You may have heard a bit about some errors and methodological issues in a paper by Carmen Reinhart and Kenneth Rogoff that has been used by austerians to argue for more austerity. The gist is that their work asserts that country’s with debt-to-GDP ratios above 90% have averaged -0.1% growth, whereas other ways of interpreting the same data set give a 2.2%. That’s a big difference. There’s an error in their Excel code that accounts for 0.3% of the difference; the rest of the difference comes from the (to me, strange) way they averaged their data. Justin Fox had a good piece on this:
In the 1990s, the consensus seemed to be that for the U.S. the inflection point was a public debt/GDP ratio of 50% — which is exactly what the country was nearing at the time. Higher than that, and the bond market vigilantes would punish the U.S. with much higher interest rates on government debt. The central teaching of what came to be known as Rubinomics was that cutting the deficit would actually stimulate the economy as it brought interest rates down.
Now, of course, U.S. public debt is up to 76% of GDP, yet the bond market vigilantes all seem to have retired or moved to Europe. In the long aftermath of a global financial crisis, with deflation a real threat, the U.S. can get away with running huge deficits with no immediate consequence. In fact, the Keynesian reasoning goes, big deficits now will lead to a better long run growth picture (and thus lower future debt/GDP ratios).
Is this reasoning correct? Well, right now the evidence would seem to support it: The U.S. is muddling through, while austerity measures have pushed Europe back into recession and most of Southern Europe into depression. For whatever it’s worth, Reinhart and Rogoff have advocated continued deficit spending too — at least for now.
The trouble with this brave new world of big data and whatnot is that people can always find some kind of evidence that supports whatever it is they want to think (I’m not suggesting that this is what Reinhart and Rogoff did here, I have no idea). Start with a conclusion, and look for data that supports that conclusion. No one will check that carefully anyway, our media is almost entirely innumerate.
I agree with Atrios that austerity is more about kicking the poor than anything else, but right-wingers will always be able to find some kind of data that shows austerity is awesome, enough to convince pundits to keep telling everyone that austerity is awesome. But it won’t probably won’t be enough to convince voters to vote for more austerity. How many divisions do the UK and European equivalents of Robert Samuelson and Joe Scarborough have?
Omnes Omnibus
I found this info about the “coding error” interesting as well:
Source. I don’t come from a data crunching background, but isn’t that more than a simple coding error? Isn’t it a fuck-up that should cause the whole paper to be tossed?
ranchandsyrup
This is how Scalia treats the law.
yam
I agree with Ed, this looks like someone wanted to make a point and worked backwards from the conclusion.
Pongo
Given the recent tragedies (and basic innumeracy of the media and the populace), this will probably not get much play, which is a shame. The authors of the original paper are acknowledging the errors, but claim there was no effort to selectively include only data that supported their hypothesis. That’s a little hard to believe (see the example of New Zealand where only one year was included in the data set–the one post WWII year that showed the slowest economic growth), but even giving them the benefit of the doubt, it doesn’t explain the incredible sloppiness of two renowned economists in managing a basic Excel spreadsheet.
This ‘substantially’ flawed paper is the basis for much of the global austerity movement causing significant suffering around the world.
Face
@Omnes Omnibus: Yes. That’s known as cherry-picking the data. Unless they had rules in place beforehand to outline what data they could exclude, what they did is scientific malfeasance.
Zifnab
:-p Sounds more like the Brave Old World to me. Can you imagine how long Arthur Laffer would have lasted on an internet web forum, if he’d tossed out his little graph in the ’10s rather than the ’80s? We’d be kicking around “Scumbag Economist!” and “Hey look at the funny line I can draw through a collection of points!” memes for years afterwards, rather than reading about the asshole as some kind of economic genius of his age.
Historically, its never been about the science itself. Always about the amount of publicity behind it. But people are louder and more aggressive in rebutting things they don’t like now, because they have access to a much wider field of opinions – many of whom are simply interested in getting the right answer, rather than playing the game.
Those data-focused middle grounders provide the partisans on the right side of the issue a lot of valuable ammo that they didn’t have 30 years ago.
Frankensteinbeck
National television and newspaper pundits really, really like kicking the poor. See Brokaw’s recent speeches for a perfect example.
Omnes Omnibus
@Face: Even if one is willing to grant that the data wasn’t intentionally cherry-picked (being generous here), leaving out over 25% of the data just fucks the thing into a cocked hat, right?
Supernumerary Charioteer
One of the reasons that this went unnoticed for so long was that Reinhart and Rogoff didn’t let anybody see their data until recently, several years after they published the finding and were pounding the op-ed pages with it. So it’s not just an issue of not checking, it’s an issue of weighting out the results that are counter to your finding, screwing up your calculations, making a case for a far-reaching policy and then not telling anybody how you arrived at your results.
Davis X. Machina
I wouldn’t count on that, at least not in the UK. Kicking the poor is popular with just about all non-poor sectors of society.
One of the enduring legacies of the Thatcher years is a complete breakdown of the idea of social solidarity. She killed off the kind of One-Nation Toryism that made it possible for a Churchill to hire a Beverige.
If the cost of putting the boot to the undeserving poor, asylum-seekers, and other lesser types is that the City prospers at everyone else’s expense, well, that’s an acceptable price.
(Lack of a coherent, clearly-better Labour alternative is a problem, too)
Irish Steel
Analogies to household finance, and the enormous sums involved in government spending, put us Spendocrats behind the 8 ball rhetorically every time.
Zifnab
@Irish Steel: Living behind the 8 ball makes you a better pool player. Dems might take a lot of shit and live a hard life, but I’ll take them over the Republican tools any day. Easy life does not make for good politicians.
