Two views of the Phillips Curve

The standard (Keynesian) view of the Phillips Curve is that a strong economy leads to higher inflation. If I’m not mistaken, Milton Friedman reversed the causation, arguing that higher than expected inflation led to a stronger economy:

There is always a temporary trade-off between inflation and unemployment; there is no permanent trade-off. The temporary trade-off comes not from inflation per se, but from unanticipated inflation, which generally means, from a rising rate of inflation. The widespread
belief that there is a permanent trade-off is a sophisticated version of the confusion between ‘high’ and ‘rising’ that we all recognize in simpler forms. A rising rate of inflation may reduce unemployment, a high rate will not.

That’s also my view of causality, although I think inflation is the wrong variable.  The model should use the rate of growth in NGDP, not prices.

Here’s Nick Rowe:

Andy Harless’ tweet (about the US economy) got me thinking.

There’s a frog-boiling aspect to this economy. The consistent lack of *rapid* improvement throughout the recovery is enabling us to reach levels of employment that might not otherwise have been attainable.

It reminds me of my old post “Short Run ‘Speed Limits’ on recovery“.  The basic idea is simple: actual inflation (relative to expected inflation) might depend not just on the level of employment (relative to some unknown level of “full employment”), but also on the speed at which employment increases.

I’ve added an epicycle to the Phillips Curve that I think makes it fit the facts better. But I added that epicycle 10 years before the facts that Andy’s tweet asks us to explain. And it’s based on an idea that make sense, and goes back further still:

It’s difficult and costly to increase employment quickly, and easier and cheaper to increase employment more slowly, even for the same cumulative increase in employment. So if demand for output suddenly increases by (say) 10%, individual firms will raise prices and wages relative to the prices and wages they expect at other firms, or raise them more than they would otherwise have done if demand for output had slowly increased by that same 10%. Even with no underlying trend growth in productivity. Even with the same average level of demand for output.

There’s another way of thinking about this question.  Inflation is not determined by economic slack, rather it’s determined by monetary policy.  When monetary policy is highly expansionary and prices rise much faster than expected, then output tends to rise rapidly.  That’s Friedman’s view.  Thus it’s no surprise that a subdued rate of inflation is associated with a slow recovery.  Phillips curve models that predicted otherwise, i.e., mainstream Keynesian models, are simply wrong.

I do believe that Andy and Nick are making valuable observations here, albeit not because they provide a useful tweak to Phillips curve theory.  It’s better to simply drop the Phillips curve and focus on other models, such as the relationship between unexpected NGDP growth and changes in employment.

Instead, Andy and Nick are showing that the natural rate of unemployment is a slippery concept.  If inflation is stable (or better yet if NGDP growth is stable), then the economy will gradually move toward its natural rate.  Because of costs of adjustment, however, the fact that unemployment is currently above (or below) the long run natural rate does not mean that monetary policy is off course, even if inflation (or NGDP growth) is exactly on target, and even if the Fed has a dual mandate.  Monetary policy is off course if expected future inflation/employment outcomes are not consistent with the Fed’s dual mandate, as was the case during 2008-16

During 1933, both prices and NGDP rose rapidly, and yet unemployment was roughly 25%.  Now you could certainly argue that even faster nominal growth would have been desirable.  But even with appropriate monetary policy, unemployment in 1933 would have been well above any reasonable estimate of the natural rate.

Some people argue that the current 3.6% unemployment rate shows that money was too tight a couple years ago, when unemployment was 4.1%.  That’s not the case.  Money might have been slightly too tight in 2017, but only because inflation was also running a bit below target.  The optimal unemployment rate might well have been 4.1% in October 2017 (due to costs of rapid adjustment) and 3.6% today.  But it certainly was not 10% in October 2009.

Even when monetary policy is producing appropriate nominal stability, it would not be unusual to see the unemployment rate gradually falling (or rising) toward its long run natural rate.



47 Responses to “Two views of the Phillips Curve”

  1. Gravatar of Benjamin Cole Benjamin Cole
    21. November 2019 at 16:01

    Well lots of good commentary here, and it may be that a slower recovery does not trigger as much inflation as a more rapid recovery.

