Gayblack Canadian Man

Foreign Policy Analysis
63rd Economic Conference (Part 1)

63rd Economic Conference (Part 1)

I’m Eric Rosen
the Boston Fed here to welcome you to our 63rd annual conference I think it’s
a really exciting topic and I’m really looking forward to the next day and a
half of hearing a wide variety of views on what we can do about some of the
geographic disparities that we see in the United States so let me start by
just talking a little bit about why I found this topic so intriguing and why I
found it so intriguing was the idea of place and the importance of place so the
typical story for many economists had been gee we should really focus on
individuals we shouldn’t focus on place we had a conference that was looking at
income inequality and one of the key elements from that income inequality
conference was to focus on education so if we educate people that you provide
them job training then that’s the right way to go that will help propel them and
they can move to where the jobs are we don’t have to worry so much about the
geographic spot that they’re located but what we found over time is that places
have been quite different in outcomes but we haven’t seen the population
shifts that would be consistent with some of that story which starts to make
us think about how important is place and do we have the right policies in
thinking about place so if you’re thinking about only individuals you’re
thinking about how do I do job training how do I make sure they get educated if
I’m thinking about place it’s probably a different set of shoes that we have to address to make
sure that that place can actually use the funds in an effective way
do you have an effective government you have an effective social infrastructure
you have effective employers that all can work together in a cohesive way to
actually generate a different outcome so from a public policy standpoint I think
it’s really interesting to think about if we’re gonna think more about
place-based policies what should those policies be and I’m hoping over the
course of the next day and a half that I’m going to learn a lot from everybody
in this audience who spend a lot more time thinking about it than I have about
how we can do this in a much better way I think there’s some interesting trends
just to highlight it used to be that we were seeing much more convergence we’re
not seeing the same kind of convergence that we were seeing before we used to
see a lot more migration from Ford rich areas we’re not seeing as much of that
migration so we have to start thinking differently as we start seeing these
different kinds of economic trends I think over the course the next day and a
half they’re going to be a variety of issues that we’re going to be addressing
that I think are really quite fascinating so why are these geographic
differences so persistent why have economists who previously were focused
on people moving starting to think very differently about regional economic
shocks and how we should address regional economic shocks thinking about
quality of education quality of health care not only what’s available in these
communities but how that affects people’s outcomes over their lifetime
and thinking about how we can think about the rural and urban divide which
in many situations has gotten much worse so I think that over the course of this
next day and a half we’re going to be trying to address a lot of these
questions and hopefully most of the papers and most the questions will also
get back to not just stating what the problem is but thinking a little bit
about what the solutions to the problem is because I think in many cases the
problems are visual and the solutions are less apparent so it I think is
really important to understand that so you might wonder why it is that the the
federal reserve’s involved in this topic in the first place did you think about
setting interest rates and monetary policy and that’s certainly one of the
key functions of the Federal Reserve but I actually think that our dual mandate
is encompassing something more in terms of welfare dynamics about trying to
change the quality of life for everybody who lives in the United States and so
we’ve been very focused over the last few years I’m thinking about how we can
make sure that New England as a whole does better so we’re fortunate that New
England is much better off than many other parts of the country but you don’t
have to go very far north of Bangor Maine to see some really striking rural
poverty you can see it in Vermont you can see if New Hampshire so you can see
it in Boston as well but I think an important insight is the poverty is not
just a city outcome it’s an outcome that is very frequently found in rural and
small cities so what what have we been doing what we’ve been trying to do
researching the area particularly thinking about how some of the
manufacturing textile migration has affected smaller cities in New England
we’re doing convenings like this conference so we’ve had a conference
this conference is focused one of the geographic disparities
I’m going to talk in a minute about the work of cities were challenged that
we’ve had which is actually a place-based endeavor to try to change
the story for some of the smaller cities in New England as well as some of the
rural areas and then I would say that the Boston Fed has always cared a lot
about labor dynamics and contributing to the National monetary policy story of
how to integrate all these things into a coherent framework so that’s why the
feds interested so nationally we’re seeing super star cities are really
attracting more people in businesses you can see that in Boston you can’t see it
out through this window let me go to the reception tonight you’re gonna look out
at the Seaport area ten years ago it was a parking lot with ocean views it’s now
basically another city the Boston fed when I first came here in 80s was around
an area that was the combat zone the combat zones been completely priced out
of the market so there is no combat zone we had been at the edge of the city
we’re now getting to be in the middle of the city so Boston’s doing really well
and when you go to the Seaport area they’re seeing new restaurants are
seeing a lot of young people you’re seeing a lot of well-educated young
people so Boston is thriving and I’m gonna argue that there are other places
in New England that have a similar story but then their rural areas that are
having much more of a challenge its older not younger its less educated it’s
not well educated incomes are growing more slowly and the health outcomes are
not nearly as good so there’s a striking disparity even in New England between
the urban centers that are doing pretty well and some of the less herb in the
areas that are not so I know a lot of people in this room have done a lot of
work thinking about the role of Education and I would emphasize that
economists have and many economists in this room have really focused on the
fact that education and the wage premium you get for getting a college education
is quite hard one of the important aspects of that is that college educated
people are not dispersed randomly across New England they actually tend to be
aggregated in some of the cities so it does create a challenge if a lot of the
wage premium is going primarily focused in communities that are successful in
drawing college-educated individuals there gonna be a paper that focuses on
health outcomes but this is looking at kind of the ultimate in health outcome
which is life expectancy and it’s looking at metropolitan versus non
metropolitan areas and when my staff showed me this chart I really found it
quite striking how that divergence had grown over time in terms of health
outcomes so it’s not just kind of an economic story it’s really literally a
quality of life and length of life story that we’re going to be addressing over
the course of the next day and a half so Geographic disparity is an important
part of New England and I’m going to make a comparison between three of the
cities and the states that they’re in one is Burlington Vermont
the other is Portland Maine and then of course Boston so if you go back to 1960
through 1980 most of these cities were not performing all that well and it was
kind of striking these were kind of gritty cities and they seemed gritty
they were cities where manufacturing base had disappeared in Maine it was
more it had been a lumber area lumber obviously wasn’t doing as well
Massachusetts and the eastern part was textiles western part was manufacturing
but as all these businesses migrated to the south or abroad you started to have
very depleted cities that weren’t doing particularly well now you’ll look at
these cities and it’s completely different story they’re young people
they’re educated the populations growing outside of the primary cities the
opposite is true so this looks at those three cities and so this is looking at
the percent of US metro area per capital personal income and you can kind of see
that as we were getting into the 70s as the manufacturing textile lumber was a
problem you can see that in all three of these cities there was a
dip so at that time Boston had dipped down to basically where the national
average was Portland and Burlington were substantially below what the national
average was but then you look over the next 40 years and there’s been a steady
increase so Boston certainly has done quite well but certainly relative to
where Burlington and Portland were in the late 70s
that’s quite a trendline that they’ve been quite successful relative to what
was happening in the United States in turn now above that average now if you
start looking at what that means for outcomes this is looking at the
population so again it’s gonna be looking at each of these three cities
and it’s comparing to what the outcome is in those read in those states taking
out the Boston metro so the Boston metro includes parts of New Hampshire there
are plenty of people who work in this building that actually commute down from
southern New Hampshire so that’s why we’ve combined New Hampshire and
Massachusetts you can see while Boston has had population growth outside of
Boston Massachusetts in New Hampshire haven’t had the same kind of growth but
it’s even more striking when you get to Vermont and Maine
so Burlington and Portland both have had reasonably good population growth at the
same time you’ve seen some reduction in population outside of those city areas
in terms of the kind of population that many businesses are looking to attract
the population between 20 and 40 in each of these cities you can see a difference
between what the primary city is doing and what the rest of the state is doing
so you can see there are more young people in Boston which is not all that
surprising as well as in Burlington Portland not so much I would highlight
an important difference between Portland and these other two cities University of
Vermont’s located in Burlington one of the key features of Burlington Vermont
is that University it’s kind of off if you haven’t been there it’s a funky fun
city and it’s partly a funky fun city because it has a lot of college students
there and one of the reasons Boston’s attractive is for the
same reason everywhere you look there’s a university in Boston and so it does
attract a different kind of population Portland Maine the main campus in Maine
is a Nora know close to Bangor and so they don’t have a college centered
primary city and so you don’t see quite the same population differences as you
see in those other two but nonetheless in all three they’re doing better than
the rest of the state looking at the share of population 65 and over you can
see that in the cities less of an older population that makes sense cities are
expensive places to live if people are on a fixed income looking for a less
costly place to live it’s not completely surprising that you’d see more older
people but it does have implications for the community when you’re
disproportionately old relative to what’s happening in the cities mm-hmm so
we had spent some time looking at these disparities and seeing what the
differences were and then we started looking at what some of the outcomes
were for various cities and so this is a chart that looks at cities starting
basically in 1960 through 2010 and asking what the economic outcomes have
been these cities were not randomly picked these are the cities with the
highest poverty rate and the lowest family income but an important context
is if you look back to 1960 there’s some that aren’t doing that well but a bunch
of them are right around the median family income of the nation as a whole
what’s striking about this chart is how little success over 50 or 60 years has
occurred so in each of these communities they’ve been losing ground relative to
the national average for a really long period of time that is really
challenging and so when we look at this chart we asked ourselves was there
something that we can do and one of the things that we started thinking about
was the value of place and so we have a competition called the working cities
challenge which is bringing in foundation
National Foundation money working with the state and working with businesses to
actually fund some of the cities the cities that won our first competition in
Massachusetts was Lawrence which is the orange line so you can see they started
close to the average overall for the United States
they’re the worst performing city in Massachusetts over this period so they
were one of the winning cities the second is Holyoke the third is Chelsea
the fourth that’s Fitchburg so three of the four are the bottom three lines on
this chart we didn’t pick them because they were the cities with the greatest
need we actually and it wasn’t the Fed that picked them it was actually an
outside steering group that did it and if you’re interested in hearing more
about the car I don’t have in my limited time enough time to talk about this but
if you type working cities or working communities you’ll hear about what we’ve
been trying to do in this area but these are the communities that we focused on
to try to tell a different story and part of the reason that we thought it
was important this is looking at the grant recipient so those four cities
that I just highlighted first the rest those cities and those lines and you can
see from 60 to 80 there was a loss of population but over the last 30 or 40
years their population has actually grown so if your solution is that cities
that are not performing well are just going to have people migrate to cities
that are the fact that these cities over 50 years have done so poorly and yet
people haven’t moved actually out over the last 30 or 40 years these
populations have grown now the population has also changed so if you
look at many of these cities it has a very different composition in terms of
ethnicity than it did 40 or 50 years ago so if you look at a city like Lawrence
which was the orange line in that previous chart it’s basically turned
into a primarily Dominican spanish-speaking community that’s not
the way it was 40 or 50 years ago but it is that’s accounting for some of this
population growth Chelsea another one of those cities which is actually just
depending on where you are in the building you can actually see Chelsea
Chelsey’s in visual distance of seeing downtown Boston but it’s been kind of a
new immigrant city for a very long period of time it’s a city that has
changed quite dramatically so each of these cities have a large immigrant
population so if your answer is just to get people to move that doesn’t look
like the solution at least for the cities that I just showed you in
Massachusetts and actually I can show you the exact same chart for Rhode
Island the same the same downward trend for the cities that were not performing
all that well have not performed over the last 50 years you see the exact same
trends in Connecticut as well so it’s not isolated to Massachusetts actually
all of southern New England has this attribute so one of the questions we
asked ourselves is Boston Portland Burlington have all been able to tell a
different story is it only the primary cities in these states that can tell a
different story or is it possible that some of these other communities can tell
a different story as well so we focused actually on communities that we thought
had the highest probability of success and the way we define that was they had
a cohesive vision for their community which is that the government nonprofit
and business all agreed on what kind of solutions they wanted to focus on
collaborated well together and could come up with cohesive projects that
would tell a different story over time we’ve done this type of competition in
Massachusetts twice once under a Republican governor wants under a
Democratic governor the governor of Rhode Island and the governor of
Connecticut both asked us to do competitions in their states as well so
we have done that this isn’t federal reserve funding it’s Foundation funding
it’s state funding and it’s business funding and we’re now doing a work in
communities challenge which were just launching in the state of Vermont we
recently launched though the cities or the communities have not been picked and
we’re hopefully going to be launching in Maine this fall so it’s a model that
we’re spending a lot of time thinking about
I think we’ve seen some positive success to date we’re certainly seeing that in
terms of the initial grants for those for winning cities it was only 1.6
million dollars it was pretty short money but they’ve leveraged it to 11
million dollar and follow-on funds so other foundations have decided to invest
in the same set of cities that’s a pretty just looking at the fact that
other foundations were willing to focus on these cities these were cities that
had not been getting national foundation money before one of the goals of this in
competition was actually to get national foundations to spend more time outside
of Detroit and Newark and very large cities that clearly have problems but
also to think about some of these other communities so this competition I think
has been successful in redirecting at least some of the foundation money to be
thinking about these communities but it’s also to get the state and business
leaders to be thinking a little bit differently about these communities as
well on our website we have an outside evaluator that talks about some of the
changes that we’ve seen and goes into a lot more detail but I think we have
already seen some success and part of our goal was to get a system change
which is to get the states to start thinking differently about these
communities and not just spread the money around generally but to think
about which communities had the highest success rate and what kind of things
that they needed to look for in order to be able to get a higher social return on
the money being spent in these communities so we need new policies the
last 50 years and many of these communities have not worked we’re trying
one way I think we’re going to hear about lots of different ways that we can
think about how to tell a different story about place but I think the
solution is not to throw up our hands and give up but actually to think about
let’s think about different solutions let’s experiment a little bit more one
of the nice things about smaller cities and communities is there a lot of them
so we can do a lot of experimentation to see what works and what doesn’t work
I think this conference is going to really focus on what more we can do
about place and I think that’s really important so I’m looking forward to
seeing all the papers and understanding what we should be thinking about
differently in turn of telling a different story in New
England I think our working cities working communities challenge really
gets at the heart of the kind of questions that this conference is asking
and we know these are tough questions and we know we haven’t had a lot of
success over the last 50 years that makes it all the more important for us
to be even more focused on success going forward so I really look forward to the
papers and all the discussions over the course of the next day and a half
thank you for being here and I think it’s going to be a really exciting day
and a half and look forward to seeing the next round of papers and with that
we’ll have Jeff and his panel come up you right good morning my name is Jeff
Thompson I’m a senior economist policy advisor here at the bank and director of
the New England health policy Center all the moderating the morning sessions and
before we kick off with our first paper and which will be presented by Chris
foot I want to give a few housekeeping items first of all just to let you know
that restrooms are located just outside those back doors either off to that
direction or back near the elevator from where you came in second note is that
brakes and sort of snack coffee items will happen out here to my left your
right and lunch will be served in the cafe which is out the back doors to your
left and my right and also to let you know that we are pleased to have many
members of the media among us today and just to very quickly mention that the
