Projecting Enrollment in a Medicaid Buy-In Program for Ohio

Executive Summary

The purpose of this report is to present a statistical model for estimating the number of Ohioans who would enroll in a Medicaid buy-in program and to use that model to study the effect of different eligibility criteria and different policies concerning premiums. The model was developed data for Ohio from the Current Population Survey (CPS).

About 10% of Ohioans ages 18-64 have a work disability, or 668,480 people. Of these, 45% showed some evidence of working or having worked recently, or at least of having wanted to work recently. Only 16% reported that they had no health insurance during the previous calendar year. Just over half of the 668,480 persons with a work disability had some form of disability income during the previous year (52%).

While a variety of different sets of rules were considered for determining what families would be eligible for an Ohio Medicaid buy -in program, the final recommendation of the Ticket to Work Study Committee of the Ohio Legislature was as follows: family income under the program would be capped at 250% of the Federal Poverty Level (FPL); and persons with a severe work disability would be given an earned income disregard of up to $20,000/ year.

The Ticket to Work Study Group also recommended that eligibility for the program be restricted to persons with assets of less than $10,000 limit (with exclusions for the following kinds of assets: primary residence, automobile, qualified retirement accounts, and medical savings accounts.

A total of 442,534 Ohioans 18 to 64 have a severe work disability and have incomes and assets that that would potentially qualify them for a Medicaid buy -in program in Ohio. The current employment rate for these individuals is only 7.9%. Under the assumptions used in developing the model, the employment rate is expected to increase to 14.7%.

The Ticket to Work Study Group in Ohio recommended the following options regarding premiums: Premiums would begin with the first dollar of income above 150% of the Federal Poverty level and would be for 10% of the income in excess of the starting level.

Assumed take-up rates in the model varied widely depending upon whether or not a person currently receives Medicaid, whether they currently have any form of health insurance, and whether they would pay a premium. Depending on their circumstances take-rates for the new program varied from a low of 5% to a high of 70%.

It is projected that enrollment in an Ohio Medicaid program might eventually reach 12,542 persons. Costs to the state for such a program will consist primarily of the new monthly premiums for people not currently on Medicaid. The state share of these premiums (42%) was estimated at about $20 million.

It is critically important to note that this report does not predict how many people might enroll in the program in the few years of program operation. Rather it projects that under a set of assumptions laid out in detail in the following material, a total of 12, 542 might eventually enroll.

The model was evaluated by comparing it's results to the actual experiences of two other states, Minnesota and Wisconsin, both of which have had operating programs for one year or longer. Based on the results, the present model appears to be conservative.

Introduction

Beginning in 2002, simply being employed will no longer be a disqualification for the continuation of Social Security Disability Insurance (SSDI) benefits or Supplemental Security Income (SSI) benefits. While being gainfully employed will continue to be a disqualification for a person making an initial application for SSDI or SSI, states may declare an employed person to be eligible for Medicaid under a new buy-in program if the person meets all of the other standards for disability that are part of the SSDI/SSI determination process.

Thus, a Medicaid buy -in program can benefit three distinct types of people:

• People who currently receive SSDI or SSI and are unable to work because they are dependent upon Medicaid for health insurance.
• People who work and currently receive SSDI or SSI, and who must be careful to keep their earned income below certain levels in order not to lose their Medicaid eligibility.
• People who work and whom Medicaid does not cover but whose disability status makes it difficult or impossible for them to obtain adequate health insurance.

The following case studies - provided by the staff of the Ohio Rehabilitation Services Commission - illustrate the need and value of a Medicaid buy-in program in Ohio. The names used are fictitious but the stories are real.

• Ann is a 39-year-old single mother of a 3-year old. Ann has multiple disabling conditions. She is on SSDI and receives a check for $873 each month. She has Medicare and Medicaid with a $409 spend-down (meaning that she must document that she has paid $409 in medical expenses each month before Medicaid pays for anything). Ann's Medicaid coverage is critical since her prescriptions alone cost more than $1,000 every month. Ann has completed computer and office administration training. She found a job paying $8.00 an hour for 25 hours a week but learned that taking the job would cause her Medicaid spend-down to increase to $806 a month. In addition, she would have to pay her Medicare Part B premium, she would lose food stamps, and her child would lose Healthy Start Medicaid coverage. Ann's response was, "Does this make any sense? I try to get a job... and am penalized to the point where I am better off on welfare!" With a Medicaid buy-in program, Ann could purchase Medicaid with the premium based on her income.
• Mary, 57 years old, worked for 35 years and raised three children. Several years ago, after losing the ability to sit, stand, talk or walk, she was diagnosed with a rare neurological condition. Her condition worsened but then went into remission. Mary receives $749 a month in SSDI and has no other income. She receives Medicare and Medicaid with a $305 spend-down. Mary has a variety of work experience. She says, "I have always been a productive person and made good money. I can't go out and get a job because I will lose my Medicaid benefits. There is no incentive at all for the person who is disabled to go out and get a job."
• Mike is 45 years old. He first developed mental illness at age 25. Mike receives $505 in SSDI each month. He has Medicare and Medicaid with a $40 spend-down. His annual medical expenses are $15,000. Mike is not working. He is interested in working part-time in a low-stress environment, possibly as a clerk or stocker. If he goes to work, rent for his subsidized housing will increase and his food stamps allocation will decrease. Even so, Mike sees the potential increase in his Medicaid spend-down as the biggest obstacle. He says that he is "between a rock and a hard place." Going to work will not improve his situation because of the resulting offsets to his benefits.