Omnes Omnibus
@Irish Steel: WRT household finance analogies, one can always bring in car loans and mortgages as beneficial deficit financing.
the Conster
This and the report that just came out on the Bush torture regime will be completely dismissed because there is no hippie punching in it.
scav
@Pongo: Problem is, it’s frightfully easy to run an analysis of “random errors” and give some probability of The Random Error in Your Favor Card coming up. (and did they learn their denying techniques from dim undergrads?) That NZ bit is indeed vedy vedy suspicious. They better hope their punditry careers are well launched, I’d think the strains of a Missa defunctorum academically might be playing.
PeakVT
@Omnes Omnibus: Read this, which is a good summary of the paper. Then read this on the causality of debt, which is opposite what R&R pushed in their op-eds.
Yutsano
@the Conster: The report needed moar both sides do it so BoBo could honour the spirit of David Broder and do a serious column on it where he concludes it’s all the hippies’ fault. Otherwise it might as well not exist.
Davis X. Machina
A second on that Rortybomb paper link.
For the people with their hands on the levers and switches, it’s no biggie. So they got the right answers from the wrong data….
Chet
Look, if Excel (!) is your primary data storage and analysis tool, then you’re not even close to being in the world of “big data.” You’re more in the world of “you know nothing, Jon Snow.”
If your RDBMS doesn’t have a terabyte-order footprint, you’re nowhere close to “Big Data.” Looking at a couple of metrics and annual GDP for an exhaustive enumeration of the countries in Europe isn’t even “a little bit of data.” Don’t get me wrong, there’s a lot of data-driven stupidity out there, but don’t lay R&R at the feet of “big data” or the community of data scientists. I know biology undergrads that wouldn’t make so elementary a mistake, and the notion that there’s a field of study out there that uses Excel (Excel! On desktop computers!) as a basis for peer-reviewed research is just astonishing. How did this ever pass review? Economics is a field with no rigor, apparently. Might as well call it “monetary theology.”
Irish Steel
@Omnes Omnibus: That’s when I whip out the enormous sums (millions! BILLIONS TRILLIONS!)
My debt buys respectable middle class accoutrements in aid of raising a family. Your debt is all T bones and caddies.
@Zifnab: Maybe I’ll take up snooker…
Another Halocene Human
@Omnes Omnibus: From the initial description I read, it was a mixture of cutting out years that disconfirmed their theory (cognitive bias at work!) AND stupid mouse/Excel tricks that excluded cells they MEANT to include, although they didn’t notice because the wrong number they got with their idiot methodology was even “better” for their case.
Of course, nobody could fucking replicate that.
The click-drag boo-boo is significant, because the fact that their findings couldn’t be repeated is what brought this all to light. If it was just methodology vs methodology it would stay an obscure argument overshadowed by ideology.
Whereas now they look like clowns.
Though trust me, scientists do toss out plenty of data (for various reasons), but tossing out data because you have a fixation on a certain result is just junk science and clown stuff. (Granted, it happens in play to pay pharmaceutical papers a lot, and that’s why they have to register with the government now so they can’t change their hypothesis after the data is in.)
And @Chet: Word.
In my experience, not all economists are Excel loving (like a freshman undergrad?!) wannabe’s, but the fact that this stuff can be presented with a straight face is what Milton Friedman and the Chicago Boys have wrought. Championing a theory that failed to predict the observed results but pleased a few greedheads in high places.
Market traders look to finance rather than economics for insight. That field seems to have a better model of the market than most of the economics field although there are exceptions like Roubini, Mandelbrot, Network Analysis experts, etc.
Emma
@Chet: This is what amazes me. I use Excel to transfer large amounts of non-financial data from one place to the next. After my experiences, I would never trust it to do any sort of analysis above the most simple.
Belafon (formerly anonevent)
@Chet: It didn’t pass review. The authors wouldn’t let anyone see the data until just recently, and when that happened, the review failed. What the paper did was confirm a lot of rich people’s biases, and so therefore it must be correct.
As for excel, no this isn’t “big data”, but considering the size of the data, Excel is more than fine.
There are lots of reasons this paper shouldn’t have had the influence it did.
DougJ
@Davis X. Machina:
Labor’s beating Tories by ten points in polls. And Lib Dem support is half of what it was at the last election. Shitty economies are not popular.
Omnes Omnibus
@Another Halocene Human:
Yeah, it would make sense to me if they excluded a year here or there in some country because a volcano had erupted or Mothra arrived and messed up the economy, but one would explain that, right?
@DougJ:
I thought you said you didn’t understand UK politics. You seem to have a good handle on this bit.
MosesZD
Point of correction, they didn’t ‘find data’ that supported their position. They cheated like Tiger Woods at a brothel.
You don’t, ever, in data analysis, do what they did in weighting the series. Unless, of course, your goal is to commit academic fraud in which you support whatever political philosophy you wish to support.
Another Halocene Human
@yam:
@Face:
I wouldn’t even call it malfeasance. It’s not science, period. It’s some sort of exercise in rhetoric, but the kind of sophistry that gives rhetoric a bad name. They say never let data get in the way of a good theory, well, these guys go a step further and modify their hypothesis retroactively to get an answer that pleases them. Well, you can always look backwards in a pile of data, cherrypick coincidences, and make statements like “no designated hitter has ever exceeded 2 HRs per game in away games in the playoffs during presidential election years when there were less than three named hurricanes hitting Florida by August”.
Biff Longbotham
A day without a Stalin-inspired quote is like a day without sunshine.
? Martin
@Another Halocene Human: It’s pretty generous calling what they had ‘data’. My kids do more rigorous data analyses in middle school.
