Empiricism and humility

by on March 15, 2017 at 1:13 am in Economics, Philosophy, Uncategorized | Permalink

Here is a Noah Smith post on those topics, and here is Adam Ozimek, both responding to Russ Roberts.  Rather than adjudicate the varying points of view here, I will stress some points of my own:

1. The political process does not select for humble versions of empiricism.  Those end up with virtually no political influence, whereas some of the more dogmatic form of empiricism may find some traction.

2. A lot of the bias in empirical methods comes simply from which questions are asked/answered.  Post Trump and De Vos, I see plenty of commentators and researchers reporting “vouchers don’t raise test scores” and virtually no “vouchers increase parental satisfaction.”  Is that empiricism?  In isolation, maybe.  In terms of reflecting the broader spirit of science, not so much.  It is also not humility.

3. I also see bias in terms of framing and contextualizing.  One empirical result is “over a short time horizon, a $15 minimum wage in Seattle hasn’t destroyed many jobs.”  Another empirical result is “rises in the prices of inputs virtually always lower input demand, with larger effects over longer time horizons.”  There is also “non-pecuniary factors of jobs adjust downward, in response to wage minimums, thereby removing the benefits for the workers from the wage hike.”  One side claims the mantle of empiricism with #1, the other side claims the mantle of empiricism with #2 and #3.  Overall the course of that debate does make me more skeptical about “empiricism as we find it,” though not about proper empiricism.  And note that the scholarly division of labor does in fact give any particular individual a sufficient excuse not to be doing the task of overall synthesis.

4. I find a very common pattern among both researchers and commentators.  They first form broadly empirical judgments about social systems, based on overall views of history, current politics (too much), and some of their relatively general empirical judgments, such as whether elasticities are large or small, or the relative crookedness of politicians vs. businesspeople, or the relative competence of voters, and so on.  Those are empirical judgments, though usually in non-formal, non-directly testable ways, and also inter-smushed with ethical judgments, for better or worse.

They then view very particular empirical debates through the broader lenses they have chosen.  For instance, views on politics used to correlate with views on the interest elasticity of money demand.  Today views on politics correlate with views on minimum wage elasticity, and so on.

It’s the kind of empiricism outlined in the first paragraph of #4 that has the greater predictive value for beliefs.  Furthermore it is sometimes (not always) the more important form of empiricism for settling many questions of policy.

5. I am sympathetic with the view that the broader empiricism outlined at the top of #4 is overused.  Yet many of the critics of that broad approach simply wish to protect the presuppositions of the academic status quo from being disrupted by the possibility of other broad paradigms.  In other words, I worry that criticizing “ideology” is too often a means of cementing in the dominant ideology in academia (and journalism), rather than an actual critique of ideology.

6. Most generally, humility is always scarcer than one might think.  Perhaps that should be one of Cowen’s Laws.

1 Dzhaughn March 15, 2017 at 1:45 am

Please note that Seattle hardly has a $15/hour minumum wage. The $15/hour rate started Jan 1, applies only to firms with >500 employees who do not contribute to health insurance. It is $13.50 for those who do. The minimum wage is $13.50 (resp. $11) for smaller firms with (resp. without) health benefits. On Jan 1 2019 everyone will be at $15.

The Washington State minimum wage is $11.


2 Anon7 March 15, 2017 at 3:31 am

Please note that the labor cost for large employers is essentially $15/hr regardless of whether it’s $15/hr without health benefits or $13.50/hr with health benefits–that is unless you believe in the health benefits fairy who provides stuff for free.


3 DBonar March 15, 2017 at 6:56 am

But if the firm is already paying 13.50$ and health benefits, it doesn’t rise to 15$ plus the health benefits. The point is that if there are firms paying at that level, the law hasn’t effected them yet.


4 Art Deco March 15, 2017 at 7:43 am

None of the characters he linked to discussed the specific value of Seattle’s minimum wage and neither did he, so to just whom do you think you’re responding?