    But think about measured inflation in the current context. Kevin Erdmann runs a monthly review of the CPI core sans housing. It runs around 1%. Housing costs are rising at about a 3% annual rate.

    So what triggered inflation in the 1970s is obviously different than what causes the rather small amount of non-housing inflation we have today.

    Any conversation about measured inflation today must have a large dollop of attention paid to housing.

    There is a further question of whether we are using voodoo-hedonics to measure housing quality, or the value of housing, and thus inflation.

    We can posit that residents of Hong Kong, Santiago or Los Angeles get a great deal of value for their housing dollar as they live in cities with wonderful amenities.

    I think that is why there are studies that show that Hong Kong has a higher per-capita GDP PPP than the United States.

    Yet this survey finds that a one-bedroom apartment in Hong Kong city centre rents for just a little bit less than average monthly net salary:

    I sometimes think that American macroeconomists, perhaps of a certain generation, are frozen in time, that being the 1970s, just as an earlier generation of macroeconomists was frozen in the 1930s.

    But it is not labor markets that are causing inflation today, nor an erosion of real living standards. It is housing markets.

  2. Gravatar of Doug M Doug M
    21. November 2019 at 17:34

    The Phillips Curve is dead. Long live the Phillips curve.

  3. Gravatar of Benjamin Cole Benjamin Cole
    21. November 2019 at 20:00

    Add on (sorry)

    Another reason to think in terms of inflation not only through the monetary policy peephole (as has been pointed out by George Selgin):

    An economy can have structural decline in supply, and relative supply. Selgin posits the agriculture economy that has a crop bust, and so prices rise. Should we blame monetary policy for inflation, in such as case? Indeed, Selgin advocates a central bank accommodate the higher prices.

    Today we have huge structural declines in the relative supply of housing in various markets, such Hong Kong, Santiago, Australia, Canada, Great Britain, parts of China, the West Coast of the US and so on. Job markets keep urbanizing, but housing markets are often frozen.

    As Kevin Erdmann has pointed out, this results in housing inflation running well above CPI core, and especially CPI core sans housing.

    Monetary policy is important, labor markets are important. But the conversation now should probably migrate to housing markets.

    By one estimate, California is short 4 million housing units. The amount of greater prosperity that could be had by going to free markets in property development is mind-boggling.

    So why do America’s macroeconomists keep fretting about labor rates?

  4. Gravatar of rayward rayward
    22. November 2019 at 05:42

    Sumner’s reference to rapidly rising prices in 1933 might be confusing as prices were deflationary, but the rate of deflation dropped significantly (from an annual negative rate of 9.79% in January to an annual positive rate of .76% in December). From an annual deflation rate of 9.79% in January to an annual inflation rate of .76% in December was a huge leap, especially after three years of consistently high rates of deflation (especially 1931-32). The annual inflation rate spiked in 1934 (up to 5.56%), and remained positive until 1938-39 when it fell back to negative territory.

    Sumner’s explanation of how expectations affect growth is clear enough, but it has an element of reading the tea leaves to it: deciphering the expectations of consumers and businesses is a job for Carnac the Magnificent. More likely is that expectations are determined after the fact: if growth was flat, then the expectations of consumers and businesses of future inflation/growth must have been low. This is not meant as a criticism (and it’s certainly not original), but it makes predictions, especially about the future, very difficult.

  5. Gravatar of Benjamin Cole Benjamin Cole
    22. November 2019 at 06:05

    Carnac the Magnificent

    Johnny Carson. Those were the days.

  6. Gravatar of Randomize Randomize
    22. November 2019 at 09:51

    Dr. Sumner,

    Do you have any recommended reading on what exactly the “natural” rate of unemployment is? It always feels like some vague hypothetical that is, rather than an actual number, some equilibrium that’s dependent on all of these inputs:

    Neutral monetary stance which is difficult to define and even harder to identify;
    Neutral fiscal policy which is also poorly defined;
    Stable supply side factors must come into play also, right?
    What else?

  7. Gravatar of ssumner ssumner
    22. November 2019 at 12:01

    Randomize, Supply side factors can cause the natural rate to change. Demand side factors cause the actual rate to change, but not the long run natural rate.

    It’s a hard variable to estimate, but still a useful concept to use when thinking about what’s going on.

