Immigration Detention: Opportunities Exist to Improve Cost Estimates
Highlights
What GAO Found
U.S. Immigration and Customs Enforcement (ICE) formulates its budget request for detention resources based on guidance from the Office of Management and Budget and the Department of Homeland Security (DHS). To project its detention costs, ICE primarily relies on two variables—the average dollar amount to house one adult detainee for one day (bed rate) and the average daily population (ADP) of detainees.
U.S. Immigration and Customs Enforcement's (ICE) Formula to Calculate Detention Costs
GAO found a number of inconsistencies and errors in ICE's calculations for its congressional budget justifications (CBJs). For example, in its fiscal year 2015 budget request, ICE made an error that resulted in an underestimation of $129 million for immigration detention expenses. While ICE officials stated their budget documents undergo multiple reviews to ensure accuracy, ICE was not able to provide documentation of such reviews. Without a documented review process for reviewing the accuracy of its budget request, ICE is not positioned to ensure the credibility of its budget requests.
ICE has models to project the adult bed rate and ADP for purposes of determining its budget requests. However, ICE consistently underestimated the actual bed rate due to inaccuracies in the model, and it is unclear if the ADP used in the budget justification is based on statistical analysis. GAO identified factors in ICE's bed rate model—such as how it accounts for inflation and double counts certain costs—that may lead to its inaccurate bed rate projections. For example, in fiscal year 2016, ICE's projections underestimated the actual bed rate by $5.42 per day. For illustrative purposes, underestimating the bed rate by $5 per day, assuming an ADP of 34,000, yields a more than $62 million underestimation in the detention budget request. By assessing its methodology and addressing identified inaccuracies, ICE could ensure a more accurate estimate of its actual bed rate cost. Additionally, ICE reported that the ADP projections in its CBJs are based on policy decisions that account, for example, for anticipated policies that could affect the number of ICE's detainees. While ICE's projected ADP may account for policy decisions, documenting the methodology and rationale by which it determined the projected ADP would help demonstrate how the number was determined and that it was based on sound assumptions.
ICE's methods for estimating detention costs do not fully meet the four characteristics of a reliable cost estimate, as outlined in GAO's Cost Estimating and Assessment Guide . For example, while ICE's fiscal year 2018 detention cost estimate substantially met the comprehensive characteristic, it partially met the well-documented and accurate characteristics, and minimally met the credible characteristic. By taking steps to fully reflect cost estimating best practices, ICE could better ensure a more reliable budget request.
Why GAO Did This Study
In fiscal year 2017, ICE operated on a budget of nearly $3 billion to manage the U.S. immigration detention system, which houses foreign nationals whose immigration cases are pending or who have been ordered removed from the country. In recent years, ICE has consistently had to reprogram and transfer millions of dollars into, out of, and within its account used to fund its detention system. The explanatory statement accompanying the DHS Appropriations Act, 2017, includes a provision for GAO to review ICE's methodologies for determining detention resource requirements. This report examines (1) how ICE formulates its budget request for detention resources, (2) how ICE develops bed rates and determines ADP for use in its budget process, and (3) to what extent ICE's methods for estimating detention costs follow best practices. GAO analyzed ICE's budget documents, including CBJs, for fiscal years 2014 to 2018, examined ICE's models for projecting ADP and bed rates, and evaluated ICE's cost estimating process against best practices.
Recommendations
GAO recommends that the Director of ICE: (1) document and implement its review process to ensure accuracy in its budget documents; (2) assess ICE's adult bed rate methodology; (3) update ICE's adult bed rate methodology; (4) document the methodology and rationale behind the ADP projection used in budget requests; and (5) take steps to ensure that ICE's detention cost estimate more fully addresses best practices. DHS concurred with the recommendations.
Recommendations for Executive Action
| Agency Affected | Recommendation | Status |
|---|---|---|
| United States Immigration and Customs Enforcement | The Director of ICE should take steps to document and implement its review process to ensure accuracy in its budget documents. |
In April 2018, we reported on U.S. Immigration and Customs Enforcement's (ICE) processes for developing immigration detention cost estimates. During our review of ICE's congressional budget justifications from fiscal years 2014 through 2018, we found a number of inconsistencies and errors in ICE's calculations pertaining to immigration detention costs. While ICE officials stated that their budget documents undergo multiple reviews to ensure accuracy, ICE was not able to provide documentation of such reviews. We recommended that ICE take steps to document and implement its review process to ensure accuracy in its budget documents. Starting in fiscal year 2022, the Department of Homeland Security instructed all components, including ICE, to use a new system to build their budget requests. According to the user manual, the system has built-in data checks to ensure accuracy when developing budget documents. The adoption of this new system should avoid the types of errors and inconsistencies that we found during our review. As a result, this recommendation is closed as implemented.
|
| United States Immigration and Customs Enforcement | The Director of ICE should take steps to assess ICE's adult bed rate methodology to determine the most appropriate way to project the adult bed rate, including any inflation rates used. |
In April 2018, we reported on U.S. Immigration and Customs Enforcement's (ICE) processes for developing immigration detention cost estimates. ICE uses two key variables-the average daily population and the bed rate-to calculate a cost estimate for the resources needed to manage the immigration detention system. ICE multiplies the projected average daily population by the projected bed rate by the number of days in the year to determine the amount necessary to operate the immigration detention system for adult detainees. We found that ICE underestimated the actual bed rate from fiscal years 2014 through 2017, by a range of $0.31 to $8.08. We noted that a difference of $5 in the bed rate could lead to a difference of tens of millions of dollars in the final budget calculation. We recommended that ICE assess the adult bed rate methodology to determine the most appropriate way to project the adult bed rate. In the first quarter of fiscal year 2021, ICE revised the methodology used to calculate the projected adult bed rate. For example, the new methodology incorporates a more robust inflation analysis. The new methodology will help ensure that ICE does not underestimate the adult bed rate and should provide more accurate budget requests for immigration detention costs. As a result, this recommendation is closed as implemented.