5 Mark March 15, 2017 at 9:00 am

The Smith post includes a discussion of Seattle minimum wage in its argument. fwiw.


6 Art Deco March 15, 2017 at 9:32 am

He never discussed specific values, so what is the complaint?


7 mulp March 15, 2017 at 12:58 pm

What is the value to you as a seller of your customer having a $15 wage instead of a $10 wage?

Or, to put it a different way, why would you prefer that your customer or client have a $10 wage instead of a $15 wage?

If you prefer the lower wage from a retailer or medical professional standpoint, why are there so few of such business people in coal country and the Rust Belt? If low wage customers were the best, wouldn’t these regions be attracting greedy business men?

If you were Trump, would you prefer $10 an hour workers as customers, or $15 an hour customers at your hotels and golf courses?


8 TallDave March 17, 2017 at 11:53 pm

Since sellers can only set prices and are generally dependent on consumers to make the actual buying decision, the correct question is: how much you willing to pay to ensure all the labor inputs to your products and services receive at least $15/hr? If high labor costs are best, shouldn’t Trump’s attacks on imports be attracting legions of socialist consumers?


9 Paul March 16, 2017 at 4:00 pm

Given the influx of tech into Seattle the market minimum wage might be going up anyway. So the $15 minimum might have been moot all along.


10 NatashaRostova March 15, 2017 at 2:00 am

Excellent post, please share more thoughts on empericism. This is my favorite post of yours in months.

The way people develop an inconsistent or overfitted philosophy of science, in order to fit their political tribe, is such an incredible abuse of the scientific method, and so many otherwise brilliant professors (of economics no less) make this same mistake. Helping people see that their empirical calibration isn’t functioning well is almost impossible. Motivated reasoning is one hell of a drug


11 Shane M March 15, 2017 at 2:04 am

+1. Good post, as well as the links.


12 Elan March 15, 2017 at 3:26 am


The obsession with claiming one’s ideology has a monopoly on the “reality based community” or making hats with the word “FACTS” on it (http://www.slate.com/articles/slate_plus/s.html) is one of the more grating parts of reading the main stream press during the Trump era.


13 tjamesjones March 15, 2017 at 5:41 am

the phrase “evidence based” has been in currency for a while. it’s close to meaningless (a) implying anybody who disagrees just doesn’t like evidence when in fact (b) all it means is that it’s possible to find evidence to support a point of view (but the point of view came first, trust me)


14 Mark March 15, 2017 at 9:15 am

This was an excellent post.

The comments on Smith’s blog are also outstanding.

I found Smith’s post very trite and shallow. For example, he posits that it’s quite simple to discern whether Rothwell or Autor et al are right but can’t bother to lay out how he does it, let alone showing how his analysis is, at every step, purely and unassailably empirical and never once resorts to a norm, a hunch, a preference, a shortcut, etc. Being smart is easy when you know you’re right.

I think – empirically – the enormously high correlation between a given economist’s theoretical beliefs and the empirical work s/he chooses to credit and accept is enough to discredit Smith’s argument.


15 Noah Smith March 15, 2017 at 3:51 pm

Wait a sec, no. I don’t think it’s simple to tell whether Rothwell or Autor is right. I think it’s really hard!! I’m trying to do it myself, and I’ve still only gotten to the bottom of about half of the points of contention.

I’m just saying, I think it’s *possible*.

A pretty clear-cut case is Borjas vs. Card: https://www.bloomberg.com/view/articles/2015-12-18/an-immigrant-isn-t-going-to-steal-your-pay-raise
But even in that case, finding who was right required some deep digging into the data. This stuff ain’t easy.


16 Thanatos Savehn March 15, 2017 at 10:33 am

Exactly so.


17 prior_test2 March 15, 2017 at 2:01 am

‘Most generally, humility is always scarcer than one might think. Perhaps that should be one of Cowen’s Laws..’

Best satire site on the web.


18 The Centrist March 15, 2017 at 1:28 pm

If anything he’s satirizing himself.


