RETIREMENT RESEARCH January 2012, Number 12-1 WHAT EXPLAINS VARIATION IN DISABILITY APPLICATION RATES ACROSS STATES? By Norma B. Coe, Kelly Haverstick, Alicia H. Munnell, and Anthony Webb* Introduction Social Security Disability Insurance (SSDI) applica- rates. Politics have little effect. Interestingly, states tions and benefit receipts vary greatly by state, which that require employers to provide temporary disability has led to concerns about potential inconsistencies in insurance have lower SSDI application rates. the way that states apply disability standards.1 This possibility has prompted numerous Congressional hearings and reports, and led the Social Security SSDI Application Decisions Advisory Board to express concern about the Social Security Administration’s ability to disentangle the at the Individual Level potential causes. This brief, using a longer time In theory, an individual’s decision to apply for SSDI is period and more comprehensive list of variables than a matter of weighing the costs and benefits of applica- other studies, explores the extent to which health, tion: one applies if it increases the expected present demographic, and employment characteristics – as value of lifetime utility. Individuals are eligible for well as state policies or politics – explain the variation SSDI if they are not currently earning more than across states. $1,000, are unable to do so for at least a year, and The discussion proceeds as follows. The first sec- have worked long enough and recently enough to tion describes an individual’s SSDI application deci- be covered.2 Workers who apply must weigh their sion and factors that may influence state-level applica- current earnings and future labor market opportuni- tion rates. The second section presents variables used ties against the future stream of SSDI benefits, plus to determine the underlying causes of the state-level Medicare coverage after two years, times the probabil- variation in application rates. The third section sum- ity of being accepted to the program, minus any costs marizes the results. The conclusion is that the health, of application.3 Thus, the health and demographic demographic, and employment characteristics of each characteristics of the individuals in each state and the state explain the largest variations in SSDI application nature of the job market would be important factors explaining the variation among states in SSDI ap- plication rates. * Norma B. Coe is associate director for research at the Center for Retirement Research at Boston College (CRR). Kelly Haverstick is a former CRR research economist. Alicia H. Munnell is director of the CRR and the Peter F. Drucker Profes- sor in Management Sciences at Boston College’s Carroll School of Management. Anthony Webb is a research economist at the CRR. This brief is the first of two adapted from a longer paper (Coe et al. 2011). 2 Center for Retirement Research State policies and politics may also affect the ap- minimize the state’s own payments through the safety plication decision in the following ways. net.8 Further, the governor’s political party affiliation could indicate potential changes in welfare policy or Health Care generosity. If individuals are aware that politics may influence program leniency, or just observe an in- States are highly involved in determining access crease in the probability of acceptance to the program, to and affordability of health insurance. Previous politics may influence the application decision. findings that Medicaid generosity influences Medi- care use suggest that the value of Medicare coverage accompanying SSDI receipt is related to policies The Variables under a state’s control.4 In addition, states may limit This project analyzes state-level data over the period the ability of insurance companies to price coverage 1993-2009.9 The dependent variable is the annual based on individual health and demographic charac- SSDI application rate by state, expressed as a percent- teristics (“community rating”) and to deny coverage age of the state’s working-age population (age 18-64) (“guaranteed issue”), and states may even mandate not receiving SSDI benefits.10 As shown in Figure individual health insurance coverage. Studies show 1, average SSDI application rates between 1993 and that these regulations have a significant effect on 2009 varied substantially, ranging from 0.5 percent in coverage, and presumably also on subsequent health Utah to 1.4 percent in Mississippi. A strong regional care access.5 component is evident, with the South having much The effect of