Administrative Renewal, Accuracy of Redetermination Outcomes, and Administrative Costs

Transcription

1 Administrative Renewal, Accuracy of Redetermination Outcomes, and Administrative Costs Stan Dorn Matthew Buettgens The Urban Institute Health Policy Center 2100 M St. NW Washington, DC October 2013

2 Copyright October The Urban Institute. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Permission is granted for reproduction of this file, with attribution to the Urban Institute.

3 Executive Summary Whenever a state Medicaid program has reliable information showing that a beneficiary coming up for renewal continues to qualify, that beneficiary should be administratively renewed, according to regulations from the Centers for Medicare and Medicaid Services that implement the Patient Protection and Affordable Care Act s broader statutory directives. Administrative renewal involves the state sending a notice that describes the basis for renewal and explains the beneficiary s legal duty to make any needed corrections. If none are forthcoming, coverage continues. It is not entirely clear, however, what constitutes reliable information that triggers administrative renewal. Beneficiaries with certain characteristics those who receive payments from the Supplemental Nutrition Assistance Program, for example, or who had prior-year income at specified levels are known to have specific, high probabilities of current Medicaid eligibility. What likelihood of eligibility is enough for administrative renewal? To help states answer that question, we analyze the impact of administrative renewal on the overall accuracy of redetermination outcomes and on administrative costs. Using a range of plausible assumptions, we find that, if a group of beneficiaries has an 80 percent or greater likelihood of current eligibility, administrative renewal will reduce the total number of mistaken redetermination outcomes. However, using administrative renewal causes the balance of mistakes to shift from incorrect terminations to incorrect renewals. Particularly as the likelihood of eligibility among a beneficiary group rises, administrative renewal also allows administrative savings by substantially reducing the number of redeterminations that must be done manually. States could thus achieve gains in overall accuracy and efficiency by applying administrative renewal to groups of beneficiaries who are known to have a very high probability of continuing eligibility. i

4 Contents Introduction... 1 A Hypothesis... 2 How the Hypothesis is Tested... 3 Identifying and assessing the importance of the relevant factors... 3 Making a range of reasonable assumptions... 7 Estimating the Effects of Administrative Renewal An example of how we obtain our results Our results Conclusion About the Authors and Acknowledgements About the Urban Institute About the California HealthCare Foundation Notes ii

5 Introduction The Patient Protection and Affordable Care Act (ACA) initiated many changes to Medicaid, including the expansion in eligibility to 138 percent of the federal poverty level (FPL), 1 which the U.S. Supreme Court effectively made optional for states. 2 Other important changes are not optional. In the past, state Medicaid programs typically required consumers to provide documentation as the first step towards establishing eligibility. Applicants often needed to provide pay stubs, for example, to prove low income. Recognizing the country s developing information technology (IT) capability, ACA Section 1413 made a major change in Medicaid s approach to verifying eligibility. Both with Medicaid and other insurance affordability programs, states must now, to the maximum extent practicable establish, verify, and update eligibility 3 based on information available in electronic form. Beginning in 2014, the Centers for Medicare and Medicaid Eligibility (CMS) will require Medicaid programs to start the verification process by exploring available data. Only if data matches fail to determine eligibility can states request documentation from consumers. To make that change possible, CMS authorized 90 percent federal funding for IT investments in Medicaid eligibility systems through the end of This shift to data-driven eligibility determination affects both initial applications and redeterminations of Medicaid eligibility. In the past, when enrollment periods were coming to an end, Medicaid programs would typically send beneficiaries a form requesting information about current household circumstances. To retain eligibility, beneficiaries needed to complete the form, attach a new set of pay stubs or other documentation, and return the form in a timely fashion. Those who failed to do so were terminated, regardless of eligibility. Such procedural terminations were a major feature of traditional Medicaid, affecting numerous eligible consumers, as explained below. In some cases, consumers who lost coverage would soon return to the program by reapplying. In other cases, people went without coverage. One study found that, six months after children and adults lost Medicaid, only 28 percent and 17 percent, respectively, returned to the program, while 43 percent and 49 percent remained uninsured. 5 Other research found that even a temporary gap without coverage substantially increased the likelihood of going without necessary health care. 6 To address this problem, CMS issued ACA regulations stating that Medicaid programs must make a redetermination of eligibility without requiring information from the individual if able to do so based on reliable information. 7 Starting in 2014, when data matches show that someone enrolled in Medicaid continues to qualify, the state must renew coverage and send the beneficiary a notice explaining the basis for the renewal. The notice must also explain the beneficiary s legal duty, enforced through criminal penalties, to correct any errors described in the notice. If no correction is forthcoming, the beneficiary s coverage continues. CMS terms this process, administrative renewal. States implementing this procedure face a practical question: When is information sufficiently reliable to warrant administrative renewal? Suppose data matches establish a 70 percent likelihood of eligibility. Is that enough for administrative renewal? Should states require an 80 or 90 percent likelihood of eligibility? Can states insist on 100 percent certainty? Should they? 1

