International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Disentangling status quo bias and zero-price effect for more robust estimation of welfare changes
Jeff Tjiong, Marek Giergiczny, Mikołaj Czajkowski, Stephane Hess

Last modified: 28 March 2017

Abstract


Many stated choice surveys, especially those in environmental economics, are characterised by the inclusion of a status quo alternative that is a ‘do nothing’ scenario and has a zero cost (in contrast with the ‘intervention’ alternatives which have a non-zero cost). In this paper, we show that the common practice of interpreting implicit prices of the choice attributes (marginal willingness to pay) without taking the alternative specific constant for the status quo alternative into account may lead to erroneous conclusions. We demonstrate that under experimental designs commonly used for discrete choice experiments applied in environmental valuation (i.e. the situation in which zero-cost is only present for the status-quo alternative) results in the intercept of the cost sensitivity being captured by the alternative specific constant for the (zero-cost) status quo. The constant for the status quo option then often takes on a positive value (even if the status quo alternative is not chosen as often as others) as it captures some of the disutility of the non-zero cost for other options. This can lead to a downward bias in marginal willingness to pay measures and apparent counter-intuitive findings where the aggregation of marginal willingness to pay leads to total willingness to pay much higher than the highest presented costs (cf. Hess & Beharry-Borg, 2012).

The technical problems with the zero-cost status quo and respondents’ cost sensitivity are not directly related to whether the status quo bias should be included in calculations of welfare measure changes, although we show that including alternative specific constants in welfare change estimations following Haneman (1984) or Small and Rosen (1981) approach accounts for the problem – it is only marginal willingness to pay measures which get affected. Leaving the status quo is an important component of the indirect utility function and as such should be included in the calculation of welfare. As shown by Boxall et al. (2009) including the alternative specific constant for the status quo has a very large impact on welfare measures as it is likely to shift welfare measures from positive to negative (a similar pattern was reported by Adamowicz et al. 1998; von Haefen et al 2005;  Meyerhoff and Liebe, 2009). This effect is hidden from most environmental valuation studies in which the status quo constant is excluded from welfare calculations and only marginal willingness to pay for the other attributes are reported. A further complexity arises in the case where respondents exhibit non-linear sensitivity to cost and where the analyst relies on linear cost effects. Indeed, this can further increase the risk of confounding effects for the status quo constant.

In addition to technical considerations, our paper refers to the theoretical reasons for the observed problems. Numerous discrete choice experiment applications in environmental valuation evidence the presence of respondents’ strong preferences for the status quo alternative. This effect is known as a status quo bias and is usually attributed to survey complexity (Boxall et al., 2009), aversion to introducing new policies and regulations in general, or as a protesting behaviour (Meyerhoff and Liebe, 2009). A parallel stream of research in the behavioural economics literature indicates that zero price has a powerful psychological effect on respondents’ choices. Instead of making choices based on comparing the relative levels of the attributes, substantial share of respondents chose the zero cost alternative, even if it is associated with much lower ‘value for money’ (Ariely, 2008). Thus, people appear to act as if zero pricing of a good not only decreases its cost but also provides additional benefits. To the best of our knowledge, this effect has not been accounted for in designing and interpreting discrete choice experiments, or in deriving the resulting welfare measures. Because of common practice of designing experiments in such a way that the status quo alternative is the only one which can be achieved at zero cost, what is currently interpreted as the status quo bias could partly be attributed to zero-cost effect. A similar albeit reduced effect may happen if the status quo alternative is the cheapest throughout while still having a non-zero cost.

Empirically, our analysis is based on Monte Carlo simulation illustrating the effects of not including zero cost level in the non-status quo alternatives and demonstrating the impact on the estimates of alternative specific constants, price coefficients and the resulting marginal willingness to pay. In addition, we use the data from empirical discrete choice experiment study to demonstrate the difference between the status quo effect and the zero price effect. The zero-price effect can be eliminated by changing the frame in which discrete choice experiment is carried out. In split-sample treatments we maintain exactly the same differences in bids, however while in the first treatment the payment is expressed in absolute terms (the resulting cost in the status quo alternative is positive) in the second treatment the payment is expressed in relative terms – above what is currently paid (the status quo is thus associated with a cost of zero). We show that by expressing bids in absolute terms we significantly lowered the tendency to choose the status quo compared to the treatment in which status quo was associated with zero cost. We also relate our findings to the extensive amount of work on non-linear cost sensitivity in other fields, notably transport, a body of work largely ignored in environmental economics.

Overall, we show that appropriate treatment of the zero-cost alternatives needs to be considered at the design, modelling and interpreting the results stage of a discrete choice experiment. We demonstrate it has serious impact on the marginal willingness to pay measures. In conclusions, we call for changing the current practice in non-market valuation using the status quo alternative.

Conference registration is required in order to view papers.