Krugman on the Phillips curve

In a recent post I praised Steven Williamson for trashing the Phillips curve.  Now Paul Krugman has a post defending the Phillips curve, which I am also going to praise.  How can I agree with such diametrically opposed views?  Simple, Williamson trashes the American version of the Phillips curve, the one using inflation and unemployment.  As you may know I view inflation as an almost worthless concept, which does more to confuse than enlighten.  In contrast Krugman discusses the original version of the Phillips curve.  Well, not the actual original version, that was developed by Irving Fisher in 1923, but Phillip’s 1958 version, which used wage inflation instead of price inflation.  Whereas price inflation is a useless concept, wage inflation is a highly useful concept, especially in business cycle analysis.  Here’s Krugman:

Start with the raw data. Here’s unemployment and increases in nonsupervisory wages since 1985:

What you see is that wage growth was low when unemployment was high, and vice versa. Now take annual averages (to avoid overlapping data) and plot the unemployment rate against the wage change over the next year:

There are a couple of possible explanations for the return of the good old-fashioned Phillips curve: anchored inflation expectations, downward sticky nominal wages. I’ll have more thoughts on that later (actually, downward rigidity and anchored expectations I think reinforce each other). But the point is that notions of how inflation works that were formed in the era of stagflation are very much at odds with the way the world has looked, not just since the Great Recession began, but since the mid-1980s. Yet stagflation still shapes both public perceptions and policy.

Later in the post Krugman criticizes the natural rate hypothesis, which is where he loses me.  I still think that’s a very useful concept, which explains why the US unemployment rate is falling despite stable (2%) wage inflation.  But overall I’m thrilled to see that Krugman prefers the wage version of the PC, he’s far more influential than me.

PS.  Notice that in 2009 wage inflation fell from 4% to 3%, whereas NGDP growth fell by 9 percentage points below trend.  There’s your Great Recession, which had nothing to do with housing/banking/lack of fiscal stimulus etc., etc.  Tight money and sticky wages create recessions.  The Musical Chairs model.

PPS.  Today the BEA raised the estimate of GDP growth once again.  NGDP growth is running 4% in 2013 (NGDI at 3.9%). RGDP at 2.6%.  I can’t wait to do my year end review; it’s been an incredible year for MM, in all sorts of ways.

PPPS.  There is some interesting recent commentary over North Carolina’s reduced UI benefits (here and here).  I’m waiting until we have more data before I chime in.  I’m an agnostic on the issue, and don’t even know if state level results would carry over to the national level.  In 2014 we may get a similar experiment at the national level.

Update:  Kebko’s lastest update on North Carolina is fascinating.

PPPPS.  Off topic, but file this under “It’s good to be the king.”  King Obama has waved what the Supreme Court has ruled is a tax, for lots of Americans.  What a nice Christmas present!



50 Responses to “Krugman on the Phillips curve”

  1. Gravatar of Brian Donohue Brian Donohue
    20. December 2013 at 11:33

    Pardon mmy density, but I understood the Phillips curve to argue (correctly or not) that higher inflation means lower unemployment.

    What does this wage inflation Philips Curve tell us? That if we just raise wages, unemployment will come down?

    That would be a sweet universe to live in.

  2. Gravatar of ssumner ssumner
    20. December 2013 at 11:36

    Brian, No, it says nothing about causation.

  3. Gravatar of Brian Donohue Brian Donohue
    20. December 2013 at 11:46

    So…it says that when unemployment is high, wage increases are low, and vice versa? Who doesn’t know that already?

  4. Gravatar of W. Peden W. Peden
    20. December 2013 at 11:59

    The phrase “Phillips curve” is frustratingly ambigious, since as an assertion that the growth rate of wages and the level of unemployment are correlated, it’s inconsistent with the natural rate hypothesis (and reality in general) whereas Friedman famously believed that changes in the growth rates were correlated and this is perfectly consistent with the natural rate hypothesis.

  5. Gravatar of kebko kebko
    20. December 2013 at 12:06

    Here’s my update with today’s November state unemployment numbers. November unemployment in North Carolina dropped another 0.6% from the October level.

  6. Gravatar of Doug M Doug M
    20. December 2013 at 12:25

    Wage inflation vs price inflation.

    How much bad policy has been implemented because of economists, central bankers, and politicians using the wrong inflation measure?