Another Halocene Human
@Zifnab: The Laffer curve was an interesting thought experiment. The problem was, only the left side of the curve was based on the experience of Western economies–the right side was a conjecture. The GOP treated the conjecture like an emergency and then waved away the evidence based side that said if you lower tax revenue you’ll … lower tax revenue. Now they’ve shifted the argument to “lowering tax revenue will grow the economy”, basically a new twist to the same old lie that you can shrink government yet improve the standard of living. It’s a big fat lie and it always was.
Oh, didn’t help that the conjecture side included a change in governmental system.
The conjecture is probably substantively wrong, from what I understand about the research on the GINI index. In other words, you can redistribute the wealth directly as income or secondarily through taxes and subsidies, and BOTH give you equivalent outcomes. So Laffer’s intuition based on the example of the USSR did not pan out.
cyntax
@Another Halocene Human:
As someone who teaches rhetoric, I approve of this distinction.
scav
If the size of the data needed to address the problem at hand and the needs of the modeling / analysis fit the capabilities of Excel, so be it. It can still be science. Science is more about the process and rigor than the mere size of the mathematical package.
MattF
It’s very easy to reach conclusions that confirm your biasses. First, any computation you do, no matter how smart you are, is full of errors. So, you will correct errors, and continue to correct errors until the data shows what you expect. Then you will stop correcting errors.
Second, there’s the problem of seeing the patterns you are looking for in random data. You find the patterns you are looking for, offer a mild apology for the small sample size (We ran out of money! Our graduate student got pregnant! Half the people who answered the questionnaire (which was only five percent of the number asked) gave self-contradictory answers!), and there you are. There’s a finite probability of finding the answer you are looking for, if you are even a tiny bit careless about excluding outliers, the data will converge to the answer you expect.
catclub
Dean Baker also points out the stupidity of looking at the debt level. The only thing that really matters is debt payments as a fraction of GDP. Half as much debt at twice the interest rate is somehow better? Different? No.
If interest rates rise and we buy back low interest debt for high interest debt, the debt level idiots will think we have improved our situation. No change.
Another Halocene Human
@? Martin: Well, sure. But these are conservative economists we’re talking about. World Book Encyclopedia and CIA Fact Sheets are a wee mite advanced for them. We don’t want to intimidate them and shatter their confidence.
Another Halocene Human
@MattF: High impact journals won’t publish studies that fail the validate the initial hypothesis. Not sexy enough.
Davis X. Machina
@DougJ: That gap will be down to 3% or less come the next general election — it always closes. Never bet against the party who’s basing their campaign on appeals to the worst in people.
I expect no majority party the next go-round, even in their reduced capacity as the other SNP (Somerset Nationalist Party), the Liberals will hold the balance of power again. If Farage wasn’t such a d-bag, the UKIP might.
Alexandra
@Omnes Omnibus:
Wow. 1951 New Zealand. Before containerization, completely dependent on oil imports, a far-flung remote country, almost exclusively tied into producing wool and feeding the British Empire comprising less than two million predomiantly rural citizens with some of the most restricted and protected set of tariffs and subsidies in the world at the time.
Terrible data point for any conclusion on modern economics.
Another Halocene Human
@scav: Excel is a shitty tool as this episode has demonstrated. Yes, I can cow Excel into doing some decent regressions, too. But one little fucked mouse click or keyboard command and Excel can hose months of data in one go (I know, back up back up). It’s also easy to silently botch your formulas, which is what happened to these guys.
I learned long ago the hard way never put anything TRULY important in Excel.
PeakVT
@catclub: That’s right. See Japan for an example.
kerFuFFler
It was just a working paper and had not undergone the review process to be published in an academic journal. Typically working papers are circulated amongst colleagues for helpful criticism before sending them in to a journal for publication. It seems odd that this paper was brought to the attention of people outside academia (politicians and journalists…) before it had even been published.
kerFuFFler
It was just a working paper and had not undergone the review process to be published in an academic journal. Typically working papers are circulated amongst colleagues for helpful criticism before sending them in to a journal for publication. It seems odd that this paper was brought to the attention of people outside academia (politicians and journalists…) before it had even been published.
The Snarxist Formerly Known as Kryptik
If the last few years have proven anything, it proves that data is meaningless, unless it bolsters the GOP’s long-held beliefs. Using it to debunk their bullshit is meaningless and pointless since they’ll just go on doing what they will do anyways and never paying a fucking practical consequence ever because they pretty much run the entire fucking show from top to bottom now and will for the next decade or so at minimum.
Redshift
If you needed an illustration of how the debunking of this paper will have no effect on our national dialogue, yesterday my senator (Warner) gave a speech where he said that debt is a bigger threat to the country than terrorism.
Sigh. I know the alternative is much worse, but I hope I live long enough to see a time when I don’t have to have such low expectations for Virginia Democrats.
Alexandra
@DougJ:
Yeah, but people aren’t convinced about Ed Milliband as PM. The Tories will run Cameron as a leader, a global statesman and all that wankery. Labour will not win by 10 percentage points… and UKIP won’t even win a seat.
? Martin
@scav:
The issue with Excel isn’t the size of the package. The issue with Excel is that it’s exceedingly easy to fall into data traps with it – because it’s shit. Data validation is crap. Repeatability is crap. Getting data into it reliably and with verification is shitty. And because it’s slow as fuck, you’re disinclined to build in those secondary checks because it grinds your workflow to nothing.
Even with smallish amounts of data, it’s much harder to fuck up in any kind of proper statistical tool. And the only real reason to dump those tools is because you’re either lazy or don’t know how to use them – and either case should disqualify you from being taken seriously on something like this.