In this report I present a statistical model for estimating the number of Ohioans who would enroll in a Medicaid buy -in program. In order to investigate the validity of the model, I also analyzed the potential future enrollment of existing Medicaid buy-in programs in the states of Minnesota and Wisconsin, both of whose programs have been in operation long enough to allow some preliminary conclusions about likely eventual enrollment.

One goal in developing the statistical model was, of course, to estimate enrollment. Without a model, it would be necessary to rely on the experiences of other states. One might argue that Ohio has about two and a half times as many residents aged 18 to 64 as Minnesota and so should have about two and a half times as many people participating in a buy-in program. There are numerous problems with such an approach:

• Ohio and Minnesota might have different income and asset limitation guidelines.
• Minnesot a might be more or less successful than Ohio in encouraging persons with disabilities to work.
• Minnesota's program may have a different set of policies for the payment of premiums.
• Only Minnesota's current enrollment is known, whereas the more critical question is what Minnesota's eventual enrollment might be.

A second, and probably more important, goal in developing a statistical model is to be able to examine the effect of different policy options on enrollment. Some of the policy options examined in this report include:

• The effect of different income eligibility criteria
• The effect of different asset eligibility criteria
• The effect of different policies on premiums and other forms of co-pays.

Concurrent with the development of my model, the Ticket to Work Study Group of the Ohio Legislature, chaired by Senator Harris, was meeting to design criteria for a possible Ohio buy-in program. The policy scenario that I use as an example throughout this report was the one that they identified as the best policy on which to base their recommendations to the legislative leadership, although an appendix to this report includes a summary of the results produced by the model under a wide variety of other scenarios.

A third goal in model development was to be able to project the cost to the state of Ohio of operating a Medicaid buy-in program. The cost calculations are more complex that simply multiplying the number of projected beneficiaries times a projected annual cost per beneficiary. Current Medicaid recipients with a spend-down who participate in a buy-in program would no longer have a spend-down, which means a loss of revenue to the state. However, some beneficiaries would pay a premium to participate, which would generate revenue.

Data Sources

I also ran my model, again using CPS data, for Minnesota and Wisconsin. The number of persons in the data file for those states was just under 300 for each state.

I supplemented the CPS data with a variety of data available from the US Bureau of the Census and the Social Security Administration (SSA).

Model Development

Table 1: Population of Persons with Work Disabilities

Work Disability: According to the CPS, 668,480 Ohioans in this age range, or 10%, reported a work disability. A person was considered to have a work disability if any of the following conditions were reported (a person might have reported multiple conditions):

• A health problem limits or prevents work (83% of the 668,480 people)
• A disability is the main reason for not working (55%)
• Retired for health reasons (37%)
• Not in the labor force because of a disability (51%)
• The person is working part time because of a disability (5%).

Published data from the U.S. Census Bureau shows that 10% of the Americans aged 16 to 64 have a work disability. (http://www.census.gov/hhes/www/disable/cps/cps200.html) A small number of Americans that age - approximately two percent - have a mobility or self-care disability that does not translate into a work disability. Data on those kinds of disabilities are not available through the Current Population Survey. This should not pose a problem for the analyses because someone with a mobility or self-care disability, but without a work disability, is not the intended target population for a Medicaid buy-in program.

Employment: Of the 668,480 people with a work disability, 45% showed some evidence of working or having worked recently, or at least of having wanted to work recently.

• 37% had worked at least once during the previous calendar year.
• 27% were working at the time of the interview and an additional 3% were looking for work. The remaining people (69%) were not in the labor force.
• About 4% said at the time of the interview that they would like to have a job or said that at some time in the preceding year they would have liked to have a job.

Note that the population of all persons with work disabilities includes persons with both serious and less serious work disabilities. The employment rates just presented are far higher than would be found among the subset of people whose disabilities are severe, such as persons who might qualify for SSDI or SSI.

