Worth 1000 words.


This is the graph that should have been posted with the “Payback” post last week.  I’d like to thank Rob Fightmaster, who helped me get this and several other graphs into a form that could be posted.  (If you can’t see it well, there is a bigger version at the very bottom.)

Update:  I forget to mention that Steve Silver helped me to produce this graph in the early 1990s.

The bottom time series shows the Great Depression.  More specifically, it shows the log of industrial production, monthly, detrended, from the beginning of 1929 to the end of 1939.  And the Depression still wasn’t over.  The two low points are July 1932 and March 1933 (the month FDR took office.)  Then you see the 4 month spike in industrial production that I often discuss in my blog.  That peaks in July 1933.

On the top is the real wage series; the nominal manufacturing wage deflated by WPI.  Also logs, with the variation multiplied three times.  Most importantly, the real wage series is INVERTED, to make it easier to see the negative correlation between W/P and IP at high frequencies, which is very much obscured using annual data.  Just looking at 1933 makes it easy to see why.  Real wages fell very sharply during that period of explosive growth between March and July 1933, and hit bottom in July.  Then real wages soared and output plunged.

I don’t want to oversell this graph; I think other factors like Smoot-Hawley and the huge rise in MTRs during 1932 also help explain the extreme fall in output during 1932.  Since I don’t plan to post my chapters covering 1933-40, perhaps I should say a word about the events that shaped the very uneven course of the recovery:

1.  As you can see IP fell sharply after the July 1933 peak, in response to FDR’s surprise decision to raise all wages by 20%.  Not surprisingly, there was a huge stock market crash as the story leaked out over three days (indeed the 3rd biggest three day crash ever.)  By late October FDR was getting nervous.  Time for another shot in the arm from more dollar depreciation.  He brought in George Warren, who I discuss in this post.  As far as I know every single economic historian, liberal and conservative, think Warren’s policy failed.  Actually it succeeded.  The dollar fell sharply between October 1933 and February 1934.  Industrial production began rising and peaked in the spring of 1934.

2.  Then FDR decided to raise wages again (I know, don’t they ever learn?)  And again he killed off a promising recovery.

3.  In May 1935 IP was still lower than July 1933.  The Supreme Court then declared the NIRA unconstitutional.  (The NIRA was the law that allowed FDR to raise wages.)  The economy took off on a two year boom.  It was helped by fast rising prices between mid-1936 and mid-1937.  The prices rose because once the international gold standard fell apart in 1936, hoarders who had feared devaluation now dis-hoarded on the reality of devaluation.  Output should have risen rapidly right up through the summer of 1937.

4.  Actually, output hit a wall in early 1937.  The Wagner Act was a response to the Supreme Court’s NIRA decision.  It made it much easier to form unions.  The union movement was emboldened by FDR’s big win in late 1936, which unions had helped to achieve, and union drives rapidly increased membership in late 1936 and 1937.  The resulting wage shock slowed the economy sharply in 1937, despite high prices.

5.  As the economy slowed, people began to no longer fear inflation, but rather deflation.  Gold hoarding increased strongly in late 1937 in response to renewed fears of dollar devaluation, causing prices to fall.  This created an identical situation to 1920-21 and 1929-30, as prices fell rapidly.  Since wages are slow to adjust (particularly with the stronger unions) real wages rose and IP fell sharply.

6.  Eventually wages did adjust, and in the spring of 1938 output began rising.  Then in late November 1938, FDR instituted the first minimum wage law (25 cents an hour) and the recovery immediately stalled.  In the spring of 1939, the economy again began recovering.  But in November 1939 he raised the minimum wage to 30 cents an hour, and the recovery stalled for the 5th and last time.

7.  Unfortunately my graph gives out at the end of 1939.  In early 1940 the economy was weak, but it took off like a rocket after the German invasion of France, which is when the “phony war” turned into a real war.  I don’t study this time period, but there were two factors contributing to recovery.  Fiscal stimulus; and (perhaps) higher inflation expectations, both of which would reduce currency hoarding.