6 These are not abstract topics. Other work shows that various data matches establish specific probabilities of current Medicaid eligibility. For example, in states that expand Medicaid to 138 percent FPL: Data showing receipt of benefits through the Supplemental Nutrition Assistance Program (SNAP) establishes a 98 percent likelihood of Medicaid eligibility for children and a 95 percent likelihood for adults; 8 Data showing prior-year income at or below 120 percent FPL establishes a 90 percent likelihood of current annual income at Medicaid levels; 9 and Data showing both prior-year income at or below 138 percent FPL and a new job begun between January 1 and May 1 of the current year establishes an 80 percent likelihood of current annual income at Medicaid levels. 10 Many states will want to select whatever probability threshold for administrative renewal yields the most accurate overall outcomes. But it is not obvious how to achieve such a result. As a matter of simple human nature, not limited to low-income health programs, many people avoid responding to notices and resist completing paperwork. 11 Accordingly, some beneficiaries will not respond to Medicaid renewal notices in a timely and appropriate fashion. Medicaid programs must therefore set default rules that determine what happens when this foreseeable lack of response occurs. With non-administrative renewal, the default, in the case of consumer inaction, is termination. With administrative renewal, the default is continued coverage. If the likelihood of eligibility for a group of beneficiaries is at or near 0 percent, more accurate outcomes result if the default is set at termination rather than renewal. 12 By the same token, if the likelihood of eligibility is at or near 100 percent, more accurate outcomes result if the default is set at renewal. 13 As the likelihood of eligibility rises from 0 percent to 100 percent, at some point along this spectrum accuracy will improve if the default shifts from termination to renewal. But at what point should this shift occur to achieve the most accurate overall balance of redetermination outcomes? The rest of this paper attempts to answer that question. It also analyzes the impact of administrative renewal on Medicaid administrative costs. A Hypothesis Suppose a beneficiary group has an 80 percent likelihood of Medicaid eligibility. That means 80 percent of its members are eligible and 20 percent are ineligible, so eligible people outnumber ineligible people by a ratio of 4 to 1. Likewise, if a group has a 90 percent likelihood of eligibility, the eligible outnumber the ineligible by a 9 to 1 ratio. With groups like these, administrative renewal provides many more chances to improve accuracy than to reduce accuracy. For each eligible person, administrative renewal prevents a possible procedural termination. Without administrative renewal, the eligible beneficiary might have failed to respond to the renewal notice and lost coverage. For each ineligible person, on the other hand, administrative renewal creates the risk of a mistaken outcome, because the ineligible beneficiary may fail to correct the state s renewal notice and so retain Medicaid. These basic facts suggest the following hypothesis: administrative renewal makes overall redetermination outcomes more accurate whenever the Medicaid agency knows that the vast majority of the group to which such renewal is applied 70, 80, or 90 percent qualifies for 2

7 Medicaid, because in such groups, administrative renewal creates many more chances to improve accuracy than to reduce accuracy. In the next section of the paper, we test this hypothesis. We also explore the impact of administrative renewal on administrative costs. How the Hypothesis is Tested Ideally, this hypothesis would be tested empirically by observing the application of administrative renewal and non-administrative renewal under the ACA to various beneficiary groups. Unfortunately, such observational results will not be available for some time. In the meantime, this paper estimates the potential impact of administrative and non-administrative renewal through the following, three-step process: 1. Identifying and assessing the importance of the factors that determine the overall accuracy of redetermination outcomes and the magnitude of certain key administrative costs; 2. Specifying a plausible continuum of assumptions for how each such factor will operate in 2014 and beyond; and 3. Showing a range of potential outcomes by calculating what would take place if all factors play out at one end of the continuum and then if all factors play out at the other end. This section of the paper addresses the first two steps in that process, which are very different from one another. The first step articulates precise formulas showing the percentage of erroneous redetermination outcomes and the percentage of manually conducted redeterminations that result from the use of administrative and non-administrative renewal, based on two inputs: the likelihood that the affected group of beneficiaries qualifies for Medicaid; and various assumptions about beneficiary behavior under the ACA and the accuracy of state eligibility decisions. By contrast, the second step, in which we develop those assumptions, is anything but precise. In particular, it is not yet known how beneficiaries will behave under the ACA. We use several strategies to address this uncertainty: We employ a broad range of assumptions about beneficiary behavior that seem plausible, based on current knowledge, so that our calculations will show a range of potential effects within which the actual results under the ACA are likely to fall; We deliberately bias several key assumptions modestly in the direction of favoring the accuracy of non-administrative renewal, so that we are likely to underestimate the potential benefits of administrative over non-administrative renewal in achieving accurate outcomes; and We explain our assumptions and calculations in great detail so that a reader who disagrees with our assumptions can substitute other assumptions and reach his or her own conclusions. The third step actually testing our hypothesis is addressed in the final section of the paper, where we calculate and compare the effects of administrative vs. non-administrative renewal. Identifying and assessing the importance of the relevant factors Non-administrative renewal In assessing the accuracy of non-administrative renewal, thereby allowing a comparison to administrative renewal s overall accuracy, several factors are key: 3

8 The percentage of eligible beneficiaries who satisfactorily return a renewal form and comply with all procedural requirements, as those who fail to do so will be procedurally terminated, despite their eligibility; The percentage of manual redetermination decisions that are made correctly by the Medicaid agency when a beneficiary completes the form and provides all requested information; and The percentage of beneficiaries, eligible and ineligible, who respond to renewal notices and cause the Medicaid program to make a manual redetermination decision. These factors are also important to administrative costs, of which manual redetermination decisions are an important component, since such decisions require public employees to analyze relevant facts and determine whether beneficiaries continue to qualify. We analyze these factors below, first in formulas and then in such formulas English-language equivalents. Formulas With non-administrative renewal, the preponderance of inaccurate outcomes and manual redeterminations within a given group can be expressed in formulas. Let P e =the percentage of the group that is eligible for Medicaid; P i =the percentage of the group that is ineligible for Medicaid; R e =the likelihood that an eligible consumer will satisfactorily return and complete a nonadministrative renewal form; R i =the likelihood that an ineligible consumer will satisfactorily return and complete a non-administrative renewal form; A t =the percentage of agency manual redetermination decisions that are made accurately; and A f =the percentage of agency manual redetermination decisions that are made inaccurately. The percentage of a beneficiary group that is procedurally terminated under non-administrative renewal, despite their eligibility, is P e x (1- R e ). 14 Other mistaken outcomes involve inaccurate decisions made manually by the Medicaid agency. For eligible and ineligible consumers, respectively, such incorrect outcomes affect the following percentages of the beneficiary group: P e x R e x A f ; and P i x R i x A f. 15 The percentage of redeterminations for the beneficiary group that involves manual redeterminations is (P e x R e ) + (P i x R i ). 4

