Assessing the demand for cigarettes and e-cigarettes among US adult e-cigarette users: a cross-sectional study using a demand system approach Authors: Lei Xu, Zezhong Zhang, Ce Shang Published: 2026 ABSTRACT AbstractIntroductionThe rapid growth of electronic cigarette (EC) use in the USA raises questions about their role in harm reduction and interaction with cigarettes. Understanding how adults who use ECs respond to price and expenditure changes is essential for effective tax policy. This study estimates own-price, cross-price and expenditure elasticities for cigarettes and ECs among US adults who use ECs using the Quadratic Almost Ideal Demand System (QUAIDS).MethodsWe analysed a nationally representative sample of 771 US adults aged 18 and older who use ECs, including individuals who use both cigarettes and ECs (dual use) and those who use ECs exclusively. Demand for cigarettes and ECs was measured using budget shares from monthly expenditures and unit prices. A QUAIDS framework jointly estimated expenditure, own-price and cross-price elasticities. To address endogeneity, state-level cigarette and EC taxes served as instruments for prices and the tobacco expenditures of other respondents were used to instrument expenditure.ResultsAmong adults who use ECs, a 10% increase in tobacco expenditure is associated with a 17%–30% higher cigarette consumption and a 5%–8% higher EC consumption. Similarly, a 10% increase in cigarette prices is associated with a 13%–19% lower cigarette consumption while a 10% increase in EC prices is associated with a 7%–10% lower EC consumption. Cross-price elasticities are generally small and not statistically significant, indicating limited evidence that price changes in one product affect consumption of the other.ConclusionsRaising cigarette taxes may be associated with reduced smoking and increased EC use, while higher EC taxes are associated with lower EC consumption without raising smoking. However, additional income is more likely directed to cigarettes, suggesting coordinated taxes on both products may be needed. Future research should assess whether EC taxes raise smoking and whether cigarette taxes promote complete switching to ECs.Keywords: Public Health, Cross-Sectional Studies, economics, methods WHAT IS ALREADY KNOWN ON THIS TOPIC Cigarettes and electronic cigarettes (ECs) are often used together, but limited evidence exists on how prices and taxes influence demand for each product, particularly among adults who use ECs. Most prior studies have analysed these markets separately, leaving gaps in understanding the substitution or complementarity between them. Cigarettes and electronic cigarettes (ECs) are often used together, but limited evidence exists on how prices and taxes influence demand for each product, particularly among adults who use ECs. Most prior studies have analysed these markets separately, leaving gaps in understanding the substitution or complementarity between them. WHAT THIS STUDY ADDS Using a Quadratic Almost Ideal Demand System model, this study provides US evidence on own- and cross-price elasticities for cigarettes and ECs. The results indicate limited evidence of substitution between cigarettes and ECs, as cross-price elasticities are generally small and not statistically significant. Overall, the findings do not support a consistent pattern of substitution or complementarity between cigarettes and ECs. Using a Quadratic Almost Ideal Demand System model, this study provides US evidence on own- and cross-price elasticities for cigarettes and ECs. The results indicate limited evidence of substitution between cigarettes and ECs, as cross-price elasticities are generally small and not statistically significant. Overall, the findings do not support a consistent pattern of substitution or complementarity between cigarettes and ECs. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These results provide limited evidence that tax-induced price changes in cigarettes or ECs substantially affect consumption of the other product. Cross-price effects are generally small and not statistically significant, suggesting minimal substitution or complementarity between cigarettes and ECs. While some estimates indicate potential associations, these are imprecise and should be interpreted with caution. The findings inform taxation and regulatory policies by emphasising the primary importance of own-price effects, while highlighting the need for further research to better understand cross-product responses in nicotine markets. These results provide limited evidence that tax-induced price changes in cigarettes or ECs substantially affect consumption of the other product. Cross-price effects are generally small and not statistically significant, suggesting minimal substitution or complementarity between cigarettes and ECs. While some estimates indicate potential associations, these are imprecise and should be interpreted with caution. The findings inform taxation and regulatory policies by emphasising the primary importance of own-price effects, while highlighting