Analysis of Refundable Tax Credit

Tax code incentives aim to expand health insurance coverage by reducing the absolute cost that individuals face when buying insurance. When a tax credit is provided, individuals subtract the amount of the tax credit from their total tax liability. If the credit is refundable, people without a tax liability receive a direct payment.

These are the nine performance dimensions against which we measured Refundable Tax Credit:

Refundable tax credits for individuals and families will have no discernable effect on total health care spending:

  • Using our model, we estimate that aggregate health care spending will increase $930 million to $4.47 billion (0 to 2 percent) with a refundable tax credit for individuals and families to obtain coverage. Read more below
  • Our results indicate that net government spending will increase $4.72 to $63.47 billion (0.5 to 6.4 percent); this reflects an increase due to the cost of the subsidy ($5.8 to $69.0 billion) and a small decline in Medicaid spending (-$1.08 to -$5.53 billion). Read more below
  • Our results indicate that individual/family out-of-pocket spending will increase $690 million to $4.67 billion (0.2 to 1.9 percent). Read more below
  • Published estimates of the magnitude of the increase in spending vary widely and depend on assumptions about the size and form of the credit, as well as the additional policy changes that accompany it. Read more below

Using our model, we estimate that aggregate health care spending will increase $930 million to $4.47 billion (0 to 2 percent) with a refundable tax credit for individuals and families to obtain coverage.

Refundable tax credits for individuals and/or families to purchase coverage will have no significant effect on total spending. The net effect will depend on the size and form of the credit, as well as on what policies, if any, are tied to the change (e.g., the elimination of the tax free treatment of employer-provided health insurance, mandates to purchase insurance) but in all of our modeled scenarios the difference in spending is close to zero.

The number of people who obtain insurance will rise as some previously uninsured individuals who are eligible for the credit purchase insurance. The increase in coverage with the credit, and the associated increase in utilization of some health services, will have little effect on overall health spending. Depending on the size of the credit and the income eligibility levels, the amount that individuals spend on health care will, for the most part, be higher among those who are newly purchasing insurance. The magnitude of this effect is unclear and may depend on the relative costs of different types of medical care, since coverage may shift utilization from some types of services to others. Previously uninsured individuals may actually offset spending increases by using primary and preventive care instead of emergency care.

We modeled a refundable tax credit with the following design:

  • Individuals and families can claim an annual Federal tax refund on money spent to purchase a nongroup insurance policy.
  • Individuals and families below income thresholds of $15,000 and $30,000, respectively, are eligible for the full amount of the tax credit.
  • The tax credit amount decreases on a sliding scale as income rises; it phases out at $30,000 (individuals) and $60,000 (families).
  • To qualify for the credit, individuals or families must purchase a policy comparable to what is currently offered in the nongroup market.
  • The amount of the tax credit that is claimed cannot exceed the amount spent on premiums.
  • The consumer pays premium costs in excess of the credit amount.
  • Individuals and families that want to purchase insurance are able to do so (that is, we designed this policy option with guaranteed issue).

We made the following assumptions in modeling the tax credit option:

  • The elasticity of take-up among the previously uninsured is 0.4 (that is, a 10 percent decrease in premium costs leads to a 4 percent increase in take-up rates).
  • Some individuals with group coverage will choose to switch to nongroup coverage to take advantage of the credit.
  • Some individuals with Medicaid will switch to nongroup coverage.
  • Some employers will drop coverage altogether. The probability that a firm will drop coverage depends on the proportion of employees in the firm who are eligible for the new tax credit.

Although this policy option would likely occur in conjunction with other policies, we modeled a refundable tax credit for individuals and families separate from other policy options.

Table 1 shows the results of our modeling on changes in national health expenditures that would occur under a policy of refundable tax credits of varying amounts.

Table 1
Effect of a Refundable Tax Credit for
Individuals and Families to Purchase Nongroup Coverage
on Total National Health Expenditures
Category of Spending Credit Amount (Individual/Family)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Change in national health spending due to increased utilization ($B) $0.93 $3.35 $4.47

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

Overall, a refundable tax credit would lead to an indistinguishable change in national health spending (range $930 million to about $4.5 billion) because relatively few people who currently are not insured become insured under this policy change. These results hold constant the income eligibility requirements but vary the amount of the credit, which results in more than a four-fold difference in total spending from the least generous to the most generous credit amount.

We tested the sensitivity of our findings to the elasticity of take-up among the currently uninsured and found that under the mid-range credit amount ($2,500 for singles and $6,250 for families) spending would increase $2.3 billion (0.1 percent) at an elasticity of 0.3 and $4.6 billion (0.2 percent) at an elasticity of 0.5.

We also tested the sensitivity of our findings to the income eligibility parameters. As the lower income threshold increases, a greater number of people would elect to switch from group (employer sponsored) insurance or Medicaid to nongroup insurance with relatively little additional impact on reducing the number of uninsured.

Our results indicate that net government spending will increase $4.72 to $63.47 billion; this reflects an increase due to the cost of the subsidy ($5.8 to $69.0 billion) and a small decline in Medicaid spending (-$1.08 to -$5.53 billion).

Based on our model, the primary financial effects of the tax credit will be on the government (Table 2). Government expenditures will increase to pay for the credit through forgone tax revenues plus additional spending (for those who did not have a tax liability), whereas government spending on Medicaid will decrease by $1.08 to $5.53 billion (0.3 to 1.8 percent) as a small number of people switch from enrollment in Medicaid to purchase of nongroup policies. The estimates of government spending are very sensitive to the size of the credit.

We also calculate the government spending per net newly insured, a measure recommended by Gruber (2008) as showing the "bang for the buck" of policies designed to reduce the number of uninsured. We estimate that under a refundable tax credit, the government would spend $2,025 to $6,365 for each newly insured person, depending on the level of the credit. The variation reflects the fact that at the lower credit amount, 4 percent of those newly selecting nongroup insurance previously had employer sponsored insurance whereas at the higher credit amount, 18 percent of those newly on nongroup insurance previously had employer sponsored insurance.

Table 2
Effect of a Refundable Tax Credit for
Individuals and Families to Purchase Nongroup Coverage
on Government Spending
Category of Spending Credit Amount (Individual/Family)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Net government spending ($B) $4.72 $26.45 $63.47
Cost of subsidy ($B) $5.8 $29.46 $69.0
Change in Medicaid spending (Federal and state combined)
($B)
($1.08) ($3.01) ($5.53)
Government cost per net newly insured ($) $2,025 $3,955 $6,365

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

Our results indicate that individual/family out-of-pocket spending will increase $690 million to $4.67 billion.

Individuals and families will have a small increase in spending on health care under a refundable tax credit from $690 million to $4.67 billion (0.2 to 1.9 percent) depending on the amount of the credit that is available (Table 3). The effect of the tax credit on spending will vary with prior insurance status, income, and the specifics of the health coverage provisions. Previously uninsured consumers and those who had been on Medicaid will experience an increase in their out-of-pocket spending for cost sharing, such as deductibles and copayments, and the increased use of health services that accompanies acquisition of insurance. Consumers with existing nongroup insurance may see their premiums decrease by 7 to 18 percent as new enrollees (who are presumably younger and healthier) buy into the market. Spending among individuals with existing, employer sponsored insurance will remain largely unchanged.

Table 3
Effect of a Refundable Tax Credit for
Individuals and Families to Purchase Nongroup Coverage
on Consumer Out-of-Pocket Spending
Category of Spending Credit Amount (Individual/Family)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Change in consumer out-of-pocket spending (deductibles, copayments, non-covered services) ($B) $0.69 $2.57 $4.67

SOURCE: RAND COMPARE microsimulation modeling results. December 31, 2008.

Some uncertainties remain regarding the effect of a tax credit on spending. A variety of incentives (tax exemptions, deductions, credits) utilize the tax code to encourage the purchase of health insurance, and we do not know how such incentives might interact to produce changes in coverage. How this policy option will affect premium pricing, particularly in the individual market, remains uncertain. We do not know how this credit will affect the products offered by the nongroup market--e.g., the extent to which private insurers may market products that offer less generous coverage after the policy is established. Further, we do not know how indexing the amount of the credit to reflect the rising cost of insurance premiums could raise the costs to the federal government. We expect administrative costs for a refundable tax credit to be small; however, we do not know the degree to which such costs will also contribute to government costs.

Published estimates of the magnitude of the increase in spending vary widely and depend on assumptions about the size and form of the credit, as well as on the additional policy changes that accompany it.

Gruber (2008) examines refundable tax credits that are available for purchasing insurance on the nongroup market. At a minimum income threshold of $15,000 for a full subsidy for individuals phasing out at $30,000 and a credit amount of $2,500 he estimates total government costs at $22.6 billion and government cost per net newly insured at $4,400; these estimates are similar to our estimates of $26.5 billion and $3,955, respectively. Table 4 shows comparisons between COMPARE modeling results and Gruber (2008) across multiple scenarios. Because Gruber uses a different base population (the Current Population Survey), it is not surprising that there are some differences in our findings. Gruber's results are in 2005 dollars whereas the COMPARE results are displayed in 2007 dollars.