9 Plain language Non-administrative renewal results in two kinds of mistaken outcomes. First, some eligible consumers are procedurally terminated. Within a group of beneficiaries subject to nonadministrative renewal, the proportion who are eligible consumers receiving such terminations is the product of two factors: the percentage of beneficiaries who qualify; and the percentage of eligible beneficiaries who fail to respond appropriately to their non-administrative renewal notices and so are terminated. The second kind of mistaken outcome involves beneficiaries who respond to renewal notices, thereby requesting redetermination, but have the Medicaid agency reach the wrong conclusion. Within a group of beneficiaries, such outcomes include both terminations and renewals: The percentage of beneficiaries who receive mistaken manual terminations is the product of the percentage of beneficiaries who are eligible; the percentage of eligible beneficiaries who respond to their non-administrative renewal notices in timely, appropriate fashion sufficient to obtain a manual redetermination; and the percentage of manual redetermination outcomes that are mistaken. The percentage of beneficiaries who receive mistaken manual renewals is a product of the percentage of beneficiaries who are ineligible; the percentage of ineligible beneficiaries who respond sufficiently to their non-administrative renewal notices to obtain a manual redetermination; and the percentage of manual redetermination outcomes that are mistaken. When a group of beneficiaries is subject to non-administrative renewal, the percentage of manual redeterminations is a sum of the following: The percentage of beneficiaries who are eligible, multiplied by the percentage of eligible beneficiaries who respond to their non-administrative renewal notices and obtain a manual redetermination; plus The percentage of beneficiaries who are ineligible, multiplied by the percentage of ineligible beneficiaries who respond to their non-administrative renewal notices and obtain a manual redetermination. Administrative renewal To determine the likelihood of mistaken administrative renewals, the following factors are essential to analyze: The percentage of ineligible beneficiaries who will correct the state s erroneous notice, as coverage will mistakenly continue via procedural renewal for those who fail to make corrections; The percentage of manual redetermination decisions that are made correctly by the Medicaid agency; and The proportion of beneficiaries who provide information in response to renewal notices, thereby causing the agency to make manual redetermination decisions. As with non-administrative renewal, we analyze administrative renewal in terms of both formulas and their plain language equivalents. 5

10 Formulas Formulas can show, for a group of beneficiaries, the proportion of incorrect outcomes and manual redetermination decisions that will result from administrative renewal. Let AR e = the likelihood that an eligible consumer will correct the information provided on the administrative renewal notice; and AR i = the likelihood that an ineligible consumer will correct the information provided on the administrative renewal notice. The percentage of a beneficiary group that is ineligible but administratively renewed is P i x (1- AR i ). Other mistaken outcomes involve inaccurate decisions made manually by the Medicaid agency. For eligible and ineligible consumers, respectively, such incorrect outcomes affect the following percentages of the beneficiary group: P e x AR e x A f ; and P i x AR i x A f. The percentage of manual redeterminations made under administrative renewal is: (P e x AR e ) + (P i x AR i ). Plain language Much like non-administrative renewals, administrative renewals results in two kinds of mistaken outcomes. First, some ineligible consumers are procedurally renewed. Within a group of beneficiaries subject to administrative renewal, the proportion who are ineligible consumers receiving such renewals is the product of two factors: the percentage of beneficiaries who are ineligible; and the percentage of ineligible beneficiaries who fail to correct their administrative renewal notices and so are renewed, despite their ineligibility. The second kind of mistaken outcome involves beneficiaries who respond to renewal notices by providing information that corrects information on the notices, thereby triggering a manual redetermination, but have the Medicaid agency reach the wrong conclusion about eligibility. Within a group of beneficiaries, such outcomes include both terminations and renewals: The percentage of beneficiaries who receive mistaken manual terminations is the product of the percentage of beneficiaries who are eligible; the percentage of eligible beneficiaries who correct their administrative renewal notices and trigger a manual redetermination; and the percentage of manual redetermination outcomes that are mistaken. The percentage of beneficiaries who receive mistaken manual renewals is a product of the percentage of beneficiaries who are ineligible; the percentage of ineligible beneficiaries who correct their administrative renewal notices, triggering a manual redetermination; and the percentage of manual redetermination outcomes that are mistaken. When a group of beneficiaries is subject to administrative renewal, the percentage of manual redeterminations is a sum of the following: 6