  8. Gravatar of ssumner ssumner
    22. November 2019 at 12:02

    Rayward, You need to look at month to month changes in 1933.

  9. Gravatar of marcus nunes marcus nunes
    22. November 2019 at 16:06

    “Even when monetary policy is producing appropriate nominal stability, it would not be unusual to see the unemployment rate gradually falling (or rising) toward its long run natural rate.”
    I would say that it´s exactly when monetary policy is producing the appropriate nominal stability that unemployment will be gradually falling to low levels ( I believe the the “natural” concept is a figment of the economists imagination).

  10. Gravatar of Michael Sandifer Michael Sandifer
    23. November 2019 at 02:47


    When you refer to “costs of rapid adjustment”, are you referring to the same costs Nick Rowe mentioned?

  11. Gravatar of ssumner ssumner
    23. November 2019 at 15:54

    Michael, There are a wide variety of costs of adjustment.

  12. Gravatar of Michael Sandifer Michael Sandifer
    23. November 2019 at 17:45


    If you’re ever motivated to do a post on that, I’d be eager to read it. I’d particularly like to know how changes in productivity play a role, if relevant.

    By the way, I have a new, potentially sillier hypothesis about monetary equilibirum conditions. It is that current earnings/price ratios for broad stock indexes should equal current nominal one year risk-free interest rate, in monetary equilibrium. I refer to this graph I recently created in Excel. This is the S&P 500 E/P yield versus NGDP:

    And here’s the 1-year Treasury yield versus the S&P 500 E/P yield:

    Divergences in these yield represent monetary disequilibrium, if I’m right. The Bretton Woods period doesn’t count, due to fixed exchange rates.

    So, r* = 1 Year T-Rates = S&P 500 E/P = NGDP growth expectations, in monetary equilibrium is the hypothesis.

    Fun fact: E/P is also the discount rate for the S&P 500, and in monetary equilibrium, would give a pretty accurate snap shot of current GDP growth. In the short-run, inflation matters. In the long-run, RGDP growth is what matters, which is why S&P 500 earnings growth has averaged about 3% since the Volker monetary regime change.

    If you want to be even sillier, you could start to claim that E/P might be the best measure of return on equities, potentially solving the equity premium puzzle in a way that would be hard for most to accept. It seems outrageously stupid or naive to think that risk-free bond returns and equity returns should have equal nominal returns in monetary equilibrium, but if you think about it, from this perspective one can argue the risks are equal and opposite, if the random chance of inflation or disinflation were equal.

    In our current inflation-targeting regime, there’s a disinflationary bias, hence equity returns being riskier.

    It is obviously time to move beyond simple calculations of correlation and squinting at graphs, and go to deeper exploration using actual econometrics.

  13. Gravatar of Michael Sandifer Michael Sandifer
    23. November 2019 at 18:31

    By the way, here’s a post by Jason Smith, in which he estimates that the S&P 500 would be about 17% higher today, without the trade wars.

    He has a really nice chart there with the trade war shocks represented as arrows, right under the effects on S&P 500 prices.

    This comports well with my calculation based on the equilibrium hypothesis, which I put in the comment section of that post. I haven’t been beaten up by a physicist over this hypothesis yet, so I hope to have it killed if it’s incorrect. I prefer not to waste my time with wrong ideas.

    I can say though, that the calculations my perspective allows for seem awfully accurate.

  14. Gravatar of Michael Sandifer Michael Sandifer
    24. November 2019 at 10:31


    Also, very serious question: Why should I look at data like this showing persistent growing demand for money, in this case, MZM, and not conclude that money’s too tight? MZM stock growth is not a perfect indicator, but it seems pretty good, and fits pretty well with output gap estimates based on my independent equilibrium assumptions.

    For example, I’ve said for a long things I think RGDP potential is about 3.5%, and this data is pretty consistent with that.

    How can demand for money grow without lowering the real and nominal GDP growth path?

  15. Gravatar of Michael Sandifer Michael Sandifer
    24. November 2019 at 10:32

    Here’s the link:

  16. Gravatar of P Burgos P Burgos
    24. November 2019 at 15:10

    What’s MZM stock? Also, @ Mike Sandifer, is growing demand for money a mechanical outgrowth of central bank policy? Couldn’t you see demand for money grow (velocity falling?) and an increase in the monetary base that keeps NGDP growing on a steady course?

