Rat Park: False for Rats but True for Humans

Rat Park was a series of experiments done by Bruce Alexander on rats and drug addiction. Bruce Alexander was skeptical of earlier experiments, in which rats were given extremely bare environments and no real alternatives to pushing the drug lever. Predictably these rats had a tendency to just repeatedly hit the drug lever. Alexander gave his rats larger environments with more toys and tunnels, even other rats. Supposedly the rats with the more enriched environment used less drugs than the ones in the bare environment. suggesting that a mere lack of alternatives was to blame. It's not that the drugs themselves are hopelessly addictive, it's just that for those rats there was nothing else to do.

Except maybe rat park didn't replicate. The Wikipedia page still incorrectly states that it failed to replicate. It's true that some attempts to replicate failed. But there were several successful attempts to replicate it, too.

Maybe the rat park effect is real, and maybe it's not. I really don't know, and I don't particularly care. I actually don't think that animal studies are particularly useful for learning about addiction in humans. Human beings have foresight and planning. A rat pressing a lever never has the thought, "This will leave me with an incapacitating hangover that will preclude my working tomorrow. My boss will catch on if this becomes a pattern. So I need to either stop completely or have a strict, disciplined schedule for my bad habit that I never deviate from." The rat doesn't think, "If I pull the lever now, I won't be able to pick up my kid from daycare and my other kid from kindergarten." Intoxication makes the typical habits of a normal life unfeasible for most people. In contrast, the rats might get intoxicated but not be incapacitated in any meaningful sense. They don't perform complex tasks, coordinated with a few fellow rats for the benefit of other rats in a vast rat economy. Drug use patterns in human beings is going to be a lot more complicated than drug use patterns in rats. Even addicts often make rational choices to abstain, even for a fairly trivial reward.

I actually think that "total drugs used" might not be the correct measure here, if we're trying to measure dysfunction due to drug use. The rats in the enriched environment might consume the same amount of cocaine (or perhaps even more), but burn off the excess energy and direct it toward "pro-social" goals. The rats in the bare environment might screw themselves up on a smaller dose. I don't know how anyone would do this. Again, rats don't really have a functioning economy, so maybe it's hard to tell a "well-functioning" rat from his dysfunctional nearly-identical twin in another cage. I'm also reminded of a comparison of two human cultural groups from a Thomas Sowell book, one of which drank more total alcohol than the other. But the group that drank less alcohol had more alcoholics and alcohol-related social problems. The group that drank more had a lot of casual drinking, but very little drinking to excess. (I wish I had a more precise reference, but this is probably from his "...and Cultures" trilogy.) So plainly, in humans if not in rats, the total volume consumed is not the proper measure of social dysfunction.

The scientific debate over rat park is interesting, and I'll follow it for any definitive signs of a clear verdict. But it's ultimately a little silly. The lessons of rat park may or may not be true for rats, but they are definitely true for humans. Human beings weigh the salient alternatives and adjust carefully to incentives, even if rats don't.

See an excellent recent piece by Scott Alexander, called Against Rat Park. I suppose that title would fit for this post, too, but I'm making a very different point. Scott discusses Ogedei Khan, a son of Genghis who was at the top of the world. He nevertheless drank himself to death. Supposedly this goes against rat park because it proves that people with very (very) enriched environments are still susceptible to drug abuse problems. I draw a different lesson. If you're at the top of the world, you can afford a costly drug habit. Ogedei wasn't about to get "fired" from being Genghis' son. Certainly some people do wreck their lives with an uncontrolled drug habit, but that prospect keeps most of us in line. "Celebrities with drug problems" is a cliche. These are people who certainly work hard and put in a lot of hours while they are working on a project (say, a movie or a concert tour). But they also have a lot of "down time" and face little social censure for indulging a drug habit. So it's not that propensity to consume drugs is a simple monotonic function of social status, declining with rising status. It's not simply that desperation drives people to drug use. For some people, a drug habit would categorically change their life. They'd have to give up most of the good things (job, family, school) in order to accommodate it. I think this is a more useful way to think about what drives some people to drugs and deters others.

Pleasure Counts

You’re working for a government bureaucracy, and your division is tasked with doing a cost-benefit analysis of the country music industry. (That's country music, not the country's music.) You take your best shot. Resources are consumed in the production of country music. We’ll have to count those. Labor, equipment, physical media like CDs and cassettes. Plus there are those mega-concerts. People travel sometimes great distances to see them, so there is the occasional auto accident. We’ll have to count those, too. Plus the gasoline consumed and the resulting carbon emissions. Your team spends a while coming up with considerations like these and tallying up the cost estimates, and it’s starting to look like a really bad deal.

Wait a minute, someone says. “Don’t some people enjoy country music?” Duh. It’s terribly obvious now. Those resources don’t just get chucked into a black hole. They go into producing something that people enjoy.  Since people are willing to shell out their own money to cover this production, you can surmise that it’s probably a good deal. The country music industry would not exist if it didn’t generate some kind of consumer surplus. Those traffic deaths? A cost paid willingly by every motorists who decides to take a trip, and this reasoning applies just as well to killed or injured motorists who weren’t headed to a country music festival. This enterprise of cost-benefit analyzing an innocuous industry was silly from the get-go, because everyone failed to consider that other people’s pleasure counts. You might think they are philistines and their preferred art is trashy, but you would probably count their pleasure in any attempt at a cost-benefit analysis.

