6. Econometric Model
7.3. Additional Personal Characteristic Variables
The analysis will now turn to additional personal characteristic variables. The analysis will be based on a model with the personal characteristics used above and also two of the supply side and quality variables. These will be regional waiting times and PMI premium inflation. Table 8 below shows the results of 8 equations that include additional personal characteristic variables that were found to be significant.
Equation 17 shows the base equations with the personal characteristic variables used in the analysis above and also regional waiting lists and PMI price inflation. The results for the personal characteristic variables are very similar to above and so need no further discussion. The coefficient on regional waiting lists is positive and significant at the 1% level and the coefficient on PMI premium inflation is negative and significant at the 5% level. Lagged public expenditure on health found to be significant in Table 7 above is not included as the introduction of lags reduces the sample size, there are also issues of collinearity if there is a relationship between public expenditure on health and future waiting lists. For all equations the diagnostics are similar, the hypothesis that all coefficients are equals to zero is always rejected at the 1% level and the hypothesis thatρ is equal to zero is always rejected by the likelihood ratio test at the 1% level. The diagnostic will thus not be discussed further.
Table 8
Random Effects Probit of Private Medical Insurance Demand with Additional Personal Characteristic Variables
Standard Error in Parentheses
* Significant at the 5% level, ** Significant at the 1% level.
Equation 18 includes a broadsheet newspaper indicator variable, taking the value 1 if the individual reads a broadsheet newspaper and 0 otherwise. The effect of reading a
Equation 17 18 19 20 21 22 23 24 Sex 0.4373** 0.4371** 0.4460** 0.3759** 0.3595** 0.4188** 0.4376** 0.2384** (0.0532) (0.0531) (0.0537) (0.0564) (0.0548) (0.0536) (0.0536) (0.0602) Age 0.0044** 0.0044** 0.0043** 0.0031* 0.0041** 0.0051** 0.0038* 0.0018 (0.0015) (0.0015) (0.0015) (0.0015) (0.0015) (0.0015) (0.0015) (0.0016) Income £000 0.0160** 0.0160** 0.0149** 0.0157** 0.0158** 0.0149** 0.0158** 0.0129** (0.0013) (0.0012) (0.0013) (0.0013) (0.0013) (0.0013) (0.0013) (0.0014) Smoker -0.1643** -0.1560** -0.1073* -0.1711** -0.1402** -0.1445** -0.1524** -0.0773 (0.0514) (0.0513) (0.0527) (0.0511) (0.0515) (0.0518) (0.0521) (0.0525) Self Employed 0.6330** 0.6301** 0.5885** 0.6299** 0.6285** 0.5338** 0.6210** 0.4834** (0.0700) (0.0699) (0.0693) (0.0698) (0.0693) (0.0717) (0.0705) (0.0702) Unemployed -0.5307** -0.5314** -0.4859** -0.5268** -0.4501** -0.5094** -0.5228** -0.3778** (0.1265) (0.1264) (0.1294) (0.1264) (0.1276) (0.1263) (0.1267) (0.1305) Conservative Supporter 0.3651** 0.3670** 0.3299** 0.3665** 0.3530** 0.3612** 0.3663** 0.3137** (0.0478) (0.0478) (0.0477) (0.0477) (0.0476) (0.0476) (0.0478) (0.0469) Broadsheet Newspaper - 0.1905* - - - 0.1782* - (0.0820) - - - (0.0811)
Owner Occupied House - - 0.4716** - - - - 0.4664**
- - (0.0527) - - - - (0.0543)
Head Of Household - - - 0.1566** - - - 0.2609**
- - - (0.0494) - - - (0.0520)
Recives Benefit Income - - - - -0.2723** - - -0.2445**
- - - - (0.0484) - - (0.0497)
Private Pension - - - 0.3936** - 0.3521**
- - - (0.0532) - (0.0497)
Married - - - 0.1115* 0.0903
- - - (0.0477) (0.0490)
Regional NHS Waiting Lists 0.0275** 0.0257** 0.0267** 0.0273** 0.0278** 0.0251* 0.0276** 0.0235* (0.0102) (0.0103) (0.0102) (0.0102) (0.0102) (0.0102) (0.0102) (0.0102) PMI Premium Inflation -0.0054* -0.0030 -0.0064* -0.0054* -0.0038 -0.0054* -0.0053* -0.0025
(0.0025) (0.0053) (0.0029) (0.0026) (0.0052) (0.0026) (0.0025) (0.0053)
Regional dummies Yes Yes Yes Yes Yes Yes Yes Yes
Time dummies Yes Yes Yes Yes Yes Yes Yes Yes
Log Liklihood -9698.44 -9695.77 -9656.88 -9693.42 -9682.42 -9671.19 -9695.70 -9605.14
Number of Obs 56436 56436 56436 56436 56436 56436 56436 56436
