Extract from. Études et Dossiers No M.O.R.E. 21 Seminar of The Geneva Association. Paris, September 2007

Transcription

1 International Association for the Study of Insurance Economics Études et Dossiers Extract from Études et Dossiers No. 333 M.O.R.E. 21 Seminar of The Geneva Association Paris, September 2007 & 25 th Anniversary of the World Fire Statistics Centre Paris, 25 September 2007 This document is free to download from The Geneva Association website, October 2007 Working Paper Series of The Geneva Association Association Internationale pour l'etude de l'economie de l'assurance The Geneva Association - General Secretariat - 53, route de Malagnou - CH-1208 Geneva Tel.: Fax: secretariat@genevaassociation.org -

2 The Geneva Association Working Paper Series Études et Dossiers appear at irregular intervals about times per year. Distribution is limited. The Études et Dossiers are the working paper series of The Geneva Association. These documents present intermediary or final results of conference proceedings, special reports and research done by The Geneva Association. Where they contain work in progress or summaries of conference presentations, the material must not be cited without the express consent of the author in question. This document is free to download from The Geneva Association website, please visit: Layout & Distribution: Valéria Kozakova The Geneva Association - Association Internationale pour l'etude de l'economie de l'assurance

3 Spillover Moral Hazard A Challenge to Fire Insurance Peter Zweifel Paris 1 1. Introduction Figure 1: Impulses determining the claims processes B Building/fire; C Contents/fire COMMON IMPULSES Risk B SPILLOVER MORAL HAZARD Risk C SPECIFIC IMPULSES SPECIFIC IMPULSES Paris 2 9-1

4 1. Introduction (cont d) Specific case: Insurance against damage to contents (of buildings) caused by fire in Switzerland Common impulses: insured s degree of risk aversion (e.g.), influences also preventive effort as well as demand for insurance w.r.t. the building (B). Spillover moral hazard: caused by public regulation of prevention of fires in buildings. Paris 3 1. Introduction (cont d) Challenges to insurers: Fire insurer: Contents insurer: Multiple line insurer: May calculate a wrong premium if neglecting the spillover moral hazard emanating from fire insurance coverage for contents May calculate a wrong premium if neglecting the spillover moral hazard emanating from insurance coverage for fire in buildings Needs to calculate the correct total premium by taking into account spillover moral hazard from line A to line B Paris 4 9-2

5 Objectives (1) Does the inclusion of information about common impulses improve the prediction of the probability of loss? (2) Does the inclusion of information about spillover moral hazard improve the prediction of the probability of loss? (3) Does public regulation of building safety create (positive/negative) external effects on private contents insurance? Paris 5 2. Review of the literature Doherty/Schlesinger (1983) and Mayers/Smith (1983) initiated the analysis of multiple risks. Gollier/Scarmure (1994) studied the case of uncorrelated risks: traditional results w.r.t. the demand for insurance have to be modified. However, the effect of a background risk on preventive effort has not been studied thus far; spillover moral hazard : (1) Coverage for risk B may cause a reallocation of prevention away from B and in favor of C (behavioral effect); (2) The prevention technology may be complementary, so a reduction of effort against risk B causes risk to C to increase as well (technological effect). Paris 6 9-3

6 Conclusion 1: A non-optimal compulsory level of insurance coverage has a spillover moral hazard effect on the probability of loss of an unregulated risk. First, it affects preventive effort directed at the unregulated risk, and second, it induces changes in the prevention of the regulated risk that influence the unregulated risk through the prevention technology. Its total effect cannot be determined analytically Paris 7 3. Types of common impulses Examples of common impulses: - Insured s degree of risk aversion - Regulation affecting different lines of insurance, - General economic situation of the country Degree of risk aversion as a common impulse Clear prediction w.r.t. demand for insurance: positive relationship Intuition w.r.t. preventive effort: positive relationship However, Dionne/Eeckhoudt (1985) and Briys/Schlesinger (1990) find an ambiguous relationship w.r.t. prevention that lower π Jullien/Salanié/Salanié (1999) show that increasing risk aversion increases prevention if its effects are felt in the lower tail of the wealth distribution Paris 8 9-4

