Probable Causation

View Original

Episode 27: Benjamin Hansen

Benjamin Hansen

Benjamin Hansen is the W.E. Miner Professor of Economics at the University of Oregon.

Date: April 14, 2020

Bonus segment on Professor Hansen’s career path and life as a researcher.

A transcript of this episode is available here.


See this content in the original post

Episode Details:

In this episode, we discuss Professor Hansen's work on legalizing marijuana:

"Federalism, Partial Prohibition, and Cross-Border Sales: Evidence from Recreational Marijuana" by Benjamin Hansen, Keaton Miller, and Caroline Weber.


OTHER RESEARCH WE DISCUSS IN THIS EPISODE:


Transcript of this episode:

 

Jennifer [00:00:07] Hello and welcome to Probable Causation, a show about law, economics and crime. I'm your host, Jennifer Doleac of Texas A&M University, where I'm an economics professor and the director of the Justice Tech Lab. My guest this week is Benjamin Hansen. Ben is the "UWE" minor professor of economics at the University of Oregon. Ben, welcome to the show.

 

Ben [00:00:26] Thank you for having me.

 

Jennifer [00:00:28] We're going to talk today about your research on legalizing recreational marijuana and how legalization affects the market for this drug. But first, could you tell us about your research expertize and how you became interested in this topic?

 

Ben [00:00:42] Yeah, thank you. So I'm an applied microeconomist, so generally I'm interested in studying human behavior and how we respond to all different incentives that we face around us. My research has mostly been focused on issues related to crime, health and risky behaviors. And marijuana policy is one of those things that it falls right in the middle of all of those topics. So it certainly remains a crime in the US, even though, you know, as we see news stories coming out, it's more legal than it ever has been and roughly twenty five percent of the United States citizens can now go purchase marijuana as simply as if they were purchasing alcohol. But in many parts of the country, it remains illegal. In fact, more people were arrested for marijuana possession last year than were arrested for all violent crimes. So it remains a hot topic within crime it also is an important health question.

 

Ben [00:01:45] You know, as we're setting tax policy, as we're setting regulations, as we're trying to prevent the externalities that we might be worried about when when there's a drug that is either legal or illegal.

 

Ben [00:01:56] So I think that there's a lot of moving parts to this that makes it a question that a lot of people will continue to be interested in and one that I've been interested in myself as I've researched other questions related to drunk driving or crime.

 

Jennifer [00:02:10] Yeah, lots of empirical questions. So as you said, marijuana legalization has been a really big policy issue in recent years, with many states either decriminalizing or fully legalizing use of this drug in at least some context. So give us a little bit more background on how this policy landscape has evolved over the past several years.

 

Ben [00:02:28] Yeah, I mean, so we could go way back early into the to the 1900s marijuana was criminalized and then that started to change with early decriminalization efforts first happening in Oregon and then it was legalized for medical use in California in 1996. I think that is really the biggest change that started and then kind of swept over the rest of the the nation that marijuana could be used for medical issues. After it was passed in California, then similar laws were passed in Oregon and Washington in 1998. Colorado soon followed thereafter. And then we've seen this spread throughout even states that typically might be very conservative, like Oklahoma now has medical marijuana.

 

Ben [00:03:20] And I think as people were using marijuana potentially for medical issues, potentially for issues that might be less medical states started to also get interested in just making it legal for recreational consumption. That's probably partly the fact that medical marijuana was always kind of a blurry distinction. If you look at many of the states where medical marijuana was allowed to be grown and to be sold, the most common person with a medical license was a male between the ages of 20 and 40 and the most common diagnosis they had was either trouble sleeping or lower back pain.

 

Ben [00:04:01] So maybe those are true medical issues or that might be the exact sort of medical issue that somebody might misreport if they just wanted to consume it recreationally. That led to the first recreational marijuana laws being introduced by Washington and Colorado in 2012. And soon after those laws were passed, other states didn't want to be left behind and so we've seen this spread all over the nation, from California to Illinois to Massachusetts. So it's spreading from coast to coast at this point.

 

Jennifer [00:04:38] What's your take on on what the political motivations are, the political drivers of legalization? Is it in many places? Is it a true concern about health and getting this drug into people's hands? Is it concern about big government and over criminalization? Is it all about tax revenue? What's what's your take on what's driving this?

 

Ben [00:05:00] I think that there's a lot of different arguments that people supporting recreational legalization would make, part of it could be the equity arguments that specifically as it relates to race, if you were to look at white people versus black people, you would find that marijuana use rates are almost the same as you cut across those demographics.

 

Ben [00:05:22] But arrest rates couldn't be more stark. So you would find that black people are anywhere from three to four times more likely to be arrested for marijuana possession, despite using the drug at almost the exact same rate as white people, so that those racial disparities are, of course, one motivation. Another one would be just the claims that the legal system is wasting too much effort on this. But I think generally all the states that have passed recreational marijuana laws have always incorporated tax revenue. In the states where low tax rates have initially been proposed, those have been usually been superseded by much higher tax rates. So I think it's pretty hard to understate the revenue motivations that most states have, especially because they usually have moved from a medical marijuana regime to a recreational marijuana regime. So they're already going from laws that are fairly decriminalized and sometimes fully decriminalized over to laws where now we're going to let this industry exist, we're going to regulate it and we're going to tax it. So I think revenue is a central motivation in probably every single state where recreational marijuana sales have been allowed.

 

Jennifer [00:06:49] But of course, in the background here, marijuana is still illegal from the federal government's perspective. So how do state and federal policy interact in this space? And to what extent is the federal government worried about people invading their state laws by traveling to states where marijuana is legal? Is that totally cool if it happens or is this something the feds are keeping an eye on?