Table 4
Comparison of Results for COMPARE and Gruber (2008)
For Four Scenarios
  Scenario
POLICY PARAMETERS Scenario 1 Scenario 2 Scenario 3 Scenario 4
Lower income threshold for singles ($) $15,000 $15,000 $25,000 $25,000
Upper income threshold for singles ($) $30,000 $30,000 $50,000 $50,000
Lower income threshold for families ($) $30,000 $30,000 $50,000 $50,000
Upper income threshold for families $60,000 $60,000 $100,000 $100,000
Credit for singles ($) $2,500 $6,000 $2,500 $6,000
Credit for families ($) $6,250 $15,000 $6,250 $15,000
Elasticity of take-up among uninsured 0.4 0.4 0.4 0.4
COMPARE RESULTS
Net newly insured (millions) 6.7 10.9 6.6 11.2
Total government spending ($B) $26.5 $73.0 $38.7 $109.5
Government spending per net newly insured ($) $3,955 $6,695 $5,880 $9,815
Gruber Results (NBER WP 13758)
Net newly insured (millions) 7.0 13.0 9.0 16.0
Total government spending ($B) $22.6 $61.5 $29.3 $79.1
Government spending per net newly insured ($) $4,400 $6,100 $5,900 $7,900

SOURCE: RAND COMPARE microsimulation modeling, December 31 2008; Gruber J, "Covering the Uninsured in the U.S.," Cambridge, MA.: National Bureau of Economic Research, Working Paper 13758, Table 4, January 2008. As of January 2 2008 available at: http://www.nber.org/papers/w13758.

Gruber and Levitt (2000) examined a policy that would provide a refundable tax credit of $1,000 to individuals with incomes less than $40,000 (with the credit phasing out at $60,000) and a credit of $2,000 to households with incomes less than $75,000 (with phase-out at $100,000) to purchase nongroup coverage. The policy considered would allow those with current group coverage to switch to nongroup coverage, but the credit could not be applied to group coverage. He assumed no change in the tax exempt status of group coverage. Gruber predicted that the cost to the federal government of this policy would be $13.3 billion in 1999 dollars. Raising the credit to $2,000 (individuals) and $4,000 (families) is estimated to cost the government almost $40 billion per year.

Reschovsky and Hadley (2004) modeled a refundable tax credit for those without group coverage that would provide a tax credit of $1,000 to individuals with incomes under $15,000 and up to $3,000 ($1,000 per adult, $500 per child) to families with incomes under $25,000. They predicted that spending by previously uninsured families who purchase health insurance under this scenario would increase from $463 per year to $2,520 per year (after the tax credit is applied). Burman and Gruber (2005) estimated that a tax credit under this scenario would cost $4.9 billion per year at the federal level but would reduce state spending by $425 million in 2004 dollars. This savings at the state level comes from the transition of individuals in public programs to private insurance.

Buchmueller et al.'s (2005) review suggests that utilization for all types of health care services increases when coverage is expanded. They note that the effect of utilization may be overstated, since utilization will vary across subgroups. For example, if coverage is expanded mostly to primarily young, healthy individuals, we would not expect a large increase in utilization.


References

Buchmueller TC, Grumbach K, Kronick R, Kahn JG, "The Effect of Health Insurance on Medical Care Utilization and Implications for Insurance Expansion: A Review of the Literature," Medical Care Research and Review, Vol. 62, No. 1, February 1 2005, pp. 3–30.

Burman LE, Gruber J, Tax Credits for Health Insurance, Washington, D.C.: Urban-Brookings Tax Policy Center, Tax Policy Issues and Options, No. 11, June 2005.

Gruber J, "Covering the Uninsured in the U.S.," Cambridge, MA.: National Bureau of Economic Research, Working Paper 13758, January 2008. As of January 2 2008 available at: http://www.nber.org/papers/w13758.

Gruber J, Levitt L, "Tax Subsidies for Health Insurance: Costs and Benefits," Health Affairs, Vol. 19, No. 1, January/February 2000, pp. 72–85

Reschovsky JD, Hadley J, "The Effect of Tax Credits for Nongroup Insurance on Health Spending by the Uninsured," Health Affairs, Web Exclusive, February 25, 2004, pp. w4113–w4123

top

A refundable tax credit will not change consumer financial risk for the overall non-elderly population, but it is likely to increase the amount spent on health care for most of the newly insured:

  • We estimate, using the COMPARE microsimulation model, little change in the median percentage of income spent on health care and the percentage of families that spend more than 10 percent of their income on such care. Read more below
  • How the tax credit affects consumer financial risk depends heavily on the consumer's insurance status prior to the credit; consumer financial risk would increase substantially for those who were formerly uninsured. Read more below
  • Previous studies have estimated similar levels of consumer financial risk under refundable tax credit policies. Read more below

We estimate, using the COMPARE microsimulation model, little change in the median percentage of income spent on health care and the percentage of families that spend more than 10 percent of their income on such care.

Consumer financial risk is the extent to which individual and/or household spending on health care compromises the ability to pay for other basic necessities. A refundable tax credit is intended to provide a subsidy with which low and middle income individuals can purchase health insurance. The effect of the credit on consumer financial risk will depend on the size of the subsidy and whether it actually covers the costs of insurance, the specifics of the health insurance obtained in terms of cost sharing, and the extent to which the provision of insurance affects the use of medical care.

We modeled a refundable tax credit with the following design:

  • Individuals and families can claim an annual Federal tax refund on money spent to purchase a nongroup insurance policy.
  • Individuals and families below income thresholds of $15,000 and $30,000, respectively, are eligible for the full amount of the tax credit.
  • The tax credit amount decreases on a sliding scale as income rises; it phases out at $30,000 (individuals) and $60,000 (families).
  • To qualify for the credit, individuals or families must purchase a policy that is comparable to what is currently offered in the nongroup market.
  • The amount of the tax credit that is claimed cannot exceed the amount spent on premiums.
  • The consumer pays premium costs in excess of the credit amount.
  • Individuals and families that want to purchase insurance are able to do so (that is, we designed this policy option with guaranteed issue).

We made the following assumptions in modeling the tax credit option:

  • The elasticity of take-up among the previously uninsured is 0.4 (that is, a decrease in premiums of 10 percent leads to a 4 percent increase in take-up rates).
  • Some individuals with group coverage will choose to switch to nongroup coverage to take advantage of the credit.
  • Some individuals with Medicaid will switch to nongroup coverage.
  • Some employers will drop coverage altogether. The probability that an employer will drop coverage depends on the proportion of employees who are eligible for the new tax credit.

Although this policy option would likely occur in conjunction with other policies, we modeled a refundable tax credit for individuals and families separate from other policy options.

Table 1 shows the median percentage of income spent on health care before and after enactment of the change in tax credit policy for the non-elderly (grouped by the size of the credit). Overall, before and after the policy change, the median percentage of income spent on health care is about 6 percent (for an individual with an income of $15,000, this amounts to $900 annually).

At the lowest credit amount, no singles or families can have their premium fully paid by the credit; at the highest credit amount, 10.3 million singles and 41.1 people are in families whose premiums can be fully paid by the credit.

Table 1
Median Percentage of Income Spent on Health Care
Before and After a Refundable Tax Credit
Among Non-Elderly
Percentage of Income Spent on Health Care Amount of Credit (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Before tax credit 6.0 6.0 6.0
After tax credit 6.2 6.1 5.8

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

Table 2 shows, at three different levels, the proportion of the non-elderly that spend more than 10 percent of their income on health care, before and after enactment of a refundable tax credit. Not surprisingly, given the relatively small number of people who are affected by this policy, the overall effect of the policy on consumer financial risk is small with slight increases at the lower credit amounts and a slight decrease at the higher credit amount.

Table 2
Percentage of Non-Elderly
That Spend More than 10 Percent of Income on Health Care,
Before and After Enactment of a Refundable Tax Credit
Percentage of Non-Elderly That Spend More Than 10% of Income on Health Care Tax Credit Amount (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Before tax credit 26.4 26.4 26.4
After tax credit 27.7 27.5 24.9

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

How the tax credit affects consumer financial risk depends heavily on the consumer's insurance status prior to the credit; consumer financial risk would increase substantially for those who were formerly uninsured.

Using the COMPARE microsimulation model, we are able to compare consumer financial risk for those who had no insurance prior to the policy change and those who were already purchasing nongroup insurance, as shown in Table 3.

