I was recently presented with an intriguing thought experiment by Jason Furman, the chair of Barack Obama’s Council of Economic Advisers. He asked me:
“Assume that on September 16 2008 you were put in charge of US fiscal and monetary policy. You are given perfect foresight about the slowness of the recovery and the lack of an inflationary spiral. You are allowed to use the same tools (maybe plus a little) so nothing highly exotic, but you can use them sooner, keep them on for longer, and dial them differently. If you did this:
“A. What is the level of output today compared to what it actually is?
“B. What is the growth rate of productivity compared to what it actually is?
“C. What is the integral of output from 2009 to the present compared to what it actually has been?”
It is an excellent question, because it focuses the mind on how we think about the macroeconomy. And it is particularly interesting because Furman has asked the same question to prominent US economists and former central bankers — so make your own back-of-the-envelope calculation before you read to the end to see the answers of the leading lights of the US economics profession.
Here is how I think about this. I will, conservatively, rule out the possibility that the productivity growth trend could be any faster because of more stimulus (even though I think it could), and will just take as given the Congressional Budget Office’s expectation for potential GDP growth — 2 per cent a year for the next four years — once the economy fully employs its resources. (So my answer to question B would be “the same”.) Then take as a starting point that the US economy is today at full employment, and that this point could have been reached earlier — let us say about five years ago, in the summer of 2013 — with additional demand stimulus.
That would be about four years after the trough of the 2008-09 recession. It is hard to see why, if 60.5 per cent of adults and 79.5 per cent of 25 to 54-year-olds could be employed (with no wage pressure to speak of) by summer 2018, the same proportions could not have been brought back to work within four years of the recovery with sufficiently fast demand growth. That would have meant about 2 percentage points more on the employment rate (or about 3.5 per cent more workers) and more than 3 percentage points (or about 4 per cent) more for prime-age workers than was actually the case in summer 2013.
How could demand have been boosted? I would have wanted a stronger fiscal stimulus, but, above all, more aggressive monetary policy. Interest rate cuts below zero should have been attempted and sustained unless they proved to disrupt money market funds excessively. The Federal Reserve should have outright targeted long-term interest rates such as the 10-year Treasury rate, in the way the Bank of Japan later decided to keep 10-year Japanese government bond yields at zero (this is technically sufficiently similar to what the Fed has done that I count it as within Furman’s rules). Forcing long rates lower would have quite directly eased the most relevant borrowing costs for both households and businesses, which should have boosted both consumer demand and (therefore) capital investment. Using these tools aggressively in the downturn would only give way to cautious normalisation when the economy reached full employment as defined above.
How much higher would output have been? Assuming nothing more than the additional workers employed, output in mid-2013 could have been about 3 per cent higher than it was (allowing for marginal workers being somewhat less productive than the average) — nearly 1 percentage point higher annually from the start of the recovery. Projecting forwards from there, trend growth (at 2 per cent) would make GDP in mid-2018 three-quarters of a per cent higher. That is the answer to question A. The answer to question C is the sum of period-by-period difference between the higher and the actual GDP experiences. Using the simplified linear paths I have used in the chart, this means about 15 per cent of annual GDP more output — or about $3tn.
That is a big number. But it is far too conservative to think that employment would have recovered five years earlier without investment rebounding sooner, too. As the chart below shows, the recession and slow recovery caused a huge shortfall in private investment. That is not surprising: with sluggish growth, there was little reason for businesses to invest in new capital. But it stands to reason that faster demand growth from my preferred macroeconomic policies would have encouraged greater investment. How much more?
Here is one way to think about this. At the start of the recession, private investment was already down from its pre-recession peak, then plunged by another 30 per cent. By 2013 it was restored to its immediate pre-recession level but left behind a large accumulated investment shortfall — capital that had not been created. It seems plausible that a demand policy returning the US to full employment by 2013 (instead of 2018) could also have boosted investment to actual 2018 levels by then, about 20 per cent higher. Assuming investment growth at the same rate as GDP after that, the rate of investment today would be about 10 per cent higher, and the capital stock more than a year’s worth of investment bigger. With investment rates normally about one-sixth of GDP and a capital stock about three times GDP, that implies a capital stock about 6 per cent bigger by today, which would mean productive capacity about 2 per cent higher (using the average GDP-to-capital-stock relationship). Over the course of the recovery, that would have added a cumulative income of more than $1.5tn. On these — conservative — assumptions, then the cost of weak macroeconomic management amounts to about $5tn or a quarter of annual US GDP.
But we are not done. If the US is not yet at “full employment”— and tepid wage pressures suggest it is not — then achieving full employment by 2013 would have meant higher GDP still. If as many of 25 to 54-year-olds worked as in 2000, which other countries have managed, the labour force would be almost 2 per cent or so bigger from then on. The cumulative extra output over the decade would have been another 10-15 per cent of annual GDP. With such growth, investment could be expected to be even higher, further upping productive capacity. The wasted economic output thus begins to approach 50 per cent of annual GDP.
And this is still conservative. It does not consider that even faster demand growth might have been met by increased hours worked. Nor does it consider that overall or “total factor” productivity (above that gained from simply having more capital) could have responded positively to a “high-pressure economy”, as businesses try to produce more efficiently to satisfy demand or cope with wage rises. Either possibility would increase further the estimated current output level and the wasted output over the decade — and possibly even the ongoing rate of growth from here.
It is instructive, then, that my lowest estimates coincide with the highest estimates collected by Furman from his economist luminaries. He told me:
“On the current level question, one person said 0%, most people said 1% (which was my answer before doing the poll), a few said 1-2%, and one person said 2%. For the current growth rate most people said it would be lower than it is . . . For the integral of the shortfall from 2008 to the present I got a variety of answers, ranged from about 5 to about 15.”
These are quite enormous differences — they imply, as Furman concluded, “vastly different mental models”. Who is right matters, most importantly, for what the right policy is in a deep recession, and less importantly, for how much leading economic policymakers should be blamed for the job they did during and after the last one. I find it striking that on the models cited by Furman, policymakers come out looking mostly blameless.
- S&P warns that a no-deal Brexit would send the UK into recession.
- My FT colleagues have produced a long read on how the Irish border became the biggest point of contention in the Brexit negotiations.