ground rules for the press coverage is that all presentations and general
discussions during the conference session are on the record any informal
or social elements the gathering conversations over breakfast or lunch
are off the record unless otherwise specifically agreed to by the parties
and we will be live-streaming the whole event so we’re all on camera and the
will be posted to the web live streamed to the website so anything said during
the conference session will be viewed by the entire public so we’ll go ahead and
start and the presenters will have 25 minutes discussants will have 20 and
then we’ll have 30 minutes of discussion for the for the general audience so with
that Chris okay great let me say at the outset that even
though Ben and I both work at the Boston Fed the views are our own and we’re not
speaking officially for the Boston Fed or the Federal Reserve System now our
goal in this paper is to set forth some basic facts about the spatial
distribution spatial evolution of labor markets over the past five decades and
that’s a pretty ambitious task in large part because there’s several
dimensions over which labor markets can vary there’s the rural versus urban
dimension center cities versus suburbs or super star cities versus less
successful cities and if we wanted to say something about each of these
potential dimensions we thought that we should take the county as our unit of
analysis so our main data set is going to come from the County business
patterns which is available yearly from 1964 to 2016 we have employment in that
data set both total employment which is essentially private non-farm and
manufacturing employment which we sometimes have to impute from an
establishment size distribution we also have payrolls which we can combine with
employment to get average earnings per job to create the data set that we work
with we’re going to combine some small counties and independent cities in
Virginia exclude exclude Alaska and Hawaii which gets us from about 3100
official u.s. counties to a balanced sample of 2,900 about 2900 counties in
our in our data set and in our work will also use some county level demographic
data from census now the first thing we’re going to do with the county level
data is group the counties into four groups based on population density and
then we’re gonna make some maps and we’re gonna choose a non-standard color
scheme for the map in the sense that we’re gonna graph the rural counties in
red and the more urban counties in blue that’s a joke by the way I’ll point out
two jokes as I go along so you know when to laugh the densest group is the single
percent top percentile in population density and they appear in dark blue on
the map which you can see is for the Northeast census region and there are
all these dark blue counties were all sort of centers of dense dense cities
the next two groups the 96 to 99th percentile and the 86 to 95th percentile
are either suburbs of big cities or the centers of smaller cities and taken
together the densest three groups account for about or account for exactly
percent of the counties but about 70% of the US population and the least dense
group in red is going to include both what we think of as rural America and
sort of outlying suburbs when jobs move in there but also because around half of
these tan counties are centers of smaller cities their growth reflects
what Jordan Rappaport of the Kansas City Fed has termed the growth of large less
crowded locations and Jordans research indicates that for cities that have
non-trivial substantial population but are still sort of have room for people
to move those are sort of cities and counties that have grown more quickly
recently now an interesting facet of both of these grasses if you look at the
top percentile in terms of density it looks like those top dense counties
those dark blue counties have done a better job sort of hanging on to their
shares of both population and employment in the national data and that’s
consistent with what we’ve seen or and actually on the other end of the
spectrum whereas we’ve seen sort of the stabilization for the top counties here
you’ve seen sort of a tick down and sort of the least dense counties in terms of
their shares and that sort of opposite effect or the opposite results they’re
sort of consistent with what’s been widely noted that big cities seem to be
doing better than rural areas in the last couple of decades since about 2000
now the fact that the densest counties are holding on to their national shares
leads naturally to the study of suburbanization because suburbanization
is really all about how the dense center of an urban area is sort of hanging on
to its employment and population versus seeing that employment and populations
spread out and we take up that question in this slide now we’re going to define
local labor markets here in the same way that it’s defined in a lot of current
macro labeled macro labor research which is as commuting zones and these zones
are approximately 700 zones that are put together by the Department of
Agriculture to reflect local labor markets and there are a lot like
metropolitan areas except the commuting zones cover every single cow
in the country now we further group or classify the commuting zones based on
the the density group of their densest County which we’re going to label as
their Center County and then we asked what’s happening to the shares of total
employment and population in that Center County for the different groups of
commuting zones now if the Center County shares are
declining what’s happening well population and employment is moving out
that suburbanization and again that way of measuring suburbanization is also
common in the urban economics literature although they often use metro areas now
the dark blue lines in each of these panels are the center County shares for
commuting zones that have one of these dark blue top 1% counties at their
center as their densest County and as you can see from most of the sample
period you see that those shares decline these declines are consistent with the
long-standing trend towards suburbanization that Ed and Matt Kahn
and others have written but interestingly it sort of flattens out
these lines kind of flatten out after 2005 both for employment or both for
population and for employment so even though there’s been a long-standing
trend something has changed something has changed relatively recently the
suburbanization in these dense seas with very dense centers is stalled you can
the next consider what about the commuting zones that have sort of the
light blue counties that cast Center County that’s in the 96 to the 99th
percentile well for population you also see sort of a flattening out
but for employment you don’t you see sort of a shallower down or decline but
there’s not much change here in 2005 now for the commuting zones with less dense
even even let in the center counties have even lesser density the CZs with
red counties at their center or the tan counties you never really see much of a
trend towards suburbanization probably because there’s not much pressure
because the center is not very dense to begin with now taken together we think
that the population data support the idea that larger cities have made a big
resurgence in the 21st century and there’s a lot of people have noted
sort of living downtown has become more attractive but the fact that you see
sort of employment decentralization employment suburbanization continue for
cities the 50 or so cities that sort of commuting zones that have sort of the
next tier level of density at their centre the fact that that’s continued
indicates that to us there’s not a lot of evidence that there’s some production
externality that’s become larger in cities that’s drawing people together in
terms of their employment rather it’s consumer amenities that are making
downtown living more attractive and that’s why you see it in population now
we next asked what can be learned by evaluating the earnings per job data
that’s available in the county business patterns
now these graph show binned scatter plots of average earnings per job by
density percentile all 100 of them for the first in the last years of our
sample period and the data are deemed with weighted from populate or their d
mean from population weighted averages of earnings so there’s a zero line in
each panel and the reason that that zero line appears above most counties is that
most people live in the rightmost part of the density distribution and that’s
where wages tend to be higher so that pulls up the weighted average of
earnings above most counties now Ben and I call this graph from crooked smile to
hockey-stick and at the start of the corrupt the
sample period you can see that there’s sort of a fairly linear relationship
between density and average earnings per job in the part of the distribution
where most people live but by 2016 that has changed and you see more of a hockey
ships hockey stick shape in which the very dense counters are sort of pulling
away from everybody else the more urban counties here are sort of pulling away
from the pack you don’t have to graph just means you can also graph the 75th
and the 25th percent to give you an idea of dispersion in
average earnings per job as a function of density we’ve done that here now if
you eyeball this graph actually it should make make clear that because
we’re graphing the 75th percentile and the 27 2015 nice to lines is the
interquartile range at each density percentile now if you eyeball these two
figures you sort of can see that the average gap looks bigger in the 1964
graph than in the 2016 graph you can actually of course check that simply by
taking the average of this gap throughout the entire distribution
that’s the gavage gap for 1964 compare it to the 1965 1966 and so on but you
could also since we’re doing nonparametric here you could also take
the average over any particular segment of this line some you know the bottom 85
85 percent tiles from 85 to 95 the top 5% and so on now here’s what you get
when you do that when you take the average gap across all of the density
percentile distribution and graph it what you sort of see from your eyeballs
actually is verified by the graph the average graph it gap is larger at the
start of the sample and 64 than it is at the end of the sample and 2016 you see a
decline here if you take it just over the bottom 85% of counties in terms of
density not surprisingly 85 is pretty close to 100 so you get a very similar
picture you also get a similar picture when you graph just the average gap over
the percentiles 86 to 95 the most interesting slide the most interesting
panel here is at this bottom right panel where we look at the average
interquartile range by year just for the top 5% and what do you find
well you don’t find much of a decline in here but you find a very large increase
and very large increase in dispersion at the rightmost part of the density
distribution that starts around night 95 now remember the top 5% those are the
light-blue county or in the dark-blue counties and that’s about where 45% of
the country lives and that to us is sort of evidence of this superstar cities
phenomenon the fact that you’re seeing dispersion in the more urban counties
that begins around 1995 now a more precise way to analyze average earnings
per job at the county levels with regressions and in our paper we run a
series of yearly regressions we don’t run one large panel regression even
though we could because what we want to do is we want to see how the correlation
of average earnings per job changes with particular potential regressors over
time so the regressions are going to be run
separately year by year the dependent variables exactly what we’ve been
looking at nonparametric ly average earnings per job and the potential
regressors our average January temperature which I know was one of EDS
favorite regressors reflects the rise of the Sun Belt also long population
density and an indicator if the county is in the top 5% of density in that year
the manufacturing share of employment the log share of persons with a
bachelor’s degree in that county which we interpolate from decennial census
data and from the American Community Survey and these were so will allow
these correlate these regression coefficients to change over time now I
don’t want to get too deep in the weeds but I do want to note that as a
technical detail we’ll also take the spatial nature of these of these of
these of this data into account in a couple of different ways one is that for
two variables log density and the manufacturing share will not only
include the own County Vale value of that variable but will also include
weighted averages of bordering County data and that’s what this WX term here
is the W is a weighting matrix that constructs weighted averages and we
specify that W matrix to have a particular form
it’s a second-order continuity matrix which sounds fancy but
it’s really easy it’s basically the idea that neighboring counties are those
which border a particular County and which border the bordering counties
that’s where you get the second order piece a second spatial detail is that
we’re also going to allow the residuals in each of these yearly regressions
these yearly slices of the data to be correlated and that’s this notation
right here that’s really important to do in order to sort of get the proper
standard errors it also helps with efficiency but we as a side product we
will estimate the strength of the correlation of residuals the clumpiness
of residuals in our regression over time and we’ll be able to use that to sort of
understand a little bit more about the data as well let it give you an idea
this is what I just said to give you an idea about how this W matrix works this
is a graph of manufacturing shares in our counties in 1980 and this is what
happens when you take the average weighted averages of nearby counties you
can see that what’s being picked up here is sort of a local area effect now a
side benefit of looking at data from a specific year 1980 is that you can see
that by 1980 the big manufacturing areas in the country were not only the Upper
Midwest but it also moved to the south in North Carolina by the way for you pub
trivia fans in 1980 is the highest state or the state with the highest
manufacturing share not Michigan or Ohio so in the interest of time I’m actually
going to skip the January temperature coefficients sorry Edie and then I am
going to just focus on the density coefficients now the the in the Left
panel the solid line shows you the linear effect of density that
coefficients just putting in Log density and the dotted line or the dashed line
here shows you the effect of the nearby density in the right panel those are the
coefficients on the dummy variable the indicator variable for whether or not
the count in the top 5% of debt of the density
distribution in that particular year and if you look at this graph you can see
how we can get a hockey stick shape in the nonparametric stuff that we showed
earlier because the top 5% here this indicator is pulling away for the rest
of the country and a nice thing about running these regressions year by year
as you can see when that happens it happens sort of around the early 1980s
the linear term here changes a little bit at that time but not much but
overall sort of you can see that just the linear effect of density declines a
lot of the action is just being picked up to the extent that density is
rewarded by the regression in terms of in explanatory power it’s being picked
up by this single indicator that indicates whether the counties in the
top 5% now as Eric showed in the opening graph or one of his opening graphs the
early 1980s were also a period in which the college premium started rising we do
have a rough education regressor in our in our very in our model as I said the
log share of bachelor degrees and interestingly and perhaps somewhat
miraculously it also starts rising in the night to early 1980s but exactly the
time the the premium estimated on individual data does not only that but
it flattens out over time although it’s sort of a little hard to see exactly
when that flattening out happens whether it’s in the 1990s or in the 2000s if you
look at the share manufacturing regressors here I’ve graphed the solid
line is the own County share and the dashed line here is the coefficient on
the neighboring County share and the solid line is interesting you can see
that the if you have a high owned County Sherriff manufacturing you’re likely to
have higher wages and of course in this particular County and of course that
could be a mechanical effect because manufacturing employers tend to pay high
wages if you’ve got a lot of manufacturing employers in your County
guess what your likely to have a high average the well
the nearby manufacturing share comes in negatively we don’t think that is a
causal effect rather what we think it is it’s reflecting the location decisions
of manufacturers they want to locate in areas that have relatively low wages so
what’s that going to do well holding constant a County’s own manufacturing
share being located then around a lot of other manufacturing share counties would
tend to reduce wages now the last set of estimates is the for the Landis that
measure the strength of the correlation in the residuals how clumpy are sort of
average earnings across different counties once you account for all of
these right-hand side variables and interestingly what you can see is that
the regressions or the residuals start out to be pretty clumpy that’s not an
official econometrics term but it declines over time until 2000
when all of a sudden the residuals become clumped together more intensely
now our hypothesis is that this is sort of another implication of the super star
cities phenomenon starting in the mid-1990s these super star cities pull
away from the other cities what would you expect to find well as we found with
the nonparametric analysis of the interquartile range you’d see on average
dispersion of residuals rise but because super star cities are really super star
metro areas or super star commuting zone you’d have a lot of the counties or all
of the counties within those labor markets move apart at the same time so
you’d have an increase in the clumpiness the correlation of the residuals as well
now I have a slide that sort of reviews the main results and how they tie back
into the spatial dimensions that we talked about earlier but rather than
review that in the interest of time I just want to touch on very briefly a
couple of points we make in the paper about the manufacturing share
essentially we move the manufacturing share from some explanatory variable and
make it the star of the show and ask what’s been going on with
manufacturing shares in the different counties that have different densities
and has often been noted there was a big rotation of manufacturing employment out
of dense counties and into less dense counties early in our sample period so
this shows you growth rates annualized percentage growth rates of manufacturing
employment in our selected years so this bar here shows you the growth rate for
the least dense group of counties the red counties between 1964 and 1973 so
even though the share of manufacturing in the United States is declining
outside of dense cities and particularly in these less dense areas manufacturing
employment is actually growing early on now what happens though is around 2000 a
gun goes off and manufacturing employment Falls very sharply everywhere
it becomes as you might say the 21st century was not a great period not just
for manufacturing in dense cities which is the case in the 70s in the eighth or
60s and 70s but for everywhere now our paper makes a couple of points
about or raises a couple of questions about this and I’ll just touch on them
briefly one is why have recent declines in the US manufacturing share have had
such a large effect on local communities as I said and as I’m sure you’re aware
many the average share of manufacturing employment in the United States has been
declining for most if not all of the post-war era but yet it seems like more
recently declines in manufacturing have has larger effects on local communities
now as we’ll hear later on in this conference part of that discrepancy part
of the reason why the more recent declines in manufacturing of her
communities worse is that the communities that have been hurt worse
recently may be different in the sense that they have lower levels of human
capital we just develop another angle in the paper a complementary explanation in
which we sort of use a spatial error model to discuss sort of
counties that are lost that lose manufacturing jobs today are counties
which are sort of packed together with other counties that have made high
manufacturing shares or whether they tend to be sort of counties that have
high manufacturing shares and surrounded by non manufacturing counties and we
make the case that more reap more moron more recently it’s the sort of clumps of
manufacturing counties that have been hurt I was worse or at least as bad as
the earlier declines in manufacturing employment were occurred and what’s
happening is because the sort of manufacturing counties are all packed
together and being hurt they have less chance to diversify we provide some
statistical analysis of that and we can come back to it in the general
discussion if you’re interested and here’s just the stuff here a second
point about the manufacturing share relates to some recent work that people
have done on within big-city inequality I won’t say much about that other than
to note that when we look at the types of jobs that have been lost in big
cities manufacturing jobs were lost in the 60s and the 70s more recently though
other types of jobs office support jobs have been lost and in the paper we talk
about how the loss of those office support jobs probably tie in to the
increase in big city in equality more than jobs lost in manufacturing early on
because the increase in big-city inequality that we’ve seen has happened
relatively recently in the 2000s the manufacturing jobs were lost way before
then but we talked in the paper about office support jobs and other sort of
non-college middle skilled jobs that have been lost that may be contributing
to hire big city in equality today so with that I think I’ll leave it there
thanks great great I’m I’m very grateful to
have a chance to discuss this this fascinating paper and I am so grateful
to the Boston Fed for hosting this conference these topics are first of a
first-order importance to our nation and it’s really wonderful to have a group of
scholars and people with strong policy interest coming together to discuss them
so I am haunted by this picture so this is a map of prime age 25 to 54 year old
and I find that definition of primates as offensive as many of you do prime
aged men males not working right throughout the US and I think it’s
important to remember as the unemployment rate gets you know to
vanishing low net levels that we still have a jobless problem in this country
we still have ten percent or ten percent of prime H men who have left the labor
force and they are not spatially uniformly distributed right they are
particularly evident in an area of America’s eastern Heartland that begins
down in Mississippi and Louisiana runs up through Appalachian ends in the in
the Rust Belt cities many of these areas more than one in four prime age men are
jobless and if we have a policy mission it is to figure out somehow or other to
reduce the enormous social dysfunction that is associated with these massive
levels of prime aged male joblessness this is the same map using the same
color scheme for 1980 there is still a regional pattern that is slightly
discernible but of course the whole level was so much lower right we’ve gone
from a world in which five percent of primary men were jobless to a world in
which 15 percent of prime aged men were jobless in the country as a whole that
just the levels are just in a totally different place and this is the same map
for women the levels of not working are much higher of course and it follows
more of a regional regional barrier and of course it’s a much more complicated
topic in many ways women who are not formally employed are typically not
miserable the way that men who are not employed typically are and moreover
they’re also doing socially useful things like taking care of kids and
doing other stuff whereas men who are not worked are typically watching
television right that’s that’s our our primary time use fact on this okay
so the this paper bike we are devote which is excellent really as a great
deal to our our knowledge as an arc and that arc relates to the overall
mark of urban history so if you think about the last 200 years in the 19th
century and early 20th century was the great movement of factories to cities
cities invented new industries like mass-produced cars and Detroit and they
grew great around automation but in the 20th century factories left and in some
sense manufacturing was never a good fit for urban density Manufacturing is
capital intensive often land intensive and once the sort of initial
breakthrough phase was over it always made sense to D urbanize things and of
course the nature of factory production has totally changed in 1910 Henry Ford
had factories that used about 200 square feet of space per worker right today
Ford and GM are at 2,000 square feet per worker and that’s it just a very
different mode of operation that requires a very different location um
late in the 20th century somewhat remarkably educated dense urban cores
have done well as consumer cities as well as places of knowledge heavy
production whoo yard and foot have put together the County business patterns
data in a usable sensible fashion that will serve many of us right I I well
remember the deep pain of hand entering the 1956 County business patterns data
in 1989 30 years ago and it’s wonderful that they’ve done this they have an
interesting arc in the paper which is they begin with urban centralization
then they talk about Matt and Manufacturing’s ongoing declined and
they come back to patterns of earnings and earnings disparity across counties I
think the basic issue here is they have two papers which are dividing their own
house within this within this paper one of which is a deep and fascinating paper
but the heterogeneous effects of manufacturing is decline on place right
that is a paper that absolutely should be written on its own and the second
paper is one which is charting the different paths of earnings and
population and employment across the u.s. using County business patterns over
a 50-year period my main comment for the for the first paper is to for the paper
on the broad paper is to always focus on employment population housing prices and
wages simultaneously right don’t split those things up those is by using those
moments of the data together that we figure out if what’s going on is a rise
in labor demand or a change in labor supply an increase in the amenities of
an area or a change in housing supply that makes it easier to to house people
bring those things didn’t use the FHFA housing price data as well there’s no
reason not to do that for the second paper I would be more focused on not
only the heterogeneity between today and the
but the heterogeneity between different areas in terms of the manufacturing
decline because I think those patterns will bear will bear rich fruit the there
are five interesting facts here in the paper currently the suburbanization of
population employment in very dense core is stopped after 2005 is this an urban
comeback or is this merely a reflection of the great recession in less dense
course the suburbanization of employment continues but the suburbanization a
population is slowed considerably is this a consumer city phenomenon or
perhaps something about the nature of housing supply the mean earnings to
density relationship moved from being a you or a crooked smile as they put it to
being an upward line so they skipped over the middle decades where it’s just
sort of a straight upward line to this hockey stick which shows this sort of
clear convexity in the relationship later on this is a great fact and I’d
love to sort of learn more about this and there’s a nice fact on my own
manufacturing being a plus but neighboring manufacturing being – for
earnings earnings dispersion across counties with similar density levels
fell throughout the 1990s but then subsequently rose in dense counties this
is an important thing to remember right that so often we you know have
conferences like this and we act as if earnings disparity across the u.s. is
some kind of a new phenomenon not least a new phenomenon by many measures in
many cases actually less than it’s been in the middle parts of the 20th century
in 1950 Mississippi was the poorest state in the Union and there were 18
states in the union with per capita incomes that were more than double that
in Mississippi today the Mississippi is still the poorest state in the Union but
there’s not a single state with a per capita income that is double that
Mississippi and they show that in this in this data the last point which is
nice is is that the the decentralization of manufacturing has been a major part
of the urban landscape since the 1960’s but starting in the 1990s manufacturing
employment started dropping as much in rural areas as in urban areas and that’s
sort of an important phenomenon part of the malaise in the eastern heartland as
the end of this manufacturing okay so this is their fact about population
employment and across these areas now I want to highlight how novel and
interesting it is to do this grouping okay it’s a really nice grouping it’s a
grouping that reflects the population weights now just tell you what what
happens if you group it by counties let’s say over the past 10 years is you
basically see a straight linear relationship between population growth
and initial pop elation density and then a decline for
the densest tenth of counties okay that’s because the densest tenth of
counties which includes a very large share of America is being split up into
sort of three groups here and there those bottom eight deciles
all of it show this strong positive relationship are being grouped together
in the 1 to 85 category okay I think both are totally legitimate ways to
break it up in some sense I am more empathetic for what they’re they’re
showing and this is showing the sort of urban comeback is is this flattening off
there in terms of the top 1% or the relatively good performance in the 86 to
nine ninety five percent area there’s also this interesting fact that the
population growth has done better in those mid-range density areas in the
past decade than the employment growth and again this is the source come back
to mid-sized cities it would be one of the reasons I we knew to expect that
okay is because prices are a leading indicator we’ve had for many years the
fact that there’s been this tilt and this is using Zillow data this is
proximity to the central business district this is across the entire
United States there’s a huge price tilt okay across the u.s. in towards central
city areas in terms of living and you know that’s why I think it’s always
important to combine the price data with the population data to actually show
this into show show that this is you know clearly something in terms of
demand for City space either reflecting jobs that are available I think more
probably the amenities there in the area we also see something like this in the
permits data so this is the path of single-family and multi-family permits
across time through 2006 right massive gap that favors multifamily / that
favors single family permits after the Great Recession a real balancing out
where single-family and multi-family looked much closer together that again
sort of suggests this rise of the of the central cities now what lies behind this
I think it’s interesting to speculate but that that the changes in technology
that were once thought to be the death knell of urban areas have proven to be
highly complementary to urban form one of these is the so-called sharing
economy and cities have always been about sharing what is an urban
restaurant but a shared dining room in a shared kitchen what is an urban park but
a shared backyard the difference is that these new technologies enable us to
share more stuff why didn’t you have ride-sharing in Boston in
1975 because you’d go to the combat zone to pick up your car and there’d be like
a dead body in the trunk and it’d be sort of an unpleasant thing to explain
this thing today we have the technology and you’re not finding the dead bodies
anymore and so we can share more stuff and this enables cities in different
ways this shows Yelp coverage relative to the actual restaurants across America
you can see as well like these forms of information sharing online are seem to
be deeply complimentary with urban form and so far from enabling us to spread
further over space many of the new technologies seem to abet face-to-face
contact or to abet the the communal use of resources I would also like to see in
this sort of broad discussion more more discussion of the joblessness problem so
this is the joblessness rates there the employment rates the pop rates for prime
age men by metro areas and non-metro areas and as you can see why is in the
late 1980s these things really tracked each other quite closely for much of the
past decade there’s been a wide gap where less skilled people have been much
more less likely to be employed outside of metropolitan area stand inside
metropolitan this is the the flip of David auteurs great fact that the that
the urban wage premium has disappeared for the less skilled it also appears to
be true that there are at least jobs of some form for Less kill people in cities
particularly in service areas whereas people outside of cities it’s harder to
imagine the future of their work um this fact I love this change from the
crooked smile to the hockey stick and I thought there were two things that
really puzzled me about it what caused the urban what caused the wage premium
for rural America to disappear during this time period why was it there to
begin with and why did it go away and secondly why is that we’ve gotten this
sort of strong convexity Asian what happened to the wage premium once
enjoyed by middle-income cities that was such a feature of the sort of middle
decades of this and now really is gone it’s just flat until you get to the
densest areas um you could see some of that here you can see this is their
their densest fifth fact I’m tempted to spend five minutes or so on the January
temperature faster but I will show you that I thought this was fascinating as
well I thought this was this is quite interesting of course you can see the
January the BA degree fact over there and see really the era of Richard
Friedman’s over educated American in the mid 1970s really showing up in those in
those middle middle years so there was a lot that I
wanted to know here and I think one of the things that’s really critical for me
to understand this is I need to see both wages and employment changes at the same
time I’m not asking for identification Cris just to be clear I’m just asking to
know if it’s looking more like labor supply shifts or more like labor demand
shifts in different in different ways and you know I think it’s these are
great things to know this is the heterogeneity of income by density I had
I mean I understand Chris’s eyeballs are better than mine for making sure that
this thing is converged over time over time I found this much more compelling
so and he also showed you the one for counties as well but you really see in
the bottom 85% of counties really the regional convergence going down you’re
going on you really see this sort of diminution of differences across space
in a really striking way and then it stops around 2000 and that’s compatible
with things that Chris Barry and I wrote about 15 years ago on the end of
regional convergence in the US which gonna show and have also shown this
shows 86 to 95 less regional convergence during the other during the early area
but then a flattening out in terms of the top 5% of counties this one’s going
just in exactly the opposite direction that one’s showing the sort of big
increase in diversity across areas and that really is reflecting the enormous
heterogeneity in which different cities reacted to the decline of manufacturing
and as such in some sense that sort of the lead-in for what is what is his
manufacturing paper is telling us is that you know 1971 two jokers put up a
billboard outside of Seattle asking the last person to leave the city to please
turn out the lights because just as no one could imagine a Detroit with a
smaller General Motors no one could imagine a Seattle with a smaller Boeing
right Seattle has of course magnificently reinvented itself
primarily because of its education levels but understanding that
heterogeneity is really sort of critical um
this is my version of income convergence has stalled this is 1982 2010 if you do
anything to correct for the measurement error created income convergence it
really becomes even a slightly positive shift as well so it really does appear
as if this you know hundred and fifty year period in which income growth was
faster in poorer places which barrel and Sally martine documented so compellingly
in 1992 that this era of income convergence documented just as soon as
they they document disappear just as soon as they documented it okay the and
from my perspective from the not-working perspective right
is a world of you know far from convergence this is a world of enormous
persistence so this is the not working ranked in 2010 and they’re not working
rate in 1980 this coefficient is more than one the r-squared is more than 80%
okay so just enormous amount of persistence in which places are our
jobless so understanding the end of regional
convergence seems important to me does it reflect the decreasing ability of
population to move across space and arbitrage real wage differences and what
is driving increasing heterogeneity of wages among the densest fifth of
counties this is just showing you the decline in migration and angiographic
sclerosis and we’re seeing this later in a later paper but there really is a
question about what extent this is individual preferences or the inability
to build in high high-cost areas um I did want to do a little bit more about
the things that we love about County business patterns right so County
business patterns has great industrial data firm sizes we know about
specialization we know about you know this fact that small firms this is
showing the places with lots of little establishments have more employment
growth in places that lots of big establishments these are the same facts
done in a variety of different ways it would be nice to for them to use more of
this I mean it’s maybe that’s a third paper but it seems like such a natural
thing I’ll go quickly over there work on the decline in manufacturing I found it
quite striking I found it quite interesting to sort of see that you you
move from a world in which manufacturing was declining in urban areas and
relocating to non urban areas to a world in which it was declining everywhere I
think actually the most effective graphs of this are actually not these but the
the 74 to 99 where you see New York and Massachusetts declining sharply in
manufacturing but North Carolina and Alabama first rising then falling
falling strong and then the next 20 years where Massachusetts New York North
Carolina and Alabama all of them have manufacturing
disappearing and you know change in manufacturing is strongly associated
with the rise in joblessness as well so this is the initial change in
manufacturing share associated with the rise of these things and then they also
show you this change in clerical work as well so we we have at least two good
explanations for why manufacturing employment is vanishing technology and
trade and applause evolve use in the long run all local low skill employment
will be in services and that American export oriented employment will be
overwhelmingly high skill in urban that is at least a plausible view of the
world this suggests the right kind of entrepreneurship like uber will be able
to generate jobs for less skilled people Boston but what the heck are they gonna
do in eastern Tennessee that seems to me like the central question of American
act not America’s economic geography and even for American economic poll
employment policy in the 21st century and you know don’t forget mill
joblessness is not some benign leisure choice right male joblessness is misery
it’s associated with you know opioid deaths again showing up in this eastern
this eastern seaboard it’s associated with a large fiscal externality federal
federal government spends are much higher when men are jobless it’s skip
over that is its associated with misery there’s you know the income gradient
with happiness is just very different from the level of an happiness
associated with prime age men not not working and particularly true if they’re
not you know they’re living alone as well and you know the it’s not
surprising this that a job is more than just an income it’s a sense of purpose
in life and it’s a social network that actually keeps you sane so just to end
that a couple of policy thoughts regional heterogeneity is not new and
they make this very clear in this paper but joblessness is a new twist
and if joblessness involves market failures either pigouvian externalities
like these fiscal externality or various Keynesian stuff that I have never
understood then this should lead us to rethink regional policies again and I
think maybe the central divide is do you actually believe in regional
redistribution right using region as a tag to give to give
money or regional targeting of Social Policy I think the argument is much
stronger for the latter than it is for the former but in any case there are
good reasons to think that America is becoming less fluid geographically than
it has been in the past there’s a sclerosis of America’s economic
geography and the traditional hope which was that poor places would become rich
by my people leaving doesn’t seem to be working partially because the out
migrants are so much more skilled than the people who leave and this is what
you’re seeing in this graph the 45-degree line is to show you the gap in
less skilled places where the out migrants are leaving those areas even
more denuded of skilled people which if you think human capital is the bedrock
on which local individual and national success rests
makes things even more problematic okay I’ll skip over that is geographic
sclerosis an excuse for revisiting place-based policies it can be the less
people are likely to move the more plausible it is to actually give
place-based aid of a variety of different forms the you know the less
likely place-based policies are to create pockets of high unemployment and
local human capital but it is still true every time
I say anything that’s not absolutely deeply hostile to anything playspace
Jason Furman will send me an email and tell me that I’m being a jerk and I’m
opening the sluice gates for billions of dollars of wasteful infrastructure
spending in places where it is inefficient and that is absolutely true
you know we should always just wake up each morning and remember de troyes
people-mover monorail and remember that when you tell government that you know
placemaking is a good reason to spend a billion dollars on infrastructure that
will never recoup its cost for user fees just think for a second just be worried
a little bit about your possibility of making a mistake I think the case is
much stronger for spatial targeting I have long thought that it is not just
unwise but just crazy to have the same types of housing policies in Boston San
Francisco Houston and Detroit right these housing markets could not be more
different it is totally plausible to subsidize housing production in Boston
with a marginal value that consumers put on housing is vastly above the physical
cost of construction right that gap does not exist in in Houston where the
private market is a great job of supplying stuff and there’s no reason to
subsidize housing in Detroit where the valuations far below construction costs
consequently we want different housing policies in different places there’s no
reason not to think that you also want different employment policies in
different places right and that may mean that if you just use a modified Bailey
Chetty formula that you want to be kinder to people who are who are jobless
in Seattle Washington or Boston right because there are fewer people on the
margin between working and not working but you know in places like West
Virginia where joblessness is is more endemic you want to do more to encourage
working so you may not want to engage in any redistribution but you may just want
to make our policies more Pro work in areas where joblessness is high at least
of thinking about whether or not we want to have spatially targeted employment
tax credits vacations from the payroll tax and high high jobless areas or
having different disability rules in West Tennessee and Boston and I would
just end by saying I’m so glad that we’re doing this topic it could not be
more central and more important and I’m so grateful for Chris and then forgive
me the chance to discuss their fascinating paper thank you thanks dad all right so now we have 30 minutes for
converse comments and questions from from the audience and we have a couple
of microphones going to be passed around by Ras and each of the aisles there so
go ahead and raise your hand if you want the mic and please make sure to before
you state your question or your comment to make sure to tell us who you are David Schleicher
yeah law school um how worried are you that about use the use of counties as
your kind of variable counties are systematically not the same in different
metropolitan areas they’re big much larger in the West as you go west across
the country and that could introduce you in your studies of population densities
and think some regional changes okay let’s go with you get a few hi Brian
Stewart from George Washington University I think one of the things we
really want to think about is differences in say earnings produced in
a place versus earnings received by residents in the place it’s the BDA data
at least from 1969 forward allow you to make some progress on this so just your
thoughts on that September night from India our Prime
Minister has recently introduced a program of y-space is important 100
smart cities and government allocated a specific amount for each city to be
developed they are not Metro based but they are
geographically spared in order to develop employment and also private
investment the second aspect of Indian development which is a large region is
giving tax credit in less and developed areas of the country so that it is not
market driven in order to allow market to keep investment but also government
spending so it is both the carbon spending smart cities and tax credit to
have a barkat driven growth within the regions which are less employment
generating where there is more poverty why don’t you respond that actually
would you want to respond to that now okay so I get her some zip code
tabulation areas that are very very very small compared to the the rural ones we
also wanted to consider that we wanted a unit that would cover the entire United
States so that we could do spatial econometrics and there are problems with
converting zip codes into zip toes and there’s also the MSAs and CBSA’s do not
cover the entire country and then the other thing is we wanted to take a
really long-term look and that was the advant
the the definitions of MSAs and CBSA has changed over time the zip ties only
really started around 2000 and with this County business patterns data it allows
us to go back to the mid 60s so it was just the best thing the idea we should
look at earnings produced in a place I think I actually did as part of the
project was little bin scatter plots I was skeptical like why is it too early
on in 1964 that the one or two percent tiles is actually making more you know
than some of the more dense counties and I replicated a lot of what I did with
the beatings by place data or sort of where the earnings were received and
sort of compared is where the people lived and I got very similar pictures
but I think that there would need to be more robustness checks along those lines
for exactly the reason you say for some of the other results in the paper I will
take on Prime Minister Modi so on one level I completely cheer for
any reorientation of Indian political economy to be supportive of cities right
the the basic structure of Indian particularly state level government is
so strongly oriented historically towards rural areas I mean primarily
because the voting rules where you know voters in rural Maharashtra would often
have 10 to 20 times more more elected officers per capita than voters in
central Mumbai so the
urban reorientation is is a great thing I think the real questions are about the
actual delivery of the smart cities program right so it’s that the the
vision of doing something for urban area is is great I think in terms of my you
know my conversation with the Indian government the problem has been more in
sort of on the ground and sort of making it making an operating I think that’s
very much where these where these things come down um in terms of tax credits
again my starting point is always spatial neutrality right I always think
that the right place begin to begin with is that the federal government should be
spatially neutral towards particular areas until you can make a case that
there’s sort of a compelling externality that is pushing you towards favor in one
area the other like getting rid of joblessness so reducing joblessness in a
level and you can convince yourself that this tax credit is a reasonably non
distortive way of actually achieving that so I would generally be wary about
locational tax credits but in a case of Indian history anything to sort of
embrace the the urban India of today rather than clinging to a rural vision
of the past as is devoutly to be wish’d great let’s take a few more I’m Katie Russ and I love the CDP 2 at
the same time when we take manufacturing chairs just as a total of the employment
recorded for a county in CBP we really under count agricultural production and
so we can get a distorted view of manufacturing shares between especially
for some rural areas and this bites in particular we’re trying to look at
patterns over time because there’s been such a big shift you know out of the
agricultural sector so I just want to kind of put that note out there for all
of us who love the CDP Katherine do you ever fix do you have a recommendation of
what to do about it yeah so instead you can use the total civilian labor force
in the denominator and re-weight that way cool thank you I David autor MIT so you know the the
question that Ed asks at the end is you know what are all these guys and in
rural Tennessee gonna do and I think there are two things to think about
going forward that sort of I don’t know the what they imply but I think they’re
important one is of course rising care needs right our population is rising
very rapidly rapidly we have a short you know small cohorts entering the labor
market they’re actually highly educated meaning that they’re gonna be less
inclined to do service work and of course our restrictive immigration
policies are adding to that so there’s gonna be enormous need for care both
actually in the places where male joblessness are high because that’s
where there’s a lot of elderly citizens um
the other factor that sort of bears on the geography of work and that I don’t
fully understand but you know the the death of distance has not occurred for
the high skill right cities have become better and better place for high school
but I would argue it may be occurring much more for the less educated a lot
more work that doesn’t need no longer meet a lot a lot more mid school work no
longer needs to be done face-to-face right can be outsourced overseas it can
be done remotely in you know back office locations and again that seems like that
presents challenges like I think the sort of polarization of employment in
cities is quite challenging but it also presents opportunities I mean it may be
possible for more opportunities to be the you know move to places that are not
terribly expensive and not extremely highly educated so in sort of thinking
about this conundrum going forward I do think those are two kind of accelerating
forces that will affect policy responses and just the strong undercurrent of
demographics David Newmark two suggestions are a
suggestion in a question I guess the suggestion is when we’re looking at
those those long-term regressions with the CVP data I kind of wanted to know
what’s what what’s the role of the individual control and what’s the role
of the location and of course when you just do is the CVP data you can’t sort
those out and obviously you can’t do CPS by County all the way back but you can
do ACS doesn’t go back very far but at least you know since you’re doing sort
of long-term the census year is plus the ACS years would let you do that I think
like in a horse out those two things I think I’d be very interesting because
there’s you know there’s compositional changes and there’s this whole the
manufacturing story you have for example is it you know being a manufacturing
worker or being near manufacturing or being near but not too near
manufacturing like that that’s driving things and the second thing that strikes
being in a lot of this discussion of male joblessness is we really are we –
what I said are we getting different male and female responses I was struck
out on yourself a few weeks ago an article in New York Times one of those
long Sunday ones about about coal country right and what are people doing
as the coal jobs disappear every sing this is not data I realize but every
single story was about a wife going back accessing various kind of programs that
gave her access to training community college and taking on some jobs and they
were care jobs to David’s point to some extent though there wasn’t one story and
it may be because the reporters actually couldn’t find anyone about a male going
back taking a job lower paid in the coal mines for sure but nonetheless a job to
make up her family and kind of animation to what extent it’s the men being
resistant to take the kind of jobs that are out there
lawyers not hiring them to what accept that pledge roll with some of this and
how different things look let’s say for women than men in these places
okay here’s one of under those I would just say on the women versus man you
know if you look at prime age male or prime age participation in the United
States over the last few years it’s remarkable that there’s in the u.s.
there’s been a recovery of prime age participation among male among women but
much less so among males and if you were doing macro labor stuff at the Federal
Reserve Bank of Boston you spent a lot of time trying to distinguish trend from
cycle in that and you know is it the fact that the men’s participation for
the nation as a whole sort of has flattened out is that actually rising
relative to a very steep trend the downward I guess this is to say there’s
a lot I think that we don’t really understand about how men and women are
differentiable differentially able to take new opportunities whether the
expansion of the service sector for example I think David has done some work
on this is much better for women because they’re just this is a more natural fit
in a way that we don’t really understand in economic so I would say obviously we
didn’t do a lot in that paper but the difference between labor force
participation and for males and females is certainly a big topic that we’re
concerned about now nationally at the Fed yeah so just just to for David so
you’re absolutely right of course I mean the dominant industry and declining
parts of America is health care and social assistance right which which
stays there when all of the exporters leave because the feds pay for it that
is certainly one plausible thing is that that will be the economic hangar but
that is still kind of a dark anchor right I mean it’s not an anchor that’s
particularly dynamic as an actor that’s correlated with future decline it’s it’s
it’s so but it’s certainly right I you know you’re certainly I mean so the
death of distance has been bad for some low end services but not for others
right so you know anything that involves you know actual interpersonal contact
and being being pleasant and taking care of you know lawns and doing stuff like
that that’s those are low end jobs that that can persist and you know if
you know many many high school people I would like more to outsource more of the
painful things in their life to to low skill people so that feels to me like
the future in in many ways ideally but I I do fear that you’re upside which was
the sort of death of distance leading to more outsourcing of the middle range
jobs about the office work that certainly is right currently but doesn’t
part of you also think that if it can be outsourced it can be mechanized and it’s
likely if that you know right now it moves to a suburb of Rochester and gets
done in an office park but eventually we figure out either how to do this in you
know in in India or more just as likely that it’s a computer simulating a human
voice doing it since we’re requiring so little from this so that’s that’s
certainly my fear is that is it basically any sort of low-skilled job
that’s that’s not involving interpersonal contact that’s not
involving delivering sort of pleasure in a one-to-one connection that doesn’t
that feels to me like that job just vanishes from the planet at some point
in time because it gets done completely by machines so I wanted to come back to this crooked
smile idea and I just wanted to note the mind look at it there’s obviously two
findings there not one right you have besides both of them and one is the fact
that these large places seem to be pulling away from places that are still
as that noted very big right in the top 10 percentile or 15th percentile of
metropolitan areas and they’re pulling away then you also show the
interquartile range that’s so actually that some of the very
big places were pulling away from other really big places and the question
becomes whether it’s one thing whether there are certain cities that were well
positioned is that it is suggested and it just happens that the well position
cities tended to be larger and there’s good reasons to expect that or is there
actually two things going on is there something some structural
reasons why certain large cities could succeed in certain ones couldn’t and
different structural reasons why middle sized places aren’t doing well and you
can even think individual levels of inequality there’s a similar question
right in the very large places well royal places have been left behind in
very large places we’ve seen substantial growing inequality between the wages low
skill at workers earn and hot what high school workers these big successful
places are earning and again the question becomes is this simply a
general decline in the general increase in the return to skills and a decline in
the demand for low-skilled labor or is there something different about I’m
structurally different about the fortunes of low-skilled people possibly
because of the growing demand for services and one-to-one interaction
between low-skilled people who are falling behind in cities places actually
if I could just jump in and answer that right I think that’s exactly right that
there’s a lot more that we could do statistically to sort of tease out all
these nonlinearities and I think that’s when we would bring in what’s going on
in commuting zones what’s going on in metro areas you wouldn’t we wouldn’t
just use counties at that point we’d say all right what do we get what do we get
for a different level of aggregation and I
help get at that issue but you’re exactly right at keene with the
observatory who could the speakers talk about the role of transportation
infrastructure transportation access for airports highways in in explaining the
trend that you’ve been opened has been discussing is there a question over here got this
over yes it’s bright with youn with Tufts
University the question you mentioned sort of tangentially the idea of
externalities and I think that my sense of it if we just sort of focus on you
know unemployment among prime age male that’s not gonna resonate in terms of
thinking about one of the positive policy solutions so what I’m thinking is
is there any way to measure these externalities meaning you know higher
health care costs higher addiction problems a higher law enforcement
incarceration followed by recidivism so those those things if you start to flesh
out the let’s say the particular burden of cost or externally externality cost
then you start to build a case that has some credibility all right we’ll take
one more gentleman over here I used to work for the FDIC have now
retired regional economists of course I face this but part of my job portraying
the individual experiences across New England which varied as you point out
greatly in doing so I found over time I began to move away from the labor market
data basically and used more the data from be ei the statewide data on GDP
which has remarkable amount of detail at the industry level it only goes back to
2000 admittedly but it nevertheless is extremely helpful because you changes
the nature of the discussion it basically reveals the individual
performances and the dynamic quality which is basically about technology the
development of technology and you can then suspend the story about global
supply and in the part of this what it is essentially is a success story and
the success story involves also casting aside a good deal of the industrial
structure that had been put in place over a considerable period of time the
remarkable thing about this area that’s been able to reinvent itself and go back
to some real core values and core capabilities that had way back in a
historical period so it is a mean I’m talking what I’m talking about its
context I don’t question your use of the data but beyond those data set there are
other data sets that can provide a lot of elimination I mean up to say a couple
of things on transportation I’m not really an expert on sort of I’ll just
say this when we first looked at the manufacturing decline in urban areas in
the CBP in the from since 1964 on I was stunned because you know I hate to admit
this but I thought that a lot of the manufacturing had left
urban areas in the 1920s and of course you know several people William Julius
Wilson others had had commented on that there was a migration outward in
the 60s and the 70s that had particularly negative effects on
communities that in that had to worked in those in those manufacturing jobs my
understanding is that a lot of that was in part due to improve transportation
the container ships and the trucks that would move the containers you can tell
I’m not an expert so I I would not be sort of able to speak so much on the
transportation effects but I do think they probably played a big role in some
of the patterns that we saw on the other part on using other data I think you’re
exactly right and the one of the things that we found sort of along the lines of
what you’re talking about in New England was that I think edge showed this graph
we have it in the paper we didn’t show it but during the Massachusetts miracle
manufacturing employment was actually declining in in Massachusetts as it was
in New York if that gets to the issue of what was going on at that time where
they sort of being pushed out because Massachusetts was such a great place to
develop other types of industries or were their trade shocks or other stuff
going on that was hitting manufacturing in in Massachusetts relatively intensely
and perhaps going to other data sets and things like that could potentially
answer that question I mean transportation than talking a second
about externalities the the transportation is of course deeply
powerful in shaping American regional changes I think the thing to be wary
about though is thinking that when you see a distress region the right thing to
do is to throw more transportation out at it I mean after all that was one of
the big things that the Appalachian Regional Commission did in the 70s in
the 80s with very few perceptible effects on the long term social problems
and indeed you know if any of you were interested in where the cutting edge is
on sort of estimating where the biggest economic bang for the buck is in
transportation there’s a recent Allan in our calacas paper that’s terrific that
does point out there are huge places for America to invest in in infrastructure
most of the highest return roads are actually in the New York metropolitan
area though those are the areas where in fact expanding lane miles would be worth
a huge amount in their in their estimates and you know just just it is
just so easy to waste billions once you start wrapping sensible transportation
investments up in the map of you know this can bring Detroit back
so you know measure measure the actual economic benefit secondly III just want
to I want to dispute the premise but then agree with you
so I want to dispute the premise which is in fact I have never had the
slightest problem when discussing with anyone in a policymaking role of arguing
that the rise of joblessness in America is an enormously pressing problem I have
never had anyone pushed back and say they know that’s obviously irrelevant ed
why would you possibly worry about about the fact that 26% of men in this county
are not are not employed why would that what and so I just think you’re just
completely wrong on this and in fact explaining externalities actually makes
things worse not better that being said in terms of the scientific agenda I am
totally in favor of measuring the action now as we do this some of my paper with
then Austin Larry Summers we try to measure it straight out fiscal
externalities and come up with some other numbers along along with other
sides and in terms of figuring out what you think inappropriate pigouvian
subsidy for work should be actually that hinges on what you think the
externalities are I will just give as a like a tidbit of like a quandary that
I’ve had around externalities are lots of these non employed men are being paid
for by people who are in their larger social network right parents you know if
more than 30 percent are living on their parents couches to what extent do we
actually think about that as being an externality or something that’s entirely
internalized to the position of making a making a decision it’s not a trivial
answer so I’ll just I’ll just end with end with that all right I think we do
have time for one more round of questions to me like you can’t
classified county density each year so you might have some fluctuation from
year to year and I wonder if you’ve thought about those counties that move
between categories and you know some of them the most flourishing may rise above
their category and some of those that are those troubled may fall lower in
population densities so I’m curious your thoughts on that actually let me jump in
on that real quickly yes we did and actually we spent yesterday taking out
all of the slides that had well if you do it this way and this year and that
way this year and they were paying pet who was painful for me to do that
but yes we had like 30 pages in the appendix that does that right but on a
serious note one of the reasons we wanted to do sort of a basic density
thing is because the definitions of rural change over time and we you know
we’re afraid of getting caught up and okay this is what’s happening to
suburbia this is what’s happening to rural America and then trying to figure
out how to sort of maintain some constancy over time so the idea of doing
density sort of the density flattest they were they’re fairly stable we could
sort of analyze things and though and do some robustness checks miraculously for
the sort of suburbanization stuff that we did it really doesn’t matter at all
whereas if we were doing population shares or other types of things that it
does so that’s a long way of saying yes yes we did worry about it I just able
Tufts University and you’ve been convinced to run for president in 2020
for president Glazer doesn’t that sound great so you have to pitch to the Trump
voters right these are the voters in these areas that have lost manufacturing
with high unemployment right what’s your pitch to them in terms of bringing back
jobs because you know targeting employment tax credits only work if
there’s more employment so how do you how do you convince them what are your
policies to bring jobs back to these for these jobless areas you are you are only
reminding me of why it is that I’ve got no business being anywhere near politics
yeah it’s certainly it’s certainly right and it hits the central quandary so as
an economist I believe that we do to more to encourage work and encourage job
creation and remember you can think about the employment tax credit as being
implemented the firm level as well you can imagine getting rid of the payroll
part of the payroll tax that firms pay workers which it should be relevant to
our models if there’s no constant downward constraint on wages if there is
a downward constraint on wages then it will actually matter the I would wrap it
up with some getting rid of the bad government policies that have you know
prevent a job creation and you’re obviously worthy in a highly productive
area right I mean that’s that’s that’s I think where I would go with this but on
what but whatever but at some basic level it is a leap of trust that if we
actually generate incentives for job creation that jobs will actually be
created and I don’t know because I can envision I mean we can see what the jobs
would be in Greater Boston right I mean I chaired a committee to try and create
with with John Paris to try and create and entrepreneurship zone in a high
poverty area in Boston of the thousands of things that lower-income kids could
do in Boston that would actually bring them bring them some form of a brighter
future whether or not working in service industries are being part of the
Boston’s export machine but I have a lot more trouble saying these are the jobs
that will come back and part that’s always the problem in pitching economic
ideas which lead to more innovation is you don’t know what it will mean to
innovate and that’s in some sense why backward-looking policies by
protectionism are so politically appealing because everyone knows what
jobs you’re trying to protect but no one knows what happens if you get rid of
those things and and try to create a world in which new jobs were created no
one knows what entrepreneurship will come come about and so I guess I would
say something like not giving up on the spirit of American entrepreneurship not
giving up on me on the ability of your region to come up with new ideas as it
once did that changed the world and it changed the economic future of America
so I would try something like that that you know finally if you go to space
I remember a battle Oh Lin it is not international trade
it is inter-regional trade and every region has its factoring or endowments
which are not homogenized we try so we are homogenizing lot of factors and
endowments and then leaving it to the market economy I mean it’s it’s it’s
become now ok the overall growth is ok but now there is a spatial problem so it
is a responsibility of the counties in the States to to find out the and to
ascertain the regional endowments which are comparative in domains which are
comparative cost beneficial and which can generate income and once you hand
hold the private investment will come I mean finally said it’s an idea but it is
to be supported at the county and state level and then the market will come and
market cannot always go to opportunities everywhere market will go where the
rates of return are highest and not the otherwise no so that’s why now that we
have reached a stage in our economic growth and a cycle especially the
developed economies like the US and others there has to be in order to get
over this problem of employment there has to be heterogeneous in the factor
endowments and generating employment and income in small small places which has
its own individuality so I’m a big supporter of eliminating the barriers to
interventional trade in India so I start mr. court I certainly support the Modi
the Modi government’s didn’t work on the on that great alright we’ll go ahead and
take a break and as a reminder there’ll be coffee out here restrooms back there
and we’ll reconvene or the kick off the next section at 10:45 you sure we’ll have 30 minutes for that and
this is a paper be presented by Jay Shambo and his Kath is co-author
Katherine Russell’s up here on the podium as well and the discussant will
be David Otter so with that I’ll hand it over to you Jay thanks very much I’m just starting the
timer I could talk about this all day so I really want to sincerely thank be
organizers of this conference for asking us to work on this paper for it it’s a
fantastic lineup of people and I’m really honored to be a part of it
what we want to talk about today is a paper on education and unequal regional
labor market outcomes and one of the reasons I was so happy to be on this
program especially after seeing who else was on it was that many of the other
people on the program have been really influential in what I’ve been thinking
about in the puzzles I’ve had in my head for the last five years and I mentioned
to David beforehand in in some sense it has summed up for me is how can a world
where blonde chard and Katz’s results exist that you get ordered or an
enhances and I’ll explain what I mean by that kind of as we go along so the paper
we’re going to talk about an end to convergence there’ll be some pictures
that actually look not unlike some that Edie put up and I’m guessing other
people will put up similar ones over the course of the next couple days
but I want to talk in particular about a growing persistence in regional labor
market outcomes and then talk a bit about trade shocks and how trade shocks
might be part of what’s going on here and then talk a little bit about policy
at the end so one thing that’s important to note is this this paper is really
kind of pulling at a lot from two different previous papers so one with
co-authors at the Hamilton project Ryan Dunn and Jenna Parsons in a paper we had
there called the geography of prosperity and then a lot in particular of the the
back half of the paper is work Katie and I did with co-authors at UC Davis
Katherine Erickson and face it’s you so I just want to be clear we’re borrowing
a lot from our co-authors in in what we’re doing here so the conference here
title is a house divided the house always been divided and I think that was
mentioned last time and I think it’s just really important to emphasize over
and over how big these regional gaps have been and how they in fact used to
be bigger and so it’s not like it’s something new that we have regional gaps
across places they’re huge today but they actually used to be bigger there
you can look at a whole set of literature talking about this
convergence over time cite one here Michener and mclean showing you know you
have this huge convergence over a hundred years and a lot of it coming
from increases in labor productivity in the places that used to be much much
poorer but then as a number of people in this room have pointed out that seemed
to stop at a certain point sometime around 1980 that stopped and so just
looking at this one way you can visualize this is looking at per capita
incomes of places different regions as a expressing them as a share of the
national income total and so you can see the south down here at you know 50
percent of the national total at one point when you had other places that
were well above the national so you had gaps almost two and a half times income
gaps across regions and these things just converged in a huge amount over the
next chunk of time some of that may have been policy and we electrified regions
that didn’t have electricity things like this but a lot more of it just seems to
be the general course of what the economy was doing but then sometime
around 1980 this seems to stop and we get this if anything slight divergence
opening up where you start to see places like New England clearly the Boston Fed
has done its job kind of taking off back again away from the rest of the country
and you can even see at the end there the southeast starting to dip back away
in the wrong direction again and so we’ve got this big
convergence you can look at it here in terms of income but you could look at
this in any number of different statistics if you agree with that the
kind of the prime age employment rates or participation rates are really
important if you look at the top 20 percent of counties today prime age
employment rates are about 15 percentage points or 16 percentage points higher
than what it is in the lowest 20 percent of counties in the Great Recession
nationally that fell about 4 percentage points so it’s almost like they’ve had
you know three or four Great Recession shocks in the lowest counties relative
to the top and we think what we did kind of as policymakers trying to deal with
the Great Recession it does feel like a lot should be done to deal with the lack
of employment in some places here’s another way to see that convergence
ending this is looking at counties and so this is median household income in
counties from the 1960 to 1980 picture looks like a perfect picture of what you
think convergence should be the counties that were the poorest in 1960 were
growing fastest over the next 20 years the counties that were richest in 1960
were growing slowest so you had this big convergence happening you flash-forward
to 1980 to 2016 and it’s basically a flat line if you look the the red
regression line there has a very slight downward tilt but if you actually run it
with weighted regression by population it’s dead flat and so we just don’t have
convergence anymore in terms of income you could look at this more broadly and
we did at the Hamilton project where you kind of pull together a whole bunch of
things pull together employment rates and household income and poverty and
life expectancy and kind of put them all together with a factor analysis and and
ask our counties doing in at different points in time and if you break them up
by quintiles the really depressing thing is if you weren’t doing well in 1980
you’re still not doing well today 71 percent of counties that were in the
lowest quintile are still in the lowest quintile and in fact 92 percent are in
one of the two lowest quintile and the exact opposite happens in that lower
right if you were in the top you’re still in the top or quite likely and at
least you’re probably in one of the top two and so what you’re seeing is again
so it’s not just income across a range of measures if you pull them together
you’re seeing a real stagnation in regional gaps places that are doing well
stay doing well places that aren’t arts so there’s just very little upward
mobility for counties so what I want to turn to then is think about the
persistence of labor market shocks and sometimes economists you see them
complain that you can’t publish null results this figure suggests otherwise
because I think it’s one of the more influential figures this is our
recreation of what was in Blanche art and cats witches they were showing from
night that what your unemployment rate was in 1975 just had zero predictive
power what it was in 1985 and so you just see
you don’t get so this would say you know we don’t worry about beat smell shocks
cuz they go away right and I should be clear there was a lot more in that paper
than that there was some really fantastic work showing how mobility how
much mobility does versus how much employment does and things like that but
this was I think really influential to a lot of people thinking about the United
States being different than say Europe in terms of the way regional market
outcomes labor market outcomes evolved and and other people have done a lot of
work bound in Holts I had a nice paper looking at this and for the 80s overall
we started to look at this again and this is where are saying well if this is
true then then how do you get these Auto Doran enhancing results that the China
shock has really persistent labor market outcomes and so we just started to roll
this forward and conveniently so 76 is where this data set can start
consistently forward so we just use 76 to 86 conveniently if you keep using the
sixes you don’t run into a recession so they make a nice set of decade jumps to
do but if you flash forward to look 86 to 96 all of a sudden there’s a bit of a
slope there it’s not a cloud anymore and then 96 206 again you start to see
something and if you squint or if you read the paper you can start to see
there’s there’s actually a nonzero coefficient there and the r-squared
starts to get up into the point to range instead of the zero range and so
suddenly you’re starting to see actually knowing what the unemployment rate was a
decade ago tells you something about what the unemployment rate is today it’s
no longer kind of washing away if you flash or to the most recent 10-year
period we’ve got here now actually that slope is around 0.6 if you knew what the
unemployment rate was in 2006 you’ve got a pretty good guess what it was in 2016
shockingly if you just run this over 30 years from 86 to 2016 you still have in
our squared around 0.3 and a slope of around 0.3 the unemployment rate shocks
aren’t going away or I shouldn’t say shocks the unemployment levels aren’t
kind of converging back to the same level nationally the way they were prior
so and I should mention there’s a great paper by Dow for sehri and Lou Connie
that that looks at a whole range of the results in that launch
in cats paper kind of rolling them forward that I highly recommend there so
one of the things we were trying to think about oh and I just did want to
mention this is true at the county level too so if you’re looking from 96 206 or
Oh 6 to 16 at the county level you again you see this you know you get a fairly
strong pattern of the height the unemployment rates get stuck once
they’re somewhere and if you are reading the the top there you can see the
persistence got stronger in the second of those two figures so what happened so
on the one hand you can talk about mobility being down I’m not going to
talk about it too much because there are too many people who know a lot more
about it than I do in this room and I know there’s a whole session on it later
in this conference and so in particular we know that for workers with lower
levels of education mobility is not as high we also know that there are a whole
range of stories about barriers to mobility so is it that we won’t build
housing in the booming areas anymore kind of Agra non ganang and show AG
story and so you just can’t workers can’t move to the booming places or is
it more of a story that you don’t have as much incentive to move anymore kind
of thinking of David otters AE a lecture from passed in January thinking that you
just don’t have the returns if you’re not a high education worker to move to
these expensive cities anymore so there are a lot of pieces around this
and other stories that might make you think part of what’s going on centers
around education right so bound and halters results when they were looking
at that kind of bunch art and cat story one of the things they were finding was
that workers with less education are less likely to move after a negative
labor market shock and so you might worry that what’s going on is low
education places something is different there again with with the auto 2019 work
on less of a premium so they have less of a reason to move somewhere else we’ve
also got the kind of second paper I was mentioning this paper draws on the paper
with Erickson natal that the China shock actually hit areas with less education
and it hit them harder so again you might think part of what’s going on
maybe the persistence is just in these low education areas and bloom it I’ll
have a paper on the China shocks showing there was a quicker pivot after
for the high education places um a lot of how I’ve thought about it comes back
to one of my favorite papers I’ve ever seen which is by John Skinner and Doug
stager that talk about great scatter plots it’ll show you that the same
places always pick up innovations it’s not that they’re doing them but the same
places that used hybrid wheat first also used beta blocker technology for heart
attacks first and so it’s just the same places grab the best innovation that
comes out and employ them over and over and over and over and so and part of
that connects back to education so it made us think that’s got to be the story
so we decide just to look at the persistence across education and there’s
no difference and we did this with regressions with interaction terms and
all that there’s nothing there it’s if you look at the highest quintile of
college education and asked what the persistence in in those places are and
you look at the lowest quintile of college education and look there they
have the same slope the same r-square it’s if anything there’s a little bit
more persistence in the places that have high high school completion relative to
low high school completion and so it’s not the case that actually the
persistence of unemployment is just in the lower education areas which got us
frankly very curious as to what was going on and one hint you can start to
see is when you look just at the x-axis here right and you see that those high
education places they’re all clustered at low levels of unemployment rates and
the the lower education places have a lot more places out there in the high
unemployment rates and so it seems like they’re both persistent in some way but
they’re persistent differently and so we thought about this some and then I think
it was Katie who had the idea that we should be looking at at transition
matrices to think through this and so that’s what we tried to do and so what
I’m showing here is if you imagine that five by five grid I showed you before
we’re just gonna focus on that one one and five five cell so if you were in
kind of a bad outcome at first do you stay there or if you’re in a good
outcome at first do you stay there and if you look at all counties in the first
column here is 1970 to 80 in the next is 1970 to 90 and what you see is about
half the counties stay in the worst unemployment outcome or in the
best unemployment outcome if we start to split that though by education groups
and so this first set of rows are for the higher education places or places
with a population with more education and the bottom rows they’re for the
places with populations with a lower level of education and what you get is
that you start to see that in particular from nineteen seventy to ninety seventy
to eighty there’s a bit more mixing up going on and I think in some sense
that’s what you were seeing in the bond chart and cats results as well but what
you start to see is that the high education places if they are in a good
unemployment rate state they stay there they are highly persistent if they’re in
a high unemployment rate bin though they don’t necessarily stay there they’re
actually reasonably likely to bounce out of that state and the exact opposite is
true for the low education places for the lower education places if you are in
a bad state you’re you’re staying there you don’t get out if you are in if for
something good happens and you have a low unemployment rate you don’t
necessarily keep that low unemployment rate and so that’s the idea is that the
average persistence for these different types of places is actually similar but
the persistence is incredibly different one gets persistence for a good outcome
the other gets persistence for a bad outcome you can roll this four to the
ninety six to sixteen groups and you get a similar type of story where the higher
education places are more likely to stick in good outcomes and the lower
education places are more likely to stick once in a bad outcome so that
persistence has a different flavor to it you can I think this figure on the the
left here is the one that really stuck with me which is this is asking if we
again take those kind of four groups the high education being either you have a
lot of college graduates or you have a good high school graduation rate low
education the opposite and we asked in this figure on the Left how likely were
you to be in the counties the quintile of the lowest unemployment rate so were
you in the good unemployment rate state and this is the bins don’t change here
this is for the groups based on their education in 1970
in 1970 it’s like you’re all around 20% which is the unconditional odds right
you’re all it’s just if are you in the lowest quintile well 20% chance that
you’re in any quintile in some sense overtime though this divides really
really sharply and you start to see that the lowest high school dropout rates
start to have a 40% chance of being in that really good State
but the yellow line is the incredibly depressing one that’s the places with
high high school dropout rates fewer than 5% of them are in the kind of good
unemployment rate state by the end of the sample you’re just if you have a lot
of non graduates from high school you’re just not in a good employment market
it’s just that’s how it’s it’s divided itself and that didn’t used to be true
as starkly if you instead asked how likely are you to be kind of in the bad
state you see a similar division the divergence isn’t quite as stark because
they started already a little bit further apart but what we’re seeing is
in some sense while in those income statistics we saw kind of a flattening
out maybe a hint of a little more divergence here you can really see the
greater divergence showing up in how places with different levels of
education are reacting to are getting stuck in different labor market outcomes
so what I want to then think about then is how trade shocks might be playing a
role here so first of all I should just state that almost certainly part of what
this story is is a story of skill-biased technological change that technology
that is augmenting the labor returns for high skill or higher education people
and then also perhaps either reducing or even replacing labor for low skill then
if that happens then you’re going to see patterns like this and I think that that
makes a lot of sense we make an argument in in the paper with Erickson and
co-authors that there’s another interesting part of the story maybe the
way trade shocks are hitting the United States and that it’s a different type of
trade shocks than we used to see and we motivate this a little bit thinking
about the product cycle and we actually borrow from Paul Krugman’s formulation
of that in his 1979 paper and the model is an international trade model and it’s
thinking about you’ve got places that are higher education
they generate innovations and new products and manufacture them for a
little while but then over time those products become more routine in their
manufacturing and as they do so then you shift them over to lower education lower
wage places and the story is you can see the exact same thing happening in the
United States you get the same product cycle playing out regionally within the
United States and that you get in some sense a shift if you’ve had that shift
take place over time though it means that when manufacturing trade shocks hit
the United States they’re going to be hitting different places today than they
used to hit so one way you can see this if you just look at manufacturing and
look at what does manufacturing correlate with in terms of what
different commuting zones look like it used to be true say in 1910 that if you
had a lot of manufacturing then you were a high patent place in you’re a high
education place rolling forward to 1960 it’s still true to some extent but not
as true on the education front and by 1990 there’s a lot less coral a lot less
strong correlation with patent activity but one of the things as manufacturing
overall is a complicated set of industries right there are high-tech new
products cycle kind of industries mixed in with the late-stage product cycle and
so one of the goals then is to think about well how could we identify the
late-stage products and you can’t just identify them based on them being
produced in the low education areas because then you’ve just kind of assumed
your answer and so what we do is we we lean on our discussant and we use the
auto Dorn enhance and trade shock and we say if China was exporting goods to rich
countries in 1990 those were late-stage products and we instead look to see how
those products evolved over time and what you see is that they in 1990 the
China shock industries were incredibly high education high patent places over
time they became less so and then by 1990 though it’s entirely reversed that
the places Manufacturing the goods that were hit by the China shock were highly
negatively correlated with education at that point if you look at this in
pictures just because they’re fun the red are the places that are highly
exposed to the truck China shock if it had happened in the decade I’m showing
you so in 1910 it is the high education high innovation
places of the United States if you move forward in 1960 that’s still
largely true although you see that activity has started to move down some
by 1980 it has really moved and by 1990 you see it’s no longer a Massachusetts
phenomenon it’s no longer a Great Lakes manufacturing phenomenon it’s now moved
all the way down here this is why if you say the China shock hit the Rust Belt
David will always correct you and say no it didn’t it hit down there and so what
we’re saying is that when the China shock hit it’s a very different trade
shock than we used to get it’s not hitting the high education places
anymore if you instead look at the Japan shock until we recreated in the paper a
shot the parallel to the China shock but for Japan’s exports to the United States
they never moved to the low education places so when the Japan shock is
hitting it’s hitting higher education places places that are more likely to
escape a negative shock than the places that got hit harder by the China shock
and I should note here since I didn’t on the earlier slide what we’re doing in
this paper that other paper is in no way kind of trying to overturn the audit
during enhance and results what we’re showing is that they’re hitting a
particular type of places audit aren’t enhance and have all the right controls
in the paper it doesn’t change the results it’s just saying if we look at
who was getting hit it’s a particular type of place so the implications here
is I think you can think of the China shock in some sense as almost
short-circuiting the domestic product cycle when things should have been
moving to the lower education areas and providing jobs there for a long time
they were immediately getting hit by the China shock so the 1975 285 trade shocks
just look a little different they’re hitting a different type of place I
should be clear it doesn’t mean it was painless for those places but it means
they were better off to start with and had better capacity to innovate a way to
move to different types of industries whereas the China shock is getting
concentrated on places that are already getting hit by technology shocks that
are that are tough for them to deal with and so when you kind of combine the
technology shocks with some of the institutional shifts in mobility and
migration these trade shocks are really being entrenched some of the regional
gaps that we see across the country so what does this mean I think you can
argue it argues for a renaissance and place-based policies I think it’s
important to note and this room probably knows more than most that people have
been working on these for a long time different groups like Brookings metro or
Upjohn have been doing really interesting work in this area for a long
time so it’s not new but I do think it’s getting more attention and I think it’s
getting more attention in part because of the politics people trying to
understand kind of Trump country in some sense although I would emphasize that I
think in in all the work I’ve seen it’s it’s not just from country right the
Bronx struggles to Gary Indiana struggles to it’s not all kind of rural
places or smaller cities but there’s really a good argument for more
place-based policies in some directions to jump to the bottom first and this is
in some sense echoing some of the things Ed said there are a lot of lessons we
can get from the past of really bad policies that we shouldn’t try you can’t
just increase the supply of higher education because people move you can’t
just subsidize capital because then the benefits don’t necessarily stick in in
the places and you don’t necessarily help the people in those struggling
places and the gaming and defining areas is a huge problem and and we’ve seen
that I think in the most recent place-based policy in the tax code but
there are things you can do on the one hand you could try to help with mobility
but that’s not a sufficient answer I think to say just move because it’s not
relevant to everybody I think and some of the work in Austin plays or in
summers David Newmark has a terrific policy proposal for the Hamilton project
that I think he’ll talk about some later about ways you could subsidize labor in
high poverty places Tim Bartok has talked about this too I think the idea
of subsidizing labor in these places with incredibly low levels of labor
participation is important in some sense what you’re trying to do is figure out a
way to increase labor demand in these areas it can’t just be about the supply
of education that’s an important piece probably is fixing especially getting
people through high school and some level of training but it’s probably not
the only thing you could do better connectivity either to universities or
to the kind of higher growth regions whether it’s through infrastructure or
broadband or programs like manufacturing extension partnership there immigration
reforms that AIG has talked about of trying to get more
immigration into those types of places but I think there are actually a lot of
things we can do other than just build two billion very very cost-benefit
negative pilot projects in these areas that would actually increase demand for
labor and help these places and I think seeing the both end of convergence on
the one hand and growing divergence and labor market outcomes is in some ways a
good argument of why we should be really pushing forward to do so so just in
conclusion the the gaps across regions have become very very persistent both in
levels of income but also an unemployment rate so it’s not just
income levels that are there we’re seeing it on unemployment rates as as
Austin Glaser and summers have pointed out we’re seeing it in labor force
participation rates as well but these economic outcomes are really dividing on
educational lines far more strongly than they used to giving us I think a real
challenge as policymakers to think of what what should be done there and that
in addition to thinking about the shifting valuation of skill and
different education levels I do think thinking how trade shocks and different
types of shocks are hitting different types of areas is important for us to
think about as economists and policy makers so I think that’s where I’ll stop
them out of time so thank you again myself great I thank you it’s a it’s a pleasure
to be here this is this is just an awesome conference what a great lineup
of papers and discussing this present company excepted and is a pleasure to
discuss this work and and let me let me serve just Freeman I think the big
question this paper is asking and this the line of work that these authors have
been doing is understanding this the relationship between shocks and
persistence and why aren’t shocks dissipating across space why are they
sticking around and I think and focusing on the China shock in particular I think
the question they’re asking and a question that I’m was asking myself as I
was reflecting these comments is was there something special about the China
shock that just made it more durable more shocking or is something special
about the places that were hit by it and I think that’s the the question they’re
asking I think it’s the right question to ask and let me say I like this
question much more than the earlier question that many authors has which was
was the China shock paper wrong and you know I think that literature has played
out so I’m happy to see this new literature about you know now that you
know you know how do we interpret it and I have to say that when we started when
Gordon Hansen David Orr and I did that project we were very much informed also
by the interpret the expectation that US labor markets are very fluid and so we
were confident that we had a good identification strategy for finding the
impacts of trade on local manufacturing of course manufacturing had to decline
given the massive increase in imports of China the question was how would that
play out our expectation was that would play out pretty smoothly so we were
quite quite stunned by what we found it was not consistent our prior so we spent
a long time kind of target hardening the paper before we even released it because
we we thought it was it was so inconsistent our priors um but I think
we’ve you know I think the evidence suggests that really something important
happened and it had lasting effects I’m so let me of my remarks I’m gonna
say five things first I just want to put up the figures again about the
persistence of unemployment because they are so startling then I want to talk
about the decline of US manufacturing put that a little bit of context then I
want to show you some evidence on just how concentrated how durable these
impacts were along a variety of margins then I want to talk about the
characteristics of China shocked places and I’m gonna be remix in things that
they said that Ed said um I hopefully will display them in a way that’s a
different and informative but it’s I’m not gonna fundamentally disagree and
then I want to talk about the relationship between the China shock and
this sort of changing geography of work and occupation so as a starting point
just this figure you know alone tells a lot that the you know there was no
predictability between the unemployment rate in a state in 1976 to 1988 later
but that’s not true if you look over the subsequent three decades so 86 to 96 96
2006 2006 to 2016 and over so we have persistence Bigley as someone say and
and so that itself is a surprise that alone is not something that you know we
grew up believing in it’s not that the old facts were incorrect they just
aren’t the same as the current facts so let’s talk now about the decline of US
manufacturing so it’s a truism that the share of US employment manufacturing has
been in long decline so here I’m actually dividing as was suggested by
overall employment not just not an agricultural employment and that affects
these trends so if you include agriculture US manufacturing plant was
never above 27 percent of all employment and and then it’s in a long decline but
you can see that it’s a pretty smooth decline and you kind of have to squint
to see the China shock there you can sort of see if you look around 2000 it
seems like the rate of decline kind of accelerates but it’s not startling
now let me just do this take away the denominator just give you the count of
jobs so this is the number of people working in US manufacturing uh and you
can see that you know it’s not at a high in the end of Second World War in fact
it reaches its peak in 1979 then it enters a long slow decline so
manufacturing loses two million jobs over the next 20 years and then between
1999 and 2007 it falls by 20 percent so it falls by almost four million jobs and
then between 2007 2010 I falls by another couple million jobs so in net
there’s a one-third decline in the number of people working in US
manufacturing in just at eleven year period now the first seven years of that
or eight years of that is is the you know is the period of extremely rapid
rise of Chinese imports following China’s accession to WTO the the next
few years of course is the recession that’s not the the China shock but I do
think it this figure is helpful because when people say well manufacturers been
in decline for a long time it’s it’s technology its technologies technology
you can say remind me of the great innovation in 1999 that only made twenty
percent of manufacturing jobs there is not one right that was clearly a trade
induced phenomenon eventually those jobs would have been automated or shipped
offshore but it just wouldn’t have happened so fast so as Jason said this
really accelerated a process um so just to remind you of how he pronounced those
impacts were here I’m just to put some figures from the work that David and
Gord and I have done in various papers so the first thing to remember of course
is just how geographically concentrated the shocks were and of course
Manufacturing’s are always geographically concentrated it’s it’s
not evenly spread across space and it’s more than just manufacturing concentrate
but the set of things done in places is highly concentrated so southern
manufacturing does not look like northern manufacturing and the areas
that were most impacted by China trade furniture and fixtures games toys and
children’s vehicles sporting athletic goods electronic components plastics
motor vehicle parts an assembly electronic computer assembly so these
were labor intensive production in which China gained very rapid comparative
advantage and so this was not the Rust Belt manufacturing in fact you could say
this was below the belt this was all southern manufacturing and it was much
was lower paid it was more labor intensive it was less tech capital
intensive it was less less skill intensive um this just shows you some
ways of measuring the consequences for various outcomes some within the labor
market some not so this shows you the impact of a kind of a one standard
deviation trade shock or actually the the interquartile range the PG 50 P 75
minutes B 25 on annual earnings of men and women so what you can see very
clearly is the effect was larger on men’s earnings particularly below the
median and so that’s important there was a lot of low education relatively high
wage manufacturing work and so when that decline it was particularly felt by men
towards the lower quartiles of the earnings distribution although
relatively high earnings for their levels of education um you can see this
more comprehensively this just shows you impact on male and female earnings by
percentile you can see looking at the blue relative red line on the on the
left the effect is larger for men across the board and this does have the effect
actually of compressing the male female female or annual earnings gap so it
causes convergence by lowering men’s relative to women’s wages when men’s
wages are higher at every percentile just about and that’s important because
in thinking about some of the kind of non labor market consequences that
played out some of them have to do we think with gender disparities and
earnings so in particular is marital status there
was a sharp decline in the fraction of young adults 18 to 39 who were currently
married not primarily stemming from divorce but from an increase in non
entry into marriage and similarly a decline in cohabitation either living
with spouses or living with partners and and let me say this is actually it’s
extremely Bukharian and in fact if you our paper called um when work disappears
in the a our insights looks at shocks by gender to earnings and what you find is
declines in male earnings holding women’s earnings constant reduce
marriage declines in women’s earnings holding man’s earnings constant
increased marriage and similarly for fertility so if you know the sort of
notion of specialization being a kind of a spur to marriages is really affirmed
and it’s the relatives that matter not just the levels um so I this you could
argue and I think something you see very clearly when you look in the data is
that areas southern intense manufacturing tensive areas were places
where men had where men had low relatively low educated men had
relatively high earnings levels and were relatively likely to be married on
manufacturing was the kind of economic foundation of a particular set of social
organizational you know you know expectations and the elimination of that
foundation had very large social effects that’s not to say that those were the
right foreign social organization but that they were it was kind of
foundational to how work and family life was organized um you see this in
children’s outcomes so fertility Falls but the fraction of children born out of
wedlock Rises and the fraction of children
living in poverty increases and the fraction of children living with married
or even two-parent households so the effects are felt you know far
beyond earnings and even some evidence of a rise in premature mortality
especially especially associated with declines in male earnings and a
substantial chunk of that although not all of it comes from drug and alcohol
poisoning much more common form of death among men than women surprisingly we do
find effect on suicides among young women these are all adults age 20 20 to
39 we’re looking at excess mortality over ten-year humilated period per
hundred thousand dollars these are not enormous numbers of course thankfully
but you should think of them as kind of the extreme margin of a broader font
phenomenon of despair that that was taught highlighting in in his comments
so you know thankfully only a very small percentage of people who abused drugs
and alcohol you know suffer fatal consequences but that you should think
of that is indicative of a lot more you know abuse going on okay so that sort of
I hope that’s you know the context of just how consequential this was so now
let me talk about China shock places so here’s how I’m gonna do that so what
I’ve done here is I’ve lined up commuting zones by their percentile in
the 1990 to 2007 China shock distribution these are China shock as we
measure Navy age to that 2013 this is the growth of the exhaustion component
of growth of imports per thousand workers and so I the percentile thing
just linearize this a bit so that the upper quantiles that’s something like
you know six or seven thousand dollars per working age adult at the lower
quintile is as close to a couple hundred dollars and this is the competitive the
the effect of these are imports predicted due to Rhinos rising impaired
advantage projected into your local labor market as a function of what
industries you have so it’s not where you it’s not how many Walmart’s you have
in what you’d be buying it’s what you work would have what we’re producing
that competes with those imports so that’s the tool I’m going to use and
that’ll just display a bunch of characteristics of places
according to their China shock percenter okay so um so this shows you the
fraction of employment is employment including over agriculture as well as
non agriculture employment in manufacturing as a function of China
shock exposure looking over 1950 to 2015 so first thing to note so the China so
the places that are going to be most exposed to on the right-hand side least
exposed on the left-hand side and what you can see is in 1950 the places that
would experience the largest China shock we’re not nearly as China as
manufacturing intensive as the places to the left of them and that’s because the
place to left were actually those would have been the Upper Midwest right those
would have been the not yet rusting belt and and then if you look in 1970 you can
see that relationships deepens what’s happening
well many you know you could argue that the US South was you know outsourcing to
China before China was available right US manufacturing taking advantage of low
wages taking advantage of limited Union threat or no union threat was moving
manufacturing south and you can see that in 1970 you can see that in nineteen
eighty and if you look between 1980 and 1990 you see this line start to get
shallower meaning there’s some decline and now I’m going to add 2000 2007 2015
and this you know rapid rotation rightward reflects the decline of
manufacturing employment in places that were exposed to the shock right so
there’s an arc here of manufacturing moving into those places between 1950 19
and 1980 and then it kind of stabilizing 1890 ok and then and then rapidly
decline this just shows you the changes in that so this is the same x-axis but
these are the changes in manufacturing intensity and you can see between
nineteen fifty-seven nineteen seventy eighty that places that are going to be
China shock to have rising manufacturing employment and then between 80 and 90
they already starts to decline and then very rapid declines thereafter so this
is just one way of seeing how dramatic this
um the other thing to know is that these places that we’re going to be exposed to
the China talk and consistent with their work
we’re less educated right so this is the non the share of adults with high school
or lower education and these places were less educated in 1950 in 1970 in 2017
and 2015 and so that’s important to know so manufacturing is moving to less
educated places and I but interestingly and it’s a really telling figure if you
look at wage growth across decades so 50 to 70 everyone’s wages grow no surprise
there seventy to eighty pretty stable in eighty to ninety these are low education
areas that have relatively rapid strong wage growth and that’s not what you
would expect right given the rising return to education this period you
would have thought what we should have predicted them to have relatively slow
wage growth the reason they don’t is because manufacturing is moving into
those locations and providing relatively high wage jobs for relatively less
educated workers and then of course in 1990 2000 that robust wage growth goes
away and then after that it’s fallen okay
now finally the st. you see a similar pattern with employment employment
population rises in these areas gonna be China shocked in the 50 through 80 90
period and then Falls really rapidly thereafter so it’s more to understand
that there’s a you know you should think of this as kind of not a things were
stable and then there was a shock but there was this this robust growth of
manufacturing movement into South and then it was rapidly eroded so this was a
kind of a multi period story okay so now I just want to relate this more
broadly to the change in geography of work so many of you seen variants of
this picture showing you the kind of polarization of employment across
occupations the left-hand side being services you know food service cleaning
security home health aides transportation repair the middle being
production office clerical administrative support and sales and the
upper part being professional technical and man
important to understand that this polarization really reflects two
different things for people with high school and lower education reflects the
movement out of middle skilled production clerical jobs and into
services for people with higher levels of education it also reflects the
reduction of the middle but a movement primarily rightward upward and
especially among women into professional technical and managerial work so really
this polarization story is about a hollowing out of opportunities for non
college workers this phenomenon has not also not played out evenly across space
so this these figures are now arrayed by population den T’s density of commuting
zones and you can see that denser places have become much much more educated over
the last six decades and as I emphasize the places that were most exposed to
trade were less educated and as this has happen as those high-density places have
become much more educated the set of jobs done by the less educated has
changed very rapidly so you can see if you look in 1970 that there that non
college workers were over-represented in middle skilled work in the 1970s in
high-density places that’s what that upward slope looks like means if I add
the 1980s you see a little bit of rotation add the 90s you see more add
the 2000s a little bit more and then by 2015 it’s gone so what’s happened here
is the non college workers used to do relatively prevalent middle skill work
in urban areas and that has utterly completely decline and it’s been offset
by movement into services nothing similar has happened for
college-educated workers they basically do the same stuff across places their
tools have changed their titles have changed but they’re actually doing very
similar work I can see we have that that’s changed a lot um what does that
erosion of middle skill work where does it come from well it really comes from
two different things one is the decline of production work and that’s the blue
series and that starts early that happens between 50 and 70 and then the
other is the erosion of clerical work for non college workers which happens
much more after the computer revolution between 80 and 2015 and so there’s
actually a simultaneous decline of two categories of non college work that were
quite important and differently so by gender okay so now I want to try to tie
these together in my remaining 343 seconds so this shows you population
density versus the China shock really these panels are all the same just about
the main thing to notice is the China shock does not occur in the most dense
places it occurs in the middle of the of the density distribution the rise in
manufacturing employment so this shows you this this history in the 1950 it was
you know much more urban in that in 1950 it slowly moves out of urban areas and
into mid density places and then it just collapses from there and so that was the
movement out of the Rust Belt into the south and then the subsequent
elimination or decline um finally the this just points out that the and
reminds you that where this occurred were not the most educated places so the
intersection of these two phenomenon is the movement of high-skilled work into
urban areas the decline of middle skill work in dense areas and then the China
shock kind of piling on to that because you already had to climb manufacturing
you already had decline in clerical work and then that went away very quickly
okay so let me conclude so the puzzle as with high regional convergence has
slowed or halted unemployment rates have become persistent
on the China shot cutting these durable impacts the question is and we know what
those were and they weren’t just earnings and so they the led to all
kinds of social enemy that is worth talking about um the question is was
something special about the China shock or is it something special about those
chalk places well I would say it’s kind of both but it’s important recognize
that the places that we’re going to experience the China shock in the 90s
forward had a positive a positive pre shine of shock in the earlier decades as
manufacturing moved in there and that was creating what was surely if an
ephemeral state because there’s no way that that manufacturing would have
stayed there four or five more decades had it not been for China’s rise it’s
accelerating the process that would eventually have occurred but of course
the rate of change matters so finally you know this is the question Edie asked
at the end of his discussion where is the land of opportunity for non-college
adults you know one reason this shock may have been so enduring is there was
no other safe harbor to move to it’s not clear where you know every says well go
pack up and move to land of opportunity well where is that place if you you know
have lessen college education and you’ve been doing industrial work for most of
your career and so I think you know in thinking about this nexus of policy
problems I think it’s important to understand one it’s you know there’s
it’s not clear where there are better opportunities but two there are a number
of forces that are going to change that story most important being demographics
and there we are entering a period of labor scarcity and scarcity of young
workers and extra scarcity of young non college workers who are going to be
required for services for maintenance for cleaning for elder care a lot of the
problem that we face is not that those jobs don’t exist but they don’t pay it
especially well they don’t offer good careers and men don’t seem to be
especially interested in doing them it’s a cultural problem um so there’s reason
for optimism labor scarcity or tight labor markets you know one of the best
policy solutions that we have all that we don’t really know how
it but the the challenges is is highly apparent thanks very much so like last time let’s go ahead and
take a few questions and then turn it back to the presenters and the
discussants so button raise your hand if you want to ask a question or make a
comment so one of the one of those sort of many reflections that comes out of
the China shock literature is that as a profession because the legacy of Adam
Smith we were all conditioned to be pro free-trade always and that’s usually the
right answer but it is also true that in 2000 2001 most people underestimated how
disrupting this event would be in it by a sizable amount and that doesn’t
necessarily mean that we shouldn’t have you know supported to China’s entry into
the WTO but we certainly should have thought about what it would do that was
bad going forward and we should have also sort of thought about which areas
are vulnerable and which areas can can adapt to the future and that’s I mean
both of your presentations were quite compelling on this is there a future
trade shock out there that we should be worried about in that’s similar how
should we pout or are we done with this have we have we already we’ve gone
through through everything or am should I be worrying about something else
that’s gonna decimate some region of the US in the future and if there is what do
you think we should be doing to actually prepare for it in a sensible way anybody
else want to get a question in there since I do wait for it I’m going to
indulge myself in there’s two questions so the first one is why should we
consider sort of the stalling of regional convergence to be a puzzle
right so you have a model you have stories about why under certain
conditions you might have regional convergence but what are we converging
toward why do you think that that should go on forever that it should become a
uniform thing I think that we do need some underlying
models that tell us what we should be converging toward and again in sort of
you know having rubbed off from Ed Rosen robic you know unemployment rates could
differ you know it doesn’t always have to be blonde short cats they could just
be a compensating differential and so I think that you know it’s wonderful that
we’ve been in the data it’s wonderful that we’re going to be thinking about
policy somewhere in there it would be nice to have a little more theory
telling us about how are these things supposed to be linked that was comment
one the second is I’m going to really be self-indulgent because I mean I talk a
little bit about comparison so the shocks that we’re talking about you know
routinization and the China shock are not local shocks to the United States
their global shocks hitting lots and lots of places so it’s one dimension
some work that I’m doing with Eric mangas into mush Mikulski HEC Paris and
because we’re looking at related questions in the context of France which
is you know looking at some of the the same chunks and so I mean it’s skip
theory and it’s gonna report a few facts that emerge from ours one is in terms
you know we talk about jobs job exposure but that’s different in your routine
ization and offshoring exposure is different than the total middle skilled
job loss and so it’s fact one that emerges from ours is it’s actually in
terms of loss of middle paying jobs it’s big cities that in percentage terms have
the big losses of jobs Paris about double what it would be in a city of
fifty to a hundred thousand to is that there’s a bias in – and maybe also why
we’re not as concerned about the the large cities which are generally going
to be similar you know the same as a dense cities is
that in in France when you lose lots of jobs in big cities it’s going to
be a um polarization everywhere but it’s going to be too high scale jobs gained
for everyone low skilled jaw or locating job gained and in the smallest city so
it’s going to be exactly the reverse to low skill for one high skull and let me
just find this oh and then when we and I think this is very sympathetic it’s sort
of taking into a common kind we’re looking at the 1994 to 2015 the story
about the time sequencing I say you know New York lost its jobs early and then
you know it’s a industrial jobs early and then some other jobs later it’s also
true in the cross-section that the jobs that you’re losing among the middle
paying jobs in the large cities is different than in the small cities that
in small cities you’re still losing those industrial jobs high in
low-skilled industrial jobs in the large city what you’re largely losing are sort
of the mid-level professional jobs and to certain extent office workers and so
on like that so alright let’s let you guys respond to those I guess I’ll start
with with Ed’s point that you know we underestimated the disruption I think
that’s right I think in terms of you know what do we do about that I think
honestly I think a big part is that we need to recognize where the shocks will
hit that the shocks are not you know the shocks we should know always weren’t
evenly distributed and understanding though the types of places getting hit
really matters because if the shocks are hitting higher education places we can
count on just the economy to deal with it in a different way and if they’re
perfectly distributed across the country right then we just say it’s the feds
problem to deal with it in all seriousness uh we just use macro policy
we lift the employment levels back and everybody’s happy but when they’re
targeted it’s different but in particular when they’re targeted at
lower education places I think the argument is there they’re even more
different and we need to to deal with that as to your question of what’s the
next shock I’m actually you know sympathetic to this view that there will
be a not never be another China right there’s never going to be another place
as big and as out of the global economy as China was when
it kind of entered the global economy not just because of WTO but just their
own choice to enter the global economy and so we’re never going to see
something quite like that but I think in general at least to me what it says is
that we should every time we’re thinking of a policy where we know they’re going
to be winners and losers that we have to think extra hard when we know the losers
are going to be people in places that are not going to recover as smoothly as
places that would recover more smoothly so that to me is is the lesson at least
I I take out of it and then I’ll just quickly before turning it over as to the
other question of you know what are we converging to and should we be shocked
by something I guess to me the argument is the level gaps are still massive and
when the level gap says we said as everyone is saying have always been
massive but when the level gaps are massive and you were converging and
mobility was higher I think there was a greater argument that policy makers
could say this is sorting itself out and we should focus on efficiency and
aggregate outcomes and not on places and I think once you had very large regional
gaps and no more convergence and less mobility it does call into question more
are there place-based policies that make sense and I think that that’s the
argument so I’m a little bit less sanguine than
Jay about whether there will be another shock or not in terms of what industries
would be most vulnerable I think we can just trace industries that have moved
from the most innovative areas to lower wage areas right now I think that might
be some mid and high skill services so if you read Richard Baldwin’s book globe
Audax you know he has a little bit of that gloom and doom for services even
perhaps college professors in the future I guess we’ll have to see um in terms of
optimality in convergence I really really like this recent paper by Pablo
foggle BOM and Cecile gout there because it actually brings some optimal tax
theory to some of the the new economic geography and it really thinks carefully
about spill overs between high skill and low skill workers including on the
amenities side and it thinks about congestion effects really really
carefully and what they find is that there is some optimal argument to be
made for subsidizing high-skilled workers in mid and smaller cities
midsize and smaller sized cities great ok I don’t want to take too much time
but on the you know should be anticipated it’s it’s interesting
actually I’ve gotten a lot of personal feedback on from the China shock paper
about who is who is to blame so so Gene Sperling assures me that it was
anticipated and that there were provisions put in place for safety
valves in the sort of trade visions that the Bush administration didn’t want to
pull the trigger on so he put you know in case of in case of China shock you
know pull this lever and then the Bush administration said oh we don’t believe
in that I’m so he assured me it was anticipate your policy in place Larry
Summers also assured me it was the anticipated I’m taking manage of him not
being here he’s not here right and that that was the price we just had to pay
because China would reform and open that went on the another thing that we
thought we knew was that the Trade Adjustment Assistance Program didn’t
work but in fact evidence from Ben Hyman suggests that it actually worked pretty
well although was way too small and hard to access so um I I think that we there
were a lot setting Gina and Larry Assad I don’t think most people anticipated
this and I don’t think we also understood that we actually had
reasonably good policy tools oh they were not set up to scale in terms of
looking forward I think there are a number of lessons first I I don’t think
there’s going to be another China shock but I think there is going to be very
disruptive impacts of automation and artificial intelligence fortunately
they’re gonna occur at the job level not the industry level and I think that’s in
some ways better because when you know when all the manufacturers in a place go
out of business that just sort of takes down lots of stuff when a subset of jobs
are impacted that has less of this sort of going out of business character and
some firms will obviously benefit but I think we should be very we should
anticipate that we’re gonna face a lot of challenges that the potential for
technological disruption is greater than has ever been the uncertainty about how
fast things can move is vast and um we the lesson is that policies can really
help and that we have a shared interest in making sure that we don’t have
sustained rising unemployment and concentrated despair so I think we
should be working hard on that I’m on on Don’s point I I guess I’m it I made
misunderstood the end of regional convergence but I would thought it’s
evidence suggested these places it’s not an amenity value of you know they have
lower wages but it’s such a shitty place it seems like they have lower wages and
everyone’s miserable so I don’t see that as as convergence in the good sense but
on the sort of the geography of this I I think your work here is really
informative and I my work is you know a so far so concentrate on the US and B as
you said it doesn’t have as much theory as some would desire and I think there’s
I think there’s a complex story to tell hopefully some of the initial facts
guide angry alright a couple questions over
here far left in front hi Amy Schwartz Syracuse University um
so I just wanted to to come back to the this gender issue which has been sort of
you know troubling me you know if there are jobs and men don’t want to take them
because they’re disutility or something I feel a little bit differently about
that as a matter of public policy than I do or public concern than I do about
there are no jobs if it’s disutility it’s different than if it’s
discrimination and we have policy solutions to deal with all of these
things potentially if it’s so I think that it’s you know this is worth
spending a little bit more time I’m kind of digging into and I think that you
know what that means is that if what we’re really gonna decide is this is
about men then asking where’s the land of opportunity for non college adults
it’s not really the question its where’s the land of opportunity for non college
men I mean I think right but if that’s the if that’s the problem and we’re
really not worried about the women in these places then then we should just
ask that I thought that used to be Alaska you know you know so but you know
I think that that if the if this is really a gendered issue then we should
figure out what that means and and ask you know is this something it’s gonna
pass in partners as this generation of prime aged men moves out of the prime
age and are the younger men different because they I don’t know have a
different utility the gentleman right next to you hi Larry Katz want one comment a couple
questions so there was a lot of pointing out of the Blanchard Cats
motivating on graph seventy five to eighty five but in actuality that was
you know we knew that was a bit of a blip you know if you’ve done this sixty
to seventy you wouldn’t have found zero correlation and you know if you actually
look at the rest of the paper you know was about the dynamics of adjustment and
you know as Davis said you know each each region has a different equilibrium
unemployment rate and a compensating differential and we were concerned about
how quickly you converge back to that and that you know we were very worried
particularly with thinking about European monetary union and other things
about places that didn’t have a lot of regional mobility shocks to particular
areas lead to very persistent deviations from their long-run equilibrium and that
was a big part and I think in the European context that certainly played
out especially when you don’t have fiscal transfers across regions and in
the US you know things have gotten worse I mean through much you know through the
first eighty years of the 20th century regional mobility seemed to be an
important part of adjustment to shocks and its really slowed down and so you
know we’re gonna hear talks about that later this afternoon so a couple
questions or puzzles about this as David Otter pointed out in the China shock
areas these are not areas that forever had been like high wage manufacturing
stuff where everyone and their grandparents had worked in there so one
of the puzzles you know some of it was how rapid the shock was but these are
areas where lots of people have just moved there twenty years ago to find the
good manufacturing job is why when the negative shock you know and your
findings and others was there so little population movement out what’s happening
to the sort of next generation is this because of the education of the areas
and so I think that’s you know is this about you know the types of jobs that
were available and more positively you know shocked areas you know just not
being very attractive to less educated workers in general not just men because
the women weren’t moving out you know very much in response either
and I think this raises the other when we look at these middle sized cities and
areas there seemed to be a big growth of jobs you know which I would say are not
not clerical is sort of not the right term yeah it’s really you know customer
service social doesn’t have to be done in person it can be done on the phone it
can be done in a chat box but it’s a set of skills from middle education and
interactive skills that don’t seem to be done very well overseas that seem to be
you know attracting more women in these areas and not men but it’s you know
there really seems to be I think this dimension that David Deming and others
have worked on on the rise of sort of interactive social skill jobs not both
in-person services you know health care but a lot of it being distance but not
being the sort of jobs that manufacturing workers or people
servicing the manufacturing sector we’re well trained for it I think that you
know when we look at the rise of employment in middle density cities this
seems to be a key part all right why don’t you guys take those two before we
take more questions sure just because I know there were a bunch of hands I’ll
try to be fast on the gender I I agree with you I mean I think especially if
the issue is people just won’t take jobs that’s a different type of policy
problem but I but I also think just even if you are looking overall at
unemployment rates and labor force participation rates and employment
population rates in these places that are kind of landing in a bad equilibrium
they’re really bad numbers for them in aggregate not just focused on men or
focused on women separately and I think I would agree with that there are big
externalities there in terms of what it means to the tax base what it means to
the ability to provide public goods and things like that that makes this this a
problem I do think that part of the solution is
remember if Betsy Stephenson is here yet but I’ve heard her talked about it a lot
of how do you deej ender some of these jobs I mean if you’ve ever been in a
hospital with 75 year old patients after surgery my dad is a large guy he’s 6-8
like the small nurses couldn’t lift him right and they would need to go find the
6th floor guy who was a nurse and you need more of them and things like that
in larious I tried to mention I completely agree I I love that paper for
all for lots of reasons not just the one picture there’s lots of great stuff in
there and and I agree with you about that I I think the one the only thing I
would flag we were saying these places getting hit were the places that
everyone had just moved to the other thing that I didn’t that’s not in this
paper but is in the other paper with with Katie is it’s also it’s not just
the low education places that got her hit harder conditional on the size of
the shock it’s also the places where the jobs were already leaving
so it’s the places that in some sense shouldn’t have been producing the China
shock stuff anymore anyway they just hadn’t gotten all the way out of it that
was brutal because then they lost the jobs and there was no way anything was
coming back and they had already been on a downward trend so I it’s not just the
places that where people are moving in and I so I think there’s there’s but I
agree with you the question I think comes back that David’s question where
were they supposed to go if it’s a mobility question what was the next
other than North Dakota for fracking like where were they going next and that
didn’t show up until the very end of that cycle so on the gender issue so
this is a little bit outside my particular specialization but I’ve read
articles recently about shutdowns in coal country and the males who are
losing their jobs their wives are going out and becoming nurses so it’s not the
male’s retraining and becoming the nurses it’s the wives becoming the
nurses but then when I look at the younger generation so I have nephews who
are training to become nurses so I wonder if some of these things may kind
of process out over time and I kind of hope they do because like my husband was
a wonderful stay-at-home father he got so much flack from that
even in California he cut a lot of flack for that and from all different kinds of
people even people who would bring stuff to the door during the day you know
assumed he must have lost his job somewhere so hopefully we’ll sort of get
over that over time very quickly so on Amy’s argument I agree but I think that
we have to recognize that if men are miserable they’re gonna make women
miserable too and we don’t have I mean look if you look at that in terms of
marriage and fertility and you know household poverty and so on so you know
even if you like you know guys suck it up you know take the job or don’t
complain it still has huge it has costs on the other gender and we don’t have
that many situations where men and women move to different places a large scale
um so I do think you know there’s a reason to want to deal with that and I
agree that you know they’re these the health service jobs there’s tremendous
opportunity and if you do go into you know if you go into a assisted care
facility you will find men in health service job so they’re mostly
international men many men from the Caribbean hopefully the norms will
change I mean flight attendants used to be Alwyn right that changed and and now
I imagine the social barrier to entering that for men has team is much lower
hopefully that will help them with health service work on the I think the
last question that Larry raised is a really important one which is you know
if many of these service jobs can be done at a distance and they don’t need
to be done the expensive cities who will have the skills to do them and who you
know how much interpersonal skill how much education what are the attributes
that can make someone successful at that and I think you know are kind of very
casual observation is right now a lot of those are articulate women but it you
know again we should if that’s a growth area and it very well could be we got to
figure out what are the skill sets needed to create you know allow people
take advantage of that opportunity so I Matthew can Johns Hopkins so it’s
coming back to Ed’s question about future shocks and I know we’re gonna
talk about migration cross this afternoon if we had different housing
markets a few if more of us were renters if we had a less durable capital stock
coming back to Ed and Joe’s paper from 2005 how much of a role would that play
when the same demand shock hits that the place would clear out and and people
with who are renters holding a diversified portfolio and not taking the
same income effect from the lost equity as the shock it’s alright let’s squeeze
in tomorrow David knew mark had a question yeah dan sickle from Wolsey College so I
found I found really provocative the comments about what might the next shock
be so here’s a maybe somewhat unusual possible thought about the next shock
and one that I certainly hope isn’t the case but perhaps is suppose we’re
entering a twenty year period of D globalization and we’re entering a
period of less trade and more trade barriers any thoughts on what some of
this literature and results in this literature might say about how that
world might play out and again it’s a world I hope we’re not getting into but
perhaps unless egg blazer becomes president in 2024 perhaps is the world
that we might be getting into Dick’s Malan CMIT first an observation and then
and then a puzzle the observation is what it means to be a high school
graduate differs substantially across the country as we know so my guess is
that the differences in effective education between what show up in the
data is low education areas and high education areas is wider than high
school graduation rates would suggest education isn’t the answer to everything
but as the result of that education high education areas respond better to shocks
education is the answer to some things the the question which is troubled
federal education policymakers for a long time is what can you do about
persistent differences in the quality of education given the strength of forces
for local control given differences in local tax bases and given the fact that
prior reforms have failed so I raise that as a policy puzzle because if you
have persistent low education areas you will have persistent based on this
analysis problems in those areas all right I know there’s other folks who
want to ask questions but in the interests of keeping the trains running
on time I’m gonna bring it back to the table up here and head to lunch in a few
minutes okay so very quickly then on the why not move and what different housing
stock lead to it I think part of the answer is honestly not everybody wants
to and and local networks of community and family and things like that I think
are one of the things but I’ll mostly defer to the whole time the whole
session on migration for people to talk about it but I do have a comment on the
D globalization and I think one of the things I think you really can’t help but
learn from the China shock literature is just how disruptive trade shocks are
right but I think what we’re seeing in the last three years is just how
disruptive trade shocks are in either direction right if you built your entire
business around a certain supply chain and that gets broken your factory shots
also not just you know can shut for either type of trade shock the
difference is a trade shock of opening brings productivity enhancements overall
to the country you actually do get gains the problem is we don’t redistribute
them very well when you have a negative trade shot going in the other direction
you just get losses that we also don’t distribute very well but they’re so
they’re gonna be concentrated again when there are no winners to compensate the
losers so I think it’s actually a really troubling thing that can show up there
and then lastly on the the education thing I agree with you about what it
means to be a high school graduate graduate differs what shocks me is just
how important just getting to like the numbers we’re using are less than high
school as the key variable there and that has converged around the country
over the last 50 years to some extent the places that used to be 50% no high
school are now 20% no high school but that’s still a huge gap and I think what
we see is it really matters being a place with a low high school graduation
rate and so I think any of the policies whether it’s distributing title one
money differently or anything that you can do to try to help these places
continue to close that gap is really important
on D globalization I think outside with J one more thing I think is relevant is
what this does to military stability and whether it would lead to a more violent
world I think there’s some evidence to suggest that is the case especially if
it’s a tri-polar world that we end up in just two quick comments on D
globalization I think if if we get a lot of manufacturing back in the United
States you know in the very short run maybe that will you know create some
temporary jobs but it’s gonna be incredibly capital intensive but it’s
not gonna be it’s never going to be mass employment the way it was on on
education I mean there’s actually you know hardening news and education the
educational attainment rates of recent us cohorts are rising really rapidly the
fraction of 25 to 34 year-olds with a who have complete high school rose from
like ninety to ninety four percent between just over the last decade so um
in the same age bracket and the number who have completed a four-year college
degree has also written by risen by several percentage points although that
may partly be the diluting of the quality of the degree but in high school
probably not so there is you know education is always the long-run best
response for some there’s a lot of attention to that so
but clearly education is not going to be the short-run solution to a lot of the
destruction we’re talking about here wonderful thanks everyone thanks to the
presenters the discussants and lunches out that way you

Leave a Reply

Your email address will not be published. Required fields are marked *