 

Ben [00:07:08] This is definitely something the feds are keeping an eye on. It's it's also something that some some state, are worried about, too. So the interactions between federal and state law are numerous and complex. So some of the first places that this emerges is just in the taxation code. So if you own a recreational marijuana retailer in Oregon or Washington, you have to pay taxes on it. but it's technically a criminal enterprise. And ever since Al Capone went to prison for not paying his income taxes, there have been statutes on on the books that basically that require criminal enterprises to also pay their income taxes. And there's some special restrictions that are also placed on them. So they're not allowed to deduct, for instance, anything except for the cost of goods sold. So that means if you own a recreational marijuana retailer in Oregon, you can't deduct your lease, you can't deduct the cost of paying your employees and many other things that normal businesses would be able to deduct from their tax liability they aren't allowed to deduct when they're actually paying their federal income taxes. So in some ways, the federal government is actually earning a lot of additional revenue right now because the recreational marijuana retailers aren't allowed to deduct things as they would if it was a legal enterprise. So that's one benefit the federal government has. But the conflict remains and a lot of the businesses could face uncertainty about how they should plan or how much they should invest in this business if they're afraid that the federal government is going to come in and raid them and in fact, you know, you can now look them up. You can see where all these marijuana growers are. And technically, they are engaging in a criminal enterprise at the federal level, growing massive amounts of a drug that is scheduled at the exact same level as heroin and methamphetamine. But a couple of years ago, the cold memo was put out that basically said, as the federal government, we're going to ignore the fact that all of these states are passing laws that technically violate our laws and we're going to prioritize our resources in such a way that we're not going to go knocking down the doors and raiding all these marijuana retailers and marijuana growers, provided that the states take proactive efforts to ensure that a couple of things aren't happening. First, they want to make sure that money isn't that is going to these marijuana retailers isn't being diverted to other criminal enterprises, so namely organized crime or gangs or something like that. They also want to make sure that states are taking proactive efforts to make sure that drugs aren't ending up in the hands of teens and young adults that aren't supposed to have legal access to it. And the third thing is that they want to make sure that the marijuana isn't being diverted from places where it's legal, at least under state statutes, to places where it remains illegal. And the degree to which states can actually take proactive efforts to enforce the last one I think that's a fairly debatable question and we've actually seen some actions in the court about this. So a couple of Colorado's neighbors, Nebraska and Kansas, actually sued Colorado, alleging that Colorado's legalization was imposing externalities on them because of the increased enforcement costs they face from people that were crossing the border and purchasing marijuana in Colorado and then bringing it back to their states.

 

Jennifer [00:10:51] And I think the other piece of it I kind of have in mind is my sense is that the federal government has more authority to intervene if there is cross state trafficking. Is that is that part of this, too, or is that sort of a secondary issue?

 

Ben [00:11:08] So they do have authority to intervene, but they would typically only intervene if it was a mass quantity of marijuana that's going from one state to another.

 

Ben [00:11:19] So for smaller levels of consumption, they wouldn't necessarily put a lot of resources into trying to prevent that.

 

Ben [00:11:28] But we certainly have anecdotal evidence of people going from Oregon to a couple of years ago, they got pulled over in Illinois by some feds and I think they had 500 pounds of weed in the back of their car at the time. They claimed it was presents for Christmas, but I don't think that that defense necessarily held up in court.

 

Jennifer [00:11:51] OK, so your paper is titled "Federalism, Partial Prohibition and Cross-border Sales Evidence from Recreational Marijuana." It's coauthored with Ketan Miller and Carolyn Weber and it was just accepted at the Journal of Public Economics.

 

Jennifer [00:12:04] You have a bunch of other work on the effects of marijuana legalization and as I'm sure everyone can tell, have paid close attention to this policy space. So before this paper, what had we known about the effects of legalizing this drug? What types of outcomes have people been looking at and what did they found?

 

Ben [00:12:21] So there's been a lot of research on medical marijuana laws, specifically as we've gone from a place where marijuana was completely criminal to now where we had this partially legal world where you could purchase marijuana for medical consumption, what effects that had on traffic fatalities, the effects that that had on marijuana consumption, the effect it had on alcohol consumption. So a lot of this is getting into the economic questions of whether or not these two goods are things that we typically substitute for each other or goods that people tend to use together, as well as goods that we would call compliments.

 

Ben [00:12:59] There's also been some research focused on teen and young adult marijuana use. Have those things increased or decreased a bit? And there's been some research looking at harder drugs-- so what happens to the usage of heroin, methamphetamine, cocaine? Then there was some research also looking at the effects that medical legalization within the United States has on drug trafficking coming across the border from Mexico. So there's been a lot of research on the on the medical side of things. And I think the general conclusion that is come out of there is that most of the medical marijuana law studies have found evidence that marijuana and alcohol look like they could be substitutes for each other, at least for the average consumer out there.

 

Ben [00:13:51] There has been very little evidence that teen drug use or teen marijuana use has increased following medical marijuana legalization.

 

Ben [00:14:01] There has been some evidence that drug trafficking from Mexico has decreased due to the competition that drug traffickers from Mexico has faced from legal producers here in the United States. And there is some evidence that harder opioids have been used less in places where medical marijuana laws have been introduced, particularly in the states that have the most liberal regimes.

 

Ben [00:14:27] So we've learned a lot from medical marijuana laws, but that doesn't necessarily tell us exactly what's going to happen with recreational markets and things are going to change in the recreational world.

 

Ben [00:14:40] I mean, so the potency of marijuana being produced has increased. Marijuana is now being advertised in ways that it wasn't before. We're now seeing new regulations, there's certainly different economies of scale. There's new products being introduced. So now people can use vaporizers, there's edibles, so there there's a lot of concerns about the externalities that marijuana could bring as we move into this new fully recreationally legal markets and there could be concerns about the cross-border spillovers that could emerge because beforehand, in order to purchase marijuana, you had to go and get a medical license from the state that you wanted to purchase in where a doctor said that you had some sort of medical condition. The restrictions on those varied a bit from state to state -- so in some states, mostly in the Northeast, these restrictions were pretty, pretty severe.

 

Ben [00:15:33] So you typically had to suffer from a terminal condition in order to get medical marijuana in other states, mostly in the West the laws were fairly liberal. Like I said, you could often get a medical license for having trouble sleeping, lower back pain, a lot of different to diagnose conditions.