Table 3
Median Percent of Household Income Spent on Health Care,
Before and After Enactment of a Refundable Tax Credit,
By Insurance Status Prior to Policy Change
Median Percent of Income Spent on Health Care, by Insurance Status Prior to Policy Change Tax Credit Amount (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Uninsured before tax credit 1.3 1.6 2.0
After tax credit (among previously uninsured) 17.2 10.0 5.0
Had nongroup insurance before tax credit 16.3 16.3 16.3
After tax credit (among previously on nongroup) 16.3 14.3 11.6

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

The two major groups of individuals who are likely to benefit from a tax credit for the purchase of nongroup insurance have very different spending patterns prior to the reform. Among those who are uninsured before the policy change, the median percent spent on health care ranges from 1.3 to 2.0 percent compared to 16.3 percent for those who had previously purchased nongroup insurance. After the policy change, the median percent of income spent on health care among those who had previously been uninsured increases whereas it generally declines for those who had previously purchased nongroup insurance.

For the newly insured overall, we estimate that the proportion paying more than 10 percent of income for health care will increase, primarily as a result of paying premiums (Table 4). For example, if the premium for an individual is $3,000 annually and that individual is eligible for the full amount of the credit, then under the scenario for the lowest credit amount, that person would pay $2,000 in premiums. This amount translates into 13 percent of a $15,000 annual income for premiums alone. If that individual increased utilization (which the model predicts would happen as a result of obtaining coverage), then some additional spending is likely to occur. The percentage of households paying more than 10 percent of income on health care declines as the size of the credit increases, from about 72 percent to about 33 percent. Some individuals who receive the credit to purchase health insurance will have previously had nongroup insurance. For those individuals, the proportion of households spending more than 10 percent of income on health care declines at the higher two credit amounts as a result of having access to a credit to partially or fully pay their premiums.

Table 4
Percentage of Households That Spend More Than 10 Percent of Income
on Health Care, Before and After Enactment of a Refundable Tax Credit,
by Insurance Status Prior to Policy Change
Percent of People in Households Spending More than 10% of Income on Health Care Tax Credit Amount (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Uninsured before tax credit 17.7 18.1 20.0
After tax credit (among previously uninsured) 71.6 50.3 32.9
Had nongroup insurance before tax credit 75.0 75.0 75.0
After tax credit (among previously on nongroup) 75.3 69.1 56.5

SOURCE: RAND COMPARE microsimulation modeling results, December 31 2008.

Previous studies have estimated similar levels of consumer financial risk under refundable tax credit policies.

Reschovsky and Hadley (2004) estimated that a $1,000 refundable individual tax credit to individuals with incomes under $15,000 and a maximum of a $3,000 credit to families with incomes under $25,000 would cover 40 percent of nongroup premiums such that the target population (low and moderate income uninsured) would still incur, on average, a $2,520 cost for premiums and services. They expected spending to increase for the target population under a refundable tax credit. For the target population prior to the tax credit, 75 percent of individuals spent less than 5 percent of their income on health care; after the credit, 60 percent will spend more than 10 percent of their income on health care. They argued that this policy puts older and sicker patients at greater risk to the extent that they pay higher premiums in the nongroup market.


References

Reschovsky JD, Hadley J, "The Effect of Tax Credits for Nongroup Insurance on Health Spending by the Uninsured," Health Affairs, Web Exclusive, February 25, 2004, pp. w4113–w4-123

top

The effect of tax credits for individuals to obtain coverage on waste is uncertain because of limited evidence, studies with mixed results, and the relatively small number of persons affected:

  • No studies directly evaluate the relationship between refundable tax credits and waste; published studies do not provide a clear direction on the likely effect. Read more below
  • Economic theory suggests that tax credits will discourage individuals from purchasing overly generous health plans, which tend to increase the use of low value care. Read more below
  • Our modeling estimates that 2.3 to 10 million persons who were previously uninsured would become insured under this policy change, which is not likely to have a large overall effect on reducing waste. Read more below

No studies directly evaluate the relationship between refundable tax credits and waste; published studies do not provide a clear direction on the likely effect.

No empirical studies directly evaluate the relationship between refundable tax credits and waste. However, several studies support the notion that the uninsured tend to get care in costly settings, such as the emergency department (ED) (Walls, Rhodes, Kennedy, 2002; Begley et al., 2006; Billings, Parikh, Mijanovich, 2000). Other studies have pointed out that while the uninsured may use EDs more often for primary care, the insured actually use more ED services than the uninsured (Hunt, 2006; Cunningham, 2006; Weber et al., 2005, Cunningham, May, 2003). A number of factors contribute to inappropriate use of EDs for primary care, and the tax credits in and of themselves do not eliminate other barriers to care. For example, individuals in medically underserved areas may not have sufficient access to primary care, even with insurance.

The literature also shows that, as cost sharing increases (as would be the case with a less generous insurance policy, such as a high deductible health plan), use of health services decreases. In particular, the RAND Health Insurance Experiment found that individuals with the highest level of cost sharing reduced their use of services across the board by approximately 30 percent—especially among low income and less healthy adults and children (Newhouse and the Insurance Experiment Group, 1993). Both appropriate and inappropriate care were eliminated in the HIE suggested that the net effect on waste is unknown. In the Swiss system, citizens can choose between different deductibles; analysts found that higher deductibles were associated with a lower probability of going to the doctor: Approximately one third of the reduction was due to increased price consciousness among patients, and the remaining two thirds was attributable to selection of healthier patients (Gerfin, Shellhorn, 2006).

Economic theory suggests that tax credits will discourage individuals from purchasing overly generous health plans, which tend to increase the use of low value care.

Keeler and colleagues (1996), who examined the effect of medical savings accounts (MSAs) (an early form of health savings account) on health spending, found evidence that high deductible health plans coupled with MSAs could reduce excess medical care use among the insured, depending on the size of the deductible. In their study of the relationship between high deductible health plans and ED use, Wharam and colleagues (2007) found a 10 percent reduction in total ED visits, which they attributed primarily to a 25 percent reduction in low severity, repeat visits. The study also reported that a disproportionate share of the reduction in high severity, first time visits occurred in the two lowest income groups (a 25 percent reduction) compared with the two highest income groups, for which there was a 1.3 percent reduction.

In contrast to other types of incentives, a refundable tax credit may encourage the purchase of lower cost policies, which in turn may lead to more cost conscious use of health care. In the group insurance market, premiums of any amount are exempt from taxes, so employers and employees have an incentive to pick very generous policies; picking such policies leads to greater use of "low value care" (care where the expected value from the service is lower than the total price) and may contribute to health care inflation. In comparison, a refundable tax credit reimburses individuals only up to the maximum of the credit and thus encourages the purchase of lower cost/less generous policies; such policies may discourage the use of low value services, because enrollees will bear a greater portion of costs. However, there is no empirical data to demonstrate that people reduce only inappropriate care when faced with high cost sharing.

A refundable tax credit may also affect administrative waste. Some portion of administrative overhead is currently devoted to care for the uninsured; this portion would decrease as previously uninsured individuals move into systems of care already in place for handling insured individuals. Conversely, the influx of previously uninsured patients into the insurance system could, at least in the short term, increase administrative overhead. The size of the net effect is unknown.

Our modeling estimates that 2.3 to 10 million persons who were previously uninsured would become insured under this policy change, which is not likely to have a large overall effect on reducing waste.

Because the number of people likely to newly obtain insurance under this policy change is small and the effects of insurance on waste are largely unknown, we would not expect to see a major change in waste (administrative, clinical or operational) as a result of tax credits.


References

Begley CE, Vojvodic, RW, Seo M, Burau K, "Emergency Use and Access to Primary Care: Evidence from Houston, Texas," Journal of Health Care for the Poor and Underserved, Vol. 17, No. 3, August 2006, pp. 610–624

Billings J, Parikh N, Mijanovich T, Emergency Room Use: The New York Story, New York, NY.: The Commonwealth Fund, Issue Brief #434, November 2000.

Cunningham PJ, "What Accounts for Differences in the Use of Hospital Emergency Departments Across U.S. Communities?" Health Affairs, Web Exclusives, [Epub July 18 2006], Vol. 25, No. 5, September/October 2006, pp. w324–w336.

Cunningham PJ, May JH, Insured Americans Drive Surge in Emergency Department Visits, Washington, DC: Center for Studying Health Systems Change, Issue Brief No. 70, October 2003, pp. 1-6.

Gerfin M, Schellhorn M, "Nonparametric Bounds on the Effect of Deductibles in Health Care Insurance on Doctors Visits—Swiss Evidence," Health Economics, [Epub April 4 2006], Vol. 15, No. 9, September 2006, pp. 1011–1020

Hunt KA, "Characteristics of Frequent Users of Emergency Departments," Annals of Emergency Medicine, [Epub March 30 2006], Vol. 48, No. 1, July 2006, pp. 1–8

Keeler EB, Malkin JD, Goldman DP, Buchanan JL "Can Medical Savings Accounts for the Nonelderly Reduce Health Care Costs?" The Journal of the American Medical Association, Vol. 275, No. 21, June 5 1996, pp. 1666-1671.

