37

A candidate to the French presidency tweeted

The fall in industrial production in Spain, Italy and France of course has a link with the euro.

enter image description here

However I have a couple of doubts on its accuracy because of its source and the arbitrary use of gauges and percentages.

Is the data accurate? The graph shows an "euro effect" according to the OP. Is this an artifact of the way the graph is made? What happened to countries outside of the euro?

Community
  • 1
Sklivvz
  • 78,578
  • 29
  • 321
  • 428
  • 99
    *Does* the graph show an "euro effect"? I don't actually see it. I see that France, Spain and Italy failed to rebound from the 2008 crisis the way Germany has, that's about it... – DevSolar Mar 21 '17 at 09:22
  • 61
    So the "euro effect" did not happen until 10 years after the "creation of the euro"? – hdhondt Mar 21 '17 at 09:23
  • 4
    You can find OECD data at https://data.oecd.org/industry/industrial-production.htm though you may need to change the parameters to get exactly what you want. The (non-euro) UK will look similar to France – Henry Mar 21 '17 at 09:25
  • 2
    Is this question about the correctness of (i) the graph, or (ii) the claim that the graph's content can be explained as a consequence of the EU forming? – Nat Mar 21 '17 at 09:31
  • 2
    @Nat the question is about whether the graph is correct; given that, whether the apparent effect is due to the choice of axis, "100" point or an actual historical occurrence. The question is not to whether any eventual effect is due to the euro, which is not even claimed by Le Pen. – Sklivvz Mar 21 '17 at 09:34
  • 4
    But the phrase "of course has a link with the euro." suggests so. As for the rise of Germany after 2005, consider https://en.wikipedia.org/wiki/Agenda_2010 . – Scrontch Mar 21 '17 at 11:03
  • 14
    A "link" means it's correlated, but https://en.wikipedia.org/wiki/Correlation_does_not_imply_causation – Sklivvz Mar 21 '17 at 11:08
  • I'll note that I've been studying various economics-related graphs for the US recently, and, in particular, graphs of "productivity" and "industrial production" tend to vary widely and disagree with each other. (And of course the major dip corresponds to the major economic slowdown that occurred worldwide in that time frame.) – Daniel R Hicks Mar 21 '17 at 12:27
  • Josh Lyman: Uh, uh, "post" - after, after hoc, "ergo" - therefore, "After hoc, therefore" something else hoc. – Lightness Races in Orbit Mar 21 '17 at 13:24
  • @Sklivvz Can you explain what is the "euro effect" supposed to be? That the euro is hurting EU economies, or is it that it's helping only Germany but hurting other countries or what? – ventsyv Mar 21 '17 at 13:44
  • @ventsyv she says it: "the fall in industrial production" – Sklivvz Mar 21 '17 at 13:47
  • 3
    The graph shows four euro economies. If she wants any hope of blaming stuff on the euro she should show non-euro economies outperform euro economies. – gerrit Mar 21 '17 at 14:30
  • 8
    @Sklivvz: What I'm saying is that I'm pretty confident that she (Le Pen) *implies* causation. Otherwise the tweet wouldn't make much sense. – Scrontch Mar 21 '17 at 15:04
  • 7
    As @Scrontch is saying, the tweet includes "_of course has a link_". If it were merely noting a correlation without implying causation, then the fact that there is a correlation wouldn't be obvious, so "of course" doesn't make sense. The inclusion of "of course" fairly explicitly asserts causation. – Nat Mar 21 '17 at 16:18
  • 4
    For a concrete example, saying "of course there's a link between the stock market and Mercury's retrograde" would suggest a belief in superstition, because while there might happen to be a correlation, that correlation's existence couldn't be obvious to anyone who doesn't believe that they're causally associated. In this case, if someone doesn't believe that the EU's formation caused a drop in manufacturing, it doesn't make sense that a link's existence should be obvious (as noted by "of course"). __*tl;dr*__- That tweet claims causation. – Nat Mar 21 '17 at 16:32
  • 2
    If you replace the 'Creation of the Euro' marker with some other event, e.g., 'President G.W. Bush is Inaugurated in U.S.A.', the graphs would be exactly the same. Is there meaning in that? – user2338816 Mar 22 '17 at 02:10
  • 5
    FYI it was debunked by Le Monde's fact-checkers: http://www.lemonde.fr/les-decodeurs/article/2017/03/21/les-manipulations-graphiques-de-marine-le-pen-sur-l-euro_5098439_4355770.html (in French) – plannapus Mar 22 '17 at 07:46
  • @Nat Instead of "of course" I'd translate what she says as "evidently" or "obviously" (*La chute ... **a évidemment** un lien*), which means very nearly the same. I agree with you: I think that (causal connection) is implied, and I think that's what she wants people to believe. FWIW I'm sure it's widely believed that the Euro is a cause of Greece's problems (because it can't devalue its currency); whether that's applicable to France, too, idk. See also [Marine Le Pen, no longer enemy of the euro](http://www.politico.eu/article/in-final-campaign-stretch-marine-le-pen-forgets-to-blast-euro/). – ChrisW Mar 22 '17 at 12:31
  • 1
    Apart from everything else, this is also one of the great examples of how to lie with graphs. Please note how the y-axis was carefully trimmed to greatly exaggerate any possible effect. – Voo Mar 22 '17 at 14:38
  • @Sklivvz I think the benefits of having a currency you can devalue are sufficiently accepted in Economics. E.g. one of the big issues in Greece is that they cannot devalue their debts. Germany benefits from this in the opposite way - the Euro is kept from getting too strong by the PGSI countries. – JonathanReez Mar 23 '17 at 10:09

