Evaluation to Determine the Effectiveness of PARIS

Evaluation to Determine the Effectiveness of the Public Assistance Reporting and Information System – Final Report

Submitted to:
Administration for Children and Families
Office of Financial Services/Division of Financial Integrity
370 L’Enfant Promenade SW
Aerospace Building, 6th Floor-East
Washington, DC 20447

Submitted by:
Health Systems Research, Inc, an Altarum Company
1200 18th St. NW, Suite 700
Washington, DC 20036

June 30, 2007

Executive Summary

I. Introduction

A. Background on PARIS Project Development
B. Potential Benefits of PARIS Participation
C. Brief Overview of the Evaluation Design and Research Questions
D. Study Methodology

II. Initial Planning and Participation in PARIS

A. How State Agencies First Heard About PARIS
B. Reasons Why Agencies Chose to Participate
C. Decision Making and Approval Process
D. Challenges to Participation in PARIS

III. Current Staffing and Administration of PARIS

A. Organizational Location of PARIS and Effect on Implementation
B. Adequacy of Current Staff Level
C. Communication/Information Dissemination

IV. Submitting and Matching Data in PARIS

A. Identifying Data Elements
B. Data Preparation and Submission
C. Challenges to Data Submission

V. Follow-up on Matched Data

A. States Approach to Conducting Follow-up
B. Percent of Matches Received Back from DMDC
C. Multiple State Matches and Geographic Reach
D. Proportion of Matches for Which Follow-up is Conducted
E. Follow-up After Receiving Matched Data from DMDC
F. Usefulness of Matched Data Received from DMDC
G. Filters
H. Use of Special Summary Reports
I. Follow-up Priorities

VI. Costs of Implementing and Maintaining PARIS

A. Developing Cost Models
B. Creating Models for Calculating Benefits
C. Populating Benefit Models

VII. Future Expansion of PARIS

VIII. Conclusion

Appendices

Listing of Tables

Table 1. Protocols Used for Each State Included in this Report
Table 2. Follow-up Activities of States Attending the 2007 PARIS Conference
Table 3. Lead Organizational Locations of PARIS
Table 4. Percent of Matches Received by States Included in This Report, August 2006 Interstate Match
Table 5. Oklahoma Match Follow-up Activities
Table 6. Types of Data Filters Used by State
Table 7. Reported Savings by States Included in the PARIS Study Sample
Table 8. Estimated Annual Savings from Medicaid Managed Care Case Closures Based Upon PARIS Matches in a Sample of States not Currently Participating in PARIS

Listing of Figures

Figure 1. Map of States with Signed PARIS Agreements, Prior to and Since September 2005
Figure 2. Map of Number of Matches in Which States Participate
Figure 3. Map of States Reporting Follow-up Activity Locations

I. Introduction

The Public Assistance Reporting Information System (PARIS) is a computer data matching and information exchange system administered by the Administration for Children and Families (ACF) to provide States with a tool to improve program integrity in the administration of public and medical assistance programs. The PARIS project is designed to match State enrollment data from the Temporary Assistance to Needy Families (TANF) Program, the Food Stamp Program, and Medicaid, with data from other participating States and from a selected group of Federal databases. Using the client’s Social Security number (SSN) as the unique identifier, the files submitted by the States are matched against:

The Interstate match, where social security numbers of public assistance clients are submitted by participating PARIS States and matched with data from all other participating States to determine if participants are enrolled in two or more States.
The VA match, which provides States with information on the eligibility of their clients for veterans’ benefits and also allows States to confirm if their clients are receiving income and medical assistance payments from the Department of Veterans Affairs (VA).
The Federal match which matches State data with information from the Department of Defense and the Office of Personnel Management to determine if clients are receiving income from any of these sources or are eligible for Federal health care coverage.¹

States participate in PARIS on a voluntary basis, and they receive no ongoing funding from ACF for participation. States are required to sign a PARIS Memorandum of Agreement (MOA), which commits the State to a minimum participation level and requires that data be submitted in a standardized format.

Effective September 30, 2005, ACF awarded Health Systems Research, Inc., an Altarum Company (HSR), a contract to conduct an evaluation of PARIS. The purpose of this evaluation is to assess the PARIS Project by examining a number of factors related to the administration, cost, and overall usefulness of the information provided. The three main goals of the PARIS evaluation are to:

Assess the effectiveness of State implementation efforts to determine what factors facilitate or hinder the execution of PARIS.
Examine how States have implemented PARIS, and how data from the various PARIS matches are used to support program integrity efforts.
Develop cost and benefit models in order to identify, where possible, consistent methods by which costs and benefits of the PARIS system can be calculated.

As of June 2007, there were 43 jurisdictions that had signed PARIS agreements.²

A. BACKGROUND ON PARIS PROJECT DEVELOPMENT

In 1993, ACF began working with State public assistance and Federal agencies to develop information-sharing projects that could effectively verify the circumstances of public assistance clients. ACF sought to identify information-sharing opportunities, and to lead and coordinate the activities that could tap into the potential for matching client participation data across States and programs.

The Department of Defense’s (DOD) Defense Manpower Data Center (DMDC) provides computer resources to support PARIS development and operation. DMDC receives all data from the State and Federal sources, and then produces the three individual match files. However, the inclusion of all three matches has been a gradual process. Initially, between 1993 and 1997, ACF sporadically provided 29 State public assistance agencies with VA compensation and benefit information for use in verifying public assistance clients’ circumstances. Then, in September of 1997, while continuing to provide VA information to States, ACF began testing the feasibility of conducting an interstate match, which provided information to 16 States regarding possible duplicate interstate public and medical assistance payments. In 1999, when DMDC became the provider of matching services, the continuity in routine matching activities began. In May 2002, PARIS implemented a third method by which States could verify client eligibility for services; the Federal match, which allows States to compare data housed in State systems with Federal information regarding active and retired Federal civilian employees and active and retired military personnel.

To oversee the ongoing development of PARIS, and to encourage other States to participate, a PARIS Board of Directors was formed in 1998. The PARIS Board provides guidance and advice to ACF, and it plays an important role in examining the ongoing matching process and suggesting ways to improve the PARIS system. The PARIS board not only provides support to ACF for the management and administration of PARIS, but also serves as a resource for other States with regard to PARIS implementation issues.

ACF also sponsors an annual meeting in the Washington D.C. area where PARIS member States and States interested in PARIS participation can come together to learn more about ways in which PARIS can support State program integrity efforts. In the past two years, ACF has sponsored officials from each PARIS member State to attend the PARIS conference. All participating States are encouraged to send representatives to the annual PARIS meeting where they can network with representatives from other PARIS States as well as meet with and learn from the PARIS Board of Directors. States that are not PARIS members may also attend at their own cost, but can obtain valuable information to assist them in making the decision to become a PARIS member.

Growth of the PARIS project has proven impressive. As noted earlier, PARIS started with just 16 States. At the time this project was initiated, there were only 26 States participating in PARIS. The Board of Directors’ aggressive promotion of PARIS, along with funding opportunities provided by ACF that encouraged States to join, have increased the number of States to 43. Additionally, some of the few remaining non-member States have expressed interest in joining PARIS in the near future. Figure 1 displays the States that have signed PARIS agreements and notes those that are new since September of 2005.

Figure 1. Map of States with Signed PARIS Agreements, Prior to and Since September 2005 D

B. POTENTIAL BENEFITS OF PARIS PARTICIPATION

The PARIS match process offers States multiple opportunities to improve program integrity and save money on improper payments. It furnishes information that can be used to make appropriate adjustments to benefits provided to a client. The matches also can lead to cost savings due to case closures, the recovery of improper payments, and the coordination of medical insurance benefits between Medicaid and other Federal-sponsored health insurance coverage. These advantages are described more fully below:

Closing cases reported as active when a participant has moved from one State to another, but has not reported the move (interstate match). Quite often interstate matches result when a participant moves from one State to another without reporting the move to officials in the original State of residence. States have set up systems to verify the residency of participants who show up on the interstate match, and to close cases in which it can be verified that the participant resides in another State.³ Case closures bring cost savings from the halting of future TANF payments and Food Stamp Program benefits, and eliminates any payments for that client to managed care organizations participating in the Medicaid program.⁴
Examination of potential dual participation for the purpose of detecting fraud and recovering overpayments (interstate match). One of the important uses of the PARIS interstate match is to identify clients who may intentionally be receiving benefits in more than one State and to take appropriate action to recover potential duplicate benefits. States work together to examine matches where active cases show up in both States to determine if benefits have been issued by multiple States. If so, either State (or both) can take action to stop the duplication of benefits, recover benefits through the issuances of an improper payment claim to the participant, and take legal action against participants for committing fraud and abuse.
Verifying income from the VA and other Federal sources (VA and Federal matches). The VA and Federal matches provide States with information on client earnings classified by type and source of income from Federal employment databases and through payments to veterans made by the VA. By using data from the VA and Federal matches, local offices can verify whether or not income was reported, if it was reported accurately, and whether or not the income should have been counted in determining eligibility. The result of this verification can lead to an adjustment of benefit levels or to the discontinuation of benefits for clients whose income levels are too high to meet eligibility requirements.
Coordination of benefits between Medicaid and other Federal insurance coverage (VA and Federal matches). One of the more innovative uses of PARIS data is to ensure that the proper agency is covering the cost of a client’s health insurance benefits. A few States use the VA and Federal matches to determine if clients are eligible for health insurance coverage provided through the VA or through the Federal military retirement system (TRICARE). If a client is eligible for both Medicaid and one or more of these Federal insurance programs, State Medicaid officials can work with these entities to coordinate the payment of benefits. This coordination results in savings to State Medicaid programs, particularly in cases where high-cost benefits such as long-term care are being provided, by shifting the cost of these services to the VA. Additionally, there is an advantage to transferring costs from Medicaid to VA for the families of veterans receiving long term care benefits, as Medicaid requires the estate of the veteran to reimburse Medicaid for all or part of the costs of the long term care. Unlike Medicaid, the VA does not have the same requirement.

Some States have attempted to document the savings that have resulted from use of the PARIS match. However, the methods by which savings were calculated have been inconsistent across the various States. This study examines the potential for developing standardized approaches to calculating cost savings and assesses the feasibility of collecting cost/benefit data when States use diverse approaches to implementing and administering PARIS.

PARIS also has the potential for increasing the number and types of files for which matches are conducted. For example, PARIS officials recently changed the file record layout to expand the number of fields to include new programs in the match. One program currently under consideration for inclusion in the PARIS interstate match is State workers’ compensation data. Adding State workers’ compensation data to the interstate match would accomplish two goals. First, it would allow States to determine if participants are receiving workers’ compensation benefits from other States, and if so, whether or not these payments were reported and treated properly at the time of certification and its impact on other public assistance benefits. Second, it would allow States to determine if workers’ compensation recipients are also receiving public assistance benefits in other States, which may be an indicator of potential fraud or duplicate payments.

A second program being considered for inclusion in the PARIS match is ACF’s Child Care Program. Data from this match would serve two purposes. First, like the VA and Federal matches, it would allow State officials to determine if an individual is receiving payments for child care that have gone unreported. Secondly, it would help State officials determine if someone was receiving child care payments from multiple States. The major barrier at this point to adding child care is that some States make payments to providers, others make payments to individuals, and a few States make payments to both (depending on circumstances). PARIS officials have recently addressed this issue through changes in the PARIS file format.

C. BRIEF OVERVIEW OF THE EVALUATION DESIGN AND RESEARCH QUESTIONS

The evaluation design for this study is cross-sectional, using interviews with a sample of PARIS officials as well as current and historic programmatic and cost data to assess the implementation, process, and effectiveness of PARIS. Effectiveness is being measured in several ways, including a cost-benefit analysis. The evaluation design is divided into two major components: an implementation evaluation and an outcome evaluation.

The implementation evaluation focuses on how well each of the States has been able to develop and implement PARIS. Organizational placement issues, communications between participating programs and those responsible for implementation, assessments of startup barriers, and the adequacy of policies and procedures for accomplishing project objectives were included in the assessment. The outcome evaluation is directed at identifying the extent to which States can effectively use PARIS data to accomplish their goals, at examining the costs and benefits of participation in PARIS, and identifying those factors that are associated with maximizing the benefits relative to costs.

The evaluation is designed to answer the series of research questions initially posed by ACF in the RFP. These evaluation questions include:

How are decisions made regarding the organizational placement of PARIS, and what are the effects of placement?
How do States decide which SSNs to submit to the DMDC for matching?
How much does it cost to initially prepare a State’s technology and program offices to join the PARIS project?
How much does it cost to submit data to the PARIS project on a quarterly basis?
How much does it cost a State to receive data back from the PARIS project?
How much does it cost to follow up and pursue a potential match?
What factors contribute to a State’s decision regarding the frequency of data submission?
What would be an appropriate cost-benefit analysis that interested non-participating States could use to assess their potential participation?
Is there a better or more justifiable approach to use to project and calculate cost savings other than a 12-month projection?
How can we ensure that the States submit their matches on the deadline date?
Are States generally satisfied with the turnaround time they obtain from DMDC?
How do States determine how, if at all, to prioritize their match results?
What type(s) of technical assistance (TA) can the PARIS project provide to the States that would encourage more participation?
What further guidance can be provided by the PARIS project that would ensure that SSNs being submitted are clean and that all appropriate information is in the record to enable another State that matches the record to promptly pursue the case without excessive cost?
What other information could be provided that would encourage additional State participation in the PARIS project?

HSR’s evaluation design gathered information from a variety of respondent groups using multiple data collection techniques. These included interviews and surveys with the following types of respondents: PARIS administrators; Medicaid, TANF, and Food Stamp program officials; staff conducting follow-up; and, where appropriate, financial officers. Administrative and cost data also were obtained through multiple methods, including an analysis of program manuals, protocols, and materials; discussions with program and financial administrators; and the use of data collection tools and worksheets completed by these administrators. Both prospective and retrospective data were collected. Prospective data were collected from entities that were new to PARIS in FY 2006, and both prospective and retrospective data were collected from States that were already participating in the PARIS program prior to 2006.

