For years, the youth vaping debate has been framed around a simple objective: reduce vaping. On the surface, that sounds entirely reasonable. Nobody wants young people who have never smoked taking up nicotine unnecessarily, and sensible policies should protect them. But there is a problem with judging policy by one number alone: what if a policy reduces vaping but increases cigarette smoking? That question matters because cigarettes and e-cigarettes are not remotely equivalent in risk, and a new working paper by Christopher Lowenstein, Jamie Hartmann-Boyce and Michael Pesko goes directly to that issue.
Their paper, posted on SSRN on 23 September 2026, brings together evidence from 11 natural experiments examining youth e-cigarette policies, including minimum legal sales-age laws, e-cigarette taxes and flavour restrictions. The authors developed a way of placing those very different policies onto a common scale by looking at how much each one changed e-cigarette use before asking what happened to youth cigarette smoking. Their overall conclusion is difficult to ignore: the evidence supports the conclusion that greater e-cigarette availability reduces youth cigarette smoking.
That does not mean vaping should be encouraged among young people, it does not mean every teenager who vapes would otherwise smoke, and it does not prove that vaping is protective in every setting. But it does raise a rather uncomfortable question for policies designed simply to drive vaping prevalence down: what happens next?
This is the question I think we too often fail to ask. A flavour ban is announced, and youth vaping falls, so success is declared. A tax is introduced, and e-cigarette use falls, so success is declared. Access becomes more difficult and fewer young people vape, so success is declared. But if some of those young people subsequently smoke cigarettes instead, have we really succeeded?
The answer depends entirely on what the policy goal is. If the goal is simply “less vaping”, then perhaps yes. If the goal is less disease and fewer premature deaths, then the answer becomes much more complicated.
That is why this paper matters. The authors are not simply comparing one flavour ban with another flavour ban or one tax with another tax. They are trying to solve a methodological problem that has made this literature difficult to interpret. Different regulations have different strengths. A minimum-age law cannot be compared directly with a tax increase of a few dollars, and neither can easily be compared with banning flavoured products. They are different interventions operating through different mechanisms. Lowenstein and colleagues instead standardise the policies according to their estimated effect on vaping and then examine the downstream effect on smoking. In plain English, when policies succeed in making vaping less available or less common, what happens to cigarettes? Across the 11 natural experiments, their synthesis points toward substitution between the two products.
That deserves considerably more attention than it receives, particularly because of the long-running “gateway” debate. For years, observational studies have shown that young people who vape are more likely subsequently to report smoking cigarettes. That association has frequently been interpreted as evidence that vaping causes smoking. But association is not direction. Young people who are inclined to experiment with one nicotine product may also be more inclined to experiment with another. Risk-taking, peer groups, family circumstances, mental health, substance use and other shared characteristics may contribute to both behaviours. More importantly, observational cohort studies asking whether a teenager who vapes today is more likely to smoke tomorrow answer a different question from natural experiments asking what happens to smoking when vaping becomes harder to obtain.
The latter can tell us something particularly important for policy. If restricting e-cigarettes is followed by an increase in cigarette smoking relative to what otherwise would have happened, then the simple gateway story becomes much harder to maintain. At minimum, it suggests that substitution is also occurring.
That distinction matters enormously. Imagine two products competing for some of the same consumers. Make one less accessible, more expensive or less appealing, and demand for the other may increase. Nobody finds that surprising when discussing petrol cars and electric cars, butter and margarine, or beer and wine. Yet nicotine policy sometimes seems to operate as though cigarettes and vaping exist in separate universes. They do not. People make choices between products, and those choices respond to price, availability, convenience, satisfaction, flavour, regulation and perceived risk.
The critical issue is that the products being substituted are not equally harmful. If a teenager who would otherwise have vaped instead chooses not to use nicotine at all, that is clearly a desirable outcome. If the same policy instead causes another teenager who would have vaped to smoke cigarettes, that is not. Both responses can occur simultaneously. The job of policy is therefore not merely to count how many people stopped vaping. It is to understand where they went.