  7. Gravatar of SG SG
    20. December 2013 at 12:41


    I don’t know if this makes any difference for you or not, but I thought I’d mention that Obama’s latest rewrite of Obamacare actually has more legal support than his earlier decisions delaying the employer mandate or the minimum coverage standards.

  8. Gravatar of Dustin Dustin
    20. December 2013 at 13:15


    FWIW I do some indication of causation in this graph – namely, dUE causing dWage.

    The indicators are:
    1) dUE is much steeper in recessions than corresponding dWage
    2) Peek dUE is consistently reached prior to dWage

    Obviously eyeballing it is not scientific, but this would fit an intuitive feel of the nature of things.

  9. Gravatar of Doug M Doug M
    20. December 2013 at 13:19

    Did the Philips curve break down in the 1960s and 1970 because of the Lucas critique, or because Samuelson replaced wage inflation with price inflation?

  10. Gravatar of Brian Donohue Brian Donohue
    20. December 2013 at 13:30

    Doug M,

    The way I learned the story, higher inflation stopped leading to lower unemployment once people started anticipating higher inflation, leading to accelerating nflation. In college in the 1980s at Chicago, we were taught about the “expectational” Phillips curve, viz. the trade-off between unexpected inflation and unemployment.

    It’s all about fooling people- money illusion or something.

  11. Gravatar of JohnB JohnB
    20. December 2013 at 13:52


    You’re getting to be like Murphy. All this attention on the econ blogosphere is making Krugman a smashing success. There is no bad press. He reminds me of that annoying kid in middle school who would bother people just to stay in the center of attention. Stop giving this guy so much of your time and he’ll scream and kick and cry and go away.

  12. Gravatar of Andrew Andrew
    20. December 2013 at 14:08

    Regarding your P.S.

    The cause of the 9% drop in NGDP growth was…..

  13. Gravatar of ssumner ssumner
    20. December 2013 at 14:08

    Brian, You said:

    “Who doesn’t know that already?”

    Everyone who mocks the “Phillips curve.” At least if they understand the original PC.

    kebko, That’s one of the strangest patterns I’ve ever seen, and something I did not anticipate. Correct me if I am wrong, but don’t we have large numbers of North Carolinians who are every bit as unemployed as in July, but who are no longer telling pollsters they are unemployed and searching, because they don’t feel they have to say that they are searching in order to keep their benefits? I can’t think of any other interpretation. In other words, not much as changed, except the way people answer questions. As you note employment may have nudged up 0.4%, but we’ll probably need a bit more time to get a handle on that. You are right that this has huge possible implications for next year’s unemployment rate, and perhaps the timing of the first interest rate increase.

    Thanks SG, I’ll take a look.

    Dustin, In my view NGDP determines wages with a long lag.

    Doug, Probably both factors contributed.

  14. Gravatar of ssumner ssumner
    20. December 2013 at 14:08

    Andrew, Tight money.

  15. Gravatar of Doug M Doug M
    20. December 2013 at 14:16

    Brian Donohue,

    I think the simpler explanation is that causation is backward. Low unemployment may cause higher inflation, but it is less likely that high inflation would cause low employment.

    Cows do not build fences.

  16. Gravatar of ssumner ssumner
    20. December 2013 at 14:55

    kebko, Let’s say that a year from now we’ve had 26 weeks max for 12 months and the national unemployment rate is 5.8%. That makes the mysterious disconnect between the unemployment rate and the huge loss of jobs even more puzzling.

  17. Gravatar of Dan W. Dan W.
    20. December 2013 at 14:59


    Do you maintain data and is it available for web viewing of NGDP and related economic data? How does today’s GDP announcement square with your model of what it should have been?

    What do you recommend as good data sources for macro-analysis?


  18. Gravatar of ssumner ssumner
    20. December 2013 at 15:29

    Dan, I don’t have data, check out the St Louis Fred site.

  19. Gravatar of Mads Llindstrøm Mads Llindstrøm
    20. December 2013 at 15:35

    I am not an economist, so maybe I am missing the obvious. But the first graph shows what I would expect from changes in supply and demand. If, for some reason, demand for labor falls relative to supply, I would expect to see 1) lower wages and 2) higher unemployment. If, for some reason, demand for labor rises relative to supply, I would expect to see the opposite. This is just like the graph shows.