In order to provide ACF with a complete picture of PARIS implementation issues, the initial study design called for interviews with officials in two jurisdictions that did not participate in the program. The purpose of these additional interviews was to determine what types of reasons and issues affected a State’s choice not to participate in PARIS. A pool of jurisdictions, including some County Governments, was created for inclusion in the study.

A number of States joined PARIS as this study was being conducted. As a result, several non-participating States that were initially selected for inclusion in the study were dropped as they joined. It became apparent during discussions with ACF and officials from the States that recently joined, that information on barriers to participation could be obtained through interviews with recent members. Therefore the decision was made to exclude non-participating States from the study, and instead focus on how newly joining States overcame any barriers to joining. However, because counties are eligible to join PARIS, and because none have, the study team decided to interview officials in one county located in a non-participating State.

To capture information about barriers, and also to determine how well new States that received grants to help them join PARIS were able to implement their program, a supplement to the evaluation design was developed. The grant program was designed to help States overcome some of the potential cost and logistical barriers that might inhibit participation in PARIS, and in its first year, funded three States to join PARIS. In the second year of the grant program, an additional five States were funded.

The grant program not only provided funds to the new States, but linked up these States to existing PARIS States as “mentors.” Mentor States provided assistance with implementation of PARIS data files, and provided ongoing support during the first year of PARIS participation. The evaluation design supplement examined the success of this grantee/mentor program in helping the year one States implement PARIS, and allowed us to collect some limited data from year two States regarding barriers they overcame to participate.

D. STUDY METHODOLOGY

Data for this report were collected through a series of site visits to individual States, where interviews were conducted with various PARIS officials. A sample of 14 participating States were included in the evaluation. ⁵ The sample frame was created by distributing participating States into categories based on a number of factors. These included:

Participating in the Grant and Mentoring Program. All States participating in the year one grant and mentoring program were included. However, during the study one of the States dropped out of the grant program, and declined to be interviewed for the study. Therefore even though this State continues to submit data, officials in this State were not included in the interviews. However, that State’s mentor was still included as part of the sample to gain some insight into the issues the grantee State faced in trying to implement PARIS.

Length of time participating in PARIS. The sample includes States that have participated in PARIS for several years, some that have participated for only a few years, and, with the inclusion of the grantees, some brand new States.

Organizational Location of PARIS within the State. PARIS can be located within a variety of organizations with a State. Some States locate PARIS within an Office of Investigator General (OIG) or Audits, while others may choose a data processing shop or an administrative branch of the health and/or welfare agency. A diverse group of organizational structures were selected.

Organizational Level of Follow-up Activity. Some States conduct follow-up activities only at the local office level, while others do so at the State level or in combination. States with all three types of approaches were included in the sample.

Once the States were selected, a recruitment process was used to convince the States to participate. All of the States selected in the final sample agreed to participate. Onsite interviews were used as the primary data collection method for generating a comprehensive description of the PARIS program along with detailed, qualitative explanations of key issues of interest to States and ACF. A number of steps were involved in the preparation of the study methodology, which are discussed below:

1. Preparatory Work

PARIS member States have multiple processes in place to create the match file, submit the file to DMDC, filter and distribute results from the match, and conduct follow-up. This involves multiple teams of individuals, all playing an important role in the processes. In order to ensure that a comprehensive approach to data collection would provide information about all operational components of the PARIS project, and to obtain the perspective of multiple respondents on key issues, HSR staff developed interview-based data collection instruments. These instruments were designed to address the study objectives and answer the evaluation questions initially developed by ACF and refined through the development of the study plan. Each measurement instrument was specifically tailored to the respondent groups discussed above, and was designed to answer all of the evaluation questions posed for that particular group.

The interview data collection instruments correspond to the respondent groups with whom they were used. The pool of interview guides consists of the following:

PARIS Project Coordinator Interview Guide
Medicaid, TANF, Food Stamp Program Administrator Interview Guide
Technical Staff Interview Guide
PARIS Data Match/Follow-up Staff Interview Guide
Financial Questions to be answered by the most appropriate individual

Each interview guide covers one or more of the following key topic areas:

Background/History of the PARIS program in the State
Administration
Staffing
Submission of Data
Follow-up
Cost of the project and its benefits

Draft data collection instruments were reviewed and approved by ACF. They were subsequently pilot tested during site visits to State agencies in Kentucky, Wisconsin, and Pennsylvania during the spring of 2006. After the pilot testing, a number of changes were made in the instruments to reflect the findings from the pretest. The guides were finalized then and sent to OMB for clearance in accordance with the Paperwork Reduction Act. OMB clearance was received in October of 2006. Copies of the final data collection instruments are included in Appendix B.

2. Data Collection and Limitations

Data collection for the 14 States began in November 2006, and the data from those visits were entered into a project database. At each site visit, two members of the data collection team were present: a primary interviewer, and a note-taker. Immediately after the completion of the interview, the note-taker prepared the information from the interview, shared it with the lead interviewer for editing and review, finalized the document, and entered the data into the project database. As there are no predetermined guidelines that must be followed in the administration of PARIS, each State uses a process they deem most effective, taking into account such factors as State resources, scope of PARIS activity, and organizational placement of PARIS. As a result, not all interview guides were applicable or relevant for all States, except for the questions for the PARIS Coordinator, which were consistently asked across each State. Other protocols were sometimes customized on a case-by-case basis, depending on their relevance to the State’s processes and the level of involvement in PARIS activities of staff being interviewed. Table 1 details the status of protocol use for all site visits included in this report.

Table 1. Protocols Used for Each State Included in this Report

Table 1. Protocols Used for Each State Included in this Report
State	Relevant Protocols
Colorado	Coordinator, Technical Staff, Follow-up Staff
District of Columbia	Coordinator, Technical Staff, Follow-up Staff
Florida	Coordinator, Follow-up Staff, Financial Manager, TANF and Food Stamp officials
Illinois	Coordinator, Technical Staff, Follow-up Staff, Financial Manager
Maryland	Coordinator, Technical Staff, Follow-up Staff
Missouri	Coordinator, Technical Staff, Follow-up
New York	Coordinator, Technical Staff, Follow-up Staff (State and NYC), Financial Manager
Oklahoma	Coordinator, Food Stamp/TANF/Medicaid Staff, Technical Staff, Follow-up Staff, Financial Manager
Oregon	Coordinator, Technical Staff, Financial Manager, Medicaid Officials
Rhode Island	Coordinator, Food Stamp/TANF/Medicaid Staff, Technical Staff, Follow-up Staff, Financial Manager
Tennessee	Coordinator, Follow-up Staff
Utah	Coordinator, Financial Manager
Virginia	Coordinator, Technical Staff
Washington State	Coordinator, Technical Staff, Medicaid Staff, Local Follow-up Coordinator

Additional data were collected from all States attending the March 2007 PARIS conference. While not all participating States attended the conference, data was collected from 38 participating States so that the results could be inferred to provide some national perspective on a limited number of issues. These data were collected to update information originally used to create the sample frame. These data are used in this report to show an overview of all States in a limited number of topic areas.

3. Data Analysis

For each major topic of analysis, we examined the open-ended interview data to capture information as representative as possible of the commonalities and diversity among State agencies with respect to each of the following key domains: background, factors affecting participation, staffing and administration, data submissions, follow-up or investigation of matches, and financial tracking and reporting. The findings from this report reflect an analysis of the key issues of importance for the further development of the PARIS project.

4. Presentation of Findings

The findings from this study are qualitative in nature, and reflect the opinions of a diverse group of respondents involved in all aspects of PARIS activity. Because the sample was designed to be representative of the diverse approaches used by States in implementing PARIS, findings are not presented as a sum or total of findings across States. Rather, findings are displayed by specific themes, with the results of interviews reflecting either the majority opinion of the respondents along with specific examples of activities or concerns presented during the interviews. Each of the sections presented reflects a component of the research questions. The results, while not applicable to all PARIS States, accurately reflect the conclusions of the various respondents within the sample frame.

II. Initial Planning and Participation in PARIS

One of the issues of interest to ACF is how States first heard about PARIS and what factors motivated each State to participate in the program. States were asked about the process leading up to PARIS participation, as well as their expectations regarding the use of PARIS data.

A. HOW STATE AGENCIES FIRST HEARD ABOUT PARIS

The majority of agencies included in the sample first learned about PARIS through national conferences, where a representative from ACF described the project and its potential benefits (such as, improved efficiency, potential cost savings, and potential for reducing improper payments). These presentations helped motivate State agency representatives in learning more about how to get involved in the program. However, it is interesting to note that while PARIS presentations were made at numerous conferences involving TANF, Medicaid, Food Stamp, or welfare fraud investigators, it was often only one of these programs that pushed for PARIS participation.

While institutional history regarding the PARIS program has been lost in many States due to the departure of staff who originally worked on the program, several States were able to pinpoint additional ways State representatives were introduced to PARIS. New York and Pennsylvania together had their own internal matching program for years and the success of these two States was actually the impetus for the PARIS program. Rhode Island was part of the pilot project for PARIS. Additionally, in one State, (Oregon) respondents reported that they learned of the potential benefits of PARIS from officials in Washington State, and after conducting an analysis of these potential benefits subsequently decided it was in their best interest to participate.

B. REASONS WHY AGENCIES CHOSE TO PARTICIPATE

States also were consistent in the reasons why they chose to participate in PARIS. In each case, respondents noted that it was a potential way to improve efficiency, validate income information collected in certification records, recover overpayments made in error, and prevent and detect fraud. Some States also reported additional reasons for participating in PARIS:

Utah had an extremely high food stamp error rate and was concerned about potential sanctions. PARIS was one strategy, along with others, to help reduce the error rates. PARIS also served to help enforce the TANF lifetime assistance limits, which require information from other States.
Both Washington and Oklahoma specifically reported that PARIS allowed them to match data with the VA in their States to coordinate Medicaid and Veterans benefits, which had the potential of reaping large cost savings.
For the agencies that were already conducting matches with neighboring States (such as New York, the District of Columbia, Virginia, Maryland, and Rhode Island), PARIS provided these States the opportunity to expand their matching program with many more States.
Florida joined in large part because of the interstate match. Given that they have a large proportion of people who live in another State during part of the year, PARIS had the possibility of reducing duplication.
Rhode Island was already involved in individual matches with Connecticut and Massachusetts and PARIS allowed them to broaden their scope in determining eligibility.
Tennessee originally joined PARIS to pursue delinquent overpayments (non-active cases that were overpaid at some point in the past) and to discover information about non-biological family members of clients whose income could affect benefits but who were not direct beneficiaries. PARIS administrators informed Tennessee that neither of these reasons are appropriate uses for the program and the State now focuses on using the match data to find addresses of people who are on the Tennessee Medicaid program (TENNCARE), but are no longer residing in the State and should therefore be dropped.
Oregon initially viewed PARIS as a method by which they could detect fraud and abuse, as well as recover payments made in error. However, after testing other applications of PARIS, they decided that it is a more cost-effective tool for cleaning up case files from unreported moves of clients, thus stopping payments and preventing future improper payments.

C. DECISION MAKING AND APPROVAL PROCESS

Some agencies surveyed reported that deciding to participate in PARIS, and gaining approval from high-level policy makers, was relatively simple and informal. There was clear buy-in from the top with little to no objections raised in both Missouri and Rhode Island. In other States, Tennessee, Maryland, and the District of Columbia, PARIS was viewed as another matching opportunity, consistent with their existing efforts, and therefore special approvals were not required. Similarly, Oregon’s decision was based upon testing whether or not the system could provide valuable information.

These experiences with ease of approval may have been the exception rather than the rule for many other PARIS States. During the course of the study, information was provided by both ACF and PARIS Board members that the approval process was often lengthy and somewhat difficult in many States. This difficulty was reflected in the amount of time ACF and PARIS board officials spent working with States to gain official approval for the States to participate.

Two State agencies, Utah and Oklahoma, reported a somewhat more formal process:

Utah had to seek the approval of the PACMIS (integrated public assistance eligibility) system steering committee. The Health Department also had to approve of the process.
In Oklahoma, there were initial discussions over the cost-effectiveness of the program, as well as who should be responsible for its administration. This was challenging because administration of PARIS required more resources than many of the partners could commit. The Office of the Inspector General decided to take the lead, but PARIS was the first time they had a grant project. Given that the process was new, the steps to gain approval were more formal. The coordinator drafted a Portfolio Management Business Case, which was submitted to the Oklahoma Department of Health Services (OKDHS) Information Technology Project Governance Board as a “business enhancement” opportunity. Ultimate approval came from the Director of OKDHS.

All of the agencies surveyed reported that the information provided by ACF regarding PARIS was very useful and persuasive. Additionally, information provided on the PARIS website related to cost savings produced by PARIS provided a good resource for State officials to use in presenting potential benefits of PARIS. However, two agencies suggested that having a structured methodology for calculating the time involved in implementation and the savings that could be generated would have been valuable for budgeting, planning, and buy-in of staff and decision-makers.⁶

D. CHALLENGES TO PARTICIPATION IN PARIS

None of the 14 State agencies reported substantial challenges to participating in PARIS from the standpoint of implementing the technical requirements and submitting data. The States had the support of the relevant policy and IT decision-makers who saw the benefits of the initiative. There were no issues with signing the PARIS agreement, and the time frame from deciding to participate to submitting the application varied from 2 weeks to 6 months. No State reported that the process was unusually difficult or cumbersome.

For the newer States (those that joined in 2005 and 2006) the grant funds were identified as the single most facilitating factor to joining PARIS. Prior to the opportunity to apply for grant funds, several of these new States cited internal priorities and limited staff availability to consider participation in PARIS. Having access to funding, as well as the help of mentor States, made it easier to set aside resources for implementing PARIS. All of the new States had interest in PARIS in the past, but having the ability to add resources for design and implementation was the single biggest factor contributing to their decision to join.

The one county from a non-participating State that was interviewed for this study expressed interest in being involved in the PARIS project, but was reluctant to join separately from the State. These officials believed that a statewide approach to participating in PARIS would be more cost effective than individual counties joining. However, while officials in the county indicated that they would encourage the State to join PARIS, they would reconsider their position on joining if the State made a final decision not to join.