19 Whatever March 15, 2017 at 2:32 pm

Cowen’s Law #x: “‘Most generally, humility is always scarcer than one might think, even after taking Cowen’s Law #x into account”.

By the way, for what it’s worth, I also think this post is excellent, one of the best in a while.


20 Ray Lopez March 15, 2017 at 2:10 am

Like the Mac Davis (he also wrote Elvis’s “A little less conversation”) song “Oh Lord It’s Hard to be Humble, when You’re Perfect in Every Way”.

But the truth is, economists should be more humble, since partial equilibrium is the best they can achieve, there are multiple saddle points in an economy, and the economy behaves largely as a nonlinear system. Many economist models cannot beat a random walk. But that doesn’t stop them from spewing forth papers at 5% significance level and pretending they have the last word. Hucksterism. At least our TC doesn’t do that (since I don’t think he publishes e-CON-omics paper).

Bonus trivia: I see GM Rogoff in his younger days had a penchant for elegant math models explaining various economic forces. But these elegant ‘closed form’ solutions are likely of little to no importance, except as a resume enhancer. Then again, that’s modern economics.


21 carlospln March 16, 2017 at 12:55 am

Ray Lopez knocks it out of the park.


22 Yancey Ward March 15, 2017 at 2:39 am

When someone tells you they are humble, do you believe them?


23 Axa March 15, 2017 at 3:26 am

First, consider the option that the prison is really humble. Then, assess the claim in the context of the person’s actions.

Assuming that everyone that claims to be humble is lying is a perfect example of misguided empiricism.


24 Amigo March 15, 2017 at 3:35 am

“This past year I’ve gotten very rich, famous, and humble.” -Amy Shumer in her latest special.


25 Art Deco March 15, 2017 at 7:37 am

No. And the only public figure I can think of who has suggested he is would be Pope Francis, a man who has much about which to be humble.


26 ChrisA March 15, 2017 at 3:31 am

Personally I don’t like the word humble to describe this approach of being cautious about drawing conclusions, it’s too suggestive of an attitude of moral superiority. I prefer to think of myself as sceptical rather than humble. Scepticism is warranted when anyone is either 1) making claims based on very short run data, or 2) arguing from conclusions, 3) using correlative data as causative or confusing a model for reality, or 4) arguing from consensus. This doesn’t mean you can’t argue for action on certain things where the outcome is uncertain. Like the school vouchers, my attitude is a bit like Milton Friedman’s on capitalism vs communism – “even if the empirical evidence says it is less efficient then I would still prefer it “.


27 Pshrnk March 15, 2017 at 9:20 am



28 Todd K March 15, 2017 at 12:17 pm

It’s getting a little rediculous seeing so many people spell skepticism wrong recently.


29 ChrisA March 15, 2017 at 3:32 pm

Its the British spelling.


30 jos March 15, 2017 at 4:03 am

As an economist who got his PhD in the 21st century,, this whole post reads as an old man, who isn’t a data native and never viewed every question as empirically testable as trying to find his way in a new world that has little use for his truth-as-stated blogging without evidence.


31 tjamesjones March 15, 2017 at 5:42 am

i actually hope you’re NOT joking.


32 Lanigram March 15, 2017 at 10:58 am

You sound like an arrogant little prick. I hope I’m there when your wings melt and you hit the earth with a dull thud.

Eventually, gravity brings us all down.


33 The Centrist March 15, 2017 at 2:03 pm

Empiricism (data nativity) has its place, no doubt. But, recall the second clause of Kant’s saying: “thoughts without content are empty, intuitions without concepts are blind”.

I see Tyler Cowen as someone interested — as he was from the early “culture” books on down — in big questions that cannot easily, if at all, be tested. Such as complacency.

Maybe Cowen is offering economic fables, in Rubinstein’s sense. Not exactly models and not exactly testable questions, but not just making it up.


34 Art Deco March 15, 2017 at 1:49 pm

You sound like a young man who doesn’t realize that multiple regression analysis has been a tool researchers have made use of for 50 years.


