The Effect of Rate Regulation on Price and Competition Sharon Tennyson Department of Policy Analysis and Management Cornell University Richard A. Derrig Opal Consulting, LLC CANE Meeting, Sturbridge, MA September 18, 2007

Impact of Rate Regulation AGENDA Regulation and Competition – The Rationale and the Reality A Little Theory and Practice Auto Insurance: Brookings/AEI Case Studies The Case of Massachusetts Implications for California

Why (now) Do States Regulate P-C Insurance Rates? Automobile insurance Homeowners insurance Workers compensation insurance Medical malpractice insurance Mandatory or socially desirable insurance Uninsured parties shift financial risk to others in society Some consumers face high prices Some markets experience price shocks

Rate Regulation as Redistribution Not equally practiced in all regulated states or lines  Some regulated states lightly regulate voluntary market rates  Residual market rates are regulated in all states and lines Implications of redistribution Price subsidies for some consumers To high risk consumers From low risk consumers To high risk consumers From insurance company capital To all consumers From insurance company capital

Efficient Redistribution? Not Likely: Not Sustainable The Problem: Competitive Market Forces Attempts to move prices a significant distance from competitively-determined prices will distort market functioning  Company responses to regulation  Consumer responses to regulation  Regulator responses to the responses

Efficiency Consequences: Supply Rate suppression distorts insurance supply  Decreased writing of voluntary coverage  Reduced innovation and quality of service  Reduced entry of firms  Increased exit of firms Rate uncertainty distorts insurance supply  All of the above effects  Price stickiness and market volatility

Efficiency Consequences: Demand Rate subsidies distort consumer behavior: Insurance demand  Increased demand from high risk consumers  Decreased demand from low risk consumers  May cause low risk consumers to forego insurance Safety incentives  Prices are less responsive to changes in losses/risk Claiming incentives  Prices are less responsive to changes in losses/risk

Implications Stringent regulation of insurance rates produces unintended effects:  Reduced competition  Higher cost inflation  Lower insurance availability  Greater market volatility Regulated outcomes may even be contrary to the regulatory objectives pursued

What is the Empirical Record? On average and over the long run, rate regulation has little effect on average loss ratios (Harrington, 2002)  Not necessary for market functioning  Creates uncertainty and costly compliance In states and markets where rate regulation is stringently applied, empirical studies find effects supportive of theory In states and markets where stringent rate regulation is dismantled, empirical studies find effects supportive of theory

Evidence from Massachusetts State regulation of private passenger auto insurance rates has created widely-recognized market problems  Exit of insurance providers, especially national firms  Small number of suppliers  Larger than normal residual market  Cost inefficiencies  Politicized ratemaking environment Less recognized problem: In the aggregate, regulation drives overall claims costs higher

Table 1 Major Regulatory Changes, Massachusetts Private Passenger Automobile Insurance YearRegulation 1971Nofault auto insurance effective 1975State rate-setting extended to all auto coverages 1977Competitive rate-setting allowed 1978State rate-setting reinstituted 1989Automobile Insurance Reform Law effective 1991Insurance Fraud Bureau began operation 1996Competitive Discounts and Deviations begin at -7.4% 2006Competitive Discounts and Deviations stabilize at -1.7% 2007Competitive rate-setting allowed 4/1/08

Table 2 Direction of Subsidies by Driver Class and Territory Compulsory Insurance Coverage 2004 Experienced Classes Inexperienced ClassesBusiness Classes Non-Boston Territories Average Premium $527.15$1,220.54$ Average Subsidy -$26.00$ $46.43 Cells Subsidized (%) 12.50%42.71%6.25% Boston Territories Average Premium $813.33$1,434.04$ Average Subsidy $253.77$520.09$32.30 Cells Subsidized (%) 64.65%72.73%36.36% Source: Authors calculations using data from Actuarial Notice 04-1, Automobile Insurers Bureau of Massachusetts, Compulsory coverages are Bodily Injury Liability (20/40), Personal Injury Protection (8,000), Property Damage Liability (5,000) and Uninsured Motorist (20/40)