One other challenge that is important to note is that some States have limited the scope of their PARIS activities so that they do not participate in all three of the matches. Almost all 43 of the participating States participate or intend to participate in the interstate match, but fewer participate in the VA and Federal match. The major reasons for limiting the scope of PARIS activities included a lack of resources to manage follow-up activities, the difficulty of understanding and working certain files, and the need to coordinate with an organization (mostly Medicaid) that are located in a different department than the Food Stamp or TANF program. Some of the issues described by States included:

One State noted that it made the decision to limit the scope of its PARIS activities because of a lack of resources with which to conduct follow-up activities. This State initially participated in all three PARIS matches, but once initial data were received, it was determined that local resources were inadequate for follow-up to be conducted on all of the files. Therefore, the State decided not to submit the Federal and VA matches and to focus only on the interstate match, with the idea that if resources become available in the future, they will increase the scope of PARIS to include these matches .
Two States noted that they did not submit data for the Federal match, because they found the data contained in the Federal file to be difficult to understand and is presented in a format that local staff could not easily use for conducting effective follow-up activities. Both States determined that a significant effort would be required to structure the results of the Federal match into a useful format for local staff to use. These States decided that current resources were not sufficient to devote time to this effort.
Two States reported that they had limited contact with their State’s Medicaid program in relation to including Medicaid staff in conducting PARIS follow-up. Both of these States noted that if Medicaid was handled as part of a consolidated record where the client was receiving TANF and/or Food Stamps in addition to Medicaid, the case was worked. However, if the client was receiving Medicaid benefits only, there was a different set of caseworkers handling these clients, and these caseworkers were not involved in PARIS activities. Both States indicated that they planned on entering into discussions with their State Medicaid officials to find a way to include them in PARIS.

Figure 2 displays the PARIS States and the number of matches (1–3) in which they participate, if they plan to submit and how many matches, or if they have not submitted as of yet and have not reported their plans.

Figure 2. Map of Number of Matches in Which States Participate D

However, when it came to using the PARIS data for follow-up, a number of challenges were reported. The major issue that came up in multiple States was the lack of time and resources for follow-up. Instead of creating a separate follow-up system for PARIS, most of the investigative or follow-up activities for PARIS were incorporated into other routine case management and follow-up activities of local staff. As a result, those responsible for conducting the appropriate follow-up were adding a significant workload to an already existing system, without adding any additional resources by which the PARIS matches could be handled.

Additionally, not all States conduct follow-up on all of the PARIS match files. Some States only participate in one or two of the matches, while other participate in all three but do not conduct follow-up activities on all. Table 2 displays the States participating in the March 2007 PARIS conference and the number of files for which follow-up was conducted.

Table 2. Follow-up Activities of States Attending the March 2007 PARIS Conference, N=39

Type of Follow-Up	Number of States	Percentage
Follow-up conducted on all three matches	14	36%
Follow-up conducted on two of three matches	8	21%
Follow-up conducted on one of three matches	7	18%
Are not currently conducting follow-up/new State	6	15%
Did not report	3	10%

Some specific examples of issues related to follow-up implementation included:

One State reported that obtaining support and time from county program personnel to implement PARIS was difficult. To address the issue, there is now a dedicated County Contact who provides technical assistance and training to the counties. This contact person champions PARIS as a way to recover dollars. For example, he provided a cleaned-up VA report so that other counties could see how much money was available to be recovered.
Another State reported that initial attempts to use PARIS data to recover funds did not prove to be cost-effective. This State had decided that PARIS should be used only as a method by which overpayment recoveries could be identified. When used solely for that purpose, the State officials determined that the cost of working the data did not result in enough of a benefit to continue its use. Additionally, attempts to duplicate Washington State’s success with coordination of benefits with the VA were not successful. However, after examining other uses for PARIS, State officials decided that PARIS data would be better used to eliminate duplicate records found through the interstate match and use these data to close cases.
Another State reported that they were limited in the amount of follow-up activity that could take place, and that expanding follow-up activities that would make more effective use of PARIS data did not fit in well with their current modernization efforts. These efforts, which have resulted in more internet-based and call center-based eligibility systems, have reduced the number of local offices and the number of staff, resulting in fewer individuals to conduct the type of follow-up needed to effectively resolve PARIS matches.
One other State reported that officials in the department where Food Stamp and TANF resided had made a decision to drop out of PARIS in 2002 because they determined that PARIS was not cost-effective and that they have limited resources by which to conduct follow-up. However, at the same time they made this decision, officials in the Medicaid program (located in a different department) determined that PARIS participation would be valuable and save money. After discussions internally within the State (including the Governor’s office) and discussions with ACF, officials in this State decided that the benefits identified for the Medicaid program warranted reconsideration of participation on the part of the Food Stamp and TANF programs. A final decision was made between the two departments and the Governor’s office that all three programs would participate.

Further discussion of follow-up issues is presented in later sections of this report.

III. Current Staffing and Administration of PARIS

This chapter discusses the organizational locations of State PARIS programs, analyzing the effect of organizational location on implementation, and noting the advantages and disadvantages of certain approaches. The chapter also discusses the processes, both deliberations and formal agreements, by which States determined the locations of their PARIS programs. The chapter also addresses PARIS staffing and the communication strategies used by States with respect to their PARIS programs.

A. ORGANIZATIONAL LOCATION OF PARIS AND EFFECT ON IMPLEMENTATION

In general, States locate their PARIS programs in one of three areas: a technical services division responsible for preparing and distributing the eligibility information, an audit or OIG office, or one of the general program offices located within a social service agency. The PARIS programs analyzed in this report are located in divisions and agencies that State officials felt were best equipped to administer and integrate PARIS activities into the existing workflow. In Florida, for example, the Department of Children and Families contains a section responsible for ongoing data matching activities, which was a natural home for PARIS. Similarly, the District of Columbia decided that their PARIS program should be located within the Eligibility Review and Investigation Division of the Department of Human Services where technical staff and investigators routinely conduct and follow up on other data matches to determine benefits eligibility.

One of the challenges faced by a few States was the coordination of PARIS activities between departments where TANF and Food Stamps resided, and the department that housed Medicaid. In these States one of two approaches to PARIS administration occurred. As required by PARIS procedures, States identified a single PARIS coordinator in one of the two departments, and that person assumed a lead role in coordinating PARIS activities between the two. However, in some States internal administration of PARIS was accomplished by the two departments undertaking PARIS activities separately from one another, with coordination of data submission and distribution done through a centralized processing center, but distribution of data and responsibility for follow-up being maintained separately in each department.

Table 3 summarizes the organizational location of State PARIS programs.

Table 3. Lead Organizational Locations of PARIS

STATE	ORGANIZATIONAL LOCATION
Colorado	Coordinated out of the Department of Human Services with major responsibilities shared by Medicaid and Systems Management
District of Columbia	Eligibility Review and Investigation Division, Department of Human Services
Florida	Data Exchange Module, Department of Children and Families
Maryland	Coordinated out of the Family Investment Administration with major responsibilities shared by Medicaid and the Office of the Inspector General in the Department of Human Resources
Oklahoma	Office of the Inspector General, Department of Human Services
Oregon	Financial Recoveries Office, Department of Human Services
Utah	Operations and Support Division, Department of Workforce Services
Washington	Cost Efficiency Unit within the Director’s Office, Department of Social and Health Services
New York	Office of Temporary and Disability Assistance
Rhode Island	The Department of Human Services
Illinois	Bureau of Research and Analysis, Department of Human Services
Tennessee	Information Technology Division, Department of Human Services
Missouri	Department of Social Services

1. Discussion on Issues Related to Developing Interagency Agreements

Even among States with centralized eligibility systems, administering PARIS requires significant interagency collaboration. Rhode Island is a notable exception in that all offices participating in PARIS are vertically integrated into the administering agency, the State Department of Human Services. Some States are able to manage interagency collaboration on an ad hoc basis, while others have developed guidelines for administering PARIS. None of the States reported developing interagency workgroups for the purpose of reviewing and refining PARIS protocols, but some States do have formal Memorandums of Understanding (MOUs) or interagency agreements in place:

In Oklahoma, Office of Investigator General (OIG) depends on a standing interagency agreement between DHS and the State Medicaid office that outlines provisions for data sharing between the agencies.
Colorado’s PARIS coordinator wrote an MOU to codify the relationship among State- and county-level agencies.
Tennessee’s PARIS program operates under a formal data sharing agreement with TENNCARE, the State medical assistance program.
Washington State has an MOU in place between the Department of Social and Health Services and the State Veterans Affairs office to facilitate the VA match and the coordination of benefits. The Washington State Department of Veterans Affairs helps the State Medicaid program by working to move veterans who are receiving Medicaid support for long-term care over to VA.
The District of Columbia, and Maryland have a unique interstate MOU, which allows them read-only access to each other’s eligibility systems. This enables them to check information on the interstate match file without going through State contacts. The District of Columbia and Virginia are in the process of setting up a similar agreement.

For the most part, State interagency collaborations have worked well, and few States reported problems with established agreements. Colorado reported issues with collaboration between the State and county-level offices, but this issue is indicative of larger problems Colorado has faced in trying to implement PARIS in a system of county-administered benefits programs, where counties have a significant amount of autonomy.

2. Advantages of Organizational Location

In almost every case, States felt that they had good reason for choosing to locate PARIS where they did. Oklahoma and Florida are among a small group of States that discussed alternative locations for PARIS, but in most instances, the advantages offered by a particular agency made it a clear choice to coordinate PARIS activities. These advantages include:

Ability to mobilize resources. One of the key issues in being able to conduct PARIS activities is the ability of the lead PARIS agency to mobilize the resources needed to conduct all of the PARIS activities. This often means coordinating with multiple programs or local offices that have significant autonomy. Both Maryland and Oklahoma reported that the organizational locations of their PARIS programs enable them to effectively direct both human and financial resources to the project activities.
Contribution to the ongoing work of the agency. Many agencies, including the District of Columbia’s DHS and Florida’s Department of Children and Families have an ongoing responsibility for conducting program matches other than PARIS to determine benefits eligibility. By incorporating PARIS into the office that administers these other match programs, PARIS gives these agencies another data matching tool. It also allows for sharing of staff across multiple activities related to data matching and program integrity.
Easily integrated. Several agencies attributed their successful PARIS implementation to the fact that the tasks associated with the program were easily integrated into their normal everyday workflow. For example, by considering PARIS one of many investigative tools, a member of the Oklahoma staff that follow-up on PARIS matches noted that most investigators in the State do not even distinguish between PARIS cases and other cases referred for investigation.
Access to Data. Some States find it useful to house PARIS within the agency that also controls the data systems for eligibility determination. For example, Tennessee administers its PARIS program out of the IT office in the State Department of Human Services, which controls access to benefits data in the State. This allows for better coordination of follow-up activities and limits the number of data entry activities needed to close or amend a record.
Motivated staff. Though not a substitute for workflow integration or available resources, a motivated PARIS coordinator and staff in some cases determined the organizational location of the PARIS program. In Oregon, for instance, where no agency was interested in taking on PARIS, the principal advantage of placing the program in the Financial Recoveries Office was that the coordinator worked in the office and had a strong desire to see PARIS become operational.

3. Disadvantages of Organizational Location

One of the principal disadvantages to any organizational location involves the limitations States place on the use of PARIS data. As discussed in Chapter I, there are many potential uses of PARIS data, including detecting and preventing fraud, coordinating benefits between Medicaid and Federal insurance systems, and removing inactive cases from a State’s benefit system. However, the ways in which States actually use PARIS data often reflect the specific priorities of the implementing agency rather than the full range of possible uses for the data. For instance, States that place PARIS in an audits and investigations division tend to emphasize using PARIS data for fraud detection and prevention and not for coordination of benefits or case file “cleaning.” In States where Medicaid plays a significant role in the program, PARIS data tends to be used for resolving residency issues and coordinating benefits, rather than for fraud prevention and recovery. Washington State is unique among visited States, in that the State’s program staff uses PARIS for addressing fraud as well as coordinating benefits and cleaning case files.

The PARIS program in one other State illustrated the disadvantage of placing PARIS in an office with a narrowly focused mission. Officials in this State knew of the many possibilities for using PARIS data, but reported that the narrow scope of their office (Financial Recoveries) made it difficult to bring people together to discuss expanded new uses for PARIS data. As a result, most of the activity of PARIS is limited to Food Stamp and TANF recoveries and reduction of duplicate case files. However, officials in this State have opened up discussions with Medicaid officials about the potential for including Medicaid data in the State’s PARIS match, and incorporate a follow-up system.

While there is no one organizational location that guaranteed successful implementation, programs with strong central coordination tend to fare better than States with decentralized models, under which significant PARIS responsibilities are divided among two or more agencies. Two States, however, reported facing a unique set of challenges in this regard.

Low compliance. One State’s program is coordinated out of the Department of Human Services, but major responsibilities are shared by the State’s Medicaid division and a separate Systems Management Division. The fact that PARIS occurs at the junction of three agencies hinders decision-making and makes it difficult for the PARIS coordinator to get a clear mandate from all three agencies. The inability of the PARIS coordinator to get the participating agencies to agree on PARIS procedures has resulted in the PARIS coordinator leaving many implementation decisions to local offices. Additionally, because of county autonomy, the coordinator has been unable to mandate county participation in the PARIS program and some counties have thus far declined to take part.
Inefficiency. The decentralization of one State’s PARIS program is due in large part to a statewide reorganization that impacted the division in which PARIS was located. The division that had responsibility for PARIS is still responsible for submitting the match file to the DMDC, but TANF and Food Stamps matches are now sent to a different office for follow-up. This change which has created inefficiencies in the State program. Officials interviewed indicated that investigators spend a lot of time conducting follow-up on cases that could easily be addressed by local case workers, such as unreported moves and other types of non-fraudulent cases. Often, when investigators ultimately refer such cases back to local offices for correction, caseworkers report already knowing about the client’s status. This State is switching to a new integrated eligibility system that officials hope will remove these inefficiencies by sending PARIS-related alerts directly to caseworkers.

