Reference Dependent Decisions on Noncommunicable Diseases

Transcription

1 Noncommunicable Diseases Prevention, Treatment and Optimal Health DONG Yaohui Toulouse School of Economics SAEe 2017, Barcelona 1 / 16

2 and Its Consequences Noncommunicable Chronic Diseases () Not infectious, long duration, slow progression, non-curable (WHO, 2015) Moral Hazard Preview s 2 / 16

3 and Its Consequences Moral Hazard Preview s Noncommunicable Chronic Diseases () Not infectious, long duration, slow progression, non-curable (WHO, 2015) Cardiovascular diseases (12% of THE), cancers (7%), chronic respiratory diseases and diabetes 2 / 16

4 and Its Consequences Moral Hazard Preview s Noncommunicable Chronic Diseases () Not infectious, long duration, slow progression, non-curable (WHO, 2015) Cardiovascular diseases (12% of THE), cancers (7%), chronic respiratory diseases and diabetes 63% of death, Major challenge to global health and development (Choi et al., 2008) 2 / 16

5 and Its Consequences Moral Hazard Preview s Noncommunicable Chronic Diseases () Not infectious, long duration, slow progression, non-curable (WHO, 2015) Cardiovascular diseases (12% of THE), cancers (7%), chronic respiratory diseases and diabetes 63% of death, Major challenge to global health and development (Choi et al., 2008) Genetic dispositions and lifestyles Tobacco, alcohol, unhealthy diet, inactivity 2 / 16

6 and Its Consequences Moral Hazard Preview s Noncommunicable Chronic Diseases () Not infectious, long duration, slow progression, non-curable (WHO, 2015) Cardiovascular diseases (12% of THE), cancers (7%), chronic respiratory diseases and diabetes 63% of death, Major challenge to global health and development (Choi et al., 2008) Genetic dispositions and lifestyles Tobacco, alcohol, unhealthy diet, inactivity Prevention: Lower the prevalence of risk factors 75% Heart disease, stroke and Type II diabetes 40% Cancer 2 / 16

7 Prevention, Treatment and Moral Hazard Moral Hazard Preview s 1 π Healthy π Ill 3 / 16

8 Prevention, Treatment and Moral Hazard Moral Hazard Preview s (Primary) Prevention 1 π Healthy π Ill 3 / 16

9 Prevention, Treatment and Moral Hazard Moral Hazard Preview s (Primary) Prevention 1 π Healthy π Ill 3 / 16

10 Prevention, Treatment and Moral Hazard Moral Hazard Preview s (Primary) Prevention Preventive care: Vaccines, condoms 1 π Healthy π Ill 3 / 16

11 Prevention, Treatment and Moral Hazard Moral Hazard Preview s (Primary) Prevention Preventive care: Vaccines, condoms Preventive effort: Healthy lifestyle 1 π Healthy π Ill 3 / 16

12 Prevention, Treatment and Moral Hazard Moral Hazard Preview s 1 π Healthy (Primary) Prevention π Preventive care: Vaccines, condoms Ill Preventive effort: Healthy lifestyle Treatment 3 / 16

13 Prevention, Treatment and Moral Hazard Moral Hazard Preview s 1 π Healthy (Primary) Prevention π Preventive care: Vaccines, condoms Ill Preventive effort: Healthy lifestyle Treatment Health insurance causes 3 / 16

14 Prevention, Treatment and Moral Hazard Moral Hazard Preview s 1 π Healthy (Primary) Prevention π Preventive care: Vaccines, condoms Ill Preventive effort: Healthy lifestyle Treatment Health insurance causes Ex-ante moral hazard Stanciole (2008) 3 / 16

15 Prevention, Treatment and Moral Hazard Moral Hazard Preview s 1 π Healthy (Primary) Prevention π Preventive care: Vaccines, condoms Ill Preventive effort: Healthy lifestyle Treatment Health insurance causes Ex-ante moral hazard Stanciole (2008) Ex-post moral hazard Aron-Dine et al. (2013) 3 / 16