Note also that if the source of the national data were not the Current Population Survey, but rather the Survey of Income and Program Participation, then the employment figures and the data on the number of people with disabilities would both have been higher, because SIPP has a more inclusive definition of disabilities.

Health Insurance: Of the 668,480 people with a work disability, only 16% reported that they had no health insurance during the previous calendar year. People might have reported having multiple forms of insurance, either because of simultaneous coverage or because of serial changes in coverage. Thus, the following percentages add to more than the percentage of people with coverage of any kind (84%).

• 32% reported Medicaid coverage.
• 25% reported Medicare coverage.
• 21% reported coverage through their job or union, and in only 29% of these cases was the coverage provided without cost to the consumer.
• 24% reported some other form of health insurance, such as CHAMPUS (federal employees), VA (veterans) or policies on which they were covered as dependents of another person.

Disability Income: According to the Social Security Administration, 188,806 Ohioans received SSDI in 1999. Based on the CPS, there were 188,882 Ohioans who received SSDI, or nearly exactly the same number.(http://www.ssa.gov/statistics/oasdi_sc/1998/table1.pdf) The estimate from the CPS, however, involves the assumption that no Ohioan with a work disability and who is under the age of 65 received Social Security Retirement. Further, the published number was for persons of all ages, and so it is necessary to assume that no Ohioan under the age of 18 or over the age of 65 received Social Security Disability. Thus, the agreement between the two numbers is a little more coincidental than the nearly exact match suggests.

Social Security Administration (SSA) reports also show that 163,039 Ohioans between the ages of 18 and 64 received SSI in 1999. (http://www.ssa.gov/statistics/ssi_st_cty/1998/98-state.pdf) This number is 10% greater than the number estimated from the CPS.

I estimated from published SSA sources that roughly 43,500 Ohioans between the ages of 18 and 64 receive both SSI and SSDI. Based on the CPS, 46,508 persons receive both, again reflecting a good match.

Even though the CPS data agreed reasonably closely to reports from the Social Security Administration on the number of people getting either SSDI or SSI or both, I adjusted the weights in the CPS so that the number of SSDI and SSI recipients matched SSA published reports exactly.

Based on the adjusted data, just over half of the 668,480 persons with a work disability had some form of disability income during the previous year (52%). A total of 46% had either SSDI or SSI, or both (as might happen when a person's work history qualified him or her for SSDI but was insufficient to produce enough in SSDI benefits that they were not also eligible for SSI). Ten percent of persons with disabilities received disability income from a source other than SSDI or SSI, such as the Veteran's Administration, a private employer, or some form of insurance. About two percent of persons with disabilities received payments under either SSDI or SSI as well as from another source.

Summary: Table 1 shows the distribution of the 668,480 people with work disabilities according to whether they receive either SSDI/SSI, whether they have any recent evidence of work, and whether they receive Medicaid. Bear in mind that no income eligibility restrictions have been place on this population. Roughly half of persons with work disabilities who receive either SSDI or SSI also receive Medicaid. This reflects the fact that Medicare, as opposed to Medicaid, typically covers persons who receive SSDI but not SSI.

Table 2: Population of Persons with Severe Work Disabilities

National Research and State Differences: Based on national research conducted by the Census Bureau (http://www.census.gov/hhes/www/disable/cps/cps200.html), 66.2% of persons with any work disability have what is characterized as a severe work disability. Only 8.3% of persons with a severe work disability are employed at a point in time whereas over half of persons with a non-severe work disability are employed at a point in time.

There are rather striking differences among the states in the proportion of people with disabilities who work. For example, based on data from the Social Security Administration on SSI recipients (Office of Research, Evaluation and Statistics, SSI Disabled Recipients Who Work, June 2000. Social Security Administration), the proportion of people with severe work disabilities who work is 6.6% for the country as a whole and 8.1% for Ohio. However, the percentage is much lower in some states (3.0% in Alabama) and much higher in others (19.9% in South Dakota). While I will return to this point later, I will note briefly here that the model I developed assumes that all of the various factors that cause states to differ in employment rates for SSI recipients will have similar proportional effects on the employment rates of persons with disabilities generally, both prior to and subsequent to the implementation of Medicaid buy-in programs.

Results for Ohio: Assuming the proportion of Ohioans with a work disability that have a severe one is the same as the proportion nationally, then there should be 442,534 Ohioans ages 18-64 with severe work disabilities.

From Table 1, 305,285 Ohioans ages 18 to 64 receive SSDI or SSI. Not everyone with a severe work disability receives SSDI or SSI, but it is probably safe to assume that only people with a severe work disability are SSDI/SSI eligible. Thus, it was assumed that in addition to the people who receive SSDI or SSI, there are 137,249 Ohioans 18-64 years old who have a severe work disability but who do not receive SSDI or SSI.