In some cases, even States with central coordination of their PARIS programs deal with organizational obstacles related to communication and workflow between the offices responsible for culling and submitting PARIS data and the offices responsible for conducting follow-up activities. One State’s PARIS coordinator is responsible for submitting and distributing PARIS files, but does not receive any information on follow-up activities undertaken by the State medical authority and the OIG. This lack of communication limits the coordinator’s ability to effectively track case resolutions.

In one county-administered State, State-level employees may speak to their counterparts in other States to resolve residency issues, but only local county offices can close cases. State-level employees, therefore, must forward case information on to county caseworkers before cases can be closed, adding an extra step to the process of resolving cases flagged by PARIS.

B. ADEQUACY OF CURRENT STAFF LEVEL

PARIS participation requires two sets of staff: one that focuses on creation and distribution of the PARIS match files, and one that oversees the use of these data for follow-up activities. Most States have divisions or agencies that are responsible for the preparation and distribution of the PARIS files, while assignment of follow-up staff varies across States, often reflecting the structure of regional and local field offices within a State.

States consistently reported adequate staff levels for project coordination and the creation and distribution of PARIS files. In New York, the PARIS coordinator works about 50 percent on PARIS, but in most other States, program coordinators spend far less time on PARIS activities. The PARIS coordinator in Tennessee reported working only about 2 hours per month on PARIS, and the Coordinator in Illinois reported that he spent very little time on PARIS in the last 2 years. Washington State was unique in that the State coordinator reported spending most of his time on PARIS. Rhode Island was the only State in which the PARIS coordinator expressed a desire to spend more time on project coordination, noting that if the program had greater staff resources, he would take time to compile and analyze statistics on the data submitted to the DMDC, the data returned, and the resolution of flagged cases.

No States reported shortages among technical staff. Other than time spent on initial programming for the data submission, technical staff members across all States reported spending only a small percentage of their time working on PARIS, typically ranging from a few hours to a few days per quarter.

However, as noted earlier, most States reported not having enough staff resources for follow-up activities.

The District of Columbia only participates in two of four annual PARIS matches because there is not sufficient staff time to follow up on all PARIS matches within a single quarter. The program coordinator is trying to fill two vacancies for investigators.
Oklahoma’s OIG was short-staffed before taking on PARIS and has not received any additional staff to handle the increased workload the program generates.
While Utah is adequately staffed to conduct follow-up activity, they do not have the resources to track these activities to determine whether the eligibility workers are following up on all of the matches.
Oregon felt that local offices could not handle the additional workload and has, therefore, tried to coordinate follow-up using existing, State-level resources.
Florida faces significant challenges as a result of their modernization efforts. One of the key components of modernization is greater use of web-based applications for clients to apply for program benefits and the use of call centers for client assistance. This has resulted in significant reductions of the number of local offices and staff that used to be involved in resolving client issues, and a much greater level of decentralization of client eligibility functions. Because of the decentralization, the central office staff simply send out PARIS matches to local offices, and to not track or have any information on the level-of-effort made by local staff to conduct follow-up.
Missouri and Rhode Island are faced with inadequate follow-up staff and therefore only conduct follow-up on interstate matches.
New York reported that current staffing levels are tight but adequate. However, the PARIS coordinator noted that current staff levels would preclude any expansions to the program.

C. COMMUNICATION/INFORMATION DISSEMINATION

From data submission through follow-up, the number of staff involved in State PARIS programs necessitates that program coordinators and core staff develop approaches for communicating guidelines and procedures. States have used a combination of procedures to communicate how data from PARIS should be processed and used. Some States communicate through revising policy and procedure manuals while others issue memorandums to local offices to instruct on how to process PARIS matches or clarify Program policies related to PARIS. For example, staff in Maryland recently implemented an “action transmittal,” which supplements their policy and procedures manual. The action transmittal describes how to process a PARIS alert and was sent to State caseworkers so that actions taken in the field would be consistent statewide. Washington State has a similar process in place, which outlines policies and procedures for eligibility and follow-up. Oklahoma decided to support its staff by providing them with both a policies and procedures manual and also inviting 20 OIG agents responsible for PARIS follow-up to attend a one-day training on how to handle PARIS cases.

Other States that worked with a more decentralized approach had to strengthen communications between headquarters staff and local offices. For example, PARIS staff in Colorado took significant measures to communicate PARIS issues with county offices, hiring a fulltime county contact person to help facilitate communication between the counties and the State office. The nature of the contact person’s communication with the counties is twofold: he is responsible for talking the counties through technical issues and training them on PARIS procedures, but he also works to convince counties of the benefits of participating in PARIS.

With regard to disseminating PARIS match data, information regarding client eligibility changes is reported to local caseworkers through an alert system in most States. These alert systems are a primary source of communication between staff working at different levels of States’ benefits and eligibility systems. While Maryland, Oklahoma, Washington, and Utah have highly automated alert systems, Florida relies on a manual system in which information on PARIS matches is faxed between offices. Illinois’s information dissemination technique utilizes the management structure at local DHS offices: State-level staff sends PARIS matches to local Financial Recovery Coordinators who are responsible for assigning the cases to local caseworkers, setting deadlines, and monitoring progress.

Information dissemination strategies are discussed in greater detail in Chapter V.

IV. Submitting and Matching Data in PARIS

This section analyzes State responses to questions regarding the submission and matching of data in PARIS. One of the key issues that PARIS officials have been concerned about is the accuracy of the data so as not to produce false positives. A false positive occurs when errors are made in entering the social security numbers into the client record. These errors can be the result of transposition of numbers, incomplete entry of numbers, or the use of placeholder numbers (e.g. 9999) because the client does not have their social security number at the time of application. In the early years of PARIS, some officials reported that false positives were a frustrating issue that required significant amounts of time to resolve. However, with the increased use of verified social security numbers, combined with more sophisticated filtering systems, none of the States surveyed reported any major problems with false positives.

Since the submission of high-quality data is essential to ensuring accurate and appropriate matches, this chapter begins with a discussion of the use and evolution of criteria States use to select appropriate data and the structure of the databases that house the data. The chapter also discusses the methods States use to ensure that data are checked for accuracy and submitted in a timely fashion. Finally, the chapter presents an overview of the extent to which States have been able to successfully submit data to PARIS for matching, including a discussion of the major challenges that have hindered this process.

A. IDENTIFYING DATA ELEMENTS

The large majority of States reported that they submit all cases that are active during the period of the PARIS match. However, some States have narrowed down the number of SSNs they submit by setting time limits for when a case is determined to be eligible for PARIS submission. For example, Oklahoma indicated they submit SSNs associated with household members who are receiving a TANF, Medicaid, Food Stamp, or Aged, Blind, and Disabled (ABD) benefit on the first day of the submission month. Rhode Island and Missouri employ similar strategies to reduce the number of SSNs submitted to the DMDC for the PARIS match. Rhode Island’s system allows the State to submit only those SSNs associated with an active benefit on the day data is submitted to the DMDC. Missouri submits only SSNs associated with an active benefit on the last day of the month preceding the data submission. In contrast, Utah considers active SSNs to be those associated with clients that receive public benefits anytime within the month of the match date.

Three States reported that they also submit inactive or invalid SSNs to PARIS:

Utah sometimes intentionally submitted known invalid SSNs to increase the likelihood of finding matches in PARIS. In these instances, eligibility workers had the discretion in deciding whether to drop invalid SSNs prior to submission.
Colorado submits all SSNs, but marks invalid or non-verified SSNs as such in the files.
New York does not validate SSNs and submits any record with a 9-digit number in the SSN field. However, this practice has been discouraged by ACF, and will likely be changed in the near future.

ACF has recognized that this is an important issue, and is in the process of working with other Federal agencies to resolve any issues related to verification of social security numbers.

1. Changes to SSN Submission Criteria

Most States reported that SSN submission criteria have not substantially changed over time. Utah did note that the same criteria is used, but all active SSNs both new and old are now submitted as opposed to just the new cases. Two States reported significant changes to their SSN submission criteria.

Oklahoma previously restricted submission to the SSNs of direct payees of public benefits programs. State officials decided to expand submission to include SSNs of all household members associated with a payee receiving public benefits because many times these additional members received part of the benefit assigned to the payee. This change resulted in a 250 percent increase in the number of records submitted. Another major change that Oklahoma made was to narrow the timeframe for determining the validity of SSNs. The previous time period considered active SSNs to be associated with household members’ public benefits anytime within 3 months of the match data. The new, stricter criteria has reduced false positives, increased the timeliness and accuracy of submitted data, and made data more comparable to those of other States.
Tennessee used to submit all SSNs on file, including inactive numbers and “deemers” (non-biological family members whose income might affect eligibility). Under guidance from the ACF, Tennessee discontinued the submission of these two types of SSNs and now submits only active, valid SSNs.

2. Data Submitted in Addition to SSNs

States were asked to identify other types of data that they currently submit to PARIS. Some States reported that, in addition to SSNs, they currently submit the termination dates of public benefits for each case. Some States also reported submitting data on additional variables:

The District of Columbia also submits data on clients’ names and dates of birth.
Maryland also submits clients’ names, but recently has decided to submit and match data based solely on SSNs to allow the flagging of individuals who may have changed names over time.

B. DATA PREPARATION AND SUBMISSION

The preparation of data for submission is a relatively straight-forward process. Most States maintain an integrated eligibility database that allows data processing officials to easily prepare and submit the PARIS files. A few States maintain a separate file for Medicaid clients that do not receive other services, and in these States the “Medicaid Only” must be added to the general public assistance file. One State facing this circumstance noted that Medicaid data had not yet been fully integrated into their integrated eligibility system, and was submitted as an add-on to the integrated Food Stamp and TANF files. Another State noted that because Medicaid only files are maintained in another department, they are not integrated at all at this point. None of these States reported using any specific edits or filters to check for data accuracy prior to creating the final database for submission.

C. CHALLENGES TO DATA SUBMISSION

Most States did not report any ongoing problems in submitting data to PARIS and did not report having missed data submission deadlines. Rhode Island and Missouri both reported minor problems in their initial attempts at data submission but these problems were addressed and have not resurfaced.

A few States described some minor problems with the data transmission process:

The District of Columbia noted that their first submission of the data could not be accepted by DMDC because a file extension was accidentally left off. However this issue has since been resolved and all subsequent data submissions have been successful.
Oklahoma indicated that, initially, it required upwards of 12 hours to transfer data to PARIS using their Direct Connect connection. DMDC suggested switching to a Transmission Control Protocol/Internet Protocol (TCP/IP), which only resulted in a slight reduction in transfer time. Oklahoma is planning to begin transferring files by virtual private network (VPN), which should reduce the transmission time to just 30 minutes. Oklahoma representatives expressed concern about the 20–30 percent failure rate associated with VPN transfers, but felt that considerable time would still be saved even with a couple of failed attempts during the 2-week window for data submission. Oklahoma also reported that they missed a submission deadline by a day as a result of internal miscommunication. Ultimately, DMDC staff was able to connect to the central, integrated system and transfer the necessary data.
Tennessee reported occasional unspecified problems with data transfer.

V. Follow-up on Matched Data

In this section, we discuss the diverse approaches States are using to follow up on PARIS matches. For purposes of this study, “follow-up” is defined as any activities on the part of the State or local office to resolve the PARIS match (in the case of the interstate match), or to utilize data from the VA and Federal matches to either verify income or coordinate health benefits. As was noted earlier in this report, there are four primary uses of PARIS data for follow-up activities, and States have taken diverse approaches to both the scope of follow-up activities and the process by which data matches are finally resolved .

A. STATES APPROACH TO CONDUCTING FOLLOW-UP

Follow-up activities can be conducted at the State level, through local offices, or a combination of both. Of all the PARIS States reporting location of follow-up activities, 44 percent conduct their PARIS follow-up activities at the State level. Thirty-six percent reported that PARIS follow-up activities occurred locally, while 19 percent use a combination of State and local staff. Figure 3 shows the States reporting the location of follow-up activity.⁷

Figure 3. Map of States Reporting Follow-up Activity Locations D

States that use a combination of State and local follow-up generally follow a pattern of having local offices verifying the income amounts presented in the PARIS file with what is in the local case record, and then conducting follow-up activities to determine, in the case of the interstate match, the current State of residence of the person matched. State-level follow-up activities almost always involve conducting more detailed investigations to determine if fraud has taken place. Additionally, State-level staff often are responsible for conducting recovery activities if it is determined that an overpayment was made and the client owes the State money.

As was noted earlier, the interstate match is used more often than either the VA or Federal match. States find that closing case files where duplicate enrollments have occurred across States is an effective and financially beneficial use of PARIS data. Additionally, the interstate match is the least complicated of the three matches. If a client shows up in two or more States, it is a relatively simple process to determine where he or she resides, and appropriately close the case.

On the other hand, the VA and Federal match files are perceived as more difficult to manage. This is because the matching follow-up requires an individual to compare reported income in the VA or Federal file with income that is reported by the client. However, some States have noted that identifying sources of income from the VA and Federal file is not a simple matter of comparing total income. For example, some veterans may be receiving payments from sources that should not be counted as income for purposes of determining eligibility. This income will show up on the PARIS match file, but it may not be clear to the caseworker conducting the follow-up activity that the source of income is exempt.

In States where follow-up on the VA and Federal matches are most successful, the States have developed tools by which the data from the PARIS files is moved into a simpler format for ease of reading by a caseworker. For example, in the State of Washington, they have developed an Excel spreadsheet that moves the data from the VA and Federal file into a format that allows local caseworkers to identify sources of income by type (allowed or exempt), and then make an easier comparison with reported income.

Another issue noted by officials is that VA payments are often reported by clients, and thus are known to caseworkers prior to receiving any PARIS match data. Local staff may feel that conducting follow-up on PARIS VA match data can be somewhat redundant and will result in fewer financial benefits if the office has a good record of capturing income data from Veterans.

One other issue that seems to be a problem in a few States is that PARIS VA data does not meet State standards as being “verified upon receipt”. While ACF considers VA data as being from an original Federal source, and verified upon receipt, a few States have policies in place that require further verification. This means that local offices must contact someone at the VA to verify the information in the file. However, finding the appropriate person to contact that will provide verification of the VA data has been problematic in these States, and often results in local offices not being willing to make the effort to conduct the follow-up activity. ACF will be working with these States in the future to determine if the State’s verification policy can be changed.