This has always been my favorite graph (or perhaps tied with that famous 6 dimension graph showing Napoleon in Russia.)  I believe it provides the key to the Great Depression.  If one can explain why real wages moved up and down, then you’ve explained the Great Depression.  Not all of the Depression, as I indicated earlier other real shocks were important.  But it explains the Great Contraction, and also why the recovery was so long and uneven.

It may be interesting to compared this real wage model (if “model” is the right term) to Irving Fisher’s famous Phillips Curve model, published in 1923 and 1925.  The first graph (at bottom of post) shows the 1921 depression, the one Austrians like to talk about.  As you can see it was quite deep, and is almost perfectly explained by Fisher’s Phillips curve model (which explains output fluctuations with a distributed lag of inflation rates.)  Indeed I imagine Fisher’s simple model would explain output during that period far better than modern new Keynesian Phillips Curve models could.

Ah, you ask, but how does it do “out of sample?”  Anyone can fit a model to past data.  I wondered the same thing, so I ran his model forward from 1923-35.  As you can see, it didn’t explain things nearly as well.  The correlation fell from .941 to .256.   But take a closer look at the second graph.  From 1923 to July 1933 the correlation is still a respectable .851.  And it even predicts a “Great Depression” after 1929.  Of course ‘predicts’ is misleading here, it is not an unconditional prediction.  After all, Fisher is famous for having argued against the view that stocks were likely to crash.  Instead, it makes a conditional prediction, based on price deflation that Fisher also didn’t anticipate.

One other thing; Fisher’s model says the economy should have recovered rapidly after March 1933.  It did for 4 months, as dollar depreciation pushed prices much higher.  What went wrong with his model after that?  Phillips Curve models don’t take supply shocks into account.  So it was unable to explain why the recovery was aborted in July 1933.  That’s why I like my real wage “model” of the Depression better than the Phillips curve model.  It explains much more.  But Fisher made a good start, and the Phillips Curve obviously should be called the Fisher Curve.

The interwar economist whose views are closest to mine is Roger Ralph Hawtrey.   His book The Gold Standard in Theory and Practice does a good job of presenting a “gold and wages” view of the Great Depression.  David Glasner is also a fan.  If you read it, make sure you do not get the first edition, from the early 1930s.  Instead get a post-war edition from 1946 or 1947, so that you get his narrative on the entire Depression.




46 Responses to “Worth 1000 words.”

  1. Gravatar of Lorenzo from Oz Lorenzo from Oz
    14. February 2010 at 11:52

    I am not actually getting any graphs using Safari or Firefox.

  2. Gravatar of Mike Griswold Mike Griswold
    14. February 2010 at 12:01

    Professor Sumner-

    The graphs you posted do not show up at all. I’ve tried both Internet Explorer and Firefox.


    Mike Griswold

  3. Gravatar of Manny C Manny C
    14. February 2010 at 12:02

    Graphs not appearing.

  4. Gravatar of david david
    14. February 2010 at 12:24

    I want a last name like “Fightmaster”.

    More seriously,
    the image is pointing to “file:///C:/Users/micro/Pictures/figure_2-1_with_title.gif”, which won’t show up, of course. Upload your graph somewhere and change the path to the path of the image.

  5. Gravatar of ssumner ssumner
    14. February 2010 at 12:53

    OK, The first thing you guys need to realize is that you need to address me like a 1st grader. David, with me and computers there is no “of course.” I’ll work on it. First I must find the meaning of terms like “upload” and “path” and “somewhere.”

  6. Gravatar of ssumner ssumner
    14. February 2010 at 12:57

    OK, I think it is now working (thanks to my wife.)

  7. Gravatar of Real Depressing Wages « It Don’t Mean Much, These Seats are Cheap. Real Depressing Wages « It Don't Mean Much, These Seats are Cheap.
    14. February 2010 at 15:03

    Did FDR’s high wage policies (and, by extension, Hoover’s) harm recovery from the Great Depression? Scott Sumner says yes: […]

  8. Gravatar of Niklas Blanchard Niklas Blanchard
    14. February 2010 at 15:23

    Wow, what a mess.

    Sorry about that, Scott. You can delete the trackback comment if you want.

  9. Gravatar of Scott Sumner Scott Sumner
    14. February 2010 at 15:32

    Niklas, This post must be jinxed by Krugman. I’m having all kinds of technical issues today. I cleaned up the garbled message in your previous comment, but left it in place so people could easily link to your blog.