16 Research Questions and Results Preview Moral Hazard Preview s Prospect Theory (Kahneman & Tversky, 1979) Prevention Treatment Policy implications Prospect Theory v.s. Expected Utility Theory (EUT) in health Health decisions violates EUT s assumptions (e.g. Oliver, 2003) Prospect Theory performs better (e.g. Osch, 2004; Bleichrodt & Pinto, 2005) Results preview Medical bankruptcy ex-post Deliberate engagement in risky health behaviors Deductible insurance: financially unfeasible can encourage prevention Inverse relationship between income and unhealthy behaviors serves as a justification of redistribution 4 / 16

17 Dependence u( ) Moral Hazard Preview s 0 W VNM Utility Preference 5 / 16

18 Dependence Moral Hazard Preview s u( ) u(w 0 ) 0 W 0 W VNM Utility Preference 5 / 16

19 Dependence Moral Hazard Preview s u( ) u(w 0 ) v( ) 0 W 0 W VNM Utility Preference Gain/Loss Preference 5 / 16

20 Dependence Moral Hazard Preview s u( ) u(w 0 ) v( ) 0 W 0 W VNM Utility Preference Gain/Loss Preference 5 / 16

21 Dependence Moral Hazard Preview s u( ) u(w 0 ) v( ) 0 W 0 W 0 Gain/Loss VNM Utility Preference Preference 5 / 16

22 Dependence Moral Hazard Preview s u( ) u(w 0 ) v( ) 0 0 W 0 W 0 Gain/Loss VNM Utility Preference Preference 5 / 16

23 Timing Pay P Choose e π(e) Ill Pay full or part of M Background Ex-post Ex-ante 1 π(e) Healthy t = t 0 t = t 1 6 / 16

24 Setup Value Function for Non-Curable Diseases point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel, Weinfurt, & Schulman, 2005) Background Ex-post Ex-ante 7 / 16

25 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) 7 / 16

26 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases 7 / 16

27 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment 7 / 16

28 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover 7 / 16

29 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases 7 / 16

30 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival 7 / 16

31 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival Remain sick 7 / 16

32 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival Remain sick cause 7 / 16

33 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival Remain sick cause Financial loss 7 / 16

34 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival Remain sick cause Financial loss Permanent damage to health 7 / 16

35 Setup Value Function for Non-Curable Diseases Background Ex-post Ex-ante point: Status quo in each period Evolution of reference point across period Pre- and post-diagnosis life expectancies (Rasiel et al., 2005) V (h, w) Curable diseases Pay for treatment Recover Non-curable diseases Pay for survival Remain sick cause Financial loss Permanent damage to health Health dimension: Life years Wealth dimension: Monetary terms 7 / 16

36 Timing Pay P Choose e π(e) Ill Pay full or part of M Background Ex-post Ex-ante 1 π(e) Healthy t = t 0 t = t 1 8 / 16

37 Treatment: Assumptions w Background Ex-post Ex-ante 0 h 9 / 16

38 Treatment: Assumptions w Background Ex-post Ex-ante 0 h 9 / 16

39 Treatment: Assumptions w Background Ex-post Ex-ante 0 h h 9 / 16

40 Treatment: Assumptions w Background Ex-post Ex-ante 0 h h 9 / 16

41 Treatment: Assumptions Background Ex-post Ex-ante w Y 0 h h 9 / 16

42 Treatment: Assumptions Background Ex-post Ex-ante w Y M 0 h h 9 / 16

43 Treatment: Assumptions Background Ex-post Ex-ante w Y M H(M) H(M), H > 0, H < 0. 0 h h 9 / 16

44 Treatment: Assumptions Background Ex-post Ex-ante w Y M H(M) H(M), H > 0, H < 0. u H > 0, u lim H 0 u H u(h(m), M) M < 0. u = +, lim M 0 M < +, 2 u H 2 < 0, 2 u M 2 < 0. 0 h h 9 / 16

45 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y Background Ex-post Ex-ante 10 / 16

46 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y M(Y ) Background Ex-post Ex-ante du dm = u H H (M)+ u M = 0 = u H H (M) = u M M 0 M Y 10 / 16