  17. Gravatar of Michael Sandifer Michael Sandifer
    24. November 2019 at 15:48

    P Burgos,

    MZM is Money with Zero Maturity. It’s the broadest measure of money, and includes notes and coins in circulation, non-bank issued traveler’s checks, demand deposits, other checkable deposits, savings deposits, and money market funds.

    To answer your second question, credibly permanently increasing money supply growth versus demand reduces the growth in demand for money, and can even reverse it. Otherwise, yes, increasing the money supply to meet the increased demand can at least help temporarily, depending on the credibility of the central bank.

    There’s no question, if you look at that graph of increasing money demand, which is still occurring, that money is tight and was tight while the Fed was raising interest rates. Every rate increase was 100% unjustified, if we set aside the Fed’s inflation target, which isn’t explicitly part of their dual mandate anyway. Their mandate is for price stability and to minimize unemployment. Depending on how “price stability” is defined, those two requirements can be contradictory.

    If you look at this chart of money demand and unemployment, you can see the clear relationship, indicating that unemployment could have fallen more quickly than it did, and could continue a ways more, if allowed.

    That’s one reason I want to know more about these adjustment costs Scott mentions, because I wonder how much those costs matter to the people who waited years to find full-time jobs.

  18. Gravatar of Michael Sandifer Michael Sandifer
    24. November 2019 at 15:49

    Oops, here’s the chart:

  19. Gravatar of Scott Freelander Scott Freelander
    24. November 2019 at 16:34


    This looks very convincing. For years, Dr. Sumner has talked about demand for versus supply of money being central and you show damanf rising. Why didn’t Dr. Sumner refer to such evidence all this time? Maybe he is skeptical about velocity measurements.

    Much more interesting is your claim about stocks. If true, maybe some way of targeting the confluence of stock and bond yields could be used to guide monetary policy.

  20. Gravatar of Michael Sandifer Michael Sandifer
    24. November 2019 at 22:27

    Thanks for the questions and comments. I’m surprised at how positive the reactions have been so far. I expected to get crucified and/or ignored/laughed at. Perhaps I still will be.

    Here’s a blog post I just finished about it.

    I apologize that if it’s poorly written, but I’ve never written much in detail about these things. I will appreciate it if someone can debunk it quickly.

  21. Gravatar of dtoh dtoh
    25. November 2019 at 00:56

    @Michael Sandifer

    1. Are you sure the S&P Traasury correlation is not just the result of arbitrage.

    2. The NDGP/S&P correlation does not look that tight.

    3. If you’re trying to correlate S&P to NGDP shouldn’t you be looking at total returns (including appreciation) instead of just yield.

    4. Is it possible we are grossly over-estimating inflation. We’re measuring the price, but we’re not accurately measuring the change in product mix. If you look at improvement in product quality, maybe we are going through a period of massive deflation.

  22. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 01:33


    1. I’m pretty sure the S&P 500-Treasury correlation is due to arbitrage. Arbitrage determines the prices of all liquid assets. I specifically point out that the correlation is due to naturally similar rates of return, and virtually identical rates of return in monetary equilibrium.

    2. The NGDP/S&P 500 correlation is very tight by social science standards, and I argue that most of the lack of tightness is due to periods of monetary disequilibrium.

    3. Total expected returns are integrated in the price in the denominator of the E/P ratio. Dividends and expected capital gains aren’t free.

    4. I think inflation estimates are pretty accurate. Scott has pointed out many times that the private billion prices project inflation rate usually tracks CPI pretty closely, and CPI and the PCE inflation also track closely.

  23. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 02:08


    By the way, look at this graph showing MZM velocity starting to fall at a faster rate in 1995. It’s correlated with that reduced lack of fit beginning in the same year on the NGDP/S&P 500 graph.

  24. Gravatar of ssumner ssumner
    25. November 2019 at 15:43

    Michael and Scott, The money supply data (and demand/velocity) seems like a waste of time. NGDP is the best summary statistic of the combined effects of changes in money supply and demand.