  10. Gravatar of Rafael Rafael
    14. February 2010 at 15:59

    Hey Scott,

    Did you see the new IMF/Blanchard paper everybody´s talking about?

    It deserves a post, doesn´t it?


    PS: Great charts.

  11. Gravatar of Leigh Caldwell Leigh Caldwell
    14. February 2010 at 17:31

    Hi Scott

    I intended to ask this when you previously mentioned this correlation, but now that you’ve posted the actual graphs seems a good time to do it.

    Do you know how the industrial production data was collected? I have a nasty suspicion that there might not have been good data on real volume production, so the statisticians may have used NOMINAL production and divided it by a price index to estimate real production.

    Of course the wage graph is also based on nominal wages divided by a price (wage) index.

    If nominal wages and production were both steady (in the short run) and the same price index was used for both graphs, this would naturally produce the striking correlation which you observe. The graphs would not be showing wages against production, but prices against prices – no surprise to see a correlation there.

    I could be wrong about this – but as you suggested yourself, the correlation is a lot closer than we would normally observe in any economic dataset. This really makes me wonder.

  12. Gravatar of Robert Simmons Robert Simmons
    14. February 2010 at 18:12

    Could you give a definition for detrending a series? Reading the wikipedia definition made my head hurt

  13. Gravatar of coray coray
    14. February 2010 at 20:04

    Have you looked at Ohanian Cole?

  14. Gravatar of Lorenzo from Oz Lorenzo from Oz
    14. February 2010 at 21:54

    I was thinking, “wow, something is really going on here”, then I read Leigh Caldwell’s comment and the former Parliamentary Research Service statistics adviser I used to be suddenly got very nervous …

    Minor editing point:
    the first addition, from the early 1930s.
    A wonderful Freudian slip, but perhaps not quite what you meant.

  15. Gravatar of Doc Merlin Doc Merlin
    15. February 2010 at 05:36

    Thanks Scott, this is exactly what I needed. It confirms some suspicions I had, and has helped me with a model I am working on.

  16. Gravatar of Macgill Macgill
    15. February 2010 at 07:34


    Hawtrey’s correct first name is Ralph, not Roger.

  17. Gravatar of Scott Sumner Scott Sumner
    15. February 2010 at 08:42

    Rafael, I’ve argued for a long time that if we are to have this sort of dysfunctional monetary policy we need a higher inflation target. But it would be much better to switch to NGDP targeting, level targeting. In that case we wouldn’t need a higher inflation target. So I half agree with Blanchard. There is also a new paper by Ricardo Reis floating around that talks about the need for level targeting, and a catch-up price rise after deflation.

    Leigh, Wouldn’t you get the opposite? Remember I am arguing that there is a strong inverse correlation between output and real wages. Suppose as you assume that nominal wages and nominal output were steady. And suppose measured P fell 10%. Both real wages and real output would appear to rise 10%, but I argue they move in opposite directions.

    I also believe the data from the interwar period are much more accurate than modern data, for the part of the economy they measure. The economy was far more oriented toward easy to measure commodities such as wheat and steel and coal. And far less oriented toward vague output like consulting services.

    The IP data is also strongly correlated with other more micro level data that are clearly real, such as freight carloadings (railroads were very important back then.) So I have a high level of confidence about the basic pattern of fluctuations that you see in the IP data. Thanks how I immediately knew the Romer and Miron IP data was wrong. However, if we had RGDP data, the fluctuations would be much smaller, as IP is unusually cyclical. Another way of making my point is that if any good business economist from the 1930s were shown the graph, he would recognize the IP cycles, and not just from IP data, but from all sorts of other data.

    Robert, You asked:

    Could you give a definition for detrending a series? Reading the wikipedia definition made my head hurt

    It’s been nearly 20 years, and I am at home without the info, but I seem to recall we regressed it on a time trend and plotted a deviation for the trend line. Which reminds me, I need to add Steve Silver’s name to the post, as he helped me do this research. The trend adjustment doesn’t make the IP series look very different, as actual IP didn’t rise very much between 1929 and 1939. I don’t recall real wages, but I think they rose more strongly, so the actual real wages would show a stronger upward trend. That’s one reason why Keynesians wrongly assumed higher real wages weren’t a problem—the level of real wages was higher in the late 1930s, and the economy was gradually recovering from the lows of 1932-33. The monthly data tell a very different story.