 

Ben [00:15:50] And so I think that leaves the door open for us, wondering what happens as we study these new recreational markets, how should we tax it? What is the effect of these different regulations? And so I think that leads to a lot of interesting questions that, you know, I'm just one of many different researchers out there trying to figure out.

 

Jennifer [00:16:09] So what are the challenges in figuring all this stuff out? What are the challenges in figuring out the effects of these policy changes on the outcomes we care about, such as in this case, how much marijuana people purchase and consume?

 

Ben [00:16:23] So some of the basic challenges are just data availability and data quality.

 

Ben [00:16:30] So we could look at information from the National Survey of Drug Use on Health with which asked some questions about recent marijuana use. We might worry that people are going to respond differently to those questions as the drug becomes legal, that maybe people might answer those questions actually more honestly than they did in the past. Likewise, we could use other proxies for marijuana use or possession. For instance, some people have looked at arrest rates and tried to look at arrests for marijuana possession as a way of measuring things like cross-border spillovers or or measuring the extent of marijuana use, but we might worry that those things themselves might be related to police enforcement or their priors about marijuana policy changing in other states rather than individuals actually changing their behavior. The other thing that we worry about is just what's happening with trends. So we have this new recreational market replacing a growing and liberal medical market. It could just be that the recreational marijuana laws are just a continuation of that medical marijuana regime that was already in place in the first place. So identifying the effects of policy can be challenging both from a data measurement standpoint, from potentially other trends that could be happening in the background or changes in enforcement and enforcement priorities.

 

Ben [00:17:53] And so I think those challenges themselves mean that researchers have to be both creative when they're thinking about how to approach solving these problems and they also have to be careful about potential data measurement issues that might conflate things that look like they're a policy impact when really it could just be a reporting impact that's also changing in the background.

 

Jennifer [00:18:19] So the setting you consider is Washington State as well, that neighboring states and the northwestern U.S.. So tell us about the policy context there. What was pre-existing policy regarding marijuana sales for medical or recreational use and what's the policy change you're going to be interested in here?

 

Ben [00:18:35] So Washington had a medical marijuana regime that had been in place from 1998 through 2014. Oregon had a medical marijuana regime in place from 1998 through 2015, and Washington began its legal sales in July of 2014 and right after this happened, the citizens of Oregon didn't want to left behind. They were concerned potentially about, you know, a growing marijuana availability gap of sorts with Washington having legal sales while they weren't available in Oregon. So Oregon immediately put a statute on their ballot in 2014 to allow for to allow for recreational sales. It passed in November of 2014.

 

Ben [00:19:27] And the additional provisions of the law wanted there to be kind of a two to three year transition period where the state was going to adopt a lot of the provisions that the federal government wanted in order to allow for recreational sales. So namely, they wanted to create a traceability system -- so this is kind of a data and reporting infrastructure that would allow growers to report when marijuana was first planted, harvested and then tested for potency and then also this traceability system would track final recreational sales when they took place, how much the sales were for, what were the taxes collected, et cetera. So that was the original plan, but in as 2015 came around, the citizens of Oregon became very impatient. They wanted their t marijuana and they wanted it now. So they effectively petitioned and demanded that the state change policy more quickly this led to emergency legislation. So this was a full emergency that was going to allow recreational sales to begin far earlier than the state had originally planned. So Oregon already had, as I said, a medical regime in place and so the emergency legislation allowed all of the medical retailers in the state of Oregon to begin selling recreationally on October 1st of 2015. So all of these retailers haven't gone through the licensing provisions that were going to be required of recreational retailers.

 

Ben [00:21:05] The marijuana grown hadn't been hadn't gone through similar licensing restrictions, and the marijuana hadn't necessarily gone through the types of potency and the test for other contaminants that eventually would be required through other regulations that were going to be imposed.

 

Ben [00:21:22] But the citizens were willing to take this risk on themselves, and that led to this big policy change that all of a sudden all these retailers that couldn't sell recreationally on October 1st of 2015, they could. And so we study how that policy shift that allowed basically everybody in Oregon to go to the local medical marijuana retailer and purchase marijuana how that affected sales, not in Oregon, because, like I said, there was no traceability data being collected. So there weren't administrative data being collected on medical sales. So we don't have any information on what was happening in Oregon in those moments, but we did have that information being collected in Washington so we can study how did this policy shift in Oregon effect sales in nearby Washington?

 

Jennifer [00:22:11] Right. So you're going to use this policy change as a natural experiment comparing marijuana purchases near the Washington state border before and after Oregon changed its law in order to measure effect on cross-border purchases. This allows you to test for cross state spillovers or externalities of state law changes that is when marijuana was legal in Washington, but not Oregon. How many people from Oregon do we think were unintentionally affected by Washington's law? To talk us through your empirical strategy, how are you using this natural experiment to measure the causal effect of marijuana legalization in one state, on other states that are nearby?

 

Ben [00:22:47] So in order to measure the effects on cross-border shopping, we split Washington into four different regions. So we considered the region that was closest to Oregon --so this is the Washington Oregon border. We consider also the Washington interior. We also considered the border along Canada and the border along Idaho. So we split up the state into those four different regions.

 

Ben [00:23:11] And then we ask, how did the trends in marijuana sales change in October 1st of 2015 when marijuana became available in Oregon? And so we estimate what was the trend drop in sales in all of those four different regions. So the key thing that we would be assuming if we just were to estimate this difference in trends by using either an interrupted time series approach or a Russian discontinuity approach, which here these things are really analogous. We're just using time as a running variable is that there are no other coincident policy shocks that are affecting sales at that time. We might be a little bit worried that that assumption might be wrong or we might be worried that there are some other seasonal shifts that happen, that people consume more marijuana in the summer than they do in the fall and as we transition over into October, maybe there's going to be sales dropping everywhere, not just because Oregon legalize.

 

Ben [00:24:13] And so that that's partly why we don't just look at the difference in the time trend before and after we actually subtract off the difference in the time trends from each other, meaning that we're comparing the drop in sales that happened along the Oregon/Washington border to the drop in sales that happened in the Washington interior. And so with that, we estimate a difference in discontinuities approach. And I think that's something that is important to be able to do if you're particularly relying on approach like a regression discontinuity in time or an interrupted time series approach, because a lot of the standard tests that we would like to do in a regression discontinuity in time, we can't.

