The response of small businesses to variation in the price of insurance: evidence from a randomized trial
Richard Kronick Louis Olsen Todd Gilmer Department of Family and Preventive Medicine, UCSD Supported by a grant from the California HealthCare Foundation
Overview
• • • • • • Policy Background Research Questions Theory Data and Methods Results Conclusions
Policy Background
• 80% of the uninsured are workers or the dependents of workers • Many uninsured workers work at a business that does not offer insurance • Small businesses are much less likely to offer insurance than large businesses • Small business owners are a politically attractive constituency
Policy Background (cont.)
• State governments have made efforts to increase the affordability and availability of coverage for small businesses
– Small group market reform – Purchasing cooperatives – Exemption from mandated benefits
• Association health plans and MEWAs
– Direct subsidies for small businesses
• MA, NY, MI • Senator Debbie Stebanow (D-MI) has proposed federal subsidies for small businesses that don’t provide insurance • Senator Kerry has proposed a refundable tax credit of up to 50% for small businesses (< 25 employees) that offer coverage
Research Questions
• If small business owners who do not currently offer insurance are offered coverage at a substantial discount, what percentage will purchase coverage? • How does the purchase rate vary as the price of coverage is varied? • What fraction of the uninsured are full-time workers at a small business that does not offer coverage?
Employer’s Decisions to Offer Insurance
• Firms act as agents for their employees • Firms offer insurance if the amount that employees are willing to pay for insurance with a given benefits package is greater than the price at which the employer can make insurance available • Assumes that wages will adjust to compensate for benefits costs
– Feldman, et. al,, 1997, Journal of Human Resources
Factors hypothesized to affect employer’s response to offer of subsidized coverage among employers not currently offering
• Price of coverage for employer and price for employee • Owner’s insurance status • Proportion of employees who are uninsured • Employee characteristics expected to affect demand for coverage (income, marital status, race/ethnicity, health status) • Business characteristics (difficulty in hiring and retaining workers; profitability; growth trajectory; number of employees)
Previous Work on the Effects of Price Reductions on Small Firms’ Behavior
• RWJ Health Care for the Uninsured Projects, 1987-1989
– Projects varied, but many provided 25%-50% premium subsidy – Helms, et. al. (1992) report low levels of market penetration
• New York State Health Insurance Pilot Program, 1988
– 50% state subsidy – Thorpe, et. al (1992) report low levels of penetration
Observational research on price elasticity
• A variety of work attempts to estimate price elasticity of demand: suggests elasticities of between -0.1 and -0.6 • Approximately 70% of small businesses offer insurance
– An elasticity of -0.1 implies that a 50% reduction in price will increase the percent offering to 73.5% – An elasticity of -0.6 implies that a 50% reduction in price will increase the percent offering to 91%
Data and Methods
• Conducted a randomized trial in San Diego from June-December, 2001 • Obtained a list of small businesses (3-50 employees) from Dun&Bradstreet • Randomly assigned employers to experimental treatments; treatments varied by how much employers and employees were required to pay for coverage • Subsidies provided by the California HealthCare Foundation
Description of the experiment
• Employers were offered a comprehensive HMO benefits package from Sharp Health Plan • Eligibility criteria:
– Between 2 and 50 FTEs – Not currently offering insurance – Have at least 2 uninsured workers with family income below 300% of FPL – Dependent children were eligible to enroll only if they were not eligible for Medi-Cal or Healthy Families
• Prices and subsidies were guaranteed for two years
Table 1: Employer Monthly Premium by Employer Experimental Treatment
Tier Level: Policy Type Tier One: Employee Only Tier Tw o: Employee & Spouse Tier Three: Employee & Child(ren) Tier Four: Employee & Family 1 $14 $22 $26 $32 Employer Experimental Treatment 2 3 4 $21 $28 $35 $33 $44 $56 $40 $53 $66 $48 $65 $81 5 $42 $67 $79 $97
Source: Sharp Health Plan - FOCUS Employer Fact Sheet.
Table 2: Employee Monthly Premium by Employee Experimental Treatment
Tier Level Tier One: Employee Only Policy Federal Poverty Level (year 2000) < 150% 150-175% 175-200% 200-225% 225-250% 250-300% < 150% 150-175% 175-200% 200-225% 225-250% 250-300% 1 $6 $11 $16 $23 $29 $43 $9 $15 $22 $31 $39 $58 Employee Experimental Treatment 2 3 4 $9 $12 $15 $16 $22 $27 $25 $33 $41 $35 $46 $56 $44 $56 $56 $56 $56 $56 $13 $17 $21 $22 $29 $36 $33 $44 $55 $46 $62 $77 $58 $78 $97 $87 $117 $146 5 $18 $33 $49 $56 $56 $56 $26 $44 $66 $93 $117 $148
Tier Tw o: Employee & Spouse Policy
Note: Shaded cells represent values that are capped to insure that sum of maximum employer and employee contributions do not exceed total premium value. Premiums f or Tiers Three and Four are not show n. Source: Sharp Health Plan - FOCUS Employee Fact Sheet.