11 The percentage of beneficiaries who are eligible, multiplied by the percentage of eligible beneficiaries who correct their administrative renewal notices, triggering a manual redetermination; plus The percentage of beneficiaries who are ineligible, multiplied by the percentage of ineligible beneficiaries who correct their administrative renewal notices, leading to a manual redetermination. Making a range of reasonable assumptions In estimating the impact of administrative renewal on mistaken outcomes, we generally select assumptions that involve relatively broad range of possible behaviors, particularly when it comes to issues without close empirical analogies from past observations. In some cases, we also err on the side of favoring the comparative accuracy of non-administrative renewal over administrative renewal. The first key assumption is the percentage of eligible beneficiaries who will respond to nonadministrative renewal notices and so avoid procedural terminations. In the past, many eligible beneficiaries failed to respond to such notices and lost coverage. Research examining this issue found that roughly 13 to 36 percent of child beneficiaries 16 and at least 30 to 40 percent of adult beneficiaries 17 who remained eligible nevertheless lost coverage annually, before the ACA. However, procedural terminations with non-administrative renewal should be less common under the ACA s new rules. For example, CMS requires, with non-administrative renewal, that forms sent to beneficiaries must be prepopulated with information known to the Medicaid agency (such as address and social security number); states cannot request information unless it is essential to redetermining eligibility; beneficiaries must be allowed to provide information in person, online, by phone, or by mail; and that beneficiaries must be allowed to cure procedural failures by furnishing missing information soon after receiving a termination notice. 18 As before the ACA, coverage will end if beneficiaries fail to respond correctly to a renewal notice. However, the ACA s simplification of non-administrative renewal procedures is likely to reduce the number of procedural terminations that result. Taking into account both the rate of procedural terminations in Medicaid before the ACA as well as the improvements made by the ACA, we assume that, without administrative renewal, between 15 percent and 25 percent of eligible consumers will fail to respond properly to state renewal notices under the ACA and so will be terminated for procedural reasons. This assumes that the ACA will have a considerable positive impact lowering the prevalence of procedural terminations below the levels observed in the past. If such a significant improvement fails to materialize, and more than 25 percent of beneficiaries receiving non-administrative renewal notices fail to respond, administrative renewal will prevent more procedural terminations of eligible consumers than we calculate, hence achieve greater improvements in accuracy than under our estimates. A second key assumption is the proportion of ineligible people who will correct the state s administrative renewal notices. In developing that assumption, we know something about those notices structure. As with all eligibility-related communications, corrections can be made by phone or on-line, so the process of providing information should not be burdensome for the consumer. Further, such notices are likely to stress the criminal punishments that can attach to failing to provide necessary correction, providing some motivation for response. Given that context, we assume that between 40 percent and 55 percent of ineligible consumers will correct 7

12 renewal notices and notify the state of facts that are inconsistent with the information on those notices. Without close empirical precedent, we have tried to choose a relatively broad range of reasonable assumptions, tilted towards a likely underestimate of consumer responses to erroneous notices. If in fact more than 55 percent of ineligible consumers correct the mistaken renewal notices they receive, administrative renewal will produce fewer mistaken outcomes than the results we report below. Our third key assumption the likelihood that caseworker redeterminations will result in a correct decision is that 96.7 percent of all manual eligibility determinations will be accurate, reflecting the 3.3 percent national average eligibility error rate found by CMS s payment error review report for fiscal year Our fourth and fifth assumptions primarily affect administrative costs but also have an incidental effect on accuracy. The fourth assumption is the percentage of ineligible beneficiaries who, under non-administrative renewal, will complete renewal forms and attempt to demonstrate their continued qualification for Medicaid, despite their ineligibility. The fifth is the percentage of eligible beneficiaries who, with administrative renewal, will respond to the state s renewal notice by providing the state with information about changed household circumstances, even though they continue to qualify. Each of these actions will cause a manual redetermination of eligibility, thus consuming administrative resources and creating the chance for a mistaken decision. On these issues, in the absence of empirical data, we make a very broad range of assumptions to test the range of administrative renewal s potential effects. We assume that between 15 and 30 percent of beneficiaries will take these actions, affirmatively responding to renewal notices in ways that run counter to beneficiaries actual eligibility status. Table 1 shows these assumptions, including how they relate to the above formulas. The table also classifies the assumptions, most of which represent opposite ends of various ranges, as either more or less favorable to administrative renewal s accuracy, compared to nonadministrative renewal. 8

13 Table 1. Key Assumptions and their Favorability to Administrative Renewal s Comparative Accuracy Assumptions that are more favorable to administrative renewal Assumptions that are less favorable to administrative renewal With non-administrative renewal With administrative renewal With all renewals The percentage of eligible beneficiaries who satisfactorily complete renewal forms (R e ) The percentage of ineligible beneficiaries who satisfactorily complete renewal forms (R i ) The percentage of eligible beneficiaries who correct the state s renewal notices (AR e ) The percentage of ineligible beneficiaries who correct the state s renewal notices (AR i ) The percentage of manual redeterminations that are correct (A t ) 75 percent 30 percent 15 percent 55 percent 96.7 percent 85 percent 15 percent 30 percent 40 percent 96.7 percent Note: Assumptions that are more favorable to administrative renewal increase the likelihood that administrative renewal will result in more accurate overall redetermination outcomes than non-administrative renewal. Assumptions that are less favorable to administrative renewal do the reverse. 9