35 NatashaRostova March 15, 2017 at 2:48 pm

This is a very depressing comment :'(

I know Causal Inference is doing some crazy awesome things. But so many findings (e.g. the minimum wage example given above), can be undetermined using different, strongly robust, economic methodologies, which imply different things.

It’s not that you can’t get an estimate for minimum wage impact in a given city, and perhaps a good one, using cutting edge Causal Inference techniques. Rather, that all you’re left with is a distribution of what we observed in a given city, at a given time, but generalizing it out contradicts other research. The reality perhaps being there is an even more complex dynamic that isn’t revealing itself to a human observer.


36 Zeitgeisty March 15, 2017 at 4:30 pm

Data-driven analysis is useful only within specific domains like predicting election results and advertising clicks

Tocqueville understood society better than the questionnaire toting sociologists of the 20th century and the data scientists of today.


37 Rich Berger March 15, 2017 at 6:27 am

I think I agree with TC, but as usual, he overcomplicates and overthinks it. Reality is far too complicated for the human mind to understand, but we can do very well analyzing small areas. We can do very well when the small area allows repeated experiments so we can see make predictions and adjust our theories after the results are in. This works very well for the physical sciences and engineering – now we can reliably fly aircraft, construct tall buildings and send rockets and probes to distant places. We can build cars that last for years and protect us in accidents. The list goes on.

When we turn to large, complicated systems, such as global climate, national and international economies and the reaction of the human body to diets, the human mind is overwhelmed. The factors involved are practically infinite and the web of causes and effects is too complex to isolate and test. Humility would recognize this state of affairs, but that is unacceptable to those who Want To Do Something.

Since the “progressive” is not results-oriented, but rather devoted to showing good intentions, the outcomes of tinkering are not really important. For example, take the “war on poverty” or the desire to eliminate racial bias through anti-discrimination laws. Poverty has not decreased, except inasmuch as more people are dependents of the government. Overt racism is way down from the 50’s and 60’s, but the sensitivity to racial slights is greater than ever.

Russ Roberts’ counsel of humility is a direct attack and threat to the entire world of the Smiths and Ozimeks of economics.


38 [insert here] delenda est March 15, 2017 at 8:17 am

I agree. I think Tyler says as much, since most of his post is basically saying that Noah’s arguments are not “empirical” in any useful way.


39 lemmy caution March 15, 2017 at 9:54 am

The poverty rate in America was reduced dramatically by the war on poverty:


If your point is that this is because the government gave poor people money, then, yes, that is what happened.


40 TakeNote March 15, 2017 at 7:42 am

It is true that opponents of school vouchers sometimes conveniently omit to mention their positive effect on parental satisfaction. But voucher advocates (As far as I can tell) mainly advocate vouchers based on their supposed positive effect on test scores, not parental satisfaction. Both sides could do with more careful consideration of each other’s positions.


41 Miguel Madeira March 15, 2017 at 8:58 am

Since when “parental satisfaction” is even relevant? Parents are supposed to choose schools thinking about what is best for their children, then what could be relevant was “student satisfaction”.


42 TakeNote March 15, 2017 at 9:04 am

I agree parental satisfaction isn’t the most important factor, but if you can make parents more satisfied and therefore happier without reducing the quality of other aspects of the child’s education, then why not?


43 Pshrnk March 15, 2017 at 9:25 am

Student satisfaction is not the most important. I was satisfied with my 7th and 8th grade experiences in a progressive early 70s school at the time. A few years later I became aware of educational deficiencies, when thrown into a school with students who came from one where they had been expected to actually learn things such as grammar, algebra, geometry etc.


44 Thiago Ribeiro March 15, 2017 at 9:58 am

The most important, I think, is tests scores. And it has already been conceded, “vouchers don’t raise test scores”


45 Miguel Madeira March 15, 2017 at 11:44 am

Exactly; my point was that, if we should compute not only test scores, but also subjective feelings of well-being, these subjective feelings should be those of the student, not of his parents.

