Assorted Wednesday links


4. "Mostly" don't agree that fetishising math models has turned economics into a pseudoscience? Is it the fetishising or the pseudoscience? I'm a little less articulate in my criticism of economics and economists: can't see the forest for all of the damn trees.

I think a lot of economics is too interested in fancy math relative to real-world applications.

I recall for my thesis, being advised to make an essentially unnecessary mathematical extension in order to improve the mathiness of the research output.

Often (OK, sometimes) in economics papers, I see unnecessary mathematical extensions which may impress some economists for its mathiness, but which end up meaning that you need a full page to explain everything to define the problem in order to turn it into a (unrealistic) mathematical model, whereas a couple/few clear sentences would have sufficed to make the point clear.

It's as though if you do not put it into a mathematical equation, that the innovation is of no value to the field. For practical purposes, this is how things tend to work these days. No matter that you need to make all manner of completely unrealistic assumptions to make the problem mathematically tractable.

I cannot think of any complex subject where "a couple/few clear sentences" is more useful than "a full page to explain everything." The value of math in economics is precisely that it forces you to think logically and clearly and to uncover any mistakes in your initial assumptions.

A journal that uses the strict, precise definitions and ditches the equations themselves would be fantastic.

Yeah, using a bunch of math has done a really great job of uncovering the counterfactuals underlying economics, like perfect information, infinitely rational infinitely self-interested actors, sweeping all deviations from reality under the rug by chalking them up to the syllogism of "preferences", ignoring everything not included in your model as an "externality" then pretending like it doesn't exist, and so forth.

Oh wait, those assumptions are still all over the place and will continue to be until economists are forced to test their models with empirical reality. Which will never happen. Because economics is not a real science, it is astrology.

E.g., if A goes up, B will go down. Often, an intuitive explanation can be offered in a couple sentences.

But to make it mathematically tractable, often you have to define, say, 10 other variables, go through 5-10 intermediary equations or steps to the final equation, and explain a long list of unrealistic assumptions that would need to hold for the basic functioning of the model to hold.

This kind of stuff abounds in economics literature. Often, if the empirics confirm that they got so much as the direction of the effect correct, the theoretical breakthrough is hailed as a success. But the direction of the effect could equally have been theorized upon in couple/few simple sentences with an intuitive explanation, with no need for a bunch of fancy math or unrealistic assumptions to make the models mathematically tractable. But if you don't have a bunch of fancy equations, prospects of getting published approach zero.

Yes, in reflecting on the responses, I can see how the mathematical statements force you to make precise definitions. But, I argue, sometimes undesirably so. For example, the mathematical focus on measures of TFP have brought us very far away from the intuitive understanding of what TFP should mean, most especially when analyzing short and medium term trends where exchange rate and commodity price fluctuations can drive huge changes in the measures.

His point is dead on. The problem with the mathiness is that it blinds economists to their own BS. I could have done without all the ancient Chinese crap, but his point is strong. He would have strengthened the argument a lot.

Mark his words, because they won't be there in a hour. Lol.

Yes, I have noticed a lot of censorship in this blog, lately.

Economics models are unusually sensitive to how rounding and precision are handled. Very few other applications are quite this sensitive -- weather prediction, some problems in particle physics, a few other things. The difference between models that produce wrong answers and models that produce right answers is how they handle IEEE floating-point arithmetic. Nearly all programming systems use the IEEE floating-point formats, but many do not support proper use of the rounding modes at all. Does the programmer of your model even know whether gradual overflow and underflow are being used?

When editing, I often make economists aware of how rounding is determined in natural sciences. You round to the closest decimal (or order of magnitude) where the entire 95% confidence interval lies.

In economics, you often have people giving results to 3-4 decimal places, where according to practices in natural sciences, you would only be presenting the information to the nearest thousand, for example. I find that economists tend to insist on providing lots of decimal places, but some number of these clients are also more forthcoming in presenting confidence intervals to accompany their "precise" calculations.

Sometimes this arises from the nature of the models, where data inputs are not amenable to confidence interval-like statistics, and in presenting 3-4 decimal places, the researcher legitimately doesn't have a clue what the statistical confidence might be. However, this can be addressed somewhat by presenting robustness checks using different modelling strategies.

I normally try to insist on not using more than 2 decimal places unless the different can be motivated. (Like, what's the point in saying there's a 12.895% change when you know the underlying data quality isn't that good? Just call it 13% and specifically mention reasons to question data quality in the specific case. But, policy decision makers seem to prefer to remain in the dark when it comes to these things, preferring "I have an exact answer, and this is what it is.")

You guys should take the first course in undergraduate physics where the proper number of decimal places to use is conveyed.

Hint: there is no fixed answer.

If your equipment error is +- 0.1 amps, then you say 3.2 amps +- 0.1 amps.

Learn the difference between accuracy and precision and use the appropriate precision.

Small steps towards a much more scientific economics.

I can't resist

5. Kind of like that old saw about receiving a million refugees per year in an EU population of 500 million. And many Europeans not seeming to understand the threat they are facing or the magnitude of what they are setting themselves up for, in more ways than one.

According to the article, the chances of hitting an asteroid in an asteroid belt is one in one billion. You're not making the point you think you're making.