Table 3 Massachusetts Private Passenger Automobile Historical Summary of Industry Discounts/Deviations YearAverage Discount Annual Change in Discount % %-1.8% %+0.0% %+2.7% %+1.0% %+2.5% %+0.8% %+0.3% %+0.2% %-0.1% %+0.1% %est'd+0.0% Source: Automobile Insurers Bureau of Massachusetts

Impact of Rate Regulation on Claims: First Cut: State-level data on average loss costs  50 states  (before and after Massachusetts effective subsidies) Hypothesis: Massachusetts’ loss costs will be higher than otherwise predicted during period of stringent regulation

Impact of Rate Regulation on Claims Second cut: Massachusetts town-level data on loss cost levels for 5 coverages  360 towns  Biennial data Hypothesis: Loss cost growth higher in subsidized towns than in other towns

State Data

Impact of Rate Regulation State by State Estimation DATA: NAIC Coverage Variations Demographics Regulation METHOD: Panel Data Regression Models Ln(Liab. Loss / W Car Years)

State Regression Estimates L st = ß 0 + ß 1 CSYears t + ß 2 MA s CSYears t + γ ’X st + ß 4 StateRegs st + ß 5 StateRegs st CSYears t + a s + T t + ε st CSYears = or Test ß 2 > 0 Control variables = traffic density, cost of hospital stay, per capita income, rate regulation, nofault, PIP limits, compulsory insurance minimums a s and T t are state and year fixed-effects Adjust S.E. for arbitrary heteroskedasticity and for arbitrary correlation within state

State Estimation Results Explanatory VariableCSYears CSYears MA x CS Years0.3781***0.3663*** CS Years Dummy ***0.0523*** Reg x CS Years0.1097*** Nofault x CS Years R-squared N1334

State Regression Estimates L st = ß 0 + ß 1 CSYears t + γ ’X st + ß 4 StateRegs st + ß 5 StateRegs st CSYears t + a s + T t + ε st Estimate identical model specification without MA data or MA interaction term Apply estimated coefficient vector to Massachusetts variable values, Obtain predicted value of Massachusetts loss costs for each year Compare Actual – Predicted value

Actual – Predicted Losses

Impact of Rate Regulation on Claims: Second Cut Panel of Massachusetts town-level data on loss cost indices for 5 coverages from AIB  360 towns  Biennial data  Pure premium index  Average class rating factor Hypothesis: Indices will grow faster in subsidized towns than in other towns

Town Data - BIL

Town Data - PDL

Impact of Rate Regulation Town Level Estimation DATA: 5 coverage Town Data AIB Territory Filings Rate Subsidies All Years Town Index – Bayesian Estimate/ Class Normalized (ACRF) METHOD: Regression Models on Town Relativity Growth

Town Regression Estimates % Δ PP it = ß 0 + ß 1 Subsidy it- τ + ß 2 PP it + ß 3 % Δ ACRF it + ß 4 % Δ Exposures it + ß 5 % Δ Density it + ß 6 Boston i + T t + ε it Use subsidy from year in which losses were generated Test ß 1 > 0 Adjust S.E. for arbitrary heteroskedasticity and for arbitrary correlation within state

Pure Premium Growth Estimates BIL PIP PDL COL COM Lag subsidy indicator *** *** ** *** *** Lag pure premium *** *** *** R-squared F-statistic2.95 *** *** 7.54 *** 7.51 *** 8.90 *** N1439 Standard errors appear below the coefficient estimates, and are adjusted to allow for arbitrary heteroskedasticity and for arbitrary correlation in errors across years within each town. *** indicates statistically significant at the 1% confidence level *

Impact of Rate Regulation California Auto Territories PROP 103 (1989) Driving History Emphasized for Relative Pricing; Mandatory and Optional Classes April 2006 Change: Territory must be less “important” than Mandatory Prop 103 Factors: Annual Mileage, Driving Safety Record, Years Licensed Effect yet to be seen but moves away from cost-based must introduce Mass-like subsides varying by insurer Rural to Urban, High Income to Lower Income, High Cost to Low Cost