Of interest to this study, it was noted that of the 14 States included in the study, all conduct follow-up on the interstate file, ten conduct follow-up on the VA file, and nine conduct follow-up on the Federal file. However, within the States conducting follow-up on VA and Federal files, some expressed frustration on how well they are able to use the information provided. Two States noted, for example, that their attempts to follow up on VA matches were frustrating and slow, and that they may discontinue their efforts in the future.

B. PERCENT OF MATCHES RECEIVED BACK FROM DMDC

One of the questions we attempted to answer was related to the percent of matches that a State receives from the total number of submissions. Specifically, we were interested to see if there is any significant variance between States in the percentage of submissions that actually result in a match. In general, there is substantial variance in the number of matches or “hits” received in each State. This variance is most likely related to specific characteristics of the State that may have an effect on either the size of their transient population or the number of States with which they frequently match. For example, the District of Columbia’s large percentage of matches is likely the result of the significant portion of its population which frequently moves between the District of Columbia, Virginia, and Maryland. Florida has a large number of matches compared with other States, most likely because it is a common destination for retirees who choose to live in Florida for the winter and in another State with a colder climate during the summer.

Table 4 below presents data collected from States that had information on the number or percent of matches received from DMDC.

Table 4. Percent of Matches Received by States Included in This Report, August 2006 Interstate Match

STATE	NUMBER OF MATCHES SUBMITTED	NUMBER OF MATCHES RETURNED	PERCENT OF MATCHES TO SUBMISSIONS
Colorado	486,643	1,403	0.29%
District of Columbia	86,929	3,483	4.01%
Florida	1,398,988	21,815	1.56%
Illinois	514,204	7,177	1.40%
Maryland	619,646	11,841	1.91%
Missouri	977,816	9,291	0.95%
New York	3,867,657	40,312	1.04%
Oklahoma	434,952	4,250	0.98%
Oregon	509,566	4,250	0.83%
Rhode Island	183,702	2,399	1.31%
Tennessee	1,330,662	6,344	1.98%
Utah	270,166	1,929	0.71%
Virginia	725,043	12,784	1.76%
Washington	1,139,154	7,718	0.68%

C. MULTIPLE STATE MATCHES AND GEOGRAPHIC REACH

Of interest to the study was the number of States for which the interstate matches occurred for any individual State and the geographic spread of matches. It is often thought that a State would most likely match an adjacent State. However, when data from the February 2007⁸ interstate match were examined, this did not always prove to be the case. Additionally, information was examined to determine if SSNs matched in more than two States. Data from the report indicated:

Some States had a surprisingly high number of matches with States that were neither bordering nor nearby. For example, Arizona, which had submitted its first match in February, had more matches with New York and Indiana than with any State in the western United States.
New York had almost as many matches with Florida as it did with Pennsylvania, an adjacent State.
Florida has its second largest number of matches with Tennessee.
Almost all States had at least one match with one other State that submitted a February interstate match. This means that almost all States had matches with 31 other States.
All States had some SSNs that matched in two other States (other than their own). This ranged from a low of seven matches to a high of 113 matches.
Seventeen of the 34 States had at least one SSN that matched in three other States.

D. PROPORTION OF MATCHES FOR WHICH FOLLOW-UP IS CONDUCTED

In addition to asking about the percent of matches each State receives from DMDC, we were also interested in determining what proportion of the matched cases receive some sort of follow-up. Again, responses varied substantially from one State to another. Some States conduct follow-up on every match they receive, while others are only conducting follow-up on one or two of the three possible matched files (e.g. interstate, VA, Federal).

Among the sample of States that are actively involved in follow-up, few were able to tell us what proportion of their matched cases were resolved. This is largely due to the fact that in most States, either “alerts” are not specifically tagged as being PARIS related, or there is no requirement in the State to report back on PARIS matches, even if they are tagged as such. In States where workers do receive PARIS-specific alerts, the alert files are rarely reconciled. Thus, once the alerts are sent to the workers, it is the responsibility of the local offices to conduct whatever follow-up is necessary, but they are not held accountable for identifying which case resolutions are specific to PARIS. In most cases, there is no monitoring to ensure that cases which need to be closed are actually closed or that benefits levels adjusted. For example, one State reported that while their workers do receive PARIS-related alerts, the workers only have a short period of time in which to resolve all of these cases, which are added in to other assigned alerts. Officials reported that it is not uncommon for workers who cannot complete the follow-up within the required timeline to prioritize the important alerts and simply file away lower-priority alerts without having worked them.

However, a few States were able to provide data on the proportion of matched cases that not only received follow-up, but eventually were referred for further administrative or criminal action. In Utah, about 25 percent of matched cases were referred for further administrative or criminal action, while only 1 percent of cases in Washington State received some sort of similar follow-up. Illinois referred 13 percent of matched cases for administrative action, but did not initiate criminal proceedings against any clients.

Because of their participation in the mentor/grantee program, Oklahoma tracked its follow-up activity for 2006. Data in Table 5 represent detailed information from Oklahoma on how all of their matched cases from a single match are handled.

Table 5. Oklahoma Match Follow-up Activities

ACTION	NUMBER OF CASES	PERCENT OF CASES
Referred to county Family Support Services Division for correction	1,047	8.3%
Closed due to insufficient evidence to support fraud	136	1.1%
Closed and referred to another State	130	1.1%
Closed due to spurious match	117	0.9%
Referred for investigation in Oklahoma	42	0.3%
TOTAL number of cases closed or referred	1,472	11.7%
TOTAL number of matched cases worked	12,517

E. FOLLOW-UP AFTER RECEIVING MATCHED DATA FROM DMDC

The process by which States follow up on PARIS matches varies greatly by State. Some examples of this diversity in approach include:

In Maryland, follow-up is conducted currently at the State level, but implementation of changes to files or notification to clients regarding closing of cases is handled at the local level. In addition depending on whether the match is Food Stamp/TANF, or Medicaid, the match is allocated to either the Department of Human Resources, OIG, or the Maryland Department of Health and Mental Hygiene, respectively.
In Oklahoma, need for follow-up on a match is determined at the State-level OIG. If the match is potentially fraudulent, follow-up is carried out at the regional OIG level; if benefits coordination is required, the match is sent to a county office.
In Missouri and Tennessee, follow-up is conducted entirely at the State level. But while Missouri’s follow-up is conducted entirely within one office, different offices at Tennessee’s State level conduct follow-up activities for matches from different programs. Food Stamps, TANF, and Medicare matches are handled by Tennessee’s Department of Health Services’ Investigations unit, while matches from Tennessee’s universal health care pilot system are handled by the OIG of their finance division.
In Colorado, Rhode Island, New York, and Utah, the State receives data from DMDC but follow-up occurs at the county or regional level.
Illinois similarly conducts follow-up at the local level, but simultaneously sends reports generated from their match file to their Bureau of Collections and to an SSA programs unit to check for SSN issues. Additionally, Illinois has special procedures for processing matches from Cook County (containing the city of Chicago); these matches are sent to a program accuracy unit for verification.
Conversely, interview data show that Oregon currently is not very involved with conducting follow-up unless expressly requested by another State. Virginia plans to conduct follow-up activities in the future, and will likely do so on a local level, but has not yet developed follow-up policies and procedures. Florida sends any matched data received from DMDC directly out to the regional offices, but does not know how many offices conduct follow-up related to PARIS. The District of Columbia has only district-level follow-up, and Washington State parses out the type of follow-up needed by type of match.

1. Time Limitations on Conducting Follow-up

In most States, reporting case conclusion to an administrator is not required of individuals involved in follow-up activities. This is true across most of the States that were interviewed. In Maryland, if a case sent for investigation is not closed within 30 days, an explanation needs to be provided and the case closed, but the case can be re-opened at another time. Illinois has an automated system that generates and sends reports of the follow-up status to the field offices 45 days after the referral date. However, there are no repercussions for the case workers if the match is not resolved within those 45 days, and the information is never sent to the State level. Only New York and Rhode Island seemed to have firm time limit requirements for follow-up activities: 60 days for New York, and the last day of the month after the match is returned (4–6 weeks) in Rhode Island. While follow-up staff in Rhode Island have no problems meeting this deadline, New York must occasionally remind local staff of the deadlines.

Even though the time to case resolution is not tracked in many States, there often is an ideal length of time during which follow-up should take place, after the case has been assigned. This time period varies from 10 days in Utah to 90 days in the District of Columbia. In Washington, Tennessee, Missouri, and Oklahoma, no time was specified, and in Colorado 60 days is the stated time limit, but only for fraud investigations. Respondents from Utah and Colorado stated that, on average, interstate matches only take about 2–30 minutes and 30–45 minutes to resolve, respectively.

2. Communication

Cross-State communication issues appear to be a point of contention for several States. Workers involved in follow-up in some States have an established network of out-of-State contacts to call; these workers generally reported no major communication issues when working a match. However, workers in other States were not as positive. They reported that some States do not provide the number of an individual contact person on their PARIS match file, but instead use either an 800 number or a generic State eligibility hotline number, which makes it difficult to find someone who is familiar with the case. In addition, calls placed to generic phone numbers can result in a conversation in which the individual answering the phone has a strictly limited amount of time to answer questions.

In order to alleviate the difficulties they experience in trying to find an appropriate contact, some States would like to see an improved system for interstate communication. For example, an interactive and secured website where residency verifications of follow-up on interstate matches could be posted between States would be helpful. A centralized client locator system that would eliminate the need to contact local offices was also mentioned as a possible improvement.

F. USEFULNESS OF MATCHED DATA RECEIVED FROM DMDC

When asked about the usability of the matched data, most States felt that, overall, the files were fairly easy to use and that the data quality was fairly high. However, there were some specific issues cited by several States. These issues include:

States have the most difficulty understanding and working with the Federal file. Many of the States feel that the Federal file is confusing and can be difficult to understand in terms of identifying sources of income and the applicable time period for which the amounts apply. Additionally, the Federal file occasionally contains truncated dates of birth that do not match the format of the field (e.g., a date of birth with four digits). Five of the 14 States included in this study are not currently using the Federal file because of these issues.
States that conduct follow-up on the VA file found it less difficult to use than the Federal file, but also reported that it requires more time to work than an interstate match if there is a discrepancy between what is reported by VA and what the client reported. Because the VA file does not present data in a manner that is easily translated into a workable format, local caseworkers reported that they must often call individual clients to discuss the reported income, and try and determine the accuracy of the data reported by the VA and whether or not it is exempt from eligibility determination. For example, some States find it difficult to divide out the numerous payment categories into those that should be counted as income and those (such as aid and attendance payments) that should not. Even the clients can be confused about sources of income, and what categories various components of their total VA payment fall into.
Two of the States included in the study reported that they use the VA database to coordinate benefits between the VA and Medicaid. This approach usually involves identifying veterans that are enrolled in Medicaid, but may also qualify for healthcare or long term care benefits from the VA. Where these veterans are identified, these two States take action to ensure that VA is picking up the costs that are appropriate for it to cover, thus saving the States dollars in Medicaid payments.
The interstate match data are only as good as what is submitted by the other States. Poorly prepared match data in one State can result in another State wasting time and effort on follow-up. For example, some States do not always submit date of birth information, while other States provide transient open and close dates for clients who go on and off of benefits from month to month, rather than the dates clients actually began receiving benefits for the first time. At least one State continues to submit inactive files, which makes the information they provide frustrating and fairly useless to work with.
Some States report that contact information is very frequently missing, both for other States and for the Federal and VA agencies. Even when contact information is provided, it is not always accurate, or useful. For example, often telephone extensions may not be included, or the contact person might not be the appropriate person to follow up with, which makes tracking down the right person very difficult.
One State reported that even when it closes a TANF and Food Stamp case, the Medicaid eligibility continues (through transitional Medicaid services) until it is actually documented that the client is no longer eligible because he or she moved. This means that some Medicaid files, which should be inactive, show up in the match.

States offered the following suggestions for improving the usability of the data files, some of which have been addressed by the new format:

Fields that are deemed to be useless should be removed in order to make the file more user-friendly. There should be a periodic assessment of the data fields, and those that are not used or result in confusion should be identified and changed when field format is changed.
There is a need for improved contact information for each State and for the Federal and VA agencies. While is would be nice to have a consistent method by which States communicate, the diversity of State approaches makes this impossible. However, having a source of information for each State that provides follow-up staff with a reliable and accurate contact would go a long way to improving communications. In addition, ACF should examine how secured websites might be used for improving contact information.
One State thought that here should be a requirement in the PARIS agreement that States shall not submit closed cases. Several States complained about others that submit closed cases that remain on the State’s master file, but in an inactive status. This makes for more work on the part of those conducting follow-up, and can result in matches from States that submit closed cases being a lower priority for local staff to work. However, ACF noted that such a requirement in the agreement is not necessary, as this issue is being addressed through other means.
Submission requirements for some data, such as start and end dates for benefits, should be clarified so that all States are submitting the same information in the same format.
Technical assistance should be provided to better understand the Federal and VA files. Most helpful would be “how to” manuals that would help States make sense of the files and allow them to train follow-up staff. Additionally, tools developed by other States to make these files more “user friendly” should be identified and shared. At the very least, better descriptions of the values should be inputted into the data fields.

Four of the States have taken the initiative of making the data files more user-friendly. Both Colorado and Washington have developed a summary report or Excel file of the VA match to make it easier to use. Colorado’s VA match summary is intended to help counties get started, whereas Washington’s detailed Excel sheet is used for every VA match. New York has developed an Access database for the regional offices to use when reporting to the State on match results. Oklahoma also uses some type of report to assist with data summarizing; however, they did not have specific details on the report.

