'The Excel Error Heard Round the World'

[ProSense]

'The Excel Error Heard Round the World'

By Steve Benen

House Budget Committee Chairman Paul Ryan (R-Wis.), at the height of the Republicans' debt-ceiling crisis in 2011, said he and his party were justified in prioritizing immediate debt reduction. Economists, he said at the time, "tell us that letting total debt rise above 90% of GDP creates a drag on economic growth."

Ryan didn't say which economists, but it didn't much matter since everyone knew exactly what he was referring to. Carmen Reinhart and Kenneth Rogoff published a report a few years ago that conflicted with the findings of many economists, but told Republicans exactly what they wanted to hear...Ryan, conservative activists, deficit hawks, the Washington Post editorial board, and the DC establishment waved around the Reinhardt/Rogoff study as definitive proof that debt reduction can't wait -- failing to reduce the deficit, or making it worse on purpose as nutty liberals like me prefer, makes an economic recovery practically impossible. Austerity measures, intended to reduce the budget shortfall, would in turn correct the problem help the economy grow.



The problem, of course, is that the Reinhardt/Rogoff study was wrong. In fact, it's wrong in a variety of important ways, which Mike Konczal summarized very well. Several scholars at the University of Massachusetts, Amherst replicated the results of the Reinhardt/Rogoff research and uncovered some serious problems...the most glaring: Reinhardt and Rogoff made a coding error in an Excel spreadsheet. Kevin Drum called it the "Excel Error Heard Round the World." (I think Kevin means that literally, since the Reinhardt/Rogoff study has helped bolster the austerity agenda in a wide variety of countries.)

How bad is it? Take a look at the chart I put together: according the Reinhardt/Rogoff research, once a nation's debt-to-GDP ratio tops 90%, the result is economic contraction. The revised research based on the same data points to 2.2% growth. It is, in other words, an enormously consequential error.

- more -

http://maddowblog.msnbc.com/_news/2013/04/16/17782034-the-excel-error-heard-round-the-world

[liberal N proud]

1. It told Republicans exactly what they wanted to hear

Error?

Maybe.

[hobbit709]

2. Just like those Faux News "errors" when they showed a (D) after the name of some crook.

[liberal N proud]

10. I think you understand

[DevonRex]

3. I always get facts from you.

Thank you so much.

[AzDar]

4. K &R

[HereSince1628]

5. A good example of the illusion of power in quatitative methods...when the result looks like

what you want you don't seriously recheck the work.

The readers see quantitation and let it go...I'm assuming this economic paper would have been peer reviewed.

Austerity is ruining nations across southern Europe and it's based on bad arithmetic.

[Skraxx]

6. "Error" Is An Exceedingly Generous Classification

[Wednesdays]

7. Excel makes lots of errors.

(A shout-out to fellow anime fans.)

[Triana]

8. Tweeted, FB'd, bookmarked....

and tweeted to @barackobama and @whitehouse (not that I think anyone reads those timelines) - may send an email to the white house with this though.

[WillyT]

9. HUGE K & R !!! - Thank You !!!

[haikugal]

11. Thank you ProSense!

Well done Rachael!! Kick!

[StatGirl]

12. Honest to God, what kind of researcher . . .

. . . uses Excel for serious analysis?

(Note: Excel is great for many things; it's probably my second favorite program. But it isn't for statistical analysis.)

[Lucky Luciano]

19. Sure, but this is not intraday tick data...

...where you may need R, Python, C whatever, or even KDB.

These kinds of stats are probably low on sample points, so Excel with maybe a lightweight add-in can do the job - if the user doesn't fuck it all up if course.

[Twiceborn]

21. Excel is fine for Analysis

A lot of people use Excel for statistical analysis. It really depends on the type of analysis you use.

Excel can even do T-test, F-Test, Chi-Square, etc., if that's what you need.

Excel is great for charts, graphs and to manipulate data in a fast, efficient manner. It can even be used for automated report writing, if you know how that is.