health care access on the SSDI ap- higher application rates and the West tending to have plication rate is theoretically ambiguous. On the lower rates. Possible explanations for the variation in one hand, individuals with access to health insur- SSDI application rates include health/demographic/ ance might be more likely to apply for SSDI because employment characteristics, state policies, and politi- they would be less likely to go uninsured during the cal factors.11 two-year waiting period for Medicare coverage.6 On the other hand, individuals might be less likely to apply for SSDI benefits because Medicare coverage is Figure 1. Average SSDI Application Rates, by relatively less attractive when they can obtain health State, 1993-2009 insurance elsewhere. Unemployment Insurance The unemployment insurance (UI) program is a federal-state partnership based on federal law and ad- ministered at the state level. The state sets the benefit structure (eligibility requirements and benefit levels) and tax structure (wage base and tax rates). Recent research finds that a more generous UI benefit delays SSDI application and that UI benefit exhaustion affects the timing of SSDI application.7 Thus, the hypothesis is that generous and/or long-lasting UI 0.5% benefits will reduce the SSDI application rate. Between 0.5% and 0.75% Between 0.75% and 1% Between 1% and 1.25% State Politics Greater than 1.25% Governors, who appoint the director of the state Dis- ability Determination Services, may wish directors Note: Washington, DC is between .5% and .75%. to be lenient in order to create political goodwill, to Source: Authors’ calculations. maximize federal income transfers into the state, or to Issue in Brief 3 Health, Demographic, and Employment • Male. States with a higher proportion of males would be expected to have higher SSDI Characteristics application rates due to their higher rates of labor force participation. State-level health characteristics come from the Center for Disease Control’s Behavioral Risk Factor • Married. States with a higher proportion of Surveillance Survey (BRFSS). The BRFSS has been married residents would be expected to have administered since 1984 and is the largest ongoing lower SSDI application rates since married telephone survey in the United States. BRFSS pro- people tend to be healthier. vides detailed data on self-rated health; health-related behaviors such as smoking and drinking; and factors • Poor. States with a higher proportion of their correlated with health conditions such as obesity, population under the federal poverty line along with state-of-residence indicators.12 Three would be expected to have higher SSDI ap- health variables from the BRFSS, all of which would plication rates. be expected to increase SSDI application rates, are used in the analysis: Variations among states and over time in employ- • self-reported fair/poor health status; ment characteristics – such as occupation, industry • smoking (ever smoked more than 100 ciga- composition, and the unemployment rate – are rettes); and expected to be associated with differences in SSDI • self-reported body mass index (BMI). application rates. Variables include: • Occupation and industry. The greater the Other important factors to be taken into account proportion of a state’s workforce employed when determining SSDI applications are the socio- in a blue-collar occupation or an agricultural economic composition and employability of potential industry, the higher the expected SSDI ap- applicants. The variables used in the analysis include: plication rate. • Age of the population. Younger populations are less likely to be insured by SSDI and less • Unemployment rate. Because greater unem- likely to have a disability that warrants an ap- ployment lowers the opportunity cost of apply- plication.13 Individuals age 50 plus face a dif- ing for SSDI, higher unemployment should ferent screening process, in which it is easier lead to more applications. to be accepted, so a state with a relatively older population would be expected to have a higher • Labor force participation rate. Discouraged SSDI application rate.14 workers may drop out of the labor force. So the lower the labor force participation rate, the • Education. States with a higher proportion of higher the expected application rate. their population with higher education would be expected to have lower SSDI application rates. The effect of low education is ambigu- State Policy ous. Individuals with less than a high school degree may be the most vulnerable, but also State policies with respect