    Isn’t E/P much higher than r?

  25. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 16:28


    While MZM is not perfect, it seems pretty good. The correlations between changes in this measure of money demand and variables of interest such as unemployment, NGDP, stock prices, etc. are high by macroeconomics standards.

    On your second question, remember that I claim that S&P 500 E/P should equal r in monetary equilibrium. Notice the great divergence in E/P and r just after the Great Recession, when longer-term interest rates began to fall in line with NGDP growth expectations. That was obviously a period of considerable monetary disequilibrium, as revealed in the much lower real and nominal GDP growth paths.

    Since then, the gap between E/P and r has closed, with the output gap as the economy has very slowly recovered.

    There’s obviously much more careful analysis required, but if you look closely at data provided and really think about it in terms of the math, it looks pretty damn good.

  26. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 16:40

    r* = r = NGDP growth expectations = S&P 500 E/P, in monetary equilibrium. The whole point of the proposed equilibrium condtion, is that it indicates whether the economy is in monetary equilibrium, and if not, how far away from equilibrium it is. If correct, this obviously means there are very precise ways to estimate r*, the output gap, RGDP potential, and what the stock market E/P should be, even when in monetary disequilibrium. It also means NAIRU is much easier to estimate.

    This is the model I’ve been using for three years, sans the E/P part, which is new. This model told me that unemployment had a signficant amount to fall, and the 3.5% RGDP potential I’ve been quoting for quite a while is well-supported in multiple metrics, if you accept the equilibrium condition.

    And I do take secular stagnation seriously. Without it, I think the RGDP potential growth rate would be about 1% higher right now.

  27. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 16:52


    If you haven’t seen it yet, I did a blog post on this last night:

    I offer a lot of graphs, and a fuller version of the argument.

  28. Gravatar of dtoh dtoh
    25. November 2019 at 17:40

    @scott Yes. Totally agree. Target NGDP.

    @michael – A couple of additional points.

    If you believe EMH, then risk adjusted after tax returns will be the same for all assets. The fact that returns are correlated is tautological.

    Even if expected appreciation is build into the price, that still doesn’t mean the e/p ratio is necessarily a good approximation of returns.

    Also, I’m not really sure what the benefit is of your proposed way of evaluating the stance of monetary policy. It’s already relatively easy to tell if policy is too tight or too loose.

  29. Gravatar of Michael Sandifer Michael Sandifer
    25. November 2019 at 18:34


    Yes, risk-adjusted rates of return should be equal under the current view, but that’s not my claim. My claim is that nominal returns should be equal in monetary equilibrium. The advantage here is in having precise ways to tell just how loose or tight monetary policy is at a given moment. My approach also offers the advantage that, in equilibrium, the e/p ratio reveals, because it is equal to, current NGDP growth.

    You say it’s already relatively easy to tell whether money’s too loose or too tight, so what’s your method? And why haven’t you shared it with the broader world, since there seems to be much disagreement?

  30. Gravatar of dtoh dtoh
    26. November 2019 at 08:13

    So Michael maybe I’m misreading. Are you suggesting that it allows you to set a more optimal NGDP target. Or that it is easier to stay on target. If it’s the latter, can’t you achieve the same thing by simply improving on the BEA’s antiquated methods for data collection. Also if you’re using e/p you still have a lag because of when companies report earnings.

    Finally, it seems to me that until the Fed is required to and committed to actually staying on target, then it is a bit of a premature and academic argument.

  31. Gravatar of dtoh dtoh
    26. November 2019 at 08:15

    And to answer your question, I’m a believer in NGDP level targeting. If you’re below the level target, then policy is too tight and the Fed needs to buy assets. If you’re over the target, policy is too loose.

  32. Gravatar of ssumner ssumner
    26. November 2019 at 09:10

    Michael, Isn’t the S&P yield a real variable while the T-bill yield is nominal? I.e., doesn’t the total return on stocks include both the earnings yield and the rise in stock prices due to inflation?

  33. Gravatar of Michael Sandifer Michael Sandifer
    26. November 2019 at 13:08


    First, NGDP targeting would be a big step forward, but remember that, sans my model or an NGDP futures approach like Scott favors, we don’t have good ways of targeting NGDP. I do think it is also possible to do a better job of targeting NGDP than possible currently, if the Fed were to adjust policy daily, based on GDP-related data on electronic transactions,for example, but we don’t have that system either. My model, if correct, would allow for the use of currently available data for guiding monetary policy.