Newhouse JP, and the Insurance Experiment Group, Free for All? Lessons from the RAND Health Insurance Experiment, Cambridge, MA.: Harvard University Press, 1993.

Walls CA, Rhodes KV, Kennedy JJ, "The Emergency Department as Usual Source of Medical Care: Estimates from the 1998 National Health Interview Survey," Academic Emergency Medicine, Vol. 9, No. 11, November 2002, pp. 1140–1145

Weber EJ, Showstack JA, Hunt KA, Colby DC, Callaham ML, "Does Lack of a Usual Source of Care or Health Insurance Increase the Likelihood of an Emergency Department Visit? Results of a National Population Based Study," Annals of Emergency Medicine, [Epub October 24 2004], Vol. 45, No. 1, January 2005, pp. 4–12.

Wharam JF, London B E, Galbraith AA, Kleinman KP, Soumerai SB, Ross-Degnan D, "Emergency Department Use and Subsequent Hospitalizations Among Members of a High-Deductible Health Plan," The Journal of the American Medical Association, Vol. 297, No. 10, March 14 2007, pp. 1093–1102.

top

Creating tax incentives for individuals to obtain coverage will not affect the reliability of the health care system, although some individuals may be more likely to obtain needed care:

  • No empirical studies have directly explored the relationship between refundable tax credits and reliability. Read more below
  • Evidence on the relationship between insurance status and reliability suggests that there may be no effect. Read more below

No empirical studies have directly explored the relationship between refundable tax credits and reliability.

No empirical studies specifically examine the relationship between a refundable tax credit for health insurance and reliability. Most of the work that has been done in this area makes cross sectional comparisons between insured and uninsured individuals.

Evidence on the relationship between insurance status and reliability suggests that there may be no effect.

There are two ways to consider the effect of a tax credit on reliability—that is, on the likelihood that patients will receive necessary care. The first is whether the policy change directly affects the way the health delivery system functions. The second is whether individuals who become newly insured are more likely to receive appropriate services.

From the system perspective, we have no evidence to suggest that a tax credit will affect the way health services are delivered. Our estimates suggest that a relatively small number of people will become newly insured under this policy change (2.3 to 10 million, or 1 to 4 percent of the total population), making it extremely unlikely that significant changes in the system would occur.

For individuals who obtain insurance under this policy, the likelihood that they will receive appropriate care may increase, although the evidence for this possibility is mixed. Some evidence suggests that health insurance improves access to care and quality of care. Several studies have found substantial differences in quality measures between uninsured and insured populations (Ayanian et al., 2000; Ross, Bradley, Busch, 2006). Ayanian et al. (2000) found that uninsured adults were less likely to see a physician when needed. Ross, Bradley, and Busch (2006) found that the uninsured used fewer recommended health services. Buchmueller et al.'s (2005) review suggests that utilization for all types of services increases when coverage is expanded. There is also evidence that expanding insurance reduces avoidable hospitalizations for children (Buchmueller et al., 2005).

For those services, such as screening or preventive care, for which access is the primary determinant of the likelihood that appropriate care will be delivered, improved access will increase the reliability of care for the individual. However, strong evidence suggests that the health care system has significant deficiencies experienced by everyone who seeks health care and there is no reason to believe that newly insured individuals would not experience those same deficiencies. The most comprehensive study of quality in the United States found that, among those with at least one visit in two years, everyone was at risk for poor quality care (Asch et al., 2006). Leape et al. (1999) found no overall differences in underuse of coronary revascularization between uninsured and insured people. Young et al. (2001) found that "insurance and income had no effect on receipt of appropriate care for depressive and anxiety disorders." Harman, Edlund, and Fortney (2004) found disparities in initiation of depression treatment between the uninsured and the insured, but no differences in quality of care once treatment was initiated.

We do not know how the reliability of care for individuals who were previously uninsured will change when they acquire insurance; reliability will depend on their health care needs, their own care seeking behavior before and after the change in insurance status, and the quality of care delivered in the places they go for care.


References

Asch SM, Kerr EA, Keesey J, Adams JL, Setodji CM, Malik S, McGlynn EA, "Who Is at Greatest Risk for Receiving Poor-Quality Health Care? The New England Journal of Medicine, Vol. 354, No. 11, March 16 2006, pp. 1147–1156.

Ayanian JZ, Weissman JS, Schneider EC, Ginsburg JA, Zaslavsky AM, "Unmet Health Needs of Uninsured Adults in the United States," The Journal of the American Medical Association, Vol. 284, No. 16, October 25 2000, pp. 2061–2069.

Buchmueller TC, Grumbach K, Kronick R, Kahn JG, "The Effect of Health Insurance on Medical Care Utilization and Implications for Insurance Expansion: A Review of the Literature," Medical Care Research and Review, Vol. 62, No. 1, February 1 2005, pp. 3–30.

Harman JS, Edlund MJ, Fortney JC, "Disparities in the Adequacy of Depression Treatment in the United States," Psychiatric Services, Vol. 55, No. 12, December 2004, pp. 1379–1385.

Leape LL, Hilborne LH, Bell R, Kamberg C, Brook RH, "Underuse of Cardiac Procedures: Do Women, Ethnic Minorities, and the Uninsured Fail to Receive Needed Revascularization?" Annals of Internal Medicine, Vol. 130, No. 3, February 2 1999, pp. 183–192.

Ross JS, Bradley EH, Busch SH, "Use of Health Care Services By Lower-Income And Higher-Income Uninsured Adults," The Journal of the American Medical Association, Vol. 295, No. 17, May 3 2006, pp. 2027–2036.

Young AS, Klap R, Sherbourne CD, Wells KB, "The Quality of Care for Depressive and Anxiety Disorders in the United States," Archives of General Psychiatry, Vol. 58, No. l, January 2001, pp. 55–61.

top

No studies directly examine the link between refundable tax credits and changes in patient experience:

  • No empirical studies directly analyze the relationship between refundable tax credits and changes in patient experience. Read more below
  • Theory suggests that the patient experience of formerly uninsured individuals will improve if they acquire insurance. Read more below

No empirical studies directly analyze the relationship between refundable tax credits and changes in patient experience.

We do not have much systematic data about patient experiences among the uninsured: Most of the routine measurements of patient experience are conducted among those with insurance (private, Medicaid, Medicare). We also do not know how patients' experiences change when they go from being uninsured to being insured. Nor do we know whether patients who newly acquire insurance, particularly the coverage likely to be obtained under a tax credit policy, are likely to change the places they go for care and thus have a different experience than when they were uninsured.

Theory suggests that the patient experience of formerly uninsured individuals will improve if those individuals acquire insurance.

The uninsured population encounters considerable difficulties in both access to and continuity of care. To the extent that this policy option would allow people to move from being uninsured to having insurance, we expect that the patient experience will improve as people have greater access to care and are able to develop relationships with primary care physicians. Those who had coverage prior to the policy change may not see any change in patient experience.

The effect of this policy on patient experience may be somewhat tempered by the ability of the newly insured to obtain care. The individuals who obtain insurance under the policy we modeled have lower incomes and may live and work in areas with limited capacity. The policy option could result in people obtaining insurance but not gaining better access to care, resulting in a poor patient experience.

This policy change would also affect people who were already purchasing insurance and who would have some portion of their premium costs offset by the tax credit. They would be unlikely to dramatically change the places they go for care. So we would expect that their experiences with financial aspects of care improve but that they would see little or no change in other dimensions of patient experience.

We can use information from the existing literature to assess the likely effect. Multiple studies confirm that people without insurance have more negative experiences of care. Schoen and DesRoches (2000) compared the experience of the continuously insured population with that of the uninsured and discontinuously insured. They found that those individuals who experienced a gap in coverage and uninsured individuals were at higher risk of going without needed care and of having problems paying medical bills, and that they rated care more negatively than those with continuous insurance. Schoen et al. (1997) compared low income uninsured adults to those with public or private coverage and found that the uninsured were less likely to have a regular provider and rated care more negatively than those with insurance. Newacheck, Hughes, and Stoddard (1996) found that uninsured children are twice as likely to lack a usual source of care and twice as likely to wait 60 minutes or more for a health care visit than are insured children. Zyzanski et al. (1998) found that high volume providers suffer from "lower rates of preventive services delivery, lower patient satisfaction, and a less positive doctor-patient relationship."


References

Newacheck PW, Hughes DC, Stoddard JJ, "Children's Access to Primary Care: Difference by Race, Income and Insurance Status," Pediatrics, Vol. 97, No. 1, January 1996, pp. 26–32.

Schoen C, DesRoches C, "Uninsured and Unstably Insured: The Importance of Continuous Coverage," Health Services Research, Vol. 35, No. 1, Pt. 2, April 2000, pp. 187–206.