46 anonprof March 15, 2017 at 10:30 am

What is the empirical basis for the assertion that parents are supposed to choose schools thinking about what is best for their children? If the difference in outcomes from various educational options is small, while the difference in parental satisfaction is large why shouldn’t parental satisfaction factor in?


47 Rags March 15, 2017 at 11:11 am

Parents have free will, but so do communities providing education. Those communities should care more about the children. First, caring more about anonymous children is in our human nature. Second, the potential positive and negative externalities are much higher for the kids. Dad already is or isn’t a bum. Sonny has a chance to go either way.


48 carlospln March 16, 2017 at 1:02 am

People don’t have free will.

But, parents, & communities DO?

Please explain.

49 Miguel Madeira March 15, 2017 at 11:42 am

“Parents are supposed to choose schools thinking about what is best for their children” was a normative proposition, not an empirical one.


50 albatross March 15, 2017 at 2:50 pm

Parental satisfaction is probably related to how well they perceive their children to be doing. That’s a reasonable approximation for how well their children are actually doing in the school (in terms of everything–are the kids happy, are they making friends, are they able to do their homework, do they seem to be understanding what’s going on).


51 Rags March 15, 2017 at 11:06 am

I was going to say “Good God, man!” The fact that test scores go down as parental satisfaction goes up should set off alarm bells. One does not balance the other. The combination illustrates an error.


52 TMC March 15, 2017 at 12:42 pm

If they go down, why does the left take so much time explaining away why they go up?


53 rayward March 15, 2017 at 7:43 am

Cowen is very good at illuminating differences in the way people see things. Consider progress. What is progress? Is it a reduction in world poverty, an increase in world literacy, a reduction in premature deaths from starvation and disease, an increase in tolerance for gays and lesbians, a reduction in the number of mothers and children who die in childbirth? Or is it an increase in the number of people who attend church regularly, a reduction in the number of abortions, an increase in the number of children born to each woman, a reduction in permissiveness toward homosexuality, an increase in the number of families structured as a patriarchy, a reduction in the number of women in the workplace? A libertarian might define progress as anything that reduces the power and size of government. A progressive might define progress as anything that reduces suffering. And never the twain shall meet. The debate about the fate of Obamacare reveals difference in the way the competing sides view progress. On one side is the view that an increase in the number of people with health insurance is progress. On the other side is the view that a reduction in the role of government especially a reduction in taxes is progress. Empiricism is meaningless if the two sides don’t measure progress in the same way.


54 AlanG March 15, 2017 at 7:57 am

Economics is a lot like baseball; there is a lot of data (for the most part) but it really depends on how one analyzes it. For many years, baseball relied on human observation but then Bill James started digging deeper and sabermetrics was born. Even then it took someone such as Billy Beane to fully apply it and win some championships. We know that shark attacks were responsible in part for Wilson’s loss of votes in seaside New Jersey counties in the 1916 Presidential elections and that droughts and floods have had impacts on Presidential elections, including Al Gore’s loss to W in 2000. The data is there; just look at it. Sometimes it is short term and sometimes long term which is why the observer needs to be careful in interpretation (and also why there is a lot of junk science being done; e.g., bad interpretation).


55 msgkings March 15, 2017 at 1:26 pm

Billy Beane never won a single championship, never even got to the World Series.


56 Adovada March 15, 2017 at 8:08 am

Good arguments for states rights and policy experimentation.


57 rayward March 15, 2017 at 8:40 am

Here is Noah Smith’s column today in Bloomberg on this topic: https://www.bloomberg.com/view/articles/2017-03-15/how-to-restore-faith-in-economics It’s understandable that Cowen would suggest humility when it comes to empirical findings, Cowen having just published a book that relies (in part) on the cyclical view of history. Smith: “So be skeptical of new economic findings [empircism] that you read about in the news. But be far more skeptical of evidence-free pronouncements from people who claim to have access to a special mode of thinking. Evidence is imperfect, but it’s a big step up from conjecture and ideology. The empirical revolution should make you trust economists more, not less.” Call me simple minded, but the fight between micro and macro strikes me as missing the distinction between what can be measured in the past and what can be predicted for the future. Sure, what has been measured in the past might provide some clues as to the future, but history isn’t necessarily determined by the details of the past. History is determined by a mix of many factors, economics being only one. Smith’s “evidence-free pronouncements” may be someone’s else’s critical observation of the past. And that would include Cowen’s.