to unemployment insur- may not have enough steady work history to ance, health programs, and disability insurance could be insured under SSDI. also affect application rates. Variables include: • Maximum weeks of unemployment insur- • White, non-Hispanic. The impact of race is ance. The longer the duration of UI, the ambiguous. States with a higher proportion lower the expected SSDI application rate. of non-Hispanic whites could be expected to have lower rates of SSDI applications, because • UI benefits/average wage. The higher the ratio, non-whites are more vulnerable. Or whites the lower the expected SSDI application rate. could have higher application rates because they are more likely to have steady job histo- • Strict regulation of private insurance market.15 ries that enable them to qualify for SSDI. States are defined as strictly regulated if they have both community rating and guaranteed issue.16 As discussed earlier, the impact on application rates could be either positive or negative. 4 Center for Retirement Research • Medicaid buy-in. States with a Medicaid buy- Figure 2. Percent of Variation in State SSDI in program have less strict earnings qualifi- Application Rates Explained by Different cations for Medicaid eligibility for disabled Factors, 1993-2009 individuals who work, allowing better access 80% 72.5% 74.8% 75.2% to health insurance outside of the SSDI pro- gram.17 Medicaid buy-in states are expected to have lower SSDI application rates. 60% • State-mandated employer temporary disability 40% insurance (TDI). TDI programs, which were mostly enacted after the Great Depression, 20% provide workers with partial compensation for wages lost due to temporary, non-occupational 0% disabilities. Holding all else constant, the Health/demo./ Health/demo./empl. Health/demo./empl. five states that mandate employer TDI should empl. + state policies + state policies + politics have lower SSDI application rates.18 Note: Year fixed-effects adds 4.4 percent, leading to the R2 of 79.6 reported in the Appendix. State Politics Source: Authors’ estimates. Due to the concern about state politics influencing the administration of this federal program, three and political factors. The descriptive statistics for the variables are included to test whether the governor’s variables in the regression and the full results are party affiliation or tenure in the job appear to have shown in the Appendix. any influence on application rates. The variables are: Before discussing the individual variables, it is im- • governor’s party affiliation; portant to note the percent of the variation explained • an indicator for reaching the term limit; and by the three groups of factors. As shown in Figure • an indicator for an incumbent governor.19 2, health/demographic/employment variables alone explain over 70 percent of the variation; introducing state policies and politics adds relatively little explana- Results tory power. Figure 3 presents the coefficients from the regres- A regression equation related state SSDI application sion analysis that were statistically significant. Most rates over the period 1993-2009 to the state health/ of the health/demographic/employment variables demographic/employment variables, state policies, have the expected signs. Poor/fair health and high Figure 3. Impact of Selected Factors on SSDI Application Rates, 1993-2009 Fair/poor health 0.06 Poor 0.03 Labor force participation rate -0.06 White, non-Hispanic 0.03 Male -0.02 State-mandated employer TDI -0.12 Republican governor -0.03 -0.20 -0.10 0.00 0.10 0.20 Notes: All results are statistically significant at least at the 10-percent level. Standard errors have been clustered at the state level. The results shown for continuous variables are for a one-standard-deviation change; in the case of dummy variables, the results show a change from zero to one. Source: Authors’ estimates. Issue in Brief 5 levels of poverty increase a state’s SSDI application Conclusion rate. While the state-level unemployment rate is not significant, the discouraged-worker effect implied by This brief has examined why SSDI application rates the labor force participation rate is important.20 The vary so much between states. Not surprisingly, positive coefficient on the percent of the population health, demographic, and employment characteristics that is non-Hispanic white reflects the steady earn- are the major determinants of this state variation, ex- ings history needed to qualify for SSDI benefits. In- plaining over 70 percent of the variation in total SSDI terestingly, states with a