House Ways and Means Subcommittee on Social Security Hearing

Chairman Johnson, Ranking Member Becerra and Members of the Subcommittee:

Thank you for the opportunity to discuss the ways in which the Social Security Administration (SSA) uses information technology to administer its programs, detect and prevent improper payments, and support anti-fraud initiatives. I am Bill Zielinski, and I am the Chief Information Officer and Deputy Commissioner for Systems at SSA. I am responsible for delivering cost-effective information technology (IT) services, and for protecting the information assets of Social Security.

Overview of Our IT operations and Online Services

SSA has many IT strengths. For example, we have a superb technical workforce and are experts at technical project management. We have designed and maintained a highly automated process for handling benefits claims and other work, including program integrity reviews. We have consolidated most of our agency's IT so that we benefit from economies of scale. We excel at designing applications that focus on users. We also have developed a rigorous, annual process to assess and prioritize future IT investments, as we always have more IT needs than available or expected resources. All agency components actively engage in this process. As the CIO for the agency, I am committed to ensuring that our IT infrastructure and services are secure, scalable, and available.

We have a proud history of using IT to support our administration of the Social Security and Supplemental Security Income (SSI) programs, and to provide substantial support to the related Medicare, Medicaid, and other government programs. These programs are immense in scope: in FY 2013, we paid over $855 billion to more than 63 million Social Security beneficiaries and SSI recipients. To support our programs, our mainframe contains approximately 14 petabytes of data, and our open, client-server IT infrastructure maintains 13 petabytes. In FY 2013, this IT infrastructure supported the processing of an average daily volume of nearly 150 million individual transactions. For the year, our IT operations supported approximately: 1.6 billion automated Social Security number verifications; 251 million earnings items; 5 million retirement, survivor, and Medicare applications; 3 million initial disability claims; 2.6 million nonmedical redeterminations; 1.5 million continuing disability reviews, including approximately 429,000 full medical continuing disability reviews; and 17 million new and replacement Social Security card applications.

Customer Satisfaction

In addition to maintaining robust IT operations capable of supporting the large demands of our programs, we are committed to building online services for the public that are simple and easy to use. We have been successful in this regard. According to the most recently released American Customer Satisfaction Index (ASCI) E-Government Satisfaction Index, we have the three highest rated--and four of the top five--e-government websites in the Federal government. Moreover, these four online services (Extra Help with Medicare Prescription Drug Plan Costs, iClaim, Retirement Estimator, and our my Social Security portal) outperformed or tied Amazon, the highest scoring e-retail website.

Supporting Increased Productivity

Our strategic investments in online services and our core IT operations have increased our productivity and efficiency--allowing us to keep up with ever-increasing workloads. For example, we currently have about the same number of employees that we had in 2007, even though our workloads have increased dramatically. In FY 2014, we estimate that the number of retirement and survivor applications will be about 30 percent higher than in FY 2007. Over the same period, the volume of initial disability claims we received increased by nearly 20 percent.

Our easy-to-use online application for applying for disability, retirement, and Medicare-- iClaim--is a huge success. Applicants file for benefits online at their own pace and on their own schedule. In FY 2013, over 1.27 million Disability Insurance (DI) claimants, or about 46 percent of DI claimants, filed online and over 1.25 million retirement claimants, or about 49 percent of retirement claimants, filed online. To compare, in FY 2008 (when we first introduced iClaim), only about 11 percent of DI claimants, and just over 15 percent of retirement claimants, applied online.

Similarly, my Social Security, is a personalized online portal that individuals can use beginning at age 18 and continuing throughout the time they receive Social Security benefits. Through this portal, individuals who register can view their Social Security Statement, view detailed information on benefits received (for up to 24 months), get a benefit verification letter, start or change direct deposit information, and change their address - all online. We will continue to expand the services provided in the my Social Security portal to enhance customer service. Currently, over 10 million people have established my Social Security accounts and used their accounts to access:

. Online Social Security Statement - 29.2 million times;

. MyDirect Deposit - 0.6 million times;

. MyChange of Address - 1 million times;

. Internet Benefit Verification Letter - 4.5 million times; and

. MyCheck Your Benefits - 20.6 million times.

Due in large part to these successful online services and our other IT initiatives, we are able to keep our administrative costs low - about 1.4 percent of the benefit payments we pay each year.

