Case Study: Modernization of the Food Stamp Program in Florida

1. Case Study: Modernization of the Food Stamp Program in Florida Christine Kissmer Scott Cody Emily Sama Martin Renee Nogales

2. Many Faces of Modernization Three commonalities: Combine technology applications, policy changes & restructuring of administrative functions Intend to create efficiencies that benefit both government & client. Apply to more than one program. Substantial variation in State efforts.

3. Food Stamp Program Modernization:4 Categories of Activities Policy changes Restructuring administrative functions Expanding application of technology Partnering with nonprofit organizations & businesses

4. 3 Basic Approaches Multi-Category/Systemic modernization� �reengineering the FSP� to create more integrated, accessible, and efficient system of providing food assistance Incremental or �modest� modernization� e.g. on-line prescreening tools or applications, pursuit of more policy simplification, engaging community partners for outreach Minimal or no modernization� e.g. enacted simplified reporting requirements but otherwise, satisfied with current program or �wait- and-see� approach GAO STUDY NOTES Selection of Florida GAO STUDY NOTES Selection of Florida

5. ACCESS Florida Goals Reduce Administrative Costs Increase Client Access

6. Key Study Questions How did application procedures change? What are the key organizational changes? How is technology used? Has modernization affected participation? How much did ACCESS Florida save the state in administrative costs? Did modernization affect other key performance measures: error rates, client satisfaction? What are the lessons for other states? ACCESS Florida included changes affecting public assistance programs such as TANF, SSI, Food Stamps, and Medicaid. We focused our case study on the specific experience of Food Stamp participants.ACCESS Florida included changes affecting public assistance programs such as TANF, SSI, Food Stamps, and Medicaid. We focused our case study on the specific experience of Food Stamp participants.

7. Overview of Study Methodology Study Focuses on July � September 2006 Site Visits and/or Phone Interviews in 7 Districts Focus Groups with FSP Clients in 3 Districts Analysis of Participation, Payment Error and Cost Data - Focus on July to September 2006, but in some cases have data beyond this point to illustrate a longer perspective.- Focus on July to September 2006, but in some cases have data beyond this point to illustrate a longer perspective.

8. Timeline 2003: Mandate from Florida Legislature 2004: Organizational Restructuring Closures Begin 2005: Web Application Call Centers Community Partners 2006: Case Study (July � September)

9. How did Application Procedures Change?

10. Electronic Application

11. Location of Application At Customer Service Center Web application on lobby PCs Outside Customer Service Center Home, office, or other place with web access Community partners

12. Eligibility Interviews Abbreviated Interview (10 Minutes) Green Track ? Case Processing Red Track ? Additional Interviews Reduced Documentation Red Track: Parent left the home in the last year Money management problems Recent fraud findings Cases currently subject to sanctions Reduced Documentation: Most expenses and assets, and some income, do not require documentation Self-service submission of documentation, either in person or by fax Red Track: Parent left the home in the last year Money management problems Recent fraud findings Cases currently subject to sanctions Reduced Documentation: Most expenses and assets, and some income, do not require documentation Self-service submission of documentation, either in person or by fax

13. What Are the Key Organizational Changes?

14. Specialization of Caseworker Functions Replaced caseworker modelReplaced caseworker model

15. Separate Intake and Processing Functions Before modernization: Clerks Case Manager After modernization: Meeter-Greeter Computer helper in some Customer Service Centers Clerks Intake Specialist Processing Specialist

16. Case Maintenance Units One for each district Database monitoring and matching Apply and lift sanctions Process Medicaid requests and manage Medicaid files

17. Community Partners

18. Community Partners (continued) Partners as of September 2006. We do not have data for later months.Partners as of September 2006. We do not have data for later months.