I'm an expert in my work organization and people call on me to produce statistics and reports all the time. People love excel since once a report is generated, it can then be sent to others who can play with the data.

Of course, I'm coming from a financial/educational work environment so I guess it depends on your type of work.

[hrmjustin]

24. Welcome to DU Twiceborn!

[Fantastic Anarchist]

26. I do productivity and operational analysis for my company ...

... using Excel. Great program with which I can do many, many things. The Directors I work with usually are amazed with the analyses I provide.

[jazzimov]

28. Welcome to DU - but Excel sucks.

Access is a lot better, but because it is so "user-friendly" it does a lot in the background in VBA that the front-end user doesn't see and can easily get "confused".

I was trained in Minitab, which is a thousand times better than Excel but still "sucks".

Still, as you said, it's good for sharing.

The bottom line is that these are all tools. The tools are only as good as the artisan using them.

[StatGirl]

33. Welcome!

Welcome to DU!

I'm in academia, and I think that makes a big difference in what is considered acceptable.

The main issue is that the analysis has to match the design of the study. So it doesn't take too much departure from simple designs -- for example, multiple-year data on countries/economies of different sizes, or repeated measurements over time within different groups with some missing values -- for Excel not to be up to the task.

And this particular error of not including all the data is less likely to happen with real stats software, because you have to actively choose observations for exclusion.

That said, I think Excel is terrific for many things, if it's in the hands of people who know something about it. I particularly like to use it to set up problems so that people can tweak inputs to get their own outputs, and it does make nice graphs. But I've seen some purely dreadful Excel workbooks from amateurs, and you probably have, too!

[Danascot]

13. Wondering whether Reinhardt and Rogoff

admitted or apologized for their "error" or are apologies just for Democrats, Liberals and wimps?

[Rex]

15. A person needs a conscience to admit to fault

something you will not find in the GOP.

[JDPriestly]

14. Thanks.

And we see the result of austerity measures in a time of economic contraction in Greece, Italy, Spain and Portugal.

This is similar to what happened in Germany after WWI. We required them to repay a war debt (or at least tried to require them to pay it) thus imposing a sort of austerity on their economy. They had a terrible inflation. Then, their economy contracted, and we ended up with Hitler and the scapegoating of Jews, Gypsies, homosexuals, the disabled, etc.

http://www.history.ucsb.edu/faculty/marcuse/classes/33d/projects/1920s/Econ20s.htm

People supported Hitler because he promised full employment. As the article explains, the new jobs were created by increasing the production of military equipment and preparing for war.

Germany had a balance of payment problem. And it could not find a source of tax revenue that did not harm its economy.

Our situation is very different from that of Germany after WWI, but we too have a severe balance of payment problem. We are willing to export, but we don't have willing buyers of enough of our goods to balance what we buy with what we sell. And our reliance on the personal income tax for the federal government and on sales and property taxes for state and local governments doesn't seem to be working well.

Do we need to harness our expenditures on imports? That's what I think, but then I am not an economist so I am interested in what economists or others respond to this post (if they do).

There are big differences between our situation and that of Germany after WWI.

But in a sense, our tremendous military involvement overseas has been, as war was for Germany, a way to stimulate our economy.

So -- what do DUers think?

Is anyone familiar with the history of Germany and Austria of this period?

[truedelphi]

30. i am not at all sure if our situation does differ from

the economic situation that Germany was confronted with. They were required by the Treaty of Versailles to repay for war reparations. Billions and billions of dollars - only by getting Hitler into power and his refusing to continue to do this, did that stop, did the German people experience relief. (Economically speaking, anyways.).

