Great and comprehensive piece. I’m personally very enthusiastic about this intervention, based onCharity Entrepreneurship’s report on tobacco taxation, which found that it’s extremely cost-effective (i.e. maybe USD 27-37 per DALY, which is >GiveWell top charities in expectation), and also on CEARCH’s own research that global health policy interventions tend to be enormously cost-effective.
Important points to add.
(1) I would push back on modelling such mere speeding up as involving simply a set number of years (year of introduction san intervention—year of introduction with intervention) in which the intervention counterfactually applies. It’s important to realize that future tax increases apply on top of the intervention—in this sense, it’s better to think of the tax as a permanent level increase, which of course brings long-term benefits.
(2) These benefits are subject to various discount rates, but generally (a) tobacco taxes are sticky (i.e. aren’t reversed), though inflation can be an issue, and (b) as you say, DALY burdens are growing due to population growth outweighing the secular decline in per capita tobacco consumption.
(3) Economic benefits are around 10% of the health benefits for tobacco, last I did an analysis on this, so agreed that the lack of incorporate in the BOTEC doesn’t alter conclusions overmuch.
(4) The main issue not talked about in this report but which you’ll find is most politically salient is the issue of regressivity—policymakers are afraid the poor are disproportionately hurt, which is a legitimate concern. That said (a) low SES populations (and youths) tend to be disproportionately price sensitive, so their consumption falls more than average (and hence they’re hurt less than they otherwise would be), and (b) you can design income-targeting lump sump compensations.
(5) Caleb Parikh and Joel Burke can speak to this better than me, but my understanding is that the Ministries of Finances tend to be the obstacle in lobbying attempts—not the Ministries of Health, which tend to be supportive. To address the former’s concern (which will be revenue-related), it’s important to point out that actually tax revenues rise with tobacco excise rates, at least over the short and medium term, due to the elasticities involved.
(6) From a consumer welfare perspective (i.e. the pleasure from smoking argument), Gruber had an interesting paper showing that, far from reducing consumer welfare, excise taxes makes the smoking population – whether actual smokers, former smokers or even potential smokers – happier, by helping them overcome their time-inconsistent preferences (i.e. helping them quit something that gives them pleasure in the short-term but which reduces their overall life satisfaction in the long-term). And this makes sense—that’s literally what addiction is.
(7) For the issue of freedom of choice - under plausible moral weights, the health benefits dominate the autonomy considerations up until the point where you transition from taxes to de jure bans; not a particularly significant issue
(8) Smuggling/black market considerations are unlikely to be an issue. While there are theoretical concerns about higher taxes causing increased black market activity, such worries are not justified by the empirical evidence. As Schwartz and Zhang (2016) find, the international experience has been that raising tobacco prices either (a) fails to raise contraband tobacco activity at all, or (b) does so only temporarily, or (c) causes a sufficiently small increase in black activity such that cigarette consumption still falls.
(9) Critically, this shallow report may be too bearish on the chances of tobacco tax advocacy. CE’s case study of 159 case studies suggests about a 27% chance of success, and while I would discount this to some extent due to selection bias (i.e. taxes being pushed in countries where it’s more likely to succeed), this would still suggest a >10% chance of success.
All in all, great work, and I am keen to seen more direct work in this area.
It looks like this estimate comes from the proportion of countries the Bloomberg consortium and World Bank worked in that passed various policies over a decade without adjusting for the counterfactual chance of policy changes without their work.
I’m curious if CE had any luck trying to estimate the counterfactual (Eg by looking at other countries, trends before BB, or diving deep on individual case studies)?
Fwiw when I looked at this a few years ago (at GiveWell, not OP) I couldn’t find any evidence of a difference in policy change by comparing countries BB worked in more vs less.
I don’t think this is good evidence against impact (policy spillovers, selection, BB’s global work, and no data before BB make it difficult to do this comparison) but it made me think unfortunately we can’t learn much about the counterfactual from the high level quantitative comparison and made me want to fall back on going deep on qualitative case studies instead.
Great and comprehensive piece. I’m personally very enthusiastic about this intervention, based onCharity Entrepreneurship’s report on tobacco taxation, which found that it’s extremely cost-effective (i.e. maybe USD 27-37 per DALY, which is >GiveWell top charities in expectation), and also on CEARCH’s own research that global health policy interventions tend to be enormously cost-effective.