47 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y M(Y ) Background Ex-post Ex-ante du dm = u H H (M)+ u M = 0 = u H H (M) = u M M BANKRUPT 0 M Y 10 / 16

48 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y M(Y ) Background Ex-post Ex-ante du dm = u H H (M)+ u M = 0 = u H H (M) = u M M ε M Y = 1 εm Y = 0 BANKRUPT 0 M Y 10 / 16

49 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y M(Y ) Background Ex-post Ex-ante du dm = u H H (M)+ u M = 0 = u H H (M) = u M Deductible D < Y M M ε M Y = 1 εm Y = 0 BANKRUPT max M u(h(m), D) Y 0 M 10 / 16

50 Treatment Demand for Medical Treatment Patient treatment problem: max u(h(m), M) (1) M Y M(Y ) Background Ex-post Ex-ante du dm = u H H (M)+ u M = 0 = u H H (M) = u M Deductible D < Y M M ε M Y = 1 εm Y = 0 BANKRUPT max u(h(m), D) M = M = + 0 M Y 10 / 16

51 Prevention: Framing w Background Ex-post Ex-ante 11 / 16 0 h

52 Prevention: Framing w Background Ex-post Ex-ante Y M Sick π(e) 11 / 16 0 h + H(M) h

53 Prevention: Framing w Y Healthy 1 π(e) Background Ex-post Ex-ante Y M Sick π(e) 0 h + H(M) h h 11 / 16

54 Prevention: Framing w Y Healthy 1 π(e) Background Ex-post Ex-ante Ref Y M Sick π(e) 0 h + H(M) h h 11 / 16

55 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Background Ex-post Ex-ante Ref Y M Sick π(e) 0 h + H(M) h h 11 / 16

56 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Background Ex-post Ex-ante e Ref Y M Sick π(e) 0 h + H(M) h h 11 / 16

57 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Gain and loss Background Ex-post Ex-ante e Ref Y M Sick π(e) 0 h + H(M) h h 11 / 16

58 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Gain and loss V (h, w) = γh + w Background Ex-post Ex-ante e Ref α(e) effort cost: Y M Sick π(e) 0 h + H(M) h h 11 / 16

59 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Gain and loss V (h, w) = γh + w Background Ex-post Ex-ante e Ref α(e) effort cost: (λ 1)(2π 1)K M π (e) = α (e) Y M Sick π(e) 0 h + H(M) h h 11 / 16

60 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Gain and loss V (h, w) = γh + w Background Ex-post Ex-ante Y M Sick π(e) e Ref α(e) effort cost: (λ 1)(2π 1)K M π (e) = α (e) Expected Utility Maximizer: K M π (e) = α (e) 0 h + H(M) h h 11 / 16

61 Prevention: Framing w Y Healthy 1 π(e) π (e) 0 Gain and loss V (h, w) = γh + w Background Ex-post Ex-ante Y M Sick π(e) e Ref α(e) effort cost: (λ 1)(2π 1)K M π (e) = α (e) Expected Utility Maximizer: K M π (e) = α (e) Loss aversion v.s. Framing 0 h + H(M) h h 11 / 16

62 Prevention Across Income Groups Background Ex-post Ex-ante Proposition 1 Y > M = the same level of prevention e Y = 0. 2 Y M e, Y 0 1 γh (M) 0 For the poor, if e Y 0 12 / 16

63 Prevention Across Income Groups Background Ex-post Ex-ante Proposition 1 Y > M = the same level of prevention e Y = 0. 2 Y M e, Y 0 1 γh (M) 0 For the poor, if e Y 0 Medical bankruptcy 12 / 16

64 Prevention Across Income Groups Background Ex-post Ex-ante Proposition 1 Y > M = the same level of prevention e Y = 0. 2 Y M e, Y 0 1 γh (M) 0 For the poor, if e Y 0 Medical bankruptcy Poorer health 12 / 16

65 Prevention Across Income Groups Background Ex-post Ex-ante Proposition 1 Y > M = the same level of prevention e Y = 0. 2 Y M e, Y 0 1 γh (M) 0 For the poor, if e Y 0 Medical bankruptcy Poorer health Less healthy lifestyles 12 / 16