Schoen C, Lyons B, Rowland D, Davis K, Puleo E, "Insurance Matters for Low-Income Adults: Results from a Five-State Survey," Health Affairs, Vol. 16, No. 5, September/October 1997, pp. 163–171.

Zyzanski SJ, Stange KC, Langa D, Flock SA, "Trade-offs in High Volume Primary Care Practice," Journal of Family Practice, Vol. 46, No. 5, May 1998, pp. 397–402.

top

We estimate a small gain in life-years among those newly insured under a tax credit:

  • We estimate an increase of 360,000 to 1,510,000 life-years, depending on the size of the credit. Read more below
  • Theory and published studies suggest that, if a refundable tax credit for insurance increases rates of coverage, the health of some groups should improve. Read more below
  • The magnitude of the effect on health may depend on the health of an individual before gaining insurance and on other socioeconomic factors, as well as on changes in access afforded by health insurance and the reliability of care received. Read more below

We estimate an increase of 360,000 to 1,510,000 life-years, depending on the size of the credit.

We modeled a refundable tax credit under which individuals can claim an annual tax refund on dollar amounts up to either their premium payment or their maximum credit amount, whichever is lower. The tax credit in the model gives individuals and families below income thresholds of $15,000 and $30,000, respectively, access to the full amount of the tax credit. The size of the credit decreases on a sliding scale as income rises and phases out at $30,000 (individuals) and $60,000 (families). We assumed that 40 percent of uninsured individuals and families who are eligible for the credit would use it to purchase health insurance. In our analysis, we used projections from a modification of the RAND Future Elderly Model to determine the increase in life expectancy attributable to a change in insurance status (Goldman et al., 2004). These estimates suggest that becoming insured leads to an average increase in life expectancy of approximately 6 months, implying a reduction in annual mortality rates of approximately 5 percent.

The effects of varying tax credit amounts on expected gains in health are shown in Table 1.

Table 1
Expected Total Gains in Life-Years with a Refundable Tax Credit
Change in Life-Years Attributable to Policy Tax Credit Amount (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Millions of life-years 0.36 0.97 1.51

SOURCE: RAND COMPARE modification of Future Elderly Model, December 31 2008.

We assumed that individuals who obtain coverage under this policy option would retain that coverage continuously until they become eligible for Medicare at age 65. We further assumed that those who obtain insurance change from the expected mortality rate observed for the uninsured to the expected mortality rate observed for the insured. We observed no difference in mortality among the population over age 65 based on their insurance status prior to becoming eligible for Medicare; therefore, we assumed that all of the effect of becoming newly insured occurs between becoming insured and age 65, a mortality difference of 9.06 percent. We assumed that there are no changes in rates of treatment or in the effectiveness with which medical care is delivered as a result of this policy change.

The results presented here are based on microsimulation analyses, which use what is known from the literature about the relationship between health (in this case, mortality rates) and insurance to estimate the effect of a policy change. However, published studies to date are generally cross sectional comparisons, and we have no experimental or longitudinal data on which to estimate how a change in insurance status affects a population of individuals.

Theory and published studies suggest that, if a refundable tax credit for insurance increases rates of coverage, the health of some groups should improve.

Consistent with our modeling results, the literature suggests that there may be a modest relationship between insurance status and health outcomes, but methodological issues have made it difficult to make an accurate quantitative estimate. The RAND Health Insurance Experiment (HIE) randomly assigned families to health plans, providing an opportunity to assess how benefit generosity affected health outcomes in a setting where health was unrelated to insurance choice. Overall, the study found that benefit generosity had a negligible influence on health outcomes for the general population, although there were some benefits for low income participants who were in poor health at the beginning of the study (Newhouse and the Insurance Experiment Group, 1993). However, since everyone in the HIE had at least minimal health coverage, that population cannot necessarily be used to gauge the effect of becoming insured.

Levy and Meltzer (2008) reviewed the literature on the relationship between health insurance and health and drew several conclusions. First, most studies are not able to establish a causal relationship between health insurance and health because they do not account for the multiple other factors that affect these two variables or the fact that health itself affects insurance status. Second, there is substantial evidence that health insurance improves health in vulnerable populations, such as infants, children, individuals with HIV, and some low income adults, but there is less evidence of this relationship for other groups. Finally, they suggest that it may be difficult to generalize the results of the studies thus far.

Hadley (2003) reviewed the literature from the past 25 years on this subject and found that the general consensus among studies was that providing the uninsured with health insurance would result in improved health. Hadley's "best guess" is that mortality would decrease between 4 and 25 percent for previously uninsured people. Our modeled estimate of 9.06 percent is contained within this range. Although Hadley cites methodological difficulties within these studies, he acknowledges the consistency in the findings of improved health with insurance in many populations and disease states.

The magnitude of the effect on health may depend on the health of an individual before gaining insurance and on other socioeconomic factors, as well as on changes in access afforded by health insurance and the reliability of care received.

Medical care represents only one of the many determinants of health (others being behavior, socioeconomic status, education, and genetics, for example); consequently, improved access to medical care via insurance changes may have only a modest effect on health. McGinnis, Williams-Russo, and Knickman (2002) suggest that only 10 to 15 percent of preventable deaths are attributable to problems with medical care. In addition, there are other important determinants of access apart from health insurance coverage.

Estimating how insurance affects health poses significant methodological challenges. In particular, insurance status and health are not independent; that is, health can directly affect the ability or desire to obtain coverage. Healthy people may be less likely to purchase insurance because they anticipate having minimal health expenditures. Sick individuals may be unable to purchase individual policies. Other unobservable characteristics may also influence both health and insurance status. Thus, when researchers attempt to study differences in health based on health insurance status, it can be difficult to discern whether having insurance really causes improved health.


References

Goldman DP, Shekelle PG, Battacharya J, Hurd MD, Joyce GF, Lakdawalla DN, Matsui DH, Newberry SJ, Panis CWA, Shang B, Health Status and Medical Treatment of the Future Elderly: Final Report, Santa Monica, CA.: RAND Corporation, TR-169-CMS, August 2004.

Hadley J, "Sicker and Poorer-The Consequences of Being Uninsured: A Review of the Research on the Relationship Between Health Insurance, Medical Care Use, Health, Work, and Income," Medical Care Research and Review, Vol. 60, No. 2 Suppl., June 1 2003, pp. 3S–75S.

Levy H, Meltzer D, "The Impact of Health Insurance on Health," Annual Review of Public Health, Vol. 29, April 2008, pp. 399–409.

McGinnis JM, Williams-Russo P, Knickman JR, "The Case for More Active Policy Attention to Health Promotion," Health Affairs, Vol. 21, No. 2, March/April 2002, pp. 78–93.

Newhouse JP, and the Insurance Experiment Group, Free for All? Lessons from the RAND Health Insurance Experiment, Cambridge, MA.: Harvard University Press, 1993.

top

Using our model, we estimate small increases in the number of people who become newly insured as a result of this policy change:

  • Given our model results, we expect that 2.3 to 10 million people would become newly insured under this policy change. Read more below
  • Published estimates of the magnitude of the increase vary widely and depend on assumptions about the size and structure of the credit; the extent to which the credit is used to purchase coverage; and estimates of how coverage will shift among individual, employer-sponsored, and public programs. Read more below

Given our model results, we expect that 2.3 to 10 million people would become newly insured under this policy change.

Coverage is expected to increase under a refundable tax credit, since consumers would face lower costs when they decide to purchase health insurance. However, the exact effect will depend on the size and form of the incentive, as well as on what coverage types, if any, are tied to the policy change (e.g., eliminating the tax exempt status of group coverage, mandates to purchase coverage). Making the tax credit refundable can help low income individuals (who pay no or little income tax) utilize the credit to purchase insurance.

We modeled a refundable tax credit with the following design:

  • Individuals and families can claim an annual Federal tax refund on money spent to purchase a nongroup insurance policy.
  • Individuals and families below income thresholds of $15,000 and $30,000, respectively, are eligible for the full amount of the tax credit.
  • The tax credit amount decreases on a sliding scale as income rises; it phases out at $30,000 (individuals) and $60,000 (families).
  • To qualify for the credit, individuals or families must purchase a policy with a predefined minimum benefit package.
  • The amount of the tax credit that is claimed cannot exceed the amount spent on premiums.
  • The consumer pays premium costs in excess of the credit amount.
  • Individuals and families that want to purchase insurance are able to do so (that is, we designed this policy option with guaranteed issue).

We made the following assumptions in modeling the tax credit option:

  • The elasticity of take-up among the previously uninsured is 0.4 (that is, 40 percent of those who did not have insurance prior to the policy change will elect to take advantage of the credit).
  • Some individuals with group coverage will choose to switch to nongroup coverage to take advantage of the credit.
  • Some individuals with Medicaid will switch to nongroup coverage.
  • Some employers will drop coverage altogether.