Impact of Rate Regulation California Auto Territories April 2006 Change: Optional Territory Frequency and Territory Severity (and 14 others) must each be forced to be less “important” (less relativity weight) than Mandatory Prop 103 Factors: Annual Mileage, Driving Record, Years Licensed DOI Study Showed 3 Scenarios that implied: Increases for 53 of 58 Counties (High +37%,Low 1.3%) Decreases for 5 of 58 Counties (High -0.5%,Low -12%) LA Zips (High – 6%, Low -11%), Beverly Hills Zips (High -10%, Low -23%) Corr Prem to LossPP, (Current 0.75, New to 0.62)

Papers R.A. Derrig and S. Tennyson (2007), The Impact of Rate Regulation on Claims: Evidence from Massachusetts Automobile Insurance, Concurrent session VD #3 S. Tennyson (2007), Efficiency Consequences of Rate Regulation in Insurance Markets, Networks Financial Institute Policy Brief,

References (Excerpt) Automobile Insurers Bureau of Massachusetts, Actuarial Notice -2: Subsidies in the Rates, Boston, MA: AIB. Various years. Automobile Insurers Bureau of Massachusetts, 2006, AIB Recommendations for 2007 Private Passenger Automobile Insurance Territory Definiitons, MA DOI Docket R , May 15. Blackmon, B.G. Jr. and R. Zeckhauser, 1991, Mispriced Equity: Regulated Rates for Auto Insurance in Massachusetts, American Economic Review, 81: Burnes, N.S., 2007, Opinion, Findings, and Decision on the Operation of Competition in Private Passenger Motor Vehicle Insurance in 2008, Massachusetts Division of Insurance Docket No. R , July 16. Conger, R.F., 1988, The Construction of Automobile Rating Territories in Massachusetts, Proceedings of the Casualty Actuarial Society, 71: Part 1, Derrig, R.A.,1993, Price Regulation in US Automobile Insurance: A Case Study of Massachusetts Private Passenger Automobile Insurance , The Geneva Papers on Risk and Insurance, 18: Derrig, R.A., and Hilary N. Rowan, 2006, Written testimony of The California Farm Bureau, California Department of Insurance, Proposed Amendment of Title 10 of California Code of Regulations, Section – Optional Auto Insurance Rating Factors, CDI File #RH DuMouchel, W.H., 1983, The Massachusetts Automobile Insurance Classification Scheme, The Statistician, 32: Rottenberg, S., 1989, The Cost of Regulated Pricing: a Critical Analysis of Auto Insurance Premium Rate-Setting in Massachusetts, Boston: Pioneer Institute for Public Policy Research. Tennyson, S., 1997, The Impact of Rate Regulation on State Automobile Insurance Markets, Journal of Insurance Regulation 15: 502 ‑ 523.

Table 4 Summary Statistics Annual State-Level Data, 50 States All StatesMassachusetts VariableMeanS.D.MeanS.D. Losses per Insured Car Traffic Density Average Cost of Hospitalization Per Capita Income Rate Regulation Dummy Nofault Dummy Addon Dummy Person Minimum Limit (000) Property Minimum Limit (000) PIP Coverage Limit Addon Coverage Limit

Table 7 Summary Statistics Annual Town-Level Data, Massachusetts Territory Years VariableNMeanS.D. Pct change BIL PP Index Pct change PIP PP Index Pct change PDL PP Index Pct change COLL PP Index Pct change COMP PP Index Lag BIL PP Index Lag PIP PP Index Lag PDL PP Index Lag COLL PP Index Lag COMP PP Index Pct change BIL AC Index Pct change PIP AC Index Pct change PDL AC Index Pct change COLL AC Index Pct change COMP AC Index Lag BIL AC Index Lag PIP AC Index Lag PDL AC Index Lag COLL AC Index Lag COMP AC Index Subsidy Indicator (lagged) Traffic Density Pct change Exposures