19. Agency Downsizing

20. How Is Technology Used?

21. Modernized Technology for Clients and DCF Workers

22. Software Interactions

23. Customer Call Center Toll-free number directs to Automated Response Unit that answers common questions Clients can choose to be routed to a live agent at the call center

24. Customer Call Center (continued)

25. Potential Impacts

26. Trends in Participation, Errors and Costs Data sources for trend analyses: DCF administrative data FNS QC error data DCF operational cost data FNS form 269 data Trends do not prove causality

27. How Has Modernization Affected Participation? When we began our analysis, we expected that some factors about modernization would increase access to and participation in Food Stamps, and that other factors would decrease access and participation. We anticipated that certain factors could increase client access to or satisfaction with the FSP. These included the new interview procedures, relaxed verification, redetermination, and reporting changes. We also expected that the move towards downsizing DCF could cause some decreased access for some clients. There were five aspects of the modernization initiative that we expected could either increase or decrease client access. (1)CCC (2)Partners (3)Service center set-up (4)Web application (5) ARU When we began our analysis, we expected that some factors about modernization would increase access to and participation in Food Stamps, and that other factors would decrease access and participation. We anticipated that certain factors could increase client access to or satisfaction with the FSP. These included the new interview procedures, relaxed verification, redetermination, and reporting changes. We also expected that the move towards downsizing DCF could cause some decreased access for some clients. There were five aspects of the modernization initiative that we expected could either increase or decrease client access. (1)CCC (2)Partners (3)Service center set-up (4)Web application (5) ARU

28. Florida FSP Caseload Leveled off as Employment Rose The caseload growth rate, which we calculated as the percentage change in the number of FSP clients in Florida since July 2001, was fairly constant on average during the ACCESS Florida implementation period In the years before ACCESS Florida, the caseload had been increasing at a growing rate each year. This change may in caseload growth rate may be attributable to ACCESS Florida�s new approach to administering the FSP. However, it�s also possible that part of this change in caseload growth may be explained by Florida�s unemployment rate, which declined between 2002 and 2006 before picking up again in 2007. We can�t know with any degree of certainty, therefore, how much of the change is attributable to modernization.The caseload growth rate, which we calculated as the percentage change in the number of FSP clients in Florida since July 2001, was fairly constant on average during the ACCESS Florida implementation period In the years before ACCESS Florida, the caseload had been increasing at a growing rate each year. This change may in caseload growth rate may be attributable to ACCESS Florida�s new approach to administering the FSP. However, it�s also possible that part of this change in caseload growth may be explained by Florida�s unemployment rate, which declined between 2002 and 2006 before picking up again in 2007. We can�t know with any degree of certainty, therefore, how much of the change is attributable to modernization.

29. Elsewhere in the Southeast, Caseloads Increased but Unemployment was Steady For comparison, we looked at the caseload growth rate in the remaining states in SERO. These other states were not affected by the modernization initiative and caseload growth continued to increase in these areas. However, these states also experienced a fairly constant unemployment rate over this period during which unemployment in Florida fell.For comparison, we looked at the caseload growth rate in the remaining states in SERO. These other states were not affected by the modernization initiative and caseload growth continued to increase in these areas. However, these states also experienced a fairly constant unemployment rate over this period during which unemployment in Florida fell.

30. Caseload Grew Less in Counties with Fewer Access Points To approach the question in a different way, we looked at the average caseload growth rate before and after modernization in counties with and without DCF office closures. Caseload growth continued in all types after ACCESS Florida, but at a much lower rate than before modernization. The difference in caseload growth between the two periods was more pronounced in counties with office closures than in those without office closures. We categorized counties by whether they had experienced DCF office closures since ACCESS Florida began. Then, we looked at the caseload growth rate by county type. To approach the question in a different way, we looked at the average caseload growth rate before and after modernization in counties with and without DCF office closures. Caseload growth continued in all types after ACCESS Florida, but at a much lower rate than before modernization. The difference in caseload growth between the two periods was more pronounced in counties with office closures than in those without office closures. We categorized counties by whether they had experienced DCF office closures since ACCESS Florida began. Then, we looked at the caseload growth rate by county type.

