Abstract
We analyze the consumption of non-life insurance across 103 Italian provinces in 1998–2002 in order to assess its determinants, in the light of the empirical literature. Using sub-regional data, we overcome an important limitation of cross-country analyses, i.e. the systemic heterogeneity due to country-specific characteristics. Individual heterogeneity is accounted for through panel data techniques. However, considering spatial units within a single market raises issues of cross-sectional or spatial dependence, either due to common nationwide and/or regional factors or to spatial proximity. We carefully assess spatial dependence, employing recent diagnostic tests, finding out that the regressors included in our specification successfully account for spatial dependence. Insurance turns out to depend on income, wealth and some demographics, as already established, but also on trust, judicial efficiency and borrowing conditions. These findings help in explaining the gap between Central-Northern Italy and the south of the country.
Similar content being viewed by others
Notes
Data from Sigma, Swiss RE, relative to 2007.
The penetration of non-life covers related to so-called personal insurance, like accident and health, is also much higher in countries, such as Germany, where some citizens are allowed to opt out of the public health insurance scheme.
Due to data unavailability, in these international comparisons we separate motor from non-motor, instead of mandatory motor-TPL from the rest (including motor own damage). In the latter case the comparisons would be even less favourable, given the relatively low importance of non-mandatory motor in Italy.
See Esho et al. (2004) for a survey.
Their results are quite similar when regressing the amounts insured, but the role of age, macroregional dummies and education is weaker.
They also estimate, but do not report, a random effects regression with results similar to the fixed effects one.
The authors are ostensibly assuming all these factors to be time-invariant.
One perhaps more convincing proxy for price is the (inverse) loss ratio, i.e. the ratio between premiums earned and claims paid for a given period, used by Outreville (1990) and Esho et al. (2004), which can be seen as an ex-post estimate for the average price paid for the services of the insurer per each unit of claims paid. Yet the reasons for inclusion of a proxy for price in a reduced-form consumption equation are unclear (see the formalization by Beenstock et al. 1988).
Having been forced to omit these two time-invariant variables from the fixed effects analysis, these two claims are based only on pooled models, so that the above-mentioned inconsistencies between the two specifications may cast doubt on this finding.
In general, the tendency apparent from the various specifications reported in the literature is for income to be positively correlated with most other regressors, so that the more omitted variables there are, the greater the coefficient of income will result. It is therefore crucial to this end to specify a reasonably complete model.
For an example in the field of insurance development, see Beck and Webb (2003), where the RE specification is accepted only on the subgroup of developing countries.
Guiso et al. (2004a), analizing the causal link between financial development and growth from a regional perspective, observe that the level of political, regulatory and financial integration reached within Italy can be considered an upper bound for that a set of countries will ever be able to attain.
Provinces correspond to level 3 in the NUTS (Nomenclature of Territorial Units for Statistics) classification by Eurostat.
Notice that the model with insurance penetration as the dependent variable is a simple reparameterization of our model.
Data on insurance premiums are collected on a provincial basis by ISVAP, the Italian Insurance authority.
In other words, at a first glance heterogeneity in insurance consumption does not seem to be only due to differences in available resources, since the average propensity to buy insurance out of one’s income is almost as differentiated.
Tests and diagnostic plots for spatial correlation as well as spatial models are based on a spatial weights matrix constructed according to the principle of queen contiguity (that is, provinces are considered neighbours if they share a common border or vertex; see LeSage 1999). According to common practice, the matrix has been row-standardized. Reggio Calabria and Messina, divided by the Messina Strait, have been considered contiguous.
We verified the appropriateness of deposits as a proxy for wealth drawing on a new database from the Bank of Italy (see Albareto et al. 2008), comparing our data on bank deposits with their estimates of household wealth for the year 1998. On a per capita, cross-region basis, the correlation between bank deposits and real assets was 0.92, with financial wealth 0.80, both significantly positive at the 1% level. Province-level data are not available.
This is collected at regional level.
This indicator comes from the database of Guiso et al. (2004b) and is relative to 1999, thus it is included as a time-invariant variable; the variability over time should be almost negligible with respect to the high cross-sectional variance.
This (like the previous one) is included as a time-invariant variable both due to unavailability of other waves and in the belief that such attitudes would show scant variability over a 6- or 7-year horizon.
Of the 20 Italian regions, two (Valle d’Aosta and Molise) contain only one province.