the need for further research to better understand cross-product responses in nicotine markets. Introduction 5.9% of US middle and high school students in 2024 and 6% of adults in 2022 reported electronic cigarette (EC) use in the past 30 days.1 2 The rising prevalence of EC use in the past two decades has generated significant public health and policy interest in understanding consumer demand for ECs and the choices between cigarettes and ECs—a harm reduction alternative to cigarettes.3,6 However, ECs are not harmless because they attract nonsmokers, particularly youth, and may perpetuate nicotine addiction.6 7 Policymakers face a dual challenge—reducing the appeals of ECs among young people while leveraging their potential in smoking cessation.8 9 Excise taxes on tobacco have been among the most effective policies in curbing tobacco use.10 As of June 2025, 36 states and the District of Columbia (DC) in the USA have imposed excise taxes on ECs.11 Growing evidence further shows that these EC taxes have achieved the desired outcome of decreasing EC use among young people.12 However, increasing EC taxes may lead to unintended consequences of increasing smoking among various populations.13 Meanwhile, states continue to increase cigarette excise taxes, which may provide incentives for smokers to transition to ECs. Therefore, there are pricing factors that either encourage or discourage adults who use combustible tobacco products from choosing ECs over cigarettes. How adults make joint choices between ECs and cigarettes in response to prices and taxes remains inconclusive.14 Another key factor in evidence-based policy-making is to understand consumers’ decision making of purchasing tobacco in response to tobacco expenditures in addition to product prices. A comprehensive set of demand elasticity measures that simultaneously consider expenditure (eg, how tobacco budget constraints impact EC and cigarette use), own-price (eg, how EC taxes and prices impact EC use) and cross-price elasticities (eg, how cigarette taxes and prices impact EC use)—is essential in evaluating the decision-making of tobacco purchases.15 16 Previous research has established that cigarette consumption is price sensitive and inelastic (increasing prices lead to less use and more tax revenues), as well as responding to expenditures (increasing income and expenditures lead to more consumption). These findings lead to widespread adoption of cigarette excise tax policies globally and the recommendation of raising cigarette taxes and prices to outpace income growth.1213 15,19 In comparison, while EC taxes are effective in reducing EC use, there is less consensus regarding whether the price elasticity of ECs is elastic and the degree to which these products act as substitutes or complements to cigarettes (ie, cross-price elasticities between cigarettes and ECs). Moreover, studies examining the expenditure elasticity of ECs are lacking despite their relevance in understanding how budget changes on tobacco products affect the consumption of the subcategories of nicotine and tobacco products (eg, cigarettes vs ECs). There is a research gap in estimating the demand system of ECs and cigarettes, simultaneously considering expenditure, own-price and cross-price elasticities. This study employs a Quadratic Almost Ideal Demand System (QUAIDS) to simultaneously estimate expenditure, own-price and cross-price elasticities for ECs and cigarettes among adults who use ECs.20 21 To address endogeneity (ie, prices, expenditures and income are jointly determined), we use an instrumental variable (IV) approach. This paper contributes to the literature by offering the first US study using QUAIDS to jointly estimate expenditure, own-price and cross-price elasticities for ECs and cigarettes among adults who use ECs, expanding on the prior literature that exclusively focuses on price elasticities. Methods Data and setting We recruit a nationally representative sample of 808 US adults who use ECs in May 2023 through Ipsos Knowledge Panel, a probability-based online panel designed to be representative of the US adult population. The Ipsos Knowledge Panel uses address-based sampling. Eligible respondents are adults aged 18 years and older who reported using ECs every day or some days in the past 30 days. Among the 808 adults who use ECs, 29 respondents are dropped due to the lack of EC price data, and 8 respondents are dropped due to the report of zero or missing expenditures on both cigarette and EC consumption. The final analytical sample consists of 771 adults who use ECs drawn from 39 US states. online supplemental appendix III, table A5 reports the distribution of respondents across US states to provide additional detail on the geographic representation of the analytic sample. The survey also collects detailed demographic and socioeconomic characteristics, including age, education, race/ethnicity, sex, home ownership, housing type, household income, marital status and region and state of residence. Measures Outcomes The primary outcomes and dependent variables of