An investigation into the robustness of willingness to pay for non-market goods in relation to subjective well-being
2016-10-31T16:05:43Z (GMT) by
This thesis investigates the sensitivity of derived monetary valuations of the well-being effects of non-market goods, by considering, in turn, the four components that contribute to these calculations. Comparisons are made to the current subjective well-being (SWB) literature by altering one component at a time in a willingness to pay (WTP) function, in addition to varying the estimator used when calculating WTP. The first component varied is the measure of income used; a more robust, up-to-date measure is proposed that takes into account household size, economies of scale and composition, plus other improvements over current income scaling approaches including equivalence scales. Secondly, anticipation and adaptation effects are examined, in order to allow for the dynamics of SWB in WTP valuations. Thirdly, due to the ordered nature of the dependent variable, the effects of both cardinal and ordinal models on WTP derivations are investigated. Such models include the ordered logit fixed effects (FE) model (see for example Das and Van Soest (1999), Ferrer-i-Carbonell and Frijters (2004) and Baetschmann et al. (2011)). The final input to be varied is whether income endogeneity is accounted for. A control function approach is used in this case. A key finding is that all four inputs in the WTP calculation contribute significantly to the estimated WTP values. However, the degree to which each input influences the variation in WTP values differs substantially. The three main contributing factors to WTP variations are the income variable chosen, anticipation and adaptation effects, and controlling for endogeneity. Additionally, this thesis proposes a new way to define WTP that is also valid for ordered estimators. Whilst the standard WTP approach relies simply on coefficient ratios at a point in time, the method proposed here uses the finding that statistically significant anticipation and adaptation effects, along with adaptation to all non-market events (except for unemployment) suggest that there are concerns with taking WTP values as yearly valuations that assume constant effects on SWB. By instead defining WTP as the summation of the value of all WTP effects over all lag and lead effect periods, the total value of an event can be calculated. This definition presents further weaknesses of the original method by showing that because of significant lag and lead effects for events such as divorce and widowhood, these events have substantially larger WTP values than first thought. This leads to the major finding in this study that rank orders are not preserved when controlling for income endogeneity, even when comparing two model specifications that are identical in all other respects. Therefore, despite absolute WTP valuations appearing more similar for linear and ordered estimators when a more appropriate model specification is in place, the rank orders are affected. This finding argues against papers such as Powdthavee and van den Berg (2011) that find that rank orders are preserved for WTP valuations when modelling SWB. In conclusion, given rank orders are generally not preserved when a small change is made in the model specification, the validity of the WTP method is significantly reduced. Finally, the above methodology is applied to education, with the main contribution to the education literature being that the positive effects on SWB from education only exist in terms of lead effects.