You use the big tools because they’re the right tools, not because they’re too big for the task.
? Martin
@kerFuFFler:
Odd isn’t the word to use here.
Poopyman
@Damn near everybody: Coming back late to the party, but this looks like a total hack job, the cold fusion of economics. Then again, calling Economics The Dismal Science does damage to the definition of science. As does this paper.
The Snarxist Formerly Known as Kryptik
@Redshift:
The scary thing is that realistically, that might be the best you’re going to get for a while, the way things seem to be heading. At least you can tell yourself that at least it’s not West Virginia, small comfort as that might be.
scav
@Another Halocene Human: And no one’s ever done crap analysis or blown data in Systat or any other large program? They’re all 100% error proof and never mis-applied? I’m not arguing its good for all things, but at some point, this seems to verging on saying there was no science done with paper and pencil because it was easy to make errors on that platform. The critical flaw is using the tool inappropriately, carelessly, if not with deliberate intent.
PeakVT
@kerFuFFler: I believe the R&R paper was presented at a conference and published in the proceedings of the conference by a big-name journal. It’s a distinction that few outside of academia would know about unless pointed out (which it has been in the case of the R&R paper). Most people (including me) would assume that published in a big-name journal equals being peer-reviewed.
kerFuFFler
Yeah, I was putting it mildly.
scav
@? Martin: All I’m saying is it”s ok to use little tools for little problems, which seems to be the flip side of what your’e saying.
Another Halocene Human
@kerFuFFler: It seems odd that this paper was brought to the attention of people outside academia (politicians and journalists…) before it had even been published.
That’s one way to put it, I suppose.
@The Snarxist Formerly Known as Kryptik: Of course. Facts harden cognitive biases in a debate, and the arguments the GOP and 1% are bringing are just an attempt to persuade those who don’t know any better–they don’t touch on their actual motivations.
@? Martin: Amen.
@scav: Excel is an easy to use tool for simple spreadsheets. It’s not well-suited to anything more complicated than that.
I have friends overseas who are VBA for Excel wizards (basically so they can do plots you can’t access from the misnamed “chart wizard”) but I wouldn’t be surprised if they import their data from CSVs because Excel + data integrity = failwhale.
Another Halocene Human
wow, I think I exceeded FYWP’s link limit by linking to other posters on this page. That’s not even a FYWP. That’s more of a SMH.
Gin & Tonic
@Chet: Excel is used every day of the year for transactions large enough to destabilize countries’ finances.
artem1s
http://www.spiegel.de/international/business/risky-economic-plan-for-japan-inspires-hope-and-fear-a-894625.html
Interesting article on what Japan is doing (not austerity). The banking class in Europe is going to lose their shit if Abe can this pull off.
SiubhanDuinne
@kerFuFFler:
Ed at Gin & Tacos this morning (his entire post is worth reading):
? Martin
@scav:
But this isn’t a little problem. This is a big fucking problem. We’re talking global economic problem. There is no bigger fucking problem than this one. The volume of data isn’t relevant, the certainty that it not be done wrong is what’s relevant. And Excel sucks at that.
Ben
Tip for everyone that thinks it is OK to use excel for more than family budgeting. It isn’t 1990 anymore. That kind of shit is used for preliminary work, but not for any serious statistical modeling. Way, way too easy to make errors. Using something like SAS or any DBMS (MS SQL)? Not so much.
Roger Moore
@Omnes Omnibus:
I think school loans are probably a better example for this particular case. The goal is to improve yourself so you can earn more in the future. Sometimes the increase in future earnings is big enough to swamp the borrowing costs, and very frequently the time to do it is when you’re in bad financial straights because you’ve lost your job and can’t find another one in your area of expertise. Deficit spending to get the economy going again is the same basic idea; we’ve lost a big chunk of our income, so it makes sense to spend money that will boost future earnings.
Haydnseek
@Supernumerary Charioteer: Exactly. Remember every math class you ever had? An answer isn’t enough. Show your work.
Omnes Omnibus
@Roger Moore: I think student loans work as well. The big thing is to be able to point out that most household budgets have long term debt and, as long as one is getting something good in return and can service the debt, it is not a bad thing.
? Martin
@Roger Moore: Simple rule: borrow for assets that increase in value, save for assets that decrease in value. Examples of assets that increase:
1) Your home (generally speaking – particularly with tax incentives added)
2) Your education (provided it is likely to lead to increased/improved employment)
3) Your health (provided it’s necessary)
There are some additional edge cases, but that’s pretty much the whole list, and there are exceptions to each of the 3 there as well. Borrowing for a car is idiotic. Even the edge cases of rare cars that appreciate in value are so volatile that it shouldn’t be done. Same with stocks. Revolving credit cards are idiotic (except when it’s absolutely necessary for health, above). Pay them off every month.
ChrisB
Don’t know if others have mentioned this but Krugman has been all over Reinhart-Rogoff on his blog. The latest is here:
http://krugman.blogs.nytimes.com/2013/04/17/blame-the-pundits-too/
scav
@? Martin: A) Did I ever say they, those specific individuals were using the correct program to task, let alone doing it carefully? Why do we assume they would have done it carefully on a different platform? You’re free to be more suspicious of Excel based analysis because of the possibility of error but people have managed to be methodical and rigorous enough to get it to do what they need.