4 Answers4

41

The raw statistics can be mined from http://stats.oecd.org which I have done using the following parameters (Sorry, it doesnt appear to be linkable!)

  • Selection: Production of total Industry
  • Frequency: Annual, last 30 years
  • Countries: France, Germany, Italy & Spain

The oecd stats system seems to put the datum point at 2010, which differs from the original question, but as all are compared equally this should not make too much of a difference.

The resulting data looks like this:

enter image description here

Producing the following chart

enter image description here

Which differs from your original, but not vastly. I suspect there is some cherry-picking of data.

As for the second part, how non-Euro nations compare, there are nine member states who do not use the Euro (Bulgaria, Croatia, Czech Republic, Denmark, Hungary, Poland, Romania, Sweden, and the United Kingdom). Not all of these countries are included in the OECD data, but most are and they can be compared to the wider Euro-zone. The chart looks as below, which does not look materially different between euro-zone and non-euro-zone nations.

enter image description here

Jamiec
  • 9,004
  • 3
  • 54
  • 64
  • 5
    The graph in the question uses as _y_ the _percentage_ of production _relative to that in year 2001_ (that's the meaning of "indice 100 en 2001" on the third line of the screenshot). This explains some of the discrepancy. – fgrieu Mar 21 '17 at 11:46
  • 1
    @fgrieu yes, the current oecd site only allows you to use 100 relative to 2010`. So perhaps the OP chart was from outdated data. – Jamiec Mar 21 '17 at 11:49
  • 6
    It looks like they equalized the euro nations in 2000 so germany comes out ahead in the article's graph because it was substantially lower in absolute numbers (~90 range vs ~115 range) at the time. – ratchet freak Mar 21 '17 at 14:16
  • You can probably get similar linkable data from e.g. https://fred.stlouisfed.org/series/FRAPROINDMISMEI – Adrian Mar 21 '17 at 16:09
  • 2
    The first chart is kind of misleading. It looks like Germany performed much worse before 2005-10 than the others and better after. I know that's not what it says, but it is what it looks like. And while it shouldn't, that is what matters. Index graphs only work well when equalized the left end of the x-axis. I know the original graph set a precedent, but I think this needs to be addressed in this answer, otherwise it just looks like it verifies not only the original graph, but also the probably incorrect deduction of the "euro effect". – imsodin Mar 22 '17 at 11:48
24

Managing Expectations: U.K. Industrial Production gives many more countries and 3 different equalization points:

1991=100:

enter image description here

2006=100:

enter image description here

2009=100:

enter image description here

and another source added the total EMU:

enter image description here

Within the Euro zone, Germany has increased production more than the other countries, even though Germany was the worst performer up to 2001.

For data back to 1955 (1919 for the USA) for numerous countries see the following link: https://data.oecd.org/industry/industrial-production.htm Make sure to check "total" under the "perspectives" pull-down, uncheck "compare variables" and use the slider to expand the time window as desired.