Although this policy option would likely occur in conjunction with other policies, we modeled a refundable tax credit for individuals and families separate from other policy options.

Table 1 shows expected changes in levels of coverage that would occur with a refundable tax credit. Depending on the size of the credit, the number of uninsured would be reduced by 2 to 10 million (by 4 to 22 percent). A small number of people will switch from group to nongroup insurance to take advantage of the credit. Another small group switches from Medicaid to nongroup insurance. Our model assumes that some employers who sponsored insurance prior to the tax incentive will cease to provide coverage after the policy change, resulting in 640,000 to 3.2 million people becoming newly uninsured.

Table 1
Changes in Coverage Resulting from a Refundable Tax Credit, by Category of Change
Coverage (in millions) Tax Credit Amount (Individuals/Families)
$1,000/$2,500 $2,500/$6,250 $5,000/$12,500
Newly insured 2.97 8.29 13.2
Newly uninsured* 0.64 1.6 3.2
Net newly insured** 2.3 6.7 10.0
Newly on nongroup 3.3 10.5 18.7
Switch from group to nongroup 0.14 1.27 3.3
Switch from Medicaid to nongroup 0.23 1.0 2.2

SOURCE: RAND COMPARE microsimulation modeling results, December 31, 2008.
NOTES: *All newly uninsured had group insurance prior to policy change.
**Net newly insured = newly insured - (minus) newly uninsured.

There are still several unknowns regarding this policy option: We do not know whether it will affect premium pricing, particularly in the individual market. We do not know how a national purchasing pool (or exchange) would affect the viability of the individual market. Nor do we know whether this credit will affect the products offered in the nongroup market (e.g., the extent to which private insurers may market products that offer less generous coverage after the credit takes effect).

Published estimates of the magnitude of the increase vary widely and depend on assumptions about the size and structure of the credit; the extent to which the credit is used to purchase coverage; and estimates of how coverage will shift among individual, employer-sponsored, and public programs.

Gruber (2008) modeled the effect of a refundable tax credit on coverage. Table 2 provides a comparison of the results of his estimates with those generated by the COMPARE microsimulation model for four scenarios. The COMPARE estimates tend to be smaller than those from Gruber, but the patterns of the effects observed when changing the amount of the credit or the income eligibility criteria are similar.

Table 2
Comparison of Coverage Results for COMPARE and Gruber (2008)
For Four Scenarios
  Scenario
POLICY PARAMETERS Scenario 1 Scenario 2 Scenario 3 Scenario 4
Lower income threshold for singles $15,000 $15,000 $25,000 $25,000
Upper income threshold for singles $30,000 $30,000 $50,000 $50,000
Lower income threshold for families $30,000 $30,000 $50,000 $50,000
Upper income threshold for families $60,000 $60,000 $100,000 $100,000
Credit for singles $2,500 $6,000 $2,500 $6,000
Credit for families $6,250 $15,000 $6,250 $15,000
Elasticity of take-up among uninsured 0.4 0.4 0.4 0.4
COMPARE RESULTS
Newly insured (millions) 8.3 14.2 10.3 17.3
Newly uninsured 1.6 3.3 3.7 6.1
Net newly insured 6.7 10.9 6.6 11.2
Newly on nongroup 10.5 20.8 14.4 30.1
Decrease in group insurance 2.9 7.5 6.6 16.1
Switch from Medicaid to nongroup 1.0 2.5 1.2 2.9
Gruber Results (NBER WP 13758)
Newly insured (millions) 7.0 13.0 9.0 16.0
Newly uninsured 2.0 3.0 4.0 6.0
Net newly insured 5.0 10.0 5.0 10.0
Newly on nongroup 12.0 22.0 17.0 32.0
Decrease in group insurance 5.0 10.0 11.0 20.0
Switch from Medicaid to nongroup 1.0 2.0 1.0 2.0

SOURCE: RAND COMPARE microsimulation modeling, December 31, 2008; Gruber J, "Covering the Uninsured in the U.S.," Cambridge, MA.: National Bureau of Economic Research, Working Paper 13758, January 2008. As of January 2, 2008: http://www.nber.org/papers/w13758.

An earlier paper by Gruber and Levitt (2000) examined a policy to purchase nongroup coverage that would provide a refundable tax credit of $1,000 to individuals with incomes less than $40,000 (with the credit phasing out at $60,000) and a credit of $2,000 to households with income less than $75,000 (with phase-out at $100,000). The policy would allow those with current group coverage to switch to nongroup coverage, but the credit could not be applied to group coverage. He assumed no change in the tax exempt status of group coverage. He estimated that 18.4 million people would take up the credit. Of this number, 4.7 million were previously uninsured (11 percent of the uninsured population), 8.6 million had nongroup coverage, and 4.7 million had group coverage. At a subsidy level of $2,000/$4,000, the number of uninsured who gain insurance would increase to 12 million. Gruber argued that it is very difficult to design a policy that will be effective at expanding insurance on a large scale while not being costly and that employers might respond to these tax credits by offering less generous plans. Burman and Gruber (2005) estimated that a $1,000 refundable tax credit to individuals with incomes under $15,000 and a maximum of a $3,000 incentive to families with incomes under $25,000 would expand coverage by 3.12 million.


References

Burman LE, Gruber J, Tax Credits for Health Insurance, Washington, D.C.: Urban-Brookings Tax Policy Center, Tax Policy Issues and Options, No. 11, June 2005.

Gruber J, "Covering the Uninsured in the U.S.," Cambridge, MA.: National Bureau of Economic Research, Working Paper 13758, January 2008. As of January 2 2008 available at: http://www.nber.org/papers/w13758.

Gruber J, Levitt L, "Tax Subsidies for Health Insurance: Costs and Benefits," Health Affairs, Vol. 19, No. 1, January/February 2000, pp. 72–85

top

Refundable tax credits for individuals to obtain coverage are not expected to change the overall capacity of the health care system:

  • We would not expect this policy option to change overall health system capacity because of the small number of people who become newly insured. Read more below
  • No empirical studies directly evaluate the effect of tax credits on capacity. Read more below

We would not expect this policy option to change overall health system capacity because of the small number of people who become newly insured.

Capacity refers to the human resources (personnel and their productivity) and capital (medical equipment, hospitals, etc.) of the health care system. There is no direct connection between capacity of the health system and tax credits to purchase health insurance or the resultant levels of coverage. Further, based on the design of the policy option we modeled, relatively few people would become newly insured, resulting in little change in the demand for services or in the amount of money available to pay for those services (see discussion under Coverage).

The size of any resulting change in utilization depends on assumptions about the extent to which demand for services will increase because of coverage, as well as the extent to which utilization patterns will change if uninsured people substitute preventive, primary, and chronic care for acute care. We estimated relatively small changes in the net number of newly insured people, so we do not expect that the resulting changes in utilization will create a major change in market conditions. We also assumed that the supply of health care resources will not adjust quickly to changes in demand.

We do not know whether any geographic areas will experience a significant change in market conditions based on significant local improvement in the proportion of people with insurance coverage whose utilization increases significantly. Nor do we know whether other trends in capacity (e.g., changes in availability of hospital supply, primary care physicians, retail clinics) will affect the ability of newly insured individuals to access care.

No empirical studies directly evaluate the effect of tax credits on capacity.

Empirical evidence suggests that increased coverage leads to increased health care utilization, which could, in turn, affect capacity. Many studies have shown that uninsured people use fewer health care services than insured people, and that changes in coverage are associated with changes in health care utilization. Buchmueller et al.'s (2005) review suggests that utilization for all types of services increases when coverage is expanded. However, there is some evidence that expanding insurance reduces avoidable hospitalizations for children (Buchmueller et al., 2005). They note that the effect of utilization may be overstated, since utilization will vary across subgroups. For example, if coverage is expanded primarily to young, healthy individuals, we would not expect a large increase in utilization.


References

Buchmueller TC, Grumbach K, Kronick R, Kahn JG, "The Effect of Health Insurance on Medical Care Utilization and Implications for Insurance Expansion: A Review of the Literature," Medical Care Research and Review, Vol. 62, No. 1, February 1 2005, pp. 3–30.

top

A refundable tax credit would present a moderate degree of challenge for implementation:

  • A refundable health insurance tax credit should be less burdensome to implement than some other insurance coverage policy options, since it would be administered through the federal tax system and since other refundable federal tax credits already exist. Read more below
  • Implementing a refundable health insurance tax credit would require expanding and improving some mechanisms used to administer existing refundable tax credits. Read more below
  • The extent of the additional complexity and difficulty would depend on a number of factors, including the scope of the credit, and would likely involve a trade-off among reaching the target population, minimizing fraud, and increased administrative costs. Read more below

A refundable health insurance tax credit should be less burdensome to implement than some other insurance coverage policy options, since it would be administered through the federal tax system and since other refundable federal tax credits already exist.