    As I understand the Krugman piece, then the point is to show that there is a long term trade off between wage-inflation and unemployment. But the wage-inflation tops at approximately 4% and bottoms at approximately 1.5% at each swing, so how could it possibly show anything about long term trade-offs. To show anything about long term trade-offs, the graph should have a decade or two with substantially higher tops and bottom, than another decade or two.

    So what am I missing? Why do Krugman think this graph is a good argument for long term trade-offs between wage-inflation and unemployment, and not just an obvious response to changes in supply and demand?

  20. Gravatar of ssumner ssumner
    20. December 2013 at 15:40

    Mads, That’s right, and pretty good for a non-economist.

  21. Gravatar of ssumner ssumner
    20. December 2013 at 15:41

    I would add that the first graph shows the effect of demand for labor changes, so it’s not purely supply and demand, but rather S&D with the assumption that demand for labor shocks dominate.

  22. Gravatar of kebko kebko
    20. December 2013 at 16:32


    It may be that the activity before July was mostly unrelated to EUI. The big reaction in regular unemployment claims didn’t come until July:

    I think there might have been some other things going on. A new Republican governor with a friendly legislature was elected in November, and a number of other items have been passed, including a Medicaid cut, and possibly including some anti-immigrant issues.

    Oddly, I don’t see any negative employee compensation by industry trends in the BEA data except a big drop in the “Management of Companies and Enterprises” category, and a small drop in the public employee categories. Nothing there indicates a 1%+ drop in total employment. There is a drop in personal income in the 1st quarter, but that drop doesn’t come from wages.
    There isn’t an unusual drop in unemployment insurance transfers in early 2013. And, seeing the size of UI transfers compared to other public transfers, I can’t imagine that changes in UI transfers would have a noticeable effect on demand, anyway. I’m not sure what the decline in the first half of the year comes from.
    As for the decline since June, it would certainly be nice to have more sources of data, but for what it’s worth, it seems to confirm the disemployment effects of the policy.
    Evan Soltas is normally so thoughtful. His Bloomberg article was disappointing.

  23. Gravatar of benjamin cole benjamin cole
    20. December 2013 at 19:03

    Excellent blogging. For several years now the Market Monetarist perpective has held more water than any other view.
    If anybody can show me a more-promising approach…

  24. Gravatar of JohnB JohnB
    20. December 2013 at 20:34

    It’s funny to see 2 well trained macroeconomists (Sumner and Krugman) disagree about every point in macroeconomic theory. You’re all a bunch of cranks!!!!! Stick to micro and micro-reasoning where people can actually reach conclusions that have some solid methodological base.

    Macroeconomists are like doctors 1000 years ago. They all claim to be right with total certainty. There are wide areas of disagreement although the areas of agreement (GDP anyone) are totally bunk as well. The best thing for patients to do back in those days was to ignore the cranks and try to get plenty of rest. Today, the best thing for politicians to do when the business cycle hits is to just cut taxes and stay out of it.

    About monetary policy, I’d rather have an automatic system like the gold standard than a system run by people with as much useful knowledge in field as medieval doctors.

  25. Gravatar of JohnB JohnB
    20. December 2013 at 20:36


    Since you advocate targeting NGDP because stable NGDP growth tends to correspond to stable growth in output and employment, do you worry that if they started stabilizing NGDP growth, the previous relationship to employment and output would break down?

  26. Gravatar of Caravaggio Caravaggio
    21. December 2013 at 02:09

    Prof. Sumner ,

    great blog as always ( even better than Lars’ .. and that’s saying a lot )..

    Is that graph about the PC , representing nominal wage increases ? If so it begs the question , where can I find evidence of sticky wages contributing to the slump ? if I don’t have an idea where real wages are going , how can I even start to assess the causation ?
    Can you provide us with some data on expected inflation and correlated wage inflation ?

  27. Gravatar of Daniel Daniel
    21. December 2013 at 02:10


  28. Gravatar of Petar Petar
    21. December 2013 at 03:46

    Cant wait for that end of year post.
    I would call 2013 the year of monetary offset/ sumner critique

  29. Gravatar of Mike Sax Mike Sax
    21. December 2013 at 04:57

    Yes, not such a nice Christmas present for those being kicked of UI-but hey, those people are part of the 47%

    I know it’s a Christmas present for you. You can’t wait for those checks to stop You’re probably getting a bounce in your step just thinking about it.