Based on the information available in the CPS, the most important variable distinguishing people with severe versus non-severe work disabilities, is probably their current employment status. Persons in Table 1 who do not receive SSDI or SSI but who work should be considered relatively less likely to counted as severely disabled and persons who do not work should be considered relatively more likely to be severely disabled. (There is a difference between a person currently working and a person being identified in Table 1 as showing recent evidence of work. Only a portion of persons identified as having recent evidence of work will be employed at any given time.)

To estimate how many people with severe disabilities in Ohio work at any given time, I inflated the national employment rate for this population (8.3%) by a factor of 1.23, which is the ratio by which Ohio's SSI employment rate (8.1%) exceeds the national rate (6.6%). The result was an estimate that 10.2% of the Ohio population of persons with severe work disabilities is currently employed, or 45,128 persons. As shown below, based on the CPS, I determined that there were 21,949 working and receiving SSDI or SSI. Thus, there should be an additional 23,823 people who were working and who were not receiving SSDI or SSI. The final figure - 113,426 people not working and not receiving either SSDI or SSI - was obtained by subtraction. I was then to calculate the following adjustments to the weights in the CPS necessary to convert Table 1 into Table 2.

Table 3: Persons Meeting Income and Asset Limitations

Eligibility Based on Income: Eleven different sets of rules were considered for determining what families would meet the income eligibility criteria for the model. The final recommendation of the Ticket to Work Study Committee of the Ohio Legislature was to use the following set of rules (designated method 11 in Appendix II):

• Family income under the program would be capped at 250% of the Federal Poverty Level (FPL). (Medicaid determination is usually based on what are known as the poverty "guidelines," which are available at http://aspe.hhs.gov/poverty/poverty.htm, whereas the more complex poverty "thresholds" are used in the CPS and in most demographic research. The differences between the guidelines and the thresholds have to do with the number of persons in the family who are under the age of 18 or over the age of 65.) In 2001, the FPL guideline for a person living alone is $8,590 and it increases to $11,610 for a two-person family and $14,630 for a three-person family.
• Persons with a severe work disability would be given an earned income disregard of up to $20,000/ year.
• Adults with severe work disabilities who live with their parents would be treated as if they were living alone for income eligibility purposes (and for asset eligibility purposes).

The last rule means that the income of the parents will not disqualify a person for coverage, which might have had the unintended consequence of forcing these individuals to live alone. Without this rule, roughly 14% of the eligible population comprises adult children living with their parents, depending on the exact combination of policy options. With this rule, the percentage of the eligible population who are adult children increases to about 20%.

Suppose, as an example, that a person with a disability has $8,400 a year in unearned income (e.g., SSI benefits) and earns $7,000 a year working part-time. Suppose the person's spouse has no unearned income but earns $15,000 a year. This person would be eligible for Medicaid under the proposed Ohio buy-in program. The earned and unearned income of the family is $30,400, which exceeds 250% of the poverty guideline ($29,025). However, after applying the earned income disregard, the countable income is only $23,400, comfortably within the income eligibility guideline. Further, the eligibility guideline affords this hypothetical couple considerable room to improve their economic circumstances through higher earnings.

In modeling what individuals would meet the income eligibility criteria, it was necessary to project what an individual's income would be if they began working and if they enrolled in the program. Individuals with no earned income in the previous year were assigned an earned income equal to the amount earned by the first person previous to them in the data file (which was in a random order) who did have earned income. This "projected income" was used in all subsequent calculations.

The model did not assume that there would be any decrease in unearned income even though unearned income might decrease as earned incomes increases. Some effort was made to investigate the impact that an increase in total income would have on model results by inflating total income for people who had earnings by a total of 25%. The effect was relatively small.

Eligibility Based on Assets: Three different levels of asset eligibility were investigated in this research - $2,000, $10,000 and $20,000 - before the Ticket to Work Study Group settled upon the $10,000 limit. However, advocates argued that Ohio - like most states that have implemented Medicaid buy -in programs - should exclude certain forms of assets, including:

• A person's primary residence
• An automobile
• Qualified retirement accounts (i.e., IRAs, 401k plans, etc.)
• Medical savings accounts.

Table 3 shows the number of persons with severe disabilities who meet both the income eligibility rules and the asset eligibility rules. As a result of applying these two restrictions, the number of Ohioans 18 to 64 with a severe work disability that would potentially qualify under the proposed program drops from 442,534 (Table 2) to 305,938 (Table 3).