So what changes if we are talking about something less innocuous? How about skiing? How about horseback riding? Lawn darts? Trampolines? There is more danger in these pass times, but most people would still agree to count the joy these hobbies bring to their enthusiasts. You might even overrule the hobbyists, deciding that they are irrationally underestimating the risk, or irrationally unresponsive to a well-estimated risk. ("You underestimated the risk of a broken arm by an order of magnitude? Overruled! And you correctly estimated the risk but are failing to respond to it the way you respond to similar risks? Also overruled! Your hobby is banned!") But you’d probably feel weird about taking a big red marker and X-ing out the “benefits” side of the ledger.

What about drugs? Alcohol? Tobacco? Pleasure still counts. Again you might think the cost-benefit analysis comes out in favor of a ban. (It takes completely unreasonable assumptions to make this work, btw.) But pleasure should still count in the “benefits” column. I feel like there is this nasty tendency to completely dismiss the pleasure people get from their vices, as if it’s not a real thing or not worthy of our consideration. There are people who indulge their vices without ever having a problem. It doesn’t interfere with their work or family life, and they don’t do it excessively enough to damage their health. Even in people who do harm to themselves, it makes more sense to think of their habit as something pleasurable with offsetting costs, rather than something that is all cost. I’ve had a few hangovers. I don’t think of them as erasing the joy of the previous night. Rather it’s a price I paid for taking things a little too far.  If we are trying to treat people decently, as fellow human beings with their own sense of agency and and their own preferences, their pleasure ought to count. Even if it seems strange to us.

I think some people might scratch their heads at my previouspost about drug prohibition. The supply and demand diagrams were drawn to show the social welfare consequences of various drug control regimes, a large part of which is the consumer surplus. Consumer surplus is the benefit to the drug users above and beyond what they are willing to pay. If someone buys heroin for $25 but would pay up to $100 for it, his consumer surplus is ($100 - $25 =) $75. This might seem like an alien concept to some people. The welfare calculation is counting it as a good thing when a heroin user buys heroin?

(My argument actually doesn’t much depend on someone having to bite this bullet. It's amazing how robust the argument for drug legalization is. You can throw out the most obvious benefits, make unrealistic assumptions in favor of prohibition, and still get a pretty strong verdict in favor of legalization. At any rate, the conclusion that a tax is better than a ban is true whether you count consumer surplus or not.)

If counting consumer surplus for drugs seems odd, return to the country music cost-benefit analysis and ask yourself what changes. The costs are higher for a risky hobby? Sure, but you should still count the benefits. The pleasure is cheap and tawdry and not real? Some would say the same of country music, I’m sure (or whatever music you like). The pleasure is still real and is experienced by a living, feeling human being. Usually when one human being is incapable of considering another’s emotions and feelings, we consider it a pathology. We describe the person who disregards others’ feelings with a word that ends in “opath”. When it comes to puritanical social alarmists dismissing their fellow citizens’ cheap vices, the lack of empathy is just as harmful. It’s worse than that, really. A sociopath is usually found out and somehow censured for their bad behavior. The puritan’s inability to empathize is often socially validated and even embodied in official public policy. This really needs to stop.

Even smart people fall for this error. In an attempt at a thorough cost-benefit analysis of marijuana legalization, Scott Alexander adds some notes on missed considerations after commenters respond:

EDIT: People in the comments have pointed out several important factors left out, including:

– Some people enjoy smoking marijuana

…

 In other words, he pretty much got a "legalize" verdict while ignoring the biggest benefit. You would get a similar result with other substances. 

Commentary On the Tax Bill Is Atrocious

I'll start by saying I have no idea if the tax bill is good or bad on net, but I will freely point out deficiencies in some of the opinion pieces I've read. Bad as those are, the garbage that gets scrawled across my Facebook feed is worse. I see even really smart people fall victim to sloppy thinking.

Tax Incidence

What bothers me most about the commentary on the tax bill is the utter lack of economic understanding, with full bore inequality-mongering and class-baiting in its place. Much of the commentary is of the variety "Corporation X will see its corporate taxes decline by $Y billion dollars! Therefore the tax bill is a giveaway of $Y billion to Corporation X. Naturally this comes at the expense of the rest of society." When I see this kind of unserious comment, I think, "Really? So you've never heard of tax incidence?" People supplying this kind of comment are doing mere accounting without any thought about how the world adjusts. You can't calculate the distributional effects of a tax bill just by recalculating everyone's tax liability under the new tax structure, assuming everything else stays fixed. Things adjust. Prices adjust, such that a tax increase on a business will tend to raise prices and a tax cut on business will lower them. No fair objecting to one and not the other; this is a perfectly symmetric statement. Scott Sumner had an excellent post on tax incidence just the other day that perfectly captures the naive view and contrasts it with how economist think about these things.

Current Rates Must Be Justified

No, you can't simply dismiss any reduction in tax rates as a "giveaway to the rich." The existing tax structure is highly progressive. (Try looking up shares of income and comparing them to shares of taxes paid. Good analysis here.) Maybe it's not progressive enough, or maybe it's too progressive, or maybe it's just the right amount of progressive. I'll play agnostic on this question, but it is a question that needs to be answered. We can't just take what exists as a given, as if the status quo government policy were automatically justified, and berate any movement in the "wrong" direction. If you think that current levels of progressivity are about right or too low, that requires some sort of argument. Perhaps the outcome of some kind of optimization problem. Even if it's an imprecise and back-of-the-envelope optimization, that would at least be a great place to start.

Given the US's very high level of progressivity, almost any kind of tax cut is likely to reduce the taxes paid by rich people more so than middle-class and poor people. That doesn't mean it's the wrong thing to do. Suppose we are in a regime where even Bernie Sanders-level progressives think the tax regime is too progressive, and we mostly all agree we should scale back top marginal rates. A dishonest demagogue could cast this change to the tax code as a "giveaway to the rich." Just so. Fixing a tax structure that's too progressive is a thing.

