Lecture Notes Health Insurance I. The demand for Health Insurance Definitions: Deductible: when the patient pays all the price for a certain range Coinsurance: the insurer pays only part of the price, the patient pays the rest Limits: coverage up to a maximum amount Indemnity Insurance: reimbursement to the patient for medical costs (often fixed price per day in hospital) Service insurance: reimbursement to the provider for medical costs Impact of these on demand? An economic theory of demand for health insurance Why do individuals choose to buy insurance and how much? Budget constraint and preferences Expected utility analysis Expected utility analysis Risk aversion Suppose we have the following situation: 1. an individual has $50,000 in money 2. there is a 10% probability that the individual will become ill and have to pay $25,000 for treatment. => 1 = “good” state 2 = “bad” state Let Mg= money income in good state Mb= money income in bad state Suppose you can insure against the loss For example: suppose you can buy $10 of insurance coverage for $1 and that you fully insure against the loss. 10% change of => Mb = $50,000 – $25,000 + $25,000 – $2,500 Mb= $47,500 Mg= 50,000-2,500 = $47,500 Regardless of which state of nature occurs Let Y = premium cost/dollar of coverage K= dollars of coverage => in general.. 10% change of getting $25,000 + K – YK 90% chance of getting 50,000 – YK Now, contingent consumption N states of nature and consumption is contingent upon which state of nature you’re in. If states are just different consumption bundles => consumer theory can handle it. Mg A 50,000 B 50,000-yk $25,000 A= Endowment B= Full 25,000 + k -yk Mb How do you attain points where Mb > $50,000 –yk By over-insuring In essence by selling insurance– if such a choice is possible which it may not be Slope = ∆Mg /∆Mb = -yk/k-yk = -y/1-y Now look at utility function and indifference curves to talk about how individuals make choices. But 1st, how do probabilities enter info utility? They should, shouldn’t they? If Pg = .9 Pb =.1 (should get a different choice than if Pg= 1 and Pb = 0 Suppose m1, m2, m3 = income in states 1 and 2 P1, P2, P3 = probability in states 1, 2, 3 2 definitions: Expected value = P1*M1, + P2*M2 + P3*M3. In our example EV = (.9) (50,000) + (.1) (25,000) = 47,500 Explain Expected utility = P1*u(M1) + P2*u(M2) + P3*u(M3) +…. Expected utility hypothesis: you choose that option with the highest expected utility = weight of ability in difference possible states of nature. Now when does an individual choose to insure? Assume that the premium is actuarially fair (book calls pure premium) i.e. reflects the true probabilities so that in our example must pay ten cents the dollar => premium = $2,500 = EV 1. Risk Aversion u (m) $25,000 47.5 50 Now look at two possibilities Don’t buy insurance => Euni = .9 u(50K) + .1 u(25K) Buy insurance => Eui = u(97,500) An individual is risk averse for when EU (ins) > EU (no ins.) Or u (EU (g)) 3 possibilities 1. EUNI < EUI => risk averse 2. EU(NI) = EUI => risk neutral 3. EUNI > EUI => risk lover So depends upon the individuals shape (preferences) of utility of money curve U U (m) Risk averse U Risk lover U (m) 25 47.5 50 u U (m) 25 50 m Have shown two things that matter in deciding whether to buy insurance 1. Attitudes toward risk 2. The insurance premium Risk averse individuals always buy insurance when the premium is actuarially fair as it was in our example. Now, just look at risk averse individuals And look at the price or premium Even with competition, insurance firms can ____ charge an actuarially fair or pure premium Why? Suppose 10,000 individuals--- all the same with the same insurance Each pay $2,500 in premiums for a total of $25,000,000 10% of these individuals will incur losses of 25,000. The company must pay out (25,000)(1,000) = $25,000,000 But it costs more money to provide insurance (i.e. transaction costs of gathering premiums, paying for losses, etc. Even if profit = 0 premium > pure premium Q: Will a risk averse individual still insure? A: Perhaps U (m) U3 M1 M4 M3 M2 1st look at EV of the gamble = M3 => EU = U3 and willing to pay (M2-M4) at most to insure the distance M3 – M4 = the additional amount willing to pay above the pure premium if prem >M2 – M4 => don’t insure Implications of this Analysis 1. as the probability of the loss gets larger => M3-M4 gets smaller => less likely to buy insurance i.e. if you are sure to pay for the expense => not willing to buy insurance. Why? 2. As the probability of the loss gets smaller => M3-M4 gets smaller => less likely to buy insurance i.e. as you become more sure that loss will not occur less likely to buy insurance. Why? 3. As the magnitude of the loss decreases less likely to buy insurance because M4-M3 decreases. 