Never tell me the odds!

#6 - Bernie Sanders displayed incredible ignorance in a recent interview with the NY Daily News on this topic pointing out further why he cannot be taken seriously as a candidate. He just all platitudes and no understanding of modern day banking.

Aside - why no mention of the renaming of the G Mason Law School to 'honor' late Justice Scalia? Does this further cement the view that this university caters only to right wing points of view? Controversial, yes; good idea, no.

"why no mention of the renaming of the G Mason Law School to ‘honor’ late Justice Scalia?"

I believe there was a post the day it happened.

"good idea, no."

They got 30 million bucks and a lot of publicity. My law school got 30 million a while ago and named it after a lawyer no one had ever heard of.

I think GM did ok.

Do you really think GMU has to re-name something to stay flush with Koch cash?

Koch! The Emmanuel Goldstein of the US left.

20 million was from an unknown person. No re-naming, he would have likely shopped it around to other schools.

Heh, maybe. Hard to argue the PR was all good.

Ever is a long time

Indeed, much as Holmes Hall at Harvard Law School caters only to students of a eugenicist persuasion.

Sanders belongs in the nut house. And, here's a novel approach. Don't allow TBTF bankers to pay people like Hillary Clinton $600,000 for 20 minute speeches.

Another guy that displayed ignorance is the author of the article. I see why. He worked with Dodd in writing the bill.

It's not his fault. Banking is too complex. No one can know it all.

Why I believe that Dodd Frank isn't working: it doesn't address the causes of the subprime mortgage crisis which touched off the credit catastrophe which ignited the fatal financial crisis. The over-large sizes of the banks or proprietary trading (however that is defined) didn't cause it: that would be structured securitizations which evolved from customary mortgage banking and packaging mortgage-backed securities which had been safely around for decades. The TBTF size issue simply concentrated the losses in large firms which failures/insolvencies the elites believed would wreck the political economy.

It seems that the author didn't realize that until they were forced to to obtain TARP money, GS and MS did not control or own commercial banking companies and their creditors did not have protection of Federal Deposit Insurance. So, if they failed it would have been a political choice to bail out the creditors, which was taken for granted.

#5: Unlike the asteroid belt, the rings of Saturn can be surprisingly dense.

> Compared with most other astronomical objects, the ice and rock particles in Saturn’s rings are extremely close together. On average, about 3 percent of the total volume of the disk is occupied by solid particles, while the rest is empty space. This may sound small, but it means the typical separation between particles is only a little over three times their average diameter. Assuming a value of 30 centimeters for the latter, the rocks would be as close as one meter away from each other. There is no hard and fast rule, however, due to density variations across the rings and the wide spectrum of particle sizes.

4. Seems about right to me. And but-for the mathiness I would've pursued a PhD in economics rather than a law degree, so I'm a victim here.

Met a guy at little league whose brother washed out of a PhD program because math. Seems sad, and while the article was pretty blah blah blah, I think econ could be a little messed up in the scoring.

Economists count coup with math, but on very narrow models, which do not generalize to policy very well. I mean yes you can point to an organ donation policy that works, don't mind the entire nations rudderless.

At least earlier on, I remember my thinking being something along the lines of "which bullshit model am I supposed to apply when faced with this type of contrived situation again?"

I remember one macro exam in grad school where a quarter of the exam was based on proving whether alpha went up or down as a result of something else changing in a non-immediate and staggered effect. Well, I can usually "see" it in the space of a few seconds or more, but I spent over an hour turning the model upside down, sideways, inside out, and every way I could possibly see to PROVE that alpha went up. I never managed, but I was sitting right under the professor's nose, and he knew I had it right straight away, and gave me pretty good marks on the question anyways. Had I not been sitting under his nose, I probably would have gotten 10% lower in the course.

In micro, I soon learned that the prof just went through the textbook. So I only went to a couple classes the entire term when I didn't fully understand the text. Turns out BOTH TIMES (!!) I showed up to ask a question, the prof explained that in fact the textbook was wrong, which explained why it didn't make sense. What a waste of time! Sit through an entire 2 hour lecture that I entirely understand and have already practiced all the relevant problems, just to ask a question, just to find out that I only failed to understand incorrect things. However, 100% of evaluation comes down to how well you answer the generally predictable set of mathematical puzzles on exam day.

I think it would be good if the field placed more weight on written explanations, but a) it's harder to come up with a marking scheme for that, and b) well, in the real world, mostly when people want an economist, they want someone to spit out a concrete number, not drawl on for 30 minutes or 20 pages about the upsides and downsides of how the specific method is likely to influence the results.

I understand that you have to understand the maths to not make retarded mistakes in using the software, but honestly, you can demonstrate that knowledge in words without math, and most of the fanciest stuff economists do for most analysis is just arranging the data properly and using a few simple commands in the appropriate software. But in order to be allowed to use that software in the real world, you have to jump through a hoop that demonstrates that you can reproduce the inner workings of the software in theoretical mathematical terms (without ever having to demonstrate any understanding of the real world stuff going on behind the datasets you might ultimately manipulate).

Like yeah, I can assemble a database and run the commands and interpret things just fine, but in economists everyone always wants to see the transcripts, and if you don't sort through all the unrealistic mathematical puzzles on exam day, well, end of story!

