-Verify mandates affected arrests among all adult whites or African Americans (panel I). In columns (1) and (2) of panel II, we interact indicators for the state-level proportion of white adults who are of Hispanic descent with the E-Verify law. The estimated coefficient on the E-Verify law can be interpreted as the effect for states with pre-treatment Hispanic populations in the bottom half of the treatment state Hispanic population distribution, while the remaining coefficients show the differential impact of E-Verify laws on white arrests for states in the 50th to 75th percentile and the 75th to 100th percentile of the Hispanic white population distribution (relative to states with pre-treatment Hispanic populations in the lowest quartile). Consistent with results from the NIBRS, we find that E-Verify enactment is associated with an 8.4% reduction in white property crime arrests in states with the highest shares (> 75th percentile) of Hispanic whites. An event study analysis in Figure 4, panel (a), shows that this decline in white property crime arrests in higher Hispanic-populated states occurred approximately one year following the adoption of E-Verify. We fail to detect any evidence of E-Verify-induced changes in white property crime arrests in states with lower shares of Hispanics (panel (c) of Figure 4).²¹ We find no evidence that E-Verify affected white violent crime arrests in states with larger shares of Hispanic whites (panel II, column (2)), nor do we find evidence that African American arrests (panel II, columns (3) and (4)) responded to E-Verify laws. Finally, in Table 9, panel III, we find that E-Verify effects are somewhat larger for mandatory, universal E-Verify laws, but are not statistically distinguishable from zero at conventional levels.²²

Table 9.

Estimated Effect of E-Verify on African American and White Arrests, UCR 2004–2015

	White		African American
	(1)	(2)	(3)	(4)
	Property	Violent	Property	Violent
	Panel I: Baseline results
E-Verify	0.001(0.019)	−0.008(0.026)	0.012(0.027)	−0.027(0.030)
	Panel II: Effect by state share of Hispanic white
E-Verify	0.027(0.024)	−0.000(0.039)	0.012(0.040)	−0.023(0.045)
50th–75th Percentile × E-Verify	−0.001(0.039)	−0.078(0.074)	0.010(0.062)	−0.059(0.073)
75th–100th Percentile × E-Verify	−0.088**(0.034)	0.034(0.049)	−0.018(0.054)	0.074(0.080)
	Panel III: Effect by type of E-Verify mandate
Public E-Verify	0.003(0.019)	−0.004(0.026)	0.012(0.027)	−0.028(0.029)
Universal E-Verify	−0.034(0.047)	−0.065(0.068)	−0.033(0.049)	−0.075(0.058)
N	1,076,699	1,076,699	1,076,699	1,076,699

Notes: Poisson estimates are generated using agency-level data drawn from the 2004–2015 Uniform Crime Reports (UCR). Each regression has controls for agency fixed effects, year-by-month fixed effects, and controls listed in Table A.2. Standard errors are clustered at the state level.

***

Significant at 1% level; ** at 5% level; * at 10% level.

Figure 4.

Event Study Analysis of E-Verify Mandates and Arrests Involving White or African American Arrestees, UCR 2004–2015

Conclusions

On January 4, 2019, 17 Republican congressmen re-introduced the Legal Workforce Act of 2019. This legislation would mandate the use of an E-Verify system across all 50 states for public and private employers. Although comprehensive immigration reform has remained one of the most contentious issues in American politics, support is growing for a federal E-Verify law. In June 2019, Senator Mitt Romney (R-Utah) introduced the Permanent E-Verify Act, which would abolish annual Congressional renewal of federal E-Verify requirements, thereby making the federal law permanent. Mounting evidence suggests that E-Verify redistributes employment from likely unauthorized immigrants to low-skilled US citizens of Hispanic descent. This study comprehensively examines the impact of state E-Verify mandates on criminal arrests involving working-age Hispanics, as well as spillovers to African Americans.

