We begin with the baseline specification of log taxed sales per capita regressed on the log real price (including taxes) as well as the real price interacted with the share of people in the state-year with Internet access as measured in the Forrester data, instrumenting for the price terms using the log of the real tax term and the level of the real tax term. We present the results from this regression in column (1) of Table 2. The estimated elasticity of taxable sales even before the rise of the Internet is, at -1.28, already at the high end of elasticities found in the previous literature. The elasticity of taxable sales with respect to the mean tax-inclusive price of neighboring states is positive, as expected, although it is not significantly different from zero.
The interaction of the state tax rate with the share of the state that uses the Internet is negative and very significant. Clearly the data indicate that the sensitivity of purchases to prices has increased substantially. The point estimates suggest that growth of the Internet in the sample from zero to its average value in 2000 corresponds with a near-doubling of the elasticity, from -1.28 to -2.09.
As a check on the robustness of the result using the CPS measure of Internet usage, the next several columns try alternatives. Column (2) uses the alternative Forrester measure of Internet usage in the state. Again the results show a large and significant increase in the elasticity of taxable sales in states as the usage of the Internet rises. The magnitudes are close. Here the elasticity rises from -1.35 to -1.97 with the growth of the Internet in the sample. Column (3) first differences the data, using the CPS measure of Internet use. The results again show the rising price sensitivity of cigarette purchases in places where the Internet is growing fastest (controlling for state and year dummies), though the overall magnitude is smaller. Here the growth of the internet corresponds with a change in the elasticity from -1.08 to -1.58. Columns (4) and (5) deal with the issue of imputed values. Column (4) restricts the sample to only those years where Internet usage is positive (i.e., no imputed zeros) by looking only at the years after 1995. Column (5) uses only years in which the CPS actually has observations (i.e., no imputed Internet usage between survey years). Both procedures yield very similar results.
Finally, in column (6), we include lags and leads of the tax rates to determine if the results are simply the result of short-run shifting of purchases. We find that the lagged and leading tax variables have a very small impact. All that matters are the contemporaneous values. Short-run timing shifts are, on average, relatively small (perhaps because the tax changes are not anticipated well) and the baseline effects are almost the same as before.
These baseline specifications, then, suggest that as Internet use grew in states, the sensitivity of taxable Internet sales in those states grew as well and that the elasticity has risen greatly. We next consider whether this evidence is consistent with alternative explanations or is likely to be tied to a rise in cigarette smuggling due to the Internet. Although it did not estimate a price elasticity, a recent medical study of 3500 smokers in New Jersey in the period surrounding a large increase in cigarette taxes points in the same direction as our findings in that it documented that the share of people reporting they had bought cigarettes online jumped by a factor of six following the tax increase (Hrywna, Delnevo, and Staniewska 2004).
The interaction of the state tax rate with the share of the state that uses the Internet is negative and very significant. Clearly the data indicate that the sensitivity of purchases to prices has increased substantially. The point estimates suggest that growth of the Internet in the sample from zero to its average value in 2000 corresponds with a near-doubling of the elasticity, from -1.28 to -2.09.
As a check on the robustness of the result using the CPS measure of Internet usage, the next several columns try alternatives. Column (2) uses the alternative Forrester measure of Internet usage in the state. Again the results show a large and significant increase in the elasticity of taxable sales in states as the usage of the Internet rises. The magnitudes are close. Here the elasticity rises from -1.35 to -1.97 with the growth of the Internet in the sample. Column (3) first differences the data, using the CPS measure of Internet use. The results again show the rising price sensitivity of cigarette purchases in places where the Internet is growing fastest (controlling for state and year dummies), though the overall magnitude is smaller. Here the growth of the internet corresponds with a change in the elasticity from -1.08 to -1.58. Columns (4) and (5) deal with the issue of imputed values. Column (4) restricts the sample to only those years where Internet usage is positive (i.e., no imputed zeros) by looking only at the years after 1995. Column (5) uses only years in which the CPS actually has observations (i.e., no imputed Internet usage between survey years). Both procedures yield very similar results.
Finally, in column (6), we include lags and leads of the tax rates to determine if the results are simply the result of short-run shifting of purchases. We find that the lagged and leading tax variables have a very small impact. All that matters are the contemporaneous values. Short-run timing shifts are, on average, relatively small (perhaps because the tax changes are not anticipated well) and the baseline effects are almost the same as before.
These baseline specifications, then, suggest that as Internet use grew in states, the sensitivity of taxable Internet sales in those states grew as well and that the elasticity has risen greatly. We next consider whether this evidence is consistent with alternative explanations or is likely to be tied to a rise in cigarette smuggling due to the Internet. Although it did not estimate a price elasticity, a recent medical study of 3500 smokers in New Jersey in the period surrounding a large increase in cigarette taxes points in the same direction as our findings in that it documented that the share of people reporting they had bought cigarettes online jumped by a factor of six following the tax increase (Hrywna, Delnevo, and Staniewska 2004).
