WEBVTT 1 00:00:04.690 --> 00:00:14.380 Luis Zavala Arciniega: Welcome to the Tobacco Online Policy Seminary, TOPS. Thank you for joining us today. I am Luis Abadar-Sinega, a research associate at the University of Yale. 2 00:00:14.900 --> 00:00:32.760 Luis Zavala Arciniega: TOPS is organized by Michael Pesco at the University of Missouri, CSAN at the Ohio State University, Michael Darnett at Johns Hopkins University, Jamie Harmon Boyce at University of Massachusetts Amherst, and Justin White at Boston University. 3 00:00:33.380 --> 00:00:50.949 Luis Zavala Arciniega: The seminar will be one hour with questions from the moderator and discussant. The audience may post questions and comments in the Q&A panel, and the moderator will draw from these questions and comments in conversation with the presenter. 4 00:00:51.190 --> 00:01:02.119 Luis Zavala Arciniega: Please review the guidelines on tobaccopolicy.org for acceptable questions. Please keep the questions professional and related to the research being discussed. 5 00:01:03.880 --> 00:01:15.499 Luis Zavala Arciniega: Questions that meet the seminar series guidelines will be shared with the presenter afterwards, even if they are not read aloud. Your questions are very much appreciated. 6 00:01:15.940 --> 00:01:27.490 Luis Zavala Arciniega: This presentation is being video recorded, and will be made available along with presentation slides on the TOPS website, tobaccopolicy.org. 7 00:01:28.230 --> 00:01:38.599 Luis Zavala Arciniega: I will turn the presentation over to today's moderator, Jamie Hartman-Boies from the University of Massachusetts Amherst, to introduce our speaker. 8 00:01:39.360 --> 00:01:55.049 Jamie Hartmann-Boyce: Thanks so much! So, today we continue our winter season with a single paper presentation by Ben Shui entitled, The Impact of E-Cigarette Regulations on Birth Outcomes. This presentation was selected via a competitive review process by submission through the TOPS website. 9 00:01:55.340 --> 00:02:04.920 Jamie Hartmann-Boyce: Ben Chui is an assistant professor of economics at the University of New Hampshire. His current research focuses on the intersection of policy evaluation and health inequality. 10 00:02:04.990 --> 00:02:17.329 Jamie Hartmann-Boyce: Dr. Andy Wang, a visiting professor of economics at St. Lawrence University, is a co-author of the study and will answer select questions in the Q&A. Dr. Shue, thank you so much for presenting for us today. 11 00:02:18.510 --> 00:02:19.649 Bingjin Xue: Thank you, Jamie. 12 00:02:20.230 --> 00:02:25.359 Bingjin Xue: So, thanks for having me. So I'm going to share my screen. 13 00:02:25.670 --> 00:02:27.360 Bingjin Xue: Give me one sec… 14 00:02:36.230 --> 00:02:38.950 Bingjin Xue: Alright, so, Jimmy, can you see my screen? 15 00:02:38.950 --> 00:02:40.470 Jamie Hartmann-Boyce: I can, it looks great. 16 00:02:40.780 --> 00:02:43.859 Bingjin Xue: Alright, perfect. Thank you, everyone, for coming. 17 00:02:43.860 --> 00:03:08.740 Bingjin Xue: So, today we are going to present our study on the impact of e-cigarette policies and regulations on birth outcomes. I'm Ben Xu from the University of New Hampshire, and my co-author, Andy Wong, is also here. He's going to help answering some Q&A questions. So, before we start the introduction of our research, I want to start with motivations and anecdotal 18 00:03:08.740 --> 00:03:16.150 Bingjin Xue: So, my friend Andy, so he's here, so, he used to be a user of both 19 00:03:16.150 --> 00:03:28.070 Bingjin Xue: combustible cigarettes and e-cigarettes. But he stopped doing so, once his wife gets pregnant. And then part of the reason is, so his wife demanded him to. 20 00:03:28.290 --> 00:03:48.899 Bingjin Xue: So his wife is a strong supporter of all the e-cigarette regulations. Well, she claims that, those e-cigarette policies, so they're helping improving the birth outcomes. And this is because e-cigarettes, they're also harmful. They contain nicotine, and then also they emit aerosols. 21 00:03:49.450 --> 00:03:52.869 Bingjin Xue: And then, as economists, Andy and I immediately felt 22 00:03:52.980 --> 00:04:05.649 Bingjin Xue: wait a minute, there could be a substitution effect, right? And the idea is, if you ban e-cigarettes, then people may switch to using combustible cigarettes, which can be more harmful. 23 00:04:06.030 --> 00:04:13.640 Bingjin Xue: So, we decided to look at the effect of different e-cigarette policies on birth outcomes, and that's our motivation. 24 00:04:14.350 --> 00:04:26.440 Bingjin Xue: So we start our, disclosure slides. So the authors, me and Dee, we have not received any funding for this paper or any other tobacco-related funding in the past 10 years. 25 00:04:27.920 --> 00:04:32.310 Bingjin Xue: So, we start our study by looking into the statistics. 26 00:04:32.780 --> 00:04:51.780 Bingjin Xue: So there is a rising prevalence of e-cigarette use in the United States in 2023, according to CDC. 6.5% of adults reported current e-cigarette use. That's 7.6% among men, and 5.5% among women. So we use these statistics 27 00:04:51.780 --> 00:04:58.570 Bingjin Xue: We try to show that maybe e-cigarette policies, they can affect a sizable population. 28 00:04:59.410 --> 00:05:05.679 Bingjin Xue: So the distribution of e-cigarette usage varies by age groups. 29 00:05:05.680 --> 00:05:20.769 Bingjin Xue: Usage among ages 21 to 24 is the highest, that reached 15.5% in 2023, while other adults groups, ages from 18 to 49, they also show high prevalence rate. 30 00:05:21.140 --> 00:05:28.749 Bingjin Xue: So if you are familiar with the fertility literature, you may realize that this age group coincides the reproductive age. 31 00:05:28.940 --> 00:05:37.489 Bingjin Xue: So, we are trying to say that e-cigarette policies, they may have an effect on pregnant population. 32 00:05:40.670 --> 00:05:45.769 Bingjin Xue: Alright, so next, we zoom into e-cigarette usage during pregnancy. 33 00:05:46.140 --> 00:05:51.610 Bingjin Xue: Pregnant, there is a high prevalence during pregnancy, so… 34 00:05:51.860 --> 00:05:57.460 Bingjin Xue: People who use e-cigarettes during pregnancy is roughly 7% globally. 35 00:05:57.460 --> 00:06:17.429 Bingjin Xue: And this rate is high, when including pre-pregnancy use. So a lot of people, when they, during pregnancy, they are going to cease smoking, they are going to switch to either the NRT, which is nicotine replacement therapy, or they are going to switch e-cigarettes, as a way to cease, smoking. 36 00:06:18.020 --> 00:06:27.849 Bingjin Xue: So a survey conducted by multiple authors, they conclude that, many pregnant women, they perceive weeping as safer than smoking. 37 00:06:28.400 --> 00:06:35.250 Bingjin Xue: There are a lot of potential risks and potential benefits associated with e-cigarette use during pregnancy. 38 00:06:35.640 --> 00:06:37.939 Bingjin Xue: So first, we talk about direct effect. 39 00:06:38.370 --> 00:06:51.900 Bingjin Xue: So e-cigarettes, although they are perceived safer, they do contain nicotine. And also, they give people exposure to aerosols, so it can be harmful to the baby and the moms. 40 00:06:52.450 --> 00:06:54.740 Bingjin Xue: However, there is no consensus. 41 00:06:54.990 --> 00:06:59.990 Bingjin Xue: On whether vaping is safer or smoking is safer during pregnancy. 42 00:07:00.810 --> 00:07:07.799 Bingjin Xue: Another channel through which birth outcomes may be affected by e-cigarette policies is through secondhand smokes. 43 00:07:08.230 --> 00:07:17.370 Bingjin Xue: So, Schilling and co-authors defined that there is a high frequency of household exposure to nicotine from partners. 44 00:07:18.210 --> 00:07:25.919 Bingjin Xue: However, on the other hand, E-squared utilization may have some potential benefits, and this is through the substitution effect. 45 00:07:26.360 --> 00:07:37.319 Bingjin Xue: So for some, vaping supports cessation, reduction of cigarette use, and it also supports them quitting, and quitting smoking improves outcomes. 46 00:07:37.700 --> 00:07:45.260 Bingjin Xue: So there can be both channels of the effects, when we analyze the effect of e-cigarette policies. 47 00:07:47.760 --> 00:07:55.719 Bingjin Xue: So, our study focuses on four state-level e-cigaret regulations. They are indoor vaping restrictions. 48 00:07:55.970 --> 00:08:08.899 Bingjin Xue: Minimum legal sales age, we consider 18 and plus threshold, especially. And then, we also consider e-cigarette tax, we consider e-cigarette retail licensure laws. 49 00:08:09.030 --> 00:08:10.740 Bingjin Xue: So we go through them one by one. 50 00:08:11.100 --> 00:08:15.960 Bingjin Xue: So, indoor whipping restrictions prohibit e-cigarette use in indoor public spaces. 51 00:08:16.350 --> 00:08:22.369 Bingjin Xue: And usually those public places considers workplaces, restaurants, and bars. 52 00:08:22.820 --> 00:08:31.250 Bingjin Xue: In our study, we consider an either-or relationship. So we consider an e-cigarette policy that's banned 53 00:08:31.370 --> 00:08:36.710 Bingjin Xue: E-cigarette use in either workplaces, or restaurants, or bars. 54 00:08:37.480 --> 00:08:44.980 Bingjin Xue: And these indoor vaping restrictions are often integrated into clean indoor air laws, or we call it smoke-free air loss. 55 00:08:46.040 --> 00:08:49.570 Bingjin Xue: The second policy we consider is minimum legal sales age. 56 00:08:49.840 --> 00:08:55.760 Bingjin Xue: The age threshold varies by states, but most of them are 18, 19, and 21. 57 00:08:56.560 --> 00:09:08.619 Bingjin Xue: Many states adopted prior to… adopted the 21 threshold prior to the federal Tobacco 21 law, which is effective in December 2019. 58 00:09:09.360 --> 00:09:17.920 Bingjin Xue: In our study, we focus on 19 plus threshold, we don't focus on 19, and we're going to give you the reasons in the next slides. 59 00:09:19.080 --> 00:09:28.300 Bingjin Xue: So, e-cigarette policies. So, in our study, we consider both volume-based e-cigarette e-cigarette taxes and price-based e-cigarette taxes. 60 00:09:29.040 --> 00:09:33.569 Bingjin Xue: The last policy we consider in this study is e-cigarette retail licensure loss. 61 00:09:33.860 --> 00:09:43.420 Bingjin Xue: So, these license laws usually require licenses if you want to sell, or if you want to sell e-cigarettes over the counter or through a vending machine. 62 00:09:45.950 --> 00:09:48.810 Bingjin Xue: So, there were some other e-cigarette policies. 63 00:09:49.080 --> 00:09:52.920 Bingjin Xue: That is not considered or analyzed in our study. 64 00:09:53.110 --> 00:09:55.460 Bingjin Xue: This includes… Flavorites. 65 00:09:56.110 --> 00:10:04.500 Bingjin Xue: So, the earliest flavor ban, so we're talking about the state-level state, flavor bans, the earliest of them is in fall 2019. 66 00:10:04.660 --> 00:10:09.860 Bingjin Xue: Which is roughly the run… the end of our study window. 67 00:10:10.090 --> 00:10:13.609 Bingjin Xue: So that's one reason why we don't consider them. 68 00:10:14.060 --> 00:10:18.510 Bingjin Xue: And another reason, is because of our study design. 69 00:10:18.660 --> 00:10:27.039 Bingjin Xue: So, if we analyze flavor band, then it will make our empirical strategy more challenging. So we're going to talk about that later. 70 00:10:28.400 --> 00:10:39.360 Bingjin Xue: So, the second policy we didn't consider is the federal minimum legal sales age, which cites the age threshold at 18 in August 2016. 71 00:10:39.910 --> 00:10:51.049 Bingjin Xue: We don't consider this, again, the first reason is this is a national-wide implementation, which limits cross-state variation and challenges over identification. 72 00:10:51.670 --> 00:10:56.830 Bingjin Xue: Another reason is we are analyzing the effects on reproductive age population. 73 00:10:57.220 --> 00:11:04.570 Bingjin Xue: And there's a smaller… it's a very small exposure share for bursaries, for those who are under 18. 74 00:11:04.730 --> 00:11:14.379 Bingjin Xue: In our data, we only have 2.3% of librosies, which are Moms under age 18. 75 00:11:14.570 --> 00:11:20.670 Bingjin Xue: So, we ignored minimal legal age, which sets the threshold, age threshold at 18. 76 00:11:21.430 --> 00:11:29.490 Bingjin Xue: We also ignored federal, Tobacco 21 law, because, again, those laws are at the end of our study window. 77 00:11:32.060 --> 00:11:48.940 Bingjin Xue: Alright, so on this slide, I show you a regulation rollout over time. On the x-axis, we have the year quarters. On the y-axis, we show the cumulative number of states with a certain type of e-cigarette regulation in effect. 78 00:11:49.100 --> 00:11:52.619 Bingjin Xue: So all the dates in this chart are effective dates. 79 00:11:53.120 --> 00:12:08.380 Bingjin Xue: And we use different colors to show different policies we consider in this study that includes the blue for the minimum legal sales age, the red for licensure, and then we have the light blue for indoor restrictions, and then the yellow curve is for taxes. 80 00:12:08.700 --> 00:12:11.849 Bingjin Xue: So from this figure, we have two takeaways. 81 00:12:12.170 --> 00:12:18.860 Bingjin Xue: The first takeaway is, so roughly all states, they start their first policy in 2010. 82 00:12:19.400 --> 00:12:26.829 Bingjin Xue: The second observation is there are increasing number of states adopt different policies, and then 83 00:12:27.070 --> 00:12:31.070 Bingjin Xue: There is no strong preference of one policy over the other. 84 00:12:31.520 --> 00:12:36.740 Bingjin Xue: And then you can see there's a SPAC in 2020, and that is the federal, Tobacco Fund 1. 85 00:12:37.870 --> 00:12:38.570 Bingjin Xue: Buh. 86 00:12:40.170 --> 00:12:49.099 Bingjin Xue: All right, so, and then we talk about the literature, and then we think about the potential theoretical framework, how eCivider policy can affect birth outcomes. 87 00:12:49.210 --> 00:13:05.190 Bingjin Xue: So, in the economic literature, so, from the theory perspective, there can be two possible effects. The first effect is what we call the substitution effect, which means if you restrict e-cigarette, it will definitely curb vaping. 88 00:13:05.320 --> 00:13:09.040 Bingjin Xue: But it can also increase smoking or initiation. 89 00:13:09.630 --> 00:13:17.709 Bingjin Xue: So, this substitution effect, so is actually evidenced… evidenced in multiple studies, in multiple studies. 90 00:13:18.330 --> 00:13:22.579 Bingjin Xue: The second potential theoretical channel is the gateway effect. 91 00:13:22.870 --> 00:13:28.129 Bingjin Xue: Which means, a lot of people, so they started using e-cigarettes. 92 00:13:28.310 --> 00:13:34.450 Bingjin Xue: But then, this utilization can introduce them to more harmful substances. 93 00:13:34.580 --> 00:13:38.649 Bingjin Xue: Including smoking, and also substance use. 94 00:13:39.590 --> 00:13:46.609 Bingjin Xue: Well, another theory is… E-cigarette policies can change people's norms and perceptions. 95 00:13:47.060 --> 00:13:54.950 Bingjin Xue: So home and public restrictions can strengthen anti-nicotine norms, and taxes can also increase perceived risk of nicotine. 96 00:13:55.920 --> 00:13:59.010 Bingjin Xue: So what is the research gap that we are trying to fill? 97 00:13:59.190 --> 00:14:02.799 Bingjin Xue: There are a few studies examining pregnant population. 98 00:14:03.310 --> 00:14:09.019 Bingjin Xue: or examine the downstream birth outcomes. An even fewer studies handle staggered 99 00:14:09.230 --> 00:14:17.890 Bingjin Xue: state policy adoption timing and policy bundling. And those are the research gaps we are trying to fill in this study. 100 00:14:19.230 --> 00:14:27.230 Bingjin Xue: Alright, so the main research question. How do different easegrity regulations affect birth outcomes? So, we are analyzing four different policies. 101 00:14:27.390 --> 00:14:31.740 Bingjin Xue: Indoor whipping restrictions, minimum legal sales age, 19+. 102 00:14:32.030 --> 00:14:35.619 Bingjin Xue: And taxes, and e-cigarette retail licensure laws. 103 00:14:36.120 --> 00:14:38.569 Bingjin Xue: We look at 4 birth outcomes. 104 00:14:38.760 --> 00:14:43.009 Bingjin Xue: So, all those birth outcomes are birth outcome rate. 105 00:14:43.340 --> 00:14:45.140 Bingjin Xue: We examine preterm birth. 106 00:14:45.600 --> 00:14:54.419 Bingjin Xue: low birth weight, and also very low birth weight, and infant mortality. Those are the four main outcomes we examine in this study. 107 00:14:56.040 --> 00:14:59.680 Bingjin Xue: Next, we move on to the empirical method we used. 108 00:14:59.910 --> 00:15:07.950 Bingjin Xue: So, overall, we are going to use this difference-in-difference design, which means we compare treated states with the never-treated states. 109 00:15:08.140 --> 00:15:17.129 Bingjin Xue: And then, if we see after the e-cigarette policy, there is a change in trend, then we are saying there can be some effects due to the e-cigarette policy. 110 00:15:17.750 --> 00:15:22.019 Bingjin Xue: However, recent literature on DID identified some key challenges. 111 00:15:22.550 --> 00:15:25.629 Bingjin Xue: The first challenge is staggered adoption. 112 00:15:26.230 --> 00:15:36.619 Bingjin Xue: So, that means if different states, they adopt the policy in different years, or different points in time, and if those effects, they vary over time. 113 00:15:36.990 --> 00:15:42.049 Bingjin Xue: Then, a canonical two-way fixed fact estimator can give you biased results. 114 00:15:42.740 --> 00:15:46.519 Bingjin Xue: Or, in other words, There can be some forbidden comparison. 115 00:15:46.980 --> 00:15:52.999 Bingjin Xue: The forbidden comparison comes from comparing a late adopter to an early adopter. 116 00:15:53.540 --> 00:15:55.070 Bingjin Xue: So what is the solution? 117 00:15:55.650 --> 00:16:01.099 Bingjin Xue: So, in our study, we used, Callaway and Santana's DID estimator. 118 00:16:01.350 --> 00:16:03.610 Bingjin Xue: What they do is very intuitive. 119 00:16:03.890 --> 00:16:09.310 Bingjin Xue: So, they are going to eliminate the impact of the forbidden comparison. 120 00:16:09.630 --> 00:16:18.369 Bingjin Xue: They consider cohort-specific every treatment effect on the treated, and then we're only aggregating the allowed comparisons. 121 00:16:18.500 --> 00:16:21.239 Bingjin Xue: So that is the key idea of identification. 122 00:16:22.590 --> 00:16:29.400 Bingjin Xue: And we have a slide, which is from the Goodman-Bacon paper, about DID, but given the time, I'm going to skip this slide. 123 00:16:30.560 --> 00:16:35.930 Bingjin Xue: So next, we talk about the second challenge of identification, which is policy bundling. 124 00:16:36.130 --> 00:16:37.400 Bingjin Xue: The idea is this. 125 00:16:37.870 --> 00:16:39.539 Bingjin Xue: The e-cigarette policies. 126 00:16:39.730 --> 00:16:42.079 Bingjin Xue: They rarely arrive alone. 127 00:16:42.400 --> 00:16:46.630 Bingjin Xue: States often adopt multiple e-cigarette policies in close succession. 128 00:16:46.910 --> 00:16:55.120 Bingjin Xue: So, in our scenario, we have 4 policies, and we observe that multiple states, they adopt multiple policies at the same time. 129 00:16:55.290 --> 00:16:59.210 Bingjin Xue: Or, they adopt the same policy about the same time. 130 00:16:59.960 --> 00:17:06.430 Bingjin Xue: So here, I'm going to present two figures to give you some evidence of policy bundling. 131 00:17:07.470 --> 00:17:12.870 Bingjin Xue: The first figure is a correlation coefficient matrix. 132 00:17:13.329 --> 00:17:24.979 Bingjin Xue: So, the X and the Y axis, I show you four different policies. Those are indicator variables, indicating whether a state has a specific policy, in effect, in that time. 133 00:17:25.690 --> 00:17:28.190 Bingjin Xue: So, here, we have two observations. 134 00:17:28.400 --> 00:17:29.360 Bingjin Xue: Number one. 135 00:17:29.590 --> 00:17:38.390 Bingjin Xue: If you look at the correlation coefficients, all of them are positive, which means if a state adopts one policy, they're more likely to adopt another policy. 136 00:17:39.100 --> 00:17:44.109 Bingjin Xue: Another observation is… This correlation can be extremely high. 137 00:17:44.680 --> 00:17:52.560 Bingjin Xue: For minimum legal sales age and indoor vaping restrictions, we have this coefficient as high as 0.55. 138 00:17:52.720 --> 00:17:57.850 Bingjin Xue: Which means they nearly adopt the same… the policy at the same time. 139 00:17:58.990 --> 00:18:02.969 Bingjin Xue: The second observation is from this stacked bar chart. 140 00:18:03.510 --> 00:18:07.500 Bingjin Xue: So, in this stacked bar chart, on the x-axis, I still give you the timeline. 141 00:18:07.780 --> 00:18:10.360 Bingjin Xue: On the y-axis, I give you the number of states. 142 00:18:10.990 --> 00:18:15.500 Bingjin Xue: So, this stacked bar chart tells you How many states? 143 00:18:15.700 --> 00:18:18.149 Bingjin Xue: Have multiple policies. 144 00:18:18.370 --> 00:18:19.779 Bingjin Xue: Ides, the same time. 145 00:18:19.890 --> 00:18:23.179 Bingjin Xue: So, for example, if you look at this yellow band. 146 00:18:23.510 --> 00:18:30.169 Bingjin Xue: It tells you the number of states with exactly one policy in effect at that time. 147 00:18:30.750 --> 00:18:33.450 Bingjin Xue: And then, if you look at this blue. 148 00:18:34.060 --> 00:18:41.390 Bingjin Xue: Bar chart, then it tells you the number of states with all four policies in effect during that time. 149 00:18:41.640 --> 00:18:44.589 Bingjin Xue: So we have, again, two tegabytes. 150 00:18:44.930 --> 00:18:51.019 Bingjin Xue: The first takeaway is you can see that the state with only one policy 151 00:18:51.530 --> 00:18:55.060 Bingjin Xue: The number of states really goes down since 2017. 152 00:18:55.560 --> 00:19:00.740 Bingjin Xue: So, if you examine this period of time, you really need to worry about policy bundling. 153 00:19:01.600 --> 00:19:11.160 Bingjin Xue: The second observation is, since 2015, we observed the start of this right stacked bar, and then you observe this blue 154 00:19:11.410 --> 00:19:12.530 Bingjin Xue: Stack the bar chart. 155 00:19:12.770 --> 00:19:20.249 Bingjin Xue: So they tell you that there are… so starting in 2015, there are more states with at least 3 policies, in effect. 156 00:19:20.800 --> 00:19:24.050 Bingjin Xue: So that gives us challenge in identification. 157 00:19:24.880 --> 00:19:28.699 Bingjin Xue: So how do we isolate the effect of a single policy? 158 00:19:29.360 --> 00:19:30.910 Bingjin Xue: The idea is this. 159 00:19:31.160 --> 00:19:32.920 Bingjin Xue: We're going to identify 160 00:19:33.310 --> 00:19:44.820 Bingjin Xue: the first e-cigarette policy enacted by a single state. So, for example, if a state enacted tax first, then we're going to label this state as a tax state. 161 00:19:45.160 --> 00:19:52.700 Bingjin Xue: And then we are going to truncate that state's series at the date our second policy became active. 162 00:19:52.840 --> 00:20:02.980 Bingjin Xue: So basically, we are going to limit the window to where the state only have one policy in effect. And that's how we isolate the effect of different policies. 163 00:20:03.640 --> 00:20:10.820 Bingjin Xue: And we're using this CSDID estimator, which allows us to compare cleanly with the never-treated states. 164 00:20:11.020 --> 00:20:28.209 Bingjin Xue: So when we compare our treatment states, we are not comparing with the other treatment states. We're only comparing with the never-treated states, the states that have never enacted any e-cigarette policy during our study period, which is 20… which is 2005 to 2019. 165 00:20:30.340 --> 00:20:39.720 Bingjin Xue: All right, so the next slide gives you what I call the identification window. So in this figure, you can see different color bar. They give you 166 00:20:41.330 --> 00:20:57.239 Bingjin Xue: the first policy of each state. So, for example, if you see a red bar, then that tells you the first policy of that state is minimum legal sales age. And if you see this green bar, it tells you the first policy of that state is e-cigarette licensure law. 167 00:20:57.440 --> 00:21:04.709 Bingjin Xue: And then we'll see the gray bar that tells you that the states already have a second policy in effect. 168 00:21:05.430 --> 00:21:11.110 Bingjin Xue: So, for some states, They enact their first policy, 169 00:21:11.740 --> 00:21:19.110 Bingjin Xue: at the same time as their second policy. So that's when you see a gray bar at the very beginning. 170 00:21:20.250 --> 00:21:26.960 Bingjin Xue: All right, so that's our key identification idea. I'm going to pause here to answer audience questions. 171 00:21:32.650 --> 00:21:40.539 Jamie Hartmann-Boyce: Sorry, my Zoom paused for a second, but thank you so much for that. Before we get to audience questions, please do keep 172 00:21:40.540 --> 00:21:55.850 Jamie Hartmann-Boyce: bringing them in, I am going to hand over to our discussant. So our discussant today is Dr. Michael Cooper, a postdoctoral scholar from the University of California, San Diego. He's published research on the effect of indoor e-cigarette bans on birth outcomes. 173 00:21:55.850 --> 00:21:57.629 Jamie Hartmann-Boyce: Over to you, Dr. Cooper. 174 00:21:58.000 --> 00:22:11.170 Michael Cooper: Thank you. So my first, clarifying question is that you just mentioned that only never-treated states are used as the comparison control group, and that leaves me wondering about… 175 00:22:11.410 --> 00:22:19.150 Michael Cooper: Pre-treatment states, if they're used as a comparison, and specifically, so looking at this graph you have. 176 00:22:19.340 --> 00:22:33.399 Michael Cooper: If you look at California, they go from zero policies to immediately multiple policies, and then get truncated out of the sample. So is California serving as a control group, or helping with identification somehow? 177 00:22:33.780 --> 00:22:44.770 Bingjin Xue: Thanks for the question. This is actually very important to our identification strategy. So, we are using the never-treated state as control only. So, for example, California, that is not part of the control. 178 00:22:45.510 --> 00:22:54.310 Bingjin Xue: And then, so we have 14 control states without any policy until 20, until Tobacco timeline. 179 00:22:55.010 --> 00:22:56.119 Michael Cooper: Okay, got it. 180 00:22:56.320 --> 00:23:01.470 Michael Cooper: And then a couple clarifying questions. So these two questions are both 181 00:23:01.980 --> 00:23:20.209 Michael Cooper: basically, are your policies just gonna be analyzed as indicators? Like, they're either on or off. Because I was thinking about, e-cigarette taxes, you know, they have different types. They have based on liquid or based on the price, and different levels, you know, how strong is the tax? 182 00:23:20.850 --> 00:23:33.590 Bingjin Xue: Yep, so that's, again, another very good question, and that's actually our next step. So, for now, we are considering an indicator variable, so it's 0, 1, and then once they switch on, they always stay on. 183 00:23:33.740 --> 00:23:45.210 Bingjin Xue: So, as far as I know, so when I examine… so our data comes from the CDC state dataset, and then I examine that data set in detail, we don't identify any states that switch back to control. 184 00:23:45.620 --> 00:23:49.150 Bingjin Xue: So, this really gives us the, 185 00:23:49.470 --> 00:24:01.069 Bingjin Xue: very reassuring policy identification idea. And then in the future, we're going to try the continuous DID to identify the intensive margin of maybe the effect of tax. 186 00:24:01.270 --> 00:24:08.499 Bingjin Xue: But maybe for a minimum legal sales age, we don't see the benefit of using a continuous DID measure. 187 00:24:09.840 --> 00:24:15.390 Michael Cooper: Yeah. And speaking of the minimum, you know, the age restrictions… 188 00:24:16.550 --> 00:24:20.350 Michael Cooper: I was just thinking that… 189 00:24:20.670 --> 00:24:38.129 Michael Cooper: You know, they only apply to a very small number of mothers. They apply to mothers that are, you know, 18 to 20 or so, but if you just have this very basic indicator that turns on for all the mothers, you might be kind of washing out the effects on this very small subpopulation. 190 00:24:38.570 --> 00:24:55.259 Bingjin Xue: Yeah, that makes sense. So, we're going to do some sub-sample analysis, but we're not going to present those results today, but definitely we're going to look at different age groups, different moms' age groups, and then hopefully we're going to see, some effect. 191 00:24:55.400 --> 00:24:57.940 Bingjin Xue: Or, moms who are young. 192 00:24:58.280 --> 00:25:01.999 Bingjin Xue: But then we'll see. We haven't done that part yet. 193 00:25:02.280 --> 00:25:11.119 Michael Cooper: Okay, got it. And I just have one more, comment, before turning it back to you, which is that it's kind of a bigger… 194 00:25:11.480 --> 00:25:15.299 Michael Cooper: Vague comment, which is that, 195 00:25:15.410 --> 00:25:31.520 Michael Cooper: you're kind of… I just wonder if you're throwing out a lot of useful identifying variation in your data by truncating these states out. You know, these are a lot of useful states that are implementing multiple policies. 196 00:25:31.630 --> 00:25:37.259 Bingjin Xue: And you end up comparing totally unregulated states like Florida and Texas. 197 00:25:37.270 --> 00:25:41.339 Michael Cooper: Two states that are very lightly regulated with only one policy. 198 00:25:41.590 --> 00:25:53.110 Michael Cooper: And then you're kind of throwing out the data from the heavy regulation states, like California, like we talked about. So I just wonder if you considered any other estimators that could account for your… 199 00:25:53.170 --> 00:26:04.569 Michael Cooper: your two problems that you've been bringing up without throwing out, so much data, with the caveat that this is very clean, what you're doing. This is a great way to isolate one policy at a time. 200 00:26:05.210 --> 00:26:08.569 Bingjin Xue: So, that's actually, another… 201 00:26:08.670 --> 00:26:22.950 Bingjin Xue: next step we are considering. So for now, we are only trying to identify a very clean first policy effect. And then, the next step is we are going to, based on the first policy, we want to analyze what's the effect of additional policies. 202 00:26:23.310 --> 00:26:32.529 Bingjin Xue: And then we're going to try to build one model to estimate, so, like you said, we're going to take advantage of, 203 00:26:32.550 --> 00:26:39.000 Bingjin Xue: all the variations. We build one model to analyze the effects of all policies, but then, in that case. 204 00:26:39.010 --> 00:26:57.310 Bingjin Xue: we are going to contaminate this estimator. So this… there is a trade-off. So we can definitely use, maybe one CSD ID will… where we're going to include some dummy variables for the other policies, and then we can include all states. So that will be one of our… our robustness check. 205 00:26:58.700 --> 00:27:01.680 Michael Cooper: Great, that makes sense. That's all I have for this part. 206 00:27:02.110 --> 00:27:02.740 Bingjin Xue: Thank you. 207 00:27:03.080 --> 00:27:14.630 Jamie Hartmann-Boyce: Wonderful, thank you. I think Andy is doing an amazing job answering audience questions, so I think go ahead and continue with the presentation. Audience, please do keep those questions coming in. 208 00:27:15.120 --> 00:27:16.699 Bingjin Xue: Alright, thank you so much, Jamie. 209 00:27:17.070 --> 00:27:19.000 Bingjin Xue: So, I will continue. 210 00:27:19.000 --> 00:27:37.349 Bingjin Xue: So, we move on to the data section, and then we're going to talk about our revisit our identification strategy, and then show our preliminary findings. So, the data we use are from the CDC Vital Statistics Restricted Use Birth Certificate Data. The sample we consider are 2005, 211 00:27:37.350 --> 00:27:42.700 Bingjin Xue: From the first quarter to the last quarter of 2019. 212 00:27:42.810 --> 00:27:45.729 Bingjin Xue: And we have county by culture panel. 213 00:27:46.200 --> 00:27:52.919 Bingjin Xue: So we can analyze our data at county by year level as well, and then our result for now is quite consistent. 214 00:27:53.630 --> 00:28:03.230 Bingjin Xue: So the outcome we considered are preterm birth rates, and then we also have low birth weight, very low birth weight, and infant mortality rate. 215 00:28:03.820 --> 00:28:16.200 Bingjin Xue: We also consider some smoking behaviors, so those include whether the mom, they ever smoke during pregnancy, or how many cigarettes they use every day. 216 00:28:16.780 --> 00:28:21.750 Bingjin Xue: The control variables we considered are 5-year age bins. 217 00:28:21.900 --> 00:28:29.750 Bingjin Xue: interacted with race and ethnicity, which include non-Hispanic whites, non-Hispanic Black, and Hispanic. 218 00:28:30.530 --> 00:28:43.880 Bingjin Xue: So, we express all those age compositions as share of county females, ages between 14 and 55 year old, which is the reproductive age. 219 00:28:44.300 --> 00:28:55.969 Bingjin Xue: We also include indicators for ACA Medicaid expansion and recreational marijuana legalizations, because studies have shown they can affect birth outcomes. 220 00:28:57.200 --> 00:29:16.289 Bingjin Xue: So this is the empirical framework we are using. We are using this difference-in-difference framework, and we are using the Calloway and Satana DID estimator. So again, this estimator are going to estimate cohort-specific treatment effect first. So basically, the treatment effect for each single state, and then you aggregate them. 221 00:29:16.290 --> 00:29:19.319 Bingjin Xue: And by eliminating all the forbidden compressors. 222 00:29:21.040 --> 00:29:25.870 Bingjin Xue: So, we start presenting some results. We start from descriptive findings. 223 00:29:26.450 --> 00:29:32.250 Bingjin Xue: So, here, we have trend plots, For preterm birth rate. 224 00:29:32.620 --> 00:29:34.609 Bingjin Xue: Per 100 lab verses. 225 00:29:35.170 --> 00:29:45.689 Bingjin Xue: So, we're trying to see if there are any pre-trends. And also, we want to see if there are any divergence in trends after the enactment of the policy. 226 00:29:46.380 --> 00:29:52.739 Bingjin Xue: Here, the black line, or the black curve, gives you the control states. The 14 never-treated states. 227 00:29:53.130 --> 00:29:59.119 Bingjin Xue: And then the other colored lines, they give you states with any policy during the study periods. 228 00:29:59.900 --> 00:30:06.730 Bingjin Xue: So, from the figure, we don't see any clear trends. And also, they seem quite parallel. 229 00:30:07.270 --> 00:30:16.090 Bingjin Xue: Another observation from the figure is you can see states that have ever adopted any e-cigarette policy, they tend to have lower 230 00:30:16.390 --> 00:30:21.400 Bingjin Xue: Preterm birth rates, which means they usually have better birth outcomes. 231 00:30:22.980 --> 00:30:28.220 Bingjin Xue: Alright, so the second figure I show you here is for the low birth weight rate. 232 00:30:28.550 --> 00:30:44.709 Bingjin Xue: So again, we have a similar observation. We see a strong seasonal trend, but fortunately, this seasonal pattern is very consistent between the control state and the treatment state. And in a DID design, this kind of 233 00:30:44.910 --> 00:30:50.360 Bingjin Xue: Synergy will be canceled off when we control for culturally fixed effects. 234 00:30:51.790 --> 00:30:56.530 Bingjin Xue: And then we move on to very low birth weight and infant mortality. 235 00:30:56.940 --> 00:31:07.740 Bingjin Xue: The pattern is very similar. Again, we observe parallel trends. At the same time, we don't have any evidence of divergence in birth outcomes. 236 00:31:07.860 --> 00:31:10.140 Bingjin Xue: So all of those are average. 237 00:31:10.280 --> 00:31:14.319 Bingjin Xue: Birth outcomes at the county and culture level. 238 00:31:15.240 --> 00:31:25.790 Bingjin Xue: And the last figure is for infant mortality rates, so we see there is a decline in infant mortality rates since 2005, and then again, we don't see any divergence in TRET. 239 00:31:27.820 --> 00:31:33.829 Bingjin Xue: All right, and then we present our formal, ATT, every treatment effect estimates. 240 00:31:34.240 --> 00:31:42.299 Bingjin Xue: So… Overall, we don't find any significant effect associated with any loss. 241 00:31:42.770 --> 00:31:45.350 Bingjin Xue: On any, birth outcomes. 242 00:31:45.540 --> 00:31:59.719 Bingjin Xue: So, for example, here, we have panel A, we're analyzing the effects of e-cigarette tax, and then in panel B, we analyze the effects of e-cigarette retail licensure laws, and then in columns 1 through 4, we have different birth outcomes. 243 00:32:00.340 --> 00:32:07.380 Bingjin Xue: The first outcome, the outcome labeled ATT, they give you the treatment effect of that policy on a specific outcome. 244 00:32:07.780 --> 00:32:17.119 Bingjin Xue: So, if I add one significant star, it means it is significant at 10% significance level, however, we don't observe any significant star here. 245 00:32:17.500 --> 00:32:24.090 Bingjin Xue: And also, the magnitude of any estimates, of any coefficients, they are very close to zero. 246 00:32:24.540 --> 00:32:30.450 Bingjin Xue: So overall, our conclusion, or preliminary conclusion, is we don't observe any effect. 247 00:32:30.640 --> 00:32:35.520 Bingjin Xue: On birth outcomes associated with e-cigarette taxes or retail licensure laws. 248 00:32:36.280 --> 00:32:55.029 Bingjin Xue: And then we move on to minimum age restrictions and indoor waving restrictions. Again, we don't observe any significant finding of any law on any birth outcomes. And also the coefficients, they are not statistically significant, they are also very close to zero. 249 00:32:56.720 --> 00:33:06.150 Bingjin Xue: And then, so we present some event study plots here. So, from the CISDID event study plots, you can see two colors. 250 00:33:06.600 --> 00:33:17.209 Bingjin Xue: The blue color to give you the pre-treatment period, so you want to see those are now effects, which shows you evidence of, no pre-existing threats. 251 00:33:17.660 --> 00:33:26.279 Bingjin Xue: And then, the red colored bars, they give you the treatment effect, or dynamic treatment effect, after the enactment of the policy. 252 00:33:26.770 --> 00:33:34.659 Bingjin Xue: So again, the overall observation is we don't see a very clear pattern of any effect. 253 00:33:34.920 --> 00:33:38.359 Bingjin Xue: Associated with ways, birth outcomes. 254 00:33:39.160 --> 00:33:47.110 Bingjin Xue: And here, different slides gives you, we organize them by birth outcomes. So, for example, the first slide is for preterm birth. 255 00:33:47.510 --> 00:33:52.520 Bingjin Xue: And then we move on to the next page. This is the effect on low birth weight rate. 256 00:33:52.900 --> 00:33:56.179 Bingjin Xue: And then the next slide is on very low birth weight rates. 257 00:33:56.290 --> 00:33:59.160 Bingjin Xue: And then, the last one is for infant mortality. 258 00:33:59.460 --> 00:34:07.730 Bingjin Xue: Although, for certain outcomes, we do observe maybe some effects in one period of time, but maybe those are from 259 00:34:07.730 --> 00:34:21.009 Bingjin Xue: some outliers, which we are still working on. We are still trying to identify what causes those, maybe, outliers. But maybe after removing them, the overall pattern, you can see there is no significant findings. 260 00:34:22.170 --> 00:34:23.749 Bingjin Xue: So, we draw the conclusion. 261 00:34:24.429 --> 00:34:33.340 Bingjin Xue: So In summary, so overpaper, we use first policy timing with truncated window. 262 00:34:33.870 --> 00:34:45.290 Bingjin Xue: To analyze the effect of different state policies on preterm birth rates, on preterm, low birth weight, very low birth weight, and infant mortality rate. 263 00:34:45.480 --> 00:34:52.450 Bingjin Xue: And we find small and statistically indistinguishable from zero effects associated with any policies. 264 00:34:53.110 --> 00:35:05.220 Bingjin Xue: So, the interpretation of the finding is, despite policy impacts on nicotine behavior seen elsewhere and seen in previous studies, we do not detect any short-term improvement 265 00:35:05.360 --> 00:35:08.160 Bingjin Xue: Or worsen effect on birth outcomes. 266 00:35:08.460 --> 00:35:26.129 Bingjin Xue: However, we want to, say that all the effects so far are still preliminary, so after we, addressed some key issues, maybe with the outliers, and then we're going to also try subgroup analysis, maybe we're going to find some other findings. 267 00:35:26.780 --> 00:35:30.379 Bingjin Xue: And… That's all for my presentation. 268 00:35:30.630 --> 00:35:33.429 Bingjin Xue: Thank you so much. I'm going to open the floor to discussion. 269 00:35:33.690 --> 00:35:38.230 Jamie Hartmann-Boyce: Wonderful, thank you so much. So again, we'll turn to our discussant first. 270 00:35:40.990 --> 00:35:55.240 Michael Cooper: Great, thanks for that, presentation. So I have a series of comments, mostly, you know, making sense of the null result here, the null effect result. 271 00:35:55.680 --> 00:36:04.780 Michael Cooper: Yeah, so all of these policies that you're analyzing, in one way or another, we think they increase the cost of using e-cigarettes. 272 00:36:05.180 --> 00:36:11.369 Michael Cooper: Right? Maybe not the direct monetary cost, but they might affect norms or things like that. 273 00:36:11.430 --> 00:36:30.939 Michael Cooper: And my main thought is just that substituting one form of nicotine for another might cancel out to no net effect on birth outcomes. So, you know, if we think nicotine is the primary harmful component for the development of the baby. 274 00:36:30.940 --> 00:36:37.199 Michael Cooper: Then maybe there is substitution going on, but it's roughly the same amount of nicotine, and that cancels out. 275 00:36:37.290 --> 00:36:49.760 Michael Cooper: to no effect on birth outcomes, which makes me wonder if you might want to look at cigarette smoking as an outcome, because I know that is in the birth records data. So did you consider that? 276 00:36:50.200 --> 00:36:58.619 Bingjin Xue: Yes, that's a very good point. Actually, we already have the result, so it's not in the slide. So, for now, we do observe a substitution effect. 277 00:36:58.620 --> 00:37:02.370 Bingjin Xue: But for most of the policies, that substitution is insignificant. 278 00:37:02.370 --> 00:37:19.969 Bingjin Xue: I remember, we only observed one significant finding for… I can't remember what policy is that, so I'm not going to, give an uncertain, answer. So we do have one significant substitution associated with one policy, but for the other policies, the effects are positive, but insignificant. 279 00:37:20.290 --> 00:37:21.130 Michael Cooper: I see. 280 00:37:21.130 --> 00:37:40.890 Bingjin Xue: So, I know in the previous literature, usually, people only consider a subset of states, because some states, they're… the revision of the birth outcome, they… they changed, at around 2010, so we are still considering, maybe, limit the states 281 00:37:40.960 --> 00:37:47.129 Bingjin Xue: Into that subgroup, that can give us, maybe a better First, it's the G-Fact. 282 00:37:47.580 --> 00:37:48.309 Michael Cooper: I see. 283 00:37:49.030 --> 00:38:07.200 Michael Cooper: And my next comment is that I think one of the channels is completely shut off. One of the channels where we think these policies affect behavior is anti-nicotine or anti-tobacco norms, because pregnant women are this unique population where there's already 284 00:38:07.430 --> 00:38:19.860 Michael Cooper: strong norms against using nicotine at all. So, now if you tell them that you can't use an e-cigarette inside, that might have no effect. They're already not going to use it inside when other people are there. 285 00:38:20.270 --> 00:38:24.469 Bingjin Xue: Yeah, so just the thought about one of the channels might not even be functioning here. 286 00:38:24.470 --> 00:38:26.430 Michael Cooper: Although there are other channels that work. 287 00:38:26.930 --> 00:38:39.079 Bingjin Xue: Yeah, that's true. So, maybe, the effect on norms and perception, if there are already tobacco policies in effect, then maybe this, added value is very small and negligible. 288 00:38:39.080 --> 00:38:40.260 Michael Cooper: Yeah. So… 289 00:38:40.420 --> 00:38:54.529 Bingjin Xue: I feel… so when we read up the paper, we're going to include a very detailed discussion of this channel, so whether we should consider it or not. But for now, we are analyzing maybe an overall effect, so this is… 290 00:38:54.530 --> 00:38:54.980 Michael Cooper: Okay. 291 00:38:54.980 --> 00:38:58.229 Bingjin Xue: A reduced from overall effect of all the e-cigarette policies. 292 00:38:58.570 --> 00:38:59.180 Michael Cooper: Yeah. 293 00:38:59.290 --> 00:39:02.959 Michael Cooper: And my next thought was about infant mortality. 294 00:39:02.960 --> 00:39:05.090 Bingjin Xue: Because I… 295 00:39:05.090 --> 00:39:24.629 Michael Cooper: In our similar research, we did find an effect on infant mortality of indoor vaping bans, but the effect was strongest later in the infant's life, so we found a stronger effect in the 100-day to 1-year range, which might make sense if there's 296 00:39:24.630 --> 00:39:34.009 Michael Cooper: More exposure to secondhand smoke as opposed to secondhand vapor, which would probably be weaker. And then that could cause respiratory problems, 297 00:39:34.200 --> 00:39:38.060 Michael Cooper: In a marginal case where the baby's health is already poor. 298 00:39:38.170 --> 00:39:44.919 Michael Cooper: So, the idea there is maybe look at infant mortality, but later in the first year. 299 00:39:45.390 --> 00:39:51.450 Bingjin Xue: Yep, that's actually another very good point. So, for now, we are considering 12, Colders. 300 00:39:51.680 --> 00:40:07.299 Bingjin Xue: after the policy. So that's 3 years post-policy. But we can definitely extend that post-period window and see if there are, any long-term effects. And actually, from the figure, like, the effect of e-cigarette tax and, 301 00:40:07.300 --> 00:40:12.079 Bingjin Xue: retail licensure law, we can already see there can be some effect in the long run. 302 00:40:12.120 --> 00:40:23.180 Bingjin Xue: But so far, we don't see any effect for indoor whipping restrictions, but again, for the last period, so if you look at panel D, the last period is turning negative, so… 303 00:40:23.410 --> 00:40:28.610 Bingjin Xue: maybe after we extend the post period, we can have something. 304 00:40:28.610 --> 00:40:35.399 Michael Cooper: That's a good point. You can kind of already, see no effect in these event studies. 305 00:40:35.550 --> 00:40:40.140 Bingjin Xue: Yeah, so… Well, the outcome here is just, death in the first year, right? 306 00:40:40.300 --> 00:40:41.669 Bingjin Xue: Yes, that's right. 307 00:40:41.940 --> 00:40:57.410 Bingjin Xue: Yeah, and actually, I do review that paper, and then actually, the main difference is the sample. So maybe, like I said, maybe when we subside to a cleaner subsample, we can get some effect. For now, we are considering all 50 states and DC. 308 00:40:57.610 --> 00:40:58.270 Michael Cooper: Yeah. 309 00:40:58.640 --> 00:41:13.879 Michael Cooper: And I just have two more comments. The next one is about local-level laws. You know, you only look at state-level laws. And in our research, we bought data on county and city-level laws. 310 00:41:13.890 --> 00:41:21.450 Michael Cooper: Because there are a lot of big counties and cities that implement these e-cigarette restrictions before it goes statewide. 311 00:41:21.680 --> 00:41:36.819 Michael Cooper: And I'm pretty sure that could lead to attenuation bias, and bias your estimates towards zero, because a big part of the population is getting treated before, it's coded as treated in your data. 312 00:41:36.980 --> 00:41:39.349 Michael Cooper: So that could be one thing that's happening. 313 00:41:39.710 --> 00:42:04.639 Bingjin Xue: Yeah, this is another very good point. So, actually, we already started analyzing county-level policies, but unfortunately, we don't have that, that, that data set you use. So for indoor whipping restrictions, we realized there are many counties that have their own policies, but for tax, we looked into, so the call is dataset. So, there are two big counties maybe we want to consider, that's the next 314 00:42:04.640 --> 00:42:16.179 Bingjin Xue: And then also for retail licensure laws and minimum legal sales age, there are a couple of counties, but not only a handful of counties, so we can easily just include them in our sample. 315 00:42:16.180 --> 00:42:22.749 Bingjin Xue: But for indoor working restriction, there are so many counties. So, maybe I'm going to… I will reach out to you. 316 00:42:22.750 --> 00:42:23.210 Michael Cooper: Sure. 317 00:42:23.210 --> 00:42:25.199 Bingjin Xue: I've identified effect at the county level. 318 00:42:25.440 --> 00:42:26.050 Michael Cooper: Yeah. 319 00:42:26.300 --> 00:42:29.399 Michael Cooper: Okay, and my last comment is… 320 00:42:29.630 --> 00:42:36.280 Michael Cooper: I just have this sense that maybe these policies only bite when they're put together. 321 00:42:36.370 --> 00:42:50.010 Michael Cooper: And maybe you can construct some outcome measure that's, like, the policy intensity or the policy count. Because we talked about it earlier when we paused for questions that… 322 00:42:50.340 --> 00:43:06.809 Michael Cooper: You know, once… once these states are getting to the strong… strongest regulatory environment, they get truncated out of the sample. So maybe there's kind of a building effect of having all four policies, that you're not seeing with your current identification strategy. 323 00:43:07.060 --> 00:43:29.889 Bingjin Xue: Yep, exactly. So, when we analyze multiple policies, maybe we're also going to extend our study window. We're going to extend the data to 2024, where all states, they already have this, Tobacco 21 law in effect. And then we can construct some intensity measure to analyze, what is the effect of two policies, three policies, and four policies. We can only… we can… 324 00:43:29.890 --> 00:43:36.789 Bingjin Xue: either just look at the numbers, or we can look at maybe the policy combinations. That can be a follow-up step. 325 00:43:37.360 --> 00:43:39.310 Michael Cooper: Yeah, that's great, that'll be really interesting. 326 00:43:39.940 --> 00:43:44.950 Michael Cooper: Those are all the comments I had prepared. Thanks a lot for a really interesting presentation. 327 00:43:44.950 --> 00:43:46.150 Bingjin Xue: Thank you so much, Michael. 328 00:43:46.820 --> 00:43:59.260 Jamie Hartmann-Boyce: Thank you so much, so please do keep your questions coming through. We have a question from Mike Pesco. Would cigarette taxes be a useful control, for example, indoor air laws affecting smoking? 329 00:43:59.780 --> 00:44:18.510 Bingjin Xue: Yes, so actually in our previous version of the paper, we already controlled for tobacco taxes. But then, for this result, we haven't controlled for that tobacco tax. But from our experience, controlling for tobacco tax or not, doesn't really change the overall findings. 330 00:44:20.550 --> 00:44:27.850 Jamie Hartmann-Boyce: Thank you, and I'm just looking at some of the questions that have already been answered in our Q&A, in case they're interested 331 00:44:28.410 --> 00:44:43.840 Jamie Hartmann-Boyce: to… for us to hear, and for you to consider as well. In particular, the last one that came through from Mia Pang asked… said, the regulations had no effects on birth outcomes. Do you think this is because the prevalence of e-cigarette use in pregnant people is too low to observe an effect? And… 332 00:44:43.960 --> 00:44:56.960 Jamie Hartmann-Boyce: Andy shared that low prevalence is likely part of the explanation, given that very few pregnant people use e-cigarettes. I wonder if you have anything you might want to add to that in thinking through the reasons here. 333 00:44:56.960 --> 00:45:12.740 Bingjin Xue: So again, our next step is to do some subsample analysis, because we have birth certificates, so we can see, what's the impact on, maybe, the youth or young adults. So we're going to release that result later. 334 00:45:13.640 --> 00:45:29.639 Jamie Hartmann-Boyce: We look forward to seeing that. And Mike's shared, just some information that county-level indoor vaping restriction data is now available through a policy data repository, so that is in the chat for anyone who might be interested in looking at that. Thank you for sharing that, Mike. 335 00:45:29.660 --> 00:45:38.289 Jamie Hartmann-Boyce: We also have a question from Sherry Warshaw. It might be interesting to also track incidents of cleft palate with e-cigarettes versus conventional cigarettes, she said. 336 00:45:40.040 --> 00:46:00.280 Bingjin Xue: So, yeah, that's, very likely. So actually, I have already started to consider that, so we're going to consider the… how e-cigarette policies interact with tobacco policies, but then it will make our identification even more challenging. We are already truncating the windows, and then if we consider 337 00:46:00.280 --> 00:46:13.269 Bingjin Xue: tobacco policy, then, I wonder if we have enough power to identify any effects. So there is a trade-off in terms of identification versus, do we consider all the policies? But maybe on 338 00:46:13.320 --> 00:46:20.379 Bingjin Xue: A very easy way to implement that idea is to control for any tobacco policies. We're going to just use dummy variables. 339 00:46:21.310 --> 00:46:23.289 Bingjin Xue: But that's a very good suggestion. 340 00:46:23.780 --> 00:46:25.929 Jamie Hartmann-Boyce: Great, thank you so much. 341 00:46:25.930 --> 00:46:48.939 Jamie Hartmann-Boyce: Okay, unless any more questions are coming in, I am going to suggest that we close out, because we're very lucky that Ben is available for Top of the Tops. So, if you have more questions, keep them coming through Q&A, but if you'd like to discuss with a speaker directly with mics enabled, you're more than welcome to attend Top of the Tops immediately following this webinar. 342 00:46:49.040 --> 00:47:00.060 Jamie Hartmann-Boyce: If you're interested, there's going to be a meeting room URL posted in the chat, and if you copy that, you will be ready to join the live discussions once this webinar concludes. 343 00:47:00.490 --> 00:47:12.010 Jamie Hartmann-Boyce: I have something in the Q&A, let me just pull it up. Oh, I have a thank you very much in the Q&A, so thank you very much, and I will hand over now back to RMC, but thanks so much. 344 00:47:12.670 --> 00:47:13.220 Bingjin Xue: Thanks. 345 00:47:17.450 --> 00:47:29.879 Luis Zavala Arciniega: We are out of time, however, if you still have burning questions or talks for Dr. Ben, please join us at the top of the talks. That will be an interactive group discussion. 346 00:47:30.020 --> 00:47:38.589 Luis Zavala Arciniega: To join us, please stop the Zoom meeting URL post in the chat, and switch rooms with us 347 00:47:38.700 --> 00:47:50.040 Luis Zavala Arciniega: once the event concludes, and we will leave the webinar room open for an extra minute after the end, and give everyone time to copy the URL. 348 00:47:51.020 --> 00:48:02.719 Luis Zavala Arciniega: Thank you to our presenter, moderator, and discussant. And finally, thank you to you all for the audience of 126 people for your participation. Have a great weekend.