66 Optimal Health Behavioral welfare analysis is challenging Assumptions A Benchmark 13 / 16

67 Optimal Health Behavioral welfare analysis is challenging Different utility functions ex-ante and ex-post Assumptions A Benchmark 13 / 16

68 Optimal Health Assumptions A Benchmark Behavioral welfare analysis is challenging Different utility functions ex-ante and ex-post Assumptions Utilitarian welfare function: Ex-post utility function 13 / 16

69 Optimal Health Assumptions A Benchmark Behavioral welfare analysis is challenging Different utility functions ex-ante and ex-post Assumptions Utilitarian welfare function: Ex-post utility function Wellbeing of the longevous: 0 13 / 16

70 Optimal Health Assumptions A Benchmark Behavioral welfare analysis is challenging Different utility functions ex-ante and ex-post Assumptions Utilitarian welfare function: Ex-post utility function Wellbeing of the longevous: 0 Welfare without insurance W = Ȳ Ȳ Ȳ (1 π(e(y ))) 0dF (Y ) + π(e(y ))u(h(m I (Y )), ζ(y ))df (Y ) Ȳ 13 / 16

71 Purely Redistributive Assumptions A Benchmark premium in the beginning of t 0 point changes P (Y ) = Y M Assume IE[Y ] M, transfer in t 0 14 / 16

72 Purely Redistributive Assumptions A Benchmark premium in the beginning of t 0 point changes P (Y ) = Y M Assume IE[Y ] M, transfer in t 0 W I = (1 π(e 0 )) 0+π(e 0 ) Ȳ Ȳ u(h(m ), M )df (Y ) = π(e 0 )u(h(m ), M ) 14 / 16

73 Purely Redistributive Assumptions A Benchmark premium in the beginning of t 0 point changes P (Y ) = Y M Assume IE[Y ] M, transfer in t 0 W I = (1 π(e 0 )) 0+π(e 0 ) Ȳ Ȳ u(h(m ), M )df (Y ) = π(e 0 )u(h(m ), M ) Proposition 1 γh (M) 0 = W I W. 14 / 16

74 Ex-ante Moral Hazard Assumptions A Benchmark Effects on Prevention Fixed Indemnity Coinsurance +/ Coinsurance could encourage prevention 15 / 16

75 Ex-ante Moral Hazard Assumptions A Benchmark Effects on Prevention Fixed Indemnity Coinsurance +/ Coinsurance could encourage prevention Even when preventive efforts are unobservable 15 / 16

76 Thank You DONG Yaohui Toulouse School of Economics 21, Allée de Brienne Toulouse, France +33 (0) / 16

77 dependent preference Violation of EUT Inconsistent predictions by EUT Appendix Violation of EUT Framing s 1 / 9

78 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 1 / 9

79 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 1 / 9

80 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) 1 / 9

81 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) Framing: point plays a key role in decision making 1 / 9

82 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) Framing: point plays a key role in decision making Loss aversion: Greater aversion to losses than appreciation of gains 1 / 9

83 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) Framing: point plays a key role in decision making Loss aversion: Greater aversion to losses than appreciation of gains Diminishing sensitivity 1 / 9

84 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) Framing: point plays a key role in decision making Loss aversion: Greater aversion to losses than appreciation of gains Diminishing sensitivity Best available description of how people evaluate risk (Barberis, 2013) 1 / 9

85 dependent preference Violation of EUT Appendix Violation of EUT Framing s Inconsistent predictions by EUT Treatment choices Example 1 Health state valuations Example 2 dependence (Kahneman & Tversky, 1979) Framing: point plays a key role in decision making Loss aversion: Greater aversion to losses than appreciation of gains Diminishing sensitivity Best available description of how people evaluate risk (Barberis, 2013) Evidence in health decision making Gambles with years of life (Verhoef et al., 1994) Health insurance decisions (Marquis & Holmer, 1996) 1 / 9

86 Violation of EUT Treatment of Lung Cancer (McNeil & Pauker, 1982) Appendix Violation of EUT Framing s Choice reversals in treatment for lung cancer Back Treatment Die in treatment LE on success LE avg Surgery (58%) 10% 6.8 years 6.1 Radiation(42%) Table: Mortality Framed Choices Treatment Survive in treatment LE on success LE avg Surgery(75%) 90% 6.8 years 6.1 Radiation(25%) 100% Table: Survival Framed Choices 2 / 9