People even managed to write code that worked in Basic without variable typing and other handy things that avoid mistakes.
kerFuFFler
I get really frustrated that people tend to focus so singlemindedly on the size of the debt rather than considering the different natures that debts can have. If someone borrowed 500 bucks and spent it on fireworks, in the end they just end up with the 500 dollar debt—–and all of their fingers if they were careful. If they bought a lawn mower and started earning a living with it the debt could be paid off over time, and the person ends up better off for having taken on that debt because it was an investment. Governments are supposed to look for opportunities to invest in our society that will pay off in the long run! Not every tax dollar spent will fall into the category of investment, but we will all be better off when we enable the government to do things like improving education to fight poverty, funding scientific research, and regulating polluters to lower health risks and other hazards.
A lot of our military spending seems just like buying fireworks—– fireworks we have to hope we never have to use.
Roger Moore
@kerFuFFler:
Not odd, just a sign that economics is dominated by political concerns rather than normal academic standards. The correct thing to do if somebody does circulate your working paper in the media is to respond to requests for discussion by pointing out that it’s a work in progress, not taking advantage of it to hit the talk show circuit. That’s the real malfeasance.
Just Some Fuckhead
I hope this doesn’t unfairly tarnish all Elvis impersonators.
Roger Moore
@The Snarxist Formerly Known as Kryptik:
Pretty much. The best example I can think of is the difference in their response to uncertainty about Iraq and about Global Warming. In Iraq, a 1% chance that they had weapons of mass destruction had to be treated as certainty to justify war. With Global Warming, anything less than ironclad proof that it’s happening is sufficient justification to ignore the problem. They will do WTFTW and seize on any justification they can find to do it.
Belafon (formerly anonevent)
@Another Halocene Human: Do you know how much code, not in Excel, that your statement could apply to? This wasn’t Excel. They would have done the same thing to Matlab. This was two guys who had a bias towards a certain result and played with the data until they found it. If they hadn’t made that mistake, they would have probably just eliminated more data.
PeakVT
@Roger Moore: It’s still not a great analogy because, in a slump, the government can borrow and spend on just about anything (or at least anything with a multiplier greater than 1) and it will help. That’s not true at the household level. SNAP and UE benefits have the highest multipliers; tax cuts for the top bracket and accelerated depreciation (which are tax expenditures) the lowest.
Roger Moore
@Redshift:
He’s probably right, but it’s because we’re grossly exaggerating the danger of terrorism, not because we’re underestimating the danger of debt. Terrorism is not a serious threat to national security, and it’s pretty much guaranteed never to be one. Debt may not be a threat now, but it’s at least conceivable that it could be a serious problem in the future.
? Martin
@scav:
Because they wouldn’t have started with a bad decision. That’s how competence is estimated. When you interview people you don’t have them do their entire job and then evaluate them after the fact. You look at their decision making process and then draw inferences from that. These guys started with a bad decision. Reasonable to conclude that more bad decisions would have followed. We can run that backwards as well – “They decided to use Excel for this? Double check their work – they probably don’t know what the fuck they’re doing.” It’s a red flag that everyone ignored.
patroclus
The major problem with this sort of analysis, quite apart from the statistical errors, is that budget deficits are, relatively speaking, not a large causative factor of economic performance generally and growth specifically. Instead, the deficits are more reflective, that is, they are a result of good (or bad) economic performance. A high budget deficit is an indication of a poorly performing economy; not necessarily a cause of it.
The reason for this confusion is because fiscal policy is clearly a causative factor and the level of budget deficits is clearly a part of fiscal policy (taxes and spending). But the deficit is less important than the spending and the taxes themselves. This is the crux of Keynes’ analysis of effective demand and why one should expand spending/investments (and hence deficits) in a poorly performing economy and tend towards restrictive policy in a robust economy.
Politicians, statisticians, laypeople and (certainly) the media confuse this all the time. Austerity does not promote growth; spending and investments do. Regardless of the impact upon an annual deficit.
artem1s
@kerFuFFler:
A lot of our military spending seems just like buying fireworks—– fireworks we
have to hope we never have to usewe always end up inventing an excuse to use.the Conster
OT, but y’all not here in Boston are missing some pretty great stuff.
Valdivia
So glad you featured this DougJ. Even if I am late to the party. I am laughing about all the excuses all the righties have made for the study. Austerity can never fail it can only be failed!
Omnes Omnibus
@kerFuFFler: @artem1s: Over the past 12 years, it has been one of the biggest sources of stimulative spending, so its got that going for it. Of course, much of that money would have provided better stimulus with a greater multiplier if it had been spend on infrastructure.
Trollhattan
@the Conster:
The president brought it. I’m there with you all in spirit.
http://www.boston.com/metrodesk/2013/04/18/people-line-for-ten-blocks-around-cathedral-the-holy-cross-attend-interfaith-service/65uhwU9x6ZMvSTjsYOHQ0M/story.html
jl
No time to read the comments carefully, so just a few comments from my point of view (trained as an economist, for better or worse).
I agree with SiubhanDuinne’s link. This paper was never peer reviewed, it was a write up of a talk at the annual AEA conference. There have been plenty of red flags about the results for quite awhile. Also agree with catclub’s pointer to Bean Baker’s points. The fact that there is no real relationship between the real burden on the economy and the ratio of public debt to GDP is one of the reasons people criticized the work as merely correlational, with no real theoretical or causal model underlying it.
I don’t understand why DougJ mentioned ‘big data’. this is very small data. The link DougJ provides does not emphasize enough that there are actually THREE problems with aggregate macro data. One is the small sample sizes that describe stable macroeconomic regimes (e.g., post WWII economies for example). But, two the fact that different policy interventions that should act differently under different macro theories are highly correlated, so very hard to disentangle their independent effect. Third, the time series of outcome variables, like unemployment, GDP have extremely high serial correlation, which reduces the effective sample size.
So, the problem is not big data, but very small data. And different theories that are observationally equivalent given the available data.
kerFuFFler
Yeah, we do that I suppose….