    Coray, The question is whether Cole and Ohanian have looked at our work, as Steve Silver and I published our study before they did. In addition, our monthly data is superior to their data. When I sent it to them last year, they seemed pretty impressed. Obviously our conclusions are similar. I don’t agree with their view of AD shocks, however. They are too RBC-oriented for my taste.

    Lorenzo, I am worried that everyone seems to accept Leigh’s point. Perhaps I didn’t emphasize strongly enough that the W/P series is inverted.

    Thanks, I’ll correct the typo.

    Thanks Doc.

    MacGill, Thanks, I am doing this stuff too fast. That’s the second name I’ve screwed up recently. I’ll change it.

  18. Gravatar of marcus nunes marcus nunes
    15. February 2010 at 09:18

    Link to Ohanian´s take on the first wage shocks:

  19. Gravatar of ssumner ssumner
    15. February 2010 at 10:36

    Thanks Marcus, Three observations:

    1. If NGDP had not fallen in half, Hoover’s wage policy would not have done any harm. It wasn’t so much a high wage policy, rather he discouraged wage cuts. If NGDP had continued growing in the early 1930s, wage cuts would not have been necessary and wages would have stayed close to their equilibrium value.

    2. If NGDP falls in half, you will have a severe depression even if wages were as flexible as in 1921.

    3. I agree that the Hoover wage policy meant that the 50% fall in NGDP resulted in somewhat lower real output, and somewhat higher prices, than if wages ahd been more flexible. So the policy did matter, and it was harmful. But it doesn’t explain the Great Contraction. A contractionary m-policy caused the Great Contraction.

    I also wanted to go back and add one additional comment to my reply to Leigh. I am pretty sure that the Fed did not deflate nominal IP by the WPI to get the real IP series. They were much more sophisticated than that, even back in the 1930s. The WPI was strongly influenced by agricultural prices back then, especially during crop failures. They certainly knew of that issue. My readings from material writtem back in that era convinced me that business economists had a pretty good sense about how to handle and interpret data. Surprisngly good.

  20. Gravatar of Richard A. Richard A.
    15. February 2010 at 12:52

    Wr = real wages
    Wn = nominal wage
    IPr = real industrial production
    IPn = nominal IP
    P = price level

    Wr = Wn/P
    IPr = IPn/P

    Fig 2.1 plots 1/Wr = P/Wn
    This can be rewritten in log form as
    log(P) – log(Wn)

    While IPr = IPn/P in log form is
    log(IPn) – log(P)

    Notice a change in log(P) causes these two series to change in OPPOSITE directions.

  21. Gravatar of ssumner ssumner
    15. February 2010 at 14:31

    Richard A. Yes, that’s what I thought as well.

  22. Gravatar of Leigh Caldwell Leigh Caldwell
    15. February 2010 at 14:39

    Hi Scott

    Thanks for correcting me – of course you are right. I knew the relationship was inverted when you first mentioned it, but missed this point when you posted the charts.

    I’m glad to be shown wrong, as the correlation you have found is a powerful one and certainly makes logical sense – I guess if anything I am just surprised that the effect operates so quickly and so precisely.

  23. Gravatar of rob rob
    15. February 2010 at 15:22

    Scott, but doesnt the cause and effect run in the opposite direction for most of this chart? Isn’t it the decrease in IP which increases real wages and vice versa? Of course, when you point to a moment when the min wage is raised and we see an immediate move I agree that IP must be reacting to wages. But isnt the reason the correlation is so close on a small scale due to real wages in the short term being more or less a function of IP?

  24. Gravatar of rob rob
    15. February 2010 at 16:01

    I should have said “the change” in real wages in the short term is a function of IP — which is merely to observe that nominal wages are sticky. Isn’t it?