Table 3: Average Employee Premium for an Employee-Only Policy, by Employee Experimental Treatment
1 Average Employee Premium f or an Employee-Only Policy Standard Deviation 25th Percentile of Firms 75th Percentile of Firms $16 $9 $8 $26 Employee Experimental Treatment 2 3 4 $24 $11 $16 $33 $32 $8 $27 $34 $36 $12 $30 $45 5 $37 $12 $28 $45 Total $30 $13 $20 $40
Note: Unw eighted N=134 firms. Source: UCSD Survey of Small Employers in San Diego, 2001.
Implementation of the experiment
• Mailed introductory letters to sampled employers • Followed up with phone calls to explain the offering, identify whether the business was eligible, and offered to schedule on-site visits with eligible employers
Data Collection
• At all businesses contacted, gathered information from owner (or knowledgeable respondent) to determine whether the business was eligible:
– Did the firm offer insurance? – How many FTEs at the firm? – Were there at least two uninsured workers?
Second round contacts
• Between 1 to 5 months after the first contact, attempted to re-contact refusals and partial refusals • Successfully gathered information on approximately 1/3 of the initial refusals and partial refusals • Using information on outcomes from the second round of contacts on converted refusals and converted partial refusals, we assigned outcomes to remaining refusals and partial refusals – assignment process was done separately by D&B reported firm size
Figure 1: Estimates of Business Eligibility for FOCUS
5,720 Letters Mailed
53% In Scope N = 3,037
15% Do Not Offer Insurance N = 855
6% Eligible for FOCUS N = 345
Out of Scope / Unable to Contact N = 2,683
Offers Insurance N = 2,182
< 2 Uninsured Workers N = 451
< 2 Income Eligible Workers N = 59
Source: UCSD Surv ey of Small Employ ers in San Diego, 2001. Note: Businesses are 'in scope' if they employ between 2 and 50 f ull-time workers.
Results
Table 4: Percent of Businesses Purchasing FOCUS, by Employer Premium for an Employee-Only Policy
Purchased FOCUS Yes No Total Employer Premium for an Employee-Only Policy $14 $21 $28 $35 $42 33% 18% 22% 14% 21% 67% 82% 78% 86% 79% 100% 100% 100% 100% 100% Total 21% 79% 100%
Pearson r = -.118; p = 0.174. Note: Unw eighted N=199 f irms; w eighted N=345. Source: UCSD Survey of Small Employers in San Diego, 2001.
Percent of Businesses Purchasing FOCUS, by Employee Experimental Treatment
Purchased FOCUS Yes No Total 1 21% 79% 100% Employee Experimental Treatment 2 3 4 25% 22% 13% 75% 78% 87% 100% 100% 100% 5 22% 78% 100% Total 21% 79% 100%
Pearson r = .024; p = 0.653 Note: Unw eighted N=134; Weighted N=345 Source: UCSD Survey of Small Employers in San Diego, 2001.
Table 5: Percent of Businesses Purchasing FOCUS, by Average Employee Premium for an Employee-Only Policy
Purchased FOCUS Yes No Total Average Employee Premium for an Employee-Only Policy $6-$17 $17-$26 $26-$33 $33-$43 $43-$58 34% 28% 17% 9% 16% 66% 73% 83% 91% 84% 100% 100% 100% 100% 100% Total 21% 79% 100%
Pearson r = -.175; p = 0.043. Note: Unw eighted N=134; Weighted N=345. Source: UCSD Survey of Small Employers in San Diego, 2001.
Percent Purchasing FOCUS by Owner Eligibility Status
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Owner is eligible f or FOCUS Owner is not eligible f or FOCUS
n=58 Purchased FOCUS
n=288 Did Not Purchase FOCUS
Note: Unweight ed Ns: Eligible owners=37; ineligible=97. Table only includes businesses that were eligible to purchase FOCUS.