Using data drawn from the National Incident-Based Reporting System, we find that E-Verify laws are associated with a 7.2% decrease in property crimes that involve working-age Hispanic arrestees. The effects are largest for males, for those who are age 20 to 44, and for E-Verify policies that extend to private employers. Supplemental analyses from the Current Population Survey suggest that increases in employment among low-skilled citizens of Hispanic descent and, perhaps, out-migration of younger immigrants, likely explain the net reduction in Hispanic crime. We find no evidence of property crime displacement in jurisdictions without an E-Verify mandate and no evidence that E-Verify affects violent crime among Hispanics. Consistent with our NIBRS-based results, analyses of the UCR show that E-Verify reduces property crime arrests among whites in states with relatively higher shares of Hispanic whites.

Using the per-offense social cost of property crime reported in McCollister, French, and Fang (2010), our estimates suggest that E-Verify generated $491 million (2018 dollars) in benefits from property crime reduction to the United States.²³ Of course, this estimate may overstate the benefit of crime avoidance to the western hemisphere if crime is displaced to border countries in North and Central America. Moreover, other important costs of E-Verify laws—such as increased compliance costs on firms, higher prices of consumer products, poorer quality job matches, and adverse health and human capital effects on immigrant families (including many children who are US citizens)—must be weighed against the benefits of crime reduction.

Supplemental Material

sj-pdf-1-ilr-10.1177_00197939211044449 – Supplemental material for The Effect of E-Verify Laws on Crime

Supplemental material, sj-pdf-1-ilr-10.1177_00197939211044449 for The Effect of E-Verify Laws on Crime by Brandyn F. Churchill, Andrew Dickinson, Taylor Mackay and Joseph J. Sabia in ILR Review

Footnotes

Acknowledgements

Churchill is grateful for support from Vanderbilt University’s Kirk Dornbush summer research grant. Sabia acknowledges support from CHEPS at San Diego State University, including grant funding received from the Troesh Family Foundation and the Charles Koch Foundation. The authors thank Kyutaro Matsuzawa and Tam Nguyen for outstanding research assistance and Isaac Baumann for excellent editorial assistance.

ORCID iD

Joseph J. Sabia

An Online Appendix is available at .

1

Tennessee and Louisiana adopted E-Verify laws that apply to both private and public employers but gave employers an alternative means of verifying employment outside of the E-Verify system (NCSL 2015). The federal E-Verify mandate applies to public employees and government contractors.

2

If Hispanic women increase informal labor market work in response to formal-sector job loss for Hispanic men (), female crime may fall.

3

ICE touts that program as a “force multiplier in the identification, arrest, and service of warrants and detainers of incarcerated foreign-born individuals.” See .

4

Other studies have more broadly explored the relationship between immigration and crime; see, for example, Butcher and Piehl (1998a,b), Amuedo-Dorantes, Banzak, and Pozo (2021), and .

5

They suggested that E-Verify could have incited violence against unauthorized immigrants. One concern about this study is that in 2011, adopted a new electronic record system, which may have improved arrest reporting (ACJIC 2011).

6

Landgrave and Nowrasteh (2019) estimated incarceration rates per 100,000 population, ages 18–54 in 2017, are 1,792 for Hispanic natives, 507 for legal Hispanic immigrants, and 1,097 for unauthorized Hispanic immigrants. In addition, found that 72% of all Hispanic incarcerations in Arizona between 1985 and 2017 involved US citizens of Hispanic descent.

7

In 2014, 86% of all NIBRS reporting agencies served local populations of less than 50,000, and 42% served local populations of less than 10,000.

8

While intercensal estimates of age- and race/ethnicity-specific population counts are readily available at the state, county, and large metropolitan area levels, this is not always the case for smaller cities and townships. Because NIBRS coverage is often not complete within counties or states, it is not appropriate to aggregate agency-level arrest counts to the county or state levels and generate age/race/ethnicity-specific arrest rates. Given these challenges, NIBRS-based studies in the economics of crime literature have generally used arrest counts (controlling for overall agency population, which we do) rather than arrest rates as the outcome of interest.

9

Approximately one-fifth of all criminal incidents involving a 16- to 64-year-old arrestee had an unknown, unreported, or missing ethnicity. Less than 1% of incidents involved an arrestee of unknown race. We found little evidence of a relationship between E-Verify and reporting of arrestee ethnicity (or race).