Taxes on Capital Should be Close to Zero

I am disappointed by some parts of the tax bill, but the reduction in the corporate tax rate is badly needed. A bit of (now)-standard macro shows that it is impossible to tax the capitalists and use the proceeds to enrich labor. Do read the full piece by Garett Jones. Here's an excerpt:

First, let me sum up a key implication of Chamley-Judd:

Under standard, pretty flexible assumptions, it's impossible to tax capitalists, give the money to workers, and raise the total long-run income of workers.   

Not, hard, not inefficient, not socially wasteful, not immoral: Impossible.

If you tax capital income and hand all of the tax revenue to workers, then in the long run (or the "steady state") you'll wind up with a smaller capital stock. And since workers use the capital stock to earn their wages, the capital tax pushes down their wages.  

Also see this excellent post by Casey Mulligan. Here's an excerpt, but you'll need to see the supply and demand diagrams to understand them:

The red area (R) is the revenue from a capital income tax.  The red and green areas (R + G) are the losses from that tax suffered by owners of the factors of production, combined for capital and all other factors.  The revenue is a LOWER BOUND on the factor owners' loss.

In the long run, all of the factor owners' loss from a capital income tax is a loss to labor (the area below the horizontal dashed line is negligible; see A below).  Therefore, in the long run, capital-income tax revenue is a LOWER BOUND on labor’s loss.

Emphasis mine. There are ways of extracting taxes from rich people, it's just that capital taxes are a particularly inefficient way of doing it. (Corporate taxes, taxes on dividends and capital gains, inheritance taxes, and other taxes that hit savings and investments.)

This also falls under the header "tax incidence". When you tax corporate profits, the corporation doesn't just keep doing everything it was doing before and shell out the money. It raises prices to offset the tax. (I literally do this calculation as an actuary so that my company has an indicated profit margin. It's a standard piece of actuarial methodology, and there is certainly a non-insurance version of this in other industries.) It is less likely to undertake a project with a marginal rate of return. It will fail to invest in the kinds of "capital goods" that make its workers more productive, thus raising their workers' marginal product of labor and thus their wages. So "capital taxes" actually come out of the pockets of workers in the form of lower wages and higher prices. If you imagine that "corporate taxes" fall on a faceless, inhuman monstrosity, as if you were simply taxing a robot, get real. It's flesh and blood humans who ultimately pay those taxes. All those price increases and wage reductions happen in reverse if we cut an existing corporate tax. (Again, if you can easily see it happening in one direction but not in the other, the onus is on you to tell us what breaks the symmetry.)

Declining to Tax is Not a Cost

I have seen that the estimated "cost" of the tax bill is $100 billion a year. This is actually the reduction in government revenue, which is very different from the cost of the bill. A tax is merely a transfer. If you take $100 billion from some people and give it to others, or decline to do so, you can't simply say that the cost of this transfer/non-transfer is $100 billion. You need more information. How much does it cost to collect $100 billion? When you collect taxes, people adjust their behavior to avoid the taxes. They do things they otherwise wouldn't do, they hire tax accountants to minimize their bills, they decline services they would otherwise have paid for, and they decline to work as often or as diligently as they otherwise would have. (Extreme example that I am pulling entirely from memory: In some Scandinavian country you have to earn $19 in pre-tax dollars to pay one after-tax dollar to a laborer who remodels your home. People thus do a lot of home handy work that they'd otherwise simply hire someone to do. Some of the value of division of labor is thus lost. Even if I'm misremembering the example, this is the kind of thing that happens with very high tax rates.) These costs are called "deadweight losses." The money transferred isn't the loss. We've taxed Peter but paid Paul. That is a wash. The loss is the labor that Peter declines to give us, the foregone sandwich because of our tax code's quirky sandwich tax, the tax accountant who should have been an engineer if minimizing people's tax liabilities weren't so lucrative, etc.  Those are deadweight losses. That is the true social cost. Economists think that the social cost of taxing a proportion X of the national income scales up like X squared. So that $100 billion transfer may be more or less costly depending on how much your tax code is currently collecting and how large the total economy is. Under our current tax structure, collecting a dollar of revenue costs about $0.33 in deadweight loss (citation needed; this is a rough figure I have seen in multiple places so I am using it, as a placeholder anyway). So, unless that $100 billion is buying something worth at least $133 billion, it's probably a good idea not to collect it.

Now if we perpetually run budget deficits such that the government can't pay off the debts when they come due, that could lead to some kind of serious financial crisis with very high social costs. It's worth estimating those. But once again, the "$100 billion" figure tells us nothing about the magnitude of that cost.

Trickle-Down Economics Isn't a Thing

Some may read the section above on capital taxes and say, "That's just trickle-down! Which we all know is wrong and doesn't work!" And indeed, "trickle-down" as it's so often depicted is silly as a basis for tax policy. The idea isn't that you give money to really rich people and big companies and hope that maybe, if they're feeling generous enough, they'll give some of it back to the rest of us. If your model for this is to follow a dollar (as if you've put a radioactive tracer on it) from the government, to the rich guy, to the rich guy's tailor/grocer/driver in the hopes that it ends up in the hands of poor people, then your model is wrong. The benefit of tax cuts comes from getting rid of the high levels of distortion and deadweight losses, not from giving lots of money to rich people and hoping they will spend it on our services. If economists are right about the magnitude of deadweight loss from taxation, it should be easy enough to cut some spending to justify some tax cuts and net out a social gain. This holds even if you think some government programs are strictly cost-justified in terms of their dollar price tag, because you have to justify the social cost of collecting the tax revenue, too. Nothing "trickles down" here, we're just eliminating wasteful spending and the destructive effects of the associated taxes.