14 Estimating the Effects of Administrative Renewal The first part of this section provides a step-by-step example of how we calculate our results, using the above formulas and assumptions. The second part sets out our findings. An example of how we obtain our results Here, we analyze a group of beneficiaries with a 60 percent likelihood of eligibility (P e = 60%, P i = 40%), using assumptions that are less favorable to administrative renewal s comparative accuracy. This example illustrates the calculations that we make for other groups of beneficiaries under assumptions that are both more and less favorable to administrative renewal. In terms of our formulas, the less favorable assumptions that we use in this example are as follows (table 1): With non-administrative renewal, renewal forms are satisfactorily completed by 85 percent of eligible beneficiaries (R e = 85%); and 15 percent of ineligible beneficiaries (R i = 15%). With administrative renewal, the percentage of beneficiaries who correct the state s renewal forms is 30 percent for eligible beneficiaries (AR e = 30%); and 40 percent for ineligible beneficiaries (AR i = 40%) percent of manual redeterminations are correct (A t= 96.7%, A f= 3.3%). Based on those assumptions and the formulas set forth above, the following calculations show how we find (rounded to the nearest percentage point) that: Mistaken outcomes would affect 11 percent of this group of beneficiaries under non-administrative renewal and 25 percent of them under administrative renewal; and Manual redeterminations would be required for 57 percent of these beneficiaries under nonadministrative and 34 percent under administrative renewal. Mistaken outcomes under non-administrative renewal As noted earlier in our explanation of formulas, mistaken procedural terminations affect the following percentage of any given beneficiary group under non-administrative renewal: P e x (1- R e ). Mistaken manual determinations, for eligible and ineligible beneficiaries, respectively, are the following proportions of a given beneficiary group: P e x R e x A f ; and P i x R i x A f. For this particular beneficiary group namely, a group where P e =60% and P i =40% and for the assumptions set out above, those three formulas are solved as follows: 60% x (1-85%) = 60% x 15%=9%; 10

15 60% x 85% x 3.3% = 1.7%; and 40% x 15% x 3.3% = 0.2%. They sum to 10.9 percent. Rounded off to the nearest percentage point, 9 percent of these beneficiaries have a mistaken termination outcome that involves procedural termination, 2 percent have a mistaken manual termination, and 0 percent have a mistaken manual renewal, for a total of 11 percent mistaken outcomes (table 2). Mistaken outcomes under administrative renewal With administrative renewals, mistaken procedural renewals affect the following percentage of any given beneficiary group, according to the formulas set out above: P i x (1- AR i ). Mistaken manual determinations, for eligible and ineligible beneficiaries, respectively, are the following proportions of the beneficiary group: P e x AR e x A f ; and P i x AR i x A f. For this particular beneficiary group and the assumptions we apply in this example, those three equations are solved as follows: 40% x (1-40%) = 40% x 60%=24%; 60% x 30% x 3.3% = 0.6%; and 40% x 40% x 3.3% = 0.5%. They sum to 25.1 percent. Rounded off to the nearest percentage point, 24 percent of these beneficiaries have a mistaken procedural renewal, 1 percent has a mistaken manual termination, and 1 percent has a mistaken manual renewal, for a total of 25 percent mistaken outcomes (table 2). (Totals fail to sum because of rounding.) Manual redeterminations and administrative costs In terms of administrative costs, the percentage of manual redeterminations with non-administrative renewal is (P e x R e ) + (P i x R i ). For administrative renewal, that percentage is (P e x AR e ) + (P i x AR i ). With this set of beneficiaries and assumptions, those two formulas are solved as follows: (60% x 85%) + (40% x 15%) = 51% + 6% = 57%; and (60% x 30%) + (40% x 40%) = 18% + 16% = 34%. Accordingly, 57 percent of redeterminations are manual with non-administrative renewal and 34 percent are manual when renewal is administrative (table 5). 11

16 Our results Using the methods illustrated in the previous section, we applied our formulas to beneficiaries known to have a 60 percent, 70 percent, 80 percent, 85 percent, and 90 percent likelihood of eligibility. We calculated the results under assumptions both more and less favorable to the relative accuracy of administrative renewal. Our findings are set forth below. Findings about mistaken redetermination outcomes Under less favorable assumptions, administrative renewal reduces overall the accuracy of redetermination if it applies to beneficiaries who have less than an 80 percent likelihood of eligibility. Figure 1. Percentage of Mistaken Redetermination Outcomes Based on Assumptions that are Less Favorable to Administrative Renewal s Comparative Accuracy Non-administrative renewal Administrative renewal 25% 19% 11% 13% 14% 13% 15% 10% 16% 7% 60 percent 70 percent 80 percent 85 percent 90 percent Likelihood of eligibility among beneficiaries being renewed How to read this figure: Among a group of beneficiaries who are known to have a 60 percent likelihood of eligibility, 11% will experience mistaken redetermination outcomes if non-administration renewal is used and 25% will experience such outcomes if administrative renewal is used (based on assumptions that are relatively unfavorable to administrative renewal s comparative accuracy); among a group of beneficiaries known to have a 70 percent likelihood of eligibility, 13% and 19% will experience mistaken redetermination outcomes if nonadministrative and administrative renewal methods are used, respectively (based on those same assumptions); etc. Note: For assumptions, see table 1. For example, if 60 percent of beneficiaries qualify for Medicaid, administrative renewal raises the proportion of mistaken outcomes from 11 percent to 25 percent (figure 1), which represents a 131 percent relative increase (table 4). If it is used with a group of beneficiaries of whom 70 percent are eligible, administrative renewal increases the proportion of mistaken outcomes from 13 percent to 19 percent (figure 1) a 51 percent relative increase (table 4). Only when 80 percent or more of beneficiaries qualify does administrative renewal begin reducing the prevalence of mistaken outcomes. Incorrect results fall 12