87 Violation of EUT Health States Valuation Appendix Violation of EUT Framing s Cohort Healthy Asthma at 40 Death at 40 Sum A 80 x B x u(asthma) = = u(asthma) = x 15 Table: Veil of ignorance approach (VEI) Dose Healthy Asthma Death Sum A p 0 1-p 1 B p 1 + (1 p) 0 = 1 u(asthma) = u(asthma) = p 3 / 9 Table: Standard Gamble Approach (SG)

88 Violations of EUT Pinto-Prades and Abellán-Perpiñán (2005) Appendix Violation of EUT Framing s Health State VEI SG Restricted in participation in major aspect of life Unable to live independently Need assistance for basic activities Table: Median valuations of each health state Back 4 / 9

89 Empirical Support for Prospect Theory? (?) Appendix Violation of EUT Framing s Figure: Certainty Equivalent Method 5 / 9 CE > expected value of the gamble = risk aversion Long duration risk aversion; short duration risk seeking Back

90 Message Framing Appendix Violation of EUT Framing s Message framing in cancer care Gain framed: emphasizes the benefits of taking the action Good things that will happen and the bad things that will be avoided Loss framed: emphasizes the costs of inaction Bad things that will happen and good things that will not happen Risk seeking in loss: Detective treatment Loss framed is more persuasive Risk averse in gains: Preventive actions Gain framed is more persuasive 6 / 9

91 s I Appendix s 7 / 9 Aron-Dine, A., Einav, L., & Finkelstein, A. (2013, jan). The RAND Health Experiment, three decades later. J. Econ. Perspect., 27(1), Retrieved from doi: /jep Barberis, N. C. (2013). Thirty Years of Prospect Theory in Economics: A Review and Assessment. J. Econ. Perspect., 27(1), Bleichrodt, H., & Pinto, J. (2005). The Validity of Qalys Under Non-expected Utility*. Econ. J.. Retrieved from Choi, B. C. K., McQueen, D. V., Puska, P., Douglas, K. A., Ackland, M., Campostrini, S.,... Others (2008). Enhancing global capacity in the surveillance, prevention, and control of chronic diseases: seven themes to consider and build upon. J. Epidemiol. Community Health, 62(5), Kahneman, D., & Tversky, A. (1979, mar). Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263. Retrieved from doi: /

92 s II Appendix s Marquis, M., & Holmer, M. (1996). Alternative models of choice under uncertainty and demand for health insurance. Rev. Econ. Stat.. Retrieved from McNeil, B., & Pauker, S. (1982). On the elicitation of preferences for alternative therapies. New Engl. J..... Retrieved from Oliver, A. (2003). The internal consistency of the standard gamble: tests after adjusting for prospect theory. J. Health Econ.. Retrieved from Osch, S. V. (2004). Correcting biases in standard gamble and time tradeoff utilities. Med. Decis..... Retrieved from Pinto-Prades, J. L., & Abellán-Perpiñán, J. M. (2005). Measuring the health of populations: The veil of ignorance approach. Health Econ., 14(1), Retrieved from doi: /hec.887 Rasiel, E. B., Weinfurt, K. P., & Schulman, K. A. (2005). Can prospect theory explain risk-seeking behavior by terminally ill patients? Med. Decis. Mak., 25(6), / 9

93 s III Appendix s Stanciole, A. E. (2008, oct). Health and Lifestyle Choices: Identifying Ex Ante Moral Hazard in the US Market. Geneva Pap. Risk Insur. Issues Pract., 33(4), Retrieved from doi: /gpp Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. J. Risk Uncertain.. Retrieved from Verhoef, L. C., De Haan, A. F., & Van Daal, W. A. (1994, apr). Risk Attitude in Gambles with Years of Life: Empirical Support for Prospect Theory. Med. Decis. Mak., 14(2), Retrieved from doi: / X WHO. (2015). Noncommunicable diseases. Retrieved from 9 / 9