Whereas income eligibility was easily determined from the CPS data, asset eligibility could only be estimated based on data from the Census Bureau on asset ownership of households. (Current Population Reports [P70-71]. Household Net Worth and Asset Ownership. U.S. Bureau of the Census.) Households in the CPS were first categorized according to what income quintile they fell into (the lowest quintile represents the poorest 20% of American households, the second quintile represents the next 20%, etc.). The percentage of households in each quintile who were assumed to meet each of the asset eligibility standards is shown in Appendix II. These percentages could not be calculated directly from the published report and they should be understood to be rough estimates.

There is a policy issue related to asset levels that has not been addressed by advocates of buy-in programs. People who benefit by being able to accumulate assets under the terms of a buy-in program may someday lose the ability to work, which could have the effect of pushing them out of a spend-down program into an ordinary Medicaid program, with dramatically lower levels of allowable assets.

Table 4: Number of Persons Projected to Hold Jobs

None of the 218,083 people in the "No Recent Work" column of Table 3 are employed, nor have any of them worked in the past year, or even expressed an interest in working during the past 15 months. The model assumes that the odds are much higher that a person in the "Recent Work" column will secure employment and thus qualify for a buy-in program than a person in the "No Recent Work" Column.

The overall employment rate of all 305,938 persons in Table 3 is only 7.9% (remember that people in Table 3 have lower incomes and lower amounts of asset accumulation that persons in Table 2, for whom the overall employment rate was 10.2%). I am projecting that after the implementation of the buy-in program, the employment rate will increase to 14.7%.

Table 5: Projected Enrollment

Spend-Down: Someone who receives Medicaid but has income in excess of the Medicaid need standard (for 2001, $460/month for an individual and $796/month for a couple) is responsible for "spending down" to the need standard. (A certain amount of income is disregarded, as is a certain residual portion of earned income. Details about spend-down computations are included in Appendix II.)

The model can correctly identify who has a spend-down but it cannot accurately account for how much spend-down the state might lose if someone moves from an existing Medicaid to a buy-in program. If a person has medical bills in excess of their spend-down amount for a given month, then he or she should exercise the spend-down. But if the person's medical bills are less than the spend-down amount, then that creates additional disposable income for the person which is never turned over to the state.

Premiums: Under federal law, states can charge premiums up to the full-cost of the insurance (roughly $5,250/year in Ohio) subject to certain restrictions. No recipient can be charged more than 7.5% of his or her income for a Medicaid premium. Further, premiums must increase with family income.

The Ticket to Work Study Group in Ohio considered the following policy options regarding premiums:

• The premium would begin with the first dollar of income above the 150%, 200%, or 250% of the Federal Poverty level, without any disregards for either earned or unearned income. (It does not make sense to charge premiums for income under 150% FPL because so many of these individuals now receive Medicaid at no charge. A premium would thus constitute a disincentive to work.)
• The amount of the premium would be 5.0%, 7.5% or 10% of the income in excess of the starting level.

The Group settled on a starting point of 150% of FPL and a premium of 10%. The detailed results in Appendix I reflect these options although the summary results in Appendix III illustrate the impact of other options.

In 2001, the Federal Poverty Level for one person is $8,590. The following analysis shows a person's monthly premiums assuming that an income of either $15,000 or $25,000, assuming that the person lives alone, and assuming that premiums begin with the first dollar after 150% of FPL.

Take-Up Rates: If someone currently receives Medicaid and has no spend-down, there would be few incentives to enroll in the new program (but if the person did enroll, it would be at no net cost to the state). However, some persons in this situation might eventually have earnings that outpace their current eligibility and so a take-up rate of zero would be inappropriate. Instead it was set at 10% for persons with no other form of health insurance and 5% for people with at least one other form of health insurance. For people with a Medicaid spend-down, the only consideration that might cause him or her not to enroll in the new program would be the premium. For persons without a premium, a take-up rate of 100% was assumed.

For people with no Medicaid and no other form of health insurance, a substantial take-up rate of 70% was projected if the person would not have any premium. For people with no Medicaid and some other form of health insurance (most typically Medicare), a fairly low take-up rate was assumed (25%) even without a premium payment.

Premiums have the effect of reducing enrollment in health insurance programs. (http://www.urbaninstitute.org/entitlements/premium.htm) I made the assumption that the following take-up rates would be reduced proportionately if the person owed a premium. I assumed the rates would be 90% of the rates shown below if the total premium amount was 1% of total family income, 80% if 2%, and so on down to 50% for a premium of 5% or more.

From the previous section, someone with a family income of $25,000 would have a monthly premium of $100.96 if the premium started at 150% of FPL and was 10% of the income above that mark. Such a person would pay a total of $1,212 in premiums over the course of the year, or 4.8% of total family income.