this study are expenditure or budget shares for cigarettes and ECs, defined as the proportion of monthly household tobacco expenditure allocated to each product. In the survey, we ask respondents to report how much they spend per week or per month on ECs and cigarettes, respectively. These variables are used to calculate the shares of tobacco expenditure spent on cigarettes and ECs (total tobacco expenditure=100%), respectively. These outcomes also allow us to estimate the price and expenditure elasticities of demand for cigarette and EC consumption in the model. Explanatory variables The explanatory variables are state-level cigarette and EC prices and taxes, as well as expenditures on ECs and cigarettes. State-level cigarette price and tax data for a pack of 20 cigarettes in May 2023 are sourced from the 2023 Tax Burden on Tobacco (TBOT).22 Specifically, while TBOT reports cigarette prices as of 1 November of each year, the inflation between May and November 2023 was relatively stable (consumer price index=304 in May 2023 and 307 in November 2023, 1982–1984=100) and there are no tax increases or implementations between May and November 2023 (except in the state of New York and D.C.). Therefore, we use November cigarette prices as a proxy for the May cigarette prices. State-level EC prices per unit (excluding Hawaii and Alaska) were sourced from Nielsen ScanTrack’s Byzza portal via the ‘Tobacco Alternatives Vapor (Category Totals)’ dataset, which reports both unit sales and average prices at the retail channel level (eg, convenience, grocery and retailer). For each state, we extract the total average price per unit at the UPC level for disposables, cartridge packs and pod systems in each channel during the 4-week period ending May 2023. Next, we compute a sales-weighted average across channels to derive a single state-level EC price per unit. This aggregated measure captures the blended retail price signal facing adults who use ECs across different store categories and provides a pragmatic, policy-relevant indicator of market prices. State-level average EC tax rates for disposables and rechargeable ECs in May 2023, standardised in dollars per millilitre, are obtained from a recent study.23 Control variables We construct a comprehensive set of control variables using self-reports, which include age in years, sex (female, male as reference), race (Black, Other races, White as reference), ethnicity (Hispanic, Non-Hispanic as reference), marital status (married, never married or other as reference), region (South, Midwest, West, Northeast as reference), house type (single-family house, other types as reference), annual household income (US$50 000–US$100 000, US$100 000+, less than US$50 000 as reference), education (Bachelor and above, less than Bachelor as reference), and employment status (full time, part time or unemployed as reference). Analytical methods The QUAIDS model To estimate the demand for cigarettes and ECs, we employ the QUAIDS model, a flexible demand framework developed by Banks, Blundell and Lewbel.20 The QUAIDS model builds on the standard Almost Ideal Demand Systems by incorporating a quadratic term, which allows for non-linear relationships between expenditure share and total expenditure. This feature is crucial for capturing the non-linear expenditure patterns observed in our data, as illustrated in online supplemental appendix I, figure A1. The QUAIDS model is widely used in the literature to estimate expenditure and price elasticities24,26 and enables us to better approximate demand behaviour as consumption patterns and budget allocations shift with changes in prices and expenditures. The mathematical specification of the QUAIDS model, including budget share equations, the price index and elasticity formulas, is provided in online supplemental appendix II. The QUAIDS model assumes ECs and cigarettes are weakly separable from other goods. That is, consumption choices between ECs and cigarettes are independent of other goods and can be analysed conditional on the total expenditure on ECs and cigarettes. Consequently, the demand for ECs and cigarettes depends on their prices and total expenditure, but is not related to other goods or the prices of other goods. The model also adheres to standard theoretical restrictions: adding up, homogeneity and symmetry, as illustrated in online supplemental appendix II. Standard demand system restrictions are imposed and/or tested following established practice. These restrictions are evaluated using Wald tests, where statistically significant test statistics indicate rejection of the null hypothesis that the restrictions hold. In contrast, failure to reject the null suggests that the data are consistent with the theoretical restrictions. Although weak separability between tobacco (defined as ECs and cigarettes) and other goods is a strong assumption, it is necessary to simplify the specification and is commonly adopted in demand studies.27 28 To test its validity, we implement a two-stage budgeting system (see figure 1 and online supplemental appendix