Anyways, I rather prefer working with words than numbers, and that's what they ultimately put me up to, so no big deal for me ...

Lol same here. Hopefully it'll all come crashing down and they'll let people who use words participate in economics discussions again.

#3: Although it makes some people very aroused.

3) Yeah, I always thought it slightly weird about girls dressing barbies and how this implied barbie spending lots of time naked. Of course, the intimate parts are not in much detail at all. Then again, maybe this teaches girls to me more comfortable about their body in a sense (assuming they aren't impacted by the ridiculous hip/bust measures the typical barbie doll implies).

I would definitely feel uncomfortable touching the "intimate parts" of a robot, much like with a doll, but I most certainly would not feel aroused. In fact, I don't think I would follow that instruction at all (or instructions to increase the electric shock upon doctor's orders, or any instructions I either don't like or simply do not fully understand, for that matter).

Generally speaking, I would feel far more comfortable dealing with robots whose design revolves around their function (e.g., Roomba), and have negative interest in ever owning a humanoid robot.

(4.) It does not bode well when the writer refers to "economics teachers", rather than professors. There are many economics "teachers" who are not professors and who do not earn $100,000 on average: lecturers, adjuncts, graduate students... A sloppy start.

"economists cannot point to concrete objects – cell phones, plastic – to justify the high valuation of their discipline" -- good gracious, this again? So only tangible objects have value? Yes -- plastic, so very valuable. Clearly CEOs -- or for that matter programmers and most jobs in the economy -- have zero value. Yes, brilliant reasoning here.

The forecasting failures are a good point, but what's that got to do with academic economists? How many economics professors spend their time on forecasting? 0.1 percent, maybe? Straw man.

Okay, to be fair it does get better... I like the points about ``Mathiness'' (Paul Romer) and having too much faith in mathematical models (he should have cited Emmanuel Derman), but again -- is his beef with university economists or Wall St economists? It isn't clear at all, and this is the article's biggest weakness. If he had focused on the empirical failure of Wall St or business economists in forecasting, okay, he'd have a point. But mixing academia with the business world just muddles the issues he sets out to analyze. I thought philosophers were supposed to be the best at writing clearly? Maybe someone should write an article debunking philosophers...

Great point about philosophy. Substitute "jargon" for mathiness and "law" for economics too

"Substitute “jargon” for mathiness and “law” for economics too"

Nobody in the last century or two has had the gall to claim that law is a science. If economists were honest about their field being a professional field (i.e. providing tools for people to use to make money) rather than a science, that would be a really good first step. Of course, economic models turn out to also be bad at making money for people as well (as detailed in the article).

No, I think the astrology analogy holds up quite well. Someday our society may turn away from the superstitions that prop up the economic priesthood. But apparently repeated, catastrophic, systemic failure of the economic profession to do the one thing they claim to be able to do is not enough to dispel those superstitions.

Too much math is one problem but bad data or misleading data is a much bigger problem. And of course the whole N=1 thing.

6) Assuming that it were deemed desirable to break up banks (seems kind of dumb to me to have "too big to fail" especially due to moral hazard, and in capital markets the size of the USA, even at 2% of the banking market I don't think it is credible to believe that this negatively impacts competitiveness or major deals - nothing would stop banks from engaging in other forms of cost sharing or divvying up deals between smaller banks as a part of a larger deal) ...

a) Can I keep my bank? It seems trivially not difficult to allow people to continue using the same bank cards, credit cards, or perhaps even online login information, etc., while having the banks take care of things on the back end for a couple/few years until new cards, etc., are issued after expiry. On the matter of ATM fees, I don't see why banks couldn't start to align in certain partnerships (think Star Alliance and One World in flights) which would allow ATM users access to a large number of ATMs within the network without additional charge (right ... as though we actually think banks would ever do anything on their own volition to reduce banking fees ... but perhaps some networks of smaller banks would see a competitive edge in doing so).

But, I like consumer choice, so I much prefer the idea of separating banks by function. They could set some assets threshold beyond which such rules would aply (hence not applying to banks which are not big enough to pose a systemic risk)

b) global business: Why would international offerings be relevant to whether the bank poses systemic risk to US financial markets? If there are more banks and more competition, wouldn't costs be lower? Are scale economies in transaction systems really that big? Couldn't a set of players just mutually invest to create some other financial body specialized in international stuff, which was not particularly risky to the US financial system? Wouldn't they naturally want to do so if the scale economies are relevant?

c) The main reason that I'm skeptical that such efforts would do much is that if financial institutions all face the same rules and follow much of the same practices, then the systemic risks posed by the entire financial system should be broadly similar (perhaps somewhat less issues with moral hazard though) regardless of size. Would the 2008 financial crisis have been mitigated much at all if a dozen smaller firms rather than a handful of major ones were at risk? I highly doubt it.