4. As an individual becomes more risk averse=> more likely to buy insurance. 5. As the price of insurance increases => buy insurance for fewer events (less insurance) where price = amount willing to pay above pure premium. P A A P1 P2 0 1 line = amount person willing to pay above pure premium AA = price of insurance (above fair premium (pure)) Rises assuming costs increases as the # of claims increases due to rising transaction costs Individual only buys for P1 < Prob. < P2 If price increases => this interval gets smaller 6. The starting income of the individual At high income levels => MU low so less willing to pay above fair premium At low income levels => MU is high again because the distance between actual and EU is less This is wrong, at least the part about income levels affecting the distance. It still may be true that lower income people are less likely to buy insurance but this is because of budget constraints not the distances between the curves. Now look at the evidence: Tables 6.1 and 6.2 We see 1. if prob. is low => use is low 2. if prob. is high => use is low 3. if magnitude is high => use is high 4. if magnitude is low => use is low => model predicts relatively well The above assumed that D for M.C. perfectly inelastic once an illness occurs. Suppose its not. Moral Hazard: the tendency for insurance to affect the individual’s behavior. i.e. the individual can affect the size of the loss under insurance. Examples: 1. fire insurance => less likely to install fire alarms, smoke detectors 2. Car insurance => may drive faster 3. Health insurance => individuals may invest in less preventative care. Why? Preventative care is not paid for but other care is. Other examples depends upon how the insurance is set up Look at 2 impacts of the moral hazard using Demand Analysis P P* MC = S D Q* Q2 Q Full Insurance Coverage: P=0 to consumer and buys Q = Q2 > Q* This is inefficient since time cost is MC = P* at Q2 MB = 0 => MC >MB With no insurance, individuals consume Q = Q* with P=P* Is this behavior rational? Yes, individual is equating MB with MC = 0 Given that Q increases, what happens to the premium? Clearly, it must rise as well. Both because Q increases and because P increase if S is upward sloping. Suppose P* = 1,000 Q* = 10 P2 S P* P2+ 2,000 D Q2 = 20 Probability of illness = .2 Q* Q2 Assume moral hazard does not cause this to change With no moral hazard and no inefficient… Pure Premium = (.2) (1,000)(10) + (.8) (0) = $2,000 With moral hazard: pure premium = (.2)(2,000)(20) + (.8)(0) = $8,000 Premium rises to pay additional costs Q: Why don’t individually obserce? Increase insurance premium and stop increase QD? A: 1st, individuals make choices on the margin. The effect of insurance is to decrease the MC to the individual 2nd: need to understand the concept that insurance groups people together => by your decrease in QD you get very little benefit. Implications of the Moral Hazard 1. QD increases with insurance ( P increases as well) 2. Premium rises => fewer people insure Recall P A A P1 P2 3 methods to decrease moral hazard problem 1. Deductibles Assume MC constant for simplicity Let Q1 = deductible amount of services Actually, deductibles are usually in dollar amounts P P* S=MC A C Q* Q1 B Q2 Look at 2 situations: 1st: Q1 < Q* => will always buy the insurance at the pure premium. Why? 2nd: Q1 > Q* => either don’t buy insurance an consume at Q* or do buy insurance and consume at Q2 How do you decide? If you do buy… Pay P* x Q1 => Extra cost = (P*)(Q1-Q*) = area a + area c Extra benefit= Area under from Q* to Qz = Area c + Area b =>But only if extra benefits > Extra costs OR if a + c < c + b or B > A => Deductibles do not reduce the amount of Q purchased if have insurance…just reduces the # of people who buy insurance. 