31. Percent Change in FSP Participation after July 2001 Source: MPR tabulationas of FNS Program Operations Data Note: Shaded area reflects the period when most ACCESS Florida changes were implemented. Southeast Region includes AL, GA, KY, MS NC, SC and TN. Large states include CA, IL, MI, NY, OH, PA and TX. Florida experienced severe hurricanes and tropical storms in August and September 2004 (Hurricanes Charley, Frances and Ivan and Tropical Storm Bonnie) and in July through October 2005 (Hurricanes Dennis, Wilma and Katrina). These storms partially explain fluctuations in the caseload growth rate in Florida during these periods as well as in other states in the Southeast Region and other large states (such as Texas). Hurricane Rita in October 2005 which did not hit Florida also affected the Southeast Region and Texas. Source: MPR tabulationas of FNS Program Operations Data Note: Shaded area reflects the period when most ACCESS Florida changes were implemented. Southeast Region includes AL, GA, KY, MS NC, SC and TN. Large states include CA, IL, MI, NY, OH, PA and TX. Florida experienced severe hurricanes and tropical storms in August and September 2004 (Hurricanes Charley, Frances and Ivan and Tropical Storm Bonnie) and in July through October 2005 (Hurricanes Dennis, Wilma and Katrina). These storms partially explain fluctuations in the caseload growth rate in Florida during these periods as well as in other states in the Southeast Region and other large states (such as Texas). Hurricane Rita in October 2005 which did not hit Florida also affected the Southeast Region and Texas.

32. FSP Caseload Growth and Unemployment Rate in Florida

33. FSP Caseload Growth and Unemployment Rate in Other Southeast States Source: MPR tabulations of FNS Program Operations data and unemployment data from the Bureau of Labor Statistics. Southeast Region includes AL, GA, KY, MS, NC, SC and TNSource: MPR tabulations of FNS Program Operations data and unemployment data from the Bureau of Labor Statistics. Southeast Region includes AL, GA, KY, MS, NC, SC and TN

34. Caseload Composition Changed Little If modernization reduced client access to the FSP, we might expect a decline in the share of the caseload that are elderly, non-English speakers, and clients with disabilities � populations that may struggle with new technology, changing access points, or more limited staff interaction Within Florida, the composition of the caseload changed little in the time before and after ACCESS Florida. There is little difference between the two periods in the share of the caseload that is elderly or disabled. And, although we don�t have data on language, given Florida�s large immigrant population we would expect to find a number non-English speakers in the Hispanic demographic group. The share of that group in the caseload also did not change substantially. The steady demographic composition of the caseload, together with the apparent interaction between the caseload and unemployment, gives no strong evidence to suggest that ACCESS Florida limited the access that eligible clients have to FSP benefits in Florida. If modernization reduced client access to the FSP, we might expect a decline in the share of the caseload that are elderly, non-English speakers, and clients with disabilities � populations that may struggle with new technology, changing access points, or more limited staff interaction Within Florida, the composition of the caseload changed little in the time before and after ACCESS Florida. There is little difference between the two periods in the share of the caseload that is elderly or disabled. And, although we don�t have data on language, given Florida�s large immigrant population we would expect to find a number non-English speakers in the Hispanic demographic group. The share of that group in the caseload also did not change substantially. The steady demographic composition of the caseload, together with the apparent interaction between the caseload and unemployment, gives no strong evidence to suggest that ACCESS Florida limited the access that eligible clients have to FSP benefits in Florida.

35. Caseload Composition Remained Virtually Unchanged Source: MPR tabulations of Department of Children and Families data.Source: MPR tabulations of Department of Children and Families data.

36. Applications Increased While Approval Rates Decreased One clear point that emerged from our analysis of Florida�s data is that the state is receiving far more applications than before, despite the leveling-off in caseload growth. The total number of applications grew by more than 10 percent after ACCESS Florida, but the approval rate declined. This may be attributable to several factors, including increased access to the FSP application by ineligible applicants or an increase in applications that are submitted without supporting documentation. From the data we obtained, we can not know the true reason that the applications increased while the approval rates declined. One clear point that emerged from our analysis of Florida�s data is that the state is receiving far more applications than before, despite the leveling-off in caseload growth. The total number of applications grew by more than 10 percent after ACCESS Florida, but the approval rate declined. This may be attributable to several factors, including increased access to the FSP application by ineligible applicants or an increase in applications that are submitted without supporting documentation. From the data we obtained, we can not know the true reason that the applications increased while the approval rates declined.

