Jason Furman on intellectual consistency

David Beckworth has an excellent podcast with Jason Furman. I don’t entirely agree with Furman’s views on macro policy—he’s more of a fan of fiscal policy than I am—but I was greatly impressed by the way he thinks about problems, particularly the way he handles data.

Furman consistently emphasized the need to be intellectually consistent, to avoid letting our policy views bias the way we interpret data. For instance, you often hear people say things like “Yes, the unemployment rate is low, but the employment/population rate is still depressed from a decade ago.” Furman points out that this argument cuts both ways:

I’m a little old-fashioned. I think the unemployment rate is the single best measure that we have for that because the number of people who want jobs at a point in time seems to vary a lot for all sorts of structural reasons. In 1983, you could have argued… or in 1982, that the economy wasn’t in recession because the employment rate or the prime age employment rate was higher than where it was a decade before, that would have been crazy. We had a 10% unemployment rate, but because of demographic changes, the employment rate was higher, not lower. So I think you want to take whatever your preferred variable is and look at what it was in 1999. Look at what it was in 1982, and ask, would it have worked in those circumstances too.

Those who read my comment section know that when commenters propose this or that metric, I often point to an earlier period of history when their preferred metric would have delivered what they would have regarded as unwelcome policy implications.

Here’s another example from Furman:

If inflation expectations go up to three at the end of this year, are you going to change your views about monetary policy or are you not? I think you probably won’t. So maybe rather than pointing to a data point that won’t change, that might change, make the real argument, which is, inflation expectations are too low. I’d like to see them rise to three or something like that. And so I think we’ve seen that over and over again. So I do predict that there’s a decent chance the breakevens are going to go up, and if they do, people are going to find all sorts of creative technical reasons as to why they’re high, rather than saying, “Actually, this said it was telling us something back in June. So now I have to continue to admit it’s telling us something.”

In other words, if you prefer a dovish policy even were inflation expectations to reach 3% in 2022, then don’t say we need easy money because the bond market shows inflation expectations at 2%. Instead just say that you’d welcome above 2% inflation. That way you have justification for your strongly held preference for dovishness even if things don’t turn out the way you expected. (This advice doesn’t impact me as I’m not a hawk or dove; I’m a stable money guy.)

In terms of current events, Furman favors a somewhat higher inflation target, which makes him a bit of a dove. If I read the interview correctly, he worries that people are making the argument for dovishness based on predictions that current inflation is almost entirely “transitory”, and that if some if it turns out not to be transitory then the credibility of doves will suffer.

My own view is that the most likely outcome is that current inflation is transitory, but there’s somewhat more tail risk of too high inflation than too low inflation during 2022-23. I believe Furman has a similar view.



20 Responses to “Jason Furman on intellectual consistency”

  1. Gravatar of Spencer Hall Spencer Hall
    5. July 2021 at 15:45

    That’s why I always point out that “nothing has changed > 100 years.

    FAIT is fraud. The FED from this point forward will be producing stagflation. What the BLS statistics considers outliers: energy, food, and rent, will never come back down. The relatively low weights of food, energy, and rent isn’t transitory (as the BLS naively / historically assumes).

    You don’t have to be a rocket scientist to observe what moves the markets and the economy. But you do have to be unbiased. The markets will turn in July without further stimulus.

    Any idiot can inflate prices. But it takes a trained eye to observe that the activation of savings is a catalyst.

    When you activate savings you have to simultaneously tighten monetary policy. That’s different from stimulating AD by the injection of money products.

  2. Gravatar of Michael Sandifer Michael Sandifer
    5. July 2021 at 16:02

    Spencer Hall,

    Markets are clearly indicating that nothing like stagflation is expected. RGDP growth is expected to be slow after the recovery, and inflation expectations have been falling for weeks. The 5 year breakeven inflation expectations never even got particularly close to 2.5%, in PCE terms.

  3. Gravatar of Rajat Rajat
    5. July 2021 at 17:24

    It was an interesting discussion. I had two main reactions.

    First, my recollection was that in relation to the second passage you extracted, Furman was referring specifically to David when he asked his question. Here’s the transcript from the start of that comment:

    Furman: And so you, I’ve noticed have placed a lot of weight on these inflation expectations. If inflation expectations go up to three at the end of this year, are you going to change your views about monetary policy or are you not? I think you probably won’t.

    Although David responded in his customarily acquiescent manner by saying he does place a lot of weight on bond market breakevens, I have always taken him to be like you in preferring monetary policy to target the forecast rather than pushing a dovish stance. Maybe he has become a bit more dovish over the years in terms of error costs, but I think not enough to warrant Furman’s accusation of inconsistency.

