How should we think about the ideal size of fiscal stimulus right now? A thread with two approaches: (1) top down (based on filling the macro hole) and (2) bottom up (based on protecting people).
Three distinct issues:
(1) When do we need money? Simple: two months ago.
(2) How long do we need money? As long as it takes, could be years, ideally would have triggers to continue after Congress is fatigued.
(3) How much per month? Rest of thread is on this question.
A top down approach would ask what the output gap is and what the multiplier is. CBO's July forecast put the output gap at 6% in Q4, at a time when they expected the UR to be 10.5% this quarter. So presumably they would say something smaller, maybe 4%. cbo.gov/system/files/2…
Alternatively, Okun's Law says the output gap is 2*(7.9% unemployment - 3.4% full employment rate) = 8%. Let's round that up to 10% which could reflect the higher "realistic" unemployment rate, a further outbreak that hurts the economy more, or a lower full employment rate.
Next we need a multiplier. Most advocates of stimulus I see on twitter tout multipliers like 1.5. On the other hand, this may be too high and the very short-term multiplier might be lower, as low as 0.5. Let's use both.
Finally we can combine these with the fact that monthly GDP is $1.8T and get the following needs for monthly stimulus for the 0.5 multiplier / 1.5 multiplier case:
If the output gap is 4%: $36b / $108b
If the output gap is 10%: $90b / $270b
If legislating for 6 months then the total ranges from $216 billion to $1.6 trillion.
I would want to err on the side of more, I worry the output gap will remain large, and that the short-run multiplier is low. So I would be at the top of that range.
Now a bottom up approach. I'll consider four elements:
(1) Health needs like testing. I'm no expert, will arbitrarily pencil in $30b/month.
(2) Unemployed.
(3) States/localities
(4) Everything else
Second, how much do we need for the unemployed? Compensation in August was $56 billion below its pre-crisis trend. In theory for that amount of money could keep worker's whole (would still have lost business income etc.). That is about $350 billion over six months.
Alternatively, 28 million on UI or waiting to get on. If you support $600/week * 4.35 weeks per month that is $73 billion a month.
PUA for gig workers etc. ends at the end of the year. Need ~$10b per month for it next year.
Continued claims falling, so likely lower than $73b for month.
I prefer $400/month given the economy.
Nets to lower but I'll stick with $70b.
Third, how much is needed for states and localities? Auerbach, Gale and Sheiner put the *revenue* shortfall at $227b over three years. There is also additional demands on spending. Let's double the number and say $450b total, if over 6 months is $75b/month.
Finally, everything else? A lot of people not eligible for UI, in fact most suffering during CARES period appears to have been people not getting UI not people getting too little UI. Mechanisms for this is checks, SNAP, child allowance, housing vouchers, etc. Call it $60b/month.
I don't have a good basis for this, but $60b per month is as much as we spend on SNAP in a normal year, so would be enough for a 12X expansion of that program. Or enough for stimulus checks every three months. Or enough to close the personal income shortfall not counting UI.
Oh, and I would allocate $0 to restoring the state and local deduction and twice as much as that to PPP.
So the bottom up approach gets you the following per month:
The problem recently has been in both goods and services. Core goods inflation has typically been about zero but in the run-up to this year had deflation. Now tariff-driven inflation.
And at the same time core services inflation has picked up.
A market slowdown in the pace of job gains, with 22K added in August, bringing the three month average to 29K.
On a percentage basis have not seen job growth this slow outside of recessionary periods in more than sixty years.
The unemployment rate rose from 4.2% to 4.3% (unrounded was a smaller increase).
Wage growth was strong and average hours steady.
All of these are consistent with a marked slowdown in labor supply (due to immigration policy) combined with a continued slight softness in labor demand (as evidenced by the unemployment rate which has been steadily rising at about 0.03 percentage point per month for 2-1/2 years.
But two reasons to be less worried than headline: (1) transitory tariffs & (2) some of this is imputed from rising stock market.
Here are the full set of numbers I'll talk about.
Particularly notable is how much lower market core has been than overall core at every horizon. Note regular core includes imputed items, notably portfolio management fees where the price goes up when the stock market goes up.
Market core is both better predicted by slack and a better predictor of future inflation. It has moved sideways this year. But given that tariffs are (hopefully temporarily) pushing inflation up that suggests that underlying inflation is going down.
The jobs slowdown is here with 73K jobs in July & large downward revisions to May & June bringing the average to 35K/month.
Not quite as bad as you might think because steady-state job growth is much lower in a low net immigration world but unemployment still gradually rising.
A small portion of the weaker jobs numbers in recent months are Federal cuts.
But the bigger issues is the slowdown in private job creation.
My latest @nytopinion attempts to answer the question, "The Tariffs Kicked In. The Sky Didn’t Fall. Were the Economists Wrong?"
Part of my argument is the economy actually has slowed & inflation has picked up, as you would expect.
Plus Trump called off some tariffs and lags.
But there are two broader lessons here:
1. U.S. economy is mostly domestic services. Trade matters but it doesn't matter as much as some of the hype might make you think. (And I confess, I do suffer from TDS, tariff derangement syndrome.)
2. Much of macro is small on a percentage basis. But small things really matter a lot.
0.5% off one year's growth rate and $1,000 per household per year forever are the same. But the former sounds small and the later makes it clear it is a large unforced error.