Table 6: Projected Program Costs

• Benefit costs
• Loss of spend-down
• Collection of premiums
• n Administrative costs

Taking the last item first, I am not in a position to project the administrative costs for ODJFS. These costs will include: (1) an increased number of disability determinations, (2) the costs associated with premium collection, and (3) the basic administrative overhead associated with running any Medicaid program, a category that covers information systems, department administration, preparation of materials for public education, and so on.

Ohio advocates for a buy-in program made the decision not to propose any form of cost sharing besides premiums (examples would include annual or one-time enrollment fees, co-pays, or deductibles). Their concern was that the administrative costs of such options would outweigh collections.

As shown in Table 6, I project that the state will lose $1.8 million in spend-down revenue (an average of $143 per person per month for people who have a spend-down). Recall from the description of this computation, however, that this is an upper limit on the loss of spend-down revenue and that CPS data cannot be used to refine this projection.

I also project that the state will gain $4.3 million in premiums collected.

In preparing my benefit cost, I was advised by the Ohio Department of Jobs and Family Services that it would be reasonable to assume that the monthly cost of providing Medicaid to a person 18 to 64 years old with a disability who was living in the community is $437/month in 2001. However, this amount is associated with some uncertainty:

• A large proportion of the people in Table 6 have Medicare or some other form of health insurance. For many of these people, Medicaid would be a wrap-around program, and therefore less expensive. However, it is unknown whether these proportions are substantially different from the population of persons now receiving Medicaid. If so, then the blended rate of $437/month might be either too low or too high.
• It is possible that people who start to work might incur additional expenses, which would make the monthly cost for all Medicaid recipients too low.
• It is being assumed that people who come onto the program who do not currently have Medicaid will cost the same amount as people now on Medicaid. There is no evidence on this point.

Total benefit costs of the fully subscribed buy-in program are therefore projected to be equal to the monthly premium ($437) for persons not now on Medicaid (10,391) multiplied by 12 plus the lost spend-down revenue minus the amount collected in premiums, or $52,015,631.

The state share of the monthly premiums is 41.9%, which makes the state's cost of the program, once it is fully subscribed, $20,412,295. If the state monthly cost is higher, then obviously program costs will exceed this amount.

Two important caveats must be mentioned:

• Under no circumstance do I anticipate this many people enrolling in the program in the first year or two of its existence. Initial take-up rate will be a function of publicity, outreach activities, and the speed with which administrative systems can be implemented. Ohio might expect a few thousand enrollees in the first 18 months of program operation.
• Long term - and by that I mean five to ten years out - systems for employment support for persons with disabilities in Ohio might improve to the extent that the employment rate for this population might increase. That would tend to increase enrollment in the program.

Comparisons to Other States

In both Wisconsin and Minnesota considerably more SSI recipients work (14.4% and 16.7% respectively) than is the case in Ohio. What determines the proportion of a state's residents who have severe disabilities who are able to work? My view is that a myriad of factors are responsible, including, to name just a few, state investment in workforce development, the quality of the health care and social services systems, and the relative size of different sectors of the state's economy (e.g., agriculture, information technology). Thus, states such as Wisconsin and Minnesota start off with much higher rates of persons with disabilities working and that they will continue to have higher rates. In other words, I am assuming that the effects of a Medicaid buy-in program within a state will be proportional to its current employment rate for persons with disabilities.

I ran my model for Wisconsin and Minnesota, making suitable adjustments for state income, asset and premium policies. The results are shown below.

Notice that there are striking differences in the projected proportion of people already enrolled in Medicaid among the three states. This appears to be due primarily to differences in the proportion of people who are on Medicaid and who have a spend-down. In Ohio, that proportion (among those eligible for the program) is relatively small whereas in both Minnesota and Wisconsin it is higher.

Minnesota's current enrollment is only 58% of what the current model projects but the state's projected eventual enrollment is 83% of what the current model projects. In contrast, Wisconsin's current enrollment is only 10% of what the current model projects and the state's own projections have not recently been updated. The original projections produced by the state of Wisconsin were carried out under a more conservative set of income and premium assumptions than were used in the enabling legislation, so the table includes a very rough estimate that Wisconsin's analysis would have been 4,000 participants had it reflected the policy options eventually enacted. Officially, Wisconsin projected 2,146 enrollees, but that was under the assumption that no earned income would be disregarded and the all program enrollees would be charged a one-time fee to enroll.

Modeling Results and Conclusions

• Premiums begin with the first dollar of income above 150% FPL or 250% Federal poverty level.
• Premiums are 5.0%, 7.5% o 10.0% of the dollar amount defined above.
• Assets are limited to $2,000, $10,000 or $20,000 after exclusion of home, care, retirement accounts and medical saving accounts.
• Eleven different income eligibility standards (see Appendix II).

