This shows the initial deviation of real per-capita GDP from its long run trend, and the average growth rate over the following ten years, for 1925 through 2005. The long run trend is based on the 1925-2005 average growth rate of real per-capita GDP of 2.3%. The points in the upper left are the ten-year periods beginning in 1931 through 1941.


UPDATE: A number of people have objected to this exercise on the grounds that the Depression and World War II period is not relevant for our current situation. I don’t agree with this. But even without them, the picture is not so different. While the postwar period up til now has never seen a persistent deviation from trend as we are experiencing now, or as rapid growth as the Friedman paper projects, the relationship between the two is clearly present. And a decade of  growth far above the postwar norm turns out to be just what you would predict on the basis of that relationship. Here’s the same graph as above, but this time using only 1947-2005.


As you can see, the relationship is a fairly strong one. The Friedman growth number does lie a bit above the regression line. But it’s still true that the current exceptionally low level of GDP relative to trend would, on historical evidence, lead us to expect that growth over the next ten years will be around 3.8 percent  — well above anything previously seen in the postwar period and close to double the long-term average.

Note that the seven points well below the line in the middle are 1999-2005, whose 10-year growth windows include the Great Recession. Without them, Friedman’s number would be much closer to the line. What do we make of that? Should the exceptionally poor performance of this period make us more pessimistic about medium-term growth prospects (it’s sign of supply-side exhaustion) or more optimistic (it’s a sign of a demand gap that can be filled)? This is not an easy question to answer. But just counting up previous growth rates won’t help answer it.


27 thoughts on “Plausibility”

    1. It’s pretty straightforward. I just calculated a long run trend of real per capita GDP from 1925 through 2005 based on the average growth rate over the whole period. Then I compared the deviation from that trend in each year, with the growth rate over the next ten years. What you see is that the only period with output as far below trend as it is now, in fact saw subsequent growth rates equal or greater to those in the Friedman paper. You don’t see it in this figure, but 1934-1936 saw growth rates around 10 percent — about what Friedman suggests for the first year of a Bernie stimulus. So the historical evidence suggests that those kind of growth rates are just what you would expect when you are climbing out of a hole as deep as we are in now. Not implausible at all.

      1. Why do economists ignore the imminent effects of WTO, TiSA, and the rest? The decision last week in the US-India solar energy WTO case is a good example of convoluted and dishonest trade related signaling behavior, and its really about progressive liberalisation of the US, not India. To those who can see this, it seems to be making it clear as a bell that US government spending on products or services is now changed in a way which benefits business goals. And to not be afraid of Sanders because even if he wins, the WTO and pending new trade deals will ensure he is powerless to funnel money to US only job creation. The adversarial case likely is what might be called political theater. Its very Obama, reminiscent of his first State of the Union Address where Obama – with Senator Schumer’s help, made it crystal clear that Obamacare- because it needed to be ‘self sustaining’ and of course, sick people are expensive, and high premiums drive away the healthy preferentially, was designed to fail. So not to worry, wealthy.

        The recent WTO decision, especially when viewed with the USTR statement: ““This is an important outcome, not just as it applies to this case, but for the message it sends to other countries considering discriminatory `localization’ policies.”

        -should also be clear.

        Basically, to get access to emergent markets (Mode Three) at the most favorable/profitable terms they want, now, the US seems to be being asked to pay in (Mode Four) jobs.

        Those jobs will be very low paying jobs considering they will be buying the services of the allegedly highly skilled employees of the winning low bidders – We have to make it happen. They will be here on L1 visas, wages likely will be between them and their employers – who likely will have located wherever regulatory conditions (especially trade deals) are the most favorable to them.

        Which may well be on the other side of the world. Its quite unlikely that long planned ‘disciplines on domestic regulation’ would not apply. This will be an entirely new world, one which is not friendly to workers at all. It would be a huge huge setback. Once it starts, it becomes virtually irreversible, and neither the President PM, whatever, of a nation nor its elected legislature would have the power to change that. That is their intent. How that works is not flexible for them. They will claim this or that fixes them, thats part of the game of passing them it seems, lying, and the norm is massive misrepresentation. Also at least 99% of the media on them is inaccurate in that and virtually every other important respect.

        A good read on the history of trade in services agreements generally is Jane Kelsey’s book, ‘Serving Whose Interests’ You can find it online to read through Google.

  1. The time period that had outsize growth in per capital GDP involved a transition from an agrarian society to an industrial society. The huge growth in productivity in agriculture moved workers from agriculture (over 20% of the population until the 1930s) to less than 1% of the population today.

