Figure III.1 shows the estimated costs of staff or contractor time needed to program IT systems for ELE. In Louisiana, the state Medicaid agency spent $83,000 on up-front programming costs, and its partner agency spent a further $22,500. Notably, Louisiana also spent approximately $310,000 on staff time to troubleshoot the automated data-matching mechanism in the six months following ELE implementation. This amount represents the cost of 4 to 10 Medicaid program staff members working to resolve matching problems, including manual processing of any ELE cases that could not be processed automatically and continually updating the matching algorithms to reduce the number of cases needing manual review. Although other programs with automated data matches also experienced a start-up period during which processes were refined, they did not report staff time expenditures of the same magnitude as those experienced by Louisiana.
Iowa Medicaid’s ELE strategy also required extensive programming work, costing $84,000 at implementation. Oregon and Alabama spent $1,600 and $6,300, respectively. New Jersey’s partner agency, the Division of Taxation, spent approximately $40,000 at implementation, and a small amount of additional funding after the first year of the program to amend the ELE question on tax return forms.23
Maryland’s Medicaid agency did not spend anything on programming at implementation. Its Express Lane partner agency, the Office of the Comptroller, incurred programming costs, but was not available to answer questions about its ELE implementation experience. Maryland’s program has continued to evolve since initial implementation, and in early 2012 the state was preparing to implement more targeted outreach processes enabled by a recently negotiated data-sharing arrangement. The state will now screen the list of tax filers under 300 percent of the Federal poverty level (FPL) who indicate uninsured dependents against the database of individuals already enrolled in Medicaid or CHIP, and will mail ELE applications only to households that do not already receive public insurance. Programming work to enable these more targeted mailings is estimated to cost approximately $25,000, although this cost may be absorbed by the Medicaid agency’s regularly contracted IT programmer.
Figure III.1. Programming Costs at Implementation, by ELE Program
Notes: Costs shown here include Medicaid and/or CHIP agency costs, and ELE partner agency costs, as appropriate. Maryland’s programming costs are not shown because we were unable to obtain data from the partner agency, which bore all the programming costs at initial implementation.
CHIP = Children’s Health Insurance Program; ELE = Express Lane Eligibility.
There are two explanations for the variation in IT costs. First, Iowa Medicaid and Louisiana have implemented ELE programs with a higher degree of automation than other ELE programs. Louisiana’s ELE program is fully automated: in 90 percent of cases, staff spend no time processing applications for ELE enrollees, and in the remaining 10 percent, they spend around five minutes per application. For renewals, which affect many more children, Louisiana’s process is completely automated. Iowa Medicaid’s ELE program is less automated, but the data match between SNAP and Medicaid data and the mailing of an ELE application form are automated. Staff only become involved when application forms are received. Second, programming costs appear to be related to the IT systems in use. Louisiana and Iowa Medicaid both use legacy Medicaid Management Information Systems (MMIS), which makes programming more cumbersome than with newer systems. In contrast, Oregon’s MMIS was implemented in December 2008, and the state’s very low programming cost for ELE reflects the relative simplicity of making changes to this modern system.
Qualitatively, states also perceived that prior data-sharing relationships made ELE easier to implement. Six of the seven programs studied in the first year of the evaluation built on existing data-sharing relationships with their partner agencies. In the case of all four SNAP partnerships, Medicaid already benefitted from a shared information technology infrastructure, client database, or other data-sharing arrangement. States uniformly highlighted these relationships as a key factor in the initial selection of partner agencies for ELE implementation. As states consider extending ELE programs to additional agencies where data-sharing partnerships must be newly forged, they may face more significant logistical and administrative hurdles, and consequently higher initial implementation costs to establish data-sharing procedures.
23 New Jersey’s third-party contractor also spent a small amount on programming, but these costs are not shown because they were absorbed by the contractor.