Romesh Vaitilingam interviews Imran Rasul for Vox

October 2010

Transcription of an VoxEU audio interview [http://www.voxeu.org/index.php?q=node/5639]

Romesh Vaitilingam: Welcome to Vox Talks, a series of interviews with leading economists from around the world. My name is Romesh Vaitilingam and today's interview is with Imran Rasul, from the University College London. Imran and I met at the European Economic Association's annual meetings in Glasgow, in August 2010, where we spoke about his research with Jerome Adda and Brendon McConnell on the impact of decriminalizing cannabis on crime.

Imran Rasul: What we're looking at is trying to measure the impact of the decriminalization of cannabis on crime. The particular episode that we study is when the police in one London borough, the borough of Lambeth, unilaterally decided to change their policing strategy towards cannabis crime, and de facto, not prosecute individuals caught in possession of small quantities that they could argue were for their own personal consumption. It's no change in how they regulated things on the supply side. But on the demand side, the police decided to relax the effort they were devoting towards that. The motivation was that they would then be able to reallocate their efforts towards other types of crime, potentially more serious types of drug crime, other types of crime more broadly.

So we've built a data set looking at crime rates for drugs crimes and six other different types of crime, by London borough, between the mid 1990s till pretty much today. So, it's spanning the 30, 31 boroughs of London, eight different crime types, looking at both rates of criminal offense, criminal arrests, and clear up rates. So, those last two measures focused more directly on conditional, and on offense being committed. Are the police actually doing something effective?

What we tend to find is that when decriminalization was introduced, the policy in Lambeth was introduced for a year. Then, after 12 or 13 months, actually reverted back to something close to what it was. That's interesting in and of itself. What we find is that the policy doesn't have an immediate impact on drugs related crime, but six months into the policy, there's a huge increase in drugs related crime. I think the order of magnitude is about 20, 25 percent, depending on exactly how you want to measure it. That effect is large, relative to what was occurring before, and long lasting. It doesn't seem to die out over time.

That's a bit of a puzzle. Because you might think that if the police have now changed their policy to no longer record as an offense something that was previously an offense the possession of small quantities of cannabis that the first order of effect would just be a purely mechanical one, a statistical one. That there should be a drop, a reduction in drugs recorded crime we find the opposite. So, in the paper, we tried to unpack that effect a little bit more, to see whether that's driven by an increase in the size of the drugs market, in general and whether we can say what would have to be true, in terms of the underlying drugs market, to explain this increase in drug related offenses.

In order to do that, we have various sets of thought experiments that we run through. The first is, supposing that you thought that the policy was already in place. That this is essentially what the police are already doing, and all they were doing is just making this official in order to protect themselves against sanctions from the public. So, all of the increase would effectively be driven by increases in consumption or the demand for other types of drugs, not cannabis.

If that were the case, then the increase in other types of drugs that we'd have to observe would be about 100 percent increase in the size of non cannabis related drugs market. So, by that estimate, it's a huge effect, holding constant the effects on cannabis. Then there's a few other thought experiments that we run through. But in all of those, we can reject the fact that there was no change in the market for drugs in Lambeth, as a result of the policy experiment. So, that seems to be potentially a cost of the police refocusing their attention onto other things. That the amount of drugs consumption seems to increase dramatically, and that's not an effect that's reversed, even when the policy is reversed.

There seems to be long run changes in the underlying structure of drugs markets, across London boroughs, as a result of the policy. So, given the underlying motivation that the police laid out for why they introduced the policy, we then look at what are the effects of the policy on other types of crime. There we find broad reductions in other types of crime. In fact, total crime, excluding drug related crime, falls by about six percent, in the long run, in Lambeth, relative to other London boroughs. This is at a time when drugs crime is generally and crime in London is generally rising. So, this is quite a big effect.