IV), which first allocates total expenditure between tobacco (ECs and cigarettes) and all other goods, then allocates tobacco spending between cigarettes and ECs. If separability holds, the unconditional own-price elasticities from this procedure should closely match our single-stage QUAIDS estimates. The two-stage budgeting process. Addressing endogeneity A key methodological challenge in estimating cigarette and EC demand is potential endogeneity in explanatory variables. First, endogeneity in total expenditure arises because unobserved household preferences or other household characteristics may simultaneously influence both expenditure and consumption behaviour. For example, households with a stronger tobacco preference allocate larger budget shares, thereby biasing the estimated relationship between expenditure and demand. Price endogeneity can also arise from correlations with unobserved factors or simultaneity, since prices are jointly shaped by supply and demand. To estimate the impacts of expenditures and prices on consumption, we use an IV approach, which has been widely adopted in the literature to address endogeneity issues.29 Following previous literature, we utilise the combination of federal and state cigarette tax per pack as an instrument for cigarette prices and state-specific EC taxes per ml as an instrument for EC prices.15 30 Both cigarette and EC taxes vary across states and are independent of individual household preferences or demand fluctuations. For the IV approach, instrument strength is assessed using first-stage F-statistics. Following conventional guidelines, F-statistics exceeding 10 are interpreted as evidence of sufficiently strong instruments, whereas lower values may indicate weak instruments. These diagnostics provide a basis for evaluating the reliability of IV-based estimates. For total expenditure, we construct an instrument using an average linkage clustering method.31 32 In the first step, households are grouped into clusters by demographic and household characteristics (including sex, race, ethnicity, marital status, household size), household income as a measure of socioeconomic status and the frequency of cigarette smoking. These variables are theoretically and empirically grounded in consumer demand models and capture heterogeneity in budget constraints and tobacco preferences.21 33 34 The number of clusters is selected using a data-driven approach that is based on the higher Je(2)/Je(1) ratio and a relatively lower pseudo T-squared statistic (see online supplemental appendix III, table A3). In the next step, the IV is constructed as the leave-one-out cluster mean of total tobacco expenditure, calculated as the average spending on cigarettes and ECs among other respondents within the same cluster, excluding the respondent itself. This approach captures the spending capacity of similar households without reflecting the respondent’s own preferences, supporting the exclusion restriction. Each of these instruments (cluster mean of total tobacco expenditure, cigarette taxes and EC taxes) is a standard approach and meets the key criteria for relevance and exogeneity, enhancing the reliability of the QUAIDS model estimates in capturing the demand dynamics of cigarette and EC consumption while mitigating bias from endogeneity. We also conduct the first-stage estimation to ensure that these instruments are not weak (see online supplemental appendix III, table A2). Other method details All analyses are conducted using Stata V.18. For the demand system estimation, we use the command aidsills, which implements the AIDS and its extensions, allowing for IV estimation and quadratic approximation of budget shares. This study’s analytic plan is not preregistered and therefore should be considered exploratory. All hypotheses, model specifications and identification strategies are developed post hoc to address the research questions. Finally, while analyses reported in the study are unweighted, we also conduct weighted analyses for specifications with instruments for both prices and expenditures. Sample size justification The final analytical sample of 771 respondents is sufficient for estimating the QUAIDS model, which involves the simultaneous estimation of multiple equations with demographic controls and IVs. The sample size is comparable to, and in many cases larger than, those used in prior demand system studies, which often rely on fewer observations for similar model estimation.26 In demand system models, statistical power depends not only on sample size but also on variation in key explanatory variables. In this study, there is substantial cross-sectional variation in cigarette and EC prices, taxes and expenditures across states and individuals, which provides strong identifying variation for elasticity estimation. Given the number of observations and the variation in key variables (eg, prices, expenditures and demographic characteristics), the sample provides adequate statistical power to estimate expenditure, own-price and cross-price elasticities with acceptable