high proportion of men have application rates. In addition, having state-mandated lower SSDI application rates – a puzzle. private TDI is associated with lower application rates, In terms of state policy variables, the only one to and the governor’s political party is also correlated have an effect is state-mandated, employer-provided with the application rate. In short, the health, demo- TDI. As many recent reform proposals argue, private graphic, and employment characteristics of a state short-term insurance policies may implicitly act as – not state policies or politics – explain most of the a pre-screening mechanism and assist in getting variation across states. individuals back to work before entering the perma- nent disability program. They may be more effective at getting their marginal claimants back to work, thus lowering total SSDI applications. The only political variable with a statistically significant effect is having a Republican governor. The coefficient suggests that a conservative political environment discourages applications. While it is interesting to see what is correlated with the SSDI application rates, it is important to put the marginal effects into context. Based on the regression coefficients, the state-level application rate would have averaged 1.0 percent between 1993-2009. If all health, demographic, and employment informa- tion were set at the best observed in the data (maxi- mum value for characteristics with a positive coef- ficient and minimum value for characteristics with a negative coefficient), the application rate would have been only 0.5 percent – a 50-percent drop. If every state were assigned the worst health, demographic, and employment characteristics, the predicted ap- plication rate increases to 1.5 percent. In short, the health, demographic, and employment variables have a profound effect on the application rates. 6 Center for Retirement Research Endnotes 1 See McVicar (2006); Bound and Burkhauser (1999); 11 We are grateful to Paul Davies of the Social Securi- and Rupp and Stapleton (1998). ty Administration (SSA) for providing the Title 2 (DI) only, Title 16 (SSI) only, concurrent Title 2 and Title 2 The $1,000 ceiling is the 2011 limit for non-blind 16 receipts by state for FY1993-FY2010. The FY1993- SSDI recipients. The limit for blind recipients is FY2000 receipts data came from paper records from $1,640. To be covered by SSDI, one must have SSA’s State Agency Operations Reports system. The worked a specified number of quarters overall and a FY2001-FY2010 receipts data are from SSA’s Payment specified number of quarters in recent years; both are Management System. a function of an individual’s age at disability onset. 12 While the BRFSS data include other health-related 3 For simplicity, our model assumes that SSDI recipi- variables that may be related to the SSDI applica- ents do not participate in the labor market again once tion rate (such as alcohol consumption, doctor visits, being accepted into the program. exercise habits, and mental health measures), these variables were not consistently available for all states 4 Cohen and Tumlinson (1997); and Pezzin and over the entire 1993-2009 period. Kasper (2002). 13 To be insured for SSDI, one must have worked 5 Buchmueller and DiNardo (2002); and Long and the required number of quarters based on age, and 20 Stockley (2009). quarters within the last 10 years. 6 This hypothesis is explored in Gruber and Kubik 14 Age is specifically in the SSDI determination (2002), who find that individuals with access to health process because the assessment of the ability to be insurance from a spouse are 26-74 percent more retrained changes depending on whether an applicant likely to apply for SSDI benefits than those without is age 50-54 (Approaching Advanced Age), 55-59 (Ad- external access to health insurance. vanced Age), or 60-64 (Retirement Age). 7 Lindner (2011); and Rutledge (2011). 15 Data on state regulations of health insurance were compiled from The Henry J. Kaiser Family Founda- 8 Iyengar and Mastrobuoni (2008) highlight this tion (2010a; 2010b), and Georgetown University classic principal-agent problem and find that states Health Policy Institute (2004). with first-term governors allow fewer applicants onto the rolls than states with re-elected governors. They 16 Herring and Pauly (2006). interpret this finding to mean that the SSDI rolls are manipulated for political purposes, but that there is a 17 These data were compiled from Kehn, Croake, and learning curve. Schimmel (2010); Croake and Liu (2009); Gruman et. al (2008); Jensen (2004, 2006); Georgia Depart- 9 Data are missing for: Wyoming in 1993, Rhode ment of Community Health (https://www.gmwd.org/ Island in 1994, Washington, DC in 1995, and Hawaii WebForms/StaticContent1.aspx); Delaware Health in 2004 because of lack of coverage in the Behavioral and Social Services (http://dhss.delaware.gov/dhss/ Risk Factor Surveillance Survey; and Nevada in 1994 dmma/); and Commonwealth of Kentucky (http:// due to lack of detailed data from Social Security on manuals.chfs.ky.gov/dcbs_manuals/DFS/VOLIVA/ SSDI-only applications. OMVOLIVA.pdf). 