Model-Derived Jobs: The Undervalued Side of EB-5

By Kim Atteberry

Recently, proposed legislation has shown that many on Capitol Hill undervalue or misunderstand the concept of model-derived jobs in the EB-5 context. This misunderstanding stems from two sources: the fitful evolution of USCIS’ job-creation policy and a general lack of trust for economic impact modeling. The purpose of this article to inform readers about both of these sources of confusion, including the quality and limitations of economic modeling in the EB-5 context. My goal is to help provide a solid basis for future legislative drafting. The EB-5 industry needs thorough, well-considered legislation and regulation in order to maintain integrity and create jobs for Americans. 

Model-Derived Jobs Defined

Originally, the EB-5 Program focused on individual investors who would capitalize and manage a new commercial enterprise (NCE) to create full-time direct jobs for qualifying workers. Once the Regional Center Program was introduced, the focus shifted to model-based jobs resulting from pooled investor capital. The shift to model-derived jobs was a difficult transition and still causes confusion within USCIS, the EB-5 industry, and congress.  

Model-derived jobs are those estimated by econometric models, such as RIMS II, IMPLAN, REMI, or REDYN (among others). An economist can examine project elements such as expenditures, revenue, or wages to estimate how many jobs a particular project may create. Jobs forecast by using a model can be categorized into three job types: direct, indirect, and induced.

It is important to understand the differences among job types.  If the industry under consideration is an auto manufacturer, then the direct jobs occur within the automotive industry. Indirect jobs would occur within the industries that supply the inputs of production for the auto manufacturer, like the steel industry, the glass manufacturing industry, and the rubber producing industry.  Finally, induced jobs are those created when workers from the auto, steel, glass, and rubber industries spend their paychecks in the local economy to purchase homes, food, and clothing.

An Insider’s View of USCIS Policy Evolution

During my time at USCIS, I educated many involved with the EB-5 program, including USCIS leaders, adjudicators, Administrative Appeals Office decision makers, Congressional staffers, and employees of other Federal agencies such as the Bureau of Economic Analysis (BEA) and the Bureau of Labor Statistics. One of the most important lessons I taught was that jobs forecast by a model include all jobs—both full-time and part-time. 

This information was problematic for some in the service because the law specifically states that the NCE must “create full-time employment for not fewer than 10 U.S. citizens or aliens lawfully admitted for permanent residence or other immigrants lawfully authorized to be employed in the United States (other than the immigrant and the immigrant's spouse, sons, or daughters).”  Further, the Immigration & Nationality Act (INA) defines full-time employment as “a position that requires at least 35 hours of service per week at any time, regardless of who fills the position.” There is no way to provide evidence that model-derived jobs are full-time or filled by lawful workers. 

Given its language allowing “reasonable methodologies,” congress clearly intended to allow model-derived jobs under the Regional Center program. In order to reconcile the conflicting statutory language, the policy within USCIS was to allow model-derived jobs for indirect and induced job creation, but not allow model-derived jobs for direct job creation. For direct jobs, USCIS insisted instead on actual qualifying direct jobs evidenced by Forms I-9 and W-2. At this stage in its policy evolution, USCIS drew a distinct line between the different job types, but unfortunately those policy decisions were only outlined in policy memoranda, and not in actual regulatory language.

At some point after my departure from USCIS, the service began to modify its interpretation of direct job qualification. One reason was that most Regional Center-affiliated job creation is created within a separate entity called the job creating entity (JCE). The EB-5 statute does not mention qualifying requirements for JCE employees; the requirements apply only those jobs created within the NCE. Thus, USCIS clarified verbally through their economists that all job creation in a JCE-structured project is considered “indirect” in terms of the EB-5 investment, and that all model-derived jobs (direct, indirect, & induced) may be lumped together and counted (except for direct construction and A&E jobs in a project with less than 24 months of construction). 

Again, the policy evolution was never clarified nor formalized, which I believe has led some in congress to believe the EB-5 industry exploits economic modeling to suit its own purposes.  For instance, those unfamiliar with the policy history look at the industry’s job-creation calculations and see an overreliance on indirect jobs. This is not true, of course, since model-derived direct jobs mathematically account for roughly half of all job creation within EB-5 projects.  USCIS has simply chosen to re-label the model-derived direct jobs, but has done so in an informal manner that has probably confused those outside of the industry.

The Value of Economic Modeling and Why Geography Matters

Recent conversations related to proposed EB-5 legislation have shown a distinct lack of trust within congress for EB-5 model-derived job estimates. There is no simple fix for this distrust, but perhaps skeptics in congress should talk to their own Congressional Research Service (CRS). The CRS works exclusively for the United States Congress, providing policy and legal analysis to committees and Members of both the House and Senate, regardless of party affiliation. CRS economists undertake impact analyses using the same models and methodologies employed by EB-5 economists. Many important legislative decisions, such as Base Realignment and Closure, the Keystone Pipeline, the Affordable Care Act have been based these same methodologies.