Things are really tight at big banks right now, with two causes:
1. Low interest rates over an extended period of time have killed profit margins for traditional banking activities (borrowing, then re-lending). Banks make a lot more money on the spread when interest rates are 1-10% higher than they are now.
2. The regulatory burden in the US has increased dramatically over the last few years. 15 years ago bank IT departments spent 90% of their time on customer-related items and 10% on regulatory compliance. Now it's closer to 80% compliance and 20% actual customer service activities, to the point where most banks have multiple "cabinet-level" departments for risk and compliance who spend all of their time internally auditing everyone to ensure they're ready for and following the latest requirements from regulators like the OCC.

So the real cost of #2 is all the lost opportunity, both for customer service enhancements and innovation, efficiency, plus added cost as a ton of money and resources are effectively wasted on compliance, much of which a rational customer wouldn't want to pay for if it wasn't being forced on the financial institutions.


The public has this pervasive mentality that whatever is bad for Wall Street is good for them. It is a failure to understand modern finance. Everyone thinks more regulations are needed, but probably can't identify a single rule. The complicated web of regulations is insane now. Compliance costs are through the roof. It's impossible to operate without a full-on legal team. That does not make the market more free.

I'm not saying that we need a totally unregulated financial sector, but more is not always better and seems certainly worse nowadays.

Until someone gets around to explaining to the voting populace that you can't have both (1) a safe, stable financial system and (2) easy access to credit for everyone, that's never going to happen.

Lo and behold -

I think a lot of the current regulatory costs are largely one-off. Definitely longer-term costs will be higher, but they are presently spending LOADS of cash ironing out all the details to ensure compliance (some of this falls my way as a translator...). I think this up-front cost of developing the systems is likely to count for most of the current spending (once in place, only a handful more datapoints on the occupation and position of the client are needed, and most of the flags and evaluation tools operate automatically after inputting the appropriate data).

What you say is very relevant, but it's not clear that this applies a whole lot more so to US banks than others - the net change, in consideration of the fact that any competitors who want to be able to do business in the US market or with US nationals, could even be in favour of American firms (maaaybe).

This assumes a static regulatory environment, which is precisely the opposite of what Dodd-Frank envisions.

So, from the early 30s through early 80s, banks were not highly regulated because Regulation Q was not a regulation, and usury laws that capped interest on both debt and bank deposits were not regulations, and prohibitions on banks owning any branch across a State line (except for New England for an area half the size of Texas) was not regulation, and the laws of the 50 States that very clearly made lots of things crimes when it comes to lending or selling to individuals was not regulation....

I grew up reading Milton Friedman and others telling me how the bank regulations were screwing me the individual because my checking was only free, my savings only earned 4-5%, I could not get a loan without having solid income and assets, and then when I didn't need credit, a personal loan would be as much as 12% instead of an implied 3% while I could get 8% on checking balances, if only the banks were deregulated.

And uninsured money market funds would NEVER EVER be at risk or be frozen in a bank run because without FDIC insurance, depositors would never put money in funds that were not 1000% safe.

I was about 20, and even with a lot of respect for Friedman's reasoned arguments, he seemed to be promising a free lunch. TANSTAAFL

The past 40 years of implementing the policies Friedman and all the other bank deregulation proved TANSTAAFL including the first mass run on the (shadow) banks since 1930 in 2008. Not to mention the zero interest in money deposited with banks and personal loans funded with 0% interest savings running at 35%-500%.

From my experience, I can't imagine that a return to the banking and finance laws of 1960 wouldn't be seen as ten times as intrusive into that sector than what exists today under Dodd-Frank.

Meh. Just let banks fail. Big PE firms and other capitalists would step in and purchase the bits that still had value and quickly get back in business. You allude to this in your (a) above. It's not like the ATMs and people and knowledge would disappear. The shareholders/bondholders/uninsured depositors would just be wiped out. As they should. All the productive assets would still be there to be scooped up and deployed.

I don't understand why everyone thinks the fix for a too cozy relationship between Washington and Wall Street is a deepening of said relationship.

I don't buy the whole rant on math, but there is some truth there, and in Romer's criticisms. My thinking - not at all close to original - is that the use of math in economic models serves two purposes:

1. To check your logic. If it turns out that the assumptions lead inevitably to the conclusion that 0 = 1 then there is something wrong, and you better figure out what it is.

2. To yield empirically testable results from the model. If your math generates an equation whose variables are empirically measurable then you can check things out statistically, just as you can logically in (1). Interest rates were supposed to go up, but actually went down? Huh!

I've always found it a bit ironic that it seems the same people who are appalled at big banks are enthusiasts for big government (i.e. more scope of federalized rather than state, local, or personal governance and functions - e.g. single payer health, retirement, education, etc.). I don't think its the "too big" aspect that they have a problem with.

Don't know who those people are, but strikes me that single payer works in practice .. while we in the US are hung up on it not working in theory.

The real end-point objective for single-payer is single-payer/single-provider, i.e. VA for All. (Medicare for All is maskirovka.)

With the provider staff of course being unionized public employees. No more of those overpaid doctors, etc. They may not get there quickly, but the logic inevitably leads there.

As to who those people are, I'd say most the Sanders camp for a start.

In most universal health care systems, a fairly high share of services are provided by private businesses which are not allowed to bill anyone except for the government. And those unionized nurses earn a lot less than in the USA.

Which health care system would you prefer?

Those of African nations where there are no third party payers, government or private.