 

Ben [00:24:54] So you usually can't test for whether or not the density is change, because time itself is going to be a uniform variable. We often can't test whether or not covariates are changing and we can in some context, but we can't on this one, just because we aren't able to measure a lot of those covariates at a high enough frequency to confirm whether or not there is stability on each side of the threshold.

 

Ben [00:25:17] And so instead, we have essentially this placebo region that we can use to try to difference off any sort of other unobserved demand shocks that could be happening at the same time. And in addition to having this placebo region, the Washington interior, we have a couple of other regions of the state that we think shouldn't also be affected, that we can use this additional placebo checks to make sure that we're really picking up this demand shock of Oregon legalization when we focus on the Oregon and Washington border.

 

Jennifer [00:25:44] And just to back up a little bit and kind of step through the thought experiment you have in mind, you're kind of imagining that you've got a bunch of people in each of these different regions of the state that are buying that live in Washington and are buying and consuming marijuana because it's legal. But then along the Oregon border, you also well, I guess along the Canadian border, the Idaho border and the Oregon border, you also have some people coming across the borders and buying in Washington, perhaps consuming in Washington or they're perhaps going home to consume. And you want to measure basically how many people there are and so when Oregon legalizes, the idea is that those people have no reason to go across the border anymore. So if suddenly you see it on, it was October 1st that the policy changed. If on October 1st, suddenly you see a big drop in sales of marijuana along the Washington or along the Oregon border, and you don't see a similar drop in other places, then that suggests those are a bunch of Oregon consumers that were buying in Washington before and are now staying home. Is that basically the story?

 

Ben [00:26:48] Yep, that is 100 percent the story. Thank you for saying it so much better than I did.

 

Jennifer [00:26:53] Perfect. OK, so. All right. So let's talk more about these border areas. So what's in the border areas of these states look like? Are there big cities on both sides? Is it mostly rural? And because you're using the interior of the state as a comparison in your difference in discontinuities analysis, and you also look at these Idaho and Canada borders, it's placebo checks. To what extent should we be thinking of these as counterfactual? How good are they as counterfactuals for the area along the Oregon border?

 

Ben [00:27:23] So each region of the state is going to be a little bit different than each other, but there will be some similarities. So if we look at the Oregon Washington border, so focusing on the Washington side, that's going to be a mix of both urban regions and some regions that are more rural. So we have Vancouver, which is going to be a fairly urban community that's very close to Portland, there's a river that separates them and there's a number of different bridge crossings. It's pretty common for people to go actually back and forth between Vancouver over to Portland.

 

Ben [00:28:01] Likewise, if you continue along the river and you you keep following it further west, you're going to run into other border communities, things are going to get a little bit more rural, but then you're going to have places like Klickitat County and Hood River where, again, these communities often cross back and forth over the river. So they are somewhat interconnected, but it isn't completely porous because there is this giant body of water separating them.

 

Ben [00:28:33] When you go over to the Oregon, the Oregon, Idaho border, you do have some communities that are very close to each other. So you have Spokane, which is a medium sized city, and then on the other side of the border, you have Coeur d'Alene, it's a smaller community, but the there's a lot of different border crossings and this is regarded as one larger MSA if you were to look at economic classifications by the federal government. So there is a lot of interstate travel that's happening, which is pretty similar to the Vancouver/Portland relationship.

 

Ben [00:29:09] When you get up to the Canadian border, things are going to be different. So you instead of interstate borders where you can freely cross and you don't really worry about anybody inspecting your car other than if I guess if you go into California, they ask if you have fruit in your car, but that's like I think the only place in the country that that really happens. But you go over to Canada and definitely you're going to have to show your passport, you're going to have to explain why you're going in, why you're crossing those borders. So the border isn't going to be nearly as porous up in Canada, but there still is a fair amount of international travel that can happen back and forth, but it won't be nearly as porous what we see along the Oregon/ Washington border and the Washington/Idaho border.

 

Ben [00:29:53] So I would say that there's a lot of similarities between those regions for the interior of Washington, I think the key similarity similarities is that there's going to be a mix of both urban regions and very rural regions, when we define that interior region that doesn't share borders with any of those other either states or countries. So you're going to have Seattle, which will be very urban, but then you can drive an hour from there and it will feel as real as almost any part of the United States that you could find in rural America.

 

Jennifer [00:30:27] And you also seen graphs in the paper and I think the other nice thing is you can you can actually look at what the trends in all these places look like and they look pretty similar before the change.

 

Ben [00:30:37] Yeah. So one of the favorite graphs I have is one where we do basically, you know, an "Intro to Micro" rescaling of the data where we just pick a base period and divide everything by that and we look at what the trend in sales sales are along the Oregon border versus these other states. And before we do any econometric smoke and mirrors or anything to try to control for trends or any of that, the trends actually match up really closely with each other until this policy shift in October of 2015. And at that point, the sales and along the Oregon Washington border just plummet.

 

Jennifer [00:31:15] Beautiful. So, yes, let's talk about the data. What data do you have on marijuana sales to be able to look at all of this?

 

Ben [00:31:23] So we obtained information from the Washington Cannabis and Liquor Board on their traceability system. So these are administrative data that record the production, processing and final sales of marijuana in the state. There's some really great parts of the data in the sense that you actually have a you basically have a production function here. Like you can see the marijuana being converted from something very raw to its final product. You know, the prices that are charged from from each from first the wholesalers to the processors and then from the processors to the retailers. So you have a wealth of information about how a good is being produced. So that means that there's a ton of other questions other than this that you can focus on related to industrial organization. Then you have for many retailers that recording the exact minute, the exact time the sale is happening.

 

Ben [00:32:23] And you also have the prices charged. So those are some really great parts of the data is that you have a wealth of information of the sales. The only challenges with the data are, one, if you were to add up the number of transactions, you end up with hundreds of millions of them.