    For the record I’m not on UI this year so it doesn’t effect me personally. However, that’s not why you support policies-just because of your immediate interests.

    Well it’s not why I do anyway. I shouldn’t speak for you.

  30. Gravatar of Mike Sax Mike Sax
    21. December 2013 at 05:15

    ‘It’s the most wonderful time of the year. It’s the hap-happiest time of the year.’

    I know your humming it when you think about these freeloaders all getting kicked off. You can admit it.

  31. Gravatar of ssumner ssumner
    21. December 2013 at 06:54

    kebko, Thanks for that info.

    JohnB, You said;

    “It’s funny to see 2 well trained macroeconomists (Sumner and Krugman) disagree about every point in macroeconomic theory.”

    Even funnier is that you did realize that in this post I agreed with him.

    You asked:

    “Since you advocate targeting NGDP because stable NGDP growth tends to correspond to stable growth in output and employment, do you worry that if they started stabilizing NGDP growth, the previous relationship to employment and output would break down?”

    That seems unlikely, although I’d expect it to weaken slightly. Most shocks are local, so there’s not much incentive to make wages stickier just because NGDP gets a bit more stable.

    Caravaggio. I would not waste time looking at real wages. But if you insist, nominal wages are at the BLS, and expected inflation can be derived from TIPS spreads. The St. Louis Fred has data on TIPS and regular bonds.

    Mike Sax, What makes you think I support the policy? Oh yes, I forget, you are just pretending to be clueless, you have some sort of secret agenda.

  32. Gravatar of John John
    21. December 2013 at 11:44


    That chart made the whole decade of the 1870s look far worse than the early 1930s. That just isn’t believable. The 1870s had falling prices as a result of the end of the civil war and the reinstatement of gold but by almost all accounts, actual production recovered well during the decade. What that graph really shows is that non-gold standard systems produce more price inflation.

  33. Gravatar of Daniel Daniel
    22. December 2013 at 04:06

    Reality 1 – gold nuts 0

  34. Gravatar of Mike Sax Mike Sax
    22. December 2013 at 06:29

    Scott, what makes me think you support the policy? Everything you’ve ever said about UI benefits till now. Like this for example:

    “For years I’ve been saying about 0.5% of the roughly 5.0% jump in unemployment was due to extended UI (albeit by much more than 13 weeks by 2009.)”

    Then again, if Nick Rowe can have evil CBers who want ever rising deflation and worsening recession in his model why can’t I have evil macroeconomists who want ever stiffer cuts to all government services and an ever worsening plight for the poor and umeployed in mine?

  35. Gravatar of ssumner ssumner
    22. December 2013 at 06:32

    Mike, Are you try to brag about how dumb your are? If not, why do you keep making such moronic statements? I really can’t figure you out. Surely no one can be this clueless.

    The statement you quoted says nothing about whether I support extended UI benefits. Read it again.

  36. Gravatar of Phillips Curve Specifications and the Microfoundations Debate | askblog Phillips Curve Specifications and the Microfoundations Debate | askblog
    22. December 2013 at 10:08

    […] Scott Sumner writes, […]

  37. Gravatar of andrew’ andrew'
    23. December 2013 at 02:21

    How bout this- cliffs are anti-economic. The government has done this to ui recipients.

  38. Gravatar of Mike Sax Mike Sax
    23. December 2013 at 19:34

    No I think you try to name call to change the subject-typical of you-gain you’ve said plenty of dumb things why be so modest. I just see it as what it is-you projecting yourself. As usual you’re incapable of answering a straight question and instead have to do the name calling-doesn’t say much about your intelligent.

    The fact that you think it adds to employment would certainly imply you support it.

    Why be coy? Are you claiming you don’t support it?

  39. Gravatar of Mike Sax Mike Sax
    24. December 2013 at 03:18

    I know Scott it’s a total jump to a conclusion-you say that extended UI adds half a point to unemployment and now that the GOP has got a chance to kill the extension to think you actually support this is just a total leap-there’s no reason in the world to think that.

  40. Gravatar of Mike Sax Mike Sax
    24. December 2013 at 03:24

    I think it goes back to that crazy flap with that writer Kaminska. A lot of times you think you’ve made these pithy points that you really haven’t made very well.

