Not all 287(g) agreements are created equal, and this heterogeneity is evident beyond the classification of the type of agreement. The MOAs in each agreement may also vary significantly. Some allow local officers trained through the program to
interrogate any individual that they encounter, while others restrict an officerβs ability to interrogate immigration status. Some 287(g) agreements also allow ICE to detain
individuals and eventually assume custody of them. The bounds of each 287(g)
agreement are important for determining what effects these agreements will have, both on the ultimate effects of the program and also the labor market effects of the agreement beyond the program. Strong levels of enforcement will magnify the effects of 287(g) agreements on immigrant worker wages, as they will increase the probability of interactions with ICE and can encourage workers to accept a wage below their true productivity. These effects may be particularly acute for undocumented workers, who have a higher fear of deportation (Arbona et al. 2010). If a 287(g) agreement is not strong, or is rarely enforced, it may not have the same effects as agreements that are
enforced. However, if it is the presence of the agreement itself, and not enforcement, that is causing effects on the immigrant population, the strength of the agreement will not matter.
To assess the strength of the effects of each 287(g) agreement, I use a variable within the TRAC data that indicates whether ICE actually assumed custody of the individual. This variable is important because it tracks whether the local jurisdiction complied with ICEβs detainer request and identifies whether the program has a strong level of enforcement. For each year in my sample, I compute the percentage of
individuals that were taken into custody by ICE after a detainer request was issued. The mean of this proportion is 0.58 for my sample, so for any jurisdiction that has a
proportion equal to or above 0.58, I characterize them as having a strong agreement, because they have a stronger effect by allowing ICE to take a greater proportion of individuals into custody after a detainer is issued. For any jurisdiction with a proportion below the mean value, I characterize them as having a low-strength agreement.
For the vast majority of jurisdictions, the number of individuals taken into ICE custody increased over time after the jurisdiction signed a 287(g) agreement. As an example of the programβs evolution over time, Figure 1 shows the number of individuals that were taken into custody by ICE in one of the largest 287(g) jurisdictions, Los
Angeles County California. Before Los Angeles County adopted its 287(g) agreement in 2005, there were very few individuals taken into custody by ICE. After 2005, the number of individuals taken into custody by ICE dramatically increased over time.
To assess the strength of the agreementβs impact on workers, I modify Equation 1 to include a variable that captures the strength of the program:
ππππππππππππ = β +ππππππππππππΞ²1+Ξ²2πππππ π πππ π π¦π¦ππ+Ξ²3π¦π¦π π π π π¦π¦ππ+Ξ²4(ππππππππππππ Γ πππππ π πππ π π¦π¦ππ) +π½π½5(π¦π¦π π π π π¦π¦ππΓ πππππ π πππ π π¦π¦ππ) + Ξ²6(ππππππππππππ Γ π¦π¦π π π π π¦π¦ππ)
+ Ξ²7(ππππππππππππ Γ βππππβπ π π π π¦π¦π π πππππ π βππππ)
+ Ξ²8(ππππππππππππ Γ πππππππ π π π π¦π¦π π πππππ π βππππ) + Xππππππππ+ βππππππ
(4)
where highstrength equals one if the individual lives in a jurisdiction with a 287(g) agreement where ICE takes a percentage of individuals into custody that is above or equal to the mean, and lowstrength equals one if the individual lives in a jurisdiction with a 287(g) agreement where ICE takes a percentage of individuals into custody that is below the mean. The definitions of all other variables are identical to those in Equation 1.
Table 12 reports the results of Equation 4. I find that both high and low strength agreements have strong negative effects on immigrant worker wages, with male
immigrants in jurisdictions with high strength 287(g) agreements earning a 7.6 log point wage penalty and male immigrants in jurisdictions with low strength 287(g) agreements earning a 4.8 log point wage penalty. These effects are similar in magnitude, and are statistically different at the 1 percent level. The effect of high and low strength agreements on female workers is almost equivalent, with female immigrant workers earning 6.2 log point wage penalty in jurisdictions with high-strength agreements and female immigrant workers earning a 7.1 log point wage penalty in jurisdictions with low- strength agreements. I formally test these effects and find that they are not statistically different. Therefore, there is evidence that male immigrant wages differ based on the strength of each 287(g) agreement, but there is little evidence that this difference exists for female immigrant wages.
However, when examining the impact of these agreements on potentially undocumented workers, there is a separation between the effect of high-strength
reports these results. Male immigrants in jurisdictions with high strength 287(g) agreements earn an additional 7.7 log point wage penalty and male immigrants in jurisdictions with low strength 287(g) agreements earn an additional 4.6 log point wage penalty. Though similar in magnitude, these effects are statistically different at the 5 percent level. The effect of high and low strength agreements on female workers is again unclear, with female workers earning a 5.1 log point wage penalty for high-strength agreements and female workers earning a 9.1 log point wage penalty for low-strength agreements.
These results show that the effects of these agreements differ based on the actual strength of the agreement if the individual is male and undocumented. For both
undocumented immigrants and immigrants in general, the effects are felt simply by having the agreement in place, rather than the strength of the agreement itself. This provides evidence that the inherent nature of the agreement, rather than the strength of the agreement, is driving the effects of these agreements on immigrant worker wages.