    The biggest problem with the Friedman analysis was that he ignored any limitations to growth, such as available pool of workers, productivity growth limitations, declining productivity of marginal workers added to the workforce, etc.

    Now lets address this statement “… just what you would expect when you are climbing out of a hole as deep as we are in now.” Have you checked private sector jobs growth recently?

    We need about one million jobs per year to keep up with working age population growth (about 85k per month). The CBO forecast used as the baseline in the Friedman study arrived at a similar figure, +77k per month. So the CBO forecast is a good baseline of just average job growth.

    Clinton added 20 million private sector jobs in eight budget fiscal years, bringing us to the highest labor force participation rate ever as women finished entering the workforce (aided in part in the 1990s by welfare reform that allowed new workers to keep partial benefits).

    During the GW Bush administration, we lost almost 2 million private sector jobs in eight fiscal years, essentially leaving us in a jobs hole of about 10 million from the nearly full employment situation at the end of the Clinton era. Over the last 6.3 Obama fiscal years, we have added over 13 million private sector jobs, essentially reducing the jobs hole to about 3 million. The jobs gains are accelerating somewhat, as we added about 5.7 million since the beginning of 2014. At current trend, we should add another 4-5 million more private sector jobs by the end of FY2017, essentially erasing the jobs hole left from the Bush years.

    This is roughly the starting point for the next president (although Friedman miraculously starts his forecast on January 1st 2017). From this level, Friedman adds +35 million jobs in ten years compared to the 9.2 million jobs in the CBO forecast needed to match working age population growth. This forecast means that Friedman projects that about 26 million additional workers can be coaxed out of the population over the ten years, over and above population growth (including immigration at current trend).

    This is an extraordinary claim. The jobs hole isn’t as big as Friedman and you claim.

    Your defense of this flawed study is indefensible. The Friedman study needs to be fixed to consider limitations to growth.

    1. “The biggest problem with the Friedman analysis was that he ignored any limitations to growth, such as available pool of workers, productivity growth limitations, declining productivity of marginal workers added to the workforce, etc.”

      And you ignore immigration.

      1. The CBO forecast does not ignore immigration; the current trend of immigration is included in the forecast.

    2. Labor force participation is endogenous–it has a lot of room to improve. In a truly robust economy, it will pull in a lot of people–just look at the late 1990s. You are limited by your inability to envision a robust economy.

      1. A lot of people, yes. An extraordinary amount of people, against all existing existing trends, and contrary to the educational incentives in the Sanders program… No !

        We have about 143 million jobs in the non-farm payroll (FRED), and should have about 147 million by the end of FY2017 based on current trend growth over the last several years. This is the employment situation inherited when the first Sanders budget will go into effect.

        Friedman claims we will add 35 million to this number in ten years under the Sanders program. This in spite of more people in school (free college), and increased number of retirees.

        Since the CBO forecast of 9.2 million over that timeframe is the minimum needed to match population + immigration, then Friedman is claiming the full employment in the Sanders economy will soar by 20-26 million over the employment level we reached at the end of the Clinton era (which is the highest labor force participation in the history of the United States).

        This is an extraordinary claim. The Friedman study compounds the era by claiming much slower household formation compared to this incredible jobs growth from our population, thus predicting a rise in real median household income to $82k, instead of the $59k in the CBO baseline.

        Friedman ignores the fact that the marginal jobs added will have lower productivity, and would likely result in more household formation; and he ignores the trend in younger people to seek alternatives to full-time employment.

        In short, the Friedman study is a pie-in-the-sky fantasy, and nowhere near anything we can expect in our real world.

        1. Hello.

          With all due respect, I think Mr. Friedman would likely be correct about the changes Sen. Sanders proposing causing a mini boomlet – were he not making the quite predictable innocent mistake of not knowing about the true direction of our nations ‘designed to hijack democracy right about now’ trade policy. Have some pity because that is the same mistake that virtually every other American who is not aware of the huge, 25 year long “progressive liberalisation” process is making too. We all are about to get an awakening. Read my last post particularly the statement by USTR Forman. I’m not expecting others to see this but we can still hope you will. America still can change this, but time is running out fast. Yell! yell like crazy.

          As an example in extreme national disconnection, look at the late Mr. Skala’s brilliantly observant paper on health care reform reality in 2009 (Video of him here: ) and his worst case scenario based on his hunches about the incoming Obama Administration: which were totally correct! ( ) Then, compare that reality- which is now compounded in its menace to democracy by three or more additional and probematic secret deals on the verge of completion.