The Burden of Government is the Spending, Not the Taxes Paid

A dollar spent is a dollar taxed. You can finance that dollar of spending with a dollar of taxes collected this year, a dollar from budget surplus from a prior year, or a dollar from a future year (plus the interest on the dollar you borrowed this year). This tax bill may make the collection of taxes mildly more or less efficient, but it won't fundamentally solve our problem unless the spending problem gets fixed. I don't know if this impugns or justifies the tax bill. If the lower tax revenue induces a future congress to cut federal spending ("starve the beast"), then the budget deficits could be good for us. If the lower tax revenues simply mean that future tax rates will need to be higher and thus even less efficient (remember the X^2 scaling mentioned above?), then the budget deficits will be bad for us. The tax bill should be connected at the hip to a spending bill. There is plenty to cut. Very few federal programs would pass a serious, honest cost-benefit test. Just start ranking programs on their cost-benefit balance, set a threshold, and start cutting. At any rate, I wish people would stop taking their eye off the spending ball to obsess over tax rates.


I wanted to write this post after Scott Alexander wrote three bad posts in a row, repeatedly doubling down on his initial claim. Some very good commenters corrected his errors, and I hope that he's digesting them. (Re-reading his second and third posts just now, I see he is conceding some important points to defenders of the tax bill.) But I worry that the good comments will be drowned out by the flood of noise-comments. Not to pick on Scott, because he is a very intelligent, honest, and thoughtful commentator. I just think he got this one very wrong, and for fairly obvious reasons. In the third post (which he opens by explaining what he's learned from his commenters, a good start!), he says, "[Y]our job isn’t to explain Economics 101 theories to me even louder, it’s to explain how the country’s best economists are getting it wrong." And he links to the IGM forum, a survey of professional economists, specifically regarding a question about the tax bill. Great, we have a poll of experts and some of their commentary, so let's see what they said.

The exact wording of the question is:

Question A: If the US enacts a tax bill similar to those currently moving through the House and Senate — and assuming no other changes in tax or spending policy — US GDP will be substantially higher a decade from now than under the status quo.

Daron Acemoglu says:   

The simplification of the tax code could be beneficial, but it is more than offset by its highly regressive nature.

David Autor says:         

Tax policy appears to have little effect at the margin on GDP growth in OECD countries.

Markus Brunnermeier says:             

It is more likely that GDP will be somewhat higher

Darrell Duffie says:

A reduced corporate tax reduction is likely to grow GDP. Whether the overall tax plan is distributionally fair is another matter.

Austan Goolsbee says:

Of course not. Does anyone care about actual evidence anymore?

Robert Hall says:

Though there is merit in cutting the corp tax and other capital taxes, with no other changes in policy, the fed gov will collapse.

Oliver Hart says:

The incentive effects are unclear to me. Some of the simplifications make sense but many of the changes look like hand-outs to the rich.

Anil Kashyap says:

doubt it will substantially change things either way

Pete Klenow says:

Expensing of investment would provide a bigger boost to the capital stock and GDP 10 years from now, per dollar of revenue lost.

William Nordhaus says:

Keynesian effect will have disappeared. Higher debt will probably outweigh lower corporate tax rates. Unlikely that nothing else will change

Larry Samuelson says:

Other factors swamp the importance of details of the tax code in determining GDP.

Richard Thaler says:

Aside from the redistribution of wealth, hard to see this changing much.

I'd summarize this by saying there are a few basic themes. 1) There's always lots of stuff going on, so with many causal factors it's hard to know the effect of a tax bill. 2) Naive empiricism. (See Goolsbee's comment, also Autor's.) As if theory tells us nothing at all. If you can't tell, I don't care for this kind of thinking. 3) Economic theory on capital taxation is probably correct, but given theme 1) the size of the effect is swamped by noise. 4) Distributional effects matter. (But these guys should have answered the question about GDP independent of their feelings on justice of distribution). BTW, I don't really get the feeling that these guys carefully read the tax bill, entered its provisions into a big simulation, and gave a considered reply after seeing the results. Dare I say some of these folks might be influenced by politics and mood affiliation? Bryan Caplan suggested once that the panel has a left-leaning political bias, though he was careful to say that doesn't mean we should dismiss their consensus (when there is one). I take this survey seriously. I'm certainly not dismissing it. But some of the comments seem flippant and reveal a shallow consideration of the question. And I don't assume the non-commenting members necessarily had a more thorough justification for their responses.

So, in short, there is no safe "Gee, I just agree with the experts" position here because the experts are all saying very different things. That's when it's time to dive into the details, or admit your ignorance and just be agnostic. Indeed, I was surprised by the number of people responding to Scott in the comments who said something along the lines of, "I don't know if the tax bill is good on net, but your argument is wrong here..." Such a comment is perfectly in tune with the IGM respondents, based on their comments.

(I'm sure the Congressional Budget Office does a better job of this thorough ins-and-outs analysis, but my impression is that they are mostly concerned with budgeting, not cost-benefit analyzing the bill? Is that impression wrong? As in, a bill under consideration can be a net loser from a government budgeting standpoint but pass a cost-benefits analysis from the point of view of society as a whole, conceivably. )

BTW, there is a second question on debt-to-GDP, and economists understandably agree that the tax bill will raise the debt. Sheer accounting is probably good enough here, save some ultra-supply-side argument.