 

Ben [00:32:40] So this is by definition, a big data project that you're going to have some you're going to have to think about how you're going and collapse the data or how you're going to aggregate it to to make it more useful for for studying. And then there's also because it's administrative data, that means that it isn't perfect. And that's probably true of any data set, whether it's originally created for research or was originally created for administrative purposes, but when you get data that's created for administration, it means that there's going to be some measurement error that happens in it.

 

Ben [00:33:14] So and that doesn't mean that the data are necessarily that hard to clean up. It just means that there's some prices in the data, some outliers that you're probably going to want to consider either trimming or dropping just because it doesn't matter how good the marijuana is, it is never being sold for two thousand dollars a gram. And likewise, you are not generally seeing any sales where people are selling it for twenty five cents a gram. So those types of extreme outliers are pretty easy to identify in the data, but it's the type of thing that you want to make sure your results are not being driven by those outliers more than anything else.

 

Jennifer [00:33:55] All right. So you clean up the data and you run your analysis, though. Let's talk about the results. What do you find are the effects of legalizing marijuana for recreational use in Oregon on purchases in Washington?

 

Ben [00:34:07] So we find after Oregon legalizes marijuana, that sales drop along the Washington/Oregon border by thirty eight percent and this drop is essentially immediate after legalization happens in the state. And that figure of thirty eight percent is when we focus on the weight --so the total weight of usable marijuana being sold, usable marijuana is the traditional marijuana that people would put in either a bong or a pipe or something like that. We also focus on some other additional categories of marijuana to then when we compare what's happening in the Washington interior, we find that that's only dropping by two percent and it's also fairly imprecise. So it could be a zero, it could be two percent, but when we difference off that two percent decrease that we saw in the interiors of the state, we conclude that sales of marijuana fell by thirty six percent following Oregon's legalization.

 

Jennifer [00:35:13] And then what happened along the Canadian and Idaho borders.

 

Ben [00:35:16] So when we also study those other two regions, we found that essentially sales were not changing. In some specifications we found small decreases and others we found small increases, which all that is consistent with these other regions were not really affected by marijuana becoming more easily available in Oregon.

 

Jennifer [00:35:36] Writes a nice placebo check. So thirty six percent is big. That's that's seems big to me. Were you surprised by how large that result was?

 

Ben [00:35:48] Initially yes. I mean, it stood out every way we could visualize it. You know, every specification we essentially tried to throw at it. We found that this estimate was really robust and I mean it in retrospect, as we go through kind of like some of the later evidence we looked at, it actually turns out maybe it wasn't as big as what we might see in other regions, but at the time, it did surprise us how large it is.

 

Ben [00:36:15] However, in public finance, there actually has been evidence that this type of cross-border shopping can be quite prevalent. So most economists that have studied this beforehand have sometimes had to go to pretty extreme measures. So when they've studied cigarettes, economists have resorted to digging in trash cans to pull out packs of cigarettes to see where they come from. And there's been pretty plentiful evidence of people crossing borders to evade taxes, but this is really the first evidence that we have that people are really responding just to the legal availability of the drug, oftentimes, meaning that they're traveling further distances and paying taxes rather than purchasing it on the "black market" that is still available in whatever region they're coming from. So that says something about the preference that people actually have for a legal drug that they're willing to travel more and pay more for it, than they would be to buy the drug illegally back home. 

 

Ben [00:37:12] It also probably says something about the salience of what is legal versus illegal, because technically, by buying this drug in another place and then bringing it back across state boundaries, they are committing potentially drug trafficking.

 

Ben [00:37:26] But that probably feels more legal to them than going and buying marijuana on the corner from from a typical drug dealer.

 

Jennifer [00:37:36] And that's really interesting. OK, so then you run a few other tests to look at what types of purchases they're most affected, as well as where geographically you see the largest impact. So tell us more about those tests and what you find.

 

Ben [00:37:49] Yeah, so we we split up the marijuana transactions. One approach was to split it up by transaction size. And we found that the largest transactions were by far and away those that were most effective and this to us signals, for one, that it's probably not that people from Washington are coming to Oregon to try to buy the marijuana, but rather it's people from Oregon that were traveling to Washington and stockpiling and buying larger amounts of marijuana that now no longer are doing so.

 

Ben [00:38:16] The other thing that we look at is different types of marijuana that people could purchase. For instance, we looked at marijuana concentrates or we looked at edibles and while we find that there were decreases in those other types of marijuana products, they weren't nearly as large as the usable marijuana categories. And that's probably because those other two products, namely concentrates and edibles, they weren't available yet when more when Oregon allowed sales at the medical marijuana retailers because those products weren't being produced in the medical market.

 

Ben [00:38:46] And then some of the final things that we did was to estimate these decreases separately for every single marijuana retailer across Washington State and then to look at the gradient based upon distance to the Oregon border. We find that by far and away the effects were concentrated among the retailers within about 20 miles of the border. And as you moved further into the interior, that the effects quickly dissipated and at that point, they essentially looked like white noise.

 

Jennifer [00:39:14] So people drive right over the border, but they're not going to there's no reason to drive really far into the state. And just to highlight that middle finding there, that you find that there's an impact on the type of marijuana or the form of marijuana that is now easily available in Oregon, but not on the other forms is another nice, really nice placebo test. Right. So basically, people will still travel to get the kinds of of marijuana that they can't get at home, but they're not traveling anymore when it's right nearby. OK, and then the next thing you do is really neat you use Google Review data to get information on who is visiting each of these stores to tell us about that data set and how you're using it to help understand what's going on here.

 

Ben [00:39:59] Yeah, so, I mean, a completely reasonable critique of what we've done is that what you've done is great, but you only have internal validity, and this doesn't tell us anything about the types of cross-border shopping we might be seeing in other parts of Washington, other parts of Oregon or on the East Coast. And I mean, I take that to heart. That's that's definitely true. So be the next approach that we took to try to figure out, well, who's going to what store and where are they coming from was to look at Google Review data.