          The difference between fact and the current media “debate” on health care couldn’t be larger.

          Please tell other people about, and especially- ask that the media intelligently cover this massive untruth problem soon! Thank you all. That boom can still happen and the world will rejoice with us.

    3. He does not ignore immigration. Counting fiscal years instead of actual years for job growth is a bit odd – and results in starting our count in the middle of a recession, excluding about 1 million jobs lost in Bush’s first year/Clinton’s last year – which adds another 2 million to the jobs hole.

      But I don’t get where he gets 1 million jobs per year. Historically, we’ve added an average of almost 1.5 million jobs per year. Going by that number and the historic jobs trend, we’re in a job hole of almost 11 million jobs.

      We have seen some drop in working-age population growth, but that isn’t discussed in this analysis. And some of it may be due to the recession.

      If you limit yourself to private payrolls, I see a historic average growth of about 1.25 million annually which would leave over 9 million missing using Paul’s numbers and the million from FY 2001.

      I ran J.W. Mason’s analysis on my own and got essentially the same result, although there were slight differences. Based on history including the war, we could expect about 4% annual growth average per capita over the next 10 years.

      I don’t think industrialization was a huge factor in the 1940s, although I could be wrong there… there was a boost in the Industrial Production Index, but it was largely a return to pre-Depression trend.

      I’m not confident the analysis applied the multipliers correctly though. It’s a question of stocks vs. flows – what causes the change in GDP is the change in government spending from the previous year, not the level of government spending above the prediction. And also should we be looking at gross government spending or net government spending? How much of the stimulative effect of the Bernie Sanders agenda will be washed out by taxes?

      Using Friedman’s numbers for spending under Sanders (regulatory+government) and his multipliers, I get 2.05% annual growth in Real GDP per Capita with an ending value of $69,855 in 2026 as opposed to the current projection of $64,903, assuming no inflation is added. This is compared to a 1.3% base growth rate (I’m not sure where Friedman gets 1.7% – it seems like a typo?). Most of this is the first year with 7% added to GDP per capita in new spending in the healthcare program… A more complicated model may smooth the growth or something. And there’s the question of any inflationary effect from this spending and the effect of the taxes. This is a larger stimulus than the ARRA, but it is worse timed and not that much larger. I don’t think my estimates of this impact are inconsistent with CBO estimates from ARRA, assuming the fiscal multiplier is now about half as large (using half, the CBO projection of Sanders spending impact would be 5.6%-15.2% in the first year with continued spending adding 0.35%-0.95% to growth in subsequent years). If CBO projected baseline growth is too low (which it seems to be based on your analysis), then this would reflect that. They projected that ARRA would have no long-term impact on GDP reflecting the fact that it was temporary spending – this is not.

      1. Thank you Jay. I’m very glad to see someone else repeat this exercise and get similar results — it’s easy to make mistakes when you are doing this stuff in real time.

        On the second part of your comment — I’ve been focused here on whether the overall growth rates are plausible, not on the analysis of Sanders’ specific proposals.

          1. Yes, I saw that.

            From my point of view, this has never been about the numbers in that paper. Jerry can defend himself, I am not going to. The question I am interested in is whether it’s plausible that higher demand — via increased public spending or some other means — could lead to a much higher level of output. Nothing I’ve written on that question depends on the estimates in the paper. Even if we end up throwing out Jerry’s numbers completely, the case for more expansionary policy will be just as strong.

  2. Paul,
    Perhaps employment alone is not enough to contribute to growth. If 100% of the working age population were employed, and they still didn’t make enough to pay all their bills, then the economy would still be depressed, since nobody would have enough money to spend on top of their bills. What I am trying to describe is 35 years of very depressed wages in America, plus household debt putting us collectively deep in the red.

    Add on top of this the fact that so much of household spending is on things that cost too much due to monopoly, such as cable, phone, TV, etc. Housing is too expensive, due to speculation.

    The entire public sphere has been gutted in the U.S. Anything that can be charged for, is charged for. With that amount of expense being carried by every adult in America, would it not be fair to say that an increase in effective income and wealth, no matter how it occurs, would be immediately be soaked up by the economy and lead to real growth?

    It seems that having a job, in and of itself, is not the issue. The fetishization of work is insane in America. The real issue is how much free time we all have to enjoy life.

    The need for wealth and income is due to the cost of living. The more big ticket items that are reduced in cost thru making them public, or destroying monopoly, or through introducing effective taxation, such as Henry George advocated, the better.