G. FILTERS

Data “filters” are specific criteria that a State uses to reduce the size of its matched data files. Filters can be a very effective tool in reducing the number of files for which follow-up must be conducted, prioritize follow-up activity, and eliminate files with incomplete or erroneous information. Filters also may help States focus their follow-up efforts on cases that stand a better chance of being resolved. For example, one State reported that they remove records from their matched data files that do not match exactly on SSN and two additional personal identifiers. This filter helps to screen out matches where it is very likely that the match was a result of a social security number being erroneously entered, rather than a match of a single individual.

The use of filters and the types of filters used varied across States. Six of the 14 States that were interviewed reported that they do not use filters at all. Table 6 summarizes the types of filters that are used by the remaining eight States.

Table 6. Types of Data Filters Used by State

STATE	TYPE OF FILTER
Oklahoma	Household member must be receiving a TANF, Medical, Food Stamp, or Aged, Blind or Disabled benefit on the first day of the submission month
	For each new match, previously matched cases are removed because they are assumed to have been worked
	Matches must have at least two demographic hits
	VA and Federal matches with less than 2 months of benefits received are removed from the file because these may have been related to Emergency Orders
Florida	Remove closed files and duplicate records from prior matches
Oregon	Filter out foster care kids because they are usually legitimate cases
Maryland	Matches already in the State’s integrated eligibility system are removed
Illinois	SSN checked to see if it was in the previous match file, and removed if so (to give field workers time to work their cases)
New York	SSN checked to make sure the match was not already resolved in a previous PARIS match
Missouri	Remove casers that have gone inactive since the match file was submitted
Rhode Island	Remove cases which have been closed in the short time since the match file was submitted
Rhode Island	Remove cases submitted by other States which are not currently active in Rhode Island

Only three States reported the percentage of matches their filters removed. In the May 2006 quarter, Florida had 24,000 matches before the filters and 18,000 remained after the filters. Therefore, the filters helped reduce their file by 25 percent. Rhode Island’s filters removed approximately 840 of their 1,200 matches (70 percent), while Maryland’s filters removed about 1,750 of its 3,500 matches (50 percent).

H. USE OF SPECIAL SUMMARY REPORTS

Maryland, Tennessee, and New York are the only States that produce and use special summary reports. Tennessee’s system generates a simple report whenever a match is not resolved within their time limit of 45 days. New York uses an automated Access database to generate reports on the progress of follow-up activities for each identified match. They also send Medicaid data reports directly to the Department of Health. Maryland uses an existing data system to create the following three reports:

A report of matches that did not generate an alert (e.g., VA matches where income had been previously known and reported)
A report of the disposition of cases that raised an alert.
A report of outstanding PARIS alerts, which are then given 20 days for action.

I. FOLLOW-UP PRIORITIES

In addition to asking about filters and reports, we also wanted to know if there were any matched cases that were deemed more important or were considered higher priority in any of the States. Most States did not give their workers specific priorities related to the PARIS alerts, though some of the people we interviewed were unsure because this type of decision would be made at the county level where the cases are resolved.

The District of Columbia does set priorities for the order in which matched cases get worked. They give the highest priority to TANF matches, followed by Food Stamps and then Medicaid. There are no particular priorities within the programs because all of the cases get addressed. Tennessee also sets priorities for their workers, focusing on resolving Food Stamp cases before all others. Eligibility workers in Oklahoma are instructed to focus on cases with no reported VA or Federal income or benefits, as opposed to cases with reported income and benefits that differ from the data in their State file, unless the discrepancy is greater than $500. Finally, although Colorado stated that they do not prioritize matched cases, they did note that they look for names that they have seen before in previous matches because they are looking for repeat violators.

Four of the States interviewed did not work files where there was a match for clients who were enrolled in Medicaid only. ⁹ This is because Medicaid is located in a different department, and up to this point had not participated in conducting follow-up with the PARIS match. All four of the State officials interviewed felt that it would be important for Medicaid to become involved in PARIS matching and follow-up. Three of the four States indicated that discussions were either ongoing or would be started shortly to convince Medicaid officials to join in the PARIS project.

One State reported that because their State has implemented simplified Food Stamp Program reporting, that follow-up on food stamp income discrepancy cases is a lower priority. Simplified reporting allows a State to eliminate the requirement for clients to report changes in income (within limits) until their next certification period. If there are minor discrepancies in income reported, this State filters out those matches so follow-up is not conducted. However, data may be retained to be made available when the certification period for that client comes up.

VI. Costs of Implementing and Maintaining PARIS

One of the key objectives of this study is to create a standardized model that States can use for calculating the cost and benefits of PARIS participation. However, because of the diverse nature of PARIS participation, a single cost/benefit model will not work for PARIS. This is because States differ in both the scope and frequency of their PARIS participation (limiting both number and frequency of matches), and cost and benefit calculations will differ among States. States are not required to match their data every quarter, so some States will only match once or twice a year. States with more frequent matching will accrue benefits more frequently. Additionally, some States limit their matching to one or two of the three possible PARIS matches. Those States that match against all three files will likely have a greater potential savings than those that do not.

In this section, we discuss the proposed models for calculating costs and benefits. These models are designed to be able to calculate costs and benefits using data that are potentially available at the State level. However, in examining State-level costs and benefits, a number of factors must be considered related to the ability of States to populate cost models. These factors include:

Only four of the 14 States interviewed had methods by which they could track the State-level cost of PARIS. Most States view PARIS as one of many tools used to examine program integrity, and therefore do not track cost by each specific tool. In all but four of the States, the costs for PARIS are either incorporated into a general budget that covers multiple program integrity activities, or are aggregated to such a higher level within the State’s accounting system, so it is not possible to track individual costs for PARIS. However, this does not mean that data are not available with some extra effort on the part of the States to isolate PARIS costs. For example, eight of the 10 States that do not track PARIS costs indicated that they could make estimates for State-level costs related to the amount of time staff devote to PARIS activities. By estimating the amount of time spent on PARIS, these States could create a cost estimate by applying that percentage to the salaries and benefits of staff working on the project.
Where follow-up activities are handled at the State level, there is more of an opportunity to track the costs. Two of the 14 States interviewed conducted follow-up at the State level, and both were able to either track costs for these activities or create reasonable estimates of costs. Three States conducted follow-up at both State and local levels, and two of the three could track or estimate State-level follow-up costs.
Nine of the States interviewed conducted follow-up at the local level. Local follow-up costs were tracked in only two of these States. These costs were tracked through a PARIS reconciliation system, which asked local caseworkers to document both the amount of time they spent on PARIS as well as the resolution of each PARIS case. One of the largest potential costs of PARIS is the amount of time follow-up staff must spend working PARIS files. Most States did not even know how many of the PARIS matches actually were worked and resolved. Local office staff interviewed noted that they do not track costs by PARIS or any other source of information, nor do most States require any type of reconciliation reporting of PARIS results. However, as was true with State-level costs, an estimation approach could to be applied to calculating PARIS costs. This would require a significant amount of tracking at the local level, for which most States indicated that they have insufficient resources to conduct.
A lack of reconciliation processes in most States makes it difficult to track PARIS results. Thus, it is difficult not only to calculate costs, but also to calculate benefits, as a result of a PARIS action. While benefit models can be developed to allow States to make reasonable estimates of benefits, populating the models require that some system for tracking results be in place. This can be accomplished by using special studies to sample and quantify PARIS follow-up results so that appropriate cost savings could be applied to the results.

Given these difficulties, the study team approached the development of cost/benefit models from the standpoint of creating reasonable models based on the ability of a State to collect the type of data necessary for the calculations. In the next sections, we describe the proposed models for calculating costs and benefits, explain the limitations of the models, and provide examples of how existing PARIS States can populate the models.

A. DEVELOPING COST MODELS

To reflect a true representation of the total cost of PARIS, cost models must be grouped into three areas. First, cost of PARIS start-up should be examined. Second, the ongoing cost of creating the master file, submitting the file to DMDC, and preparing and filtering the file once it is returned should be included. This second area also should include ongoing maintenance costs, as well as administrative costs related to the project. A third area includes the cost of staff time devoted to follow-up activities, such as conducting coordination of benefits or trying to capture overpayment recoveries.

1. Start-up Cost

It is important for States that are considering joining PARIS to understand the costs involved to program and implement the PARIS data matching protocols. Start-up costs encompass those costs related to the programming of the initial PARIS files to match the required PARIS format. Additional start-up expenses also might include the costs of incorporating follow-up activities into policy and procedure manuals, training staff on how to conduct follow-up, and setting up tracking systems for reconciling PARIS results.

2. Ongoing State-level PARIS Costs

All of the States interviewed reported that the ongoing cost of creating PARIS files, submitting the data to DMDC, and preparing and filtering the results are relatively small. One State, which conducts an extensive coordination of benefits effort with the VA and TRICARE reported more significant costs for these activities. However, for most States, ongoing costs of PARIS operations are not commonly tracked, as most States incorporate PARIS activities into a more general, higher-level compliance activity. None of the States interviewed had a tracking system that would allow one to track costs specifically to PARIS activities. However, some States were able to provide estimates of the amount of time needed to create the files, submit the match, and filter and distribute the results.

3. Cost of Local Follow-up

As was noted earlier, two of the States interviewed tracked local costs. However, only one of these States conducts follow-up on all three matches. When examining costs for local follow-up in the State where all three matches are worked, we noticed that costs do vary by type of match. In general, it took local staff longer to work an average Federal match than it did for them to work an interstate or VA match, thus creating a variance in the average cost per case worked between the three matches. Matches that were tracked in the second State were only interstate matches, and thus there was more consistency in the average cost per case work. The methods use by both States, however, can provide a framework for calculating local costs, even if they were calculated using a sampling method. This could be done by devoting resources to conducting a time and results study based on a representative sample of PARIS matches. Once an average cost per match is identified, this average can be applied to all PARIS matches.

4. Coordination of Benefits and Cost Recoveries

Two of the States have special projects by which they track PARIS activities related to coordination of benefits and establishment of financial recoveries. Because these costs have their own coding as part of a special project, the PARIS coordinators have access to accounting records to rack the costs from month to month. This allows the PARIS coordinators to produce reports to justify the ongoing funding of these projects.

B. CREATING MODELS FOR CALCULATING BENEFITS

Ideally, a standardized benefit calculation model would not only allow an individual State to calculate benefits for their own purposes, but would allow ACF to conduct comparisons across States. Additionally, they would allow ACF to make national estimates of savings to document the success of PARIS. While the findings of this study support the idea that a consistent methodology can be implemented, there are some challenges that will need to be addressed as the implementation proceeds.

For example, the diversity in approaches used by States to implement PARIS makes it extremely difficult to develop a single model for calculating benefits that can compare one State with another. Some States only work all three matches, while other States only work matches for one or two of the programs included in PARIS. Some only work interstate matches and close cases based upon unreported moves, while others use PARIS as an income verification system and use the data to verify eligibility. The frequency of State submissions for match also impact the total savings that can be accrued from PARIS.

1. Factors Influencing a Standardized Approach to State Benefit Calculation

When examining factors that influence the ability of a State to calculate benefits, a number of issues must be considered by each State. The VA and Federal matches match income data for clients within a State, and thus are more like most other State matches that verify income. For example, States often use data matches with other income sources, such as “the Work Number” or the National Database of New Hires to verify that income in not going unreported. Therefore, States can use these data to either reduce the levels of benefit or to terminate a client because their income is over the eligible level. However, documenting and linking the change in benefit or the termination to a PARIS match is not generally done by States, thus complicating the State’s ability to calculate a benefit associated with PARIS.

With regard to the interstate match, the diversity of State approaches to conducting follow-up may affect the ability of a State to populate the models to produce consistent results across States. Factors related to the amount of time State and local staff spend conducting follow-up activities, the ability of States to document benefit levels reduced or eliminated, and the actions other States take to resolve duplications all have an impact on benefit calculation. These factors are discussed in more detail below:

Level of effort. One of the key issues related to the State’s ability to utilize a standard method of calculating benefits is the ability of an individual State to devote resources to both conducting follow-up activities and documenting results. Half of the States in the sample indicated that resources for conducting follow-up activity are limited, and not all matches can be worked. Additionally, working PARIS matches compete with other priorities related to program integrity,¹⁰ and the percentage of matches worked by an individual State can vary from quarter to quarter. Because of these limitations, it is not possible to create a model that identifies savings for a particular quarter, and then applies these savings to future PARIS match data.
Frequency of match participation. Another issue relates to the frequency of participation in the four matches offered each year. As was noted earlier, some States match quarterly, others semi-annually, and others annually. As a result, States that match quarterly and have a high match rate with States that submit files less frequently find that in some quarters the matches on which follow-up is to be conducted are fewer than others. As a result, it is difficult to extrapolate data from a single quarter, as it will vary depending on the frequency of matches submitted by other States.
Difficulty in resolving matches. One other issue noted by some States is the ability of two individual States to work together to resolve interstate matches. As was discussed previously, some States experience communication problems related to identifying proper contacts within a matched State or being able to resolve matches in a timely manner. As a result, matches with “problem States” may be less of a priority for resolution, and therefore not necessarily worked or resolved.
Calculating benefit usage. In some States determining the actual dollar value of the benefit, such as the value of food stamps received or Medicaid capitation benefits paid, may be compounded by whether or not the benefits were actually used. When a participant moves and does not report the move, they are still issued benefits. However, because the participant has moved, they may never redeem the benefits (in the case of Food Stamps or TANF). As a result, the benefits remain on the States’ books until a point in the future when the time limit for cashing the benefits occurs, and the account is closed. States differ on how they count the value of unredeemed benefits. Some States will terminate a client as an unreported move, and will use the value of the benefits from their issuance records to record a savings, even if the client never actually used the benefits. In contrast, other States will only count the savings if a client used the benefits.

A number of States attempt to calculate cost savings due to PARIS. Of the 14 States included in this study, five had prepared cost savings estimates. Four of the five States prepared cost savings estimates based upon the results of the interstate match, three had built estimates based on savings from income verification using the VA or Federal match, and one had prepared an estimate of savings from coordination of benefits between VA, TRICARE, and Medicaid. These estimates were reviewed by the project team to assess: (1) the methods used for calculating benefit levels of clients terminated; (2) the length of time associated with the benefit reduction or elimination; and (3) how well these estimates matched up with the methodology proposed by this study. Table 7 displays the reported savings from the States.