You find different effects on different types of crimes. So, that's a potential benefit of the policy. So then, to look more closely at what the police are actually doing, we look at what happens to the number of arrests and what happens to clear up rates as a result of the policy. Remember, for drugs crime, we find a big increase in drug offenses. Exactly at the same time, we find a big collapse in the number of individuals who are arrested for drugs related crime and a big collapse, ultimately, in clear up rates for drug related crime.

So those three bits of evidence really seem to suggest that when the policy is introduced, the police essentially reallocate their resources away from drugs related crime and potentially focus on other types of crime.

On those other types of crime, as I mentioned, offenses fall. But in addition, arrests for those other types of crime significantly rise. So, the police, on some measures, seem to do better and have an effective reallocation of their effort away from drugs related crime towards other types of crime. That causes a reduction of about six percent in total crime, excluding drugs.

So then, the final part of the paper says, "Well, we've got an increase in drugs crime, a decrease in other types of crime. Is there a way that we can see how people value these net effects?" So, we look at what happens to house prices. House prices are nice, because you can measure those within a borough. You can really do this at the postcode level, so look at whether there are differential effects on house prices by type of house, or by neighborhood within Lambeth.

Places where the drugs market might be particularly active might be differentially affected than other parts within Lambeth, where the drugs market doesn't have such an effect. So, in the long run, we find a reduction in house prices in Lambeth, holding everything else equal, as a result of the policy. So, despite there being overall reductions in crime, people seem to put disproportionate weight on the increase in drugs related crime. That more than offsets any weight they associate with falls in other types of crime.

That, presumably, is because when you have an expansion of the drugs market, people care about drugs crime, but also other things related to the drugs market that also occur, other forms of antisocial behavior that might be going on. This effect seems to be larger in parts of Lambeth, where the drugs market is larger to begin with so places close to Brixton Tube Station, for example. So, what we're trying to do in future work is to unpack what these other types of activities may be that are associated with the drugs market, such as road accidents, the consumption of alcohol, and potentially other complementary products to drugs, and so on and so forth. So, that's one direction for future work that we're looking to go in.

Another set of results that we have, which are spinning off in another direction, are to try and understand the effects of this policy on the restructuring in the drugs market in London as a whole. So, one check that we do is using a method called synthetic control methods, which is essentially to try and ask: Is there any other borough in London that seems to have a similar spike in drugs crime, when the policy in Lambeth is introduced?

The answer is no, with the exception of one other borough. One other borough, at the time when Lambeth decriminalizes, has a similar increase in drug related offenses. That other borough turns out to be Haringey. It wasn't clear to us why this should have been. Perhaps it's just a statistical chance that this would occur. We have 30 boroughs. Inevitably, you might expect a similar increase in crime to be observed in one of them. However, Lambeth and Haringey are orders of magnitude larger the effects there than any other borough, making us suspect that this needs to be looked at a little bit further.

So our second reaction was, well, perhaps the police in Haringey are doing the same as in Lambeth, except they're not announcing it. They're letting all the media attention focus on Lambeth, but they're essentially doing the same thing. If that's the case, then the police in Haringey should also be having, we should be observing a reduction in other types of crime in Haringey and all of the other effects we've found in Lambeth. That turns out not to be the case. In Haringey, although drugs crime rises, every other type of crime also rises. Police arrests don't change. Clear up rates don't change.

So it seems as if in Haringey, it may be and this is speculation, which is why it's not in the first part of analysis--there may some interlinkage between the drugs markets in Lambeth and that in Haringey, where we might think of Haringey as being an important port of entry of drugs into London as a whole. There are a number of connections that we want to pursue. One is the obvious connection between the tube lines between Haringey and Lambeth. The second is, Haringey is well known to be a port of entry of hard drugs, Class A drugs, into London as a whole.

I live quite close to Haringey. So, just by chance, I read a story in my local paper, which highlighted the fact that I think six months ago was the first time that armed police were permanently deployed onto the streets of some parts of London. The two locations in which they'd been deployed were Brixton, the center of Lambeth, and Haringey, in response to an increase in gun related crime. So, I'm sure these underlying networks that the police are aware of, we're just trying to now probe the data that we have in that direction, to try to better understand how the industrial organization of the market for drugs as a whole changes, when one location in particular decides to decriminalize.