17 from 14 percent to 13 percent of beneficiaries, if administrative renewal is used with a group 80 percent of whose members are eligible; from 15 to 10 percent, if 85 percent are eligible; and from 16 percent to 7 percent, if 90 percent are eligible (figure 1). These represent 9, 34, and 56 percent relative reductions in mistaken outcomes, respectively (table 4). With assumptions that are more favorable, administrative renewal begins improving overall accuracy as soon as the proportion of eligible beneficiaries reaches 70 percent (see figure 2): If 60 percent of beneficiaries qualify for Medicaid, administrative renewal increases the proportion of mistaken outcomes from 17 percent to 19 percent (figure 2); but If 70 of beneficiaries qualify, administrative renewal cuts the proportion of mistaken outcomes from 20 percent to 14 percent (figure 2), representing a 26 percent relative reduction (table 4). As the percentage of eligible beneficiaries rises above 70 percent, the comparative advantage of administrative renewal grows, finally yielding a 79 percent relative reduction in mistaken outcomes when used with a group of beneficiaries 90 percent of whom qualify for Medicaid (table 4). Figure 2. Percentage of Mistaken Redetermination Outcomes Based on Assumptions that are More Favorable to Administrative Renewal s Comparative Accuracy Non-administrative renewal Administrative renewal 17% 19% 20% 14% 22% 24% 25% 10% 7% 5% 60 percent 70 percent 80 percent 85 percent 90 percent Likelihood of eligibility among beneficiaries being renewed How to read this figure: Among a group of beneficiaries who are known to have a 60 percent likelihood of eligibility, 17% will experience mistaken redetermination outcomes if non-administration renewal is used and 19% will experience such outcomes if administrative renewal is used (based on assumptions that are relatively favorable to administrative renewal s comparative accuracy); among a group of beneficiaries known to have a 70 percent likelihood of eligibility, 20% and 14% will experience mistaken redetermination outcomes if nonadministrative and administrative renewal methods are used, respectively (based on those same, more favorable assumptions); etc. Note: For assumptions, see table 1. As figures 1 and 2 illustrate, regardless of what assumptions are used administrative renewal increases overall accuracy once the vast majority of beneficiaries 80, 85, or 90 percent are known to qualify. 13

18 Using administrative renewal also changes the nature of most incorrect outcomes from terminations to renewals, however, even if the total number of incorrect outcomes declines. For example, table 2 shows that, under unfavorable assumptions, administrative renewal reduces the proportion of incorrect outcomes from 16 percent to 7 percent when such renewal is used with a group of beneficiaries who have a 90 percent likelihood of continuing eligibility. But all the mistakes that result from nonadministrative renewal involve terminations of eligible beneficiaries, whereas most of administrative renewal s mistakes involve renewals of the ineligible. The consequences of such mistakes are asymmetric. An incorrect termination of Medicaid eligibility typically means that a beneficiary who should continue to receive Medicaid instead becomes uninsured, at least temporarily. As explained earlier, this may result in reduced access to health care, a later reapplication for Medicaid coverage that consumes administrative resources, or both. An incorrect renewal of Medicaid means that someone who should receive subsidies in a health insurance marketplace or an offer of employer-sponsored insurance that the ACA classifies as affordable will instead continue in Medicaid. How state policymakers regard these two results may influence how they view administrative renewal. Neither mistaken procedural terminations prevented by administrative renewal nor mistaken procedural renewals caused by administrative renewal should lead to federal sanctions under CMS s payment error regulations. Provided the state followed applicable procedures and legal requirements, CMS will not find error if, because of factors unknown to the state, redetermination resulted in a mistaken outcome. 20 Maximizing the use of administrative renewal may lessen the risk of federal error findings by reducing opportunities for human error. Nevertheless, state policymakers who want only the eligible to be renewed and only the ineligible to be terminated, consistent with the wise use of administrative resources, will limit administrative renewal to circumstances in which it is likely to maximize correct redetermination outcomes. This analysis suggests that such circumstances involve beneficiary groups with an 80 percent or greater likelihood of eligibility. Findings about administrative costs and savings By reducing the number of redeterminations that are conducted manually, administrative renewal lowers administrative costs. The extent of that reduction depends on the characteristics of the affected population. These savings grow when more beneficiaries are eligible. 21 For example, under assumptions less favorable to administrative renewal (table 5), administrative renewal reduces the number of manual redeterminations by 40% if the beneficiaries to whom such renewal applies have a 60 percent likelihood of eligibility; 48% if beneficiaries have a 70 percent likelihood of eligibility; 55% if beneficiaries have an 80 percent likelihood of eligibility; and 60% if they have a 90 percent likelihood of eligibility. One should not confuse administrative costs with total costs. Per beneficiary, health care costs greatly exceed administrative expenses. A state may be able to lower its total Medicaid spending by erecting procedural barriers that raise administrative costs but prevent eligible consumers enrollment. Many policymakers, however, see efficiency as its own important objective, because it lowers the proportion of taxpayer dollars spent on administration and increases the proportion spent helping people in need. 14