10

Using the 2004–2015 NIBRS, we find that less than 5% of all white arrestees are of Hispanic descent.

11

Using agency-level Hispanic population produces nearly identical estimates.

12

Ban-the-box laws prohibit employers from asking employees about their criminal histories at initial job screening and have been found to increase statistical discrimination against African American and Hispanic young adults (Doleac and Hansen 2017, 2020). Recent work by suggested that ban-the-box laws increase crime among Hispanic males because of diminished labor market opportunities.

13

We also experiment with restricting our UCR analysis to the treatment and control states from the NIBRS as well as using the NIBRS data source itself. The pattern of results we uncover suggests that any differences in estimates obtained from the UCR and NIBRS are not attributable to sample selection.

14

We include state-specific linear time trends as controls in the spirit of the specification estimated by Orrenius and Zavodny (2015). In addition, in results reported in , we examine the sensitivity of findings to the use of the Outgoing Rotation Groups (ORG) instead of the Basic Monthly Survey (BMS).

15

We also calculated p values using a wild cluster bootstrap standard error approach (Cameron, Gelbach, and Miller 2008; ), which did not qualitatively change our policy conclusions (column (6), p value = 0.06 for property crime and 0.79 for violent crime).

16

Table A.7 shows coefficient estimates on lead and lagged effects of E-Verify laws on criminal incidents involving a Hispanic arrestee. In , we explore whether our findings are driven by any particular state’s E-Verify law. Across samples, our results provide consistent evidence of a 5.1 to 8.7% decline in property crime incidents involving working-age Hispanic arrestees. Estimated effects on violent crime are much smaller and statistically insignificant in all cases.

17

The findings in , which use the ORG as compared to the BMS, show larger declines in employment for likely unauthorized Hispanic immigrants.

18

In , we show the estimates for naturalized citizens of Hispanic descent and all Hispanic US citizens. The results for naturalized Hispanics are more similar to non-citizen immigrants than to US-born Hispanics.

19

The results of event study analyses, shown in Figure 2, suggest little difference in pre-treatment trends in the respective demographic shares and evidence that the decline in the share of the state population made up of low-skilled Hispanics follows the enactment of E-Verify. Moreover, event study analysis from the ORG (shown in ) displays a similar pattern of results.

20

Results from the ORG (shown in Table A.10 and ) are qualitatively similar.

21

Panels (e) and (f) of show event study analyses for property and violent crime arrests involving African American arrestees.

22

We find little to no evidence that E-Verify mandates have an impact on labor force participation among African Americans and non-Hispanic whites, consistent with null crime effects reported above. We do uncover some inconsistent evidence that E-Verify may attract low-skilled, non-Hispanic whites to E-Verify states, consistent with attraction to low-skilled job opportunities.

23

Data on property crimes committed from 2004–2015 are obtained using the FBI’s Crime in the United States reports (https://ucr.fbi.gov/crime-in-the-u.s/2016/crime-in-the-u.s.-2016/topic-pages/tables/table-1). We then use the 2004–2015 UCR’s Arrests by Age, Sex, and Race files to calculate the share of property crime arrests involving men ages 16 to 64. To generate an estimate of the number of crimes committed by men ages 16 to 64, we multiply the crime counts in the 2004–2014 period from the FBI’s Crime in the United States report with the share of property crime arrests involving men ages 16 to 64 from the UCR’s Arrests by Age, Sex, and Race files. Next, we estimate the number of crimes committed by Hispanic men ages 16 to 64 by multiplying the above crime estimate with the percentage of arrests involving Hispanic male adults (https://ucr.fbi.gov/crime-in-the-u.s/2014/crime-in-the-u.s.-2014/tables/table-43). Using our estimate from , column (6), we estimate 85,555 fewer property crimes following the enactment of E-Verify mandates. Finally, we used McCollister et al.’s (2010) $5,739 per crime cost of a property offense to obtain a total E-Verify-induced property crime benefit of $491 million.

References

[ACJIC] Alabama Criminal Justice Information Center. 2011. Crime in Alabama 2011. Accessed at https://www.alea.gov/sites/default/files/inline-files/CrimeInAlabama-2011.pdf (August 1, 2020).