37. Client Impressions Rated most recent DCF experience positively Mixed opinions of assistance available in lobby Generally aware of new self-service components Preferred the �personal touch� of the caseworker model Liked the web application Struggled with use of call centers We conducted twelve focus groups with clients, recent applicants, and non participants over the course of this study. One of the questions we asked was how they would rate their most recent DCF experience. Most clients rated their experience highly, and either equal to or better than their most recent other interaction with another government agency (such as the DMV) or a previous interaction with DCF. Along with their positive general opinions, clients had some reservations about the new approach. Many clients described frustration with using the CCC, and some said that they missed the personal approach of the former caseworker model. Some clients were upset about recent experiences at DCF which they felt were unpleasant. Clients also liked some new modernized components, such as the web application and, in some cases, the assistance and resources available to them in the lobby. In summary, clients had mixed opinions about the different components of modernization, which would not be inconsistent with what we�d expect from clients experiencing any major change in the way customer service is delivered. Overall, however, clients generally had positive impressions of the modernized approach. We conducted twelve focus groups with clients, recent applicants, and non participants over the course of this study. One of the questions we asked was how they would rate their most recent DCF experience. Most clients rated their experience highly, and either equal to or better than their most recent other interaction with another government agency (such as the DMV) or a previous interaction with DCF. Along with their positive general opinions, clients had some reservations about the new approach. Many clients described frustration with using the CCC, and some said that they missed the personal approach of the former caseworker model. Some clients were upset about recent experiences at DCF which they felt were unpleasant. Clients also liked some new modernized components, such as the web application and, in some cases, the assistance and resources available to them in the lobby. In summary, clients had mixed opinions about the different components of modernization, which would not be inconsistent with what we�d expect from clients experiencing any major change in the way customer service is delivered. Overall, however, clients generally had positive impressions of the modernized approach.

38. Have the Changes Affected Payment Errors? With streamlined and mechanized approaches to processing applications replacing a system that was previously more driven by personal interaction, the issue of payment errors becomes increasingly important. Some factors, such as agency downsizing, new technology, and relaxed verification have the potential to increase the error rate. For example, fewer staff touching more cases could lead to inadvertent oversights in eligibility and benefit determination, especially when relaxed verification means that staff are collecting less information from clients. Other factors, such as simplified reporting, new ways for clients to report changes, and specialized worker functions, could either increase or decrease errors. With streamlined and mechanized approaches to processing applications replacing a system that was previously more driven by personal interaction, the issue of payment errors becomes increasingly important. Some factors, such as agency downsizing, new technology, and relaxed verification have the potential to increase the error rate. For example, fewer staff touching more cases could lead to inadvertent oversights in eligibility and benefit determination, especially when relaxed verification means that staff are collecting less information from clients. Other factors, such as simplified reporting, new ways for clients to report changes, and specialized worker functions, could either increase or decrease errors.

39. Payment Errors in Florida Increased From 2004-2006 We analyzed the average monthly error rate for the years before and since ACCESS Florida and found that Florida mirrored the error rate decline seen in other states, especially other states like Florida, until 2004. After 2004, Florida�s error rate increased while the error rates elsewhere continued to decline. It is important to note that, while this increase coincides with modernization policies, it also coincides with several hurricanes in 2004 and 2005. During these years, the state ran large disaster benefit programs, which disrupted regular operations and diluted staff time and resources. We can not know for sure what share of this increase in errors, if any of it, would have occurred in the absence of the disasters. We analyzed the average monthly error rate for the years before and since ACCESS Florida and found that Florida mirrored the error rate decline seen in other states, especially other states like Florida, until 2004. After 2004, Florida�s error rate increased while the error rates elsewhere continued to decline. It is important to note that, while this increase coincides with modernization policies, it also coincides with several hurricanes in 2004 and 2005. During these years, the state ran large disaster benefit programs, which disrupted regular operations and diluted staff time and resources. We can not know for sure what share of this increase in errors, if any of it, would have occurred in the absence of the disasters.