19 Table 2. The Percentage of Mistaken Redetermination Outcomes within Various Beneficiary Groups, Based on Assumptions Less Favorable to Administrative Renewal s Comparative Accuracy Non-administrative Renewal Administrative Renewal Likelihood of eligibility within Mistaken Mistaken Mistaken terminations Total Mistaken renewals Total renewals terminations beneficiary group mistaken mistaken Procedural Manual Manual Manual Procedural Manual (P e ) outcomes outcomes terminations terminations renewals terminations renewals renewals 60 percent 9% 2% 0% 11% 1% 24% 1% 25% 70 percent 11% 2% 0% 13% 1% 18% 0% 19% 80 percent 12% 2% 0% 14% 1% 12% 0% 13% 85 percent 13% 2% 0% 15% 1% 9% 0% 10% 90 percent 14% 3% 0% 16% 1% 6% 0% 7% Note: See table 1 for the assumptions that are more and less favorable to administrative renewal s comparative accuracy. Procedural terminations and renewals take place when beneficiaries do not respond to renewal notices. Totals may not add because of rounding. Table 3. The Percentage of Mistaken Redetermination Outcomes within Various Beneficiary Groups, Based on Assumptions More Favorable to Administrative Renewal s Comparative Accuracy Non-administrative Renewal Administrative Renewal Likelihood of eligibility within Mistaken Mistaken Mistaken terminations Total Mistaken renewals Total renewals terminations beneficiary group mistaken mistaken Procedural Manual Manual Manual Procedural Manual (P e ) outcomes outcomes terminations terminations renewals terminations renewals renewals 60 percent 15% 1% 0% 17% 0% 18% 1% 19% 70 percent 18% 2% 0% 20% 0% 14% 1% 14% 80 percent 20% 2% 0% 22% 0% 9% 0% 10% 85 percent 21% 2% 0% 24% 0% 7% 0% 7% 90 percent 23% 2% 0% 25% 0% 5% 0% 5% Note: See table 1 for the assumptions that are more and less favorable to administrative renewal s comparative accuracy. Procedural terminations and renewals take place when beneficiaries do not respond to renewal notices. Totals may not add because of rounding. 15

20 Table 4. Relative Impact of Administrative Renewal and Non-administrative Renewal on the Percentage of Redetermination Outcomes that are Mistaken Likelihood of eligibility within beneficiary group (P e ) Mistaken Outcomes, under Less Favorable Assumptions Administrative renewal Relative difference Mistaken Outcomes, under More Favorable Assumptions Administrative renewal Relative difference 60 percent 11% 25% +131% 17% 19% +13% 70 percent 13% 19% +51% 20% 14% -26% 80 percent 14% 13% -9% 22% 10% -56% 85 percent 15% 10% -34% 24% 7% -68% 90 percent 16% 7% -56% 25% 5% -79% Note: See table 1 for the assumptions that are more and less favorable to administrative renewal s comparative accuracy. See tables 2 and 3 for the percentages of total redetermination outcomes that are mistaken, for each beneficiary group, under assumptions that are less and more favorable to administrative renewal s comparative accuracy. Table 5. Relative Impact of Administrative Renewal and Non-administrative Renewal on the Percentage of Redeterminations that are Manually Conducted Likelihood of eligibility within beneficiary group (P e ) Manual Redeterminations, under Less Favorable Assumptions Administrative renewal Relative difference Manual Redeterminations, under More Favorable Assumptions Nonadministrative renewal Nonadministrative renewal Nonadministrative renewal Nonadministrative renewal Administrative renewal 60 percent 57% 34% -40% 57% 31% -46% 70 percent 64% 33% -48% 62% 27% -56% 80 percent 71% 32% -55% 66% 23% -65% 85 percent 75% 32% -58% 68% 21% -69% 90 percent 78% 31% -60% 71% 19% -73% Note: See table 1 for the assumptions that are more and less favorable to administrative renewal s comparative accuracy. Relative difference 16

21 Conclusion If a state Medicaid program knows that, within a particular group of Medicaid beneficiaries, 80 percent, 85 percent, or 90 percent of enrollees continue to qualify, then applying administrative rather than non-administrative renewal is likely to increase the overall accuracy of redetermination outcomes. Some ineligible people will fail to correct the state s renewal notices and retain coverage, but procedural terminations will be prevented for a larger number of eligible people. Administrative costs will fall considerably, since many fewer redeterminations will need to be conducted manually. If it is applied to beneficiary groups where the likelihood of eligibility reaches these high levels, administrative renewal is likely to yield overall gains in accuracy and efficiency. 17

22 About the Authors and Acknowledgements Stan Dorn and Matthew Buettgens are a senior fellow and senior research associate, respectively, at The Urban Institute s Health Policy Center. They are grateful to the California HealthCare Foundation for its support of this research and the insightful advice of Catherine Teare, our project officer. In addition to Catherine Teare, the authors thank Lisa Dubay, John Holahan, and Steve Zuckerman of The Urban Institute, and Beth Morrow of The Children s Partnership, who reviewed earlier drafts of this report and provided helpful feedback. None of these individuals or organizations are responsible for the opinions expressed in this report, which are the authors responsibility. About the Urban Institute The Urban Institute is a nonprofit, nonpartisan policy research and educational organization established in Washington, D.C., in Its staff investigates the social, economic, and governance problems confronting the nation and evaluates the public and private means to alleviate them. About the California HealthCare Foundation The California HealthCare Foundation, based in Oakland, California, works as a catalyst to fulfill the promise of better health care for all Californians. We support ideas and innovations that improve quality, increase efficiency, and lower the costs of care. Notes 1 Five FPL percentage points are subtracted from modified adjusted gross income (MAGI) in determining Medicaid eligibility, under rules that go into effect in Accordingly, the net income standard for Medicaid eligibility, in states that implement the expansion, is 133 percent FPL, but the gross income standard is 138 percent FPL. 2 National Federation of Independent Business v. Sebelius, 132 S. Ct (2012). 3 ACA 1413(c)(3)(A). 4 4 CMS. Federal Funding for Medicaid Eligibility Determination and Enrollment Activities final rule, April 19, 2011 Federal Register (76 FR 21950). See also CMS. Guidance for Exchange and Medicaid Information Technology (IT) Systems, Version 2.0, May 2011, MIT/Downloads/exchangemedicaiditguidance.pdf. 5 Benjamin D. Sommers Loss of Health Insurance Among Non-elderly Adults in Medicaid. Journal of General Internal Medicine, 24. no. 1 (2008): C. Schoen and C. DesRoches. Uninsured and Unstably Insured: The Importance of Continuous Coverage, Health Services Research 35, no. 1 pt. 2(April 2000): CFR (a)(2). 8 Stan Dorn, Laura Wheaton, Paul Johnson, and Lisa Dubay. Using SNAP Receipt to Establish, Verify, and Renew Medicaid Eligibility (Washington, DC: The Urban Institute, April 2013) 9 Stan Dorn, Matthew Buettgens, Habib Moody, and Christopher Hildebrand. Using past income data to verify current Medicaid eligibility. Prepared by the Urban Institute for the California HealthCare Foundation, October The same study is the source of the subsequent bullet and the finding in its accompanying end note. 10 To continue through the final months of the calendar year, data showing a combination of prior-year income at or below 120 percent FPL, and no new job begun between January 1 and September 1 of the current year, and quarterly wages earlier during the current year above 80 percent FPL establishes a 70 percent likelihood of current annual income at Medicaid levels. 18