I guess I was just thinking about our enormous nuclear arsenal. Such a frickin’ waste especially when you consider the expense of storing, guarding and maintaining them.
Trollhattan
@Just Some Fuckhead:
At your Vegas wedding, accept no powdery souvenirs from the minister. Just sayin’.
scav
@? Martin: You’re assuming that by not using excell they wouldn’t have started with a bad decision and so everything in the analysis would have come out hunky dory and correct and they’d have written an entirely different paper that would have been properly peer reviewed and transparent and no dodgy NZ magic and no scary pronouncements about the correctness of their conclusion,despite the errors in their model, all that is predicated and understood by the initial choice of a program. Clearly I am misunderestimating the beardie sneekieness of The evil Spreadsheet. Forget The Database, That’s so last decade.
Or is it just that scientific peer review can ditch looking at all the logic and data issues now and focus on the single critical issue of software choice? This hairball really has failures through several steps of the academic / scientific process but software choice is the most critical.
Roger Moore
@kerFuFFler:
This. We need something closer to a national balance sheet that shows us both our debts and our assets. That way we’d be looking not just at the effects of borrowing and spending, but also at things like the value of natural resources being exploited vs. royalty payments.
Anoniminous
People need to adjust their interpretive regime.
Economics is a Dismal Science but it’s a really good Cargo Cult.
jl
On other hand, empirically, if one has an open mind (OK, that is a stretch in economics, but work with me here for awhile) the situation is not as dire as people think. Empirically, you can do more than just run regressions on official national income and product accounts time series (or cross sections of time series). You can look at
out of sample predictions and forecasts,
survey and program enrollment and payments data on flows between different states (very good for labor markets and some financial markets),
You can look at time series of central bank and legislative spending actions as initial estimates, revisions and final estimates of financial and national economic accounts data unfold to get a better idea of how actual government and central bank policy reaction functions really work.
This stuff is very hard to find in U.S. macro journals. Have to go to international, U.K. and European journals to see much of it.
Redshirt
I’m terrified of Big Data. Seriously. I worked in Big Data too long not to fear the consequences of ever more powerful and complex hardware/software.
Just one example: Facial recognition software that can scan crowds and match every person to a photo id in a database.
JPL
@the Conster: This link says it all.
With you in spirit.
Mandalay
@Chet:
Ecomomics is not really a field at all. It is chunks of history, sociology, psychology, finance, accounting and statistics masquerading as a field in its own right.
The world has recognized that “geography” is not really a meaningful field of study, and the sooner we do it for economics as well the better. That might mean that people who want to crunch numbers should be mathematicians and statisticians rather than “economists”.
the Conster
@Trollhattan:
The governor brought it too – big time. The mayor, who badly broke his leg last week, checked himself out of the hospital on Monday afternoon and managed (with a visible wince) to stand from his wheelchair to give his speech which was almost completely intelligible. But when the president looked right into the camera to address the injured in the hospitals, well, I got something in my eye. He’s visiting them all now.
jl
@Mandalay:
” How did this ever pass review? ”
It was never peer reviewed. It was a non-peer reviewed conference paper.
Gin & Tonic
@the Conster: Menino, almost completely intelligible? That must have taken a superhuman efort, much more so than standing on a broken leg.
Omnes Omnibus
@Mandalay:
Given that economics is an older academic field than psychology or sociology, I find your theory intriguing.
Eric U.
@Omnes Omnibus: yes, the paper should be retracted. Unfortunately, the damage has been done
The Raven on the Hill
Good discussion over at Crooked Timber, with very funny lead-in from Clippy. Honest scientists have a way of being short with people who cook their data; favorite remark in comments: “This paper is pure, self-serving fraud that doesn’t rise to the level of astrology.”
weaselone
It seems that for several posters powerful data analysis software serves as Porsche substitutes. The problem with the study was the Tools running the study(R&R) not the tool(EXCEL) used to crunch the numbers. Yes, EXCEL is pathetic compared to what is available out there, but it is more than adequate for many tasks. The problem is that too many groups are feeding in garbage and too few are actively checking for garbage coming out regardless of whether they are using huge, throbbing or puny, shriveled data analysis tools.
I was always curious about how Bernie Madoff was able to get away with his crime for years after the authorities were 1st tipped off by Markopolis despite the fact that the fraud could easily be detected using mere EXCEL. I guess everyone was busy laughing at the size of Markoplis’ software package while Bernie made off with a few more billions.
artem1s
@kerFuFFler:
yea, we did manage to keep a few fingers off the button with those. But I always got the impression with Star Wars that if Raygun and his pals had ever gotten it to work they would have used it as an excuse to make pre-emptive strikes on the looney theory that we could keep the Commies from retaliating. I remember clearly the cold war, chickenhawk opining at not being able to nuke the enemy whenever they wanted. And nuclear winter was as universally disdained as pinko academic propaganda, just as much as climate change is reviled today.
burnspbesq
@Zifnab:
Well, actually, no. If you believe in the diminishing marginal utility of wealth, the Laffer Curve is logically inevitable. The question is where the inflection point lies, and the Piketty-Saez work on optimal tax rates pretty conclusively establishes that the inflection point is at a tax rate higher than any rate currently being imposed anywhere on earth.
jl
@The Raven on the Hill:
Thanks for link to Crooked Timber. It was nice to see MS Mr. Paperclip making intentional humor (as opposed to unintentional, as is his wont).
Sadly, there is a history of prominent economists doing poorly vetted analysis that has big policy implications, and then being passive aggressive when it is called into doubt (to be very polite about it).