  25. Gravatar of cucaracha cucaracha
    15. February 2010 at 20:01

    “…While IPr = IPn/P in log form is
    log(IPn) – log(P)

    Notice a change in log(P) causes these two series to change in OPPOSITE directions.”

    You should consider that price index changes are in part derived from changes on industrial prices themselves, ie, changes in log(P) contribute to changes in log(IPn) in the same direction.

    The same probably doesn’t happen with wages as fast because of a certain lag between changes in log(P) and changes in nominal wages.

    Anyway, fixing wages is not the better way to redistribute income and increase consumption as hiring ends up discouraged because of the higher fixed costs.

    It is better to have a progressive federal income/wealth taxation scheme and more public spending to generate more jobs when it comes to recessions.

  26. Gravatar of cucaracha cucaracha
    15. February 2010 at 20:17


    the best way…

  27. Gravatar of Scott Sumner Scott Sumner
    16. February 2010 at 07:15

    Leigh. The speed of the correlation may be slightly misleading. There is a natural tendency to read the causation as running from real wages to output, and to some extent it does. But real wages change for two reasons, nominal wages shocks and price level shocks. It is possible that monetary policy impacts NGDP, which causes P and Y for change at the same time. As P changes, so do real wages, assuming nominal wages are sticky. So some of the correlation may be due to ordinary Phillips curve effects, with real wages merely being a symptom. That’s not necessarily me view, but it is how a skeptic might read the evidence. But even so, all I would really need to do is explain why P changed, and I can do that with a gold market model of the price level.

    rob, See my previous answer. I don’t think IP causes P to change in the same direction. Indeed for any given NGDP, a fall in output should raise prices. The key causal factors is the change in NGDP, which makes P and Y move in the same direction. That’s just standard macro theory, I am not claiming anything out of the ordinary. The only wrinkle is that I assume wages are stickier than prices. Modern Keynesians don’t make that assumption, but it was clearly true for interwar data.

    cucaracha. See above for my response to your first.

    Public spending is a horrible way to address recessions. Monetary policy is far more powerful, and far cheaper as well.

  28. Gravatar of Lord Lord
    16. February 2010 at 10:47

    I interpret this as, given sticky nominal incomes, prices affect real incomes directly which then affect IP with a year to a quarter lag. Once real incomes cease rising a turnaround comes but up to a year later. Prices had already started turning up when the 33 devaluation caused them to really jump. So were the wage hikes to reduce the stickiness of wages to respond, to prevent them from over responding, trying to reestablish the previous price level as nearly as possible, or to moderate the rate of recovery?

  29. Gravatar of StatsGuy StatsGuy
    16. February 2010 at 11:45

    “Public spending is a horrible way to address recessions. Monetary policy is far more powerful, and far cheaper as well.”

    If you don’t mind the distributional consequences and carry trade bubbles… (Or, the Fed could simply finance a portion of public spending directly, which would work wonders except for the bad incentives this gives to congress.)

    BTW, I’m growing increasingly skeptical of the excess cash balance mechanism you posit as the key transmission mechanism. US banks apparently are now sitting on 1.3 trillion in reserves, even as credit to businesses declined by 14%.


    I am certain you will point out the flawed causality here and how the credibility from an NGDP targeting regime would fix it. Sure, but I’m talking specifically about the cash balance mechanism, which does not claim to require NGDP targeting. How big does the cash reserve need to be before velocity recovers? And when that happens, do we really think it will be a gradual equillibrium shift? Hmm.

  30. Gravatar of StatsGuy StatsGuy
    16. February 2010 at 11:53

    Well, I’ve expressed my skepticism of this data before – namely, the gaming of expected changes (both starting and stopping) to wage policy using inventory management (and inventories were relatively larger in early Taylorist economies than today).

    Also, I think a large part of the correlation could be related to hiring practices and wage structures. When industrial production picks up, firms hire back cheaper (non management, unskilled) labor. This means that the _average_ wage goes down. If this were driving the effect, then we might observe this in the lead/lag structure. I can’t tell based on the graph what that structure might be, but maybe you can post the data for inspection?

    Again, I don’t doubt the effect, but the chart likely overstates it a lot.