The results are shown in Appendix III. The box that is highlighted in the table corresponds to the policy options endorsed by the Ticket to Work Study Group and which were used in preparing the detailed tables shown in Appendix I.

With the exception of California, whose buy-in program is new, poorly publicized (as of late 2000), and not surprisingly poorly subscribed, Ohio would be the most populous state with a Medicaid buy -in program for persons with severe disabilities. In considering how to limit its financial exposure, advocates and legislators should bear in mind that there are three powerful influences on eligibility and take-up rates:

• The earned income disregard
• Income and asset limits
• Premium policies

Appendix I Detailed Results for Ohio

Table 1:Population of Persons with Work Disaiblities (Ohio Residents Ages 18 to 64)

Table 2:Population of Persons with Severe Work Disabilities (Ohio Residents Ages 18 to 64)

Table 3:Persons Meeting Income and Asset Limitations (Ohio Residents Ages 18 to 64) with a Severe Work Disability

Table 4: Number of Persons Projected to Hold Jobs Under New Buy-in Program (and Percentages)

Table 5:Projected Enrollment

Table 6: Projected Program Costs

Amount of Spend Down Lost - $1,824,796
Amount of Premiums Collected - $4,297,676

Total Benefit Cost of Program - $52,015,631
State Share of Benefit Cost of Program - $20,412,295

Appendix II Model Parameters

Table 1

Ages of eligible persons = 18 to 65

Number of persons 18 - 65 = 6,887,990

Criteria for identifying a person as having a work disability (persons aged 62 or older who identify themselves as retired are excluded)

• Reports a health problem which prevents or limits work
• Reports retiring or leaving a job for health reasons
• Reports disability/illness main reason for not working previous year
• Reports being not in the labor force because of a disability
• Reports health/medical reasons for working part-time

Number of recipients of federal disability incomd (assumes nobody under the age of 18 or above the age of 65 is on SSDI. Also assumes nobody under the age of 65 received SS Retirement).

• Social Security Disability Income (SSDI) recipients = 188,806
• Supplemental Security Income recipients = 163,039

Criteria for identifying a person as having a recent work history

• Reports working anytime during previous year
• Reports being employed or unemployed (i.e., in labor force)
• Reports wanting a regular job, either full-time or part-time
• Reports having spent any time looking for work previous year

Number of persons in Ohio with work disabilities, as defined above = 668,480

Table 2

Proportion of persons with work disabilities who have severe disabilities (national data applied to Ohio) = 0.662

Employment rate for persons on SSI
Ohio = 8.1
National = 6.5

National employment rate for persons with severe disabilities = 8.3

Ohio employment rate for persons with severe disabilities [10.3 = 8.3 * (8.1/ 6.5)] = 10.3

Number of persons in Ohio with severe work disabilities (442,534 = 668,480 * .662) = 442,534

All persons receiving SSDI or SSI are presumed to be severely disabled

Persons with work disability but not on SSDI/SSI are weighted according to their current employment status in reducing the population to 442,534
Weight multiplier for persons currently employed = 0.151052
Weight multiplier for persons currently unemployed = 0.552002

Table 3

Persons who have recent work experience and had earned income in the previous year have their personal incomes inflated by the amount shown to the right. All of the increase is assigned to earnings. Their unearned income is kept constant. Varies 1.00 - 1.25.

Persons without recent work experience, or without earned income in the previous year, are allocated an amount of earned income equal to that of the first person previous to them in the data file who had earned income the previous year. These allocations are made prior to the earnings inflation (described above). People who get earned income allocated are presumed to have constant unearned income.

All income adjustments are made prior to determining income eligibility.

Income eligibility is satisfied when adjusted family income is less than or equal to 250% of the Federal Poverty Level for the family as a whole. For Ohio, adjusted family income was calculated in each of the following ways: Vaqries Methods 1 - 11

1. Disregard 100% of the person's earnings
2. Disregard 50% of the person's earnings
3. Disregard up to $10,000 of the person's earnings
4. Disregard up to $10,000 of the person's earnings and up to $10,000 in earnings by anyone else in the household.
5. Disregard up to $20,000 of the person's earnings and up to $20,000 in earnings by anyone else in the household.
6. Disregard up to $30,000 of the person's earnings and up to $30,000 in earnings by anyone else in the household.
7. Disregard up to $10,000 of the person's earnings and, if the person lives with parents, up to $10,000 of other earnings.
8. Disregard up to $20,000 of the person's earnings and, if the person lives with parents, up to $20,000 of other earnings.
9. Disregard up to $30,000 of the person's earnings and, if the person lives with parents, up to $30,000 of other earnings.
10. Disregard up to $10,000 of the person's earnings and, if the person lives with parents, treat him or her as a one-person household.
11. Disregard up to $20,000 of the person's earnings and, if the person lives with parents, treat him or her as a one-person household.