Amuedo-Dorantes

Catalina

Bansak

Cynthia

. 2012. The labor market impact of mandated employment verification systems. American Economic Review 102(3): 543–48.

Amuedo-Dorantes

Catalina

Bansak

Cynthia

. 2014. Employment verification mandates and the labor market outcomes of likely unauthorized and native workers. Contemporary Economic Policy 32(3): 671–80.

Amuedo-Dorantes

Catalina

Banzak

Cynthia

Pozo

Susan

. 2021. Refugee admissions and public safety: Are refugee settlement areas more prone to crime? International Migration Review 55(1): 135–65.

Anderson

D. Mark

. 2014. In school and out of trouble? The minimum dropout age and juvenile crime. Review of Economics and Statistics 96(2): 318–31.

Baker

Scott R.

2015. Effects of immigrant legalization on crime. American Economic Review 105(5): 210–13.

Becker

Gary S.

1968. Crime and punishment: An economic approach. Journal of Political Economy 76(2): 169–217.

Bertrand

Marianne

Duflo

Esther

Mullainathan

Sendhil

. 2004. How much should we trust differences-in-differences estimates? Quarterly Journal of Economics 119(1): 249–75.

Bohn

Sarah

Lofstrom

Magnus

. 2012. Employment effects of state legislation against the hiring of unauthorized immigrant workers. IZA Discussion Paper No. 6598. Bonn, Germany: Institute of Labor Economics.

10.

Bohn

Sarah

Lofstrom

Magnus

Raphael

Steven

. 2014. Did the 2007 Legal Arizona Workers Act reduce the state’s unauthorized immigrant population? Review of Economics and Statistics 96(2): 258–69.

11.

Bohn

Sarah

Lofstrom

Magnus

Raphael

Steven

. 2015. Do E-Verify mandates improve labor market outcomes of low-skilled native and legal immigrant workers? Southern Economic Journal 81(4): 960–79.

12.

Buchmueller

Thomas C.

DiNardo

John

Valletta

Robert G.

2011. The effect of an employer health insurance mandate on health insurance coverage and the demand for labor: Evidence from Hawaii.American Economic Journal: Economic Policy3(4): 25–51.

13.

Butcher

Kristin F.

Piehl

Anne Morrison

. 1998a. Recent immigrants: Unexpected implications for crime and incarceration. Industrial and Labor Relations Review 51(4): 654–79.

14.

Butcher

Kristin F.

Piehl

Anne Morrison

. 1998b. Cross-city evidence on the relationship between immigration and crime. Journal of Policy Analysis and Management 17(3): 457–93.

15.

Butcher

Kristin F.

Piehl

Anne Morrison

. 2008. Crime, corrections, and California. California Counts 9(3): 1–23.

16.

Cáceres-Delpiano

Julio

Giolito

Eugenio

. 2012. The impact of unilateral divorce on crime. Journal of Labor Economics 30(1): 215–48.

17.

Cameron

A. Colin

Gelbach

Jonah B.

Miller

Douglas L.

2008. Bootstrap-based improvements for inference with clustered errors. Review of Economics and Statistics 90(3): 414–27.

18.

Cameron

A. Colin

Miller

Douglas L.

2015. A practitioner’s guide to cluster-robust inference. Journal of Human Resources 50(2): 317–72.

19.

Card

David

Dahl

Gordon B.

2011. Family violence and football: The effect of unexpected emotional cues on violent behavior. Quarterly Journal of Economics 126(1): 103–43.

20.

Castillo

Monica

Schulman

Janie

. 2011. Ready or not, here they come: State E-Verify laws and what employers should know. Morrison and Foerster Employment Law Commentary 23(8): 1–8.

21.

Chalfin

Aaron

. 2013. What is the contribution of Mexican immigration to US crime rates? Evidence from rainfall shocks in Mexico. American Law and Economics Review 16(1): 220–68.

22.

Chalfin

Aaron

Deza

Monica

. Forthcoming. New evidence on Mexican immigration and crime in the United States: Evidence from a natural experiment in immigration enforcement. Criminology & Public Policy.

