Professor of Practice at Harvard. Teaches Ec 10, some tweets might be educational. Also Senior Fellow @PIIE. Was Chairman of President Obama's CEA.
Sep 14 • 8 tweets • 2 min read
The breathless debates over whether real wages are up since pre-COVID (& not sure why it's a debate--they are up) are all missing a key point: households/workers are mostly getting bigger pay increases than the data suggests.
🧵on repeated cross sections vs. longitudinal data.
The average data people cite is for a different group over time. And the groups are, to a first approximation, the same in in terms of age/experience/seniority.
If you followed the same group of people they would get a bigger raise because their age/experience/seniority rises.
Sep 13 • 13 tweets • 4 min read
We had 2 consecutive unqualifiedly good CPI reports. I was hoping for a 3rd but this one is only qualifiedly good. Not a huge concern but some.
I'm focused on core CPI which grew at a 3.4% annual rate after 2 months <2%.
Note, I'm not worried about headline inflation. It was very high in August and the 12-month rate rose from 3.2% to 3.7%.
IS a good measure of how August was a difficult month for households with an 11% (sa) increase in gasoline.
But IS NOT a good predictor of future inflation.
Sep 3 • 15 tweets • 4 min read
A nice write-up of the stunning increase/level of the deficit by @JStein_WaPo. In fact the increase is so stunning that you have to think that at least some of it is transitory noise. But some of it probably is not just transitory noise.
A 🧵. washingtonpost.com/business/2023/…
Note: My numbers below depart from the official numbers in that they do not count a ~$400b cost in FY 2022 for student loan debt relief that never happened nor do they count a similar-sized saving that will likely show up in the FY 2023 numbers as a result of the SCOTUS decision.
Aug 29 • 9 tweets • 3 min read
The immaculate loosening of the labor market continues as the quits rate hits a landmark 2.3% in July, which is where it was prior to COVID.
The quits rate is a good predictor of inflation--and a good benchmark for how workers are perceiving the labor market.
The job openings rate also has fallen from a peak of 7.4% in March 2022 to 5.3% in July. It is still 1.4 standard deviations above pre-COVID (although some argument that it was trending up prior to COVID so may be less tight then it seems).
Aug 21 • 19 tweets • 4 min read
My @WSJopinion argues that the Fed should make a "hawkish pivot" to a higher inflation target. Specifically get inflation <3.0%. Then shift to a 2-3% range, stick with that, and re-emphasize the price stability side of the mandate.
A 🧵 summarizes. wsj.com/articles/the-f…
1st: What is the right blank slate inflation target? Inflation is costly & inconvenient. But it also plays a useful stabilizing role that minimizes the severity of business cycles by making it easier for businesses to cut real wages, sectoral adjustments & negative real rates.
Aug 10 • 10 tweets • 4 min read
The CPI came in basically just as expected--the second unambiguously good month in a row (following a few months that most likely should be interpreted as improvements as well).
Core CPI at an annual rate:
1 month: 1.9%
3 months: 3.1%
6 months: 4.1%
12 months: 4.7%
As expected 12-month CPI rose from 3.0% in June to 3.2% in July. Tells you less about July 2023 (when inflation was low, 0.2% m/m) than about July 2022 dropping out of the sample (when inflation was even lower, 0.0% m/m).
Rough view of what the Fed should do this year based on core PCE in 2023-H2:
1. Above 3.5%: Hike more than once.
2. At 3.5%: Hike once.
3. Below 3.5% & UR < 4%: No more hikes.
4. Below 3.5% & UR > 4%: Cut rates.
This is roughly consistent with the reaction function implicit in the June FOMC Summary of Economic Projections (note full-year core PCE of 3.9% means ~3.5% in H2). So to the degree real activity and inflation deviate from what they expected they should react accordingly.
Jul 12 • 12 tweets • 4 min read
The headline is CPI inflation fell to 3.0%. But mostly this is the huge June 2022 # dropping out & what is almost certainly a transitory 17% fall in energy prices over the last yr.
But underneath the #s show moderation, with the lowest core reading of the inflationary period.
Overall what is happening is that service price growth is gradually slowing--as was predictable and predicted given the lags in the shelter measure. And goods price growth which was unusually (and likely transitorily) high in recent months turned back down again.
Jun 28 • 12 tweets • 4 min read
I don't understand Chair Powell's claim that "policy hasn't been restrictive for very long".
The Goldman Sachs Financial Conditions Index reached its current level of restrictiveness over a year ago--and has eased considerably since the fall.
The Fed funds rate does not directly affect the economy. Instead the FFR--and expectations of it--affect other variables that do matter for the economy. For example, most of the increase in mortgage rates was in the first half of 2022, more than a year ago.
Jun 14 • 4 tweets • 1 min read
Pleased to see that the Fed continues to be data dependent: they are expecting higher inflation & lower unemployment so raised their median expectation for rates by 50bp. I wish they had gotten started on that process today, it may not be enough, but a reasonable place to land.