23 11 For example, when a new employee must complete a form to sign up for a 401(k) retirement savings account, 33 percent participate. In firms where new employees must complete forms to opt out of such accounts, participation levels reach 90 percent. David Laibson. Impatience and Savings. National Bureau of Economic Research Reporter, (Fall 2005): 6 8. In another example, drivers on opposite sides of the New Jersey-Pennsylvania border have diametrically opposed patterns of enrollment into various types of auto insurance, because the two states laws automatically place motorists into different types of insurance unless they complete a form requesting the other category of coverage. Cass R. Sunstein and Richard H. Thaler. Libertarian Paternalism is Not an Oxymoron. The University of Chicago Law Review. 70, no. 4 (Fall 2003): In such a group, everyone or nearly everyone is ineligible, so those who fail to respond to renewal notices will tend to be ineligible. Fewer mistakes will result if they lose rather than retain coverage. 13 With such a group, those who fail to respond will tend to be eligible, so fewer mistakes will result if the default is set at renewal, rather than termination. 14 We use x rather than * to indicate multiplication. 15 To simplify our calculations, we assume that a manual redetermination has the same probability of error, whether it involves an eligible beneficiary, an ineligible beneficiary, administrative renewal, or non-administrative renewal. Varying the probability of mistaken manual redeterminations with these changing contexts could produce different results, depending on the extent of the assumed variation. 16 One study found that among children enrolled in Medicaid and CHIP in , 12.6 percent lost coverage during the average year even though they continued to be eligible. Benjamin D. Sommers. From Medicaid to Uninsured: Drop-Out among Children in Public Insurance Programs. Health Services Research 40, no. 1 (2005): Another study found that, among children who were uninsured in 2008 but qualified for Medicaid or CHIP, 35.5 percent had been enrolled in Medicaid or CHIP the previous year but then lost coverage. Benjamin Sommers. Enrolling Eligible Children in Medicaid and CHIP: A Research Update. Health Affairs 29, no. 7 (2010): A study that examined adult Medicaid beneficiaries in concluded that even among adults who were in unusually stable eligibility categories that made them always eligible, 29 percent lost Medicaid during the average year; altogether, 43 percent of Medicaid adults lost coverage during the average year, with no statistically significant difference between those whose incomes rose and those whose incomes fell; over a two-year period, 53 percent of adult beneficiaries whose incomes declined nevertheless lost Medicaid; and that, controlling for other factors, Medicaid disenrollment was 75 percent more likely for adults than for children. Sommers, Loss of Health Insurance Among Non-elderly Adults in Medicaid. 18 CMS. Medicaid Program; Eligibility Changes Under the Affordable Care Act of 2010; Final Rule. Federal Register 77, no. 57 (March 23, 2012): 17144, If in fact the accuracy rate is lower, then our comparison will understate any comparative advantage achieved by administrative renewal in preventing erroneous outcomes, since administrative renewal will reduce the number of caseworker redeterminations. CMS, FY 2011 Louisiana Medicaid: Payment Error Rate Measurement (PERM) Cycle 3 Summary Report, (Washington, DC: US Department of Health and Human Services 2012). 20 As CMS explained in March 2012, States that follow procedures outlined in the regulations will not be cited for a PERM error for lack of further documentation. PERM regulations issued in 2010 provide that PERM will measure errors relative to the State s own policies and procedures as long as those policies and procedures are consistent with Federal policy and regulations. CMS, Medicaid Program; Eligibility Changes, 17144, With non-administrative renewal, an eligible person is more likely than an ineligible person to complete the state s renewal form and request a manual redetermination. Such redeterminations are thus more frequent, with nonadministrative renewal, if eligible consumers are more common. The reverse is true with administrative renewal. When sent a renewal notice stating that the Medicaid program has found the beneficiary eligible and describing the basis for the state s finding, an eligible person is less likely than an ineligible person to correct that notice and trigger a manual redetermination of eligibility. As a result, redeterminations grow less frequent, with administrative renewal, when eligible consumers are more numerous. In sum, as eligibility becomes more likely, manual redeterminations and resulting administrative costs rise for non-administrative renewal and fall for administrative renewal, thus increasing the relative savings achieved by administrative renewal. 19