    Also, there are some clear anomolies here. For example –

    April 1931-July 1932. The lines clearly diverge. Why? Could this be public spending sustaining wages even as deflation intensifies? But that wouldn’t make sense, since public sector jobs were probably higher than average wage and so this would increase average wages but the inverted line is flat to slightly rising while industrial production is falling rapidly.

  31. Gravatar of StatsGuy StatsGuy
    16. February 2010 at 12:01

    Leigh Caldwell:

    “I could be wrong about this – but as you suggested yourself, the correlation is a lot closer than we would normally observe in any economic dataset. This really makes me wonder.”

    Even leaving gaming of inventory level aside, one would observe this if there was a mechanical relationship due to wage structure and hiring practices. If low-skilled low-pay employees are the first to be let go when orders decline, and the first to be hired back when orders increase (as is the case in a Taylorist manufacturing firm), the relationship we observe above would occur automagically.

    Not that I have any better data to play around with, but the graph above is weak proof for the posited causal mechanism. Sorry.

  32. Gravatar of ssumner ssumner
    17. February 2010 at 07:46

    Lord, I don’t see the lag you see. Prices and output began rising at the same time, around late March 1933. I see changes in NGDP as causes both changes in prices and output, and roughly simultaneously.

    Statsguy, Stabilizing NGDP through monetary policy doesn’t have important distributional implications, nor does it create carry trade bubbles. Carry trades are most associated with Japan, which had a dysfunctional monetary policy.
    And even if it did cause those problems, they are far less severe than the problems associated with a rising national debt.

    You said;

    “Sure, but I’m talking specifically about the cash balance mechanism, which does not claim to require NGDP targeting.”

    I have always acknowledged that if the central bank suggests base injections are temporary (as they have done here) they won’t be inflationary. In any case the payment of interest on reserves makes it a moot point. Even if they are permanent, they won’t be inflationary if the Fed pays a high enough interest rate. I don’t see the connection this has with the excess cash balance mechanism, which assumes zero interest currency.

    You said;

    “Well, I’ve expressed my skepticism of this data before – namely, the gaming of expected changes (both starting and stopping) to wage policy using inventory management (and inventories were relatively larger in early Taylorist economies than today).”

    Yes, and I pointed out that your explanation can’t explain why nominal wages were uncorrelated with IP during the 1920s, but suddenly started being correlated after 1933. Nor does the composition bias explanation help.

    I don’t know why output was so low in 1932, but my hunch is that bad supply-side policies played a role. But also recall that this isn’t really a complete model, and I certainly don’t claim that real wages and IP “should be” inversely correlated. Usually there is no correlation. This is a completely ad hoc model, if you can even call it a model. BTW, you misinterpreted 1931-32, real wages were actually falling. The real wage graph is inverted.

    Statsguy#3, Your comment would only be true if we generally observed this pattern. Then you could say “of course they are correlated, it is all due to X, Y and Z.” But we don’t generally observe this correlation, it is specific to the 1930s. The graph is not meant to prove some sort of causal interpretation, if that had been my intention, then I would agree with your criticism. By itself, the graph proves nothing about causality. My book (if I ever finish) is supposed to provide evidence of causality.

    I lost all my data, so I can’t post it. The IP data is from the Fed, and shouldn’t be hard to find. The WPI data comes from various issues of the Federal Reserve Bulletin, I don’t know if it is online. The manufacturing wage data is from Employment Hours and Earnings, United States, 1909-84, from the BLS.

  33. Gravatar of Lord Lord
    17. February 2010 at 09:38

    I am looking at P in Figure 8.2 reversing in mid 31 which leads to real wages peaking then in Figure 2.1, while IP doesn’t hit bottom until mid 32. That also leads me to identify earlier peaks(falls) in real wages to later peaks in IP. Even the 37 real wage downturn seems a quarter ahead of the IP downturn. Now the effect may well have been different between the two different kinds of shocks, and I was somewhat surprised real wages stabilized before IP, but perhaps I shouldn’t be since consumer staples and non-durables would be favored over durables.