Or those of France, Germany, Canada, Japan, Sweden, where almost 100% of medical care is provided by private sector providers who are paid by a third party payer that might be government (UK, Taiwan) , or entirely private insurers (Germany, Switzerland).

In Africa "individual responsibility" goes so far as requiring patients to source drugs, bandages, medical devices and supplying them to medical practitioners who probably only have some medical training, unless the care is provided by European and US charities or governments.

Why isn't Africa the conservative's absolutely ideal healthcare market?

Why would you chose the failed French or Swedish healthcare over African healthcare?

Surely the lack of government intervention in African healthcare makes it the absolute best free market healthcare possible.

It depends on what you mean by works in practice. Does it result in new drugs or treatments? No. Does it result in best quality of care? No.

It does keep costs down. A monopoly employer, especially if they are the government, can get away with paying doctors and nurses very little. It can ration services.

So if you measure success by convenience to the government - and lots of opportunities for graft in anything the government does - then sure. If you measure it by treatment for the population, then no it doesn't.

It's not actually clear that the net result is fewer drugs or treatments. Quite plausibly, the newer drugs and treatments happen to congregate where the most money is, and would have happened regardless, just distributed a little differently across borders. Of course, that implies an exceedingly strong US pharma lobby, but we all know that already exists.

I'm open-minded RE: single-payer, but you just aren't making sense. "Newer drugs and treatments happen to congregate"... what are you talking about? The R&D spend for these things is undertaken because of the profit potential. By legislating prices for drugs down there will be less R&D, how can you dispute that? Still might be a better deal for most people, but no need to distort things.

Mike - yes, that is the natural counterargument and I did not take the time to pre-empt it.

The point about "congregating" is that a lot of research efforts are done by people who would have been in medical research regardless of $120k a year or $300k a year, but obviously they choose to earn $300k a year in the US instead of $120k a year at home.

I understand that much of the following flies in the face of orthodoxy, but consider a few things which place in doubt the extent (if not the entire direction) of the anticipated R&D-enhancing effect of private markets.

a) Even under a universal pay system (which, importantly, does not apply to a diversity of pills), the private drug developer still has huge opportunities for profits in a public system.

b) Consider the amount of money dedicated to diverse marketing expenses from advertising to sales commissions and darker stuff like what essentially amounts to undisclosed paid-for expert promotions. Much of the premium price for drugs in the USA goes straight back into essentially unproductive expenditures (and to be clear, I absolutely understand that a certain degree of marketing expense provides an unambiguous value in providing market actors with information to buy products).

c) A lot of money for clinical trials is driven by unnecessary repeats, in an effort to game the statistics until they have a couple outcomes with the "desirable results". This might still occur in a different system, but the extent of the profit motive exaggerates this.

d) A lot of R&D expenditures are geared towards very marginal improvements by tinkering with existing drugs, rather than seeking genuine breakthroughs (which often occur through more curiosity-driven or ethics-driven researchers in public research institutes and universities). Here, they often try to game the patent system, sometimes offering drugs which are no more effective than the alternative (sometimes even less so), but which they can then spend millions or billions on marketing as "better" (J&J and Risperadone is an example here).

e) But then there are non-pharma things, like medical equipment and improved treatment methods. On the matter of medical equipment, I feel strongly assured as a result of many conversations with a friend who works in public sector medical procurement that corporations are able to ensure that they make lots of money while satisfying the procurement processes. On the matter of improved treatment methods, this rarely results from the profit motive (I assert) and instead results from experienced doctors just kind of realizing that something might work better, trying it, and sometimes finding out that the new way is better.

It would indeed be ridiculous to assert that the profit motive is irrelevant. However, I have pointed to many reasons that water down the technological gains of the enormous economic resources plowed into the sector. Also, even in a universal system, virtually all product developers are private, and must face market discipline in addition to having a strong motive to develop more competitive products.

I argue that the main effect of the private system in the USA is to concentrate these efforts in the USA, and likely have fairly limited effects on global medical technologies advancement (I could even stretch the case to argue that the net effect is negative, but I do not actually believe this is the case, but that could change under TPP and other "excessive" patent enforcement efforts which make it very difficult for foreign ventures to get off the ground and pursue broader research endeavours).

Key: outside of the universities, technology developers are still private in universal systems.

4. The difference between most mathematicians and most economists is that the former have no ax to grind, the latter a Paul Bunyan ax to grind, the former using math to solve nature's questions, the latter using math to solve political questions.


Everyone needs to learn economics, if only to detect which way they are being lied to their political masters and betters at that specific instance.

That saying about "lies, damn lies, and statistics" makes perfect sense if you interpret statistics to mean government statistics, ie. economics.

#4 -- I mostly agree. Can anyone point me toward an article of similar length that makes the contrary case?
I think you in advance.

Making the contrary case, try this one. I don't always agree with Noah Smith, but he's always fun to read.

That link was bad I think this one will work -

Having read the article, I am not very impressed. Essentially, he is saying that economists are in demand because they are conversant with statistics. Why not hire a statistician, who did not waste time studying arcane (and in my opinion) useless areas of economics (such as macroeconomics)? I would be interested in hearing why Tyler disagrees. That would make an interesting post.