  3. Krugman’s trck was to leave out the Depression and WWII. He’s a shady bastard. I’ve been saying it for years. Watch him like a hawk. Total Democratic Party hack.

    1. I’v e been wondering a lot recently why there’s such a strong convention in macroeconomic discussions to pretend the WWII period never happened.

      1. Probably because it’s sort of a unique event?

        If you’re forecasting world population growth would you want to include WW2 in your forecast?

        1. Sure… why not? There’s a possibility of other disasters that kill millions of people and growth in the decade was about 9.5% according to the UN estimate, much higher than WW1, 6.3%, and not that different from the proceeding decade. Both of those numbers are pretty close to the trend… and growth kinda came back to trend after. It isn’t a huge, obvious deviation. Even the 60 million or more killed in WWII over 6 years does not amount to much when 10-20 million people are born each year. (Growth did seem to accelerate thereafter, although the growth rate peaked in 1970 and population added peaked in 1989, according to the US Census estimates).

          The point is that you should use all the historic data available, because we generally have a pretty small dataset and lots of uncertainty.

        2. I think this is exactly backward. As Krugman himself says, natural experiments are much more useful than regressions in establishing causal relationships. And the depression and World War II are the biggest natural experiments we have on aggregate demand.

    1. His argument there is basically that we should expect demand shocks to have permanent effects. It’s a pretty radical claim — definitely not orthodox — but I’m sympathetic to it. My response is: Why doesn’t it work both ways? If a period of exceptionally high unemployment causes skills to atrophy, people to leave the workforce, etc. (and I agree with him that it does), then why won’t a period of exceptionally low unemployment have the opposite effects? E.g.:

      1. My two cents:

        I think that we should divide the determinants of growth in 3 different groups:

        1) Supply side determinants;
        2) Cyclical demand side determinants;
        3) Supercyclical demand side determinants.

        1) “Supply side” determinants, essentially productivity growth due to technology and population growth; I personally think that productivity growth is essentially exogenous.

        2) Cyclical components of growth: this essentially represents the percentage of people working; for example if today 50% of people are working and tomorrow 55% of people are working, ceteris paribus I would expect a growth of 10%.
        I think that this kind of cycle follows a “Goodwin” dynamic, however I reread Goodwin’s original paper and I think that it has the big problem of being formulated in absolute, “real” quantities for capital, while for a cyclical model a relative measure of capital would be better (also Goodwin original paper doesn’t have disinvestiment, that at least in relative terms exists in reality).
        Ideally, we could see total growth as a theorical more or less straight upward line that represents productivity growth and 100% of people working (“potential” GDP), with a bouncy wave below the line that approximates the theorical line at peaks, then falls, then reaches another peak etc., representing effective GDP.

        And this is the main problem: can we expect all peaks to be at the same height relative to theorical growth?
        For example, if the first peak is 90% of potential, the second 80% of potential etc., we would see real growth diverging from potential growth.
        This deviation of “peaks” from “potential” is what I call:

        3) Supercyclical demand side determinants of growth.
        I say demand side because supply side effects are already accounted for in the “potential”.
        I think that various effects, for example high levels of debt, or an increase in rents, can cause this divergence.

        However, “potential” GDP is usually estimated as a trend of the peaks – that is, it is assumed that when the economy peaks, it reached potential (something that in my opinion is impossible because, since I believe that cycles are caused by a Goodwin dynamic, at potential the wage share would be 100%, that obviously is impossible in a capitalist system).
        Therefore a change in these “supercyclical” determinants of growth are attributed to a change in the growth of potential, hence “histeresys”, “secular stagnation” etc. .
        This is for example a big problem in the EU because the ECB is estimating lower and lower “potential” for the economies of Italy, Spain etc., and thus assuming higher and higher levels of “natural” unemployment.

        But the theoric problem comes from the fact that usually economists (at best of my understanding) only think in terms of (1) supply side long term determinants or (2) cyclical demand side determinants, and never of (3) supercyclical demand side determinants.

        1. These are difficult questions. I am coming to the feeling that you can’t just talk about “the supply side” and “the demand side” as two sets of factors affecting the same outcomes. They represent two incompatible frameworks for thinking about the economy.

  4. If we’re going to ignore the data from 1925 t0 1947 because of Depression, War, and Structural Transition (Agrarian to Industrial), then we should likewise ignore the effects of the “Great Recession.

    The rational for the removal of data points from a long term trend must be applied equally.

    What happens if you remove the 1999 to 2005 “outliers”?

Comments are closed.