Simple rules: they raised their core PCE inflation forecast by 30bp. That alone is consistent with about a 45bp increase in the FFR (or a 15bp increase in the real FFR). Then the 40bp decline in the unemployment rate would suggest some additional tightening on top of that.
Jun 13 • 4 tweets • 2 min read
Some of the slowdown in CPI-based measures reflects slowing inflation (albeit slowing less than expected).
But some of it is about a narrowing gap between CPI-based measures and the PCE-based measure that the Fed is focused on.
CPI was 2pp above PCE, now is less than PCE.
You see a similar story, albeit less stark, with core measures of inflation that exclude food and energy.
Core CPI has fallen a decent amount from its peak but core PCE has not fallen nearly as much.
Jun 13 • 5 tweets • 2 min read
The CPI-based Ecumenical Underlying Inflation measure was 3.5% in May, down from 4.0% last month. This is the median of 21 measures (7 measures over 3, 6 and 12 months each) all re-meaned to match PCE inflation.
Note how much variation there is between these different measures. This is specific to the CPI where shelter has a large weight. Look at last month's PCE-based Ecumenical Underlying Inflation measure, much more consistency as you look down the columns.
Inflation forecasting is hard. But I feel confident in these two predictions:
On July 12th BLS will announce the 12-month CPI went down.
On August 10th BLS will announce the 12-month CPI went up.
The reason I'm confident is the same reason that changes in 12-month inflation are not very imformative, it's because these changes tell us more about what happened a year ago than what is happening now.
Jun 13 • 4 tweets • 2 min read
One of the upcoming positives for inflation (especially the CPI where housing has more weight) is that new rent growth has slowed dramatically. Here is the measure I'm using, a simple average of a seasonally-adjusted version of Zillow and Apartment list.
You can see how different that is from all rents.
Jun 13 • 11 tweets • 4 min read
Core inflation (which excludes food and energy) rose at about a 5% annual rate for a sixth straight month. But other measures of underlying inflation were much lower reflecting the outsized importance of shelter and used cars (again) this month. Some relief should be coming.
Here is the standard core concept. It has been relatively steady at an annual rate:
Generally quite a good predictor of future inflation because it excludes volatile components. But...
Jun 9 • 7 tweets • 2 min read
Over the last 12 months the budget deficit has been 8% of GDP (6% ex student loans) while the unemployment rate has averaged 3.6%.
Historically this low an unemployment rate is associated with a balanced budget.
I prefer to look at the primary deficit (which excludes interest), especially for comparisons across long time periods with very different debt levels and inflation rates.
Same story: primary deficit is 6% (4% ex student loans), historical association is a primary surplus.
Jun 7 • 5 tweets • 1 min read
Two broad ways to be a dove on monetary policy:
1. Getting to 2% inflation is easy, just wait for it to happen on its own.
2. Getting to 2% inflation is hard, not worth it when 4% inflation isn’t so bad anyway.
For some reason I hear vastly more of 1 than of 2.
I’ve said that getting to 2.0% inflation will plausibly require an unemployment rate of 6.5% for 2 years.
I see people explaining “here is why you’re wrong and it will be easier than that because of lagged rent, nonlinear Phillips curve, Beveridge shift, etc.”
In June they will likely raise GDP & inflation forecasts & lower the urate.
IF they don't change the FFR too will shout "we are data independent."
Magnitudes: if the inflation forecast goes up 0.2pp and the unemployment rate forecast goes down 0.2pp that, by itself, *ought* to result in *at least* a 50bp increase in the 2023-Q4 FFR dot.
That calculation assumes their conditional policy stance was right in March...
Jun 2 • 7 tweets • 3 min read
Another very strong headline jobs number of 339,000 in May. Bringing total jobs to 1.3 million above CBO's pre-pandemic forecast.
But the unemployment rate rose from 3.4% to 3.7% which is the 94th percentile of historical one-month changes.
Note that the household measure of jobs fell by 310,000. But the difference from the payroll survey is entirely different definitions, when BLS adjusts to comparable definitions the household survey showed 394,000 jobs added. bls.gov/web/empsit/ces…
May 31 • 7 tweets • 2 min read
Fiscal policy suffers from several biases whose relative magnitudes vary over time:
One tweet for each:
Expansionary bias: Voters and the people they elect have short time horizons and like to benefit today at the expense of the future. This bias is a staple of political economy models and popular discussion.
May 31 • 5 tweets • 2 min read
New JOLTS data tells a contradictory story. The quit rate fell and is now almost back to where it was pre-COVID. In contrast job openings ticked up and remain highly elevated.
Overall there are 1.8 openings for every unemployed person, down slightly from 2.0 peak in Mar-2022.
And here is a graph of job openings per unemployed worker, a metric that has both some a priori theoretical appeal and has been at the center of a range of recent work including Ball, Leigh and Mishra, Bernanke and Blanchard, my work with Willie Powell, etc.