Persistence of Drug Use

Here are some numbers from the 2014 National Survey on Drug Use and Health by the Substance Abuse and Mental Health Services Administration. (Henceforth the SAMHSA survey.) It's available here. (Pages 120 and 2071; it is a very large pdf.)

[Click the table for a clearer view.] As a first-blush look at drug use persistence, we can take some ratios. How many lifetime users used in the past year? How many past-year users had an "abuse disorder", suggesting some kind of problem controlling their behavior? I don't want to pretend this will tell us which drugs are truly addictive and which aren't, but I thought it might be suggestive.

A few things jump out at me. The once-and-your-hooked thing is utter nonsense and isn't true of any drug. Compare past month use to lifetime use, or past year abuse disorder to lifetime use. These are all very small numbers. It looks like heroin and tobacco are particularly addictive by these measures. No surprise there, I hope. No other class of illegal drugs seems to come close.

Persistent use does not equate to addictiveness, so we need to be careful here. Looking at past month users per lifetime users, marijuana stands near the top of the list of illicit drugs (though alcohol and tobacco have it beat). Marijuana isn't addictive in the same sense that heroin and cocaine are addictive; at most it is "habit-forming" in the same way that video games are habit-forming. 19% of lifetime marijuana smokers imbibed within the past month, compared to 9% for lifetime heroin users. Somewhat paradoxically, the high persistence of marijuana use may be related to its non-addictiveness. If users can pick it up and set it back down easily, why not? On the other hand, someone who once had a cocaine or heroin habit might realize that imbibing now is making a categorical choice about their lifestyle and plunging back into the abyss.

There is a very high persistence for alcohol use, but a relatively small number of people with an abuse disorder. I think the concept of controlled but persisting recreational use needs to be introduced into the drug policy conversation. Plainly there are a lot of casual drinkers. These people will never have issues with their drinking and some of them may never even get drunk. Surely there are analogs to this behavior pattern for the other drugs. Maybe someone smokes a bowl once a week, but puts it down for the rest of the week. Maybe someone uses methamphetamine/Adderall to concentrate and meet a deadline, but cycles off it after the project is done. See my previous post on Unbroken Brain, which mentions that many heroin users are on a very strict schedule of once a week. This is probably not frequent enough to develop a tolerance or persistent withdrawal symptoms. The point here is that controlled use of a substance is possible, and even periodic use does not inexorably devolve into full-blown addiction.

I really wish that they had more questions about frequency of use, maybe a "past week" question and then a "number of days in the past month" question. Once a month is really nothing. Even if someone stayed in the "used in the past month" category for years, it wouldn't necessarily indicate a drug problem. A "past month" user could still be an infrequent user. It might be policy-relevant to know who is using a few times a week or on an almost daily basis. And once again, even frequent use doesn't necessarily mean the person has a problem.

I wonder if the criteria for an "abuse disorder" might be overly broad. 12% of people who used marijuana in the past year had a "marijuana abuse disorder?" Yeah, right. 6% for hallucinogens? Bull. Hallucinogens aren't habit forming. People who like tripping balls quickly learn that they can't take, say, LSD or psilocybin on consecutive days because it stops working. I wonder if people who get arrested and sent to forced treatment for some of these non-harmful substances are automatically counted in the total? From the link:

The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), no longer uses the terms substance abuse and substance dependence, rather it refers to substance use disorders, which are defined as mild, moderate, or severe to indicate the level of severity, which is determined by the number of diagnostic criteria met by an individual. Substance use disorders occur when the recurrent use of alcohol and/or drugs causes clinically and functionally significant impairment, such as health problems, disability, and failure to meet major responsibilities at work, school, or home. According to the DSM-5, a diagnosis of substance use disorder is based on evidence of impaired control, social impairment, risky use, and pharmacological criteria.

So there are gradations of abuse disorders. If I was hung over and missed a class, does that count as "mild?" I'm guessing that the distribution of "severity of abuse disorder" is like almost every other social problem: the mild cases are the most common, moderate less so, and severe even less so. Again, it would be good for the SAMHSA survey to have finer gradations. The mild, perhaps even moderate, abuse disorders might not be "policy-relevant" in the sense of causing massive social harms. Incidentally, I hate this little trick. Someone defines a problem by pointing to the worst possible cases, but then lower the threshold to include stuff no reasonable person would worry about in order to get a big, scary number. I'm tempted to take the 6% for hallucinogens or the 12% for marijuana and simply subtract it from the other numbers as a "bullshit adjustment factor." (Not really. I'd rather just say, Here's the best we can do with bullshit numbers, and here are the caveats. Making "adjustments" to fundamentally bad data doesn't seem valid.) I'd go so far as to say that the "abuse disorder" figures are almost completely useless, or at best than merely suggestive of what they claim to measure. This annual SAMHSA report frequently uses these numbers to say, "Only [worryingly small percentage] of people with an abuse disorder actually sought treatment in the past year." I'm sympathetic to the idea that some severe addicts who need help can't get it, but my overall impression is that most of these people don't need serious help.

I think there is a selection bias at work here, so I would be hesitant to compare the legal drugs to the illegal ones. I strongly suspect that the kinds of people who use illegal drugs are distributionally/demographically different from the kinds of people who use legal drugs. Someone who tries heroin or cocaine is likelier to be highly impulsive, low-conscientiousness, habit-forming, and risk-neutral or risk-seeking. That certainly isn't to say all illegal drug users are like this, just that whenever you take a tiny sub-sample of the population who by definition are law-breakers, you're going to pick up a larger fraction of people with social and behavioral problems. Dip your ladle into the pool of "alcohol users" (~2/3+ of the entire population) and you basically get a representative subset of the population. Dip your ladle into the pool of "heroin or cocaine drug users", and you're going to get lots of very nice, normal people plus a disproportionate number of anti-social types. Given that, I think the "substance abuse disorder" figures probably overstate how dangerous these substances would be in a regime of legalization. The kinds of impulsive risk-seekers who are likely to have drug problems are already using drugs; they aren't deterred by the law. Those nice, normal people who don't form nasty habits, we don't need to worry about. Perhaps a few of these people might become occasional pot smokers or even heroin users, but very few of them are likely to become full-on addicts.