58 libert March 15, 2017 at 9:08 am

Critical observation of the past would include empirical facts about the past, yes? And if they do, they aren’t “evidence free pronouncements”. I think he’s just calling for people to support their claims with evidence rather than ideology, which I’m totally on board with.


59 wiki March 15, 2017 at 9:32 am

It is way worse than that. If you publish things that are too far outside the elite consensus your work is more likely to be desk rejected or subjected to harsh scrutiny than those that support elite views. Mistakes that the elite make that they don’t view as important slide through.

Worst of all, should you come up with work that violates your own prejudices and you bravely try to pubish it but in ways that involve nuance without a clear right or wrong, neither the scientific establishment nor the popular press is likely to be very supportive if the underlying narrative isn’t very clear. Especially if it seems to be damaging for all sides. This is not a problem though if you are high status enough and can therefore get brownie points for being complex. Anyone else though, and the answer is often indifference.


60 The Engineer March 15, 2017 at 9:34 am

The things we care about, public policy and such, are complicated, multivariate affairs. A multivariate phenomenon is naturally going to have a lot of possible explanations, and most “empiricism” is just going to be a convenient selection of variables that produce results aligning with a pre-determined narrative.

What you get is what we have: different narratives supported by different “facts”, and a devolution into arguments about what the facts of the situation actually are.

If we are honest, a lot of these public policy and other complex issues are in fact unknowable.


61 Dimitrios Halikias March 15, 2017 at 10:47 am

These sorts of discussions remind me always of Sheldon Wolin’s phenomenal (though rarely read) essay, “Political Theory as a Vocation.” The big data revolution in empirical social science is not so different from what he called the “methodism” of mid-century American political science:

“At this point a protest might be made that too much is being read into the idea of method. Methods per se do not presuppose a philosophical view of things, but are neutral or instrumental, analogous to the technician in being indifferent to the purposes of their master. Such an argument is not only wrong but superficial. In the first place, the elevation of techniques has important cuurricular consequences. The requirement that students become proficient in an assortment of technical skills preempts a substantial portion of their time and energy. But more important, training in techniques has educational consequences for it affects the way in which the initiates will look upon the world and especially the political portion of it. ‘Methodism’ is ultimately a proposal for shaping the mind. Social scientists have sensed this when they have noted that research methods are ‘tools’ which ‘can become a way of looking at the world, of judging everyday experience.’

In the second place, the alleged neutrality of a methodist’s training overlooks significant philosophical assumptions admittedly incorporated into the outlook of those who advocate scientific inquiry into politics. These assumptions are such as to reinforce an uncritical view of existing political structures and all that they imply. For the employment of method assumes, even requires, that the world be of one kind rather than another if techniques are to be effective. Method is not a thing for all worlds. It presupposes a certain answer to a Kantian type of question, What must the world be like for the methodist’s knowledge to be possible?

…This is but to say that there are inherent limits to the kinds of questions which the methodist deems appropriate. The kind of world hospitable to method invites a search for those regularities that reflect the main patterns of behavior which society is seeking to promote and maintain. Predictable behavior is what societies live by, hence their structures of coercion, of rewards and penalties, of subsidies and discouragements are shaped toward producing and maintaining certain regularities in behavior and attitudes. Further, every society is a structure bent in a particular and persistent way so that it constitutes not only an arrangement of power but also of powerlessness, of poverty as well as wealth, injustice and justice, suppression and encouragement.”


62 Art Deco March 15, 2017 at 1:42 pm

Shorter Wolin: It’s a problem because people concerned with empirical questions will take me less seriously.