Quality Has Improved

Our efforts to improve processing times and increase productivity have not come at the expense of our quality. Quality is integral to all of our processes, including our disability claims process. For instance, we have developed and implemented the electronic Claims Analysis Tool (eCAT), a web-based application, to help State disability determination services (DDS) examiners apply policies correctly throughout the disability decision-making process. eCAT uses "intelligent pathing," which prompts users to consider the appropriate questions based on the unique characteristics of each case. This documentation is particularly useful for future case review because it enables an independent reviewer to understand the examiner's actions and conclusions throughout the development and adjudication of the claim. We fully implemented eCAT last year and made it mandatory for use in every DDS.

We are also piloting our Electronic Bench Book (eBB) application. eBB aids in documenting, analyzing, and adjudicating disability cases at the hearings level in accordance with the Social Security Act and our regulations. We expect that eBB will improve the accuracy and consistency of our disability decision process.

Another example of where we use IT to support our programs is our development of the Disability Case Processing System (DCPS). DCPS will replace the 54 different systems that support the DDSs with one national system based on state-of-the-art technology. This system will incorporate eCAT and other tools designed to improve quality and productivity. Additionally, DCPS will allow us to systematically implement policy changes in a faster way, and it will promote more consistency among the DDSs.

Applications that Support Payment Accuracy

We also develop applications that allow our employees to more efficiently gather information and identify improper payments. For example, in FY 2012, we implemented the Access to Financial Institutions (AFI) program nationally. This program allows our employees to automatically and electronically gather financial account information directly from financial institutions. Historically, having financial accounts in excess of the allowable resource limits is the leading cause of improper payments in the SSI program. Because AFI is more efficient than a paper-based process, we are able to verify financial account information--and thereby reduce improper payments--on more SSI claims and post-entitlement actions.

We also develop applications that allow beneficiaries to directly update their claim information--which results in more accurate benefit payments--without the need to contact a field office. For example, in FY 2008, we implemented the SSI Telephone Wage Reporting System (SSITWR), an automated toll-free number that makes it easy for SSI recipients to update the wage information on their records. Our studies indicate that wages submitted through SSITWR are highly accurate and we confirm their accuracy using our data exchanges. Based on the success of SSITWR we recently created a mobile wage reporting application.

Data Exchange and Electronic Verification Services

We have thousands of data exchange agreements with Federal, State, Local, and Foreign governmental entities. Data received from external exchange partners allows SSA to pay benefits accurately, efficiently and timely. Examples of data received are income, assets, incarceration status, medical evidence, and benefit payments received from other government programs.

By efficiently sharing data with other agencies and private organizations through our electronic verification services (where allowed by law), we help them to efficiently administer their programs and reduce the number of field office visits and 800 number calls to verify benefit information. This improves customer service and allows us to redirect our resources to our other critical program work.

SSA's Use of Predictive Models to Support Program Integrity

We take seriously our responsibility to maintain the public's trust through effective stewardship of program dollars and administrative resources. We use our IT operations and technical expertise in support of this critical strategic objective. Specifically, we use statistically valid predictive models that enhance key agency program integrity functions while ensuring that agency resources are used in the most cost effective and efficient manner possible. I will briefly summarize the predictive models we have used successfully to curb improper payments.

CDR Predictive Model

Beginning in FY 1993, SSA began developing a series of predictive models to ascertain the likelihood that a full medical continuing disability review (CDR) would result in a finding that a disability beneficiary has medically improved and is no longer eligible for benefits. Our predictive models for CDRs use a multitude of variables to provide an aggregate score that predicts the likelihood of medical improvement and cessation. Our use of the predictive models has allowed us to be extremely cost effective in prioritizing full medical CDRs with our limited resources. In FY 2013, we conducted approximately 429,000 full medical CDRs. We estimate that the money spent on CDRs saves, on average, $9 for every dollar invested, including savings accruing to Medicare and Medicaid.

SSI Redetermination Model

SSI redeterminations are reviews of all of the nonmedical factors of eligibility to determine whether a recipient is still eligible for SSI and still receiving the correct payment amount. All SSI recipients are subject to periodic redeterminations. Every year SSA schedules redeterminations for the cases most likely to have payment error. To do this, we use a statistically valid scoring model--the SSI Redetermination Model. This model, which we first implemented in the late 1970s, predicts the dollar amount of likely overpayments for every SSI recipient and, having such information in hand, ensures that we select SSI cases to be reviewed efficiently and in a highly cost effective way.