I'd like to see caffeine on here as a sort of test case. It's not really harmful, often quite the opposite, but definitely addictive.

Prohibition advocates often cite the pervasive use of tobacco and alcohol as an argument for keeping other drugs illegal. Certainly you can see that the legal substances are more widely used. But people don't just pile-on, they trade-off. People wouldn't just start imbibing all manner of substances on top of what they're currently doing. People face constraints like money, time, limited willingness to endure damage to their own health, and negative interactions between substances (some simply unpleasant or buzz-killing, some positively deadly). This argument by drug warriors never made sense to me. It seems more likely that people's willingness to tolerate self-harm is the limiting factor, and expanding their means of accomplishing it doesn't change the total harm done. There just aren't that many people who think, "I'm not going to destroy myself with alcohol or tobacco. But if meth and crack become viable options, I'm in!" Or rather, I should say there are such people but for the billionth time this kind of person is not responsive to legal penalties. Drug prohibition doesn't work for them.

That's all I got for now. What do you see?

“To the highest bidder” As a Sneer

There is a ridiculous sneer that goes something like: "Under capitalism, everything goes to the highest bidder!" I was reminded of this trick when I saw a recent tweet by Bernie Sanders. The sneer makes it seem as though under capitalism "the rich" just get their way with everyone because they have more money. This is kind of absurd even when applied to consumer goods. Sure, the rich have a lot of disposable income, but it's limited. Their purchasing power is smaller than the combined purchasing power of the middle and lower classes. People set up shops and stock them with goods for poor people, too. It's not like "the rich" totally dominate them by buying all the stuff and leaving nothing behind. Besides, most rich people save a lot of their income. They don't consume, which leaves all that stuff for the other people to consume. It gets bid upon by lower-income individuals.

The "To the highest bidder!" sneer, nonsensical as it is when applied to consumer goods, makes less sense when applied to capital goods. "No fair! Under an auction system, radio spectrum will simply go to the highest bidder!" Well, I say, usually it's a good thing when something get allocated to its highest value use. If you don't really give it much thought, you might envision this world where a few rich moguls call all the shots. The Sumner Redstones and the Rupert Murdochs of the world just use their massive wealth to decide what the media looks like. And this naive view gets it exactly backwards. The "rich" part most likely came second. These men grew rich (or richer) by selling people what they wanted to buy. You or I may think their product is meretricious, but they make their fortunes by insightfully predicting what these unwashed masses want to see. And if they try too hard to "call the shots," someone else will create a product that is more in tune with consumer demand. Even if they are willing to squander a large sum of their fortunes producing the media that they think consumers should want, unless they sell a product that consumers actually want they will go out of business. Even if the media mogul is an incorrigible old coot bent on a social mission, s/he will eventually be voted out by the board of directors or compelled by their heirs to step aside. These kinds of people usually have partners, and these partners aren't willing to lose money forever on some rich guys vision of social engineering. Even the rich get outbid.

Imagine someone who says, "I'm willing to buy this plot and build a factory on it. I'll pay $1 million." A second bidder says, 'I'm willing to buy the same plot and build a factory on it. I'll pay $10 million." You shouldn't conclude that the second buyer simply has 10 times as much wealth backing him, and is just blowing his stash on frivolous new ventures. You'd probably conclude that the second bidder has a higher valued use for the property. Someone, perhaps the bidder himself bidding his own money, or perhaps a series of venture capitalists and other investors, did some due diligence on the proposed "new factory" project. They decided that the resulting revenues would justify a bid of up to $10 million. It's not that they start off richer. Rather it's that they expect to produce something that they can sell to consumers for lots of money. This isn't a case of "rich guys throwing their weight around." It would be more accurate to describe this as consumers hiring capitalists to build a factory and produce consumer goods for them. In reality, the consumers don't know about the factory builders' plans and don't even realize they want the factory's output. But the capitalists have market research, management experience, and some intangible je ne sais quoi business acumen for the next big thing. And they might be wrong. Maybe the $1 million dollar bidder's idea is better. But capitalism lets a million little flowers bloom. There may be cases where the better idea gets "outbid" by someone with inferior ideas but a superior bankroll. But this is unlikely to be the case consistently, for the millions of big and little auctions that happen in our vast economy. Over time and over a large enough sample size, the better ideas outbid the worse ones because they are better ideas.  Not simply because a few rich guys call the shots. The "rich people's ventures perpetually outbid the less-funded but more insightful and meritorious ventures" scenario is paranoid in the extreme.

The sneer makes even less sense when applied to non-rival goods. That is exactly the case with the Sanders tweet. There are times when bandwidth is scarce. If we're all trying to watch Netflix at 9:00 p.m., or we all want to binge-watch the new Punisher show the night it comes out (okay, sorry, enough about me), then bandwidth becomes scarce. My use of the internet slows down your use of the internet. If we're just reading text articles and stupid blogposts all day (okay, sorry, seriously, enough about me!) we have plenty of bandwidth and we aren't hogging network resources in any meaningful sense. This notion, that this marginal internet  user will be perpetually outbid by "rich people" in a way that makes the internet unusable, is nonsense.