    It goes to the fact that you’re a pretty lousy communicator. I mean a good communciator is someone who is able to handicap the fact that X amount of people don’t read well. Yet, you’re trying to pretend that I’m the only one in the world that would presume that because you constantly have derided UI benefits-particularly the extension-as a bad policy that now that the extension has been cacelled you likely support it.

    I’d be interested in a kind of natural experiment where how any would read that quote of yours has somehow not inferring that you support ending the UI extension. I think you’d find that if that’s not what you support the problem is you not me. My comprehension is fine. You might try to make yourself clearer if you’re allegedly always being misunderstood.

  41. Gravatar of ssumner ssumner
    24. December 2013 at 11:01

    Mike, You said;

    “The fact that you think it adds to employment would certainly imply you support it.”

    You’ve just rejected the entire field of economics. By your logic all of economics 101 is wrong. 100% of it. Cost-benefit analysis? Never heard of it? So let me ask again, is this an act on your part?

    Brad DeLong thinks it adds a half point to unemployment, do you think he supports it?

  42. Gravatar of Mike Sax Mike Sax
    24. December 2013 at 15:24

    Well Brad Delong also doesn’t think I’m nearly as stupid as you want to give me credit for. Indeed, one reason I don’t let your personal insults faze me is because no one else accuses me of such things in such heated rhetoric. David Glasner and for instance are able to have a sane conversation. For what it’s worth the only people I seem not to have a civil economic discussion with is you and Major Freedom. Maybe you were in his civility class.

    You have certainly discussed UI a fair amount of times over the last 4 and a half years you’ve written the blog. I’m familiar with cost-benefit analysis. However, in all the posts I’ve seen you discuss UI you’ve only discussed the costs side.

    So I guess I should just have presumed that you see benefits in it even though you never mentioned them? That would not have been a ‘moronic statement?’

    I think it’s reasonable at least to assume as I did that you don’t see any benefits in it. Now of course maybe I’m wrong but this isn’t really my fault-if you think the point is important you should have qualifed it: ‘I think that it may cutoff about half a point of employment however, it does have these positive benefits to it so it’s not black and white.’

    As you didn’t do it all I could do is infer one way or the other. I guess I could have said ‘Scott often claims that UI raises unemployment significantly but of course he must know about its benefits as well he just never writes about them’ but why would that somehow be more reasonable than just going with what you’ve actually said and inferring accordingly?

    Again, if you don’t want to be mischaracterized on it then it’s on you not me to make sure you’ve clarified it and you can’t really act as if my reference is outrageous.

  43. Gravatar of Mike Sax Mike Sax
    24. December 2013 at 15:45

    ‘you can’t really act as if my inference is outrageous.’

  44. Gravatar of Mike Sax Mike Sax
    24. December 2013 at 15:50

    Scott I’ll tell you what. Let’s get a fair, impartial third party to administer an entry to Econ 101 and we’ll see how I do. I’d be wiling to d a friendly wager. Or I guess considering how you evidently feel about me-here I go inferring again-maybe it wouldn’t be so friendly.

    I’d guess I’d do fine in it. I aced Macro and Micro back in college back in 1995 and back then I didn’t even have much interest in the subject.

  45. Gravatar of Geoff Geoff
    24. December 2013 at 19:05

    Mike, seriously, quit being such a drama queen little b^&tch. Man up already and stop being so weak minded.

  46. Gravatar of Mike Sax Mike Sax
    25. December 2013 at 08:14

    Thanks for making my point for me Major.

  47. Gravatar of Daniel Daniel
    25. December 2013 at 09:50

    Mike Sax,

    Coming from a man who says talk of “monetary offset” is actually code for letting poor people starve, the irony is delicious.

  48. Gravatar of Geoff Geoff
    25. December 2013 at 11:43

    Mike, thanks for admitting that my point that you’re a drama queen little b&*ch is in fact your point all along.

  49. Gravatar of Mike Sax Mike Sax
    26. December 2013 at 16:02

    Daniel that’s a good way to put it. Maybe I’ll use it.

  50. Gravatar of The Phillips Curve is Horizontal – Credit Capital Advisory The Phillips Curve is Horizontal - Credit Capital Advisory
    23. March 2016 at 00:41

    […] criticized the approach of analyzing the direct relationship between unemployment and inflation. Scott Sumner has also attacked this approach stating that, “whereas price inflation is a useless concept, wage […]

Leave a Reply