23.

Cunningham

Scott

Shah

Manisha

. 2018. Decriminalizing indoor prostitution: Implications for sexual violence and public health. Review of Economic Studies 85(3): 1683–715.

24.

[DHS] Department of Homeland Security. 2018. What is E-Verify. Accessed at https://www.e-verify.gov/about-e-verify/what-is-e-verify (October 1, 2019).

25.

Doleac

Jennifer L.

Hansen

Benjamin

. 2017. Moving to job opportunities? The effect of “ban the box” on the composition of cities. American Economic Review: Papers & Proceedings 107(5): 556–59.

26.

Doleac

Jennifer L.

Hansen

Benjamin

. 2020. The unintended consequences of “ban the box”: Statistical discrimination and employment outcomes when criminal histories are hidden. Journal of Labor Economics 38(2): 321–74.

27.

Donohue

John J.

III Aneja

Abhay

Weber

Kyle D.

2019. Right-to-carry laws and violent crime: A comprehensive assessment using panel data and a state-level synthetic control analysis. Journal of Empirical Legal Studies 16(2): 198–247.

28.

East

Chloe N.

Luck

Philip

Mansour

Hani

Velasquez

Andrea

. 2018. The labor market effects of immigration enforcement. IZA Working Paper No. 11486. Bonn, Germany: Institute of Labor Economics.

29.

[FBI] Federal Bureau of Investigation National Press Office. 2015. FBI releases 2014 crime statistics from the National Incident-Based Reporting System. Accessed at https://www.fbi.gov/news/pressrel/press-releases/fbi-releases-2014-crime-statistics-from-the-national-incident-based-reporting-system (October 1, 2019).

30.

Freedman

Matthew

Owens

Emily

Bohn

Sarah

. 2018. Immigration, employment opportunities, and criminal behavior. American Economic Journal: Economic Policy 10(2): 117–51.

31.

Good

Michael

. 2013. Do immigrant outflows lead to native inflows? An empirical analysis of the migratory responses to U.S. state immigration legislation. Applied Economics 45(30): 4275–97.

32.

Goodman-Bacon

Andrew

. 2021. Difference-in-differences with variation in treatment timing. Journal of Econometrics. https://doi.org/10.1016/j.jeconom.2021.03.014

33.

Gould

Eric D.

Weinberg

Bruce A.

Mustard

David B.

2002. Crime rates and local labor market opportunities in the United States: 1979–1997. Review of Economics and Statistics 84(1): 45–61.

34.

Hanson

Gordon H.

2006. Illegal migration from Mexico to the United States. Journal of Economic Literature 44(4): 869–924.

35.

Heaton

Paul

. 2012. Sunday liquor laws and crime. Journal of Public Economics 96(1): 42–52.

36.

Kerwin

Donald M.

McCabe

Kristen

. 2011. Labor Standards Enforcement and Low-Wage Immigrants: Creating an Effective Enforcement System. Washington, DC: Migration Policy Institute.

37.

Kubrin

Charis E.

2014. Secure or insecure communities? Seven reasons to abandon the Secure Communities program. Criminology and Public Policy 13(2): 323–38.

38.

Landgrave

Michelangelo

Nowrasteh

Alex

. 2019. Criminal immigrants in 2017: Their numbers, demographics, and countries of origin. Cato Institute. Accessed at https://www.cato.org/publications/immigration-research-policy-brief/criminal-immigrants-2017-their-numbers-demographics (October 1, 2019).

39.

Lee

Jongkwan

Peri

Giovanni

Yasenov

Vasil

. 2017. The employment effects of Mexican repatriations: Evidence from the 1930s. NBER Working Paper No. 23885. Cambridge, MA: National Bureau of Economic Research.

40.

Levitt

Steven D.

2004. Understanding why crime fell in the 1990s: Four factors that explain the decline and six that do not. Journal of Economic Perspectives 18(1): 163–90.

41.

Lin

Ming-Jen

. 2008. Does unemployment increase crime? Evidence from US data 1974–2000. Journal of Human Resources 43(2): 413–36.

42.

Lindo

Jason M.