Although there is no direct evidence of the operational feasibility of a refundable tax credit for health insurance, the credit would be administered through the federal tax system, which already provides for refundable tax credits. By making use of both public (i.e., the Internal Revenue Service--IRS) and private (i.e., taxpayer) resources used to administer the tax system and building on the experience from existing credits, a refundable health insurance credit should be less burdensome and costly to administer than other health insurance policy options of comparable scope that require entirely new institutional structures.

Two refundable credits that are part of the federal income tax are most relevant to the implementation of a refundable health insurance tax credit: the earned income credit (EIC), which is for low-income workers and their families, and the health coverage tax credit (HCTC), which is for dislocated workers (Batchelder, Goldberg, and Orszag, 2006; Holtzblatt, 2008).

The EIC is a refundable credit designed to supplement wages through the tax system. The EIC provides a credit of up to 40 percent of earned income for claimants with two or more qualifying children (34 percent of earned income for those with one qualifying child), with a maximum amount of $4,824 in 2008 for families with two qualifying children. The EIC is targeted toward lower- and middle-income taxpayers and phases out completely at $41,646 (in earned income or adjusted gross income, whichever is greater) for a married couple with more than one qualifying child ($36,995 for a married couple with one qualifying child).

Wage earners without qualifying children, including single individuals, are eligible for a much smaller credit of up to 7.65 percent of earned income, with a maximum of $438. This component of the EIC phases out at lower income levels than the credit for families with children. For example, for a married couple filing jointly, the credit phases out completely at $15,880 of adjusted gross income in 2008 (IRS, 2009).

The EIC is viewed as a relatively successful program that reaches a substantial share of its intended population (Holt, 2006). In 2007, the most recent year for which data are available, about 24.6 million taxpayers received the EIC (IRS 2007b). About 3.4 million returns received EIC to offset income tax liability before credits, 5.3 million returns received EIC to offset other taxes, and 21.6 million returns received refundable EIC (IRS 2007a). (The subtotals do not add to the total since some returns received more than one form of EIC.) Moreover, as a result of reforms undertaken in recent years, previous concerns about compliance problems, which resulted in erroneous or intentionally fraudulent claims for the credit, have abated, although they have not been entirely resolved (Batchelder, Goldberg, and Orszag, 2006).

The HCTC is a much smaller program that pays 65 percent of premiums for qualified health insurance coverage for (1) workers who lose employment because of trade liberalization or (2) retirees who lose pension payments because of employer default on pension obligations. It provides benefits to a monthly average of 16,000 households (Dorn, 2008). Even though the HCTC is a much smaller program than the EIC, it is relevant to a refundable health insurance credit by virtue of its availability to those who are not employed.

Implementing a refundable health insurance tax credit would require expanding and improving some mechanisms used to administer existing refundable tax credits.

Holtzblatt (2008) has identified three distinct but related issues that need to be addressed by administration of a refundable health insurance credit or comparable health insurance tax subsidies: whether the program reaches the intended population, how the program can be structured to provide insurance to households that lack the cash flow necessary to purchase it in advance, and whether eligibility for the program can be verified. The first two issues concern the ability of a health insurance credit to encourage its intended beneficiaries to purchase health insurance. The third issue concerns the extent to which availability of program funds can be limited to those who have acquired qualifying health insurance and whose income levels are appropriate for the amount of credit received. The EIC and the HCTC employ mechanisms that can be useful in addressing all three issues. Neither type of credit, however, is a complete or entirely adequate model in these respects.

While implementation of a new health insurance credit could build on experience with the EIC and the HCTC, it is likely that additional design features would be needed to ensure, first, that the credit would reach nonworkers (unlike the EIC) and those who do not file a return (unlike both the EIC and the HCTC); second, that it would provide a more effective advance payment mechanism than either existing credit; and, third, that it would achieve both of these goals at as low as possible cost in erroneous and excessive payments of the credit. In general, these additional features would add complexity and difficulty to implementation of the health insurance credit compared with administration of the EIC in its current form, but the credit’s features could accomplish some simplification compared with the mechanism for the HCTC. We discuss these issues in more detail below.

Mechanisms for reaching the intended population. To claim the EIC, it is necessary to file a tax return, whether or not tax is due. Studies of EIC participation rates for a variety of populations indicate that a relatively high percentage of eligible workers claim the federal EIC and that this percentage has increased over time (e.g., in 1999 as many as 83 percent of all eligible households, and 94 percent of those required to file tax returns) (Holt, 2006). The EIC is available only to taxpayers with earned income and, therefore, only those who are employed.

The health insurance credit, on the other hand, would be intended for both workers and nonworkers, and many of the latter do not file tax returns. According to Holtzblatt (2008), there were about 29 million who did not file in 2007, of whom more than 28 million had no wage income. An additional 8 million of the 37.5 million who filed reported no income tax liability and also had no wage income. Thus, a health insurance credit would require some mechanism to reach those who do not file returns.

The HCTC is designed specifically to provide those who are not employed with a tax subsidy to purchase health insurance. However, the program has not been entirely successful at reaching its intended recipients. The participation rate for the HCTC is much lower than that for the EIC: According to one estimate, only about 22 percent of individuals eligible for the HCTC actually receive the credit (Dorn, Varon, and Pervez 2005). Reasons given for this low participation rate include the complexity of the application process and participant cash flow problems (participants have to obtain insurance and pay the first three months of premiums before becoming eligible for the credit) (Dorn, 2008; Hevener and Kerby, 2008). Another possible problem is that public awareness of the HCTC is limited because of the small scale of the program.

It is also the HCTC’s small scale that makes it difficult to say with confidence how readily and effectively its mechanisms could be adapted to administer a broader health insurance credit available to unemployed as well as employed individuals (Holtzblatt, 2008). Moreover, like the EIC, the HCTC is available only to tax filers. However, there are reasons to believe that a broader health insurance credit could make use of the experience from the HCTC while avoiding some of its problems. For example, demonstrating eligibility for the health insurance credit should be somewhat simpler than for the HCTC (e.g., there would be no need to establish that the beneficiary has suffered job displacement due to international trade), which could in turn make it possible to simplify the application process.

Greater spending on public relations campaigns and other outreach efforts could help increase utilization of the health insurance credit by its intended beneficiaries above the relatively low participation rates for the HCTC. Some of those eligible for the health insurance credit would neither be employed nor be obligated to file tax returns, and outreach programs would also need to be designed to reach this population. However, it is possible that federal, state, and local agencies, as well as health care providers that are not directly involved in administering the health insurance credit, could participate in outreach programs. In addition, public awareness of the health insurance credit should be greater than for the HCTC because of the larger scope of the proposed program, and economies of scale in publicizing the program should be possible.

Program structure that allows for advance payment of credit. A health insurance credit would require a significant number of eligible participants to claim the credit in advance, and it would need to allow advance payment of the credit outside of an employment relationship. Workers with a qualifying child claim up to 60 percent of the EIC in advance through reduction of wage withholding by employers. The U.S. Government Accountability Office (2007) has estimated, however, that only 3 percent of eligible taxpayers claim the EIC in advance. A health insurance credit would need to improve substantially on this low participation rate.

The HCTC can be paid in advance and thus provides a possible model for a health insurance credit. Though the total participation rate is low, slightly over half of those receiving the HCTC take advantage of advance payment. As noted above, a possible reason for the HCTC’s overall low participation rate is that taxpayers must be enrolled in a qualified health insurance plan to be eligible for the HCTC, which in turn requires that up to three months of premiums may need to be paid in full before the HCTC payments start. Another reason for low enrollment may be that coverage of preexisting conditions can be excluded for up to 12 months (a barrier that would be eliminated in legislation under consideration in the 111th Congress). These limitations may reduce incentives to enroll for those who are in fact eligible (Dorn, Varon, and Pervez, 2005). The administrative mechanism for the HCTC is also quite cumbersome and expensive, requiring multiple transactions by the IRS and the Treasury Department per beneficiary each month, both with beneficiaries and insurance plans (Dorn, 2008).

The advance payment problem could be addressed by allowing individuals to apply for a determination that they qualify for the program without already being enrolled in an insurance plan. This is the approach taken by other health subsidy programs, such as Medicare, for which enrollees make partial premium payments. The problem regarding preexisting conditions could be overridden legislatively in the case of the health insurance credit, either by an outright prohibition or by prohibiting insurers participating in the credit program from denying coverage of preexisting conditions.

