Relative risk aversion around the world

Studies in Banking and zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA FiRaRct! 6 (1988) 127-128. North-Holland zyxwvutsrqponmlkjihgfedcbaZYXWVUTS RELATIVE RISK AVERSION AROUND THE WORLD Further Results George G. SZPIRO* zyxwvutsrqponmlkjihgfedcbaZYXWVUTSR Hebrew University of Jerusalem, Jerusalem, Israel Jean-Francois OUTREVILLE* UNCTAD Special Programme on Insurance In recent papers [Szpiro (1986) and this issue], an approximate demandfor-insurance function was used to estimate the degree of relative risk aversion as a function of GNP, for 2 :‘*i,izction of 15 countries. For most countries it was shown that constant relative risk aversion (CRRA) cannot be rejected. Insurance data for another set of countries, including 11 developing countries has become available, and the results for all 31 countries are presented in table 1. The methodology is identical to the one in the previous paper and in the references cited therein, except that a correction for secondorder autocorrelation is now made whenever necessary. As can be seen from panel 1 of table 1, the hypothesis of constant relative risk aversion can not be rejected for 24 countries of our sample at the 95% significance level, and for 29 countries at the 99% level? Panel 3 shows that the degree of relative risk aversion lies between about 1 and 5.’ For the ‘world as a whole’, mean h* is 1.14, and the mean degree of RRA is 2.89. Pooling all 686 observations, and correcting the data of each country for first-degree autocorrelation, we receive a degree of RRA of 2.5. Our results give more support for the hypothesis of CRRA and a coefficient of relative risk aversion greater than two, as expected. *The opinions expressed in this paper do not necessarily reflect those of the author’s respective institutions. ‘It must be noted, zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA however, that the standard errors for the estimate of h* are sometimes quite large, so that this test is not very powerful. 2Care must be taken when making inter-country comparisons, since observation periods differ. References Szpiro, George G., 1986, RRA around the world, Economics Letters 20, 19-21. Szpiro, George G., 1988, Insurance, risk aversion and utility, Studies of Banking and Finance, this issue. 0169- 6939/88/$3.50 0 1988, Elsevier Science Publishers B.V. (North-Holland) Table 1 Panel 1 Australia Austria Bolivia Brazil Canada Cyprus Denmark Finland France Germany Guatemala Holland Ireland Israel Italy Japan Korea Malaysia Mexico Morocco New Zealand Norway Portugal Singapore Spain Sweden Switzerland UK USA Venezuela Zambia Period Method” 1950-1980 19661984 19661983 1962-1982 1957-1981 1970-1983 1950-1980 1950-1980 1955-1984 1950-1979 19751984 1972-1979 1970-1980 1950-1980 1951-1980 1952-1980 1969-1984 1972-1983 19661983 1971-1982 1950-1979 1950-1979 1960-1983 1965-1983 1955-1980 