precision. Patient and public involvement Patients and members of the public were involved in the development of the survey instrument used in this study. Specifically, adult EC users participated in cognitive testing during the survey design stage to evaluate the clarity, relevance and interpretability of the questions. Their feedback informed revisions to the survey prior to data collection. Patients and the public were not involved in the statistical analysis, interpretation of results or reporting of this study. There are no plans to directly disseminate findings to study participants, as the study relies on de-identified survey and secondary data. Results Table 1 shows summary statistics of all variables. For outcome variables, adults who use ECs allocate 80% of the monthly tobacco expenditure to ECs and 20% to cigarettes, with a combined monthly expenditure of US$135 across both products. For explanatory variables, the average cigarette price was about US$9 per pack of 20, whereas the average EC price was US$6 per unit. For smoking and EC use frequencies and history, among the 771 adults who use ECs, 541 (70.17%) report daily EC use, 230 (29.83%) report someday EC use. Additionally, 161 (20.88%) report daily cigarette smoking, 114 (14.79%) report someday smoking, and 422 (54.73%) report no current cigarette use. Moreover, 90.40% report ever smoking, 35.67% report current use of both cigarettes and ECs (dual use), 54.73% report exclusive EC use with a history of smoking; and 9.60% report exclusive EC use with no history of smoking. Table 1Summary statistics (N=771)VariableMean (percentage)SDOutcome-related measures Cigarette expenditure share0.200.32 EC expenditure share0.800.32Explanatory variables Cigarette price (US$)9.091.57 EC price (US$)6.190.73 Total expenditure (US$/month)134.76303.53 Cigarette tax (US$)2.871.07 EC tax (US$)0.500.81Control variables Age45.2314.60 Female0.590.49 Hispanic0.110.31 Race White85.15% Black7.13% Other7.72% Marital status Married40.34% Never married35.80% Other23.86% Education Bachelor degree or above28.79% Some college or below71.21% House type Single-family house63.16% Other36.84% Household income <50k45.14% ≥5 k, <100k30.60% ≥100k24.26% Region Northeast13.62% Midwest27.89% South42.02% West16.47% Employment Full time52.53% Other47.47%Cigarette and EC use prevalence Smoked EC in the past 30 days Daily70.17% Some days29.83% Smoked cigarette in the past 30 days Daily20.88% Some days14.79% Not at all54.73% Non-smoker9.60% Use status Dual use35.67% Exclusive use: EC (with history smoking)54.73% Exclusive use: cigarette (never smoking)9.60%Among the 275 dual users, the mean cigarette expenditure share is 55.40% and the mean EC expenditure share is 44.60%. Among the 422 exclusive EC users (with a history of smoking) and 74 exclusive never-smoker EC users, the cigarette share is 0% by definition, and the EC share is 100%. For states without EC taxes, the value of EC taxes is zero. Conditional on states with positive or non-zero taxes, the average EC taxes were US$0.82.EC, electronic cigarette. Summary statistics (N=771) Among the 275 dual users, the mean cigarette expenditure share is 55.40% and the mean EC expenditure share is 44.60%. Among the 422 exclusive EC users (with a history of smoking) and 74 exclusive never-smoker EC users, the cigarette share is 0% by definition, and the EC share is 100%. For states without EC taxes, the value of EC taxes is zero. Conditional on states with positive or non-zero taxes, the average EC taxes were US$0.82. EC, electronic cigarette. Table 2 summarises the elasticity estimates (expenditure, own-price, and cross-price elasticities for cigarettes and ECs) for adults who use ECs under various model specifications, including baseline models without using IVs and those using IVs for either prices or expenditures or both. Models (1) through (4) are applied to the entire sample of adults who use ECs, whereas column (5) is restricted to individuals who report using both cigarettes and ECs (ie, dual users). Our preferred model specifications are models (4) and (5), where both prices and expenditures impacts were estimated using IVs. Nonetheless, the results are very similar across different model specifications. Table 2Price and expenditure elasticities of demand for cigarettes and ECsQUAIDS models(1)(2)(3)(4)(5)Expenditure elasticities Cigarettes1.72***1.73***2.44***3.08***1.29***(1.58 to 1.85)(1.59 to 1.86)(1.53 to 3.35)(1.71 to 4.44)(0.49 to 2.09) ECs0.84***0.84***0.68***0.54***0.57(0.81 to 0.86)(0.81 to 0.86)(0.48 to 0.87)(0.26 to 0.82)(−0.60 to 1.74)Own-price elasticities Cigarettes−1.61***−1.31***−1.94***−1.75**−0.80*(−2.34 to −0.88)(−2.23 to −0.38)(−2.87 to −1.02)(−3.08 to −0.43)(−1.45 to −0.16) ECs−0.98***−0.91***−0.89***−0.71***−0.28(−1.14 to −0.81)(−1.11 to −0.70)(−1.10 to −0.68)(−1.05 to −0.36)(−1.20 to 0.64)Cross-price elasticities (consumption to price) (Cig, EC)−0.11−0.42−0.50−1.33−0.49(−0.84 to 0.62)(−1.35 to 0.51)(−1.44 to 0.44)(−2.93 to 0.28)(−1.11 to 0.14) (EC, Cig)0.140.070.21**0.17−0.29(−0.03 to 0.30)(−0.14 to 0.28)(0.01 to 0.42)(−0.13 to 0.46)(−1.24 to 0.67) Observation771771771771263 Price IVNoYesNoYesYes Expenditure IVNoNoYesYesYesModels (1)–(4) include the whole sample of EC users, and model (5) only includes respondents who report non-zero expenditures on cigarettes. We use model (5) for sensitivity analysis. 