63 Jonathan March 15, 2017 at 10:47 am

On humility and empiricism

Michael Lewis on Darryl Morey: “People who didn’t know Daryl Morey assumed that because he had set out to intellectualize basketball he must also be a know-it-all. In his approach to the world he was exactly the opposite. He had a dif­fidence about him—an understanding of how hard it is to know anything for sure. The closest he came to certainty was in his approach to making decisions. He never simply went with his first thought. He suggested a new definition of the nerd: a person who knows his own mind well enough to mistrust it.”


64 pyroseed13 March 15, 2017 at 10:50 am

This is one of Tyler’s best posts, and my own views on this issue are broadly similar. I read Russ’s post and thought to myself, “I agree that that economists can be too confident in their predictions at times, but then should we just discount all empirical evidence?”, and then I read Noah’s post and thought “We should do empirical economics, but not every issue can be resolved by appealing to data and models.” Here’s the way I see it: We can use the empirical side of economics to test claims, but our ability to determine which claims should be given the most weight in policy-making are ultimately going to depend on our values and priors.


65 albatross March 15, 2017 at 3:17 pm

What’s needed is a gauge of how certain people in a field are about a given result or claim, so that outsiders (voters) can decide how much weight to give that result or claim. If economists are pretty overwhelmingly confident that, say, raising the minimum wage to $15/hour won’t cause more than a 1% increase in overall unemployment rate, then that would be useful. But of course, people (including academic economists) have a strong incentive to phrase their claims in the strongest form possible, as it’s more convincing. If Alice says “The evidence is mixed and not so easy to interpret, but overall the best prediction we can come up with is that raising the minimum wage will have very little effect on unemployment,” and Bob says “It is absolutely clear, and known to every economist, that raising the minimum wage will put people out of work,” outsiders are likely to feel like Bob is more convincing than Alice.

What I really want here is a prediction/betting market. 100 economists each, with expertise in the relevant area of research, put a substantial bet (say 1% of their annual salary) on the claim that if the minimum wage is raised to $15/ hour, the unemployment rate one year later will be no more than 1% higher than on the date the new minimum wage passes. They gain 1% of their annual salary if they’re right and lose 1% if they’re wrong. As an outsider, that would give me some confidence that they were giving me their honest opinion.

Alternatively, they could go the Sam Wang route, and agree to eat a bug on live TV if their prediction turned out wrong.

Without making an explicit testable prediction and having some skin in the game, it’s just too tempting to amp up your claimed certainty to Krugmanian levels in order to strengthen the case for your preferred policies.


66 Max March 15, 2017 at 10:50 am

Why do people take Noah Smith seriously? He is almost always incorrect in one way or another.


67 Per Kurowski March 15, 2017 at 11:15 am

Here are the Basel Committee’s risk-weights used for determining the capital requirements for banks: For dangerous AAA rated = 20% – For innocuous below BB- rated 150%. I have to ask, what kind of shoddy empiricism is that? Don’t you care? Are bank regulations really so irrelevant?



68 Lanigram March 15, 2017 at 11:28 am

Best post ever – it addresses the issue at the core of human “thinking”. I like Jonathan Haidt’s analogy of the elephant and the rider to describe our thought and behavior. The elephant – the part of the brain we share with other primates – has been honed by tens of millions of years of evolution. The rider – the neocortex (?) – is a recent innovation that has evolved to serve the elephant. The elephant is in charge and it either likes or dislikes – approach or withdraw. The rider fabricates post hoc narratives to justify the elephant’s “decisions”. Though the rider can, slowly, train the elephant.

Haidt’s analogy does a good job explaining motivated reasoning, confirmation bias, mood affiliation, and all the other ways to describe our predictably irrational behavior.


69 Lanigram March 15, 2017 at 11:33 am

I meant to write:

“Though the rider can…train the elephant, he cannot make it do what it does not want to do.” – Haidt

I also want to say there are many excellent comments on this post as well, except for the arrogant “data native”.