Verification of program eligibility. Administration of a refundable health insurance tax credit would also require mechanisms to verify eligibility. The EIC has had mixed experience in this respect. Unlike government transfer systems, which generally require program eligibility to be established before payments are made and often require third party verification, the tax system follows a "self-assessment" model to determine eligibility. Concerns about claims of the EIC by ineligible taxpayers or in excessive amounts resulted in several statutory and administrative changes in the past two decades, and there is evidence that subsequent reforms have reduced error rates. Nevertheless, according to one estimate, roughly 23 to 28 percent of the EIC claims are paid in error. This compares with an error rate of less than 6 percent for food stamp benefits, although participation in the food stamp program is considerably lower than for the EIC (Holtzblatt, 2008). Moreover, unlike the EIC, a health insurance tax credit would be available only to those who purchase health insurance. Verifying that an individual has purchased insurance, and thus has satisfied the purpose of the credit, would create additional enforcement and compliance difficulties and costs.

Establishment of eligibility for the HCTC involves an extensive application process, including verification of enrollment in a qualified plan and that the beneficiary qualifies for the program by virtue of job displacement or default on pension obligations. According to one source, the HCTC has largely avoided fraudulent claims, but its administrative costs are high, administration of the program is cumbersome, and only a limited universe of insurance plans is eligible to participate in the program (Dorn, 2008).

The extent of the additional complexity and difficulty would depend on a number of factors, including the scope of the credit, and would likely involve a trade-off among reaching the target population, minimizing fraud, and increased administrative costs.

Experience with the EIC and the HCTC indicates that there is a trade-off between coverage and compliance. Streamlined administration mechanisms are likely to expand the number of people covered by the program, but they may also tend to increase payments of the credit to unintended beneficiaries and to result in excessive payment amounts. Maximizing coverage while also minimizing erroneous payments is likely to require increased expenditures for program administration.

One design feature that should simultaneously promote increased coverage, improve compliance, and lower costs would be to make the substantive rules for the health insurance credit as simple as possible, so that recipients would be able to apply the rules more accurately and easily (Dorn, Varon, and Pervez, 2005). If possible, it would be desirable to coordinate the rules for the credit with simplification of the Internal Revenue Code’s complex and multiple definitions of household and dependent status for income tax purposes (Graetz, 2008).

Reaching the target population. Successful implementation of a health insurance tax credit could involve several components. Dorn, Varon, and Pervez (2005) have argued that a health insurance tax credit would better reach its target population if the credit (1) limits premium costs for the low income uninsured; (2) does not require full premium payments while applications are pending; (3) provides access to coverage that beneficiaries value, including care for preexisting conditions; (4) is combined with outreach that uses easily understandable, multilingual materials and proactive enrollment efforts; and (5) features a simple application process involving one form filed with one agency. Each of these components of a credit program, as well as an expanded advance payment system, would tend to increase per beneficiary costs by comparison with those of the EIC and the HCTC. In the case of outreach programs and an expanded advance payment mechanism, this increase could be offset, or more than offset, by administrative economies of scale, especially by comparison with the HCTC.

Minimizing fraud. Compliance concerns can be addressed, although not eliminated, by using increased third party verification of taxpayer provided information, insurer verification of insurance purchase, and employer verification of income. Avoidance of fraudulent use of advance payment might also require cooperation of insurers, in addition to employers, as these payments are made (i.e., as insurance is purchased). Regarding an advance payment mechanism, insurers might also serve as points of contact for intended beneficiaries of the health insurance credit who do not currently file tax returns, or the IRS or another government agency could serve that function. These elements of a refundable health insurance credit program would generate increased costs and complexity in administering the credit, both for the third parties and the IRS.

Earnings verification is already conducted to some extent for the EIC, but this mechanism would need to be augmented to make a refundable health insurance credit available to non-wage-earning households. Insurers might also be asked to serve this function. Alternatively, it might be necessary for this verification to occur after the fact through the existing tax information reporting system, perhaps supplemented by information provided by insurers. Since household income can vary throughout the year and insurance must be consistently maintained month-to-month, it might also be necessary to adopt somewhat simplified income rules, including use of prior year income, to establish eligibility for the health insurance credit. Each of these measures, however, would likely result in some increase in error rates.

Administrative costs. A review of the history of the EIC found that administrative costs of the program ranged from 1 to 3 percent of program costs (Holt, 2006). This amount does not, however, include costs incurred by private parties. By one estimate, the HCTC administrative costs represent 12 percent of program costs, higher than those for the EIC, as a result of the HCTC’s smaller scale and the costs of its more elaborate advance payment mechanism. This estimate also does not reflect costs to private parties. The IRS estimates that increasing enrollment in the HCTC could achieve economies of scale: For example, tripling enrollment would increase administrative costs of the program to the government by only 40 percent, according to one estimate (Holtzblatt, 2008).

Administrative costs per beneficiary should be lower for a broader health insurance credit than for the HCTC, but they could be further lowered for the IRS and the Treasury by providing that payments be made to insurance companies on a quarterly (or other, less frequent) basis, with an interest adjustment for the payment delay (Dorn, 2008). It might also be possible for "payments" of credit amounts to be made by the government to insurers by allowing the latter to offset such amounts against estimated or other ongoing tax payment obligations.

Creation of a refundable health insurance credit would impose substantial new responsibilities on the IRS, and its budget would need to be increased to fulfill these responsibilities. The IRS has a relatively low budget compared with the magnitude of the revenues it collects and the programs it administers, largely because of the self-assessment model on which the tax system is based. IRS operations are structured primarily around the annual accounting and filing period for tax returns, while administration of the refundable credit would involve a more frequent schedule of operation and also require adaptation of IRS procedures for that purpose (Holtzblatt, 2008). If, as Dorn, Varon, and Pervez (2005) recommend, a single agency administers the health insurance credit and that agency is the IRS, the mission of the IRS would be extended to encompass those who do not file returns, who currently may have no contact with the IRS, even indirectly (as in the case of recipients of the EIC) through employers.

Many of these concerns would apply to government subsidies for health insurance administered by agencies other than the IRS, however; and such agencies would also encounter difficulties that the IRS would not face. By comparison with the IRS, agencies providing means-tested assistance typically have more limited contact with moderate-income families, but they may have greater contact with very low income and non-wage-earning families.

A full assessment of the operational feasibility of the refundable credit must take into account these and other differences between administration by the IRS and by other government agencies. Accordingly, such an assessment requires a comparison of the strengths and weaknesses of the IRS relative to other agencies that might administer new health insurance subsidies, and of use of the tax system for that purpose relative to alternatives, including other existing programs (Holtzblatt, 2008).

References

Batchelder L, Goldberg F, Orszag P, "Efficiency and Tax Incentives: The Case for Refundable Tax Credits," Stanford Law Review, Vol. 59, No. 1, October 2006, pp. 23-76.

Dorn S, "Comment," in Aaron HJ, Burman LE, Using Taxes to Reform Health Insurance: Pitfalls and Promises, Washington, D.C.: The Brookings Institution, 2008, pp. 199-207.

Dorn S, Varon J, Pervez F, Limited Take-up of Health Coverage Tax Credits: A Challenge to Future Tax Credit Design, New York, N.Y.: Commonwealth Fund, Publication 869, Issue Brief, Vol. 15, October 2005.

Graetz MJ, 100 Million Unnecessary Returns: A Simple, Fair, and Competitive Tax Plan for the United States, New Haven, Conn.: Yale University Press, 2008.

Hevener MBH, Kerby CK, "Administrative Issues: Challenges of the Current System," in Aaron HJ, Burman LE, Using Taxes to Reform Health Insurance: Pitfalls and Promises, Washington, D.C.: The Brookings Institution, 2008, pp. 147-170.

Holt S, The Earned Income Tax Credit at Age 30: What We Know, Washington, D.C.: The Brookings Institution, Policy Brief, February 2006.

Holtzblatt J, "The Challenges of Implementing Health Reform Through the Tax System," in Aaron HJ, Burman LE, Using Taxes to Reform Health Insurance: Pitfalls and Promises, Washington, D.C.: The Brookings Institution, 2008, pp. 171-198.

Internal Revenue Service (IRS), 2007 Statistics on Income, Individual Income Tax Returns (Complete Report), Table 2, 2007a. As of December 7, 2009: http://www.irs.gov/

Internal Revenue Service (IRS), 2007 Statistics on Income, Individual Income Tax Returns (Complete Report), Table 4, 2007b. As of December 7, 2009: http://www.irs.gov/

Internal Revenue Service (IRS), Earned Income Credit: For Use in Preparing 2008 Returns, Washington, D.C.: U.S. Department of the Treasury, Publication 596, 2009. As of November 12, 2009: http://www.irs.gov/

U.S. Census Bureau, "State and County QuickFacts," rev. September 4, 2009. As of November 12, 2009: http://quickfacts.census.gov/

U.S. Government Accountability Office, Advance Earned Income Tax Credit: Low Use and Small Dollars Paid Impede IRS’s Efforts to Reduce High Noncompliance, August 10, 2007. As of November 12, 2009: http://www.gao.gov

top