    Second, my approach is better than NGDP targeting, in that it wouldn’t have the weakness in that if there were a productivity boom large enough, such that real GDP growth alone exceeded the NGDP target, adhering to the target would cut real growth. It may seem unlikely that this would ever be a problem, but recall that Scott favors an NGDP target of around 4% or perhaps a bit higher and we certainly had years in the late 1990s in which real growth under such a scenario could have been cut short.

    On your question about lagged reported earnings for stocks, P still moves moment-by-moment, so it isn’t an issue. In the long-run, earnings track real GDP growth.

  34. Gravatar of Michael Sandifer Michael Sandifer
    26. November 2019 at 13:14


    No, both empirically and theoretically, E/P is a nominal variable. Earnings, however, closely track real GDP growth. For example, a quick calculation reveals that, between 1982 and the present, the average growth rate in S&P 500 earnings was about 3%, the same as average RGDP growth.

    Theoretically, this makes sense, because higher inflation, for example, should and does initially boost stock prices, but persistently high inflation, after inflationary expectations have set in, should and will depress stock prices, as we saw in the 70s.

  35. Gravatar of Michael Sandifer Michael Sandifer
    26. November 2019 at 13:19

    Oh, and also, theoretically, it makes sense that earnings would closely track NGDP growth, because the 500 companies in the S&P 500 have saturated their markets, on average, and so depend solely on RGDP growth for increased earnings.

    Because the E/P depends on trailing 12 month earnings, this means that P will increase with rising inflation expectations, with E lagging until the next round of quarterly earnings reports. Similarly, P would fall with disinflationary expectations, while E lags until the next earnings reports.

    That just illustrates my point, that E/P will be constant in monetary equilibrium, ceteris paribus. Of course, ceteris paribus is very important here, but that condition does normally hold, and there are other equilibrium conditions that apply.

  36. Gravatar of Michael Sandifer Michael Sandifer
    26. November 2019 at 13:27

    Actually, that last paragraph was mistated:

    “That just illustrates my point, that E/P will be constant in monetary equilibrium, ceteris paribus. Of course, ceteris paribus is very important here, but that condition does normally hold, and there are other equilibrium conditions that apply.”

    Over short periods that will be true, but important to make sure nominal interest rates stay equal to E/P.

  37. Gravatar of JImP JImP
    26. November 2019 at 13:35

    Any bets on the idea that, had the advice of this web-site been followed Trump would never been elected?

  38. Gravatar of Michael Sandifer Michael Sandifer
    26. November 2019 at 13:40

    I should also say that, if we ever have a situation in which AI technology, for example, begins to allow ever increasing productivity growth rates, which some anticipate, simple NGDP targeting wouldn’t work well at all. But, if I’m right, nominal interest rates will still equal E/P in equilibrium.

    I don’t expect that above to happen in my lifetime, so it’s a very theoretical concern, but still.

  39. Gravatar of Scott Freelander Scott Freelander
    26. November 2019 at 16:12


    If I read your stock and GDP chart correctly,the stock yield shows real output growth hitting a wall in the 70s, is that correct? That shows you lost your argument against Dr. Sumner about real output limits in the mid to late 60s. ;P

    This is very interesting. That is what is expected if AD was temporarily exceeding the real GDP potential, and then bouncing back down below potential repeatedly.

    If that is right the gap in stock yields and GDP during and since the last recession shows us the output gap. Wow

  40. Gravatar of dtoh dtoh
    27. November 2019 at 03:17


    “Any bets on the idea that, had the advice of this web-site been followed Trump would never been elected?”

    No. He would have run as Democrat and beat Jeb Bush.

  41. Gravatar of dtoh dtoh
    27. November 2019 at 03:23



    1. Get better daily NGDP info. Not that diffi
    2. Even if p is current market price, it’s still largely based on old lagging e so it’s not really current.
    3. If 4% is too low a target for NGDP, set a higher number. Scott’s 4% number is politically influenced.
    4. I’m not sure I buy the argument that S&P earnings is a good proxy for RGDP growth. I’d probably argue that it’s biased toward faster growing segments of the economy and that it’s biased toward companies gaining market share.

