Conference Paper BHPS-2009 Conference: the 2009 British Household Panel Survey Research Conference, 9-11 July 2009, Colchester, UK
All in the family: is family smoking causal?
The World Health Organization (1999) predicts that worldwide mortality from tobacco will rise from about four million deaths in 1998 to about 10 million deaths a year by 2030. While most of this increase stems from tobacco use in low income countries, deaths in developed countries are still significant. The European Partnership to Reduce Tobacco Dependence (2001) observes that: “Unless more is done to help the 200 million European adult smokers stop, the result will be 2 million European deaths a year by 2040.” Using data from the first 15 waves of the British Household Panel Survey, we investigate whether and how a youths decision to smoke is influenced by the behavior of his parents and siblings. We focus in particular on trying to understand whether a child is (causally) more likely to smoke when his or her parent smokes and, for children with older siblings, when his or her older sibling smokes. We pay particular attention to the question of causality. Although there is a positive association between the smoking behavior of youth, parents and siblings, its causality remains an open question in part because it is difficult to find valid instruments. We propose and use two new approaches. First, we use temporal distance between parent and child to predict the probability a parent starts to smoke and then use the price of cigarettes to predict the parents decision to continue to smoke (Lillard 2008). As a stylized fact, the majority of smokers in all developed countries begin to smoke between the ages of 13 and 19 (Kenkel, Lillard, and Mathios 2004). After age 19 the probability a person starts to smoke falls sharply. We exploit this stylized fact to predict which parents smoke as a function of cigarette prices and taxes they faced in these critical years. Because the second generation faces their decision to start about a generation later, we argue that cigarette prices/taxes in the early years are orthogonal to the cigarette prices at the time the children started. We also investigate the relationship between the smoking behaviour of siblings. Here we use role-model hypotheses to argue that sibling influences will vary with the sex composition of ones older siblings. Under the assumption that the sex of ones older sibling is exogenously assigned, we explore whether one is more or less likely to smoke if one has a same sex or opposite sex sibling who smokes.