Like our other predictive models, we continually review and improve our SSI Redetermination Model; in FY 2011, we expanded the model to include SSI living arrangement information, which enhanced the effectiveness of the model in selecting the most productive SSI redeterminations. In FY 2013, we completed over 2.6 million SSI redeterminations; by targeting the highest priority cases, the SSI Redetermination Model helped us to recover or prevent $3.4 billion in SSI overpayments.

Pre-Effectuation Review (PER) Model

The law requires us to review at least fifty percent of all State Disability Determination Service (DDS) initial and reconsideration disability allowances, and a sufficient number of CDR continuances to ensure a high level of accuracy. In FY 2011 (the most recent year for which information is available), we reviewed over 500,000 allowances and 8,400 continuances. To ensure we target for review those cases with the highest risk of decisional error, we have developed and continue to improve our PER Model, which predicts the probability of error and dollar amount cost of erroneous DDS allowances. In FY 2011 alone, completed PER reviews resulted in preventing the release of $752 million in improper DI, SSI, Medicare and Medicaid program payments.

Continuing Disability Review Enforcement Operation Predictive Model

The Continuing Disability Review Enforcement Operation (CDREO) identifies DI beneficiaries who appear to have substantial earnings after disability onset, through an automated matching of our current DI beneficiaries with the Internal Revenue Service (IRS) reported earnings posted to our Master Earnings File. We recently developed a predictive model to identify which alerts are most cost effective. We implemented the model nationally in June 2013 after piloting the model in 2011 and 2012 by analyzing our CDREO alerts and prioritizing which alerts should be reviewed first.

Medicare Part D Subsidy Redetermination Model

SSA has primary responsibility for redetermining whether a beneficiary is eligible for a Medicare Part D Subsidy. To help us prioritize which cases to review, we have developed the Medicare Part D Subsidy Redetermination Model, which predicts those Medicare Part D cases most likely to have an incorrect subsidy amount. In FY 2013, the model identified the most productive 25 percent of Medicare Part D subsidy cases for redetermination; we estimate that these cases contain about 60 percent of all incorrect subsidy amounts.

This model helped us prioritize the roughly one million Medicare Part D subsidy cases for redetermination, which resulted in the correction of about $800 million in Medicare subsidy payments in FY 2012 alone.

Supplemental Security Record (SSR)/OCSE Wage Profiling Model

To help us prevent and detect SSI improper payments, we use the SSR/OCSE Wage Profiling model--a predictive model that uses data from a quarterly Office of Child Support Enforcement (OCSE) wage data match to determine which SSI recipients have received wages that are likely to result in significant SSI overpayments. From FY 2000 through FY 2013, the cases selected under this model have resulted in the recovery or prevention of $1 billion in SSI overpayments.

SSR/IRS 1099 Income Profiling Model

Similar to the SSR/OCSE model, the SSR/IRS 1099 Income Profiling Model uses data from SSA's quarterly IRS 1099 data match to determine which SSI recipients have received unearned income that is most likely to result in significant SSI overpayments. From FY 2000-FY 2013, the cases selected under this model have resulted in the recovery or prevention of $740 million in SSI overpayments.

Representative Payee Misuse Models

The Representative Payee Misuse Models allow us to more effectively target reviews of representative payees to detect, deter, and prevent misuse of beneficiary funds by representative payees. Based on recommendations from a National Academy of Sciences study commissioned by Congress, we developed statistical models to identify cases that had the greatest likelihood of detecting beneficiary funds misuse. These models target both individual representative payees and representative payee organizations. The models are able to detect cohorts of cases with a misuse rate at about twenty times the overall rate occurring in the universe of all beneficiaries served by representative payees.