70 David Condon March 15, 2017 at 11:43 am

These kinds of posts make me sad that philosophy of science has almost no traction outside of philosophy itself. Although Russ Roberts makes claims about the limits of sense experience, at no point does he make any claims that there are alternative sources of information or even might be. Then Noah Smith invents a word, intuitionism, without ever explaining why he’s not just saying rationalism. Now you’re suggesting Russ Roberts isn’t a dogmatic empiricist on the basis of those charges even though suggesting as Russ Roberts does that we shouldn’t make claims until we have strong evidence about those claims is a very empiricist thing to say. Perhaps Russ Roberts thinks theory should be developed more independently of data, or perhaps you think he practices in that manner, but there’s nothing here to suggest that. If people can’t keep straight a word as commonly used as empiricism, then just imagine the problems with keeping other philosophical terms straight.


71 Art Deco March 15, 2017 at 1:44 pm

Then Noah Smith invents a word, intuitionism,

He did not. John Rawls made use of the term in the introductory chapter of A Theory of Justice.


72 Noah Smith March 15, 2017 at 3:54 pm

Holy cow, that awful philosophy camp that my parents sent me to in junior high ended up paying off! 🙂


73 F.F. Wiley March 15, 2017 at 11:58 am

I agree with those complimenting this post and especially TC’s emphasis on framing.

The biggest drawback to minimum wage hikes imo is that they deter entrepreneurship. It’s hard to let your animal spirits out when officials can sweep in at any time and upend the reward/risk mix. (The margin between reward and risk, or prices and costs, is precariously thin for most businesses.) I doubt this unintended consequence of minimum wage hikes – or, more accurately, the mere possibility of minimum wage hikes – has ever been measured empirically or with any accuracy. I’m also guessing Noah Smith has never started a business and would never think to look at minimum wages from the (potential) businessperson’s perspective.


74 albatross March 15, 2017 at 2:47 pm

Re #1: Success in politics is about convincing people. Confident assertions about stuff you’re really not all that sure about work well for convincing people–much better than careful statements that acknowledge places where you don’t understand something perfectly.

Politics selects for that. So we get lots of very confident assertions that only an idiot could disagree with some empirical claim that is honestly not well-understood, whether that’s the claim that Iraq has WMDs, or that raising the minimum wage will have no impact on employment, or that greater government spending is the one certain way to get us out of this recession, or whatever.


75 msgkings March 15, 2017 at 2:52 pm



76 Thanatos Savehn March 15, 2017 at 4:47 pm

Every time I think we’ve turned a corner and have decided to base our decisions on evidence rather than wishes something like this ruins my day: https://www.statnews.com/2017/03/07/alternative-medicine-hospitals-promote/

Maybe the postmoderns are winning their war against “the tyranny of science” after all. Maybe most people really are more concerned about living bias-confirming narratives than maximizing their survival. Maybe sheeple are real.


77 Troll me March 16, 2017 at 12:45 am

2) Is it not natural that education results should be used to evaluate education quality, as compared to how pleased the parents are? The education system is there to educate, not to please parents.

I’ve seen parental satisfaction prioritized in some schools, with situations like failing to increase an overloaded student’s TESOL score by the “necessary” 20 points leading to problems. By the educational outcome method, the 4 point increase in the score (after just a few additional classes) compared to the previous test was a success. But by the parental satisfaction method, this was a failure.

3) The framing of bias is curious. Seattle’s $15 minimum wage issue is the ONLYmajor minimum wage issue to report on empirically right now. It’s recent. So … you want them to report long-term effects about a recent thing?. For the case of wages, the econ 101 observation that you share, is in fact rejected by nearly 100 continuous years of observation of both wages and labour demand rising at the same time (exceptions in recent decades at the lower end of the wage spectrum excepted, possibly related to the fact of minimum wages not keeping pace with inflation). People are not widgets.



78 tjamesjones March 16, 2017 at 5:59 am

i think 4 is your strongest point


79 Troll me March 16, 2017 at 3:47 pm



Leave a Comment

Previous post:

Next post: