Latest / The Capitalism and Freedom in the Twenty-First Century Podcast / Business Cycles, Growth, Misallocation, and AI with Pete Klenow | Hoover Institution

Transcript

Business Cycles, Growth, Misallocation, and AI with Pete Klenow | Hoover Institution · The Capitalism and Freedom in the Twenty-First Century Podcast

0:09
This is the Capitalism
0:10
and Freedom in the 21st Century Podcast, an official podcast
0:13
of the Hoover Institutions economic policy working group
0:16
where we talk about economics, markets, and public policy.
0:19
I'm Jon Hartley, your host Today my guest is Pete Klenov,
0:23
who's professor of economics at Stanford University in one
0:26
of the world's leading authorities on economic growth.
0:29
He's made really fundamental contributions in, you know,
0:33
areas like misallocation on s, prices on development,
0:37
accounting in many other areas.
0:39
It's a real honor to have you on, Pete.
0:41
I think economic growth is one
0:43
of the most important areas in economics
0:44
and it's amazing to be able to talk to really one
0:48
of the field's leading scholars.
0:51
Well, thank you so much for having me on.
0:53
I'm excited to talk with you.
0:55
Well, before we get into
0:56
what I think is economic's most important question,
0:58
what causes economic growth?
1:00
I wanna get into your background.
1:01
So you grew up in the Bay Area
1:03
and you got your PhD from Stanford.
1:06
I mean, what influenced you
1:07
to become an economist in the first place?
1:10
I would say in the first place, I was totally attracted
1:12
to its relevance to policy.
1:15
In particular things like trade policy,
1:17
but also industrial policy,
1:19
whether we should favor certain sectors and,
1:21
and kind of like what causes economic growth.
1:22
I was already as an undergrad at Berkeley, by the way,
1:26
although I did get my PhD at Stanford at Berkeley,
1:29
I was a business major and,
1:30
and I was very drawn to, you know,
1:32
technological progress generated a lot of growth and,
1:35
and in fact, even from high school,
1:37
there was like a high school project where they wanted us
1:40
to put like by date major political events on one side
1:45
of the, the chronology on the other side, like other events
1:48
that are not necessarily politics.
1:50
And I filled that up with technological innovations.
1:52
So I was always kind of taken with technological innovations
1:55
and their importance for economic growth.
1:57
And so I was motivated even as an undergrad then
2:01
to think about the, that as some
2:02
of the most important questions for economists.
2:05
But I was just very drawn
2:06
to economics from its policy relevance, including levers
2:10
that could affect growth rates,
2:11
but also, you know, all kinds of policies.
2:13
And so it seemed like a powerful paradigm in terms
2:15
of its use of data, but also math and statistics.
2:19
And then it had like a, it had like a public, you know,
2:22
debate and element to it that I, I found appealing as well.
2:26
Very good. And yeah, I mean, it, it, it,
2:31
I, I think the story is similar for me.
2:33
I mean, you know, this idea that, you know,
2:35
people in academia could argue so much over, you know,
2:38
these ideas I always felt sort of attracted me in a sense
2:42
that, you know, these, it must be interesting.
2:44
There must be work to be done if there's
2:45
so much disagreement over all these issues.
2:48
Yeah, that's a great way to put it.
2:49
So, so you did your PhD at, at Stanford
2:52
and you know, at that time, my sense is, you know,
2:55
the big macro folks around at that time, you know, were
2:59
in the department, folks like, you know, Bob Hall,
3:02
folks like John Taylor, folks like even Joe Sticklet
3:05
who sort of does both micro
3:07
and macro theory, I think was around too.
3:09
Who are your big influences then,
3:11
and who sort of influenced you to, to choose growth
3:15
and to choose or to choose macro?
3:16
You do a lot of business cycles,
3:19
business cycle work as well.
3:20
Do you know, going into your Stanford eCom PhD program
3:23
that you definitely wanted to do macro and,
3:25
and coming out of it, you, you felt the same way?
3:28
Or, or what, what sort of led you, I guess,
3:31
in your undergrad years to, or
3:34
or in your PhD to doing macro?
3:36
Yeah, I was always drawn to macro
3:37
'cause they seem like the most important questions,
3:40
but of course if macro didn't exist,
3:41
I was really drawn to economics.
3:43
So I think I would've done another field of economics,
3:45
probably trade, maybe public finance.
3:47
'cause those are very related to policy questions too.
3:50
But I was, had no question coming
3:52
in that I wanted to do macro.
3:53
I was drawn to Stanford partly by Hall and Taylor
3:56
and ended up working with both of them.
3:58
And both of them were appealing partly
4:00
because they, you know, brought in new techniques,
4:04
but at the same time, and, you know,
4:05
were empirical focused at the same
4:07
time that they were model focused.
4:08
But also that, you know, that it was already clear
4:12
that there was these emerging freshwater saltwater
4:14
and they were kind of outside that they were kind
4:16
of doing research that appealed to both the freshwater camp
4:20
and the saltwater camp, which is more
4:21
relevant for business cycles.
4:23
And, and my research is, you know,
4:24
since graduate school has,
4:26
has shifted away from business cycles to more growth.
4:28
But like you said, I I have kept, you know, a hand in,
4:31
in business cycle literature partly
4:33
motivated by Holland Taylor
4:35
Makes perfect sense.
4:37
And yeah, you know, I, I think about that, you know,
4:40
Robert Lucas saying, you know, once you start thinking about
4:42
growth, you can't, you know, think about anything else.
4:44
I know a lot of your early work, you know, certainly related
4:47
to, you know, business cycle seeking prices.
4:49
And is it fair to say that you've moved to,
4:51
to doing more growth over time?
4:53
Is that, would you say that that's fair
4:55
to say about your sort of the trajectory
4:57
of your research career that, so you,
4:59
you've shifted from business more less business cycle work
5:02
to, to more, more growth
5:04
and you know, you think about the, you know, difference in,
5:07
you know, GP per capita between say, you know, the US
5:11
and you know, the Sub-Saharan African country.
5:13
I mean it's, you know, probably, you know, 20 x
5:15
or something like that versus, you know, a say a a one,
5:19
you know, bad business cycle might cause you know,
5:20
a 5% drop in GDP or something like that.
5:23
Was that sort of the, the thing that, is that fair to say
5:26
that you've been increasingly more attracted
5:28
to growth questions as your career's progressed?
5:31
I think that's true. Yeah.
5:32
And I think it's, it's partly what exactly what you said,
5:36
which is, you know, you mentioned Lucas.
5:37
Lucas also kind of famously calculated
5:41
that the welfare costs of business
5:42
cycles were not that large.
5:43
And there's been a lot of research since then finding
5:46
that they're larger than his simple calculation might've
5:49
suggested, you know, things like workers not being able
5:51
to perfectly insure some people thrown out of work
5:53
and recessions and other people keeping their jobs.
5:56
But even taking in that into account, just like you said,
5:59
the, the gaps in income between say the US and and,
6:03
and some countries in Sub-Saharan Africa is enormous
6:05
compared to business cycle fluctuations.
6:07
So just the, the magnitude
6:08
of the economic question involved has, you know,
6:11
that pull kind of pulled me away from,
6:14
from business cycles over time.
6:16
And, and then,
6:17
and then also just the, the rate at which economic growth
6:20
occurs in any country, not just the level differences
6:22
between rich and poor, but just the rate at which growth
6:25
occurs in any given economy, including the United States.
6:27
Is it slowing down? Why is it slowing down?
6:30
Is there something the government
6:32
can and should do about it?
6:34
Those are really powerful questions To me, just the miracle
6:36
of compounding means that, you know, the difference
6:39
between growing, you know, 2% a year,
6:42
we double incomes per every 35 years versus 1% a
6:45
year every 70 years.
6:46
So the stakes are really high
6:49
for average growth in rich economies too.
6:52
So, so I wanna talk about just business cycles for,
6:54
for a quick sec before we get into growth and,
6:56
and what causes it and, and your work
6:58
here on business cycles.
6:59
You, you've made some really seminal contributions on sick
7:02
prices and there's a, a famous 2004 paper that
7:07
you wrote in the JPE with Mark Bills
7:09
and is titled, you know, some evidence on the
7:11
importance OFS prices.
7:12
And I think, you know, up until that time, you know,
7:15
there hadn't been a whole time empirical work done
7:17
on ONS prices.
7:18
And my sense is that you sort of went deeper into, you know,
7:23
the CPI underlying data
7:24
and found that, you know, siki prices were, were,
7:26
or prices were more sticky than previously thought.
7:28
I think ECMO Sisson came after and
7:29
and sort of found that they were more sticky than,
7:32
than even what your evidence suggests.
7:34
But, you know, I I guess at that time, and,
7:37
and correct me if I'm wrong, sort of,
7:39
there's these new Keynesian theorists that sort
7:41
of came about in the eighties and nineties
7:44
and they sort of just put sticky prices in the model in the,
7:47
in real business cycle models
7:49
because, you know, there wasn't really room for policy.
7:51
And that, that's my understanding from sort of the founders
7:54
of, of new canteen modeling.
7:57
And, and then I guess sort of later they sort of, you know,
8:01
empirical economists like yourself sort of dug into the,
8:03
you know, these questions about, well,
8:05
how sick our prices actually
8:06
and could they be sticky enough to actually contribute to,
8:12
you know, business cycles.
8:13
And, and you know, the, I'm curious
8:16
what your thoughts are on that and,
8:17
and like what was going on in the lurch at the time when
8:20
you're writing this paper and, and, and why was it important
8:23
or why has it been important?
8:25
Yeah, I actually wanna step back and,
8:26
and really key on something you said, which is that,
8:29
you know, people were theorizing about mechanisms in
8:33
business cycle models, which is,
8:34
I feel like there's been over last decades micro data
8:38
revolution in macroeconomics.
8:41
And you probably started on the labor side, you know,
8:44
where people were looking at wage data from household
8:46
surveys and consumption data from household surveys.
8:48
Bob Hall did some of that.
8:49
Mark Bills you mentioned earlier, a frequent collaborative
8:52
of mine did a bunch of that work.
8:54
So this was a of that
8:56
because it said, well, first of all, I'm very sympathetic
8:58
with the view that that monetary policy has real effects,
9:02
that it doesn't just flow into inflation immediately.
9:04
That, that it has impacts on things like real interest rates
9:06
and unemployment rates and,
9:08
and real investment consumption for, for several years.
9:11
And the evidence for that, you know, is,
9:13
is basically based on time series.
9:15
So it's hard to, hard for the literature
9:17
to prove definitively,
9:18
but the consensus was certainly that,
9:20
that, that it had real effects.
9:22
And so that's one of the reasons we,
9:23
that the literature wanted to move away from
9:26
the real business cycle models
9:28
where the business cycles were efficient
9:29
and prices were perfectly flexible.
9:31
And there have been studies using micro data,
9:35
but for select products
9:37
or firms like, you know, magazine newsstand prices
9:40
or LL bean catalog prices
9:42
or saltine prices at, in the grocery store.
9:45
So what we, my work
9:46
with Mark Bills did is they were fortunate enough
9:48
to get access to the micro data, the CPI,
9:51
the consumer price index, which covers, you know,
9:54
we use the non housing portion, but that's still 70%.
9:57
So just imagine what kind of coverage that was compared
10:00
to the existing papers that were, that were much narrower.
10:02
And the, and there were surveys like I mentioned earlier,
10:05
where they ask firms how often they change their prices,
10:10
but there's no substitute for looking at
10:11
how much they actually do change their
10:12
prices by looking at prices.
10:14
And so Mark
10:15
and I were able to document for a much broader set of goods,
10:18
basically all goods
10:19
and services outside housing, 70% of out
10:22
of consumer spending, how often prices change
10:25
for new cars versus used cars versus
10:28
close versus grocery store items.
10:30
So it, like you said, it then
10:33
provided some discipline
10:34
to theorizing about why monetary policy is powerful
10:38
and how monetary policy should optimally be set.
10:41
Because if those frictions, if those lack
10:43
of pricing flexibility are the key frictions that mean
10:47
that monetary policy can,
10:49
can move the economy in real terms,
10:50
there are also a reason why we might need them to do that.
10:54
That the market might itself not produce the efficient
10:58
equilibrium because prices, wages aren't perfectly flexible.
11:01
So the economy can have things like under utilization
11:04
of labor where people are unemployed
11:05
who shouldn't really be unemployed.
11:07
It's just a market imperfection that creates that.
11:09
So that's why documenting the extent to which that's true,
11:12
the nature of it is quite important for our understanding
11:15
of business cycles and what monetary
11:17
and fiscal policy should do about business cycles.
11:20
Yeah, you know, I, I'm sort of,
11:23
I've become a bit anti sticky prices recently, like in,
11:26
in the sense that one,
11:29
the reason why they did sticky pricess instead
11:31
of sticky wages, if we go back
11:32
to Kane's in the general theory,
11:34
he actually talked I think a lot more about sticky wages
11:37
than he did about sticky prices. He didn't really talk,
11:39
I'm told I'm not because I've been told that, but yeah.
11:42
And, and you know, you sort of have to read it
11:45
and it, it's, it's not an easy book to read.
11:47
But, but what I think is, is important is
11:50
that the reason why, you know,
11:52
macro theorists chose sticky prices instead
11:55
of sticky wages in their models in the late eighties
11:57
and nineties was that models with sticky prices are easier
12:01
to solve than, than sticky wages
12:02
than than models with sticky wages.
12:04
And so, so one, I think
12:07
what what's interesting now is I sort of have a,
12:10
a paper sort of documenting some of these facts that
12:12
like since the time that you've written this, you know,
12:16
your your famous 2004 paper, I think it's fair to say that
12:19
the prices have become less sticky.
12:22
Retail prices have become less sticky.
12:23
And part of this is, you know, the growth of e-commerce,
12:26
you e-commerce prices, you know, adjust, you know,
12:30
more frequently, we now,
12:31
which in the past couple years had the advent
12:33
of electronic shelf labels, you know, Walmart a very large,
12:38
almost I think half of Walmart stores have,
12:40
will have electronic shelf labels by the end of the year.
12:42
But increasingly, you know, stores
12:46
and retailers are, are using electronic shelf labels
12:48
and they can adjust prices more quickly
12:52
with, with those as well.
12:53
So, you know, it is, you know, we've now in the US gone from
12:57
around like almost 0% e-commerce at the beginning
13:00
of like the 21st century, around 2000 to I think
13:04
around like 25% of all retail goods
13:06
or so at least 20% are, are all e-commerce in China.
13:10
I think it's like 50%.
13:12
But I mean, is it fair to say that, you know, maybe 60 wages
13:15
or informational rigidities might might be more important
13:17
than, than price rigidities?
13:18
I mean, than, I mean it's, it'd be interesting to, you know,
13:22
if if it's the case that you know, wage if prices become
13:25
much more flexible or almost perfectly flexible
13:27
with online prices and it's not perfectly flexible
13:32
'cause there's still some sort of labor decisions involved
13:35
with changing these prices.
13:36
But if we still got recessions
13:38
and prices, you know, were much more flexible.
13:41
I I guess it might sort of dismiss I guess some
13:44
of the importance of, of, of,
13:45
of sticky prices for macro models.
13:47
If that's, that's true.
13:48
I don't know if you, you if, if you spend
13:49
Much About this
13:50
One, one quick thing about whether prices have become more
13:53
flexible is that it's, I think one study
13:57
for example found that it's true within product categories
14:00
that they become more flexible, just as you've said.
14:02
And some of this evidence isn't as recent as some
14:04
of the trends you, you were mentioning, you were saying,
14:07
you know, that that Walmart becoming, you know, having half
14:09
of their products be, you know, the, the,
14:12
the price is electronic label on the, on the shelf.
14:16
That might be more recent than the study
14:18
I'm about to refer to.
14:19
But the study about to refer to found
14:20
that there was an offsetting effect from shifting toward
14:24
services that services have become more
14:25
important and they're stickier.
14:27
So that's a force making prices more sticky in the economy,
14:30
just a compositional change in the, in the components
14:33
of spending that's tended to work against the,
14:36
within sectors becoming more flexible.
14:39
So that's relevant,
14:40
but again, that might not incorporate the most recent
14:43
evidence that you're referring to.
14:45
And then the other thing I was gonna say is I've always had
14:47
this prior that wages were more sticky than prices.
14:51
That that kind of stylized view
14:53
that you change price once a year,
14:55
that's kind of fallen away.
14:56
That the evidence is that prices change more
14:58
frequently than once a year.
15:00
There's still this active debate about
15:01
how you think about temporary price discounts.
15:04
I'm mostly persuaded that they're not used very much
15:08
as a form of price flexibility.
15:10
But there's, you know, one interesting study about u the UK
15:14
finding that the frequency of sales went up
15:17
in the COVID recession in the UK worked
15:20
by Alexi Kristoff is at the Bank of Canada,
15:23
the co-author of mine.
15:25
So, so I don't think it's completely, you know, unrelated
15:29
that temporary price discounts are purely
15:31
for price discrimination, have nothing to do
15:33
with business cycles, but I think that's largely true.
15:36
And so that, so, but,
15:40
but that means the prices are are
15:42
more sticky than just like the posted price
15:44
that includes the temporary price discounts.
15:46
When you filter those out,
15:47
you get some stickiness be somewhere
15:49
between six months and a year.
15:51
But for wages they seem to change once a year
15:53
or sometimes even less frequently than that.
15:55
And there's been work about whether bonuses are, are a form
15:58
of flexibility, but it looks like they're largely a
16:01
cyclical, they're largely for performance reasons
16:03
and not like very sensitive to the business cycle.
16:06
Somewhat surprising 'cause you'd think the profitability
16:08
of the firm would be very cyclical and
16:10
therefore the bonuses would be very cyclical.
16:12
But it doesn't appear to be really a genuine
16:15
form of wage flexibility.
16:16
So my sense is that wages are stickier than prices,
16:19
which also matters for things like our wage change.
16:22
The leading indicator for inflation,
16:24
should you be looking at wage changes, expecting them
16:27
to change first and prices to change.
16:28
Second might be the other way
16:30
around it might be prices adjust faster
16:33
and then wages catch up.
16:34
But again, the part of the literature
16:38
has grown in understanding not only getting micro data
16:41
better and better on how much prices wages are changing
16:43
along with prices, but also just like thinking about the
16:46
interactions between those two.
16:48
'cause the firm's wages haven't changed,
16:49
then they have less incentive to change their prices.
16:52
So they, the wage stickiness feeds into price stickiness.
16:56
Makes perfect sense that, that's a fascinating way, way
17:00
to think about, I, I wanna just shift into economic growth
17:03
'cause you know, this is so important, you know,
17:04
why are countries rich and poor, you know,
17:07
the differences in, you know, living standard
17:10
between you advanced economies versus, you know,
17:13
developing frontier ones that, you know, it's so stark.
17:17
I wanna just like, you know, walk through, you know,
17:19
growth theory and,
17:21
and you know, get into, you know, misallocation, you know,
17:23
and I guess a quick, a very short theory
17:25
or a very short history of, of growth theory, you know, sort
17:28
of starts with, you know, mouth this, you know,
17:32
is there arithmetic growth or, or geometric growth and,
17:35
and are these, are there Malthusian trapps?
17:37
And I think there, there's evidence that this was sort
17:39
of true before the, you know, 1,617 hundreds.
17:42
And then, you know, what happened was sort
17:45
of modern growth came along and,
17:46
and we started seeing, you know, geometric rate in,
17:50
in a very serious, meaningful way, you know,
17:53
starting in the 16, 17 hundreds in the UK and, and, and,
17:56
and other countries, you know, thereafter.
18:00
So, you know, one theory is, you know, there's, you know,
18:05
when you have, you know, Bob Solo comes around
18:07
and you have, you know, the solar growth growth model
18:10
and you have you sort of capital labor
18:12
and obviously, you know, technology, you know, your TFP
18:17
and, and then you sort
18:18
of have the sort of successors of that.
18:20
You have endogenous growth, you know, this idea that
18:23
you know, that that ideas, you know, flow between countries
18:27
and that at some level, you know, the, your stock of ideas
18:31
or your a your technologies in part depend on, on, you know,
18:33
the number of people that you have and,
18:35
and with more people you'll get more, you'll get more,
18:39
you know, ideas and, and more growth.
18:42
But then it turns out that, you know, at some level,
18:44
you know, we've had ideas spreading for long periods
18:47
of time now, but for, you know, clearly we still have mass,
18:52
you know, massive living differentials.
18:53
And I think since endogenous growth theory, you know, a lot
18:56
of theorists have tried to, and,
18:59
and growth accounts have tried to explain why is it
19:01
that some countries are still rich
19:02
and poor, even though, you know, the,
19:04
the great ideas coming outta Stanford today are still,
19:06
you know, accessible, say over the internet in in places
19:09
like, you know, Africa, which are,
19:10
you know, still very poor.
19:12
So here's where we get in, in a lot of the past sort
19:14
of 20 years, I think in, in growth is really focused on like
19:17
in, you know, various explanations of, you know,
19:20
largely political economy related, you've got, you know,
19:22
institutions, you've got culture,
19:26
you've got geography, I, I'd like, you know, institutions
19:30
to some degree or, you know, different ways of putting
19:32
that, you know, maybe economic freedom.
19:33
But you, you have a, a really amazing,
19:37
you know, take on on this.
19:38
And first I wanna highlight some
19:40
of your work on development accounting
19:42
and sort of trying to expl one, I want you to sort
19:45
of explain what development accounting really is
19:48
and why that's helpful to understanding this issue of,
19:51
you know, what's actually causing growth.
19:53
And then I wanna really talk about misallocation,
19:55
which is I think your sort of really semi contribution and,
19:59
and I would argue, you know, a Nobel Prize winning won or,
20:02
or, or potentially Nobel Prize winning won, you know,
20:05
what exactly misallocation means and, and,
20:09
and how this sort
20:10
of misallocation framework explains why countries are
20:13
so rich and poor, even though, you know, ideas can, can move
20:15
around the world in the way
20:17
that endogenous growth theory says,
20:19
says it should be, you know, causing growth.
20:21
But somehow empirically we don't find this.
20:23
I'm, I'm just curious, so development, accounting, walk us
20:26
through your contributions there.
20:29
Okay, so we're very motivated by some calculations.
20:32
We, meaning me and Andreas Rodriguez, Claire
20:34
and some of the work and Mark Bills
20:36
again, and some of the other work.
20:38
And the, we were very motivated by this seminal paper
20:42
by Manky Greg Manky, David Romer
20:46
and David Weill.
20:48
And they just had this beautiful calculation
20:51
where they said, we can explain roughly, you know, 80
20:55
to a hundred percent of income differences using just
20:58
differences in physical capital, meaning like equipment
21:00
and structures plus human capital.
21:03
Some measure, some proxy for human capital that was based on
21:06
schooling attainment, in particular
21:08
secondary schooling attainment.
21:09
And they said, look, this makes perfect sense
21:12
because ideas technology, if it's disembodied,
21:15
it can flow if it's embodied in equipment that can flow
21:18
that's tradable and can flow across countries.
21:21
So the thing that really can't flow is people,
21:23
and if people in a certain country lack human capital,
21:26
that's gonna lower their, their income.
21:29
So they, that doesn't explain why poor countries might have
21:32
less physical capital,
21:33
but it would explain why they, why they're poor,
21:37
despite the fact that ideas could flow.
21:39
So they, you know, kind of a simplistic version of
21:41
what they said was that ideas
21:43
and technology are the same everywhere
21:45
and that's generating growth that's common everywhere.
21:48
But that all these differences in the levels
21:50
of development are due, due
21:51
to differences in mostly human capital.
21:54
So part of our reaction to that was just, you know,
21:58
how exactly they used schooling attainment pri,
22:02
secondary schooling attainment is a proxy.
22:04
We're like, well what about primary schools
22:06
and what, you know, some of these countries,
22:07
people are only going to primary schools
22:08
or a large fraction of the population depending
22:10
on what year you're looking at.
22:12
So that seemed like an omission.
22:14
So we want, and of course, you know, high school
22:16
and college, you know, all of the, all three levels.
22:18
And so we took all those into account,
22:20
but then the key thing we did is say,
22:22
rather than just assume human capital is proportional to
22:26
how much secondary school you get, instead,
22:28
there was a large literature in labor economics that tried
22:31
to say what was one year
22:32
of educational attainment worth in terms of higher wages.
22:36
And if one year
22:37
of education is worth about 10% higher wages,
22:40
that suggests if a whole country got one year more
22:42
of schooling, average wages
22:44
and average incomes in the country might grow about 10%.
22:47
So putting that discipline on it turned out
22:50
to mean a much smaller number, much smaller contribution
22:52
for human capital
22:54
and a much bigger kind of residual that might reflect,
22:58
you know, other stuff like technology differences
23:00
or misallocation of, of resources,
23:02
which we're gonna talk about in a bit.
23:04
So it opened up the door that that maybe half
23:07
of income differences between rich
23:08
and poor countries might reflect something other than
23:11
physical and human capital.
23:12
Although having said that, some
23:14
of the research since we wrote that has said, Hey,
23:17
that we'd looked at quantity of schooling, let's try
23:19
to incorporate quality of schooling.
23:21
So they get a bigger human capital number from that.
23:23
They also, this is work, a lot of it's done by David Lakos
23:27
and Todd Schulman,
23:29
but other people are involved in the work.
23:31
Nancy Chen, Ben Ma, Tomaso pio.
23:34
And one of the things they're incorporating is
23:36
that people don't just learn in school,
23:37
they also learn on the job.
23:39
And there might be learning a lot more on the job in rich
23:41
countries and in developing countries.
23:43
That one I think is a great insight
23:46
and great evidence they brought to bear.
23:47
But I'd also like to see more, more fleshing this out
23:50
because like how much you learn in the job could also be a
23:53
function of the technology.
23:54
Like if the technology's advanced,
23:55
you got a lot more stuff to learn.
23:57
If it's simplistic, you can learn it in much faster time.
24:00
So it's separating out learning on the job from technology
24:03
is not so obvious to me.
24:05
But anyway, that's, that's kind of my view on like
24:08
what development accounting is, what our current thinking is
24:11
and the relative importance.
24:12
But let me stress why this matters
24:15
because one reaction more applied micro types
24:18
or even, you know, average humans,
24:20
not just applied micro economist,
24:22
but any, any average person could say, well
24:24
what does it matter?
24:25
You know, what fraction of income differences do are due
24:29
to technology versus things like education versus
24:32
experience on the job?
24:33
What does that matter? What does it,
24:34
what is this useful for?
24:36
And I think the, the answer is that it's helps us prioritize
24:41
more research, including the applied micro studies
24:44
that look at individual interventions related
24:46
to agricultural technology versus health versus education
24:50
versus technology more generally.
24:52
So how much research effort we should put into those
24:56
different areas should be a function of how important those
25:01
proximate factors are
25:02
for income differences across countries.
25:04
To me it's the same as as saying we wanna know
25:07
what people die from
25:08
and that helps prioritize what we should be doing, you know,
25:12
you know, making, doing research for health improvements on
25:16
Absolutely.
25:17
I mean, and it, it's so, so critical and so important.
25:20
I mean, you know, I think back to the, you know,
25:22
the thank room wallpaper
25:24
and I mean sure, you know, you could, we could argue,
25:27
you know, it's like we don't quite have the kind of rigor
25:31
of, you know, causal, causal inference maybe today.
25:33
But the idea that you can sort of, you know,
25:35
understand like, well, you know,
25:37
a country has this many people, you know, this is
25:39
what the output is, you know, per person, you know, the,
25:42
you know, country has this much in the stock of sort
25:44
of machines, machinery, you know, and,
25:48
and what, you know, what's sort of the what, what's left,
25:52
what's missing there, you know, the TFP is, you know,
25:53
the sort of the residual side of, you know, this is the a
25:58
or you know, and I think one economists famously said,
26:01
it's sort of a measure of our ignorance
26:02
that we don't really fully understand what, you know.
26:04
Yeah. EFP really is.
26:05
And I think that's a great way of, of putting it.
26:07
But I think just even in an attempt, like, you know,
26:09
that paper and success of papers like yours trying
26:12
to understand, you know, or human capital, you know,
26:15
education, you know,
26:16
what are education levels across country, trying, trying
26:18
to understand across countries, you know, how, you know,
26:21
what are factor allocations
26:23
and how does it line up with, with, you know, output.
26:26
It's, it's really amazing.
26:27
I think this, you know, it's a nice segue into, you know,
26:30
talking about misallocation.
26:32
So, you know, misallocation and manufacturing TFP in China
26:35
and India as I you're most side of paper.
26:38
And, and this is where the, the whole
26:40
shift in the field towards saying misallocation comes,
26:43
it's really a pathbreaking paper for, for those
26:45
that aren't aware, you know, the, the idea, you know,
26:50
and you please explain it better than I'm sure you can
26:52
explain it far better than I can, you know,
26:53
I think my understanding is the idea is that, you know,
26:56
output is lost because, you know,
26:58
factors aren't being efficiently allocated across firms.
27:01
And you study China and India and, and,
27:04
and firms in both these countries and,
27:06
and essentially argue that, you know, if resources were,
27:10
you know, allocated across, you know, factors as efficiently
27:14
as, as they are in the US then you know, your TFP and,
27:18
and your growth rates would be maybe like, you know,
27:20
40% higher in China or 50% higher in India.
27:25
Obviously there's assumptions, you know, about
27:27
what the production function is
27:29
for firms in these countries, but I want you to explain it
27:32
'cause I'm sure you can explain it
27:33
better, better than I can.
27:35
I know there's a lot about, you know, small firms
27:37
and their productivity in, in these countries,
27:39
but really, you know, it's pathbreaking work by yourself
27:42
and Chiang, she and,
27:43
and really want you to explain it in,
27:46
in your words and we can
27:48
Okay.
27:49
Talk a little bit about what might be the sources
27:49
of misallocation and other things like that.
27:52
Yeah, let, let me say one background thing.
27:54
Before going into how we thought about misallocation,
27:58
what motivated us to look at the background thing is that,
28:02
you know, if you said 50%
28:04
of differences in income were not due to physical
28:06
and human capital there,
28:08
that other 50% is a measure of ignorance.
28:10
We want to figure out what what's causing that.
28:12
And one challenge to a technology story is,
28:16
like I mentioned, you can import the equipment, you can,
28:19
you can have foreign firms if they, if they know how
28:22
to use the technology, they can operate in your country.
28:24
Now there's barriers to all of those things.
28:26
There's, there's tariffs that,
28:27
that restrict equipment imports, there's
28:30
restrictions on FDI, sometimes it's legal restrictions,
28:32
sometimes it's just the cost of adapting
28:34
to a different environment.
28:36
But then there's also potentially complementarity
28:41
between human capital and technology.
28:42
Like your willingness to bring in technology might be a
28:44
function of whether the workers have the skills
28:47
to use that technology.
28:48
So you could have this coordination failure
28:50
where people aren't acquiring skill
28:51
because there's not a lot of demand
28:53
for the skill in the local economy.
28:54
'cause the technology is not very skill intensive
28:57
and that there's not an incentive to introduce a lot
29:00
of skill intensive technology
29:01
because skill is relatively scarce in a developing country.
29:04
So anyway, so that's kind of a broad thing about the role
29:07
of technology and in, in development accounting.
29:09
But still people have pointed out, like David Weill
29:12
and other people pointed out, if you, if 50%
29:15
of income differences were due to technology,
29:17
then some countries would have to be
29:18
centuries behind on average.
29:20
Now you can look at individual, you know, subsistence farms
29:24
and say they're a century behind or something,
29:26
but like for the whole economy, it just, you know,
29:29
casual evidence would say they're not,
29:31
they're not using technologies from a century
29:33
ago or, or older.
29:35
So that kind of left a challenge of like
29:38
what else could contribute
29:39
to these differences in income across countries
29:41
that aren't trivial directly to average human capital
29:44
and physical capital or technology being used by workers.
29:47
And there was actually a really nice paper by Diego Ian
29:50
and Richard Rogerson that did just a calculation,
29:53
an illustrative calculation that said,
29:55
imagine there were three types of firms,
29:58
big efficient firms, medium sized firms,
30:00
and then inefficient small firms.
30:02
And they're small because they're inefficient
30:04
and imagine you tax the big firms
30:06
and subsidize the small firms.
30:07
And they had in mind lots of different policies
30:11
or market failures that might have this flavor.
30:13
And they calculated that
30:15
this simple calculation like a a 20% tax on the big big
30:19
and a 30% subsidy in the small could lower aggregate TFP
30:23
20%, 25%, 30%.
30:25
And I was stunned by this. I was like, really?
30:28
I, we thought kind of misallocation across firms
30:32
or even across sectors was relatively small on the order
30:35
of 1% of GDP loss that you would get from that.
30:38
So their calculations stunned me.
30:40
So that made me say, well let's look in the data ang, Tisha,
30:44
you mentioned the collaborator on this.
30:45
We said, let's look in the data
30:47
and see if there's, there's evidence consistent with a lot
30:49
of misallocation of resources.
30:51
And we were also mentioned, you,
30:52
you you mentioned small firms.
30:53
We were also motivated by the fact
30:56
that firms were much smaller in
30:57
India than the United States.
30:59
We was like, maybe that's part
31:01
of why India's relatively poor.
31:02
There's too many small unproductive firms
31:04
and not enough of these world class larger firms.
31:07
And the evidence certainly goes that way, that there's a lot
31:10
of small unproductive firms in India compared to the US
31:14
but there's still the, the issue
31:17
of whether those firms are the optimal size,
31:20
like maybe there's no misallocation, there's just a bunch
31:23
of workers who lacked the skill to work at bigger firms,
31:27
but they're efficiently allocated.
31:28
So it's really still just a human capital question.
31:31
So where Chang and I came in was like, well,
31:34
does it look like when they hire a worker of the same skill,
31:38
does it look like they get less output out
31:40
of those workers even controlling
31:41
for skill at smaller firms than bigger firms?
31:44
Or just more generally, you know,
31:45
do some firms get a lot more output per per unit
31:49
of input than other firms?
31:51
And if so, that could generate lower aggregate output.
31:56
'cause you're not efficiently allocating
31:57
inputs across firms.
31:59
If the return to, to the same worker is high at a big firm
32:03
and low at a small firm, then by just reallocating
32:05
that worker, you don't change
32:07
inputs at all in the whole economy.
32:08
The worker is still there before and still there
32:10
after the reallocation, but aggregate output goes up.
32:13
And so we were motivated by the rest
32:16
and rogerson calculation to say, does it look like there's,
32:19
as if there's taxes on big firms or barriers, big firms face
32:23
and subsidies on small firms
32:25
or advantages small firms get in terms of less regulation,
32:28
less taxes and so on That or,
32:31
or it could be market failures like, like maybe the,
32:34
the big firms are starved for credit.
32:36
You might think big firms are, are well financed
32:38
and have no problem getting credit,
32:39
but not the necessarily the fast growing ones.
32:41
So some of the moderate sized ones might be, you know,
32:44
have great opportunities to invest,
32:46
but not a great way to get external financing so that they,
32:50
so that they actually are starved for capital and, and don't
32:54
and have a high marginal return to capital,
32:56
but they can't get access to the capital.
32:58
Whereas the small firms might not have a
32:59
lot of great opportunities.
33:00
So they might actually have good access to capital relative
33:03
to the opportunities they have.
33:05
So it could be market failures,
33:06
like the financial system doesn't
33:08
perfectly allocate capital.
33:09
And maybe that's a bigger deal in
33:11
places like India and China.
33:12
China you have things like state owned banks
33:14
that are deliberately steering the allocation of capital.
33:18
Now that allocation could be efficient.
33:20
They could be like have this great x-ray vision into
33:23
what the, the highest rate of return is
33:25
and they're better than the private sector
33:27
of the private banks at lending money.
33:29
But the evidence doesn't support that.
33:31
Its supports that they tilt resources towards state owned
33:34
enterprises that might not be using it more efficiently.
33:37
They might have political motives for some of the allocation
33:39
of capital that they pressure state owned banks to do.
33:41
And we were just pointing out that
33:43
that has efficiency costs.
33:45
'cause if you're gonna funnel capital
33:47
to a lower return investment project, then it's then
33:49
that's gonna lower output per per unit of capital.
33:52
And the same thing with human capital.
33:54
If you encourage state owned enterprises to hire a bunch
33:56
of workers for political reasons,
33:58
that's gonna lower the value of what output
34:01
that they produce relative to the private firms
34:04
that are more productive, that are starved if you will have
34:06
to compete too much for those workers
34:09
because those, those state owned enterprises receive
34:11
subsidies from the government to continue
34:13
to employ lots of workers.
34:15
So that would drag down aggregate productivity in places
34:18
like China where there's big role for state owned banks
34:22
and state owned enterprises more generally.
34:23
And then in India where you do still have a big role
34:25
for state owned banks, arguably misal allocating capital,
34:29
the prevalence
34:31
of low productivity small firms in India opens the door
34:35
to the idea that that misallocation that they,
34:38
they have too many resources, too much capital
34:40
and too much labor is that these unproductive small firms
34:43
and if it were freed up, they could produce more output
34:45
with the same workers of the same skill level.
34:47
That's really the open question.
34:49
And so what Chang
34:50
and I calculated using this micro data on manufacturing
34:53
firms in India
34:55
and compared to the US was
34:56
that there was much more dispersion in revenue
34:59
per unit of input.
35:00
You can think of it as profit rates across firms
35:03
that suggested that the small firms especially may have been
35:08
operating in, in, in persisting because of tax
35:11
and regulatory benefits or market failures.
35:13
Maybe the big firms have higher revenue per unit of input
35:16
'cause they have higher markups price markups or markdowns.
35:19
So it could be market power and market failures
35:21
or it could be government interventions
35:24
and some of which by the way the go, some
35:25
of the government interventions may be to reduce inequality.
35:28
But we were pointing out they, they appeared to have
35:31
or at least consistent
35:32
with those having major impact on the ability
35:34
to even just produce a lot of output per year of input.
35:37
So something holding back India
35:39
and China in an important way could be misallocation
35:42
by state owned banks, tax
35:44
and regulatory policies that favor small
35:46
and formal firms against bigger formal firms.
35:49
And also things like state owned enterprises in China
35:52
that get special subsidies and so persist
35:54
and operate with too much capital and too much labor.
35:57
They'll be better reallocated to the private sector
35:59
that's more efficient with it.
36:00
Is it possible that, you know, things like regulation?
36:04
You know, I think about about India.
36:07
I have a, a study
36:09
or a paper that looks across countries and and run.
36:12
We ran a bunch of surveys across countries trying
36:14
to figure out what exactly the occupational licensing
36:17
prevalence was across countries.
36:18
And we found out that India was actually one
36:19
of the most licensed countries.
36:22
About 40% of its over 40% of its labor force as a license
36:28
compared to say, you know, 20, around 25% or so for the US
36:33
and developed countries.
36:35
A lot of that I think has to do with the, you know,
36:37
so-called licensing RA in India.
36:39
I'm just curious, you know, what some of these, you know,
36:41
frictions may be, you know, behind misallocation, you know,
36:45
why is it that you know that there isn't, you know,
36:49
a better allocation of of factors, you know, is it,
36:52
is it possible that labor market regulation, you know,
36:55
zoning other, you know, is it possible
36:59
that government policies are, are causing misallocation?
37:03
I'm, I'm curious, how do you think about sort
37:05
of the causes of misallocation?
37:07
I think for sure government
37:08
policies are contributing to it.
37:10
So some that I mentioned directly
37:12
but are state owned enterprises,
37:14
which are prevalent in China, much less important in India,
37:18
but still not trivial.
37:19
Then there's state owned banks
37:21
that are steering the allocation of physical capital
37:23
for non-market, not rate of return reasons, which again,
37:26
could be benevolent and could be wonderful,
37:28
but the evidence that that exists mostly suggests that it's
37:32
for political reasons that they, they misallocate it
37:36
and endure the output consequences of that
37:38
for maybe for political reasons.
37:40
But other ways in which the government is involved is,
37:42
like I mentioned, that, that it's hard for them
37:44
to enforce taxes
37:45
and they choose not maybe to enforce some taxes
37:48
and regulations on small firms.
37:49
They exempt them from it
37:50
and they apply things like you mentioned the functioning in
37:53
the labor market long before our study.
37:56
Richard Rogerson, again, who I mentioned before,
37:58
but also Hugo Oppenheim had an important paper describing
38:01
what firing restrictions do to the allocation
38:04
of labor across firms.
38:05
'cause obviously we're kind of sympathetic, you know,
38:07
it's tough to be thrown out of work,
38:09
to be fired at your job,
38:10
but knowing that the firm, the firm knowing that it has
38:13
to pay a big severance package when it lays off a worker
38:16
or it has to get government permission in the case of,
38:18
of India in certain years in certain sectors discourages
38:22
'em from hiring people.
38:23
And to begin with, because one
38:24
of the things we've work learned from the work
38:26
by John Halter Ringer and others
38:27
that have used census data in the US
38:29
and other countries, there's just a
38:31
tremendous amount of churn.
38:33
The market is constantly reallocating inputs across firms.
38:37
And so if you throw sand in those gears,
38:39
then it's just not gonna reallocate as efficiently.
38:42
And so a well-meaning policy, which is to kind of
38:45
shield workers for some of this churning
38:47
and disruption can have efficiency effects
38:50
because it means firms are
38:52
that are productive aren't even bothering
38:54
to hire their workers as a precaution against
38:56
having to fire them later.
38:57
And that means they're, they stay behind at another firm
39:00
that might also already have them
39:01
and keep them just to avoid the firing costs.
39:03
So it's, it's generating misallocation this this policy to
39:07
put limits on firm's ability to, to fire workers.
39:11
So there's other government policies you could talk about.
39:14
There's differences in availability, let's say India,
39:18
differences in availability of electricity.
39:20
Some people are getting it for free,
39:22
some people are paying full price and then water.
39:25
There's water rights and certain agricultural sectors
39:28
where there's water rights
39:29
and there's not a, a well-functioning
39:31
market to reallocate those.
39:32
Imagine you're a very efficient farmer, you would like
39:35
to expand, but you can't
39:36
because you don't have the permits for the water or land.
39:39
If the land markets,
39:41
if the government doesn't enforce land contracts very well
39:44
or puts restrictions on ability to sell land or to,
39:47
or to even rent out your land in certain sectors, then
39:50
that's gonna cause the misallocation of labor, I'm sorry,
39:53
land across farms, across manufacturing firms.
39:57
You know, when you talk to these manufacturing firms
39:59
that are very efficient and productive and profitable
40:02
and say, why don't you expand?
40:03
They often say things like, well, we'd have
40:05
to get the permits for acquiring land
40:09
or we'd have to hire more workers.
40:11
And that's difficult to do in this environment
40:14
where there's firing restrictions and you know, or,
40:18
or access to electricity.
40:20
So all of these government
40:22
policies could be contributing to it.
40:23
And you can step back and more broadly say things like lack
40:26
of infrastructure, lack of, you know,
40:28
like if the roads are too crowded or there's,
40:31
or interstate taxes for that matter,
40:33
that might discourage trade across states within India,
40:36
but also underdeveloped infrastructure, which is less
40:39
of an issue in China, that is in India.
40:41
That could mean that efficient firms don't expand as much
40:44
because they can't sell far away,
40:45
they can't sell the far away markets
40:47
because it's too costly to ship long distances
40:50
because of the state of the infrastructure.
40:52
So government decisions about the quality
40:55
and quantity of infrastructure can also
40:56
affect misallocation.
40:59
And then there's, you, you ask what what causes,
41:02
there's also market failures, though.
41:03
It's not just government steering capital,
41:05
but the fact that the private market might not,
41:10
you know, if they know there's, it's, it's hard
41:12
to observe the quality of firms in a certain sector.
41:15
They're going to naturally wanna lend less to that sector E
41:18
but the the return might actually be high.
41:20
They just can't tell which firms in
41:22
that sector are the best ones to allocate to.
41:25
So that, that ends up under providing
41:27
capital to that sector.
41:28
And there's potentially a role
41:29
for the government to improve on that.
41:31
So there's lots of potential market failures.
41:33
I'm, I personally think there's lots
41:34
of pricing power out there related
41:36
to my sticky price research is
41:38
that firms are setting their prices
41:40
and keeping them stable for long periods of time.
41:42
That's not consistent with a flex.
41:44
You don't see that in commodity markets.
41:46
You don't see that in well organized markets
41:47
that prices are stuck at some level.
41:50
So I think that's a byproduct of market power
41:52
that prices are sticky
41:53
and price stickiness to some extent is pervasive.
41:56
And same thing with wage stickiness.
41:57
So I, I think there's labor market power
42:00
and I think there's product market power that's pervasive
42:02
and, and it looks like it might play an important role in
42:07
dragging down productivity
42:08
because again, in the United States, there might be
42:12
market plenty of market power and plenty of monopsony power,
42:15
but it seems to be less dispersed across firms.
42:17
So there's evidence that we have
42:19
that there's the differences across firms are not as stark
42:21
and that's what's leading to misallocation.
42:23
If all the firms are raising their markup themselves,
42:26
that has distributional implications
42:27
that might increase the profit share
42:29
and reduce the share of income going to labor.
42:31
Similarly with monopsony power,
42:33
if the firms are getting more
42:34
and the workers less, that has really important
42:36
distribution implications.
42:37
But the efficiency of the allocation
42:39
of workers across firms is related
42:41
to whether some firm has more ity power than other,
42:44
or some firm has more monopoly power than other,
42:47
then the firms with a lot of monopsony and, and,
42:51
and monopoly power are producing too little
42:53
and hiring two fee workers with too little capital.
42:56
So that's what leads to misallocation.
42:59
Got it. So I, I guess I wanna talk about just
43:02
one li you know, potential limitation to, you know,
43:05
development accounting misallocation in general.
43:07
And that is, you know, what,
43:09
what comms would call mis specification.
43:11
You know, a lot of this, you know, measuring misallocation
43:14
or development accounting in general assumes
43:16
that we understand
43:17
or know what the, the pro production function is, you know,
43:21
an aggregate production function for a country
43:24
or you know, with a production function
43:26
for a firm looks like.
43:28
And you know, production function for those
43:30
that aren't aware kind of assumes that, you know,
43:33
output is a, you know, a function of, you know,
43:36
various factors like, you know, capital or labor
43:40
or you know, human capital or, or,
43:43
or land and, and so forth.
43:45
And that there's also some like technology augmenting,
43:48
you know, thing as well, we, you know, usually use a
43:52
or TFP technological pack of progress,
43:55
but you know, we're, we're often, you know, economists are,
43:57
are making assumptions, but what that, you know,
44:00
function is, and, and, and misallocation
44:04
and measurements of it really,
44:05
and all development accounting really
44:08
rests pretty heavily on this.
44:10
I'm curious, you know, you know, is that a challenge?
44:13
I mean, how, how do you understand, you know,
44:15
what the production function really is in real life?
44:18
Yeah, it's definitely a challenge.
44:19
So as you say,
44:21
if we wanna know whether firms using too little labor,
44:24
say relative to capital, like mixed distortions,
44:26
lap capital versus labor versus intermediate goods they
44:29
buy from other firms.
44:31
Not enough to say some firms use more capital relative
44:34
to labor than other firms
44:35
because that cap firm might be using more
44:37
capital intensive technology.
44:39
There might be nothing inefficient about that.
44:41
I remember Dave Donaldson, who's a great economist at,
44:43
at MIT kind of throwing down the gauntlet to me
44:46
and saying, how can you say there's misallocation?
44:48
I mean, we don't say people in Japan eat more rice than
44:51
people in the United States, therefore they're eating too
44:53
much or we are eating too little.
44:55
There could be just differences in preferences.
44:57
So similarly, firms just could have different technologies,
44:59
just like in consumers have different preferences.
45:01
We don't say you're consuming too little
45:03
or too much necessarily.
45:04
We, you know, there's obviously health considerations we
45:06
could bring in, but in information.
45:07
But, but he was basically saying,
45:09
how do you know this isn't all efficient?
45:11
And I have one kind of broad response
45:14
and then one more detailed response.
45:15
The broad response is there is that that matters
45:19
for these mixed distortions,
45:21
how much capital relative to other inputs.
45:23
But a lot of the dispersion in, in a lot
45:27
of the misallocation looks like it's due
45:28
to differences in revenue over total costs.
45:31
So if you can choose your mix of costs,
45:32
how much is intermediates, how much is capital
45:34
and labor completely freely and completely efficiently,
45:37
and there's no other distortion in the economy,
45:39
then your revenue per unit of input should be the same.
45:42
It's like the market should equate the rate of return.
45:44
And so profit rates should be the same.
45:46
So the fact that revenue per unit
45:48
of input costs is the same, we don't need to do anything to,
45:51
to weight them, but just look at your choices.
45:53
So that's kind of like assuming your choices
45:56
of your input bundle,
45:57
your input mix is completely efficient.
46:00
So even if we assume that we still see lots of evidence
46:02
for misallocation, in fact the bulk of it seems to be, well,
46:05
it looks like a scale distortion rather than a mixed
46:07
distortion that some firms have too little capital labor
46:10
and intermediates and other firms have too much
46:13
capital, labor and intermediates.
46:15
And the more detailed comment to kind
46:16
of push back a little bit on whether we need
46:19
to know whether there's heterogeneity in, in,
46:23
in production technology
46:25
that could rationalize the input choices firms make.
46:28
One thing that some studies have found is
46:30
that there's difference in prices.
46:31
I mentioned difference in prices for electricity
46:34
and water in India,
46:36
but there's also, you know,
46:37
California for that matter, right?
46:39
There's a nice paper by Han Na
46:40
and co-authors about misallocation of water,
46:43
which looks like it's especially problematic in India
46:46
and California.
46:47
And part of that you can just see in the fact that they're,
46:50
you could just say, oh no, these are just technology
46:52
differences, that's why they use a lot of water.
46:54
But they say, look, they have price differences,
46:56
they get cheap access to water relative to say households
46:59
or relative to some other firms.
47:01
I don't wanna make enemies among almond growers in the,
47:03
in California right at this moment.
47:05
But the point is, if we see
47:06
that some firms get preferential access
47:08
to inputs at cheaper prices, then that's not just, hey,
47:12
they have a different technology, they face the same prices
47:14
and made different choices so it's all efficient.
47:17
Instead it's like, oh, the market isn't equating the cost
47:19
of this resource that should be the same
47:21
across these buyers.
47:22
It's either price discriminating
47:24
or they have preferential access for some other reason.
47:27
So that's evidence consistent with input misallocation
47:30
that doesn't require us to know exactly
47:33
what the production structure of a firm is.
47:35
Asin that, that's a really a great point.
47:39
I wanna just sort of segue into sort
47:42
of talking about development as a field.
47:44
I feel like recently there's been a big debate on Twitter
47:46
and elsewhere about micro development versus
47:49
macro development.
47:50
Okay. And I feel like you're a, a a, you know,
47:53
big representative of macro development.
47:55
And interestingly, I think over the past, you know,
47:58
I think fair to say, over the past 20, 25 years,
48:01
micro development has been very, very dominant.
48:05
You know, randomized controlled trials, you know, the,
48:08
the rise of those, you know,
48:09
several Nobel prizes being given, you know, estro Duflo,
48:13
Apogee Banerjee
48:14
and Michael Kramer, you know, the,
48:15
the them really being the big pioneers of using RCTs
48:18
and development in various countries, you know,
48:21
like Kenya and elsewhere.
48:23
But recently there was some, I, I think a bit
48:26
of blowback on Twitter that, you know, that really,
48:29
you know, it's wonderful, you know, that RCTs can get,
48:32
you know, very good precise answers,
48:34
but often these are, you know, for smaller sort
48:36
of questions versus, you know, in,
48:38
in macro development you're trying
48:40
to really tackle big questions, you know,
48:42
why are countries rich and poor and,
48:44
and trying to get a, you know, a very big understanding
48:46
of why that's the case.
48:48
But obviously this comes with a lot less
48:51
precision sometimes, you know, not as great identification.
48:54
I'm curious, you know, what is the case
48:55
for macro development in, in your eyes as a field?
48:59
Well, it might surprise you by,
49:00
by defending micro development at the beginning
49:03
of my answer, which is that I view them as compliments.
49:06
So for example, some of our evidence is consistent
49:09
with misallocation,
49:11
but without knowing exactly what the sources of
49:14
that misallocation are,
49:15
there's no clear policy implication in terms of getting rid
49:17
of particular government regulation or, or,
49:20
or, you know, in trying to encourage the market
49:23
to be more competitive without knowing
49:25
what the distortions are, we can't,
49:27
we can't actually do that.
49:28
And so applied micro people tend to study a narrow question
49:31
that enables 'em to get at particular individual policies
49:34
and how they're contributing to misallocation.
49:36
So, and that's the way they're complimentary,
49:37
that the macro evidence sets priorities in terms of what,
49:41
what the big inefficiencies are,
49:43
or the, the big undersupply is in the
49:45
market is of human capital.
49:46
Is it physical capital? Is it the allocation of resources?
49:48
The micro research can compliment that
49:50
by saying individual policies
49:52
and what they would, what they would do.
49:54
Another analogy I like,
49:55
I'm trying in this one, let me see what you think.
49:58
The first time I've tried this is you could say, look,
50:02
you know, I'm not, I'm not a a life expectancy a demographer
50:06
or, but just a health researcher can tell you decompose,
50:10
you know, what, how, what's the most
50:11
important thing for a life expectancy?
50:12
But imagine nutrition's really important and, and exercise,
50:16
but that we have a hard time quantifying exactly which
50:19
exercise, exactly how much there's research that comes out.
50:23
But it's is, and same thing with food,
50:25
it's inherently limited because often we don't have random
50:28
variation and often, you know, the effects are long term
50:31
and it's gonna take a while to see them,
50:32
and we're just not tracking enough people for long enough.
50:35
So macro development kind of has that problem, which is
50:39
it's working on something that arguably
50:41
is incredibly important.
50:42
You know, it led to hundreds of millions
50:44
of people being lifted out of poverty of these government.
50:46
I think these government policy reforms in places like China
50:49
and India, I think that's true.
50:52
But at the same time, I can't tell them exactly
50:54
which policies did what.
50:55
I can tell some estimate of what,
50:57
what moving away from state owned enterprises did.
50:59
But there's lots of details in there that I can't address
51:02
that applied micro people could better address.
51:04
So I guess I would say, you know,
51:06
macro is working on these big things like, like nutrition
51:08
that are hard to get answers on,
51:10
but might be really important for people's health, you know,
51:13
quality of life, life expectancy.
51:15
But at the same time, you still want the people,
51:17
the dermatologists, the, the people dealing
51:20
with some narrower issue that's still important
51:23
to people's quality of life.
51:24
Still worth investing resources into figuring out are
51:26
people, you know, should people be, should vaccines be free?
51:30
Should bed nets be free? Are bed nets effective?
51:32
Are they cost effective? Should, should the government
51:37
subsidize kids to, to stay in school, parents
51:40
to keep their kids in school
51:41
or to take them to doctor's visits?
51:43
All of those seem really use worthwhile.
51:45
All those are really worthwhile.
51:47
So I'm a huge fan of banerjee to flow creamer
51:50
and the entire body of work on RCTs and micro development.
51:53
I just don't see it as, as either or.
51:56
I see macro development
51:58
as being really important in its own right.
52:00
And, and them being complimentary subfield.
52:05
That's a great point. You know, that how,
52:07
how they can be complimentary.
52:08
My last sort of question for you
52:10
and you know, p is, you know, really talk about ai,
52:13
you know, I think it's very much, you know,
52:16
technology is very intimately related to economic growth
52:19
and it's widely considered to be, you know, the sort
52:23
of leading driver of, you know, economic growth.
52:27
Obviously, you know, political economy
52:28
and o other factors can, you know, cause
52:33
technology to be less easily adopted or,
52:36
or less easily used.
52:38
I'm just curious what your general thoughts are on, on AI
52:41
and economic growth and,
52:43
and you know, what some, what we should be thinking about it
52:47
and, and how growth economists like yourself
52:49
are thinking about it right now. Okay.
52:51
I think obviously incredibly
52:52
topical, incredibly important.
52:54
So I'm glad you brought it up
52:55
and it's a great subject to, to end on.
52:58
So I think, you know, one
53:00
of the challenges we face in thinking about AI is that,
53:03
you know, it might be different than other
53:04
technologies we've seen in the past.
53:06
We can think about robots, we can think about software
53:09
IT in general and changes that that's brought to, you know,
53:13
that might make, may have made bigger
53:14
firms bigger and so on.
53:16
So I think one of the challenges with AI is just
53:19
that it might be different.
53:20
And so the things we can carefully study is
53:22
what happens in the past
53:24
and they may not a perfect guide
53:25
to what's gonna happen with ai.
53:27
So that's one comment reaction.
53:29
Another very broad comment is that,
53:31
and this I think is getting less common, is
53:32
that there was a a lot of discussion about whether growth
53:35
was slowing down and whether that was inevitable.
53:37
They were just kind of running out of ideas.
53:39
And ideas are getting harder and harder
53:40
and more expensive to find, like every doubling
53:43
of semiconductor microprocessor capacity requires more
53:47
and more billions of of research.
53:49
And eventually we're gonna hit some sort of wall or,
53:51
or we only have so many people to work on these things.
53:53
And so progress has to slow down technologically just
53:56
to come diminishing returns.
53:58
So I think at the same time, some people were saying that,
54:02
other people were saying, I'm worried technology's gonna
54:04
take all the jobs away, including ai.
54:06
And I'm like AI taking jobs from people.
54:08
And I said, those aren't consistent
54:09
because if, if we get a lot out of the work done using AI
54:12
that we used to get done during with people,
54:14
then income per per productivity will soar.
54:18
Income per unit of labor input will soar.
54:21
Now of course there's major distributional questions
54:23
that we should be really concerned about.
54:24
Who's gonna get the income from that?
54:26
We could talk about that at length,
54:27
but I think you want to know what I think
54:28
about the growth potential.
54:30
And I would say that too much of the work has been focused
54:33
for my assessment.
54:35
Too much of the work has been focused on
54:37
what tasks in just the general job market will
54:42
AI do and what will that, how that'll affect employment
54:45
and wages and different occupations.
54:47
Obviously that's incredibly important
54:48
and I can see why people are focusing on that.
54:50
But for the long run growth question,
54:52
I think a critical thing is whether AI can make research
54:55
more productive and, you know, I'm using it,
54:58
everybody I know is using it and,
55:00
but we don't have a, a magnitude yet of how,
55:02
how much it's gonna contribute.
55:04
You know, I would love to see things like more therapeutic
55:07
drugs being discovered with the help of ai.
55:10
I mean, I think that could also affect our view of it.
55:13
If we start seeing concrete benefits that affect everybody.
55:17
If AI helps come up with treatments for Alzheimer's,
55:20
that would be huge, right?
55:21
Or if it mitigates, you know, customized drugs
55:24
to attack certain types of cancers,
55:26
that would just be huge for human welfare.
55:28
And my general reaction would be if it can do that,
55:32
the growth rate can accelerate
55:34
and we should figure out how to, how to redistribute the,
55:37
the income from it to make sure most people benefit from it.
55:40
So I'm kind of a, a a a c cautious optimist
55:44
because I see, you know, the people who work on it say you,
55:47
you know, if you think it's impressive now you think
55:49
what it's done now, the growth rate it's
55:50
shown, it's still coming.
55:52
So I don't know enough about it
55:54
to know whether it's gonna hit a wall,
55:55
but everybody who works on it seems to be saying, of course,
55:57
those people are also selling shares in their stock.
56:00
So you have to be careful, but everybody who's working on it
56:03
is claiming this is gonna lead to,
56:05
to rapid productivity growth.
56:06
Of course, I have to mention the work
56:08
by my colleague here at Stanford
56:09
and the business school, Chad Jones,
56:12
and some of it's with Chris ti,
56:14
who's also in the business school here,
56:16
and they're really emphasizing bottlenecks.
56:18
And Ben Jones at,
56:20
at northwestern Kellogg is also emphasizing this.
56:22
There's a subset of activities that we really need, right?
56:27
That then those will become, get more
56:30
and more of the income in the economy.
56:31
So it's kind of like saying if AI makes software free, it's
56:34
so efficient, we can, it can do anything we want with it.
56:37
That doesn't mean software takes over the whole economy any
56:40
more than water
56:42
or air Being virtually free means those take over the whole
56:45
economy and those are essential
56:46
things, water and air, right?
56:48
But that doesn't mean we can use them to make everything.
56:50
So if AI gets so good,
56:52
it might just make software incredibly cheap and, and, and,
56:56
and we might get 2% higher income outta that.
56:58
Or we might get, if instead software can be used to do a ton
57:01
of different things that it's not doing now,
57:03
then maybe we'll get a 20% burst in income over the
57:06
next 10 years out of that.
57:07
That would be great. And again, my instinct is
57:10
just try really hard to, to deal
57:13
with the distributional question.
57:14
So as many people can benefit from that as as possible.
57:17
But anyway, the, the, the Jones
57:18
and TenneT view is that we're not gonna get an explosion
57:22
of growth in our lifetime, which is one
57:26
of the more pessimistic takes on it.
57:29
But I, I have to say, and they're,
57:30
and they're transparent about this, they don't really know
57:34
how many bottlenecks there are.
57:36
And I would again, go back to if in research we can really,
57:40
they're, they're kind of assuming research uses
57:43
all goods in the economy.
57:44
So if there's some bottleneck that's a bottleneck
57:46
to research too that might not be right, might be
57:48
that AI can make research so productive
57:51
that we can solve all kinds of human, you know,
57:53
we can stop global warming, we can,
57:56
you know, cure Alzheimer's.
57:57
We can, and dementia more generally, we can, you know,
58:00
cure a bunch of cancers.
58:01
That would be incredible. That's what I'm hope,
58:02
that's the cautious optimist in me is that it's gonna lead
58:05
to explosion of research, productivity and growth.
58:08
I said just the other day, some open AI tool I think
58:11
solves some very old map problem.
58:13
So yeah, I'm, I'm constantly optimistic like yourself
58:15
and I, you know, as Bob Solo used to say, you know, AI
58:19
or, you know, internet was everywhere,
58:21
but except in the productivity statistics,
58:24
I think it's fair enough to say that
58:25
that could be the case with ai.
58:26
At least for now though we've seen some slightly, you know,
58:29
above I think average GP growth prints,
58:32
but it could, you know, certainly get a lot higher.
58:37
You know, this stuff, you know,
58:39
gets more fully integrated into the economy.
58:41
It's a real honor to have you on Pete.
58:42
I really think that thank you allocations, you know,
58:45
a Nobel Prize winning idea
58:47
and it's a real honor to be able to talk to you about it
58:50
and all your other terrific contributions here.
58:53
Thanks so much. This was really fun.
58:55
This is the Capital and Freedom, the 21st Century podcast,
58:58
an official podcast of the Hoover Institution's economic
59:00
policy working group where we talk about economics,
59:02
markets, and public policy.
59:04
I'm John Hartley, your host.
59:06
Thank you so much for joining us.