(Thinking of my own internet consumption. I'm an actuary married to a physician. And yet we shopped for the best data plan. We basically picked the cheapest plan at 1 or 2 Gb a month, I don't remember which. We didn't just buy up as much data as we could get. The notion of throttling other users to dedicate bandwidth to our phones never even occurred to us. I seriously doubt that we're atypical, either. I know other doctors and actuaries, and they shop around for the best value, too. I seriously doubt that even corporate executives try to purchase the "throttle the network and send it all to me!" plan. I try to imagine the "rich people buy up everything and leave nothing for anyone else" scenario playing out, and when I try to picture it it just seems unbelievably paranoid.)

Very early on, economist Ronald Coase explicitly advocated allocating radio spectrum to the highest bidder. He had in mind a rationale similar to the one I give above. Basically, allocating to the highest bidder means allocating to the highest valued use. And historically that has worked out pretty well when tried. Things worked poorly when we've deviated from it. In an allocation-by-fiat rather than allocation-by-auction regime, spectrum gets allocated based on somebody's idea of what's socially beneficial, or it gets allocated by political favoritism (which is exactly how LBJ and his wife made their fortune). If someone has a novel idea for the use of spectrum, the bureaucrats who allocate it can simply scoff and say it will never work. So a new technology may take years or decades to get off the ground. (Again, read the Grumpy Economist piece.) This doesn't happen under a "to the highest bidder" regime. The upstart business with the hot new idea runs some numbers, figures out what are the likely revenue scenarios, and buys the spectrum if it can justify the cost. It doesn't have to plead with bureaucrats, beg politicians to change the relevant laws, or ask permission from its competitors to operate. In these kinds of auctions, it is ultimately the consumers who are bidding with their purchases, not the business owners themselves. There would be no money for them to bid with in the first place unless they can win over customers.

Political Popularity as a Constraint

Suppose you favor Policy A, and you are arguing with someone who favors a less optimal Policy B, but you both agree that either choice is superior to a third Policy C.

You insist that Policy A is optimal. Your opponent insists that if we try to implement Policy A, a populist backlash will turn it into Policy C. If we instead opt for Policy B, we won't get the backlash. We might even gradually drift toward Policy A, but we need B as a stepping stone. You say that's all very well, but plainly A is better than B if only we get it implemented. Questions of realpolitik aside, if A is better than B we might as well say so. Your opponent again cautions you that you can't simply pick policy in a vacuum. Those brutish masses clamoring for bad policy are part of the world, and we can't simply assume them away. "A brief stopover in Policy A followed by collapse into Policy C" is no progress at all.

I recently saw something like this play out between Don Boudreaux and Rick Manning. It was on the question of free trade. Boudreaux favored unilateral elimination of tariffs (call it "Policy A"). Manning favored trade agreements that bilaterally (or multi-laterally) eliminate tariffs (call it "Policy B"). Both were duly afraid of high tariffs, protectionism, and economic nationalism (call these things "Policy C"). Boudreaux was taking the position that even unilateral elimination of tariffs would benefit the nation that does it, regardless of whether its trading partners reciprocate. It's not a prisoner's dilemma, in which both parties need to cooperate to get the reward. Either party can unilaterally make itself better off by eliminating its own tariffs or import quotas. Manning's argument was that unilateral elimination of tariffs leads a nation's citizens to feel like they're getting a bum deal, and protectionism raises its ugly head. It's frustrating to watch. Boudreaux is saying, "Sure, there are political constraints that lead us to pick bad policy. But Policy A is the best possible choice. I might as well say so." And Manning is essentially saying, "Political considerations are hard constraints on what we can actually do. Better choose a path of trade liberalization that doesn't lead to populist backlash." These are my own paraphrases, not actual quotes. (BTW, I thought Manning was rude and made borderline sexist remarks. "I can't support an import ban on chocolate because my wife would object!" And in case you didn't get this knee-slapper the first time, he repeats it. His dismissal of Boudreaux's graduate assistant's scholarship struck me as incredibly rude. But I absolutely take his point about acknowledging the realities of politics.)

I found this interesting. It is usually the libertarians who are reminding leftists and conservatives of political constraints. "You don't just get Obamacare as envisioned by its authors, rather you get Obamacare as implemented by a future Republican congress, and implemented in states with Republican leadership and Republican governors. You can't just say it's good policy in a vacuum, assuming away all of those nasty people who don't agree with you." The political process imposes hard constraints on what kinds of policies are actually feasible. If your policy to correct market failures requires an all-wise and infinitely compassionate bureaucrat to actually implement it, it's going to fail. Even if it requires merely above-average-yet-still-mortal intelligence and civic-mindedness, good luck finding enough competent and self-motivated people to run your bureaucracy. In fact, forget that. Good luck getting your grand scheme passed through the political process without being hopelessly mangled into something else. A political favor here, a carve-out for government unions there, a concession or two to legislators who are up for reelection this year, and the grand scheme to optimize society gets morphed into something nobody actually wanted.

(Note that this post is not about Obamacare. I could write an identical example about neoconservative nation-building being hamstrung and underfunded by Democratic opposition and eventually taken over by a future Democratic leadership and perhaps abandoned by an exhausted or apathetic voting public. If you can imagine such a thing happening. This constraint is bipartisan.)