95% CIs are reported in brackets.*p<0.1, **p<0.05, ***p<0.01.EC, electronic cigarette; QUAIDS, Quadratic Almost Ideal Demand System. Price and expenditure elasticities of demand for cigarettes and ECs Models (1)–(4) include the whole sample of EC users, and model (5) only includes respondents who report non-zero expenditures on cigarettes. We use model (5) for sensitivity analysis. 95% CIs are reported in brackets. *p<0.1, **p<0.05, ***p<0.01. EC, electronic cigarette; QUAIDS, Quadratic Almost Ideal Demand System. For the entire sample of adults who use ECs, the estimates suggest that a 10% increase in tobacco expenditures is associated with a 17.2% to 30.8% higher cigarette consumption and a 5.4% to 8.4% higher EC consumption. Own-price elasticities indicate that a 10% increase in prices is associated with a 13.1% to 19.4% lower cigarette consumption and a 7.1% to 9.8% lower EC consumption. These results indicate that cigarette demand is price-elastic among adults who use ECs, whereas EC demand appears relatively price-inelastic. Cross-price elasticities between cigarettes and ECs are mostly non-significant. The only exception is model (3) (expenditure instrument only), where EC demand is positively associated with cigarette prices (0.21, p<0.05), suggesting potential substitution between cigarettes and ECs among adults who use ECs. However, our preferred specification (model 4 with both prices and expenditures instrumented) suggests that there is no statistically significant cross-price relationship between ECs and cigarettes, as the estimated cross-price elasticities are not statistically different from zero. The unconditional price elasticities (online supplemental appendix III, table A4) are nearly identical in magnitude to our main QUAIDS estimates, reinforcing the stability of our findings. Model (5) restricts the sample to individuals who report using both ECs and cigarettes (ie, dual use, defined as positive expenditures on both products). For this group, the expenditure elasticity was 1.29 (p<0.01) for cigarettes and is not statistically significant for ECs (0.57, p>0.1). The price elasticity is estimated to be −0.80 (p<0.1) for cigarettes and is not statistically significant for ECs (−0.28, p>0.1). Moreover, weighted regression results from the preferred IV specifications are consistent with the unweighted analyses and support the same conclusions (online supplemental appendix III, table A1). The analysis also controls for factors influencing the budget share allocated to cigarettes, summarised in table 3. Age is positively associated with cigarette budget share, indicating that older individuals allocate a higher proportion of their tobacco expenditure to cigarettes. On the contrary, owning a single-family home, being married and having full-time employment are associated with reduced cigarette budget share. Table 3Coefficient estimates from QUAIDS models examining the demographic and socioeconomic determinantsDependent variable: cigarette budget share(1)(2)(3)(4)(5)Age0.002**0.002**0.003**0.003**0.004**(0.001)(0.001)(0.001)(0.001)(0.002)Female0.0040.004−0.019−0.040−0.032(0.022)(0.022)(0.030)(0.040)(0.064)Hispanic0.0200.0150.0140.0010.038(0.052)(0.052)(0.062)(0.080)(0.150)Race (base: White) Black0.099**0.097**0.092*0.0810.027(0.044)(0.044)(0.052)(0.068)(0.070)Other−0.017−0.015−0.040−0.060−0.098(0.042)(0.043)(0.052)(0.068)(0.174)Marital status (base: non-married) married−0.050**−0.050**−0.076**−0.101**−0.090(0.023)(0.033)(0.032)(0.044)(0.091)Education (base: some college or below) bachelor and above−0.029−0.029−0.0130.002−0.016(0.024)(0.024)(0.031)(0.041)(0.050)House type (base: other) single-family house−0.064***−0.064***−0.076***−0.087**−0.109(0.023)(0.023)(0.028)(0.037)(0.077)Household Income (base: <50k) ≥50k, <100k−0.018−0.018−0.0020.0140.051(0.025)(0.025)(0.031)(0.042)(0.057)≥100k−0.039−0.041−0.0080.0190.033(0.033)(0.033)(0.044)(0.060)(0.100)Region (base: Northeast) Midwest0.0030.0220.0130.0600.136(0.043)(0.047)(0.052)(0.079)(0.107)South−0.035−0.013−0.045−0.0130.059(0.043)(0.047)(0.052)(0.075)(0.086)West0.0410.0530.0600.1020.118(0.043)(0.044)(0.053)(0.078)(0.119)Employment status (base: part-time/no work) Full time−0.043*−0.042*−0.071**−0.097**−0.105(0.023)(0.023)(0.033)(0.046)(0.088) Constant−0.000−0.036−0.302−0.617**−0.259(0.077)(0.084)(0.206)(0.338)(0.670)*p<0.1, **p<0.05, ***p<0.01.QUAIDS, Quadratic Almost Ideal Demand System. Coefficient estimates from QUAIDS models examining the demographic and socioeconomic determinants p<0.1, **p<0.05, ***p<0.01. QUAIDS, Quadratic Almost Ideal Demand System. Discussions This study contributes new evidence on the demand elasticities of cigarettes and ECs among US adults who use ECs, focusing on how spending on cigarettes and ECs responds to changes in income, prices