I received a BA in economics before the sharp turn to mathematics. I think economics is very useful in understanding many aspects of out world, but mostly in a qualitative sense and almost exclusively as regards microeconomic phenomena.

Pure statisticians are far less likely to hone in on the policy-relevant issues, I imagine. At the very least they will need some direction from an ... economist. Or, we can put them all under the direct control of politicians, who are sure to direct the analysis in the appropriate manner.

And just in case mathiness is not enough to create a veneer of sophistication economists will often use metaphors to piggy back on the prestige of medicine. The use of these types of metaphors was rampant during the financial crisis. Typically a high priest from either Harvard or the Fed would on a prime time news show and deliver the sermon with a grave look: "The economy is in the ICU, but is recovering." "The Federal Reserve needs to deliver strong medicine immediately". "Bank credit is the lifeblood of the economy". The new astrologers want the little people to know that they are not unlike the heart surgeon that saved Aunt Mollie.

Look, you can either have shock therapy or a poison pill. We're doing science here.

There is one thing I have never heard a good explanation for from the "break up the big banks" people, even accepting the assumptions undergirding their conclusion that it would be a good idea:

Failure is correlated. 50 banks with 2% market share each, investing in the same ways as 5 banks with 20% share each, would face the same risk of failure. We'd just need 40 smaller bailouts when the crisis came, instead of 4 big ones.

I'm sure it'd make a difference at the margins, but not anywhere near as big a difference as you'd think, and since open-ended bailouts (the issue breaking up the banks is supposed to solve) is a real problem, we should do something else instead to fix it.

The key words are "investing in the same ways". You would hope that 50 banks would enable some sort of diversity in outlook, policy, personnel and hence risk taking. Five banks means five sets of edicts from On High limiting the freedom of the employees to do what they think is sensible. One central strategy. Fifty banks means fifty sets.

That hope would be misplaced. There just aren't that many independent business lines and they still all tend to lose money at the same time when there is a panic.

A couple decades before there was a crisis in big banks, there was a crisis in savings and loans, that involved precise many similar small players facing similar incentives and making similar bad decisions for similar reasons.

As others have commented.

Financial risks tend to be correlated, since a big enough shock in one sector (real or due to malfeasance/mismanagement) will spread to others. It's really hard to insulate your investment strategy from other strategies, even if both strategies are responsibly created and there is no overlap in asset ownership.

What is needed is some way to definancialize the economy, so (a) smart people don't waste their talents in zero-sum moneymaking competitions and (b) letting the N biggest banks fail wouldn't crash the economy, for any reasonable value of N. Unfortunately, it is really tricky to propose something that would do that and be consistent with liberty, and even trickier to propose something that would also be possible to enact.

It would at least reduce moral hazard, since each one of them could more credibly believe that they might be the one who is allowed to fail. But I think what you're saying is hugely relevant.

We could even have a policy of literally rolling the dice to determine which banks to let live and which to let die and that alone would make all of them far more risk-averse than they are now.

I really like that idea :) Among other things, it would take political favouritism out. However, it would make it more difficult to "punish" those banks that were most irresponsible - perhaps it would be better to ignore the role of luck, and just allow a certain percentage of banks in the worst positions to fail.

The reason we have big swings is because markets are highly interconnected and exhibit positive feedback loops. This was true before bank consolidations and will be true after boneheaded attempts to break up banks. Only idiots think some government teardown of banks will stop future correlated asset selloffs.

6.-- Interesting. But what about this this : "that means that betting markets think there is a 30% chance Trump fails to get 1237 bound delegates but still wins on the first round."? How is that possible, at such a high probability? Very few delegates are not bound during the first round, and they don't seem to be massively inclined toward Trump. Are the market nuts?

Also I have observed that there are arbitrage possibilities on (e.g. on the market on "who will be the next president?", the total price of the "buy no" on Trump, Cruz, Clinton, Sanders is $2.95, but since at most one can be president, you're sure to get at least $3 if you buy the four "no"s). I have refrained to use them, because I have heard it is illegal to bet on this markets for an American resident, and because the site may close or otherwise keep your money with a non-zero probability. But this makes me doubt that this markets are serious predictors of anything, especially when you try to infer from them
some conditional probability.

"the total price of the “buy no” on Trump, Cruz, Clinton, Sanders is $2.95, but since at most one can be president, you’re sure to get at least $3 if you buy the four “no”s"

I believe that takes a percentage of trades. Yes here it is:

"Whenever you sell a share for a higher price than you paid, we charge a 10 percent fee on your profit. "

There are other possibilities as well. For one thing, the delegates could change the rules to unbind all delegates, and Trump could come in with a bound majority but still lose on the first ballot. It seems odd, but it is hard to underestimate how many people hate Trump. Some people think if he comes in only a few short he will pick up the Carson delegates and win that way because people who like betting like the models that show that.

I agree the market is silly and there are arbitrage opportunities, however there are much better ones just writing call options against insolvent companies.

I would like to see a breakdown of how comfortable each candidate's supporters are with touching a robot's private parts, assuming I can stop laughing about the idea of a robot having private parts. No good reason, but it's as good a question as a lot these questions are.

#4. I come at this from the complete opposite side.