1950-1979 1950-1980 1973-1979 1950-1980 1967-1984 1972-1984 AR1 AR1 AR1 AR1 AR1 AR1 AR1 AR1 AR1 AR1 OLS AR2 AR1 AR1 AR2 AR2 AR1 AR1 AR1 AR1 AR2 AR1 AR1 AR1 AR1 AR1 AR1 AR1 AR2 Q!,S AR2 “Method Panel 3 - h* T(h) Z(h)b 6 T(b) 0.58 3.11 0.19 1.58 -0.60 - 0.67 1.87 0.03 2.05 1.34 0.43 - 0.07 4.02 1.24 - 0.54 1.12 2.03 - 0.47 1.34 5.06 2.11 0.92 2.17 0.61 1.54 0.08 1.69 0.16 1.73 0.95 2.44 2.13 1.71 1.11 0.26 0.86 0.57 0.36 0.35 0.58 0.54 1.34 48.38 3.53 0.53 0.70 0.57 1.11 0.86 0.88 2.77 0.71 1.02 1.09 0.26 0.35 0.59 0.76 10.67 1.24 0.24 1.10 0.20 1.23 0.73 2.23” 1.86’ 2.93d 2.42’ 2.76d 1.81’ 0.63 0.43 0.02 0.86 0.45 2.20’ 0.21 0.93 1.71 0.39 1.47 1.56 0.08 1.07 1.50 1.54 1.56 0.91 0.08 0.59 0.21 1.31 0.02088 0.02228 0.00517 o.w93 0.01420 0.00785 0.02149 0.01382 0.02759 0.02063 0.00153 0.0253 1 0.02980 0.01655 0.00176 0.00408 0.00973 0.00768 0.00665 0.01300 0.00550 0.01819 0.01385 0.00754 0.01119 0.01393 0.022 10 0.01958 0.02720 0.01934 0.00441 7.3 16.0 11.1 14.3 20.2 13.3 25.5 14.6 7.4 6.5 17.7 8.2 2.8 7.0 2.6 4.5 4.9 16.9 8.2 22.6 1.7 6.2 8.5 15.3 17.9 10.2 9.4 2.2 10.9 7.5 4.9 n T(n) Rsq Lower limit D-W AR1 *3LS: Ordinary least squares, AR 1: First-order autoregressive process, AR2: Second-order autoregressive process. RRA Upper limit 4.39 8.92 - 2.27 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDC 0.986 1.63 2.63 - 0.00475 - 5.24 0.984 2.62 3.44 2.42 2.06 - 0.00850 2.09 3.00 - 0.00247 -4.15 0.806 2.09 1.53 4.09 5.12 -6.91 1.64 8.983 3.32 -0.OOi21 - 2.28 0.98 1 1.78 4.71 8.80 3.11 - 0.00301 2.03 3.15 4.12 - 0.00249 - 5.66 0.680 2.48 1.59 1.68 1.93 - 10.55 0.995 1.47 -0.01281 - 3.92 0.986 2.45 2.36 3.19 4.57 - 0.001$34 - 5.56 0.995 P.34 2.59 3.58 1.90 - 0.01065 - 9.24 0.995 1.60 2.93 3.79 - 0.00705 2.23 -6.14 0.865 1.34 2.64 3.26 4,ll -0.00047 2.03 1.42 2.32 -0.01777 - 3.20 0.92 1 0.95 - 3.01 0.924 1.63 2.00 4.06 0.96 -0.01492 - 7.43 0.983 2.37 2.93 3.87 5.11 - 0.00427 -- 3.90 0.995 0.81 1.51 -0.OU218 1.40 0.39 - 3.34 0.997 1.59 4.89 8.53 2.93 - 0.00084 - 5.01 0.97 1 1.85 2.06 3.09 4.64 -0.00315 - 6.79 0.987 2.27 3.98 4.95 3.27 -0.00193 - 7.64 0.987 2.08 3.78 4.87 6.29 -0.00137 - 5.34 0.684 2.24 1.X) 2.44 1.53 - 0.00684 -4.06 0.967 I.52 1.19 2.48 - 0.00463 0.41 - 3.21 0.985 1.35 2.89 4.87 1.85 - 0.0063C - 5.20 0.980 1.54 2.18 2.94 4.07 - 0.00470 4.99 6257 - 5.26 0.989 1.62 3.92 -0.00151 -11.07 0.993 1.88 1.80 2.08 2.41 - 0.00538 - 3.79 0.978 1.42 3.92 5.85 2.80 - 0.00355 -7.11 2.24 0.993 2.16 2.76 3.55 -0.00801 1.4” - 2.08 0.707 3.04 0.53 3.95 -0.01382 - 2.59 1.85 0.993 1.86 2.84 5.06 - o.OoY57 - 3.86 0.757 1.28 1.34 1.94 2.97 - 0.00997 1.32 - 2.88 0.45 1 1.85 3.15 5.81 -0.00140 1.14 Weighted average Pooled data 686 Obs. Panel 2 2.89 0.010544 19.70 - 0.00220 -11.69 0.333 1.16 bNumber of standard errors that II* is removed from h= 1. ‘h* different from h = 1 at the 95% significance level. dh* different from h = 1 at the 99% significance level. 2.17 _ 2.48 2.85