Table 7. Reported Savings by States Included in the PARIS Study Sample

STATE	SAVINGS	STATE FISCAL YEAR FOR WHICH SAVINGS WERE IDENTIFIED	MATCHES INCLUDED IN SAVINGS ESTIMATE
Colorado	$62,611	2006	All Matches
Oklahoma	$779,349	2006	All Matches
New York	$45,644,400	2006	Interstate
Rhode Island	$631,351	2005	Interstate
Washington	$788,868	2006	VA

2. Examining Current Efforts by States to Calculate Benefits

While ACF encourages States to identify and report benefits of PARIS participation, there is no requirement for a State to do so. If a State does calculate benefits of PARIS participation, they often do so for a number of reasons. Most often, States calculate benefits of PARIS to justify and support their use of resources for participating in PARIS, support funding of PARIS activities through savings transfers from programs (Medicaid, TANF, Food Stamps) to PARIS, documenting and enforcing client recoveries, or to maintain an ongoing estimate of savings to present to State legislative committees, the Governor’s office, or other oversight groups.

Because each State calculates their cost savings based upon their own needs, there is some variance in approach to the factors we examined. Some of the approaches used by States include:

Of the States calculating cost savings from closed cases (interstate match) three of the four used estimates of average benefit level for each client category to determine the level of savings per case closed. For example, if the average Food Stamp benefit in the State was $150, that amount was applied to all closed Food Stamp cases. Only one of the four States determined the actual amount of benefit being received by the client whose case was closed, and summed that total into their benefit calculation formula.
Of the four States calculating benefits from case closure using the interstate match, three of the four annualized the benefit savings by multiplying the individual client monthly benefit (either average or actual) by twelve months. Only one State calculated benefit levels to the end of the client certification period, and then totaled up the savings for each client.
Of the three States that calculated savings based upon recoveries, two of the three calculated the benefit level by using the dollar amount that was established as a claim for client repayment. This meant that if a client owed $3000, the benefit calculation used the $3000 whether or not the client ever paid this amount. The rational for using this number was that there may be a long time between establishment of a recovery and collection, and by “getting the recovery on the books”, the State has established a dollar amount as a target for taking any and all action for recovery. For example, the recovery may take the form of attaching a tax refund, lottery winnings, or other amount that would normally be owed the client in the absence of the recovery claim. Only one of the three States used actual amounts recovered for their calculation.
Two States reported savings based upon reducing benefits because of income discrepancies identified using the VA or Federal file. Both of these States used the actual amount of the benefit reduction to calculate their benefits.
The State calculating benefits from coordination between Medicaid, VA and TRICARE used actual savings by calculating the amount of payment that Medicaid had been making and summing it through the end of the certification period.

All of these approaches have advantages and disadvantages. Because this study has the unique purpose of trying to determine the best method by which ACF can calculate the costs and benefits of PARIS, a number of factors needed to be examined that individual States did not take into consideration. In preparing the benefit models for the interstate match, the following factors were considered.

What methods best reflect the impact of PARIS on State-level benefits? One of the concerns regarding establishment of a State-level benefit methodology is the ability of States to link a PARIS activity with a consequence. In order to claim a benefit from PARIS, the State must be able to link the benefit reduction or termination to the PARIS match. As was noted earlier, most States do not reconcile their matches, so they do not know how many terminations or benefit reductions are attributed to PARIS. The model proposed for this study must identify a practical method to collect data on PARIS match results. This means that the study team had to balance the accuracy of having a system by which all matches are reconciled with a system that uses sampling and estimates to assess the impact of PARIS across a State. While the former is the most accurate method to calculate benefits, we believe that reasonable estimates can be made using sampling. By using a method that works in more States, the ability of ACF to estimate PARIS savings becomes more consistent and practical.
What benefit levels should be used to calculate savings? Again, to answer this question, one must balance what data are available in the most number of States, as compared to a few States that may have excellent data because of the sophistication of their data systems. Because the goal of ACF is to estimate PARIS savings in the most consistent manner across the largest number of States, the most practical approach should be used. This means that using average benefit levels by program and client classification¹¹ is likely to produce the best results across the largest number of States.
For how long should benefit calculations be made? Again, the most accurate method for calculating benefits would be to calculate how long the remaining certification period is for a client whose benefits have been terminated or reduced. However, only one of the 14 States included in the study had the capability to pull this information from their data system. We therefore believe that it is reasonable to calculate savings estimates for a 12-month time period.

This does not mean that States should stop calculating benefits to meet their own needs. However, for purpose of making national estimates of the cost and benefit of participating in PARIS, use of a consistent method will help to resolve confusion between different approaches, eliminate improper comparisons between States, and provide States that have not calculated costs and benefits with a model that can be used.

Creating a standardized method for calculating cost and benefits of PARIS requires some flexibility in approach. By developing models for specific activities rather than an overall single model, the applicable model(s) can then be applied by each State individually to allow for an approach to suit their own customized implementation of PARIS. The models can then be adjusted to meet frequency and scope issues within an individual State, and thus create a cost-benefit level for each State. As a result, four separate potential benefit models have been created:

A model that calculates savings related to closing cases from unreported moves on the part of clients
A model for calculating savings related to the recovery of overpayments
A model for calculating savings as a result of benefit reduction as a result of clients not reporting all income
A model to calculate savings as a result of coordination of health benefits with Medicaid.

A description of each model, along with issues considered in developing these models, is discussed below.

a. Model for Cost Savings Related to Case Closure

Most State officials use the interstate match as a mechanism by which they can close cases of clients that no longer reside in their State. By terminating the case as a result of a PARIS match, the State avoids paying benefits to someone no longer eligible for services. In examining the potential savings as a result of closure of cases, a number of issues were examined. These include:

Determining what percentage of PARIS interstate matches result in a case closure. It must be remembered that many of the matches worked on the interstate file do not result in the closing of a case. If a local caseworker conducts a follow-up activity and determines that the client is properly residing in their State, then it is up to the other State to take an action, and the case usually remains open in the State of proper residence. Because costs are associated with working a file, even if a closure dose not take place, it is important to determine the extent to which clients continue benefits even if they are found on the PARIS match.
Determining the amount of benefits saved for Food Stamps and TANF. The assumption behind calculating savings related to benefit termination is that PARIS matching would catch an ineligible client who moved to another State but did not report the move much earlier than would normally have been done by waiting until re-certification. In an ideal benefit model, a State would be able to determine the actual dollar amount of the benefit level of every client terminated through the use of a PARIS match, multiply that amount by the number of months remaining on the certification period, and thus demonstrate savings through avoiding payments to the client.

However, while most State data systems have this type of information, no State tracks the information as it ties to a particular information source. Many States do not reconcile their PARIS matches, so State officials do not know how many terminations take place as a result of information provided through follow-up on the PARIS matches. This means that routine information on the benefits of PARIS matches may not be available.

Calculating Medicaid savings. Medicaid savings can fall into two categories: savings created from avoiding payments to physicians for services provided on a fee-for-service basis, and halting monthly capitation payments to health plans for States using managed care. For calculating benefits for managed care savings, a relatively straight-forward method of multiplying the monthly capitation rate by the months remaining in the certification period would produce the benefit calculation. For fee for service clients, it becomes more complicated. If a client has moved, the likelihood of a Medicaid fee-for-service payment being made is very low. However, because States must control Medicaid costs, and often limit enrollment based upon projections of costs, the fact that someone remains on the rolls means that an obligation is created that the State must consider as a potential payment. Therefore, the savings created by closing the case can be used to either add new clients to the program or reduce program obligations. The most reasonable method for calculating a benefit would seem to be using an average utilization rate for different categories of eligibility (pregnant women, children, adults, etc.) and then using billing information to create an average monthly cost for each group. The monthly cost would then be multiplied by the remaining certification period, and thus the savings estimated.
Creating appropriate time periods for calculating benefits. In the model discussed, the most accurate representation of savings would be to total the value of the benefits and multiply that value by the time remaining in the certification period. However, in examining data at the State level, it would be very difficult and costly for State officials to determine the remaining certification period of each individual terminated through a PARIS match. In the past, State officials have calculated benefits making the assumption that benefits would accrue for an entire year. While this is not likely, it may be the most reasonable approach to take in calculating the appropriate time period for which to accrue monthly savings. One could assume that in some cases, such as Medicaid certifications, benefits might actually accrue for an entire year. For TANF and Food Stamp benefits, making the assumption that benefits would have accrued for a year without PARIS intervention is more problematic, but is a reasonable alternative given the limitations of available data.

Therefore, we are proposing that the methodology described above be used to create models that aggregate data to the average benefit level for all clients, rather than using individual client benefit levels. Once an average benefits amount is calculated the State would use these averages to calculate benefits for clients that are terminated via a PARIS match. The basic formula would be:

Food Stamp/TANF average monthly benefit level multiplied by the number of PARIS terminations, followed by multiplying this number by 12 months worth of savings.
Medicaid Capitation Rate¹² multiplied by the number of Medicaid recipients terminated, with that sum multiplied by 12 months, equals savings for States with Medicaid managed care.
For States that use fee-for-service payments in their Medicaid program, one needs to combine the average Medicaid monthly reimbursement by category¹³ with the average cost based on unitization rates. This cost number is then multiplied by the number of Medicaid clients in each category terminated times 12 months equals savings.

b. Model for Calculating Benefits Related to Overpayments and Recoveries

Creating a model for calculating benefits for overpayment recovery would seem to be more straightforward, but it is limited by the way in which some States calculate overpayment savings. Overpayment recovery is based on two distinct processes: establishing the overpayment and collecting the overpayment. For most States, there is a clear number available for establishing the overpayment. However, collecting these overpayments can be problematic, particularly where a client has moved out of State. Four of the agencies interviewed for this project estimate that an overpayment could stay on the books for up to ten years prior to it being abandoned and written off.

In order to come up with a reasonable calculation of the benefit related to overpayment recoveries, we decided that using the potential recovery amount established when an overpayment is calculated would reflect the potential benefit of PARIS data being used for financial recovery. One could multiply the total dollar amount of the established recoveries by the percentage of recoveries actually made, but one would then have the problem of identifying the percentage that could be attributed to a PARIS match. Therefore, by using data when the initial recovery is established, the State has the potential to credit the recovery to a PARIS action.

c. Reducing Benefits as the Result of Improper Income Reporting

Perhaps the most complicated benefit to calculate is the amount of benefit reduction that occurs when income reported in the Federal and VA match is not accurately reported by the client. In each of the States visited, officials were asked if they had any information on the number of clients who had benefits reduced because of PARIS, and while several indicated that they believed there were benefit reductions, none could provide any data in this regard. In talking with three local offices, all reported adjustments of some sort had been made using PARIS data to verify income. However, none could provide any information on the number of PARIS matches that resulted in reductions, nor the amount of any reduction.

Reductions can take place in one of three ways. First, the client can be determined to be eligible for all programs for which they were originally certified, but the level of benefit should be reduced because of unreported income. A second reduction can take place where the client is determined to be ineligible for one or more programs, but eligible for another. As an example, a client may be found to be ineligible for TANF, but eligible to remain on Food Stamps and Medicaid at the same benefit level. Finally, a client may be found to be ineligible for a program such as TANF, and still be eligible for Food Stamps with a reduced benefit level.

To calculate totals for benefit reductions, we are recommending that a special study be conducted with a random sample of PARIS matches resulting in benefit reductions. To conduct this type of study, we recommend that a State ask local offices to identify the SSNs of clients from a single VA and Federal matches where any action affecting benefits was taken. The State would then draw a random sample of cases and conduct an analysis of benefit level changes for the sample. Once an average benefit level change is determined, that dollar amount would be multiplied by the total number of clients composing all the clients for that match that had a benefit reduction, and that result would be multiplied by 12 months.

d. Coordination of Benefits with Medicaid

Only two of the States interviewed are coordinating benefits between Federal and VA matches and Medicaid. Calculating coordination of benefits with Medicaid basically involves determining the extent to which benefits being paid for Medicaid clients should be paid by another source. The primary example of benefit coordination is between Medicaid and VA for long-term care costs. The dollar value of coordination of benefits is generally determined by an office that specializes in this activity, and the total benefits can be determined by multiplying the total dollar amount transferred to the other insurance coverage by 12 months, and totaling the amounts for all clients for which coordination of benefits have occurred.

C. POPULATING BENEFIT MODELS

One of the major challenges that will be faced by PARIS in the future is the population of these cost and benefit models. As has been explained, there are a number of difficulties in populating the cost and benefit models. Even with a consistent approach to calculating benefits, the nature of PARIS makes it difficult for States to devote the level of resources needed to implement the models in a consistent and accurate manner.

As indicated before, the study team believes that data are available at the State and local levels to populate these models. However, because a State may be estimating costs differently now, or because of a lack of resources, many States are not likely on their own to employ the models without some motivation and assistance. What is more likely to have support of the States is an independent assessment of costs and benefits, where an outside entity can come into a State, gather financial information, and populate the models.

The study team recommends that ACF consider an independent assessment of costs and benefits in a limited number of States. States could be grouped into the four categories of benefits noted in the models, and the independent agent could develop the appropriate measures to make the calculations. This could include a number of activities, including:
Review accounting records and conduct time studies to determine both the amount of time individuals in each State spend on PARIS activities, and develop an average cost per case follow-up activities.
Developing sampling methodologies for identifying a large enough sample of client actions so that savings identified through sampling can be used to create a valid estimate of savings.
Identify average benefit levels for all categories of action, including terminations of Food Stamp and TANF benefits. Develop categories for Medicaid savings and assist the State with identifying these categories in their data system.
Conduct the analysis of the data once the models are populated. Adjust the savings based upon issues identified in this report, including variance in size of match based upon frequency of data submissions, limited scope of PARIS activities, and limitation on data
Develop cost savings estimates for programs not included currently in a State’s PARIS activity to show potential savings if scope were to be expanded.

The advantage of this approach would be an unbiased cost benefit report that can then be used to (1) show the potential savings of PARIS through real examples; (2) be used as a sales point for new States to join; (3) help to encourage existing PARIS States to expand the scope of PARIS activities; and (4) possibly encourage all States to use a similar model so that in the future, costs and benefits can be viewed with consistency in approach and results.