I did my PhD in applied math and eventually published a paper in an econ journal. I think that makes me more of an "economist" than e.g. someone with a degree in econ that doesn't work or publish in the field. My contribution was largely mathematical. Some economists had been working with a mathematical model for years and I was able to come along and improve on it. I was a lot more interested in math than econ.

The mathematicians, scientists, and engineers that I more frequently collaborated with were always a bit in awe of the economists. It's the ultimate glamour field and we were envious. But the economists were pretty respectful of mathy people, especially in micro. On the other hand, a lot of the economics results seemed pretty arbitrary to us. As many different mathematical models could be proposed and analyzed, we always found it hard to discern why a particular model was deemed valid and published. The standards in science and engineering (experiment, practicality) are more objective. Charm and persuasion is pretty important in econ, which I don't think many dispute. Certainly much more important than in e.g. Computer Science.

Math in social science isn't so much a way to discover truth as much as a way to check for internal consistency.


Please point me to the academic field where university faculty, in a quest for tenure, don't produce ridiculous papers using absurd jargon that doesn't matter to the outside world? In my mind, that's all this argument says.

The problem isn't the economists. It's that everyone else thinks they can understand the economists work, without adequate understanding of the strengths and weaknesses of models, data and technique. Very few actual economists speak about economics with the conviction of politicians. We call anti-vaxers complete cranks who don't understand science of statistics or anything. Why don't we call the people who misinterpret economics research or models equally demeaning names?

The astrologers are the talking heads on television who think they can second guess Fed decisions when they lack an understanding of the subtlety of economic models, and they don't have access to the same data.

"Please point me to the academic field where university faculty, in a quest for tenure, don’t produce ridiculous papers using absurd jargon that doesn’t matter to the outside world?"

Good observation, but I think the point here is that a collective delusion has gripped this particular academic discipline and not the others, such that we get false conclusions out of its practitioners more often. The point of the article isn't that economists write lots of inane papers, which they of course do, but rather that they are too focused on mathematics, making much or all of their work useless and wrong. There's plenty of biology articles analyzing the mundane details of the lives of fire ants I'm sure, but that's not the same as saying most of those articles (as well as the important ones) are wrong because of a failure of method.

I have an idea of why the mathiness took over. OUR Economics department uses Calculus as a pre or co-requisite course to keep out the lunkheads. Eventually, all economics students who go on think that math course was foundational, rather than an enrollment barrier. Our Econ department wants to require Differential Equations, now. the Math department snorts with derision - they don't appreciate being turned into gatekeepers for another major and claim the economists just need a laminated sheet of paper with plug and chug equations.

Presumably economists realized they had suffered a communications break down some time in the Seventies. You could not get the Left, the Keynesians and the neo-Classical economists to agree on even basic things. Are big deficits good or not? Does debt matter?

So instead of arguing with each other about what "is" means, they opted to go for mathematical models instead. That way they do not argue with each others' assumptions but with each others ability to construct a model.

Also, of course, physics-envy. They did not want people to think they were sociologists who owned a tie.

Yes, I think the application of physics/chemistry notions of equilibrium are troublesome.

I don't really feel like explaining it right now, but consider the dynamic equilibrium across a cell membrane. Things are going both ways. Much economic modelling assumes an immediate (or perhaps slightly staggered/delayed) unidirectional effect. I put the blame on Leon Walras for taking things in this direction, and the field has never recovered since. (It DOES make macro modelling a whole lot easier, of course.)

#6: Former aide to Friend of Angelo Chris Dodd defends their handiwork. He offered not one compelling reason in the course of the entire article. In essence it was: some people might find it inconvenient for an interim period of time if we broke up the megabanks. Enough.

The government approves all bank mergers and acquisitions. The big banks were forced to acquire failing banks and securities companies by the Federal Reserve. Sanders talks out of both sides of his mouth. Of course, he may honestly not know what is going on in the government. He probably has been asleep for 20 years.

Agree. Sanders actually voted against TARP (approved by the senate, signed into law by GWB). He just hates Wall Street because it's a symbol of capitalism and it's easy to get people angry. So now we have to break up the big banks to not give them bailouts that he wasn't going to give anyways.

IIRC at the time, the problem was not so much that the banks were 'big' is that they had characteristics that required skills outside of the FDIC's set. The FDIC had ample practice at brokering the purchase of failed deposits-and-loans institutions and superintending them on occasion. Universal banks with a complicated trading book were novelties to them. Also, about 2/3 of Citigroup's deposits were domiciled abroad. And, of course, apart from Citigroup, Wachovia, and Washington Mutual, the most troubled institutions were not banks. You had the mortgage maws, an insurance company (AIG), diversified securities firms (three of the five bulge bracket firms were insolvent and at least one of the remaining was distressed), the implosion in commercial paper &c.

"The big banks were forced to acquire failing banks and securities companies by the Federal Reserve"

Incorrect. The FDIC manages this.

Read much?

In ordinary circumstances, they do. However, some of the salients of the financial crisis were deals brokered by the Federal Reserve. These involved universal banks acquiring securities firms (the Bear, Stearns purchase and the merger of Bank of America and Merrill Lynch).

#4: "the ubiquity of mathematical theory in economics also has serious downsides: it creates a high barrier to entry for those who want to participate in the professional dialogue, and makes checking someone’s work excessively laborious. Worst of all, it imbues economic theory with unearned empirical authority."