2nd: Coinsurance P P* S Pc Q* Q1 Q2 Pure premium: (P* -Pc)(Q1)(.2) < (P*)(Q2)(.2) Let Pc = coinsurance price => even with insurance must pay some of the price 1- Moral hazard problem is less 2-pure premium is lower with coinsurance => more people buy insurance 3rd: Prepaid plans like HMOs and PPOs focus on Drs and patient incentives not just patient thru coinsurane or deductibles. Adverse Selection: Consider 2 groups of people 1st group: prob. of illness =.8 2nd group: prob. of illness =.2 Suppose that the insurance company cannot distinguish between individuals in the 2 groups. Results? Assume equal number of individuals in both groups=> company observes a group who prob. of illness =.5 and bases its premium upon that. Let Mg = 10,000 MB = 2,000 u u (m) 2,000 3,600 6,000 8400 10,000 m Pure premium for Group 1 (high risk) = (.8)(8,000) =6,400 M = 10,000 -6,400 = 3,600 Pure premium for Group 2 (low risk) = (.2)(8,000) =1,600 M= 10,000-1600 = 8,400 Both would be willing to buy at average premium of $4,000 (m= $6,000) In our graph, Group 2 does not buy but Group 1 will always buy. Why? Group 2 may buy dependent upon several factors but most important is how different the risk levels for the 2 groups. Conclusions: Adverse selection 1. causes fewer low risk individuals to buy insurance and more high risk individuals to buy 2. Premium must rise if this is true, more low risk individuals drop out Controls? 1. experience ratings but perfect experience rations = no insurance. 2. exclusions for pre-existing conditions 3. decrease premium the longer insured 4. unwillingness to pay deductible and coinsurance may signal risk status Conclusions for the Chapter 1. forced coverage for all expenses is inefficient. Both high and low prob. events should likely not be covered. Why? => 100% coverage not optimal 2. moral hazard and adverse selection problems: % of ind. ded Co ins. 3. Public Policy: Major medical Size of exp. National health insurance? Coerced coverage for all individuals discrimination Other issues 1. Differential Health Insurances Suppose Health insurance reimburses hospital expenditures but not physician services Drs D* Mc = mc* D1 Hosp. H* H1 If decrease P of H => substitutes hospitals for Drs and inefficient since original iso-cost represented true costs. This is a service policy => results in overuse of those services which are reimbursed. An indemmity policy keeps the relative prices of the two goods the same since it reimburses for all medical expenditures This does cause D more MC to increase but does not change relative prices => no technological inefficiency Dr MC = MC* MC= MC1 H Note: figure 6.9 indicates allocative efficiency but this is incorrect Service benefit insurance creates more problems. i.e. inefficiency while indemmity insurance does not. 3 Problems Increased use of insured services Point where MP = 0 Increase demand for quality which is inefficient Tax Advantages Health insurance as a fringe benefit is not taxed Look at the individual who has two choices 1. Get a $300/month raise (BL2) 2. Get health insurance benefits (BL3) worth $300/month M + 300/Pc M/Pe M/Ph M + 300/Ph Health insurance Q: Why ever choose (2)? Since it cuts off part of BL? A: Tax benefits– suppose $300 is taxed but health insurance is not +> for 1 actually face BL4 => Plan 2 Makes everyone better off but does cause inefficiencies since forces some individuals to use more health insurance…then optimal Note: there is one type of ___ that may not be better off— the individual would choose no health care and depends on tax rate if ind. A would be better off The Market for Health Insurance Public Policy: 2 Questions 1. is intervention justified? 2. what type of intervention? Efficiency in 2 senses Supply Side: 1st- firm technological efficiency (use resources to min. cost of production) 2nd- Industry: does each firm produce at min point on LRAC? [suppose not any reason why this might be okay?] Demand Side Allocative efficiency MB=MC? Note: will basically take the same approach for all the other markets as well. A) Demand Market: Recall that market demand is determined by: 1. price of insurance 2. prob. of loss 3. magnitude of loss Income of the consumer Risk aversion Price elasticity ~ -1 => increase P of 10%, decrease QD by 10% Firm Demand Look at 3 different types of insurance 1. Blue Cross/Blue Shield= non profit 2. other commercial plans= profit 3. Independent plans= prepaid plans (HMOs); self insurance; service contracts Look at the changes embodied in table 11-2, p. 237 Trends: 1. increas in % of Pop. covered but slight especially in later years 2. decrease in BC/BS and big increase in Independent =>market demand is relatively inelastic but firm demand is elastic due to substitutes and competition Differences in 1. type of benefit 2. price 3. extent of coverage (Coinsurance, Deductibles) 4. reimbursement 5. reputation Predictions: 1. price will vary as the product varies 2. the product will change over time as pref. change (or as costs change) Now look at efficiency Is there an information argument that consumers find buying insurance inefficient since costly to gain information about competing co’s? Probably not. 1. large benefit item=> pays individuals to gain info 2. insurance often bought by groups and cost/person of gaining information is less. => information probably not a problem (note table 8-2 suggests it is for individual policies) Now look at Benefit/Premium Ratio Benefit = average benefit paid for by group Premium= price of insurance for that group If B/P ratio = 1 => premium = price Premium: If B/P ratio < 1 => price > pure premium as B/P ratio decreases, price increases If industry competitive expect to see B/P ratio close to 1 If monopoly=> B/P ratio would be low Look at table 8-2 to see how this has worked Note: book concludes that a fair amount of competition exists in the health insurance market, especially in the later years. Community Rating Why don’t we just put everyone into the same basket, charge them the same premium and get the same benefit? = community rating This is what Blue Cross did Problems: Suppose we have 2 large goals 1. efficiency 2. redistribution so low income individuals can afford medical care Look at how community rating affects both of these goals Assume 2 groups: High risk and low risk What you are trying to do is cross subsidize the high risk group. But 3 problems: 1) Inefficiency: low risk will be paying too high a price => may choose to self-insure even though for cost they should not. 2) Is the high risk group the one that we want to subsidize? Blue Cross subsidized the old but are they low income? Evidence suggests that Blue Cross actually subsidized the middle to high income. 3) Is community rating the efficient method of subsidizing? No, because it distorts choices by others => just use direct subsidies to achieve the goal. Competition ensured the demise of community rating. Low risk groups would leave the Blue Cross system with more options and this is what happened. The uninsured Look at table 11.3 / 11.4 (p. 241-42) Why do people choose no insurance? What does our theory tell us? P increase or decrease (prob.) Loading costs Lack of competition Working uninsured Book discusses 3 major reasons 1. see figure 11.4 (p. 242)—Basically firm has limited exp. Rating => must pay i1 not i0 => can’t compete 2. Pre-existing conditions may keep out certain industries with high % of such people—Book discusses beauty shop workers (temporary, young, etc.) 3. Attitudes Solutions—mandated coverage? Separate insurance from work Conclusion: D relatively competitive especially in recent years => allocatively efficient 2 points to support this: price is close to pure premium and demise of community rating is probably a result of increased competition in the market. B) Supply: look at 2 issues n determining the technological efficiency of production of health insurance 1. economies of scale = right # of firms in industry 2. each firm produces at min. cost Note: in normal model, competition ensures these 2 things but may have (1) information problems and (2) non-profit firms like BC BS. (1) Economies of Scale: book notes that there are many firms (> 1,000) in the insurance industry Empirical evidence seems to suggest that costs/claim decreases as the insurance firm gets larger. This appears to be true for commercial firms and BC BS. Problems How do you measure costs? General problems with all these quality and type of service varies => may get bias. The type of policy matters as well For example: group v. individual policies. 2nd is likely to be more costly to administer => may get additional bias. (2) Internal efficiency Theoretical Small information problems => Competition and profitmax will result in internal efficiency Currently doesn’t exist in a large sector especially w.r.t. BC & BS for 2 reasons: 1. BC & BS (BC est. by hospitals directly) have some monopoly power (competitive advantages) due to: BC & BS non-profit => favorable tax treatment but premium increases are regulated. [note: lost federal tax exempt status in 1986] Blues do not compete with each other => legal collusive arrangement between them. BC (hospital portion) receives a discount on hospital charges that most commercial insures do not. Why? One possibility is that hospitals are trying to increase utilization of their expense services. BC provides more comprehensive coverage than most hospital