|
| United States Immigration and Customs Enforcement | The Director of ICE should take steps to update ICE's adult bed rate methodology by incorporating necessary changes based on its assessment, and ensure the use of appropriate inflation rates and the removal of family beds from all calculations. |
In April 2018, we reported on U.S. Immigration and Customs Enforcement's (ICE) processes for developing immigration detention cost estimates. At the time of our review, ICE based its projected adult bed rate on historical costs since fiscal year 2009. Specifically, to estimate what ICE's projected adult bed rate would be two years into the future, ICE averaged the year-over-year percentage change in costs and multiplied the current bed rate by this figure twice. ICE officials told us that the average of the year-over-year percentage change served as its inflation rate and more accurately reflected the escalation of detention costs. However, we identified a number of factors in ICE's bed rate model that led to inaccuracies, including using incorrect inflation factors and mixing costs for family and adult facilities. We recommended that ICE incorporate changes to the adult bed rate based on its methodology assessment, including using appropriate inflation rates and removing family beds from calculations. At the end of fiscal year 2018, ICE removed the cost of family beds from its calculations of the projected adult bed rate. ICE provided spreadsheets that demonstrated that the cost of family facilities was no longer included in its calculations of the projected adult bed rate. Furthermore, in the beginning of fiscal year 2021, ICE began using a new methodology to calculate the projected adult bed rate. As part of this new methodology, ICE now calculates inflation rates using a linear regression analysis and bases the inflation factors of each of the bed rate components (e.g., bed/guard costs, health care) on the most recent 5 years of actuals. ICE provided spreadsheets that demonstrate how the model calculates and incorporates the inflation rates. In addition, according to ICE's Adult Bed Rate Calculator User Guide, ICE provides the inflation rates for each bed rate component to the Department of Homeland Security and the Office of Management and Budget for review and further discussion. Removing the cost of family facilities and revising the method for calculating inflation rates should help ICE project a more accurate adult bed rate. As a result of these changes, this recommendation is closed as implemented.
|
| United States Immigration and Customs Enforcement | The Director of ICE should take steps to determine the most appropriate way to project the ADP for use in the congressional budget justification and document the methodology and rationale behind its ADP projection. As part of that determination, ICE should consider the extent to which a statistical model could be used to accurately forecast ADP. |
In April 2018, we reported on U.S. Immigration and Customs Enforcement's (ICE) processes for developing immigration detention cost estimates. During our review of ICE's congressional budget justifications (CBJ) from fiscal years 2014 through 2018, we could not identify a clear methodology that ICE used for developing the average daily population (ADP) to calculate detention-related budget needs. ICE officials told us the final ADP figure was based on policy decisions. We recommended that ICE take steps to determine the most appropriate way to project the ADP and document the methodology and rationale behind the ADP projection. We suggested that ICE consider the extent to which a statistical model could be used to accurately forecast ADP. During the course of our review, ICE officials explained the agency's statistical model that could be used to project ADP. According to officials, ICE had been using the model since 2014 to estimate the amount of funding the agency would need for detention costs for the remainder of the fiscal year. We found that this type of model was a reasonable method to forecast ADP, and that the modeling choices were generally consistent with accepted statistical practices and appropriate for the data and application. Since then, ICE provided documentation that it also used this model to forecast the ADP included in its initial budget requests submitted to the Office of Management and Budget during the fiscal years 2019 through 2022 budget cycles. ICE's use of this model should help the agency project a more accurate ADP on which to base its budget request for detention costs. As a result, this recommendation is closed as implemented.
|
| United States Immigration and Customs Enforcement | The Director of ICE should take steps to ensure that ICE's budget estimating process more fully addresses cost estimating best practices. |
In April 2018, we reported on U.S. Immigration and Customs Enforcement's (ICE) processes for developing immigration detention cost estimates. We found that ICE's cost estimate for immigration detention resources did not fully meet best practices outlined in GAO's 2009 Cost Estimating and Assessment Guide. As defined in the guide, the characteristics of a reliable cost estimate are comprehensive, well documented, accurate, and credible. Our assessment found that ICE's 2018 cost estimate substantially met the comprehensive characteristic, partially met the well documented and accurate characteristics, and minimally met the credible characteristic. We recommended that ICE take steps to ensure that its process for estimating immigration detention costs more fully addressed cost estimating best practices. In June 2023, we assessed ICE's budget estimating process to determine if it more fully addressed cost estimating best practices given the changes that ICE implemented to address our other four recommendations. Our assessment focused on the three characteristics of a reliable cost estimate which ICE's process did not meet or substantially meet during our 2018 review: well-documented, accurate, and credible. We found that the "well-documented" characteristic improved from partially to substantially met, but the other two characteristics did not improve. In November 2023, ICE began implementing a new model to estimate the projected adult bed rate and performed a risk and sensitivity analysis on the model. In January 2024, we again assessed ICE's budget estimating process and found that the "accurate" and "credible" characteristics improved to be substantially met. According to ICE officials, this model was used to develop the fiscal year 2025 cost estimate for immigration detention. The new model and other changes ICE has made to its process should help ICE develop a more reliable estimate for immigration detention costs. As a result of these changes, this recommendation is closed as implemented.
|