As mentioned earlier, there are a number of recognized and widely-accepted modeling tools available for EB-5 job creation estimation. Using any of these models, the first step in performing an economic analysis is to determine the geographic area under study, the size of which depends on the goals of the study. For example, when examining the impacts of Hurricane Katrina, the BEA likely utilized single county, multi-county, single state, and multi-state geographies to determine the extent of the storm’s economic impact.  It is important to note that the smallest study area possible is a single county because of economic data limitations. 

For an EB-5 analysis, we use different approaches depending on the type of project, local variances, and the type of submission. If we are looking at the impacts of a single hotel project, then we typically utilize local commuting patterns as the impact area. If we are looking at the impact of a major seaport project, we utilize multipliers that cover the multi-state hinterland of the port (the inland area where imports are distributed and from which exports are collected). If we are examining the potential impact a new Regional Center, we use economic multipliers that cover the entire scope of the Regional Center geography.

Economic impact modeling is widely-accepted and is the best method to estimate job creation in the EB-5 industry. Unfortunately, it is not possible to show empirically that indirect/induced model projections are accurate since we have no real-world data to compare against. However, on the direct job side, I have seen numerous examples where the model comes quite close to the actual staffing projections provided by the project developer. The following table provides a brief, redacted analysis of several of these projects across various industries.


Model Direct Jobs

Actual Direct Jobs

Quick Serve Restaurant



Hotel #1



Hotel #2



Assisted Living Facility




Recall from earlier that the model always includes part-time and full-time jobs, and that there is no way to separate the categories. For this reason, this table compares the model against the developers’ lists of all positions, not just the full-time jobs. Clearly, impact analysis is not perfect; however, the table shows that the estimates are reasonable.

My Concerns about Future EB-5 Legislation

A couple of concepts related to job creation were included in recent versions of draft legislation.  Specifically, one provision would have required that at least 10 percent of qualifying job creation be direct jobs specifically. A second provision would have required certain jobs be created within the Targeted Employment Area (TEA). While I applaud new legislation for the program, Congressional authors must understand the impacts of such language.

The proposed 10 percent direct job requirement did not specify whether model-derived direct jobs would be acceptable. In fact, there was additional language stating that the Secretary of Homeland Security may ask for additional evidence to prove that these direct jobs had been created, further complicating the issue. The proposed language did not indicate what type of evidence would be acceptable—expenditure/revenue numbers or actual employee I-9’s. If the draft legislation intended to allow model-derived direct jobs, then it needed to say so explicitly, noting that expenditure/revenue evidence would be sufficient since it is not possible to prove model-derived direct jobs with I-9’s. Furthermore, if model-derived direct jobs were to be acceptable, the 10 percent rule would have been moot. As I’ve already discussed, about 50 percent of all qualifying jobs are actually model-derived direct, and renamed as “indirect” through a USCIS policy decision.

If instead Congress intended to allow only actual direct jobs (as had been the USCIS policy while I was at the Service), then Congress needs to know that their language would have disqualified most infrastructure projects from using EB-5 capital. Large infrastructure projects are typically completed through a complex bid process with multiple layers of sub-contractors.  It would be virtually impossible to comply with the I-9 requirement because the labor is removed from the control of the NCE and/or the JCE. Thus, investors would be unable to prove any direct job creation at all, and the 10 percent rule would effectively disqualify all of the indirect jobs, too. It is for this reason that I suspect lawmakers undervalue or misunderstand the importance of the indirect jobs resulting from EB-5 capital investment.

The TEA-centric job-creation language was also problematic. This provision would have required petitioners to prove that jobs were created within the very few census tracts incorporating the TEA. As noted earlier, the smallest study area for existing impact models is the county level. Thus, model-derived jobs cannot be estimated at the census tract level. The only way to comply with the proposed requirement would have been with actual direct job data—not model-derived job creation. If Congress had intended to allow model-derived direct jobs in the draft, then it was counter-productive to place other restrictions that would have prohibited the use of modeling. 


The EB-5 program was never intended to be a local-level endeavor; the goal was to benefit the entire country through widespread job creation. One of the best ways to measure this benefit is through calculating indirect jobs, where a construction project in Iowa may create indirect jobs in a steel plant in Pennsylvania. Instead, much of the recent debate on reauthorizing the program focused on how projects benefit the local area – specifically in terms of direct job creation within a tightly-defined TEA. I believe this is an overly narrow view of job creation, runs counter to the original intent of the regional center program, and ignores the limitations of modeling. It further demonstrates that some lawmakers undervalue or misunderstand indirect jobs. Clearly, the EB-5 program needs updated legislation and just as importantly, new regulations. Thoughtful drafting can prevent long-term interpretation conflicts.  

Kimberly Atteberry

Kimberly Atteberry