We have no war reparations of course, but we do have on going wars which under the George Dubya doctrine of having our cake but not raising taxes for it, those wars are exponentially expensive. "The Three Trillion Dollar War", a recently published book on this situation, goes into the seriousness of the problem. After all, as our wars cost some three trillions of dollars when you don't factor into them the losses other than actual military costs, and these losses might total another 2 to 3 trilliions of dollars! But our society really dopes need to consider such things as how much a family and ultimately society is losing when a returning veteran is unable to work on account of injuries sustained in the war.

Then there is the whole matter of the fifteen to sixteen trillions of dollars that the Bernie Sanders insisted upon audit of the Fed Reserve showed us has been "lent"out by Bernanke to his friends at Big Finance companies across the globe. Experts tell us some four and a half trillion most likely will not be repaid. So that is in a sense why the austerity programs are becoming popular with our bought and paid for politicians. And of course the media assists them in this all the time. Turn on the TV, and you'll hear at least one Talking Head every few hours talk about how Social Security is simply not manageable, in terms of expense. The One Percent has to get their ill gotten gains somewhere, and they prefer the hides of us old people rather than their being hauled down to size.

[WillyT]

16. Hey Pro !!! - Check Out The Front Page Of HuffPo's Business Section !!!

More Info: http://www.huffingtonpost.com/business/

[ProSense]

20. Thanks, and

the "error" has a face: Republicans' poster boy for bullshit budgets.

[1StrongBlackMan]

27. Translation into non-conservative speak ...

We weren't wronggggggg! The reviewers just read it wrong. Besides, we corrected the error we didn't make in the study the reviewers reviewed. The re-study we did, two years later, clearly shows that the error we didn't make in the study that the reviews reviewed, really wasn't that bad ... we came close. You liberals are just so stuck on facts and accuracy!"

[SunSeeker]

17. K&R

[UnrepentantLiberal]

18. Does Sarah Palin have a budget plan?

Don't get me started on Jack Kemp and Dan Quayle.

[Spitfire of ATJ]

22. IOW: They cherry picked the intelligence....

Want to hear something funny?

The rich listen to these guys.

[santamargarita]

23. It's all Microsoft's fault...

Republicans can't make mistakes, they are a mistake.

[1StrongBlackMan]

25. I don't claim to be numerically/statistically literate, but ...

What I take from the Chart is that economies perform best when the Debt to GD is >30%; but always grows with government sending. Right?

[SmileyRose]

29. that's not a "coding error"

it's this

[eridani]

31. Another summary link

New Research Undermines The GOP’s Austerity Agenda

http://thinkprogress.org/economy/2013/04/16/1875631/new-research-blows-a-hole-in-gops-austerity-agenda/

In the debate over government spending, the central data point wielded by fans of austerity is the claim that once a country reaches a debt load over 90 percent of its economy — a threshold the United States is approaching — economic growth goes into a tailspin. That argument came from a 2010 study by Carmen Reinhart and Kenneth Rogoff. After surveying a wide number of countries, they found that, on average, once the 90 percent mark is crossed, economic growth slows. Though the paper always had problems that kept many economists from embracing it, that didn’t stop it from becoming “the most influential article cited in public and policy debates about the importance of debt stabilization” as Slate’s Matt Yglesias put it.

There were already problems with the Reinhart-Rogoff study, but up until now, other researchers haven’t been able to replicate or pick through its numbers. A new paper finally has, and as Mike Konczal over at Next New Deal reports, it dug up some truly mortal flaws.

First, Reinhart and Rogoff excluded the post-war years for certain countries that enjoyed robust economic growth despite debt levels well over 90 percent. They also chose a skewed method of weighting the data: for example, New Zealand’s single year of terrible growth while over the 90 percent threshold wound up counting just as much as Britain’s 19 years of healthy growth. And they even incorrectly input at least one Excel spreadsheet formula, wrongly excluding several countries form their calculations.

[renate]

32. there are links to the spreadsheet here

http://www.guardian.co.uk/commentisfree/2013/apr/16/unemployment-reinhart-rogoff-arithmetic-cause

(in the seventh paragraph). I'd sure be interested in a statistician's evaluation.