    Second, I was quite taken aback by this comment from Furman:

    You in 2019, put down a really elegant framework for nominal GDP targeting. If we were following it now, we would already have lifted off interest rates. And we’re going to, with extreme likelihood, overshoot the nominal GDP target we were on.

    That simply cannot be right using a NGDP level target path, as US NGDP has only just exceeded the last peak from Q4 2019. Even looking at quarterly rates, I am not sure if it is right. Do you have any thoughts as to what he meant?

  4. Gravatar of ssumner ssumner
    5. July 2021 at 20:47

    Rajat, Regarding the first point, I noticed that too. I wondered if Furman was responding to something David had said in an earlier conversation, as it wasn’t clear from the transcript itself.

    On the NGDP question, many people are predicting very rapid growth in Q2, and perhaps he expects us to overshoot within a year or two.

  5. Gravatar of Spencer Hall Spencer Hall
    6. July 2021 at 06:02

    re: “Markets are clearly indicating that nothing like stagflation is expected”

    Define stagflation. From 1955-1964, the rate of inflation, based on the Consumer Price Index, increased at an annual rate of 1.4 percent. R-gDp averaged 3.6%

  6. Gravatar of Spencer Hall Spencer Hall
    6. July 2021 at 06:26

    The 5.4-percent increase in the CPI for the period ending August 2008 fooled Bernanke too.

    That’s what N-gDp level targeting is all about. “FAIT implies that the FOMC’s short-term inflation objective will change over time—possibly even from meeting to meeting.”

  7. Gravatar of Michael Rulle Michael Rulle
    6. July 2021 at 06:37

    I never really understood ——-and still don’t——-why the unemployment rate is a better indicator of employment than the employment rate. Furman does say the changing difference between those numbers are a function of “structural reasons” as people’s situations change. I am not sure what that means—-although it implies some kind of particular thing happens on the margin which makes it less possible to influence desired outcomes thru policy. Sounds a bit mysterious.

    Of course one should not change their policy views by changing their benchmarks —as that is just introducing randomness into policy making. So that is a good point by JF—-although one would hope that is obvious to policy makers.

    I believe Scott is a proponent of usefulness with regard to technical issues (such as employment rate versus unemployment rate) ——but as I said —-I don’t know why we generally use unemployment. Is it more useful to use unemployment?

    Higher left tail risk than usual on inflation says Scott. The forward 1, 5 and 10 year B/E rate has slipped a bit lately —-but the longer trend has move a lot. The 5 year hit 2.75 and is down to 2.5. Powell clearly believes the baseline of last year is influencing forward opinions to the upside. Plus, we need to factor in AIT—-which means shorter term forward rate (1-2 years?) should be higher than 2%

    All the more reason Scott’s view on higher tail risk seems reasonable. Does not mean it is right of course——but it is reasonable. We will see.

  8. Gravatar of ssumner ssumner
    6. July 2021 at 08:42

    Michael, You said:

    “I never really understood ——-and still don’t——-why the unemployment rate is a better indicator of employment than the employment rate.”

    No one claims it is. The claim is that it better measures labor market disequilibrium, which is mostly true.

  9. Gravatar of Spencer Hall Spencer Hall
    6. July 2021 at 11:49

    The economy should do well considering the child credit cash, “a $250 monthly payment for each of your kids ages 6 to 17, from July through December.”

  10. Gravatar of Scott H. Scott H.
    6. July 2021 at 11:58

    I think the unemployment rate versus employment/population rate example is a poor one given what’s been going on with that debate recently. The issue there isn’t so much one of intellectual consistency but one of intellectually entertaining the idea that recent gov’t policy has changed the relationship between these two measures and what they are supposedly measuring.

    If I remember correctly, even the good Dr. Sumner also bailed on unemployment rate in favor of looking at the employment/population rate in a recent post here.

  11. Gravatar of postkey postkey
    7. July 2021 at 01:59

    5% to 10% ‘inflation’?
    “I argue that the velocity of circulation of broad money should return towards its long-run norms. These norms are appreciably higher than actual velocity in 2020 and in early 2021, when the pandemic caused people to defer expenditure. Pursuing further the logic in the 14th June special e-mail, I suggest that equilibrium national income should be understood as the quantity of money multiplied by equilibrium velocity, where equilibrium velocity is the inverse of the desired ratio of money to national income/expenditure. I also propose – for the sake of argument – that, when making projections for three important countries/jurisdictions (the USA, the Eurozone and the UK) we assume that,
     by the end of next year velocity will be 5% beneath its average value in 2018 and 2019, and
     money growth will proceed from now at an annualised rate of 5% (i.e., it rises each month by just above 0.4%).
    What does the exercise say about the likely rise in nominal gross domestic product between the first quarter (Q1) of 2021 and Q4 2022?
    The answer is that nominal GDP has to rise by 30% or so in the USA, and by about 25% in the Eurozone and the UK. By extension, annual inflation over the next 18 months to two years will lie in the 5% – 10% band in the USA, and will exceed 5% in the Eurozone and the UK.”

