The Cost of Inaction: Economic Modeling of Delayed Harm Reduction

In 2022, a team of health economists at the University of Michigan published a modeling study that should have been a bombshell. They simulated the effects of different U.S. regulatory scenarios for e-cigarettes over a 50-year horizon, comparing a 'status quo' scenario (moderate regulation, limited harm-reduction messaging) with a 'harm maximization' scenario (aggressive flavor bans, restrictive regulation that treated vaping identically to smoking) and a 'harm minimization' scenario (risk-proportionate regulation, honest communication, incentivized switching). The results: the 'harm maximization' scenario—which closely resembled the policies several U.S. states and the FDA were implementing—was projected to result in 2.5 million additional premature deaths over 50 years compared to the 'harm minimization' scenario. The study passed peer review, was published in a leading journal, and was largely ignored by the policymakers whose decisions it modeled. The cost-of-inaction literature—economic and epidemiological modeling that estimates the mortality and morbidity consequences of delaying harm reduction—is substantial, consistent, and systematically neglected in policy debates that privilege the precautionary principle over the consequences of precaution.

The economic modeling of delayed harm reduction follows a consistent methodology across studies. Researchers construct simulation models that track a synthetic population of smokers over time, modeling transitions between smoking, vaping, dual use, and cessation under different policy scenarios. The models incorporate the best available evidence on the relative risks of different products, the effectiveness of vaping for smoking cessation, the rates of youth initiation under different policy conditions, and the natural history of smoking-related disease. The outputs are estimates of life-years lost or gained, premature deaths, and healthcare costs under each scenario. The models are necessarily simplified representations of complex reality, and their absolute predictions should be treated with appropriate uncertainty. But the relative comparisons between scenarios—which policy approach produces better or worse outcomes—are robust to reasonable variations in the model assumptions, and the direction of the results is consistent across studies conducted by independent research groups in multiple countries.

The most striking finding across the modeling literature is the mortality cost of policies that treat vaping and smoking as equivalent. When vaping products are taxed at the same rate as cigarettes, when their marketing is restricted as severely, when they're subject to the same retail access limits, the models consistently predict that fewer smokers switch, more vapers relapse to smoking, and population smoking rates decline more slowly. The mechanism is straightforward substitution economics: when you eliminate the price, convenience, and information advantages of the safer product, consumers shift toward the more dangerous one. The magnitude of the effect is substantial: a 2023 study estimated that equalizing taxes on cigarettes and e-cigarettes across all U.S. states would result in approximately 200,000 additional smoking-attributable deaths over 30 years compared to a risk-proportionate tax framework. The policies that are intended to protect public health by 'discouraging all nicotine use' are, according to the best available modeling, increasing mortality by discouraging the switch from smoking to vaping.

The modeling also reveals the asymmetric consequences of the two types of policy error. A 'false positive' error—permitting a product that turns out to be more harmful than current evidence suggests—produces harms that are concentrated among the people who use that product, many of whom are former smokers who would otherwise have continued smoking. A 'false negative' error—restricting a product that turns out to be substantially less harmful than smoking—produces harms that are concentrated among the smokers who would have switched but were prevented or discouraged from doing so. The expected mortality cost of the two errors is not symmetric: the false negative (restricting a truly reduced-risk product) causes more deaths than the false positive (permitting a product that's riskier than expected), because the baseline comparator for the restricted product is continued smoking, which carries an enormous mortality risk. The precautionary principle, applied symmetrically to all nicotine products without regard for the differential baseline risk, is not precautionary at all. It's a gamble that the unknown risks of new products exceed the known risks of cigarettes—a gamble that the modeling suggests is a losing bet.

The most provocative finding in the cost-of-inaction literature concerns the temporal pattern of mortality effects. Policies that restrict vaping access produce immediate harms—smokers who would have switched continue to smoke today, and some die of smoking-related disease this year. The benefits of such policies—if the restricted products turn out to be more harmful than current evidence suggests—are deferred by decades, because the diseases that vaping might cause (respiratory illness, cancer) have long latency periods. The result is a temporal asymmetry that makes restrictive policies politically attractive: the costs (increased smoking mortality) are diffusely distributed and difficult to attribute to the policy, while the putative benefits (prevention of hypothetical future vaping-related disease) are easy to claim and impossible to disprove in the short term. The cost-of-inaction modeling suggests that this temporal asymmetry has a body count: every year of delay in implementing evidence-based harm reduction is a year in which smokers who might have switched continue to die from the products they're being discouraged from leaving.

The modeling literature has limitations that should be acknowledged, not ignored. The models are only as good as their input parameters, and several key parameters—the long-term health effects of vaping, the net effect of flavors on youth initiation and adult cessation, the responsiveness of switching behavior to policy changes—are uncertain. The models can't predict black-market responses to restrictive policies, which could either amplify or mitigate the mortality effects. And the models, by design, can't capture the cultural and psychological factors that influence smoking behavior in ways that may not be responsive to the policy levers the models simulate. These limitations mean that modeling results should inform policy rather than dictate it—they provide a quantitative framework for thinking about trade-offs, not a precise prediction of policy outcomes. But they also mean that policymakers who dismiss modeling because it's uncertain are implicitly relying on their own mental models, which are less rigorous, less transparent, and almost certainly more biased than the formal models they're rejecting.

The cost-of-inaction literature delivers an uncomfortable message: the precautionary approach to nicotine regulation, however well-intentioned, is not consequence-free. Every year that harm reduction is delayed, every policy that makes switching harder, every communication that obscures the risk differential between smoking and vaping—these actions have mortality costs that can be modeled, estimated, and compared to the putative benefits of precaution. The costs are not hypothetical. They're measured in smokers who continue to smoke because the safer alternative was made more expensive, less accessible, or less appealing than the product that's killing them. The precautionary principle asks: 'what if we permit these products and they turn out to be harmful?' The cost-of-inaction literature asks a different question: 'what if we restrict these products and they turn out to be substantially less harmful than smoking?' Both questions are valid. The modeling suggests the second question's answer involves a much larger number of preventable deaths.