Romesh: What lessons do you think you could take, generally, from this research for the policy of decriminalization of cannabis? I mean it strikes me that it's one borough where there evidently are serious spillovers to other parts of London. You might have a policy that would decriminalize cannabis within London, or you might have a policy that decriminalized in the UK, which might have very different effects.

Imran: Sure. So, in the paper, we were able to compare the effects of the localized decriminalization in Lambeth, vis à vis the nationwide reclassification of cannabis from Class B to class C in the UK that occurred between 2004 and 2009. So, in de facto terms, in terms of how the police were policing cannabis related crime on the demand side, the two policies were implemented in a very similar way. At a nationwide level, you can shut down some of the channels that we find in the Lambeth studies, such as drugs tourism and then, just look at whether the police are better able to reallocate their efforts toward other types of crime. The effects are much more imprecise, but generally, we find that when there's a reclassification of cannabis from B to C, in the UK as a whole, drugs related crime doesn't rise to anywhere near the extent that it did when Lambeth unilaterally had the same policy. But at the same time, the police are able to reallocate their efforts towards other types of crime, to some extent. So, we think that the paper has some implications for thinking about whether drugs policy should be coordinated between jurisdictions or not. Certainly, if you want to decriminalize, you shouldn't be the only place that does it. It seems to be these very large spillover effects, and the effects seem to persist in the long run.

Romesh: Both the Lambeth policy and the national policy were reversed. Was that because of a feeling that it had a damaging impact in the short term, and they couldn't almost invest the political capital in keeping it going for longer?

Imran: There's no one explanation you can find on either dimension that says this is the key reason why the policy was stopped. There seems to be a buildup of gradual concern related to both policies. Ultimately, political decisions were made to reverse both policies. If we focus in on the Lambeth experiment, in particular, that was initially announced to be a trial that would run for six months. What we find is though the increase in drugs crime that have been documented didn't occur in those first six months. It occurs at the end of this six month period. So, what happens at the end of the six month trial period? Well, that's when the policy is deemed to have been a success by Brian Paddick, who was actually the commander of the police in Lambeth and started to experiment with this policy. He actually instigated MORI, I think, to conduct an opinion poll of local residents during this first six month window. Where, individuals were asked about whether or not they thought the policy had been a success, or whether they were aware of the policy. What do they think about this general change in policing and such?

On the whole, people were deemed to be cautious, but relatively supportive of the policy, if it allowed the police to reallocate their efforts toward other types of crime. In the data, we observed no change in drugs crime in that first six month window. At the end of that six month window, when the policy was announced to be more permanent, that's when we observe a spike in drugs crime that seems to occur. So, there is some signal that's sent, either to consumers or suppliers, or to both, that now there's going to be, because there's going to be a permanent change, that that leads to changing in the underlying structure of the market.

As soon as the policy goes permanent and you observe this spike in drugs crime, that's when negative stories start to build up about the effects of the policy. There are reports that children are showing up into school, showing the after effects of having consumed cannabis, and so on and so forth. That feeds into political pressure and eventually the policy is reversed, seven months after it was announced to be permanent.

What's interesting is that when head teachers were asked about this, most of them said that they didn't see any effects. But these are very small samples. You only have a very small number of schools in Lambeth, but we're going to try to look at the educational outcomes of children who live in Lambeth, close to Lambeth, at the time that the policy is wide, in future research. But the police themselves were very much against a reversal of the policy. They felt as if they were more efficiently allocating their resources with the policy, which is part of the initial aim. We exactly picked that up in these different trends and different types of crime.

In terms of a broader policy implication, though, you think about how do people value this policy, in terms of house prices, summing together all of the effects on different types of crime, it seems that people don't want to live in neighborhoods in which they're the only one that decriminalizes. Does that also reflect nationwide? That's very hard to say.

Romesh: Imran Rasul, thank you very much.

Imran: Thank you.