Important points to add.
(1) I would push back on modelling such mere speeding up as involving simply a set number of years (year of introduction san intervention—year of introduction with intervention) in which the intervention counterfactually applies. It’s important to realize that future tax increases apply on top of the intervention—in this sense, it’s better to think of the tax as a permanent level increase, which of course brings long-term benefits.
(2) These benefits are subject to various discount rates, but generally (a) tobacco taxes are sticky (i.e. aren’t reversed), though inflation can be an issue, and (b) as you say, DALY burdens are growing due to population growth outweighing the secular decline in per capita tobacco consumption.
(3) Economic benefits are around 10% of the health benefits for tobacco, last I did an analysis on this, so agreed that the lack of incorporate in the BOTEC doesn’t alter conclusions overmuch.
(4) The main issue not talked about in this report but which you’ll find is most politically salient is the issue of regressivity—policymakers are afraid the poor are disproportionately hurt, which is a legitimate concern. That said (a) low SES populations (and youths) tend to be disproportionately price sensitive, so their consumption falls more than average (and hence they’re hurt less than they otherwise would be), and (b) you can design income-targeting lump sump compensations.
(5) Caleb Parikh and Joel Burke can speak to this better than me, but my understanding is that the Ministries of Finances tend to be the obstacle in lobbying attempts—not the Ministries of Health, which tend to be supportive. To address the former’s concern (which will be revenue-related), it’s important to point out that actually tax revenues rise with tobacco excise rates, at least over the short and medium term, due to the elasticities involved.
(6) From a consumer welfare perspective (i.e. the pleasure from smoking argument), Gruber had an interesting paper showing that, far from reducing consumer welfare, excise taxes makes the smoking population – whether actual smokers, former smokers or even potential smokers – happier, by helping them overcome their time-inconsistent preferences (i.e. helping them quit something that gives them pleasure in the short-term but which reduces their overall life satisfaction in the
long-term). And this makes sense—that’s literally what addiction is.
(7) For the issue of freedom of choice - under plausible moral weights, the health benefits dominate the autonomy considerations up until the point where you transition from taxes to de jure bans; not a particularly significant issue
(8) Smuggling/black market considerations are unlikely to be an issue. While there are theoretical concerns about higher taxes causing increased black market activity, such worries are not justified by the empirical evidence. As Schwartz and Zhang (2016) find, the international experience has been that raising tobacco prices either (a) fails to raise contraband tobacco activity at all, or (b) does so only temporarily, or (c) causes a sufficiently small increase in black activity such that cigarette consumption still falls.
(9) Critically, this shallow report may be too bearish on the chances of tobacco tax advocacy. CE’s case study of 159 case studies suggests about a 27% chance of success, and while I would discount this to some extent due to selection bias (i.e. taxes being pushed in countries where it’s more likely to succeed), this would still suggest a >10% chance of success.
All in all, great work, and I am keen to seen more direct work in this area.
Interesting thoughts Joel. Is the analysis in (9) public / could you point me towards it?
(I work at open phil but only made a tiny contribution to this report; I’m just curious)
You can find CE’s Research Report here: https://3394c0c6-1f1a-4f86-a2db-df07ca1e24b2.filesusr.com/ugd/26c75f_2081c09f8f20405e89105ac88c01ec6d.pdf
Thanks, this looks like a helpful report!
It looks like this estimate comes from the proportion of countries the Bloomberg consortium and World Bank worked in that passed various policies over a decade without adjusting for the counterfactual chance of policy changes without their work.
I’m curious if CE had any luck trying to estimate the counterfactual (Eg by looking at other countries, trends before BB, or diving deep on individual case studies)?
Fwiw when I looked at this a few years ago (at GiveWell, not OP) I couldn’t find any evidence of a difference in policy change by comparing countries BB worked in more vs less.
I don’t think this is good evidence against impact (policy spillovers, selection, BB’s global work, and no data before BB make it difficult to do this comparison) but it made me think unfortunately we can’t learn much about the counterfactual from the high level quantitative comparison and made me want to fall back on going deep on qualitative case studies instead.