Maximum amount of assets allowed, after excluding home, car, retirement savings and medical savings accounts. - Varies $2,000 - $10,000 - $20,000

It was assumed that everybody currently on Medicaid would meet the asset limitation criteria because they meet more stringent criteria now and in acquiring new assets would be able to stay beneath the limit.

For people not on Medicaid currently, assets were determined based on current, not inflated, family income because it was assumed that a family would meet the criteria at the time of joining the program and then manage their assets so as to stay under the limit.

Following are the percentages of people who were assumed eligible with the asset limitation criterion set at $2,000/family, by income:

• Under $14,500 = 90%
• $14,500 to $26,572 = 75%
• $26,573 to $40,541 = 66%
• $40,542 to $62,741 = 50%
• $62,742 or more = 25%

Following are the percentages of people who were assumed eligible with the asset limitation criterion set at $10,000/family, by income:

• Under $14,500 = 95%
• $14,500 to $26,572 = 89%
• $26,573 to $40,541 = 83%
• $40,542 to $62,741 = 77%
• $62,742 or more = 71%

Following are the percentages of people who were assumed eligible with the asset limitation criterion set at $20,000/family, by income:

• Under $14,500 = 100%
• $14,500 to $26,572 = 96%
• $26,573 to $40,541 = 92%
• $40,542 to $62,741 = 88%
• $62,742 or more = 84%

Table 4

The following steps were taken to estimate employment rates.

1. Within each of the four groups of people with a recent work history defined by Medicaid status and disability income status, compute the proportion who are currently employed.
2. Multiply that rate by 1.33. (do not let it exceed 0.9).
3. For each of the four corresponding groups of people with no recent work history, set the employment rate equal to 7.5% of the rate for people with a recent history of working.

Table 5

People who already receive Medicaid are examined to see if they have a spend down.

All spend down calculations are based on a current Medicaid recipient's current income, not projected income under the buy-in program.

To calculate the spend down, first determine if the person has one.

1. Subtract $20 from all monthly unearned income for the person except SSI payments.
2. Subtract $65 from all monthly earned income for the person and divide by 2 (or subtract $85 if person has no unearned income).
3. Add together the remaining earned and unearned income and compare to the Medicaid need standard.
4. Any amount in excess of zero means the person has a spend down.

One problem with this methodology is that the person may not have medical bills in excess of their spend down in a given month, and so the model probably overstates the spend down revenue that would be lost to the state under a buy-in program.

Persons eligible for Section 1619a or 1619b are presumed to be taking advanatge of it, and so any calculated spend down for them is set to 0.

Premiums are calculated based on a defined percentage of the person's total income (not the family's income) in excess of a set percentage of the FPL for the family.

% of FPL at which premiums begin - Varies 150, 200, 250

% of income above cutoff due as a premium - Varies 5.0, 7.5, 10.0

Take up rates are calculated as follows:
For persons with no health insurance (except perhaps Medicaid), no spend down and no premium:
With Medicaid = 0.10
Without Medicaid = 0.70

For persons with some other health insurance (and perhaps also Medicaid), no spend down and no premium:
With Medicaid = 0.05
Without Medicaid = 0.25

Persons with a spend down (which means they are already on Medicaid). = 1.00

For persons with a premium, the above takeup rates are adjusted
1% - 0.9
2% - 0.8
3% - 0.7
4% - 0.6
5% - or more 0.5

Table 6

Loss of spend down is over-stated because it assumes every recipient has at least a certain minimum in medical bills every month.

Premiums assume perfect collection efficiency.

No administrative costs are assumed.

Monthly costs of benefits per recipient - Varies $437

Appendix III Results for Different Policy Scenarios

Method 1: Disregard 100% of the person's earnings

Method 2: Disregard 50% of the person's earnings

Method 3: Disregard up to $10,000 of the person's earnings

Method 4: Disregard up to $10,000 of the person's earnings and up to $10,000 in earnings by anyone else in the household

Method 5: Disregard up to $20,000 of the person's earnings and up to $20,000 in earnings by anyone else in the household

Method 6: Disregard up to $30,000 of the person's earnings and up to $30,000 in earnings by anyone else in the household

Method 7: Disregard up to $10,000 of the person's earnings, and if the person lives with parents, up to $10,000 of other earnings.

Method 8: Disregard up to $20,000 of the person's earnings, and if the person lives with parents, up to $20,000 of other earnings.