The results for a specification where 18 regional fixed effects are included instead of the 5 macroregional ones, not reported, are quite similar and can be obtained from the authors upon request. The fact that the behaviour of the model does not change substantially between the two cases is taken as evidence in favour of the robustness of our specification.
We are therefore assuming that the inclusion of (macro)-regional and time effects does make the residual individual effect to be uncorrelated with the regressors. One further restrictive hypothesis our model rests upon is that all regressors be strictly exogenous, which is needed for consistency of the RE estimator.
A structural a-priori reason to suspect the presence of serial correlation is the existence of pluriannual contracts.
For the RE-HC model a robust Wald test is done as discussed in Wooldridge (2002, Ch. 10).
It must be noted that this test has power against the presence of random effects as well as against serial error correlation.
The results for the model with regional dummies are also generally consistent with the former, the main difference being that the coefficient for the share of the agricultural sector on GDP, while retaining the same sign, is not statistically significant any more.
We have also tested for differences in wealth elasticity across the two macroareas, without finding statistical evidence in favour of it.
The (lagged) loss ratio of the property sector, included in an alternative specification as an ex-post estimate of risk conditions, proved not significant. See also the discussion in Sect. 3.2.
While fixed effects estimators are inconsistent for fixed T in dynamic panel data models, Pesaran (2004, 6) shows that the CD test continues to hold also in this context, as long as the disturbances are symmetrically distributed.
References
Albareto G, Bronzini R, Caprara D, Carmignani A, Venturini A (2008) The real and financial wealth of Italian households by region. Household wealth in Italy 2007
Anselin L (1988) Spatial econometrics: Methods and models. Kluwer Academic Publisher, University of California, Santa Barbara
Arrow KJ (1971) Essays in the theory of risk bearing. Markham, Chicago
Baltagi BH (1995) Econometric analysis of panel data. Wiley, UK
Baltagi BH, Li Q (1995) Testing ar(1) against ma(1) disturbances in an error component model. J Econom 68(1):133–151
Baltagi BH, Song SH, Jung BC, Koh W (2007) Testing for serial correlation, spatial autocorrelation and random effects using panel data. J Econom 140(1):5–51
Beck T, Webb I (2003) Economic, demographic and institutional determinants of life insurance consumption across countries. World Bank Econ Rev 17(1):51–88
Beenstock M, Dickinson G, Khajuria S (1988) The relationship between property-liability insurance premiums and income: an international analysis. J Risk Insur 55(2):259–272
Bera AK, Sosa-Escudero W, Yoon M (2001) Tests for the error component model in the presence of local misspecification. J Econom 101(1):1–23
Bivand R (2009) spdep: spatial dependence: weighting schemes, statistics and models. R package version 0.4-34
Borch K (1986) Insurance and giffen’s paradox. Econ Lett 20(4):303–306
Briys E, Dionne G, Eeckhoudt L (1989) More on insurance as a giffen good. J Risk Uncertain 2(4):415–420
Browne MJ, Chung J, Frees EW (2000) International property-liability insurance consumption. J Risk Insur 67(1):73–90
Croissant Y, Millo G (2008) Panel data econometrics in R: the plm package. J Stat Softw 27(2), http://www.jstatsoft.org/v27/i02/
Cummins JD, Outreville JF (1987) An international analysis of underwriting cycles in property-liability insurance. J Risk Insur 54(2):246–262
Cummins JD, Weiss MA (2000) Analyzing firm performance in the insurance industry using frontier efficiency and productivity methods. In: Dionne G (ed) Handbook of insurance, chap 24. Kluwer, Boston, pp 767–829
Elhorst JP (2009) Spatial panel data models. In: Fischer M, Getis A (eds) Handbook of applied spatial analysis, chap 2. Springer, Berlin
Enz R (2000) The s-curve relation between per-capita income and insurance penetration. Geneva Papers Risk Insur Issues Pract 25(3):396–406
Esho NA, Kirievsky A, Ward D, Zurbruegg R (2004) Law and the determinants of property-casualty insurance. J Risk Insur 71(2):265–283
Falciglia A (1980) The demand for non-life insurance: a consumption-oriented model. Geneva Papers Risk Insur 17:45–53