7 4. Empirical analysis 4.1 Contents insurance: Insurable risks and relation to building/fire insurance Regulation of building/fire (B) and contents (C) insurance: - B is a public monopoly in 19 (out of 26) Cantons - In 2 Cantons, the public B insurers also have the monopoly for coverage of risk C -Binsurance compulsory in most Cantons - C is supplied by private insurers in most Cantons - C insurance is compulsory in 8 Cantons Preventive effort against risk B also affects C Vice versa, preventive effort against risk C also affects B The prevention technology exhibits a high degree of complementarity Paris Data and variables containing information about multiple risks Observation period , major Swiss insurer X spec : X spil : X comm : DLOSS: risk specific impulses, i.e. characteristics of the insured and his property; regulation of C; spillover moral hazard effects; common impulses; dependent variable, indicator of probability of loss π, = 1 if at least one claim (contents due to fire occurs during a given year; ε: normal error term, not i.i.d. Paris

8 4.3 Spillover moral hazard information I DCOMP_B: = 1 if B insurance is compulsory in the Canton According to Conclusion 1, the sign of the coefficient of DCOMP_B cannot be predicted However, the public monopoly is capable of forcing an excessive level of prevention P B Given complementarity in prevention technology, there may well be beneficial spillover moral hazard Paris Spillover moral hazard information II Table 2: Loss probability in contents fire insurance: static and dynamic model Variable d X spil DCOMP_B X comm NUMRISK BEFORE1 (I) Static: RE C without BEFORE1 Coefficient t a ME b (II) Dynamic: RE C with BEFORE1 Coefficient t a ME b ,**,*: Significantly different from 0 at the 1%, 5%, 10% level a: t-statistics, b: marginal effect calculated at variable means, c: random effects probit model, d: for complete estimation results see Table A1 of the appendix Paris

9 4.3 Spillover moral hazard information III Interpretation In Cantons that have compulsory fire insurance for buildings, the likelihood of a loss in contents/fire insurance is 1 to 1,4 percentage points lower (other things equal) spillover moral hazard effect! With each additional risk insured, the likelhood of a loss in contents/fire insurance increases by 0.2 to 0.3 percentage points likely moral hazard effect! The fact that there was a loss in the year before is associated with a likelihood of loss in contents/fire of 1,5 percentage points likely adverse selection effect! Conclusion: A beneficial spillover moral hazard effect between insurance lines can be identified in the case of regulated building fire insurance lowering the probability of loss in contents insurance. Complementarity in prevention technology seems to dominate behavioral effects Paris Common impulses information IV Risk aversion (and hence level of general preventive effort) can also be proxied by the loss experience w.r.t. other C risks covered (fire, water, glass, theft) DOUBFIRE contains the loss experience w.r.t. the 4 lines, with fire given double weight NOFIRE contains the loss experience w.r.t. 3 lines, excluding fire Paris

10 4.3 Common impulses information V Table 5: Dynamic model with loss experience from other contents risks (VII) RE C with DOUBFIRE (VIII) RE C with NOFIRE Variable Coefficient t a ME b Coefficient t a ME b d X spil DCOMP_B X comm NUMRISK DOUBFIRE NOFIRE * ** Paris Common impulses information VI Interpretation I Compulsory fire insurance of buildings again exerts a (beneficial) moral hazard effect The number of risks covered (possible indicator of risk aversion rather than moral hazard) loses statistical significance when loss experience is constructed in a better way (DOUBFIRE) moral hazard effect prevails When loss experience excludes fire (bad indicator of preventive effort), statistical fit deteriorates moral hazard effect decisive Paris

11 4.3 Common impulses information VII Interpretation II How do BEFORE3 (risk specific) and DOUBFIRE (multiple risks) compare? Estimated marginal effect of DOUBFIRE on π C is 0.072, that of BEFORE3, Including multiple risks information is relevant! Paris 17 Conclusion 2: The information about the common impulse risk aversion (and hence general preventive effort) contained in the loss experience of other contents risks contributes to the explanation of the probability of loss in contents fire insurance. It can and should therefore be used in risk classification Paris

12 5. Conclusion and outlook For ensuring future competitiveness, insurers would be well advised to use all the risk information readily available in the calculation of their premiums. Here, two types of additional information are considered, spillover moral hazard and common impulses. Compulsory coverage and mandated prevention against fire risk in buildings in Switzerland (B) may be sources of spillover moral hazard, while an increased number of other risks covered, riskspecific loss experience, and other (multiple-risk) loss experience in contents (C) insurance constitute instances of common impulses. Econometric evidence with regard to the probability of loss in C insurance points to the existence of beneficial moral hazard emanating from compulsory B insurance. It also demonstrates the value of multiple risk information. These findings are relevant for insurers doing multiple lines business, especially when closely related to a regulated line of insurance. Paris