Martin Feldstein once claimed to have found that Social Security produces massive productivity losses, only to have that result debunked when an error was found in his data processing. He later did a revised analysis, with the same conclusion, only to have that result changed due to an error. Dean Baker writes about that episode in his Beat the Press blog today.
But the myth that Social Security produces huge productivity losses lived on for years.
the Conster
@Gin & Tonic:
Who knew all this time all he needed was the hard stuff?
Chris
@patroclus:
The trouble at least in our country becomes the number of people who, to borrow from Krugman, “see economics as a morality play” and support austerity not because they think it works (they don’t care if it works) but because it’s the just punishment on a sinful population for allowing [socialism, runaway debt, poor work ethic, insert favorite right-wing-half-understood-bogeyman here].
Suffern ACE
So what was this paper that was different from their book? In the book they were looking at all kinds of national debt, not just government debt. Did they rewrite their book eliminating private sector business and consumer debt to come up with a magic number of where government debt ought to be?
Mandalay
@Omnes Omnibus:
I have no idea why; I think your point bolsters my argument. Just because we have had a taxonomy in a certain form in the past does not justify its continuation, and advances in psychology and sociology (for example) render economics less relevant as a discrete field of study, as do the massive advances in the capture and analysis of economic data, which have much stronger ties to statistics and mathematics rather than economics.
And just because a topic is new does mean that it is not a perfectly valid and coherent field of study (e.g. Linguistics, AI).
The only reason “economics” exists now as a field of study is that it is so entrenched. The world eventually accepted that “geography” was a bogus field (in terms of taxonomy, not the issues it looks at), and I think economics should, and will, eventually go same way.
Mnemosyne
@weaselone:
I think they’re making the kind (and possibly unwarranted) assumption that the researchers made honest mistakes because of the limitations of Excel and not that the researchers deliberately cooked the numbers and then blamed the program.
FWIW, it appears to be a both/and situation — they cooked some of the numbers but also screwed up their Excel formulas in a dumb way by not clicking and dragging far enough.
Eric U.
@Mnemosyne: Excel was fine for what they did, they just didn’t use it right and didn’t care to check their work — if it wasn’t simply fraudulent in the first place.
I think the only problem with the Laffer Curve is that the location of the inflection point is not commonly known and all the righties ignore the fact that the curve is not monotonic. It clearly shows that if you lower the tax rates too much revenues go down
Omnes Omnibus
@Mandalay: I disagree with your take on this. I think economics looks at society from a particular angle that other disciplines do not. And vice versa. No discipline will ever manage to sum up and explain human behavior, but the more different approaches that are taken the more likely we are to get somewhere. Also too, this is a good reason for cross-disciplinary work and education.
FWIW I am not an economist by any stretch of the imagination – I took Macro and Micro in college and that is it.
jl
@Suffern ACE: Looks like WP ate my reply to you. and no time to retype. Krugman addressed the differences between the book and conference paper under attack in is blog recently.
Edit: since edit seems to work, in brief: more historical context in book, more causal theory, less sweeping conclusions in book
raven
@Omnes Omnibus:
Econ
Buy low, sell hi
Guido Sarducci
Omnes Omnibus
@raven: If one is high while selling, doesn’t that violate the basic principle of “don’t get high off your own supply?”
patroclus
@Chris: Agreed – there are certainly austerity supporters who believe that it’s a morality play rather than a policy argument. But the problem is even broader – everyone considers themselves an expert on economics kind of the same way that everyone thinks that they’d be a better football coach or baseball manager than anyone else. So every conceivable factor whatsoever can be used to justify opinions on economic policy – religious, moral, scientific, personal experience, old wives’ tales, you name it, it can be used.
And economics as a discipline doesn’t evolve in the way other disciplines evolve – as omnes says, it’s been around for so long that there aren’t any major new discoveries that either are made or that can be made. Fiscal policy is either expansionary or contractionary – the analysis doesn’t change with each new study. And monetary policy is similar – once the monetary regulators became self-aware post WWII (like SkyNet), we haven’t learned all that much about the monetary aggregates and what works and what doesn’t. The tools are the same – open market operations, the discount rate, reserve requirements and capital adequacy have the same sort of expansionary/contractionary effects as taxes and spending. And we know what they are.
And what we know is that expansionary policies are appropriate in times of poor economic performance; not austerity measures.
raven
@Omnes Omnibus: I can’t remember.
Roger Moore
@Eric U.:
Am I the only one who looked at their Excel spreadsheet and thought that their “research” would have been more appropriate for a school science fair than a serious academic publication? I can imagine a High School sophomore making the methodological choices they did, but they seem like the kind of thing that would be unacceptable for an undergraduate and laughable for a serious researcher.
Chris
@patroclus:
And to make matters worse, when people find themselves looking for experts to trust on economics, they often don’t think of economists, they think of businessmen. Because “they live in the real world not the ivory tower,” “they make lots of money so they must know how the economy works,” etc. Which, since businessmen as a class tend to be represented in society via things like the Business Roundtable and Chamber of Commerce, gives the righties even more clout.
And you’re right about things being known – at least, I don’t know if everything in economics is known, but certainly we know enough for things like how to weather a recession.
James E. Powell
@kerFuFFler:
It seems odd that this paper was brought to the attention of people outside academia (politicians and journalists…) before it had even been published.
Odd? More like typical. For the politicians and journalists, this paper had no influence on what policies they advocated.
They advocate austerity (for others) because that’s what the ruling class wants them to advocate. They are in favor of whatever the ruling class wants. They always were and always will be.
As it was with Laffer, the “expert economist” support for policies is only part of the propaganda. They only need it because they can’t really say “We are going to work for the benefit of the ruling class and screw everybody else.”
patroclus
@Chris: Indeed – Republicans, especially, often campaign on the basis of being good at business means being good at economics. Romney’s entire campaign was based on that.