DavePhD
  • 103,432
  • 24
  • 436
  • 464
  • 7
    These are nice graphs, but I think your answer needs to have text explaining how these graphs prove, or disprove, the claim. – Jack Aidley Mar 21 '17 at 12:34
  • 1
    @JackAidley Germany was the worst performer up to 2001, then it rose to the top of the Euro zone by 2011 and stayed there. I think I answered two of the OP's questions 1. "The graph shows an 'euro effect' according to the OP. Is this an artifact of the way the graph is made? " and 2. "What happened to countries outside of the euro?" – DavePhD Mar 21 '17 at 12:40
  • Crucially, though, none of them show other Euro countries. – Sklivvz Mar 21 '17 at 13:33
  • @Sklivvz I thought you wanted "countries outside of the euro". – DavePhD Mar 21 '17 at 13:34
  • @DavePhD yes, but then you state "Within the Euro zone, Germany has increased production more than the other countries, even though Germany was the worst performer up to 2001." which requires some evidence, for example a EU average and more EU countries to compare. – Sklivvz Mar 21 '17 at 13:39
  • 1
    @Sklivvz Ok, I added another graph with the total EMU – DavePhD Mar 21 '17 at 13:56
  • @Sklivvz and I added a link to a very complete data set that includes almost every country. Ireland increased production much more than any other country if you start anywhere in the 1975-1995 time period. Korea was the only one close. – DavePhD Mar 21 '17 at 14:46
  • 1
    "even though Germany was the worst performer up to 2001" ? I would suggest "because Germany was the worst performer up to 2001" would be closer. It's easier to improve if you start from a low base. – Martin Bonner supports Monica Mar 22 '17 at 10:54
  • If you think about it just for a couple of *seconds*, the ridiculousness of "Germany was the worst performer" should become apparent. This is the largest economy of Europe we're talking about. "Worst performer" *compared to what*? – DevSolar Mar 22 '17 at 15:22
  • @DevSolar originally, when I wrote that, I meant in the 1991-2001 time interval compare to France, Italy and Spain. Now (after Sklivvz's comment) I mean in the 1991-2001 time interval compare to France, Italy, Spain and the combined EMU. – DavePhD Mar 22 '17 at 15:29
  • Yes, but you have to be *very* careful not to mix up industrial *production*, and industrial production *increase*. (An issue that is unfortunately widespread in economics...) – DevSolar Mar 22 '17 at 15:32
  • @DevSolar I don't think I'm confusing them. The graphs are production (not production growth) normalized such that all the countries production is "100" at the reference date. On a given interval, the ratio of production on the final date to that on the initial date quantifies production growth over the given interval, and in that sense I am saying German was worse 1991-2001 compare to the other 3 and the total EMU. – DavePhD Mar 22 '17 at 15:47
  • But the claim and the OP's question is about production, not production growth. ;-) – DevSolar Mar 22 '17 at 15:48
  • @DevSolar "fall" is negative growth. – DavePhD Mar 22 '17 at 15:53
  • Err.... and while writing what I wanted to say at this point, I realize that it doesn't sound half as convincing written down than it did in my head. :-D Never mind me. – DevSolar Mar 22 '17 at 15:57
  • Those are some *very* confusing graphs. They show how each country is doing *relative to itself in the past* but then puts them all on the same axis? I mean, take the US. Production change is less than Germany, but there's more industrial production in the US, so a smaller relative change represents a larger gross change. US production being 10% greater might be significantly more gross production than German production being 25% greater. So Germany might have more production relative to itself, but that doesn't mean that gross production has increased in Germany more than in the US. – Bacon Bits Mar 22 '17 at 16:30
2

Remark: my previous answer having been deleted, I rewrote it completely in order to address the issues. I hope that's fine.

The OP asks three questions:

  1. Is the data accurate?
  2. The graph shows an "euro effect" according to the OP. Is this an artifact of the way the graph is made?
  3. What happened to countries outside of the euro?

The first two questions have been addressed in the French media (sorry for posting links in French, but obviously the topic is covered mostly by french news). Here are the analyses I'm aware of, together with a short summary:

  • lemonde.fr accuses Le Pen of manipulating data by cherry-picking the year 2001 as reference year. They interpret the difference in industrial production as a consequence that Germany recovered faster after the 2008 crisis.
  • L'Obs says that the graph is accurate but misleading. The article acknowledges that Euro plays a significant role in the observed difference, because countries cannot devaluate their currency anymore. But it also emphasizes that other factors explain the difference.
  • arretsurimage.net (paywall) agrees with the two previous studies for the most part, insisting on the questionable causation link between the graph and the claim.
  • les-crises.fr criticizes (vehemently) the above-mentioned analysis given in lemonde.fr, explaining in particular why the choice of the reference year is appropriate. The article also points out some mistakes in this analysis.