  34. Gravatar of scott sumner scott sumner
    18. February 2010 at 10:44

    Lord, I wouldn’t make too much of the graphs, as there is no expectation that the correlations would be perfect. I think output in 1932 might have also been depressed by supply-side factors. In my view the proper nominal indicator is NGDP, not the price level. I only use P because it is more available for that period. Annual NGDP fell very sharply between 1931 and 1932. Where shocks are easily identifiable, as in 1933, there don’t appear to be any lags.

  35. Gravatar of If you want to know why the great depres… « The Sagamore Journal If you want to know why the great depres… « The Sagamore Journal
    22. February 2010 at 12:09

    […] the great depres… If you want to know why the great depression was so great check out this post by Scott Sumner (he’s kind of like Paul Krugman but from Chicago… and without a Nobel […]

  36. Gravatar of TheMoneyIllusion » The Great Depression: The Movie TheMoneyIllusion » The Great Depression: The Movie
    11. April 2010 at 13:33

    […] The first graph in this earlier post shows the entire Depression in one […]

  37. Gravatar of The White House’s capitulation on deficits | Rediffer.Com The White House’s capitulation on deficits | Rediffer.Com
    16. July 2010 at 22:47

    […] http://www.themoneyillusion.com/?p=4220 […]

  38. Gravatar of A Response to “How Obama Saved Capitalism and Lost the Midterms” Part 2 « tracingtheconsequences A Response to “How Obama Saved Capitalism and Lost the Midterms” Part 2 « tracingtheconsequences
    4. November 2010 at 21:41

    […] http://www.themoneyillusion.com/?p=4220 […]

  39. Gravatar of New Deal 2.0 – Because Falling Down the Stairs Once Just Wasn’t Enough « tracingtheconsequences New Deal 2.0 – Because Falling Down the Stairs Once Just Wasn’t Enough « tracingtheconsequences
    8. December 2010 at 20:33

    […] was first alerted to this narrative of the Great Depression by Scott Sumner via this post of his, which has far more detail and is worth reading if one has the time.  Everything makes so […]

  40. Gravatar of Is economics a science? (liquidity trap edition) Is economics a science? (liquidity trap edition)
    18. March 2011 at 09:31

    […] reflect this scenario?  Were big wage hikes expansionary in the Great Depression?  Scott Sumner has crushed that view, with an eyeball scan of the numbers, along with some simple […]

  41. Gravatar of Has the Keynesian IS-LM model made good predictions lately? — Marginal Revolution Has the Keynesian IS-LM model made good predictions lately? — Marginal Revolution
    14. October 2011 at 01:34

    […] That claim, at least for the Great Depression, has been knocked down fairly conclusively by Scott Sumner.  If there is no comparable test on today’s data, it is because we have grown that much […]

  42. Gravatar of Tyler Cowen | kushima.org Tyler Cowen | kushima.org
    26. January 2013 at 05:34

    […] reflect this scenario? Were big wage hikes expansionary in the Great Depression? Scott Sumner has crushed that view, with an eyeball scan of the numbers, along with some simple narrative. 5. If I Google […]

  43. Gravatar of Daniel Moerner Daniel Moerner
    30. January 2013 at 00:38

    Is there any way you could re-post the graphs from this post? This is exactly the graph I’m looking for, and the images appear to be dead.

  44. Gravatar of Noah Smith on Keen | Letters to a Post-Apocalyptic Dictator Noah Smith on Keen | Letters to a Post-Apocalyptic Dictator
    2. June 2013 at 01:55

    […] data, which I originally misinterpreted, is what first forced me to pay attention to Keen (just as this piece of evidence is what forced me to tune into Scott Sumner). We have a strong empirical relationship. It needs […]

  45. Gravatar of British Living Standards are Rising BECAUSE Real Wages are Falling | uneconomical British Living Standards are Rising BECAUSE Real Wages are Falling | uneconomical
    12. September 2013 at 00:59

    […] [Inspired by, and see also: old Sumner posts on real wages in the Great Depression.] […]

  46. Gravatar of Financial Acceleration and the Paradox of Flexibility | Skyler Lehto Financial Acceleration and the Paradox of Flexibility | Skyler Lehto
    12. June 2015 at 12:28

    […] periods when interest rates are floor-bound (i.e. supply shocks in a liquidity trap). To that end, here is the historical data showing inverted real wages and output during the Great […]

Leave a Reply