  12. Gravatar of Spencer Hall Spencer Hall
    7. July 2021 at 07:22

    What the FED has to be concerned with is the indiscriminate distribution of purchases between the banks and the nonbanks (Paul Meeks’ RPDs, reserves for private deposits, 3rd edition of “Open Market Operations” published in 1974).

    The O/N rrps have a limit. They will turn an expansive policy into a contractive policy. The current rise in the U.S. $ reflects this.

  13. Gravatar of ssumner ssumner
    7. July 2021 at 10:43

    Scott, It really depends on the purpose. The low level of employment shows that we need to create many more jobs. The unemployment rate better measures how hard it is for the unemployed to find a job (i.e. disequilibrium.)

  14. Gravatar of Spencer Hall Spencer Hall
    7. July 2021 at 14:44

    Re: ““I argue that the velocity of circulation of broad money should return towards its long-run norms”

    Hardly, the FED remunerates IBDDs (destroying velocity). And precautionary savings, the demand for money, is up. Also, more money is being dissipated in financial investment.

    I.e., the U.S. is turning Japanese “Japanese households have 52% of their money in currency & deposits, vs 35% for people in the Eurozone and 14% for the US.”

  15. Gravatar of Spencer Hall Spencer Hall
    7. July 2021 at 16:44

    Re: “I argue that the velocity of circulation of broad money should return towards its long-run norms”

    The interest rate spike in 2018 is the antecedent and paradigm.

    1) % Deposits vs. large CDs on “Assets and Liabilities of Commercial Banks in the United States – H.8”
    2) the ratio of M1 to the sum of 12 months savings

    As George says: “When interest rates go up, flows into savings and time deposits increase” (and counterfactually, velocity falls)

  16. Gravatar of Gene Frenkle Gene Frenkle
    8. July 2021 at 20:56

    The Trump supporters are the funniest using every different metric to put Trump in the best light—they poo-pooed Obama’s unemployment rate by pointing to the LFPR while ignoring the LFPR when Trump was president…and then argue Obama never had 3% GDP growth while Trump did IF one uses 4 consecutive quarters by which standard Obama had over 3% GDP growth.

    That said, considering the baby boomers didn’t reach 65 until 2011 AND manufacturing jobs and union membership peaked in 1979 (so work got easier while wages and benefits got worse along with people living healthy longer)…the LFPR should have been higher from 2001-2010…so I would argue that the unemployment rate was a suboptimal metric to use during that period and the LFPR never recovered to the high in early 2000.

  17. Gravatar of ssumner ssumner
    9. July 2021 at 08:44

    Gene, The unemployment rate also never fully recovered in the 2000s.

  18. Gravatar of Gene Frenkle Gene Frenkle
    9. July 2021 at 11:44

    sumner, I noticed that in Japan, UK, Germany, Canada, and US the unemployment rates were very low in 2019, so it wasn’t just America with very low unemployment rates. Is that evidence that central bankers are getting better at their jobs and that macroeconomics has entered a “golden age”? Or is it evidence that the economies and institutions are mature and they are on autopilot, i.e., Harvard MBAs and LSE are churning out good executives?? I think we can rule out politicians doing a good job because all of the countries have very different politics and they all had low unemployment rates.

  19. Gravatar of Kester Pembroke Kester Pembroke
    10. July 2021 at 05:43

    The measure of inflation is how much a person can buy with their income. If that isn’t changing yet nominal numbers are going up then you have an inflation.

    Otherwise you have a redistribution of output amongst the population.

    The Productivity Paradox is the primary mechanism of redistribution. From a firm’s point of view, using the infinite horizon capability of mainstream economics, it makes no sense to improve productivity. If you improve productivity in your field, then you may get a short term boost to market share and earn a lot of money, but then all your competitors will do the same thing and you will get into price competition which forces the price lower as the margin is used up chasing the demand that is limited by the wages paid. The result is that the consumer ends up with more for less, and the firm ends up relatively back where it started.

    So the best approach for firms is to agree with all their competitors to carry on doing what they are currently doing using the same level of productivity. An oligopoly or cartel.

    So you have a mainstream economics that believes in infinite horizon capability for inflation control (hence the Ricardian stuff), but appears to forget about it when it talks about the competitive mechanics that justify their desire to maintain wage caps.

  20. Gravatar of ssumner ssumner
    13. July 2021 at 08:34

    Gene, There is a global business cycle, so it’s not surprising that many nations saw low unemployment in 2019.


    “The measure of inflation is how much a person can buy with their income.”

    No, that’s RGDP. Inflation measures how much people can buy with a fixed quantity of money.

Leave a Reply