It was beyond the scope of this study to collect all of the data from each State to populate the models. However, in order to demonstrate how the models would be populated, we have created a simulation of populated models in Appendix A. These models are created from hypothetical States using data that are similar to some of the data examined when the study team reviewed State calculations. They should not be considered actual calculations of a particular States cost or benefits, but are presented for informational purposes only.

VII. Future Expansion of PARIS

As noted earlier in this report, there are a number of opportunities for expansion of PARIS in the future. Because of changes in the file format, additional programs can be added as the PARIS community decides there is a need for expansion. In addition to new programs, there is also room for expansion of how existing PARIS data are being used by some States. Described below are some of these opportunities, and some issues to consider when making expansion decisions.

Expansion of programs included in the PARIS match. PARIS is currently considering expanding the PARIS match to include State Workers Compensation payments for the interstate match. By adding workers compensation to PARIS, States would be able to use the data to verify income that might not be reported because the source is from another State, or detect duplication of payments (fraud) that might occur. However, there are some issues that must be resolved before it can be fully embraced by the PARIS States. First and foremost, Workers Compensation is often handled by a State Government Department separate from where all other PARIS activities take place. This means that State officials must negotiate with the separate department to incorporate their data into the PARIS match and make arrangements for follow-up activity to take place. Second, training will need to take place in order to ensure that State and local officials can read and understand the file that is created, and clearly identify how income is displayed, time limits, and any other limitations on the data.

At the PARIS conference, States attending were asked if they were interested in or planned to participate in the Workers compensation match, if it were to be made available to them. Of the 35 States responding to the question, “Do you plan on participating in the Workers Compensation Match that will be available in August of 2007?” 46 percent responded that they would not participate, 31 percent indicated that they would, and 23 percent were uncertain as to whether or not they would participate.

Also noted earlier is the potential for expanding PARIS to include child care payments in the database. This process will likely need to be refined by ACF and the PARIS Board of Directors to both demonstrate the value of the match as well as how best to conduct follow-up before many States decide to participate.

One method by which expansion can be promoted is through the publishing of PARIS success stories on the PARIS website. By posting successes of States participating in the Workers Compensation and Child Care program matches, other States will see the value of participating and may well join.

Expansion of the number of States included in PARIS. It is a stated goal of the PARIS board of Directors to include all States in the PARIS match. As of the date of this report, there were still nine States that had not signed PARIS membership agreements. As indicated earlier, some of these States have indicated an interest in participating. For example, Hawaii has attended the PARIS conference for the past 2 years at their own expense, and has indicated that they are likely to join in the near future. Additionally, a report by the California Legislative Analyst office prepared in March of 2007 indicated that California would benefit greatly from joining PARIS, and information sharing meetings have taken place between ACF and California officials.

As has been noted, there is the potential for significant benefits to all States joining PARIS. For example, if one were to examine a single benefit that could potentially result from PARIS participation, it is in the area of closing Medicaid cases where a client enrolled in managed care has moved, but not reported the move. If a client moves, and does not report the move, the State continues to make payments to the Managed Care Organization (MCO). By identifying these unreported moves, the State can realize a savings from closing Medicaid managed care cases where capitation payments are being made.

To provide a rough example of the potential for savings, the study used data from the Kaiser Foundation State Health Facts website (www.healthfacts.org), to make a very rough estimate of potential savings for some of the non-participating States. The method used for calculating the savings is the method proposed in the cost-benefit model discussion previously. No attempt to calculate the cost of realizing these benefits was made, as it is unknown as to how each of these States would conduct follow-up, so only potential benefit amounts are displayed. It must be remembered that not only are these data from several years ago, but match rate data are merely estimates of potential matches based upon the experience of other PARIS States.

Table 8 displays some of the States not currently participating in PARIS, the estimated percentage of all Medicaid recipients enrolled in managed care in June of 2005, and the potential savings if these States experience the same match rate for unreported moves as the average of the study population. These figures represent a very rough estimate of the potential for savings in Medicaid managed care payments for the States listed, and is presented for demonstration purposes only. No claim is being made that States will actually realize these savings. However, the methodology used to calculate this savings was done using a conservative approach of a 1 percent match rate for the entire State’s managed care population¹⁴ and 50 percent of the matches being cases where the client has moved from the State cited.

While these savings are very rough estimates, the potential savings in only one area of PARIS, can be a strong motivator for States to consider joining. In any case, States should be able to use this approach with their current data to estimate potential savings from PARIS in this and other program areas, and to develop the necessary decision packages or justification for joining the PARIS family.

Table 8. Estimated Annual Savings from Medicaid Managed Care Case Closures Based Upon PARIS Matches in a Sample of States not Currently Participating in PARIS

STATE	PERCENT ENROLLED IN MANAGED CARE (2005)	NUMBER OF LIKELY MATCHES	NUMBER OF CASES POTENTIALLY CLOSED	TOTAL POTENTIAL SAVINGS
State A	50%	32,318	16,159	$25,983,929
State B	48%	13,358	6,679	$10,339,178
State C	92%	2,620	1,310	$2,641,079
State D	31%	4,905	2,453	$4,944,483
State E	80%	1,490	745	$1,323,320
State F	61%	320	160	$395,520

Expansion of the scope of PARIS in current PARIS States. As has been noted, there is great potential for expansion of PARIS within the existing States. Several States do not match or conduct follow-up on some of the PARIS files. Only two of the 14 study States reported using PARIS for coordination of benefits between Medicaid, VA and other Federal health insurance programs. The most common reasons cited by States for not expanding the scope of PARIS include lack of resources to develop follow-up systems, lack of follow-up staff, and a lack of understanding of how to use the PARIS data to the maximum. By supporting efforts to expand the scope of PARIS within existing States, ACF can capitalize on the excellent work of other States in modeling productive uses for PARIS data. As was noted in the section on expansion of PARIS matches, the PARIS website can be an effective tool in encouraging States to expand their PARIS efforts to increase their total savings.

VIII. Conclusion

The PARIS project is viewed by participating States as a beneficial effort that helps improve program integrity. The findings of this study support this belief. PARIS provides States with a tool that has the potential for not only saving money, but also helping to improve the overall integrity of public assistance programs. With the PARIS project almost doubling in size since this study started, there are even more opportunities for States to match with new partners. While some States are currently limiting their applications of PARIS, the potential for both expanded use of existing data as well as growth in the types of programs participating in PARIS is excellent. The sharing of ideas among States, along with States’ development of creative uses for PARIS data will help promote expansion in the future and will help PARIS to reach its full potential as a tool to preserve program integrity.

Footnotes

¹ Federal health coverage could include health benefits through such sources as coverage for current Federal employees, military health coverage through TRICARE, or benefits obtained through military retirement.

² The term "jurisdictions" is use to describe entities participating in PARIS, as there are currently 41 States, the District of Columbia, and Puerto Rico. In addition, counties within non-participating States are eligible to join, although none have as of June 30, 2007. However, within the PARIS community, the term "PARIS States" rather than "PARIS jurisdictions" is used, and therefore also used in this report.

³ All States have appeals processes that give clients who are removed from the program the opportunity to appeal the decision,. In some cases, a client may be removed from the program, but be reinstated later if they successfully appeal the State's decision. Appeals of decisions to terminate a client based on PARIS data are usually only successful if the client is living in the State from which the termination occurred, but did not respond to initial inquires related to the PARIS match.

⁴ In some States a problem was identified related to termination of Medicaid cases because of rules that allow Medicaid benefits to continue even when other program benefits have been terminated (Transitional Medicaid Services). This problem occurred in two States where TANF and Food Stamps were housed in a different department than Medicaid. Officials at local offices could terminate TANF and Food Stamp cases where it was shown that the client lived in a different State, but Medicaid could not make the same termination until a separate determination was made that the client was no longer eligible for Transitional Medicaid Services. The result is that capitation payments continued to be made to the managed care plan until the separate determination process was made.

⁵ States included in the sample were Washington, Oregon, Tennessee, Utah, Colorado, Rhode Island, Virginia, District of Columbia, Florida, Missouri, Oklahoma, New York, Maryland, and Illinois.

⁶ One of the objectives of this study was to develop a standardized methodology for calculating costs and benefits of PARIS. States have expressed interest in having such a standardized approach to help with comparing results across states and to calculate costs and benefits of potential PARIS expansion activities without having to "re-invent" a process.

⁷ At the 2007 PARIS conference, 36 of the 38 States reported location of follow-up activities. Some newer States have not implemented follow-up activities as of March 2007, but are included in these data because they have solid plans in place for conducting follow-up through one of the three methods. Two States participating in the PARIS conference have not been conducting follow-up as of yet, and have not made a decision related to its location.

⁸ For purposes of displaying total matches, August data were used because August is the month in which all PARIS States are encouraged to provide data, thus providing more of an assurance that all States in the sample would be included. For examining number of multiple State matches, February data were used, as these data included newer States and increased the total population from which matches could result.

⁹ In all four States, Medicaid field staff certify and handle cases where a client receives Medicaid benefits only. If a client receives benefits from TANF and/or Food Stamps as well as Medicaid, the case is handled by the social service rather than the Medicaid agency. This is because eligibility for Medicaid was determined as part of adjunctive eligibility. As a result PARIS matches involving Medicaid and at least on of the other two programs were worked by the TANF or Food Stamp eligibility worker.

¹⁰ Such as other data matches discussed earlier in this report or other caseload-related activities such as outreach and processing applications.

¹¹ Client classification is important in calculating Medicaid savings. For example, pregnant women, children, individuals with disabilities, or the elderly may be eligible for different capitation rates under managed care or higher reimbursement rates under fee-for-service. We believe that it is not unreasonable for a State to identify client categories for Medicaid case closures.

¹² For States using a primary care case management model, the cost of the case management fee would be applied rather than a capitation rate.

¹³ Some States may want to break out clients by category, such as pregnant women, children, etc. while others may wish to use an average monthly reimbursement amount for all Medicaid clients.

¹⁴ Actual rates for matches in the study States ranged from a high of 4.1 percent to a low of .29 percent. Seven of the 14 study states had match rates of 1.3 percent or higher.

Figure Text Descriptions

Figure 1. Map of States with Signed PARIS Agreements, Prior to and Since September 2005

This map is an outline of the United States of America and a Territory, Puerto Rico. In addition to the 48 contiguous States, outlines of Alaska and Hawaii are included at the bottom.
The table has a key below it containing three boxes.
The first box is light yellow and indicates that a State filled with this shade was a PARIS State prior to September 2005.
States shaded in light yellow are: Alaska, Arkansas, Arizona, Colorado, Georgia, Idaho, Indiana, Louisiana, Maine, Michigan, Minnesota, Montana, New Mexico, Nevada, Oklahoma, South Carolina, Wyoming, West Virginia
The second color is orange and shows that a State filled with this shade has been a PARIS State since September 2005.
States and Territories shaded in orange are: Connecticut, Delaware, District of Columbia, Florida, Illinois, Kansas, Kentucky, Maryland, Massachusetts, Mississippi, Missouri, Nebraska, New Jersey, New York, North Carolina, North Carolina, Oregon, Pennsylvania, Puerto Rico, Rhode Island, South Dakota, Tennessee, Utah, Virginia, Washington, Wisconsin
The third box is outlined in black with no shade and indicates that a State with no shade is not a PARIS State.
These States are not shaded: Alabama, California, Iowa, North Dakota, New Hampshire, Ohio, Texas, Vermont

Figure 2. Map of Number of Matches in Which States Participate

This map is an outline of the United States of America and a Territory, Puerto Rico. In addition to the 48 contiguous States, outlines of Alaska and Hawaii are included at the bottom.
The table has a key below it containing six boxes.
The first box is pink and indicates that a State filled with this shade submitted one match.
States and Territories filled with pink are: Missouri and Puerto Rico
The second box is blue and indicates that a State filled with this shade submitted two matches.
States filled with blue are: Arizona, Connecticut, Delaware, Kansas, Massachusetts, Nevada, New York, Pennsylvania, Rhode Island
The third box is green and indicates that a State filled with this shade submitted three matches.
States filled with green are: Alaska, Arizona, Colorado, District of Columbia, Florida, Illinios, Indiana, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, Montana, Nebraska, New Jersey, New Mexico, North Carolina, Oklahoma, Oregon, South Carolina, Tennessee, Utah, Virginia, Washington, Wisconsin
The fourth box is orange and indicates that a State filled with this shade submitted no matches.
States filled with orange are: Georgia, Idaho, South Dakota, West Virginia, Wyoming
The fifth box is yellow and indicates that a State filled with this shade submitted no matches, but plans to submit 2.
The State filled with yellow is: Minnesota
The sixth box is purple and indicates that a State filled with this shade submitted no matches, but plans to submit 3.
The State filled with purple is: Maine

Figure 3. Map of States Reporting Follow-up Activity Locations

This map is an outline of the United States of America and a Territory, Puerto Rico. In addition to the 48 contiguous States, outlines of Alaska and Hawaii are included at the bottom.
The table has a key below it containing five boxes.
The first box is pink and indicates that a State filled with this shade submitted one match.
States filled with pink are: Alaska, Arizona, Arkansas, Connecticut, Delaware, Louisiana, Oklahoma, Oregon, Michigan, Montana, Nevada, South Dakota, Wisconsin, Wyoming
The second box is orange and indicates that a State filled with this shade submitted two matches.
States and Territories filled with orange are: Colorado, Florida, Illinois, Kansas, Kentucky, Mississippi, Nebraska, New Mexico, New York, North Carolina, Pennsylvania, Puerto Rico, Virginia
The third box is green and indicates that a State filled with this shade submitted three matches.
States filled with green are: Indiana, Maryland, Massachusetts, South Carolina, Tennessee, Utah, Washington
The fourth box is blue and indicates that a State filled with this shade has not yet set up a follow-up system.
States filled with blue are: Georgia, Idaho, Maine, Minnesota, West Virginia
The fifth box is purple and indicates that a State filled with this shade has not reported.
The State filled with purple is: New Jersey