 

Ben [00:40:30] So, I mean, many of us might go out there and review things on Google. And in case you're not that careful with your privacy settings or maybe you even just want public, your review will post publicly and then you can click on your any review that you've done and then you can see your review history and so with that, we can measure we could potentially try to back out where the customers are coming from for every marijuana retailer. So we this wasn't necessarily our first approach. Our first approach was to look at the Yelp data that existed, mostly because those data were already far more easily scraped and aggregated and other other researchers had done that.

 

Ben [00:41:12] However, we found that there weren't many marijuana retailers on Yelp and there were very few reviews --so that data meant that we couldn't use Yelp. So then we thought, well, maybe we could use "Leafly" or some of these marijuana specific websites just because they have the most reviews, but then it turned out that you could see all the reviews, but you couldn't see where anybody was from.

 

Ben [00:41:32] So after this then we turned to the Google Review data, which, like I said, the great part about it is that you can look at people who have reviewed the marijuana retailer, see where else they reviewed and then impute where they're from. The only thing that was particularly challenging about this is that there's no convenient way to automate the scraping of Google Review data. And that's because Google, you know, rightfully views this as intellectual property, that they probably don't want other people collecting the data for. So we did so manually-- so we just went to every marijuana retailer along the Washington/Oregon border every marijuana retailer along the Oregon/Idaho border, a random sample along the in the Washington interior and then we also found marijuana retailers along the Oregon/Idaho border and marijuana retailers in Massachusetts. And so for each of those we could then we took a sample of the ten most recent reviews just because it was pretty time intensive to do this manual scraping.

 

Ben [00:42:39] And then we were able to predict what fraction of their recent visitors came from local regions versus what fraction were coming from out of state.

 

Jennifer [00:42:50] So just to clarify here, so you've got you know, Bob Smith reviews a marijuana retailer in along the Washington/Oregon border and then you click on his profile and he's also reviewed a shoe store near his house, then a local restaurant and stuff like that and you can use that to see he actually lives in Portland or wherever. That's basically what you're doing, OK? So so when you collect all the data, what do you find?

 

Ben [00:43:16] So with that, we could then estimate what fraction of reviewers are coming from these different regions. So we find if we look at Washington, Oregon in recent times, that only about nine percent of of the retailers were coming from Oregon, despite the fact that the population for Portland was much larger than the population for Vancouver, Washington. However, when we looked at other parts of the state, like the Washington/Idaho border, then we found that although the majority of people were coming from local regions, that there was a very sizable fraction that were coming from out of state. So if I look at the exact number, I believe it was let's see, we had sixty two percent of recent reviewers in Washington along the Washington/Idaho border were coming from Washington, but then that we found twenty five percent of those were coming from Idaho and a total of thirty eight percent were coming from from regions outside the state.

 

Ben [00:44:23] And so we can use that, comparing the fraction of reviews that are coming from local populations to estimate a parameter that tells us about the de facto legalization that a nearby purchasing opportunity brings to regions where marijuana continues to be illegal. So we did this for, as I said, the Washington/Idaho region. We also did it for the Oregon/Idaho region there we found for these Oregon marijuana retailers that 81 percent of the reviewers were coming from Idaho and and only about 10 percent were coming from Oregon itself.

 

Ben [00:45:06] When we did this for Massachusetts, we found that fifty four percent of the reviewers were coming from out of state. And for the out-of-state retailer or for the out-of-state reviewers, the modal reviewer was coming from New York, but then there was still a sizable fraction coming from Connecticut, New Hampshire and Vermont. So then we compare these review frequencies relative to their population frequencies to estimate the de facto legalization that these nearby opportunities offer.

 

Jennifer [00:45:35] And so that leads perfectly to what I was going to ask you next. So what does all of this tell you about the extent to which marijuana is de facto legalized in states that haven't yet actually changed their laws to legalize the drug?

 

Ben [00:45:50] Yeah, so and I we have to put some caveats on this, because it could be possible that people are less likely to review marijuana retailers if they fear legal prosecution. So if I'm coming over from Idaho, where marijuana remains an illegal drug both at the state and federal level, and I could be worried that my review could be used as evidence against me that might lead people to be less likely to leave reviews in the first place. So that would suggest that any selection bias that we see here would probably make these estimates lower bounds.

 

Ben [00:46:23] Moreover, if individuals are actually engaging in stockpiling behavior where they go when they purchase marijuana, but then they expect that marijuana to last them for a longer period of time, this is going to ignore any differences in transaction sizes that could exist across consumers as well. And we did find some evidence of that, at least along the Oregon Washington border and so it's entirely plausible it could exist in other places. So with those caveats, when we estimate de facto legalization, we find that the that along the Idaho/Washington border that marijuana is effectively 70 percent as legal as it is for people in Idaho versus the people that live in Washington. Or another way of saying this would be that the Google reviewers in Idaho are 70 percent as likely to leave a review as those who live in Washington when you compare them relative to their local populations. When we do the same thought experiment for the Idaho/ Oregon border, we find that marijuana being fairly nearby, which it isn't really that close. They're having to drive almost an hour, in some cases an hour and 15 minutes if they're living in Boise, but marijuana for them is 50 percent as legal as it is for people living in people living in Oregon and in Massachusetts, we find that marijuana is about forty five percent is legal for people in nearby states as it is for those that reside in Massachusetts.

 

Ben [00:48:01] And when we circle back to the original natural experiment that we did, which was along the Oregon Washington border, we find that the effective de facto legalization that recreational marijuana in Washington offered to those along the Oregon/Washington border was only about 20 percent, actually a little bit less than that. And we view this as probably largely being driven by the very liberal medical marijuana regime that existed in the first place in Oregon that created a substitute policy or substitute way to purchase marijuana for many of the people who lived along the Oregon/ Washington border, but in Oregon.

 

Jennifer [00:48:40] So the crossed state shopping seems to be higher in other places.

 

Ben [00:48:44] Yes, it does.

 

Jennifer [00:48:46] So as we talked about at the beginning, one reason that states care about these cross-state purchases is that there's big money at stake. So if people travel to another state to buy marijuana, that other state gets the tax revenue from that purchase. So what can you say about how much Washington benefited from this cross-border shopping from Oregon?