Number of Individuals 16258 16258 16258 16258 16258 16258 16258 16258
Average Obs per Individual 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5
Wald 577.26** 582.47** 639.60** 575.14** 608.83** 608.55** 568.39** 707.05** 1.3020 1.2970 1.2579 1.2945 1.2879 1.2835 1.3001 1.2269 (0.0484) (0.0484) (0.0477) (0.0480) (0.0488) (0.0486) (0.0484) (0.0473) 1.9174 1.9127 1.8756 1.9193 1.9039 1.8998 1.9162 1.8468 (0.0464) (0.0463) (0.0447) (0.0459) (0.0465) (0.0462) (0.0464) (0.0436) 0.7862 0.7853 0.7787 0.7849 0.7838 0.7830 0.7859 0.7733 (0.0081) (0.0082) (0.0082) (0.0081) (0.0083) (0.0083) (0.0081) (0.0082) LR of 5905.36** 5872.32** 5771.69** 5890.50** 5844.16** 5856.36** 5904.51** 5654.48** 2 χ2 χ σ ) ln(σ2 ρ 0 = ρ
broadsheet newspaper are more likely to purchase PMI. The inclusion of this variable makes PMI premium inflation insignificant although the estimated coefficient is still negative.
Equation 19 introduces a indicator variable taking the value 1 if the individual lives in an owner occupied house and 0 otherwise. As in Besley et al (1999) the effect of living in an owner occupied house has a positive and significant effect on PMI purchase. The size of this effect is similar to that of being male instead of female, thus the effect is quite large in terms of the purchase decision. This variable may however be picking up more than simply the effect of living in an owner occupied house. The reduction in the income coefficient, which was generally stable across most specifications, suggests that the variable may be picking up an income effect with those individuals with higher incomes more likely to have owner occupied homes.
The head of household indicator variable taking the value 1 if the individual is regarded as the head of the household and 0 otherwise has a positive estimated coefficient that is significant at the 1% level. For individual paid policies this is to be expected.
Equation 21 includes an indicator variable taking the value 1 if the individual receives benefit income and zero otherwise. The estimated coefficient is negative and significant at the 1% level. There is a possible issue of collinearity here as the unemployment and benefit variables are strongly correlated. The estimated coefficient on the unemployed indicator variable is only slightly reduced and remains
highly significant, this will be due to the benefit variable including more than just unemployment benefit and so the separate variables are picking up separate effects.
It was argued above that the Conservative Party supporter variable was acting as a proxy for attitude towards private sector provision. Equation 22 uses an additional variable that may be used as a proxy for an individual’s attitude towards private provision of services. The indicator variable private pension takes the value 1 if the individual has a private pension and 0 otherwise. The estimated coefficient is positive as expected and the effect is significant at the 1% level. Those individuals with a private pension are more likely to favour private provision of services and the results for Equation 22 show that those with a private pension are more likely to purchase PMI. Equation 23 shows that being married has a positive effect on PMI purchase with this effect significant at the 5% level.
Equation 24 includes all additional personal characteristic variables. All the variables except the married indicator variable remain significant but the inclusion of all these variables affects the estimated coefficients of the core personal characteristics. This is probably due to issues of collinearity. For Equations 17 to 24 regional waiting times remain significant suggesting that this result is not sensitive to different equation specifications. PMI premium inflation is not significant in all of the equations and is more sensitive to the specification of the equation. This may be due to the issues discussed above about not having an individual specific premium price and some of the personal characteristic variables picking up price effects.