10 The denominator is the number of residents age 18 Five states enacted employer disability insurance 18-64 in a state as of July 1 from the U.S. Census mandates prior to the first year of data included in Bureau. From this figure we subtract the number this analysis: California (1946), Hawaii (1969), New of beneficiaries, obtained from the Social Security Jersey (1948), New York (1949), and Rhode Island Administration Statistical Bulletins (SSA 1994-2009), (1942) (U.S. Social Security Administration 2010). since current beneficiaries are not at risk of applying. Issue in Brief 7 19 The political variables come from National Gover- nors Association (2011) and Council of State Govern- ments (2007). 20 This finding is not explained by colinearity. If we estimate the relationship without the unemploy- ment rate, the labor force participation rate remains significant; if we estimate without the labor force participation rate, the unemployment rate remains insignificant. 8 Center for Retirement Research References Long, Sharon K. and Karen Stockley. 2009. “Health Insurance in Massachusetts: An Update on Insur- References for the data sources used in this brief are ance Coverage and Support for Reform as of Fall available in the full paper (Coe et al. 2011). 2008.” Washington, DC: Urban Institute. Bound, John and Richard V. Burkhauser. 1999. “Eco- McVicar, Duncan. 2006. “Why Do Disability Benefit nomic Analysis of Transfer Programs Targeted on Rolls Vary Between Regions? A Review of the People with Disabilities.” In Handbook of Labor Evidence from the USA and the UK.” Regional Economics, edited by Orley C. Ashenfelter and Da- Studies 40(5): 519-533. vid E. Card, 3417-3528. Amsterdam: Elsevier. Pezzin, Liliana E. and Judith D. Kasper. 2002. “Medi- Buchmueller, Thomas and John DiNardo. 2002. “Did caid Enrollment Among Elderly Medicare Ben- Community Rating Induce an Adverse Selection eficiaries: Individual Determinants, Effects of Death Spiral? Evidence from New York, Pennsyl- State Policy, and Impact on Service Use.” Health vania, and Connecticut.” The American Economic Services Research 37(4): 871-892. Review 92(1): 280-294. Rupp, Kalman and David Stapleton., eds. 1998. Coe, Norma B., Kelly Haverstick, Alicia H. Munnell, Growth in Disability Benefits. Kalamazoo, MI: and Anthony Webb. 2011. “What Explains State W.E. Upjohn Institute of Employment Research. Variation in SSDI Application Rates?” Working Paper 2011-23. Chestnut Hill, MA: Center for Rutledge, Matthew S. 2011. “The Impact of Unem- Retirement Research at Boston College. ployment Benefits Extension on Disability Insur- ance Application and Allowance Rates.” Working Cohen, Marc A. and Anne Tumlinson. 1997. “Under- Paper 2011-18. Chestnut Hill, MA: Center for standing the State Variation in Medicare Home Retirement Research at Boston College. Health Care.” Medical Care 35(4): 618–633. Gruber, Jonathan and Jeffrey Kubik. 2002. “Health Insurance Coverage and the Disability Insur- ance Application Decision.” Working Paper 9148. Cambridge, MA: National Bureau of Economic Research. Herring, Bradley and Mark Pauly. 2006. “The Effect of State Community Rating Regulations on Premi- ums and Coverage in the Individual Health Insur- ance Market.” Working Paper 12504. Cambridge, MA: National Bureau of Economic Research. Iyengar, Radha and Giovanni Mastrobuoni. 2008. “The Political Economy of the Disability Insur- ance. Theory and Evidence of Gubernatorial Learning from Social Security Administration Monitoring.” Working Paper Number 70. Moncali- eri, Italy: Collegio Carlo Alberto. Lindner, Stephan. 