It's interesting to think about how this same political hobbling might apply to libertarianism. A grand mobilization to shrink government could conceivably backfire, and the populist backlash could leave us with more government (or anyway worse government) than we had before. I'm not sure how much I really buy this idea, but it's definitely something libertarians need to think about. Maybe legalizing drugs suddenly, rip-off-the-band-aid style, would lead to a temporary surge in overdoses while people figure out which drugs are truly dangerous and in what dosages. Maybe this would lead to a crackdown and leave us with even worse drug policy than we currently have (supposing that's possible). Or maybe we'd have blown the one chance in a generation to accomplish drug liberalization. The traumatic experience will take even milder liberalizations off the table, even if a temporary rise in overdoses is a perfectly acceptable loss from a utilitarian point of view. Most members of the voting public just don't have the patience or the bandwidth to process these policy questions. Stark yet misleading examples loom large.

Likewise, maybe a temporary experiment with open borders leads to a few minor terrorist attacks. Opponents of open immigration say, "See!" The voting public, oblivious to the utilitarian calculation, over-weights a few murders without comparing them to the millions of lives lifted out of poverty. They say, "Yes, I see. Close the borders down!" Again, the experiment could ruin immigration reform for a generation.

Maybe a tax reform includes lots of future cuts to government spending, but if those cuts don't actually happen we'll just end up with higher and less efficient taxes in the future. And maybe no sane person believes those cuts will actually happen.

I still think I should state clearly and honestly what the best policy is, without regard to political considerations. I'd love to see open borders, or even nearly open borders with basic background checks. I'd also love to see something more likely, like expanding immigration from ~1 million a year to ~2 million a year. But I think vastly liberalized immigration verging on fully open borders is better policy, and I might as well say so. I also think that complete and instantaneous drug legalization is probably better than a slow draw-down of existing penalties and barriers. I sometimes argue in favor of these other policies, like punitive taxes or user licensing, that achieve what current law tries to achieve more effectively and humanely and cheaply. "Drug legalization light" is probably more politically palatable and less likely to cause a backlash. But I still think that full legalization without guardrails is the better policy, and I feel I should say so honestly. If I were in a position to actually make a political deal and shape policy, I'd change my tune.

I also think it's possible to change the minds of the voting public. I know it's crazy, but I've seen it happen. If we say "Policy A is better than Policy B" often enough and clearly enough, maybe the voting public will come around. We shouldn't have to settle for "sub-optimal but palatable" policies. So there is a niche for people like Don Boudreaux, who ignore the political constraints and openly speak their minds.

__________________________________________________________

I feel like two different concepts crept into my discussion of political constraints. One is popularity. Given enough political will, this constraint disappears. If there is political will for immigration reform, drug legalization, or tax reform, it will happen. The other concept is intrinsic government inefficiency. The incentive structure for government is inherently different than it is for voluntary institutions. You can't just "quit" a bad government program; you have to continue paying for it and perhaps submitting to it or being punished by it. All the political will in the world doesn't fix this problem. Unless the voting public becomes infinitely vigilant, government institutions will not be as responsive to their "customers" as voluntary institutions. Bureaucracies will seek to maximize their budgets while individual employees seek to minimize their workload. Some gunner-bureaucrats may seek to maximize their authority and acclaim, perhaps in pursuit of a nicer job or political career. Private bureaucracies have these problems, too, but they more quickly run up against budget constraints and the vetoes of shareholders and ultimately their customers. I meant for this post to be about the first kind of constraint, but lapsed into talking about the second kind.

Trolls: An Important Niche In the Internet Ecosystem

As much as I despise people who deliberately or negligently derail a perfectly good thread, I want to give trolls their due. Note that my definition of "troll" is perhaps more expansive than the official one. There are plenty of people who aren't deliberately trolling but are so incapable of controlling their outburst that they might as well be doing it on purpose; these folks are just as deserving of our opprobrium as self-conscious trolls.

First of all, an unscripted opinion is a valuable thing. The tamer, more refined conversations (like you might see on Steven Landburg's The Big Questions blog or Scott Alexander's Slate Star Codex) are sometimes too sanitized. That very nice person who is trying to counter your argument might have had initial thoughts such as "You're an idiot" or "You're an asshole." It's good etiquette to suppress these reactions and give a more considered response. But doing so conceals some information about your interlocutor's mental state. It's occasionally useful to have this information, and trolls provide it freely. The problem with giving this information up too freely is that it leads to escalation, even if done delicately. "Oh yeah? Well I think you're an asshole!" So it's probably best that we politely keep these things to ourselves, but use trolls to reveal hidden information.

Also, even really bad counterarguments can be useful. It's like a game of "'Why?' Boy," in which you keep asking yourself how you know something to get at deeper and deeper layers of understanding. This process is named after the stereotypical small child who constantly asks "Why?" to each one of your successive explanations. Even if you yourself fully understand the details of your argument, all the way down the rabbit hole to the deepest layers of comprehension, not everyone will grasp it on the first pass. A troll who is playing "'Why?' Boy" with you forces you to refine and recast your argument, making it simpler to understand and digest. Don't pass up the opportunity to respond to a very bad argument. It may not do the troll any good, but an onlooker might be impressed and actually learn something.

I hope there are more elegant ways of getting information about people's emotional reactions, or refining your arguments, such that we can do away with trolling. At least for the second thing, you can practice on your own. Just imagine someone blurts out a flippant response and try to respond to that. I try to do this, and it pays off. I often anticipate a counterargument ("anticipate" meaning "to plan for", not merely "to predict"), so I have a ready answer. And, of course, sometimes I say something wrong and someone corrects me, often politely but sometimes not. I learn from these exchanges, too. (Sometimes, it's fun to play, "What if I'm the one being a troll?" It forces you to be more charitable to someone who appears to be spouting gibberish; always entertain the possibility that they aren't but that you are.)

I'm not quite saying "Hail trolls!" More like, "If we're stuck with these shitty behaviors, we might as well make the best of it."