Economics of Labor Markets, Public Workshop on Competition in Labor Markets 2 of 7


DR. MARINESCU: All right.Good
morning, everyone. I’m very happy to be here
today to talk to you about some of the key advances
in the economics of labor markets in respects in
which this is related to antitrust enforcement. So I’m really happy
that the prior remarks introduced the
topic already. So indeed recently we’ve
seen a growth of labor antitrust enforcement
and, in a way, you could say that this is an
infinite growth because we’re starting
from almost zero. So while labor antitrust
enforcement exists, it is, as has been said
before, muted compared to product market
antitrust enforcement. And this creates a
litigation gap that we discussed in my paper
with Eric Posner. And so, you know, looking
at this, you might think that this implies that
obviously labor markets are competitive. Obviously there is no
issue with competition in the labor market. So today, I want to
convince you that, in fact, labor markets are
not perfectly competitive and it is worthwhile to
look at the ways in which anticompetitive practices
affect workers in the labor market. So why should we
care about this? Why are we
interested in this? For many reasons, but one
recent reason is that there has been slow wage
growth since the last recession and
even before that. And this has ignited
government interest in addressing impediments to
wage growth, which include a lack of competition in
the labor market, among many other factors. So just to illustrate
the growing interest in antitrust from the DOJ
and the FTC, as has been already mentioned, in
2010, the DOJ sued Silicon Valley companies for
no-poach agreements. Also to me, very
interestingly, in 2016, they issued an antitrust
guidance for human resource professionals
detailing the kinds of behaviors by human
resource professionals that are acceptable or
unacceptable from an antitrust perspective. So I think that’s very
important to put it out there for everyone to see
because it’s not always intuitive what is and
isn’t acceptable in this area. More recently, in 2018,
FTC’s chairman said in a congressional hearing,
quote, “We’ve told the staff that they’re
supposed to look at potential effects on the
labor market with every merger that they review.”
And so, on the issue of mergers and their impact
on the labor market, with respect to potential
anticompetitive effects, I invite you to have a look
at my paper with Herbert Hovenkamp, which has
recently been published. All right. So this is — just to set
the stage, now let me introduce to you a key
economic concept that is going to help you think
about competition in the labor market, namely the
labor supply elasticity. So what is the labor
supply elasticity and why does it matter to
understand competition in the labor market? So think of it this way. If you think about this
elasticity — another synonym for it might be
sensitivity — one way you can think about it is to
think about a worker who’s currently in a job and ask
what would it take for this worker to quit. More specifically, how
much of a wage decrease would the worker endure
before they would quit their current job? So why is this
relevant here? Well, if you think about
it, if the elasticity is low, meaning that the
worker isn’t very sensitive to such a wage
decrease, that creates an opportunity for employers
to pay workers less because the workers will
not quit or not easily quit for a given
wage decrease. So if workers — in this
perspective, if workers have low sensitivity
to wages, AKA low elasticity, that means
that employers are able to pay them less and less
below their actual contribution to
production and still have workers not quit or few
workers are quitting. Oops, I think I just
lost my slides here. All right. I think I remember what I
want to say next, which is so that’s one way of
looking at the problem. And I want to tell you
about another way of looking at the problem. By the way, this first
way, looking at the quit elasticity, there’s a
whole literature in labor economics addressing that. Another way of looking at
it, so I told you a story about workers
leaving their jobs. Now, let’s think about
workers wanting to go to another job, so more like
the hiring side of this. So there we can think
about the application elasticity. So when workers are
looking for a job, how sensitive are
they to wages? And in particular,
much can a firm who’s trying to recruit underpay
workers before nobody will come or very few
workers will come? So that’s the side
of the elasticity of application. So when we recruit
workers, how much do workers care about wages? How sensitive are
they to wages? So the basic conclusion
from this concept is that the elasticity of labor
supply is a good measure for competition and that
lower elasticity means lower competition. From these two examples,
you can see that if workers have low
elasticity, that means that I can pay them less
and not have them quit in drones. I can pay them less and
still get workers apply to my vacancy. So that’s the
key idea here. So the lower the
elasticity, the lower the state of competition
in the labor market. Conversely,
think about this. Think about a perfectly
competitive labor market. That means that nobody
can short-change me as a worker because there’s
somebody else next door who will immediately take
me up at the competitive wage level. So in that sense, the
elasticity is infinite in an extreme case. Any small change in wages
will make me quit and go to the next best option
next door, and there’s many, many of those by
assumption of a very competitive labor market. So let’s think about the
determinants of this elasticity. What makes the elasticity
lower, which implies lower competition
in the labor market? So there are many factors
that you can think of, and I invite you to think of
examples from your own practice. But I want to highlight
here two particular factors. One is job
differentiation. So jobs are different
from each other and one very important dimension
for workers, as my work has shown, is distance
from their home. So workers tend to prefer
jobs that are close to home. And that limits the
labor supply elasticity because, you know, there
might be a relevant job that is quite far. So if the job that’s
closer to me underpays me, I might still take it
because the faraway job isn’t really equivalent,
from my perspective, because it’s too far