Siminski

Peter

Swensen

Isaac

. 2018. College party culture and sexual assault. American Economic Journal: Applied Economics 10(1): 236–65.

43.

Lott

John R.

2018. Undocumented immigrants, U.S. citizens, and convicted criminals in Arizona. Accessed at https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.3099992 (October 1, 2019).

44.

McCollister

Kathryn E.

French

Michael T.

Fang

Hai

. 2010. The cost of crime to society: New crime-specific estimates for policy and program evaluation. Drug and Alcohol Dependence 108(1-2): 98–109.

45.

Michaud

Nicholas D.

2010. From 287(g) to SB 1070: The decline of the federal immigration partnership and the rise of state-level immigration enforcement. Arizona Law Review 52: 1082–133.

46.

Miles

Thomas J.

Cox

Adam B.

2014. Does immigration enforcement reduce crime? Evidence from Secure Communities. Journal of Law and Economics 57(4): 937–73.

47.

[NCSL] National Conference of State Legislature. 2015. State E-Verify action. Accessed at http://www.ncsl.org/research/immigration/state-e-verify-action.aspx (October 1, 2019).

48.

National Immigration Law Center. 2011. The history of E-Verify. Accessed at https://www.nilc.org/wp-content/uploads/2015/12/e-verify-history-rev-2011-09-29.pdf (October 1, 2019).

49.

Orrenius

Pia M.

Zavodny

Madeline

. 2015. The impact of E-Verify mandates on labor market outcomes. Southern Economic Journal 81(4): 947–59.

50.

Orrenius

Pia M.

Zavodny

Madeline

. 2016. Do state work eligibility verification laws reduce unauthorized immigration? IZA Journal of Migration 5:5. https://doi.org/10.1186/s40176-016-0053-3

51.

Orrenius

Pia M.

Zavodny

Madeline

. 2017. Creating cohesive, coherent immigration policy. Journal on Migration and Human Security 5(1): 180–93.

52.

Orrenius

Pia M.

Zavodny

Madeline

Gutierrez

Emily

. 2018. Do state employment eligibility verification laws affect job turnover? Contemporary Economic Policy 36(2): 394–409.

53.

Öster

Anna

Agell

Jonas

. 2007. Crime and unemployment in turbulent times. Journal of the European Economic Association 5(4): 752–75.

54.

Park

Elena

Friedman

Debra S.

2008. E-Verify for federal contractors and subcontractors required startingJanuary 15, 2009. Labor and employment alert: News concerning recent labor and employment issues (Cozen O’Connor). Accessed at http://www.cozen.com/admin/files/publications/Labor111708.pdf (October 1, 2019).

55.

Raphael

Steven

Winter-Ebmer

Rudolf

. 2001. Identifying the effect of unemployment on crime. Journal of Law and Economics 44(1): 259–83.

56.

Rosenblum

Marc R.

2011. E-Verify: Strengths, weaknesses, and proposals for reform. Washington, DC: Migration Policy Institute.

57.

Rosenfeld

Richard

Fornango

Robert

. 2007. The impact of economic conditions on robbery and property crime: The role of consumer sentiment. Criminology 45(4): 735–69.

58.

Sabia

Joseph J.

Mackay

Taylor

Nguyen

Thanh Tam

Dave

Dhaval M.

2021. The unintended crime effects of ban the box laws. Journal of Law and Economics. Forthcoming.

59.

Schnepel

Kevin

. 2018. Good jobs and recidivism. Economic Journal 128(608): 447–69.

60.

Treyger

Elina

Chalfin

Aaron

Loeffler

Charles

. 2014. Immigration enforcement, policing, and crime: Evidence from the Secure Communities program. Criminology and Public Policy 13(2): 285–322.

61.

Urban Institute. 2017. State immigration policy resource. Accessed at https://www.urban.org/features/state-immigration-policy-resource (January 2019).

62.

Zhang

Yinjunjie

Palma

Marco A.

Zhicheng Phil

. 2016. Unintended effects of the Alabama HB 56 immigration law on crime: A preliminary analysis. Economics Letters 147(October): 68–71.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

2.00 MB