But the problem is deeper than that too. Economics has evolved (some might say devolved) into sub-specialties, there are bank economists, equity economists, agricultural economists, there used to be labor economists, the big field these days are statistical forecasters who supposedly can predict the future based on econometric models, microeconomists and many many more. There just aren’t many jobs for general macroeconomists – Krugman is one of the few exceptions. So, even, when you look to an “economist,” you aren’t necessarily going to get good advice from someone knowledgeable for the answer you seek.
Moreover, in the U.S., fiscal policy is determined by Congress; not the Treasury and certainly not by economists. That makes every aspect of fiscal policy susceptible to politics generally with all the idiotic biases that politics entails (which contributes to what I said earlier about everyone being an “expert”). So, even though the U.S. has slow growth right now, we just stupidly imposed tax increases (FICA and upper incomes) and we just stupidly imposed sequestration – both of which are demonstrably contractionary. Luckily, the Fed is not as politically constrained and it’s been pumping $85 billion per month into the economy so as to peg the short-term rates really low, but overall, current overall policy just now is a mishmash of some contractionary policies and some expansionary policies which continues to cause a mishmash of confusing not-particularly-expansionary results so we continue to have slow growth and (historically) high unemployment. And, of course, high deficits (although they are declining); which leads to calls for more austerity.
? Martin
@scav:
I’m just saying that by starting with a bad decision, more bad decisions should have been predictable. I’m not saying that a good decision is protection against future bad decisions.
Converse, not contrapositive.
Chris
@patroclus:
Well, to mirror what people on the other thread are saying about the media, I don’t know what our solution to fiscal policy – at least that last part – would be. Sure Congress setting policy opens up the process to a lot of stupid politicization, but “politicization” of that kind is a natural side effect of the democratic process.
(And to your first paragraph – I do object to Republicans thinking that being good at business means being good at economics. I’d add, though, that I think there’s a large chunk of our businessmen, who tend to be the ones who like to promote themselves on Fox News, who don’t even understand business either).
Tonal Crow
@Eric U.:
Limits, how do they work?
Tonal Crow
This is very similar to the bullshit that the more-advanced climate-change denialists spew, for example, “global warming has stopped because it was warmer in 1998”.
David Koch
I don’t know where you’ve been Doug, but “hard working white folk” have been voting against their economic interests for decades.
El Cid
Let’s not forget that the basic lesson was that all the people to be affected by their recommended policies, many of whom would be hurt very badly, weren’t worth fucking doublechecking your spreadsheet.
Fuck you, poors, you’ll get a carefully checked argument about whether or not you’ll be hurt by something when you get rich.
Until then, live with fucking arbitrage, bitches!
Another Halocene Human
@Belafon (formerly anonevent): As I said earlier, I respectfully disagree. By making a boo-boo like that, they made it a big story. Had they just engaged in reverse-engineering of their methods (h/t to gin and tacos), then it would just be partisans on both sides flinging poo.
Because their motivation is ideological.
But now our side can paint austerians as buffoons. I call that a win.
Another Halocene Human
@? Martin: Actually, not peer reviewed was the red flag, because nobody could get their hands on the data.
Another Halocene Human
@scav: Are you on the Excel dev team or something? Why so defensive?
I use Excel every single day at work (and often at home off of work), but I’m still going to laugh when somebody uses it for something inappropriate.
Another Halocene Human
@the Conster: Menino seemed clear enough on the radio. He made me cry.
Unsympathetic
Data disproves Republican economics? SHOCKING
danielx
Response to Krugthulu’s blog entry of the 17th. I like it muchly.
Chanson de Roland
Well, now that the two papers which are the fundamental support for austerity as a strategy for economic recovery have been totally discredited and the price of gold has collapsed, I suppose that austerians will have to return to the tried-and-true religious basis for inflicting the otherwise gratuitous pain of austerity: That we are sinful folk who have behaved badly, so pain is our just lot and our only road to salvation.
But I’ve always found two problems, actually three, with foregoing justification for austerity: (1) As an atheist and a Neo-Keynesian, the explanation has no religious validity for me, and, as economics, it is utter nonsense; (2) Like all religious doctrine, there is no way to generate testable propositions so that one can test the idea of austerity as salvation to determine whether it is true; and (3) The most sinful folk, those whose actions brought about the instant economic disaster, the one-percenters and their agents, haven’t suffered and are not being asked to suffer to even the slightest degree.
On the first two problems, I don’t see any solution for them. However, I suppose that the answer to the third problem is that the one-percenters and their lackeys don’t give a damn about the next world but are concentrating on having a great time in this world by any means necessary, no matter what the consequences for others. And let me say that they are doing an excellent job of achieving that objective.
Chet
@Gin & Tonic: Maybe that’s why they do. As almost everyone notices after perhaps 5 minutes of using Excel, it has a decimal precision problem: Excel only approximately stores values.
OmerosPeanut
Hiding poor scholarship – let alone outright errors – by keeping data and the code used to interpret the data private is at least as big a problem for medical research. Frankly, I’m not sure how we’re going to make progress in data openness either.
Ideally, researchers would consider results from private data to be less significant than those from published data with published code for the analysis in the same way that a randomized controlled trial is viewed as giving stronger evidence than an observational study. Since the incentives are all wrong for that to work on its own, I’d love to see some major journals start requiring data and code to be published online in order to be accepted for publication. But then we run full on into the problem that the editors of these journals are exactly the people who have most benefited under the current system and new journals created to push this goal would, at best, take many years to become prestigious.