and taxes. Our findings highlight that, among adults who use ECs in the USA, cigarette use appears more responsive to changes in overall tobacco spending than EC use, and ECs are inelastic to tobacco expenditure. As 90% of respondents are former or current smokers, additional tobacco expenditures are more likely directed to cigarettes than ECs. Consistent with the law of demand, own-price elasticities are statistically significant and negative for both cigarettes and ECs among adults who use ECs, suggesting that higher taxes on ECs and cigarettes are associated with lower consumption of both cigarettes and ECs within this population. It is important to note that these price elasticity estimates apply specifically to adults who use ECs and thus reflect demand responses in this population, rather than elasticities for the general adult population. Nonetheless, the magnitudes and directions of our EC price elasticity estimates are generally consistent with patterns reported in prior studies examining people who use ECs or similar subpopulations.1315 18 35,40 Our cigarette price elasticity estimates are higher than those reported for the overall US adult population, which generally range from −0.3 to −0.5.12 15 17 38 41 42 Earlier studies used data collected before ECs were widely available, when substitution opportunities between cigarettes and ECs were more limited. As ECs have become more prevalent and serve as potential substitutes, consumer sensitivity to cigarette prices may have increased.43 44 Our estimates apply to adults who use ECs and are based on observed real-world purchasing behaviour in contrast to prior studies that rely on hypothetical purchase scenarios; these differences in sample composition and study design may help explain the higher price elasticities observed relative to the general adult population. Our findings also provide insights into the revenue impact of tobacco taxation among adults who use ECs. Recent literature indicates that both cigarette and EC excise taxes are largely passed through to retail prices, suggesting that tobacco tax revenues are determined by price elasticities.15 45 Our findings further suggest that price responsiveness for both ECs and cigarettes is close to proportional, indicating that increases in prices are associated with comparable proportional reductions in consumption. These results suggest that higher taxes on cigarettes and ECs may be associated with lower consumption among adults who use ECs, with implications for tax revenue depending on demand responsiveness. In particular, cigarette demand appears price-elastic, implying that higher cigarette taxes may reduce tax revenue, whereas EC demand appears relatively price-inelastic, suggesting that higher EC taxes may increase tax revenue. Overall, tax revenue effects are likely to differ across products and depend on the underlying demand elasticities. The cross-price elasticity estimates do not provide strong evidence of substitution or complementarity between ECs and cigarettes among adults who use ECs. Across model specifications, cross-price elasticities are generally small and statistically non-significant, suggesting limited responsiveness of EC consumption to cigarette prices and vice versa. While some point estimates indicate a potential positive association between cigarette prices and EC use, these estimates are imprecise and should be interpreted with caution. Overall, the findings do not support a clear asymmetric substitution pattern between ECs and cigarettes in this population. One possible explanation is that individuals who already use ECs may have more stable consumption patterns and are less sensitive to cross-product price changes, particularly when both products are accessible. Although our analysis focuses on a distinct population—adults who use ECs, our findings of limited cross-price responsiveness are broadly consistent with prior studies.1315 19 45,47 For example, a recent study on the relationship between heated tobacco products (HTPs) and cigarettes found that higher cigarette prices were associated with increased HTP sales, whereas higher HTP prices were not significantly associated with cigarette sales.48 Similarly, an experimental study reported that cross-price elasticities between ECs and cigarettes were not statistically significant among individuals who use both products (dual use), though some effects were observed among those who use ECs exclusively.49 Unlike some prior studies documenting unintended consequences of EC taxes, we did not find evidence that higher EC taxes are associated with increased cigarette smoking among adults who use ECs. One potential explanation for discrepancies in the literature is differences in outcome measures. Studies reporting that EC taxes or prices are associated with increased cigarette smoking often rely on survey prevalence data, whereas studies finding no impact or negative impact tend to use sales data.16 Considering our findings also rely on consumption measures rather than the prevalence measure, the combined evidence suggests