2011. “How Does Unemployment Insurance Affect the Decision to Apply for Social Security Disability Insurance.” Dissertation. Ann Arbor, MI: University of Michigan. APPENDIX 10 Center for Retirement Research Table A1. Descriptive Statistics Mean Standard deviation Minimum Maximum Dependent Variable (Percent of Working-Age Population) Total SSDI application rate 0.83 0.24 0.06 1.65 Health, Demographic, and Employment Variables Health Fair/poor health 0.15 0.03 0.08 0.25 Ever smoke 100+ cigarettes 0.47 0.05 0.25 0.61 Overweight or obese (BMI) 0.59 0.06 0.42 0.71 Age Profile Age under 18 0.26 0.03 0.19 0.37 Age 18-25 0.11 0.01 0.07 0.16 Age 25-50 (omitted) 0.35 0.02 0.29 0.44 Age 50+ 0.28 0.04 0.14 0.38 Education Profile Less than high school 0.15 0.05 0.05 0.33 High school degree (omitted) 0.34 0.05 0.20 0.48 Some college 0.42 0.06 0.23 0.57 Post-graduate 0.09 0.03 0.03 0.28 Other Demographics White, non-Hispanic 0.76 0.16 0.16 0.99 Male 0.49 0.01 0.46 0.52 Married 0.55 0.05 0.27 0.65 Poor 0.12 0.04 0.05 0.26 Occupation Service occupation 0.43 0.03 0.33 0.53 Blue-collar occupation 0.25 0.04 0.08 0.38 Other occupations (omitted) 0.32 0.05 0.21 0.58 Industry Agriculture and physical industries 0.29 0.05 0.11 0.42 Professional industries (omitted) 0.71 0.05 0.58 0.89 Labor Force Unemployment rate 0.05 0.02 0.02 0.13 Labor force participation rate 0.67 0.04 0.55 0.76 State Policy Variables Length of UI benefits (weeks) 31.66 9.27 26.00 66.33 UI benefits/average wage 0.37 0.06 0.20 0.55 Strict health regulation 0.13 0.33 0.00 1.00 Medicaid buy-in 0.37 0.48 0.00 1.00 State-mandated employer TDI 0.10 0.30 0.00 1.00 State Politics Variables Republican governor 0.54 0.50 0.00 1.00 Governor at term limit 0.29 0.45 0.00 1.00 Incumbent governor 0.39 0.49 0.00 1.00 Source: Authors’ calculations. Issue in Brief 11 Table A2. Regression Results for SSDI Applications, 1993-2009 Health, Demographic, and Employment Variables Coefficient Standard error Fair/poor health 2.087 *** (0.580) Ever smoke 100+ cigarettes 0.267 (0.210) Overweight or obese (BMI) 0.034 (0.370) Age under 18 -0.247 (0.480) Age 18-25 -0.649 (0.510) Age 50+ 0.234 (0.430) Less than high school -0.153 (0.400) Some college -0.313 (0.370) Post-graduate -0.664 (0.550) White, non-Hispanic 0.200 * (0.110) Male -1.785 ** (0.730) Married -0.445 (0.330) Poor 0.799 ** (0.320) Service occupation -0.422 (0.380) Blue-collar occupation 0.467 (0.510) Agriculture and physical industries 0.448 (0.400) Unemployment rate 1.087 (0.860) Labor force participation rate -1.393 *** (0.470) State Policy Variables Length of UI benefits -0.003 (0.000) UI benefits/average wage 0.081 (0.210) Strict health regulation -0.003 (0.030) Medicaid buy-in 0.008 (0.020) State-mandated employer TDI -0.117 *** (0.030) State Politics Variables Republican governor -0.026 * (0.010) Governor at term limit 0.029 (0.020) Incumbent governor -0.020 (0.010) Constant 2.454 *** (0.650) Observations 862 R-squared 0.796 Note: * significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent. Robust standard errors clustered by state are in parentheses. Also included are a set of year dummies (excluding 1993). Source: Authors’ calculations. RETIREMENT RESEARCH About the Center Affiliated Institutions The Center for Retirement Research at Boston The Brookings Institution College was established in 1998 through a grant Massachusetts Institute of Technology from the Social Security Administration. The Syracuse University Center’s mission is to produce first-class research Urban Institute and educational tools and forge a strong link between the academic community and decision-makers in the public and private sectors around an issue of Contact Information Center for Retirement Research critical importance to the nation’s future. To achieve Boston College this mission, the Center sponsors a wide variety of Hovey House research projects, transmits new findings to a broad 140 Commonwealth Avenue audience, trains new scholars, and broadens access to Chestnut Hill, MA 02467-3808 valuable data sources. Since its inception, the Center Phone: (617) 552-1762 has established a reputation as an authoritative source Fax: (617) 552-0191 of information on all major aspects of the retirement E-mail: crr@bc.edu income debate. Website: http://crr.bc.edu © 2012, by Trustees of Boston College, Center for Retire- The research reported herein was performed pursuant to ment Research. All rights reserved. Short sections of text, a grant from the U.S. Social Security Administration (SSA) not to exceed two paragraphs, may be quoted without ex- funded as part of the Retirement Research Consortium. The plicit permission provided that the authors are identified and opinions and conclusions expressed are solely those of the full credit, including copyright notice, is given to Trustees of authors and do not represent the opinions or policy of SSA, Boston College, Center for Retirement Research. any agency of the federal government, or the Center for Retirement Research at Boston College.