Love conquers all but nicotine

Jan van Ours, Ali Palali 16 October 2015

a

A

According to the World Health Organisation, tobacco use is an epidemic. In the past decades a large amount of evidence has piled up about the adverse health consequences of tobacco use. The gloomy health effects of tobacco have stimulated many governments to influence its use through cigarette tax policies or tobacco control policies. One of the main aims is to help smokers to quit. Therefore, it is important to understand the dynamics in quit decisions of two-smoker couples. The decision of one partner to quit smoking may induce the other partner to quit smoking as well. Whether or not this happens is a relevant topic for policymakers in their fight against tobacco use. If spousal peer effects in quitting-to-smoke decisions exist, then anti-smoking government policies get ‘two for the price of one'.

There are several mechanisms for spousal peer effects. The first is possible bargaining between the partners. The partner who quit smoking might try to convince the other to quit smoking as well. The reason is not always clear. The partner who quits first can do so because he or she wants to protect the other from the adverse effects of smoking. However, it is also likely that he or she thinks that quitting will be hard if the partner persists in smoking. Whatever the reason is, the spouse who decides to quit first can have an interest in the partner quitting smoking as well. The second mechanism is learning. Partners can learn from the smoking decision or the quit decision of one another. If there is such a partner-caused accumulation of information, then the decision of one partner might affect the other. The third is spill-over effects. One partner can consider the quit decision of the other as an incentive to quit smoking.

Partners are similar to begin with

Analysing the effect of the quitting decision of one partner on another is not an easy task. Partners tend to be similar, because they get together through selection in the marriage market (assortative matching). This means that those who smoke are more likely to match with smokers, and those who are more likely to quit are likely to match with potential quitters. Several studies from different fields of social sciences find that individuals partner through an assortative matching process, and therefore share similar personalities, behaviours, proclivities, and risk attitudes (Humbad et al. 2010, Leonard and Mudar 2003, Canta and Dubois 2015, Powdthavee 2009, Abrevaya and Tang 2011). Humbad et al. (2010) state that partners show similar personality traits and that these similarities are mostly due to assortative matching. Leonard and Mudar (2003) show that similarities between partners are not limited to personality traits but can also be found in observable behaviours such as drinking habits. Canta and Dubois (2015) find similar results for smoking behaviour; there is a significant correlation between the cigarette smoking patterns of partners. Therefore, any attempt to study true spousal peer effects in the decision to quit smoking has to take the correlation due to assortative matching into account. According to Clark and Etilé (2006), assortative matching on lifestyle preferences can be picked up by correlated individual effects in the male and female smoking equations. Conditional on this correlation, the authors find that smoking behaviours of partners are statistically independent, that is, that there are no spousal peer effects. On the other hand, McGeary (2015) finds evidence for spousal peer effects by investigating the quit decision of partners.

Our analysis

We investigate whether there are spousal peer effects in the decision to quit smoking using a unique data set from the Netherlands (see Palali and Van Ours 2015 for details). The data includes information on the smoking behaviour of 812 couples. Table 1 shows that 75% of males and 61% of females have ever been smokers. Among those who have ever smoked, 35% of males and 46% of females had quit smoking.

Table 1. Percentage of females and males in couples who ever smoked cigarettes (on the left) and who quit smoking conditional on ever smoking (on the right), in %

Table 2 presents the distribution of females and males in couples based on starting and quitting smoking, showing that smoking and quit decisions of partners are highly correlated. In this table, we define three groups for both females and males – those who start and quit using tobacco, those who start and do not quit using tobacco and those who do not start using tobacco. In almost 50% of the couples (15+19+14) both partners follow the same starting-quit behaviour. We also see that the percentage of couples in which the male ever uses tobacco but the female never starts (14+11) is considerably higher than the percentage of couples in which the female ever smokes but the male never starts (5+6).

Table 2. Distribution of females and males in couples based on starting and quitting smoking (%)

Finally, Figure 1 plots the couples according to the years in which males and females quit smoking. In this figure we only plot couples where both partners quit smoking to avoid confusion due to censored observations. The figure shows that even though there is considerable number of observations around the 45-degree line, there is a lot of variation in terms of quit years.