Focarelli D, Savino C, Zanghieri P (2004) Chi assicura la salute e i propri beni in Italia? Prime riflessioni sulla base dell’indagine sulle famiglie italiane della Banca d’Italia. Quaderni di impresa assicurativa 3:179–197
Grace MF, Skipper HD (1991) An analysis of the demand and supply determinants for non–life insurance internationally. Technical report, CRMIR, Georgia State University
Gruber J, Poterba J (1994) Tax incentives and the decision to purchase health insurance: evidence from the self-employed. Q J Econ 109(3):701–733
Guiso L, Jappelli T (1998) Background uncertainty and the demand for insurance against insurable risks. Geneva Papers Risk Insur Theory 23(1):7–27
Guiso L, Sapienza P, Zingales L (2004a) Does local financial development matter? Q J Econ 119(3):929–969
Guiso L, Sapienza P, Zingales L (2004b) The role of social capital in financial development. Am Econ Rev 94(3):526–556
Guiso L, Sapienza P, Zingales L (2008) Trusting the stock market. J Fin 63(6):2557–2600
Hau A (2008) When is a coinsurance-type insurance policy inferior or even giffen? J Risk Insur 75(2):343–364
Hoy M, Robson AJ (1981) Insurance as a giffen good. Economics Letters 8(1):47–51
Hussels S, Ward D, Zurbruegg R (2005) Stimulating the demand for insurance. Risk Manag Insur Rev 8(2):257–278
King RG, Levine R (1993) Schumpeter might be right. Q J Econ 108(3):717–737
Leisch F (2002) Sweave: Dynamic generation of statistical reports using literate data analysis. In: Härdle W, Rönz B (eds) Compstat 2002 —Proceedings in computational statistics. Physica Verlag, Heidelberg, pp 575–580, http://www.stat.uni-muenchen.de/~leisch/Sweave
LeSage JP (1999) The theory and practice of spatial econometrics. Tech. rep., University of Toledo (US)
MacMinn R, Garven J (2000) On corporate insurance. In: Dionne G (ed) Handbook of insurance, chap 16. Kluwer, Boston, pp. 541–564
Millo G, Piras G (2010) splm: econometric models for spatial panel data. http://www.R-Forge.R-project.org/projects/splm/, r package version 0.1-21/r76
Mossin J (1968) Aspects of rational insurance purchasing. J Polit Econ 76(4):553–568
Outreville JF (1990) The economic significance of insurance markets in developing countries. J Risk Insur 18(3):487–498
Outreville JF (1996) Life insurance markets in developing countries. J Risk Insur 63(2):263–278
Outreville JF (2000) The retention capacity of insurance markets in developing countries. In: Dionne G (ed) Handbook of insurance, chap 23. Kluwer, Boston, pp 749–766
Pesaran MH (2004) General diagnostic tests for cross section dependence in panels. Working paper 1229, CESifo
Pinheiro J, Bates D, DebRoy S, Sarkar D, the R Core team (2009) nlme: linear and nonlinear mixed effects models. R package version 3.1-92
Pratt JW (1964) Risk aversion in the small and in the large. Econometrica 32(1–2):122–136
Prosperetti L (1992) Le determinanti della domanda di assicurazione danni nelle province italiane: un’analisi econometrica. Diritto ed economia dell’assicurazione 2–3:387–417
R Development Core Team (2009) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org, ISBN 3-900051-07-0
Schlesinger H (2000) The theory of insurance demand. In: Dionne G (ed) Handbook of insurance, chap 5. Kluwer, Boston, pp 131–151
Skogh G (1989) The transaction cost theory of insurance: contract impediments and costs. J Risk Insur 26(4):726–732
Sweeney GH, Beard TR (1992) The comparative statics of self-protection. J Risk Insur 59(2):301–309
Szpiro GG (1985) Optimal insurance coverage. J Risk Insur 52(4):704–710
VVAA (2007) Non-life insurance in Europe. Tech. rep., Comitee Europeen des Assurances
Wooldridge JM (2002) Econometric analysis of cross-section and panel data. MIT Press, Cambridge
Acknowledgments
We would like to thank for useful comments and suggestions at various stages of this work Giacomo Pasini, Annapaola Lenzi, Elisa Tosetti, Roberto Cannata and participants to the following conferences: Spatial Econometrics Association, Cambridge 2007; European Regional Science Association, Liverpool 2008; Applied Statistics Conference, Bled 2008; Italian Congress of Econometrics and Empirical Economics, Ancona 2009; and to internal seminars at DEAMS, University of Trieste.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Millo, G., Carmeci, G. Non-life insurance consumption in Italy: a sub-regional panel data analysis. J Geogr Syst 13, 273–298 (2011). https://doi.org/10.1007/s10109-010-0125-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10109-010-0125-5