plans. => monopoly power for BC & BS Why don’t they just drive other, less competitive firms out of the industry? Because they are not profit-max. They use their competitive advantage to benefit others (by increased costs of production => inefficient) Possibilities: Consumers - not much support for this Hospitals - some support for this Physicians - fair turnout of empirical support for this Conclusion 1. Economies of scale exist 2. BS/BC may be internally tech. inefficient 3. However, increased competition in the past decade, especially by HMOs, etc. has decreased the ability of BC & BS to be inefficient => prospect for the future looks good. Market Competition in Health Care Basically want to look at 2 issues Why did competition evolve now and not before? What is the nature of the new competition? I. Why did competition evolve? A. Impetus from several sectors of the market for increased competition (+) Federal initiatives fueled by concern with rising expenditures on Medicare => implemented several plans Increased supply of physicians by: Subsidizing construction of new medical schools Subsidizing medical education for physicians and all health professions => Increase competition among physicians by increasing supply. HMO acts decreased expenditures by stimulating HMOs 1973 HMO Act Employers with more than 25 employees had to offer HMO option in area Federally qualified HMOs exempt from restrictive state practices 1979 amendment to CON legislation Loosened restrictions on building hospitals => HMOs found it easier to build their own hospitals if desired. => increased comp. especially when hospitals began to have excess capacity Medicaid Changes- basically eliminated the patients right of provider choice => states could negotiate with “efficient” providers New hospital reimbursement: DRGs talked about before. => decrease occupancy rates in hospitals, etc. Note: by this time, comp. already taking effect. (2) private sector Basically business wanted to decrease costs of health ins. Benefit programs for a number of reasons: recession, foreign competition, etc. Solutions: self-insurance, deductibles and coinsurance, pressure on insurers for efficiency, insurance coverage for low cost substitutes for hospital care. Impact: Business concerns translated into insurance concerns. Why? Hospital utilization decrease => excess capacity in hospitals developed [occupancy rate decreases] => hospitals became more willing to participate in alternate delivery systems Note: the same kind of things had happened before but had not resulted in increased comp. Why? Anticompetitive practices by physicians. => of even more importance (3) Application of anti-trust laws to health sector Previously not applied to health sector=> before when above conditions held competition by such things as: Denying hospital privileges to participating physicians Denying licences Limit advertising Why? Service industry exemption => decisions which changed this recently. Goldberg vs. Virginia State Bar: price fixing not legal for service industries 1978 Supreme Court denied the use of anti-competitive behavior by engineers => service industry exemptions lifted or at least decreased Advertising has 2 impacts on the market: 1st- decrease price of medical care…Why? 2nd- decrease variance of price of medical care…Why? 1st: increase information available to the public => increase elasticity of any individual suppliers D Curve MC Pna Pa Da MRA P is higher with no advertising and lower with advertising 2nd- recall our theory of how consumer’s search for best quality and best buy. Do it by spending resources. Get more variation in price for: large budget vs. small budget goods; when search costs are higher => advertising (as long as it contains info decrease search costs => get less variation in the price between difference producers) 2 impacts of Advertising 1. consumers have info on different products price and quality=> Decrease price in market because n increases/ 2. decreased consumers search costs => decreased variation in prices Frequency Pa Pna Price 3rd possible impact is to remove barriers to entry for competing firms includes: new Drs and new HMOs/PPOs. Empirical Results appear to support the theory (1) P decreases (2) elasticity increases => more substitutes. (3) variation decreases Also concern with negative effects of advertising i.e. not informative but induces individually to buy more low quality services But empirical work finds no reduction in quality of services with increased