that any substitution of cigarettes when ECs become more expensive may primarily occur on the changes in initiation or participation, rather than how much they use them. We also discussed earlier that our study population is limited to adults who use ECs, which may explain the difference between our findings with previous literature on unintended consequences of EC taxes. This study has several limitations. First, although tested, the weakly separable assumption may be restrictive. Second, cigarette and EC expenditures are self-reported, which may be subject to recall error or misreporting despite the use of survey items. Third, Nielsen ScanTrack data used to construct state-level EC prices primarily capture mainstream retail channels (eg, convenience, grocery and mass retailers) and may not fully reflect prices in vape shops or online markets, where prices are often lower.50 As these outlets account for a substantial share of EC purchases in the USA, this may introduce measurement error in price variables. In addition, the absence of statistically significant cross-price elasticities limits inference regarding substitution towards EC in response to cigarette tax increases. Furthermore, our sample is limited to adults who use ECs in the past 30 days, so estimates reflect intensive-margin responses (changes in use or spending among current users) rather than extensive-margin outcomes (initiation or cessation). Excluding non-users may lead to more conservative elasticities if marginal users are more price-sensitive, limiting inference for population-wide tax impacts. Finally, although we employ an IV approach to address potential endogeneity, causal interpretation of these estimates relies on strong assumptions, including key assumptions required for valid IV analysis. While first-stage results suggest that the instruments are strongly correlated with endogenous variables, the exclusion restriction cannot be directly tested and may be violated if state-level taxes or cluster-level expenditure patterns are correlated with unobserved factors influencing consumption. In addition, the cross-sectional design limits our ability to account for time-varying unobserved heterogeneity. Therefore, our findings should be interpreted as associations informed by an IV framework rather than definitive causal effects. Future research using longitudinal data or alternative identification strategies is needed to strengthen causal inference. This is the first study to use QUAIDs to estimate demand elasticities for ECs and cigarettes, yielding several policy implications. Higher cigarette taxes are associated with reduced smoking among adults who use ECs, while evidence of increased EC use remains limited. Higher EC taxes are associated with lower EC use and are not significantly associated with increased smoking among adults who use ECs. However, as income or expenditure increases, individuals in this population tend to allocate a greater share of additional spending to cigarettes than to ECs, which may reflect the stronger addictive potential of cigarettes. This raises concerns that higher expenditure may be associated with greater cigarette consumption among adults who use ECs. From a policy perspective, this pattern highlights the importance of considering how cigarette and EC taxes are structured within a broader regulatory framework, rather than relying on cigarette taxation alone. However, our results do not provide direct evidence of cross-product substitution in our preferred model, and further research is needed to assess whether people who use ECs—particularly those who use both cigarettes and ECs—fully transition away from cigarettes or quit nicotine use without relapse. Supplementary material The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Footnotes Funding: This research is funded by the National Cancer Institute. The current research is also sponsored by OSUMC Cancer Control and Prevention Program Pilot Funding. Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: This study was reviewed and determined exempt by the Ohio State University Institutional Review Board, as it used de-identified secondary survey data. Data availability free text: The data that support the findings of this study are available from the corresponding author on reasonable request. The data are not publicly available due to restrictions related to data use agreements and participant confidentiality. Certain data components, including Nielsen ScanTrack data, are subject to third-party licensing agreements and cannot be shared directly by the authors. Patient and public involvement: Patients and/or the public were not involved in the design or conduct or reporting or dissemination plans of this research. Data availability statement Data are available on reasonable request. Data may be obtained from a third party and are not publicly available. References Associated Data Supplementary Materials Data Availability Statement Data are available on reasonable request. Data may be obtained from a third party and are not publicly available.