  42. Gravatar of Michael Sandifer Michael Sandifer
    27. November 2019 at 09:30

    Scott Freelander,

    I would be careful interpreting the S&P 500 yield in this context, as RGDP bounced around a lot during the Great Inflation. That means that, instead of nominal changes moving in concert and dominating E/P, there were offsetting moves between real and nominal components. I hope that’s clear. That said, I agree with your ultimate conclusion. Scott won that debate. It appears RGDP was at its limit well before 1970, as he claimed. This is also the view put forward in Mishkin’s textbook.

  43. Gravatar of Michael Sandifer Michael Sandifer
    27. November 2019 at 09:41


    1. I agree. Why the Fed doesn’t simply get electronic GDP-tagged sales data daily is beyond me.

    2. That’s an empirical question. I have charts using monthly data for E/P also, and the results don’t change. I can send you the Excel file, if you like.

    3. Setting a higher NGDP target helps. It’s one reason I favor setting an NGDP target of at least 5%. I’d probably set it at 5.5%, just to keep the 2% inflation rate for now. However, if the techno-optimists are right, and productivity growth permanently accelerates, and then at an accelerating rate, then simple fixed-level targeting won’t work.

    4. I wouldn’t say S&P 500 earnings are a “good proxy” for RGDP growth. Earnings are a lot more volatile, but the average long-run compound rate of growth are very similar. The equilibrium condition I propose allows for current RGDP estimates in other ways.

  44. Gravatar of Michael Sandifer Michael Sandifer
    27. November 2019 at 13:43

    It was a poor choice of words from me above to say that earnings closely track RGDP, when that’s obviously not true. They track on average.

  45. Gravatar of Scott Freelander Scott Freelander
    27. November 2019 at 23:23


    Your ideas are very exciting, but they’re also very simple. It worries me that it might be too good to be true. Why wouldn’t this have been thought about decades ago, at least?On the other hand, I can’t find a flaw in it, but maybe Dr. Sumner or another expert can.

    Maybe it is just difficult for a lot of people to imagine the world in constant monetary equilibrium.

    Also, I am not convinced you solved the equity premium puzzle, but I guess it doesn’t matter for the purpose of monetary policy. The equilibrium conditions seem to hold, and so e/p is an important indicator.

  46. Gravatar of Michael Sandifer Michael Sandifer
    30. November 2019 at 01:34

    Scott Freelander,

    I’ve had the same thoughts about why not many others have discussed economics this way, and I’ve seen none take some of this thinking to its logical conclusion. As you say, maybe not many people spend time thinking about what the world would actually be like in monetary equilibrium, and also I don’t think that many economists pay attention to variables like E\P. It does give me pause that the ideas are simple, but not adopted in hundreds of years of economic modeling and research.

    That said, the more I explore the data implications and data fit, the less concerned I am. For example, look at this FRED graph of MZM money demand versus a CPI-based measure of real interest rates. Tell me that this measure of demand for money doesn’t matter.

    Also, it’s easy to dsicover what my model predicts about yield curves, with a 5% NGDP level target, versus the implied NGDP growth rate currently. For example, if we had a 5% NGDP level target now, the 30 year Treasury rate would be about 8.85%. This would be a very similar yield curve to that one that existed in the latter half of 1994, before the velocity of MZM money began falling at a faster rate in 1995.

  47. Gravatar of Michael Sandifer Michael Sandifer
    30. November 2019 at 08:23


    I wonder, by the way, why do I never hear economists perform calculations to determine what the yield curve should be, given various average NGDP growth rates? For example, right now the yield on a 1 year T-bill is 1.6%, and it isn’t much higher for a 10-year T-bond(1.78%). Even if we assume only a 3% NGDP growth rate over the next ten years, the 10 year yield should be over 2%(2.016%). Even if we assume only 2% NGDP growth over 10 years, the 10 year yield should be 1.88%. This is completely ignoring my equilibrium condition proposal.

    How can money not be seen to be tight right now, even using conventional theory?

Leave a Reply