We use two data sets to measure happiness, the General Social Surveys (GSS) from the U.S. and from Canada. The U.S. GSS is a nationally representative survey in the United States that has been administered to 1500 to 2500 households in most years since 1972; we use data from 1973 (the first year where state identifiers are available) through 1998. The Canadian GSS is a nationally representative survey of Canadians that has been administered sporadically since 1985; we use all available surveys that include a happiness question (1985, 1986, 1989, 1990, 1991, 1996, and 1998). Both surveys ask a variety of standard economics questions, but their use has mostly been in other disciplines, since the survey’s main focus is on questions not traditionally used by economists: attitudes towards current events or political parties; religious devotion; and psychological measures such as happiness. It is the last measure that forms our key dependent variable.
In particular, in each year the U.S. GSS asks respondents “Taken all together, how would you say things are these days — would you say that you are very happy, pretty happy, or not too happy”? The Canadian GSS question asks “Would you describe yourself as very happy, somewhat happy, somewhat unhappy, or very unhappy”, and there is also an option for “no opinion”. Since only a very small share of the sample responds that they are very unhappy or no opinion, we combine those responses with somewhat unhappy to form our unhappiness category. read
Another advantage of both surveys for our purposes is that both surveys have been carried out for many years. Over the time periods covered there have been enormous changes in the real excise tax rates charged by the states and the Canadian provinces, absolutely and relative to each other. It is these changes that provide the identifying variation for our model. Data on state cigarette excise taxes come from the publication The Tax Burden on Tobacco. We use state excise tax values as of February of each year, as the GSS data were collected over the February-April period. Data on Canadian tobacco taxes were collected by Gruber, Sen and Stabile, and incorporate both federal and provincial excise and sales taxes on cigarettes. We use the tax rate as of the month of the survey, since the Canadian GSS was collected in various months of the year over time.
Table 1 shows the means and standard deviations of the variables in both data sets. The first three columns show the GSS data from the US, and the second three columns show the data for Canada; in each set of columns, we first show the means for all respondents, and then separately by smokers and nonsmokers. We use three dummy variables as our dependent variables for measuring happiness, corresponding to the three possible answers to the happiness question above. Over our entire sample, in the United States 32% of respondents report themselves to be very happy, 55% are pretty happy, and 12% are not very happy.