That is exactly the sort of question Australia should be asking as well. We have built a vaping policy around restricting access, reducing product choice, limiting flavours and pushing legal supply into a tightly controlled system. Much of the political discussion then asks whether vaping prevalence falls. But that is not enough. What happens to smoking? What happens to illicit purchasing? What happens to dual use? What happens to people who would have switched completely from smoking? What happens when the legal alternative becomes less attractive while cigarettes remain widely available? Those questions matter because the purpose of tobacco policy should not be to win a statistical contest over which nicotine product has the lowest prevalence. It should be to reduce harm.
There is also an important lesson here about the difference between intentions and outcomes. A government may sincerely introduce a flavour ban to reduce youth vaping. That tells us the intention. It tells us nothing about the final health outcome. Policies interact with real people and real markets. Consumers substitute, illegal suppliers respond, manufacturers adapt, prices move, products change and people obtain nicotine from somewhere else. The important question is not what regulators hoped would happen. It is what actually happened.
That is why natural experiments can be so valuable. They use real-world changes in policy and compare what follows. They are not perfect. This new paper is still a working paper and has not yet gone through final journal peer review. It is also synthesising observational natural experiments rather than randomised controlled trials, so causal interpretation still depends on the assumptions and quality of the underlying studies. The policies, populations and periods involved are heterogeneous, which is the very reason the authors developed their standardisation method. Those limitations should be taken seriously, but they should not be used as an excuse to ignore an inconvenient pattern.
If multiple natural experiments point in the direction that restricting vaping can increase cigarette smoking, policymakers need to grapple with that possibility. It is not enough to say cigarettes and vaping are both tobacco products and therefore reducing either must be beneficial. Health outcomes do not work that way. A cigarette involves combustion. A vape does not. That difference matters, and once you accept a hierarchy of risk, substitution becomes a central public-health issue.
This is where I think much of the vaping debate has lost sight of the actual objective. We have become very good at asking whether youth vaping has fallen, but much less interested in asking whether youth smoking rose, fell or stayed the same at the same time. Yet if the most dangerous product is cigarettes, surely smoking should be one of the first outcomes we examine.
This paper effectively asks us to change the order of the questions. Not simply, “Did the policy reduce vaping?” but, “Did the policy reduce vaping, and what happened to smoking as a consequence?” That extra sentence changes everything. It means a policy cannot automatically be called successful because one behavioural indicator moved in the desired direction.
We have seen the same mistake elsewhere. Legal cigarette sales fall, therefore, smoking must have fallen. Vape sales fall, therefore, nicotine use must have fallen. Seizures increase, therefore illicit supply must have fallen. None of those conclusions necessarily follows. You have to look at where demand went. That principle applies just as strongly to young people.
And it also exposes a deeper problem with a zero-nicotine policy objective. If the objective is zero youth nicotine use, then vaping, nicotine pouches and smoking all count as failures. But if the objective is minimising future disease, then the type of product matters enormously. Those are different goals. They can point toward different policies. Pretending otherwise does not remove the trade-off. It merely hides it.
I do not read this paper as an argument for youth vaping. I read it as an argument for something much more basic: measure the outcome that actually matters. If a vaping restriction prevents a young person from ever using nicotine, good. If it moves somebody from vaping to cigarettes, not good. If it causes no meaningful change because consumers move into an illicit market, that matters too. Public health needs to be able to hold all three possibilities in its head at the same time. That requires moving beyond slogans and looking at complete behavioural responses.
The authors’ conclusion that e-cigarette availability appears to reduce youth cigarette smoking will undoubtedly be uncomfortable for some people. But uncomfortable evidence is precisely the evidence that deserves the closest attention. Policy should never be designed around ensuring that the evidence agrees with our preferred intervention. The intervention should change when the evidence tells us it is producing the wrong outcome.
That brings me back to the question I think should accompany every future announcement of a successful vaping restriction: you reduced vaping. Fine. Now tell me what happened to smoking.
Too complex for bumlet n ilk