2. LITERATURE REVIEW
2.2 Price Policy
2.2.2. Behavioural economic studies of price policy Using an experimental
220.127.116.11. Purchase task studies Studies suggest that consumption data generated from hypothetical purchase tasks generates demand curves which conform to the exponential
model of demand and are comparable to those generated in self-administration studies (Few, Acker, Murphy, & MacKillop, 2012; Grace et al., 2014, 2015a; MacKillop et al., 2008; MacKillop & Tidey, 2011). These tasks can also assess individual differences in substance
use using the demand metrics derived from demand curves (Murphy, MacKillop, Tidey, Brazil, & Colby, 2011). Demand metrics derived from the Cigarette Purchase Task (CPT) have shown robust convergent and divergent validity in adults (Chase, Mackillop, & Hogarth, 2013; Few et al., 2012; Grace et al., 2014; MacKillop et al., 2008; MacKillop & Tidey, 2011) and adolescents (Murphy et al. 2011). Studies have found significant positive correlations between derived demand metrics and smoking-related variables such as cigarettes per day and nicotine dependence, with the strongest correlations found for Q0 [r = .23 -.68] and Omax [r = .27-.88], and less so for breakpoint [r = .02-.41] (Chase et al., 2013; Grace et al., 2014; Mackillop et al., 2016; MacKillop et al., 2008; Murphy et al., 2011). Chase et al. (2013) performed a multiple regression with Omax, Q0 and breakpoint and found that all three metrics significantly predicted variation in nicotine dependence; Omax was the strongest predictor.
A potential limitation of using hypothetical purchase tasks is their reliance on estimated, self- reported consumption, which may not correspond with actual observed behaviour (Jacobs & Bickel, 1999). However, temporal discounting studies requiring
participants to choose between a delayed and immediate reward have shown that choices for hypothetical rewards correspond with choices for actual rewards (Bickel, Pitcock, Yi, & Angtuaco, 2009; Johnson & Bickel, 2002; Lagorio & Madden, 2005; Madden, Begotka, Raiff, & Kastern, 2003; Madden et al., 2004). Similarly, studies comparing hypothetical and actual reward conditions on an alcohol purchase task showed that there were no significant differences in performance between the two conditions, and high correlations between conditions [r = .87] (Amlung, Acker, Stojek, Murphy, & MacKillop, 2012; Amlung & MacKillop, 2015).
Only one study has compared actual and hypothetical conditions using a CPT. Wilson, Franck, Koffarnus, and Bickel (2016) found that daily purchasing behaviour on a hypothetical cigarette purchase task was not significantly correlated with real and potentially
real weekly purchase tasks [r = .41-.43; p>.05]. However, it was stated that this could be due to the different time horizons used; the hypothetical purchase task asked about cigarettes per day while the real and potentially real purchase conditions required participants to purchase one week’s worth of cigarettes. Additional studies using identical prices and consumption periods would be needed to confirm whether hypothetical cigarette consumption would reflect actual consumption. However in the CPT, as in alcohol purchase tasks, choices are being made for familiar goods organised in discrete and well-understood units (such as price in dollars and cigarettes either as single cigarettes or familiar, purchasable pack sizes) which should allow for accurate estimation of consumption (MacKillop et al., 2012). These studies suggest that the CPT may be a valid and reliable way to measure demand for cigarettes and the derived measures may be helpful for predicting which smokers will benefit most from tobacco excise tax increases, and which smokers may be better reached using alternative strategies.
Demand curve analysis provides a rich understanding of how changes in price would influence the decision to smoke, and the use of monetary price offers a clearer application to public policy. Purchase tasks can be useful in examining public policy issues because the same magnitude, range and density of price change cannot be examined in actuarial studies of existing market data (Hursh & Roma, 2013; MacKillop et al., 2012). Several studies have highlighted applications for public policy including predicting individual responses to excise tax increases (Grace et al., 2014) and behavioural interventions (Secades-Villa, Pericot- Valverde, & Weidberg, 2016) and left-digit price effects (MacKillop et al., 2014; MacKillop et al., 2012).
Grace et al. (2014) demonstrated the application of the CPT to public policy by establishing that CPT-derived demand indices can predict changes in smoking behaviour in response to excise tax increases. This study found that a measure of local elasticity based on
the regression slope for simulated demand at prices close to the market price of cigarettes (ranging from NZ$0.64 to NZ$0.85 per cigarette) predicted decreases in smoking after the tax increase. Secades-Villa et al. (2016) assessed whether CPT indices predicted treatment outcomes among smokers receiving combined cognitive behavioural therapy and contingency management. This study found that higher elasticity was associated with more days of
continuous smoking abstinence after controlling for cigarettes smoked per day, years of smoking and nicotine dependence. These studies suggest that the demand indices derived from the CPT may have clinical utility and can be predictive of behaviour change
(MacKillop, Miranda, et al., 2010).
MacKillop et al. (2012) used a range of 73 prices organised highly densely around current market price and identified that a major influence on price effects was left-digit bias: large magnitude effects on consumption based on the transition from one whole dollar amount to the next (e.g. $4.80 to $5.00 per pack). These results suggested that tax increases that traverse whole dollar pack prices would be more successful in reducing cigarette consumption. Another study found that changes in reported motivation to quit smoking at transitions across whole dollar prices per pack were approximately three times larger than changes in motivation at other price increases (MacKillop et al., 2014). This study also found that left-digit transitions had a much greater effect at market-relevant prices than at very low prices. Together, these studies highlight the importance of considering both the absolute amount and the relative position of a price increase to achieve the greatest reductions in smoking.
2.2.3. Summary and application to New Zealand. Price policy is internationally