Figure 1. Scatter plot of couples according to the years in which males and females quit smoking

Notes: In the figure each dot represents a couple consisting of two quitters. Since there is no quit year for someone who persists in smoking until the survey time, we exclude such observations from this figure. Note that those observations are not excluded from our analysis, but assumed as censored observations.

Our data include information about starting age of smoking as well as quitting age. Using this information, we estimate mixed proportional hazard models, which allow us to study how transitions to smoking status and non-smoking status are affected by observed and unobserved individual characteristics. First, we estimate smoking dynamics for both partners separately assuming that a quit decision of one partner is exogenous to the quit decision of the other partner. The first 2 columns of Table 3 present the results. The table shows that the quit decision of one partner has a positive effect on the quit rate of the other partner.

However such estimation assumes that there is no correlation in the smoking behaviour through unobserved characteristics, that is, one partner's decision to quit smoking is orthogonal to the decision of the other partner. This is unlikely to be the case due to the assortative matching underlying partnership formation, and common external shocks to the household. As indicated earlier, quitters are more likely to partner up with potential quitters to begin with. In order to control for correlated behaviour in the decision to quit smoking, we perform a joint maximum likelihood estimation of partners' starting and quitting behaviour in which we allow for spousal correlations in unobserved heterogeneity. The results are presented in columns 3 and 4 of Table 3. We find that the significant positive affect disappears once we control for correlated behaviour between partners. As such, the similarities in the smoking behaviour of partners are likely due to assortative matching during partnership formation and common household shocks. They are not due to bargaining, learning or spill-over effects.

Table 3. Parameter estimates of the partner effect for males and females in couples

Note: Based on information about smoking behaviour of 812 couples; absolute t-statistics in parenthesis.

We conclude that the behaviour of two partners is correlated, and while there may be cross-partner effects in behaviour, this likely doesn’t extend to the decision to quit smoking. Apparently, love conquers a lot and perhaps all, except for nicotine addiction.

References

Abrevaya, J and H Tang (2011) “Body mass index in families: Spousal correlation, endogeneity, and intergenerational transmission”, Empirical Economics, 41: 841-864.

Canta, C and P Dubois (2015) “Smoking within the household: Spousal peer effects and children’s health implications”, BE Journal of Economic Analysis and Policy, 15, forthcoming.

Clark, A E and F Etilé (2006) “Don't give up on me baby: Spousal correlation in smoking behaviour”, Journal of Health Economics, 25: 958-978.

Humbad, M N, M B Donnellan, W G Iacono, M McGue, and S A Burt (2010) “Is spousal similarity for personality a matter of convergence or selection?”, Personality and Individual Differences, 49: 827-830.

Leonard, K E and P Mudar (2003) “Peer and partner drinking and the transition to marriage: A longitudinal examination of selection and influence processes”, Psychology of Addictive Behaviors, 17: 115-125.

McGeary, K A (2015) “Spousal effects in smoking cessation: Matching, learning, or bargaining?” Eastern Economic Journal, 41: 40-50.

Palali, A and J C van Ours (2015) “Love conquers all but nicotine; spousal peer effects on the decision to quit smoking”, CEPR, Discussion Paper 10860.

Powdthavee, N (2009) “I can't smile without you: Spousal correlation in life satisfaction”, Journal of Economic Psychology, 30: 675-689.

a

A

Topics:  Health economics

Tags:  Smoking, Tobacco, Cigarettes, taxes, smoking cessation, Peer Effects, quitting, spousal peer effects, health, Netherlands, assortative mixing, Matching, tobacco control, sin taxes

Professor of Applied Economics, Erasmus School of Economics, Erasmus University Rotterdam, The Netherlands Professorial Fellow at the Department of Economics, University of Melbourne; CEPR Research Fellow

Research Fellow, CPB Netherlands Bureau for Economic Policy Analysis

Events