Table 6: Weighted Logistic Regression Results Predicting the Probability of Purchasing FOCUS
Parameter Estimate (1) (2) Variable Name Employer Premium Payment f or an Employee-Only Policy -0.024 (0.022) -0.029 (0.016)* 1.88 (0.47)*** -0.21 (0.68) 134 0.13 -0.027 (0.010)*** 1.88 (0.47)*** -0.18 (0.64) 134 0.13
Average Employee Premium Payment for an Employee-Only Policy
Total Premium Payment for an Employee-Only Policy
Ow ner is FOCUS-Eligible
Intercept
Unw eighted N Pseudo R2 * P < .1; ** P < .05; *** P < .01. Note: Robust standard error in parentheses. Weighted N = 345. Source: UCSD Survey of Small Employers in San Diego, 2001.
Figure 2: Predicted Probability of Purchasing Insurance, by Premium Cost and Owner Insurance Status
0.8
0.7
Probability of Purchase
0.6
0.5
0.4
0.3
0.2
0.1
0 $20 $30 $40 $50 $60 $70 $80 $90 $100
Monthly Premium for an Employee-Only Policy
All Small Businesses
Ow ner is Insured
Ow ner is Uninsured
Estimated Elasticity
• 73% of businesses in San Diego offer insurance • Market price for a policy like FOCUS was approximately $145 for a single employee • At $60 for a single employee, we estimate 20% of the remaining businesses would offer, bringing the total offering rate to 79% • A 57% reduction in price is expected to increase the offering rate by 7.8% -- an elasticity of -0.14
Figure 3: Distribution of San Diego Small Businesses Not Offering Insurance by Number of Uninsured Workers
90 80 70to 9 5 60 50 4 40 30
3 20 10 to 19
20 to 50
Zero Uninsured Workers
East West North
10 0 1st Qtr
2
2nd Qtr
3rd Qtr1 Uninsured Worker 4th Qtr
Source: UC SD survey of San Diego employers, 2001. Note: Unweighted N = 492 businesses. Weighted N= 10,636 businesses.
Can the results be generalized to other settings?
• Information costs -- not all eligible businesses were aware of the offer • As a result of income related premiums and requirement that worker be uninsured in order to be eligible, at some businesses, not all workers were eligible • If owner was insured, business was not eligible • Location on the border with Mexico may result in less interest in insurance than elsewhere
Limitations of the analysis
• We were required to estimate the number of businesses that were eligible to purchase
– Reasonable upper and lower bounds are that between 17% and 24% of eligible businesses purchased
• Because employee income and employee premiums are highly correlated, we are unable to separately estimate the effects of income and premium • Experiment was not double-blinded
Conclusions
• Small employers are not THE problem
– Including dependents, full-time workers at small firms that do not offer insurance account for 21% of the uninsured
Conclusions (cont.)
• Uninsured workers in small firms are concentrated in a very small number of firms: of the 40,000 small firms in San Diego, 3% of the firms account for over 50% of the uninsured workers in small businesses
Conclusions (cont.)
• When offered insurance at $60 per month for a single employee (a 57% subsidy), approximately 20% of eligible firms that do not offer insurance will start offering • At $20 per month, 40% purchase • Either employees do not value even very low priced insurance, or employers are not acting as agents for employees (and/or compensating wage differentials are more theoretical than real)
Summary
• The small employer market is a smaller target than is sometimes thought • Even very substantial subsidies have limited effect on employer decisions • Programs that subsidize insurance for small businesses are unlikely to have a substantial effect in reducing the number of uninsured
Figure 4. Distribution of the Uninsured in San Diego, by Employment Status of the Head of Household
Full-time Worker in a firm with < 51 employees that does not offer insurance 21%
Non-Workers 14%
Part-Time Workers 12%
Self-Employed 9%
Full-time Worker in a firm with < 51 employees that offers insurance 16%
Full-time Worker in a firm with 51+ employees that offers insurance 21%
Full-time Worker in a firm with 51+ employees that does not offer insurance 7%
Source: Em ploym ent status of the uninsured from 2001 C alifornia Health Interview Survey. Uninsured part-time workers, non-workers, and children are assigned to the insurance and employment status of the family head, using analysis of data from the m erged February and March 2001 C urrent Population Survey. Uninsured full-tim e workers are characterized by their own employm ent status.
Figure 1: Percent of San Diego Businesses Offering Insurance, by Firm Size
100% 90% 80%
87.4%
90 80 70
48.9% 60.2% 65.7%
80.8% 74.7% 73.5%
Percent Offering Insurance
70% 60% 50% 40% 30% 20% 10% 0%
60 50 40 30 20 10 0 East West North
2
3 1st Qtr
4 2nd Qtr
Number of Full-Time Employees
5 3rd Qtr
6 to 4th Qtr 10
11 to 50
Total
Source: UC SD survey of San Diego employers, 2001. Note: Unweighted N = 2,830.