40. Payment Errors in Florida Decreased after 2006(based on official payment error rate)

41. How Much Did ACCESS Florida Save in Administrative Costs? A major impetus for Florida�s modernization initiative was the state legislature�s requirement that DCF decrease its budget, so we examined the question of administrative costs. We expected that most parts of the initiative, such as agency downsizing, an increased reliance on client self-service, and policy changes that reduce the amount of contact clients must have with staff, would likely to reduce administrative costs by increasing efficiency. It seemed that other changes, such as the call centers, recruiting and training partners, and developing a web application and streaming software could either increase or decrease costs, depending on how those changes are implemented. While these changes are costly to develop, they might ultimately create efficiencies that outweigh the costs of modernization. A major impetus for Florida�s modernization initiative was the state legislature�s requirement that DCF decrease its budget, so we examined the question of administrative costs. We expected that most parts of the initiative, such as agency downsizing, an increased reliance on client self-service, and policy changes that reduce the amount of contact clients must have with staff, would likely to reduce administrative costs by increasing efficiency. It seemed that other changes, such as the call centers, recruiting and training partners, and developing a web application and streaming software could either increase or decrease costs, depending on how those changes are implemented. While these changes are costly to develop, they might ultimately create efficiencies that outweigh the costs of modernization.

42. DCF Operating Costs Fell Substantially DCF�s operating costs declined substantially. Costs were relatively stable from 2001 through 2003, and then dropped in 2004 to 2006. Most savings came in the category of personnel expenses. Overall, the state reduced its operating costs by more than 30 percent between 2001 and 2006. DCF�s operating costs declined substantially. Costs were relatively stable from 2001 through 2003, and then dropped in 2004 to 2006. Most savings came in the category of personnel expenses. Overall, the state reduced its operating costs by more than 30 percent between 2001 and 2006.

43. Certification Costs Declined� We observed the opposite pattern when examining certification costs. These fell by half in Florida between 2001 and 2006, while certification costs elsewhere declined by only a few percentage points over the period. This change in certification costs in Florida is particularly notable because certification costs comprise the largest share (55 percent in 2006) of all administrative costs for the state�s FSP program. We observed the opposite pattern when examining certification costs. These fell by half in Florida between 2001 and 2006, while certification costs elsewhere declined by only a few percentage points over the period. This change in certification costs in Florida is particularly notable because certification costs comprise the largest share (55 percent in 2006) of all administrative costs for the state�s FSP program.

44.  Their goal is a web-based front�end system that uses FLORIDA engine. Clients will still have to fill-in the application but data from the web will be displayed alongside current FLORIDA data.

45. Streaming � State in process of developing the Access Management System that will interact with FLORIDA and eliminate the need for streaming. The Client Registration part of AMS was completed October 2007. The Application Entry part was scheduled for completion by January 2009 but will be later. ACCESS Florida - Updates

46. Call Center � ARU simplified to help clients. It will also be expanded to give clients more data options. One-quarter of callers are getting their information from the ARU (up from 10%). Tool developed to forecast caseload and the number of agents needed. Currently do not have enough staff to provide coverage. ACCESS Florida -- Updates

47. ACCESS Florida -- Updates Web Application � new version rolling out February 2009 and will include question about origin of application. Document imaging system rolled out. Scanning component that was developed by the SunCoast was rewritten to .NET for more flexibility. Now adding FAX functionality to it.

48. For more information: www.fns.usda.gov/oane/

49. OPTIONAL SLIDES Do not copy

50. What Are the Lessons for Other States?

51. Lessons for Other States The timing and order of changes can impact successful implementation during initial stages Florida�s communication strategy appears to be effective Local flexibility appears to enhance services More resources for fraud prevention, benefit recovery, and quality control may help offset any increase in error rates Types of worker skills and recruiting efforts likely will change under a modernization model

53. Several possible directions for future FNS studies Does modernization make a difference in who is reached by the Food Stamp Program? What reporting requirements and performance standards are important to a modernized Food Stamp Program?

54. FNS Research Agenda Florida case study first of three modernization research efforts. Second is Enhancing Food Stamp Certification: Food Stamp Modernization Efforts, The Urban Institute, Oct 2006 � June 2009 Comprehensive inventory on modernization in all 50 States and DC (plus some counties) Describe and compare practices. Identifying modernization best practices. Further intensive case studies -- assessment of data and opportunities to evaluate impacts on access, errors, and costs.

55. Assessing Modernization Impacts: A Research Challenge Challenge to untangle effects of multiple programs & external factors. States using combinations of techniques. Lack of comparison sites. Limited or no data (e.g., application withdrawals/incompletions)

56. Change In FSP Participation In FL, MS, AL & Rest of SERO (includes disaster FS benefits)