Question 1: is the data accurate?

All the sources agree that the data is accurate.

Question 2: The graph shows an "euro effect" according to the OP. Is this an artifact of the way the graph is made?

The sources disagree about this point. Most emphasize the fact that it's easy to misinterpret this kind of graph, at least. In detail:

  • For lemonde.fr, it is an artifact caused by the choice of the reference year.
  • The article in L'Obs is much more nuanced, and explains why Euro has indeed a beneficial impact on Germany's industrial production and a negative one on the three other countries. However the article also details other factors which are likely to have an impact as well (my understanding is that the direct causation link is rejected).
  • arretsurimage.net (paywall) also questions the causation link between Euro and the fall in industrial production.
  • les-crises.fr considers that Euro has clearly an impact, but mentions it would be a simplification to say that Euro is the only cause.

Underlying question: does Euro benefits to Germany at the expense of other countries ?

It's not very clear in the OP's question, but given that the question is tagged "economics" and "Europe", I will assume that the OP is also interested in evidence showing whether the claim is true or not (since the fact that the presented graph is inconclusive does not imply that the claim itself is wrong).

L'Obs explains that the main impact of Euro is on the price of imports and exports,so one can look at this indicator instead of industrial production. The graph below shows the current balance account expressed as percentage of GDP (no need for a reference year):

current balance account expressed as percentage of GDP, OECD

This graph is from the OECD website again (you need to select the parameters). It does not show any clear sign of decline caused by Euro for France, Spain or Italy (in particular France and UK perform similarly, although the latter is not part of Euro). However Germany starts to improve dramatically around the time where Euro is introduced, and keeps on improving later on.

Erwan
  • 470
  • 3
  • 8
  • See also [How to proceed when implicit and explicit claims diverge](https://skeptics.meta.stackexchange.com/q/3916/2703) -- I think this OP is only interested in whether the graph is accurate, and doesn't want to read about any "implied" claim (e.g. that there's a causal link, nor that it's bad for France, etc.). – ChrisW Mar 24 '17 at 10:28
  • 1
    @ChrisW thanks for the link, it kind of helped me understand my own frustration with the question ;) To me the question of the existence of an artifact or not calls for studying alternative evidence: if it corroborates the conclusion then the artifact is unlikely. – Erwan Mar 25 '17 at 02:21
  • You're welcome, and welcome to this site. This site has defined [a lot of rules](https://skeptics.meta.stackexchange.com/questions/tagged/faq); many questions are closed (by users) and answers deleted (by moderators), even popular answers. [This chat room](http://chat.stackexchange.com/rooms/55959/discussion-between-sklivvz-and-chrisw) helps to explain the moderator's dissatisfaction with answers to "implicit claims". – ChrisW Mar 25 '17 at 02:33
1

The data seems accurate. I re-calculated the OECD data to 2001 = 100 but didn't truncate the y-axis (truncating the y-axis overemphasizes any relative increase or decrease):

De-Fr-Es-It

I also created a second sheet by removing Spain and Italy and adding Switzerland (non-Euro/non-EU) with a similar development as Germany and the UK (non-Euro/EU) with a similar development as France:

De-Fr-CH-UK

If you want to create your own charts use two Excel sheets: one with the OECD industry production data of all OECD and some others countries downloaded from OECD.Stats and the "chart maker".

If you want to replace a country in the chart maker, change the country name and copy the respective column of the country in the OECD industry production data sheet to one of the columns on the left (e.g. replacing Italy). The chart as well as the re-indexed data on the right will be changed automatically.

If you want to choose another base year, choose one year on the right and set it to 100. You have to modify the functions of all other years manually though (takes two minutes).

enter image description here

klanomath
  • 319
  • 1
  • 4
  • Upvoted for not truncating the zero. That always makes shifts seem more dramatic, and is a known way to bias change into something that seems more dramatic. – Edwin Buck Mar 23 '17 at 17:01