Has economics in fact earned any empirical authority?

4. I suggest that mathiness can be very helpful in economics if left to mathematicians who focus on a particular industry or subject. I have in mind Austin Frakt, who describes his educational background as being in physics and engineering and whose PhD is in statistical and applied mathematics and who describes himself as a health economist and researcher. Trouble arises when economists use mathiness to give their research/findings an aura of objectivity. Nobody believes it, except for those who already drank the cool-aid.

5) "Space is big. You just won't believe how vastly, hugely, mind-bogglingly big it is. I mean, you may think it's a long way down the road to the chemist's, but that's just peanuts to space." - That other British Adams writer.

4. Not persuasive. He makes the common error of focusing on macroeconomics (and even then mostly Wall Street economics) with minimal reference to microeconomics. So right off he is ignoring where 90% of the useful stuff gets done.

Funny, on many levels, that he goes out of the way to refer to Paul Krugman as a "Nobel Laureate". Wait a minute, isn't that like calling him a Nobel Laureate in Astrology?

Yeah DSGE models are a game but so is most of academic publishing in the social sciences and humanities.

I totally agree that math has become too much of a barrier to entry in the formal study of economics.

Romer's mathiness critique is more subtle than he understands.

Economics blogs outscore blogs in all the humanities put together. And popular economics books sell more too. Why is that? Maybe the philosophers and sociologists should try harder instead of ranting about economics.

#4. Mathiness is a problem in macroeconomics because at the macro level an economy behaves more like an organism than a machine. Macro has been built on incorrect physical models ever since Léon Walras cribbed the formulas that formed the basis of his general equilibrium theory from a popular physics textbook of the time.

An economy is not a machine. It can not be understood through reductionism, nor modeled as if it has specific inputs which lead to specific outputs. It is a complex adaptive system, and as such its future direction is unpredictable. It is dominated by feedback and network effects, sensitive to initial conditions, and opaque to much mathematical analysis.

Economists should be studying ecosystems, not differential equations. If they must do math, they should be learning chaos theory and information theory. Instead of building predictive models, they should be learning more about the dangers of tinkering with a complex system.

Ecology went through a mathematical phase, and was quite interventionist for a time. We thought we understood how ecosystems worked. But who knew that introducing six wolves into Yellowstone park could, through a complex series of feedbacks and unintended consequences, cause the direction of rivers to change? We once tried to intervene to stop forest fires, only to discover that the result was fewer small fires but the occasional devastating megafire. Sound familiar?

An economy isn't a machine in equilibrium, or even many local equilibria. It's better thought of as a complex computing machine with software we don't understand. The measures we apply math to, such as GDP, inflation, M1 and M2, aggregate employment and so on are nothing more than abstractions - emergent properties of an underlying system we don't understand and can't understand.

Relationships which hold today, or even which have held for years when you go data mining for them, could end tomorrow. What looks like a stable equilibria is just a point of meta-stability in an ever changing system. Because the economy is adaptive, trying to predict its future behavior by studying how it behaved in the past is bound to fail. Because it is highly sensitive to initial conditions, the way the economy responded to a shock in 1960 might not only be completely different than how it would respond to a similar shock today, but it might even respond completely differently if you could re-run 1960 over again with the same shock, only with very tiny differences in details. How much can you learn from past behavior if that behavior represents just one path out of millions that could have been taken if things had been very, very slightly different?

If you are an economist, take some time to study complexity theory and information theory. While you're at it, take a class or two in ecology and learn what other scientists have discovered about messing around with complex systems.

In the end, you will hopefully develop some humility regarding what you really understand about how the economy works at its deepest levels, and even more humility regarding your ability to predict what it will do in the future. And most of all, you might stop yourself the next time you are thinking of telling some politician how he or she can intervene in the economy to improve it.

You should add some citations, expand a few arguments and send that off to some journals (if you have the appropriate letters beside your name). Or maybe simplify it slightly (still add a few citations) and send it to some economics-oriented news outlets. Worst case scenario is they'll give it a quick read, not get back to you, and maybe take it slightly into consideration in future publications.

Most of my intuitive critiques came from studying lots of microbiology and stuff about dynamic equilibria across different types of membranes. The use of "equilibria" in economics isn't remotely satisfying if you have much exposure at all to the natural sciences. In the ecology I studied, the math was roughly similar to what you see in economics (game theory stuff, when you're not just doing intro stats stuff), but it was more to demonstrate theoretical concepts, and case studies generally show the theoretical tools to be very inadequate for fully understanding species interactions and complex feedback loops in an ecosystem.

My thesis advisor did some pretty realistic equilibrium modelling in relation to a) monopoly pricing and quality in antibiotics and also b) in optimal plotting and pesticide application in a 2-crop + 2-pesticide + 1 pest model with pesticide resistance (he was the dynamic equilibrium and Matlab programming guy in the department, but is very aware that he's not a scientist). But the conditions under which economic equilibrium modelling in its current form can have realistic applications for the real world are few and far between.

#6 could the FDIC say to a big bank "we cannot insure your deposits anymore you are to big?"

Comments for this post are closed