 

Ben [00:49:05] So from Oregon itself, we estimate that this increase their tax revenues by roughly four percent if we factor in the additional cross-border shopping that's happening along the Washington/Idaho border this number would increase by another six to seven percent, suggesting that upwards of 11 percent of the revenue that they've taken in could be coming from cross-border shoppers.

 

Ben [00:49:30] However, the benefits that local communities could have from cross-border shopping could be magnified even further. So the case in point of this emerges actually along the Idaho and Oregon border. So if you were to look along that region of those two states, the Oregon population is very small-- so in some communities, there's only hundreds of people and the largest community, Ontario, you know, there's about 6000. But nearby you have Boise, which has a couple of hundred thousand people. But Oregon is somewhat unique and that they offer cities an additional three percent excise tax that they can earn if they allow recreational if they choose to allow recreational marijuana sales and they want to enact that three percent additional excise tax.

 

Ben [00:50:16] So a small community of Huntington, Oregon, that has just over 400 people living in it saw this as an opportunity a couple of years ago and so they actually opened up a couple of marijuana retailers and so these quickly became famous throughout the state of Oregon as the busiest retailers in all of the state. And so first thing, when they open in the morning, there's a line of usually about 50 people waiting outside, all with Idaho plates waiting to buy marijuana in this small city of four hundred people and they quickly doubled their city budget based upon the additional revenue that they were taking in from these cross-border shoppers. Ontario saw all the revenue that this small city of Huntington was getting, and they chose to pursue a similar option themselves, allowing for three additional marijuana retailers. And the role of cross-border shoppers in these motivations is highlighted by the fact that some of the businesses are actually owned by business owners from Idaho that are starting these marijuana retailers in Ontario to cater to clients that are presumably coming from Idaho.

 

Ben [00:51:23] And one of their grand openings actually saw brief national headlines because Snoop Dogg came to perform at the grand opening and the citizens of Ontario had no idea who this is and so they actually put out a newspaper article titled "What is a Snoop Dogg" to to inform their citizens what was happening with this with this free concert. And they actually briefly asked for support from the state to have the National Guard help them out with crowd control.

 

Jennifer [00:52:01] Never a dull moment in a small area. Yeah, so this this is a very active area of policy and state legislators are looking for guidance on how to amend their laws. So what do you tell them? What are the policy implications of these new findings?

 

Ben [00:52:19] So, I mean, some of it is that cross-border shopping itself is going to create additional incentives to legalize for the states that do so sooner with places where marijuana is illegal around them. So we saw the additional revenues that Washington has gotten, which has added up to to nearly 80 million dollars of additional tax revenue that they've brought in. Likewise, if we look at other states on the East Coast, Massachusetts could be benefiting from this now, as could Illinois and places like New Jersey. We could expect that they could get even more in tax revenue if they legalized, particularly before New York City does.

 

Ben [00:53:00] So I think one thing is that the incentives of cross-border shopping are going to effectively create a race to legalization, because if I don't legalize and my neighbor does, I'm still going to have to deal with the externalities of marijuana because people are from my region are going to go to someplace else to buy it. But then that other region is actually going to get all the revenue to perhaps try to enforce or offset those externalities. So at some point in in regions like Idaho, if you're saying, well, effectively marijuana is legal, despite the fact that we socially don't approve of it, we might as well at least get this revenue that right now we're just exporting over to other states.

 

Ben [00:53:45] Moreover, I think that this means it will be somewhat challenging to try to study natural experiments by just using geographic boundaries. We're going to have to carefully think about what do these boundaries really give us if I compare or marijuana legalization in Washington and I want to think, well, Washington legalized sooner so I can use that as a treated state. And I can compare that to Oregon as the place that legalize is later. As a counterfactual, I could really worry that the that Oregon is actually being partially treated and perhaps in a way that's really, really meaningful. And that means any estimates that you get out of this could be biased toward zero. And so far, a lot of the research that's looked at the health effects of marijuana, either the effects on teen drug use or the effects on traffic fatalities, most of the papers that have come out on this have found effectively statistical zeroes and maybe the effects of zero or maybe these cross-border shopping effects are making it difficult for us to come up with counterfactuals. And, you know, this is just one measure of cross-border shopping that we're actually able to pick up in this study. It could be happening in mass on other levels, too, if you have growers that are producing marijuana. So, for instance, here in the state of Oregon right now, the authorities estimate that every single year there's millions of pounds of surplus marijuana that the growers aren't able to sell. And right now, they say that they destroy it when they can't sell it.

 

Ben [00:55:15] But it's entirely possible that they're packing they're packing up that marijuana into into trucks and then shipping it across state lines, over into other places, meaning that it could be that the rest of the country is also being partially affected by policies that are just happening in individual states.

 

Jennifer [00:55:31] That is definitely a problem for these natural experiments that it's been really neat to see these policies just like slowly roll out across the country. And that's exactly what researchers want to be able to, to really measure the impact of of broader access to this drug on all kinds of things. But, yes, it does make it more complicated really quickly. So so what other work has come out since you wrote this paper related to the effects of marijuana legalization? It does seem like the evidence on these policy changes has been coming out pretty rapidly in a short time. But as I mentioned the beginning, you have been keeping track of it all. So what what else is new?

 

Ben [00:56:11] So, I mean, there's a bunch of things that I've seen come out so myself, Mark Anderson, Joseba and Dan Reese, we wrote a paper looking at the effects of marijuana legalization on teen marijuana use. There we found again that medical marijuana laws had no effect on increasing teen marijuana use and we found some evidence that perhaps recreational marijuana was associated with a decrease in teen drug use.

 

Ben [00:56:36] I mean, I think more importantly, we didn't find it was going up, but the point estimate was actually negative and it was significant at marginal levels. So that was one study that's come out.

 

Ben [00:56:48] There's also been some some new research by Ketan Miller and Boyoung Seo. So that has looked at marijuana substitution in marijuana's either substitution or complimentarity with alcohol, such as marijuana prices drop and it becomes more easily available in Washington state, are we seeing liquor taxes going up or are we seeing liquor sales going up? Are we seeing them going down? And they tend to find evidence that it is going down. There's a couple of other papers that are looking at potentially its complimentarity with E cigarettte use.

 

Ben [00:57:21] One other interesting paper that is still in working paper form is by one of my colleagues here at Oregon. Ed Ruban has studied the effects of marijuana legalization on crime in Denver and they're one of the things that he has found really challenging is that it turned out that right. As marijuana legalization is about to happen in Colorado, Denver actually changed how they reported a bunch of different crimes from misdemeanors up through felonies.

 

Ben [00:57:48] And so if you look at the timing of it, it looks like all of these crimes, including drug possession, including disorderly conduct, including including graffiti or other types of nuisance crimes, it looks like all these things are jumping up.

 

Ben [00:58:04] And there's actually been a bunch of crime papers that have published with those data right now that are saying, oh, my gosh, marijuana is destroying Denver. And maybe it's true. But if you look at the city of Denver and all the reports they put out, they say, all right, in right as marijuana policy is about to shift, we changed how we reported crime. So you really can't interpret any data pre and post this.

 

Ben [00:58:27] If that's the type of thing that makes it really hard for us to study what's happening in a natural experiment, so I think it's really important to study what's happening from a police reporting standpoint or from what other things could be changing, too. So I have a paper I'm working on with an undergrad who finished her honors thesis and in that study where we're specifically focused on what's happening to to marijuana searches in Seattle, and we find the day that marijuana becomes legal to possess in 2012, searches for drugs fall by 60 percent. But then we also find that exact same day citations for not having your license and registration current also jump up by about 70 percent is it really true that people are like, yes, I can violently consume marijuana, I'm not going to pay my license, I'm not going to register my car, or is this just police had more stuff to do with their time, and so they started citing people for other marginal crimes. So I think thinking about how police respond to this shift in policy, that was a big part of what they do. I think that's really important as well.

 

Jennifer [00:59:41] Yeah, I totally agree. That's really interesting. But please, please don't change how you collect your data at the same time that a big policy goes into effect. It makes all of our lives easier if they are separate events. OK, so what what's the research frontier here? What are the next big questions that still need to be answered?

 

Ben [01:00:02] So there's big questions in public finance about optimal taxation. So, I mean, I can't tell you how many times we've been asked, well, what is the optimal tax rate? And so then I try to pull out a napkin and draw a picture of the Laffer Curve, you know, and at least so far, a lot of the estimates have come out of the literature say like, look we could probably push tax rates even higher than what they are today, and that's not the argument that's being made in California right now. So a lot of the the whole marijuana industry is alleging that high taxes is the reason that prices are so high and that they need to cut taxes in order to get rid of the black market, which for me is just the strongest evidence that the marijuana industry is actually a legitimate one now that they have a lobby that saying that taxes are too high and that they have to reduce taxes to save the children. Right.

 

Ben [01:00:58] So - but I think looking at optimal taxes and certainly that it could be what are rates? It could be where do you tax marijuana? Do you tax it by price? Do you tax it based on potency? Also, what the welfare implications are based upon the average person who smokes a lot of marijuana tends to have lower income and there could be important welfare effects there. And there was recently some research about that generally with syntaxes and certain - that came out, I think, in the - I forget if it was JP or QJE, but there was a whole paper about that with syntaxes and certainly that applies to marijuana, too.

 

Ben [01:01:36] Like I said, these questions about the effects of marijuana legalization on health, is it going to increase drug use? Is it going to decrease drug use? Is it going to substitute for alcohol? Is it going to be a complement with it? Or are traffic accidents going up because of it? There's a whole book of by Alex Berenson that says that marijuana legalization is responsible for the increase in homicides that we've seen nationwide. So is it related to violent crime? Does it actually cause psychosis and like big mental health problems or does it substitute for these other drugs that have larger externalities? I think those continue to be big questions.

 

Ben [01:02:14] There's I think certainly questions about how households are going to react to this, especially if there's disagreements about marijuana consumption and what are the effects on domestic violence. You know, and there's a bunch of different mechanisms that could come into play there, especially with its potential substitution with alcohol or complimentarity or just about household disagreements about how budgets could be spent so I think that could be an emerging question for the future. But like I said, I think we're going to see a lot of movement and this is going to be both on the micro side trying to study how individuals respond. There's going to be some things on the law and economic side, trying to figure out how district attorneys, how police are going to respond to this. There's going to be questions in industrial organization, what are the right regulations? What are the types of regulations we see? And so we have everything from in Washington state, organizations are required to be vertically different than each other, meaning if I grow marijuana, I have to sell it to a retailer who can then sell it versus in Colorado, you're actually required to be vertically integrated because you have to grow eighty percent of the marijuana you sell in your store.

 

Ben [01:03:25] So you have completely different regulatory regulatory environments that that could affect the economy of scale marijuana. And then I think there's there's the ultimate question of, well, what happens if we federally legalize and we enable more capital to be put into this perhaps advertising changes or becomes more aggressive or companies right now that aren't putting resources into trying to produce marijuana, maybe they dump a lot more into it. So certainly I could imagine big tobacco could want to enter into the marijuana market and engineered in ways that it isn't being produced right now. So, yeah, there's a ton of open questions,.

 

Jennifer [01:04:06] Lots of open questions, a good space for enterprising grad students to start working. My guest today is Ben Hansen from the University of Oregon. Ben, thanks so much for doing this.

 

Jennifer [01:04:25] You can find links to all the research we discussed today on our website, Probablecausation.com. You can also subscribe to the show there or wherever you get your podcasts to make sure you don't miss a single episode. Big thanks to Emergent Ventures for supporting the show and thanks also to our Patreon subscribers. This show is listener supported. So if you enjoy the podcast, then please consider contributing via Patreon. You can find a link on our website. Our sound engineer is Caroline Hockenbury with production assistance from Elizabeth Pacotti. Our music is by Werner and our logo is designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.