advertising (may even increase Quality) Spillover effects on non-advertisers => P decrease in markets where some advertise even though all do not P decreases even though quality does not => appears to be with reason for concern. B) Competition from alternate forms of health care providers. Like HMOS/PPOs Large increase in two market share. Serve app. 15% of the population in 1987. Average annual % increases = 19.6% From 1980-87=> large impact on the market Want to look at: advantages of HMOs, problems with HMOs, empirical evidence on HMO performance Advantages of HMOS Patients do not choose the provider at the time of illness- long term relation impact. Hospital efficiencyL for FFs (cost based) hospital reimbursement => Drs had no incentive to be concerned about hospital costs. HMO Advantages (cont’d) Now dr. does have an incentive either as owner of HMO or given incentives by HMO. Explain Cost minimization: HMO has an obvious incentive to min costs since fee does not depend on the amount of services provided Dr. productivity increases since HMOs use more complementary services like Dr. assistants and more of an incentive to lobby for damages in state practice acts Preventative care- since HMO as an insurer and has a long term relationship => cost effective preventative care will be provided by HMO. Explain No incentives to duplicate facilities unless cost effective Incentive to use cost effective generic drugs Incentives to innovate in care: technology, location, benefits,etc. HMO Problems: Biggest problem is with quality of care. To illustrate assume: patient has no info on qty of services provided =>no info on quality of treatment. (fee HMO = flat monthly fee, no other impact) HMOs incentive? Max profit or minimize private costs? Private costs: Wx X; where X = # of services provided Wx= C (cost of) => HMO minimizes costs by decrease in Quality (x) to zero. Just as in analysis of medical malpractice. Are there any problems with the analysis? Yes: consider the following reasons why this might be a problem. Assume that the HMO does not have the ability to set x=0. Why not? Repeat dealings or reputation: if HMO is in business for long term, then 2 effects: patients may have repeat dealings and leave with inadequate care, or patients may be able to gain info easily from old patients. Impact of both? Consumer Choice Assume 2 kinds of consumers well and ill informed. Well: consumers cause competition and increase quality even for ill informed as long as the HMO can not distinguish between the 2 types. Medical Malpractice System: if x < x* => sue and obtain judgment => gives incentive. Explain. Insurance Incentives: Suppose decreased care now (say preventative) increase in needed services later and services with large info. => HMO bears the cost of insuring against this => will provide such cost effective care. Explain. Spillover effects in For-Profit HMOs The book claims that for profit HMO Drs owned by Drs => each Dr has an incentive to monitor other Dr in organization since decrease in their profits due to lost reputation => control own quality due to profit incentive Problem: suppose large # of Drs in HMO => benefit to any Dr of monitoring is low (extra profit is split up as a large #) cost high => not likely to do it. This is a typical moral hazard problem Solution: HMO, who has better info, needs to control Dr’s actions since it has a large incentive. Explain. (3) Empirical Evidence Look at 3 issues: quality, expenditures/utilization, biased selection (a) quality: little empirical research (b) Expenditures/utilization Utilization decreases: length of stay, hospital admissions Expenditures: per day decrease, per admission decrease Table 12. 3 (p. 270) HMO = GHC Admission rates decrease, hospital days decrease, visits increase, preventative visits increase. Discuss But, per capital expenditures appear to increase. May be a short-term impact only. C) Bias due to Selection Problems 3 possible kinds of problems 1. Healthier patients with lower expenses may be more likely to join HMOs => get lower utilization, lower exp and higher quality only because of self-selection. 2. Sicker patients may be more likely to choose HMO since coverage is more comprehensive. 3. HMOs may locate in areas with higher exp, higher utilization, and lower quality since they can be more competitive. Empirical evidence suggests that self-selective bias is not a large problem. Some get + bias and other get – bias Some get significant bias in correction procedures