SSA's Increasing Use of Big Data and Data Analytics

In addition to our successful predictive models, we are increasingly using data analytics to make our processes more efficient and more productive. Recently, i360gov.com recognized the efforts of our Office of Disability Adjudication and Review (ODAR) in using data analytics to operate "one of the largest administrative judicial systems in the world." n1 ODAR has developed extensive and rigorous data analytics capabilities that allow it to identify patterns and areas for further examination of policy compliance and consistency. As i360gov.com noted, ODAR uses an "analytic feedback process" to lead to better results in the appeals process:

ODAR now captures key claims data, visualizes the results, analyzes those results and delivers feedback to managers and appellate judges, so the organization can change the policy, system, or advise personnel to take corrective steps based on what the data uncovers. The ability to analyze large and complex data sets using case analysis tools, data visualization, clustering analysis and multiple variable models allows ODAR to efficiently tackle the complex challenges faced daily in adjudicating disability appeals. Overall, ODAR has gained more consistency and accuracy in the processing of all appeals, along with the ability to process more claims, more quickly as well.

Building upon our successes in using predictive models and data analytics, we are undertaking a special initiative to expand our use of data analytics to enhance our ability to detect and prevent disability fraud. Specifically, we will apply analytical tools that can determine common characteristics and patterns of fraud based on data from past allegations and known cases of fraud. We will apply these tools when reviewing initial applications or existing data on beneficiaries for potential fraud or other suspicious behavior. With these tools, we expect to be able to identify suspicious patterns of activity in disability claims and prevent fraudulent applications from being processed. During this fiscal year, we plan to pilot these analytic tools and demonstrate their value.

We already have been proactive in using data analytics to detect and prevent possible fraud on our my Social Security portal. While the detected level of potentially fraudulent activity on my Social Security is low, as our Acting Commissioner has stated repeatedly, "we have no tolerance for fraud." In FY 2013, we created a new integrity review system and established a Fraud Analysis and Coordination Team (FACT) unit to analyze - using data analytics - suspicious behavior and potential fraud in our online services. The FACT unit takes necessary steps to mitigate any losses to SSA and to our customers.

Similarly, we are working with OIG on our data analytics projects to combat potential disability fraud. OIG's participation is valuable due to their knowledge of actual fraud cases that will inform our development of analytics processes. We continue to assess the potential extent to which data analytics will help in the fight against fraud. We will keep you apprised of our progress as we develop and pilot potential tools and applications.

Need for Adequate, Sustained Funding

We have a long history of delivering results in administering our vital programs in an efficient, cost-effective manner. We have been successful, in large part, due to our highly trained employees who are dedicated to serving our customers and being good stewards of the Social Security and SSI programs, and because we have made significant and strategic investments in IT that have allowed us to keep up with ever-increasing workloads. However, as the agency's CIO and Deputy Commissioner of Systems, I want to underscore the importance of receiving adequate, sustained resources to fund long-term strategic improvements to our IT infrastructure and the applications we use to administer our programs and to conduct program integrity and anti-fraud activities.

Adequate funding enables us to invest in tools and technology, which are vital for delivering quality service. Technology benefits our customers by providing more options to do business with us over the Internet or through self-service options. We must build upon the success of our online tools and my Social Security, which provides Internet users a secure way to do business with us. As we perfect these self-service options, we can add more business functions to them, which free our employees to focus on complex work and the customers who most need our help.

However, when our resources are significantly constrained, it may prove difficult for us to do anything more than maintain our current infrastructure or make marginal improvements. While the recent appropriations act helps, investments in IT require timely, adequate, multi-year funding. As I noted earlier, we have been recognized for our use of data analytics to administer one of the largest administrative judicial systems in the world. Our success in that regard was based in no small part on the infusion of resources from FY 2008 through FY 2010 when we received the full President's Budget request for those years and additional resources from the Recovery Act in FY 2009 and FY 2010.

Conclusion

Thank you for this opportunity to update you on how we use our IT operations to help administer our programs and for program integrity. We have a successful history of using IT to leverage our limited resources. Over the past few years, limited resources have challenged us to maintain our high-quality service to all of our customers. We believe we can do more. With adequate and sustained funding, we can hire and retain a highly skilled workforce, invest in the technology that will help us work smarter and faster, and deliver a quality return on investment to the American people.

n1 See Leaning In, http://www.i360gov.com/whitepapers/leaning-in-government-s-push-to-leverage-big-data/. (Last visited February 19, 2014)

Read this original document at: http://waysandmeans.house.gov/UploadedFiles/2_26_14_SS_Testimony_Bill_Zielinski.pdf