from where I live. So that’s one aspect,
differentiation. Of course, the jobs could
differ in many other dimensions. And the second aspect that
I want to talk to you about, and is highly
relevant to enforcement, is job availability. And in particular, I want
to make an important distinction between
jobs that are nominally available, that I’m going
to be able to capture through looking at the
number of vacancies in a certain labor market. So these jobs you might
think are available. But I want to turn your
attention to the fact that even if those jobs are
nominally there, there are vacancies open, that
doesn’t fully represent the state of competition
in that labor market for a number of reasons. First of all, if you have
a no-poaching agreement, well, there might
be a vacancy. But workers who are
coming from competitors are going to be
essentially not able to take up that vacancy
because the two firms agree not to poach each
other’s employees. So the opportunity for a
given worker to move to a competitor is
going to be lower. So even though vacancy
is there, in fact the chances that the
worker will move to the job of the competitors
is lower because of the no-poach agreement. The no-poach agreement
is one thing. More severely curtailing
is the non-compete because the non-compete applies
to all competitors, not just a competitor that
we agreed with, but any competitor that satisfied
the condition in the non-compete. So now I have all these
jobs around me that potentially I
could apply to. But in fact, my non-
compete might say that within two years, I’m not
allowed to work for any competitors within a five-
mile, 10-mile radius. Even if there’s jobs
there, in reality, those jobs are not for
me because of the non-compete. So again, this reduces
job availability, compared to what you might think
in the presence of non-competes. And the third factor I
want to talk to you about is labor market
concentration. So that too decreases
effective job availability. So, for example, in a
given labor market, you might have, who knows, a
hundred Starbucks jobs. It looks like there’s
a lot of jobs. But if they are all
controlled by the same entity, the degree of
competition between those two jobs is very
different because, as I go to one job, maybe I got an
offer from two of those jobs. But it’s still Starbucks. So they’re not going to go
into a bidding war with themselves in
order to hire me. So that’s another way
in which labor market concentration decreases
job availability compared to the exact same
situation with the same number of jobs but where
every employer would be just posting one vacancy. So the upshot of this is
that lower labor supply elasticity means lower
competition and that restrains and other
features that I just talked to you about
lower the labor supply elasticity — is expected
to lower it and to lower the competition in
the labor market. So now, I want to tell you
about a new paper that I just finished a draft of
with my coauthor Jose Azar and Steve Berry. So in this paper, we use
a state-of-the-art model from industrial
organization, which is the branch of economics
that you would find most commonly in antitrust
litigation because it deals with issues of
pricing in the product market. But now, I want to import
those tools into the labor market and
specifically what I’m going to do is to use
application data, so looking at workers, what
kinds of jobs they apply to. So this is a big data
kind of project. Here you go, I
said the buzzword. So in the model, we have
the worker first choosing which market they’re
going to apply to. So I define a market by
an occupation at the SOC-6 level of detail and
the commuting zone. So such a market would
something like accountants and auditors
in D.C., okay? So at that first level of
decision, I decide as an accountant, am I
going to apply to accountant/auditor job
in D.C. or maybe in the, let’s say, Philadelphia
area, where I’m from? So based on the wages
in all those different markets, I’m going to
choose which particular market to target, okay? So that’s the
first stage. Then in the second stage,
I chose which market is the most worthwhile to me. I decide which particular
job opening to apply in that market. Let’s say I chose
Washington, D.C. accountants
and auditors. Now I’m going to choose a
particular vacancy as a function of the wage
and all the other job characteristics. So with this setup, we are
able to derive one very important measurement,
which is the market level elasticity of
applications. So that means how much
the market level wage, boosting that, how much
would that increase job applications to
the whole market. So again, if we think
about accountant and auditors in D.C., if I
increase the average wage in all vacancies, how
many more people are going to come from other
markets and apply to my market? This elasticity is highly
relevant because it speaks to the ability of a
hypothetical monopsonist, an employer who would
monopsonize this market, to lower wages. And so, if the elasticity
is low, that means that the hypothetical
monopsonist could lower wages quite a bit without
fear that the workers would go elsewhere. All right, so this is what
we are going to measure here. And in the next slide,
I’m going to show you results, looking at
different occupations, estimating their market
level elasticities and comparing those estimates
with the critical elasticity from the
hypothetical monopsonist tests. And the logic there is
that if the elasticity is low, and in this case
below that critical elasticity, then the
hypothetical monopsonist would be able to
profitably decrease wages by 5 percent, okay? So that’s the
fundamental test here. So let’s look
at the results. I’m sorry. This might be a little bit
hard to read in terms of the details. So you have there the most
common occupations in the U.S. and they are ranked by
their median elasticity. So for each occupation,
what you see there is a box plot. And I want to — because,
you know, in the interest of time, I want to direct
your attention to the median, which is the
little, you know, line in the middle of the box. So that’s the median. It means that 50 percent
have elasticity higher than that, 50
percent below. And this is estimated
for each occupation. Now, why does an
occupation have different elasticities? Because remember I’m also
considering the commuting zone. So some commuting zones,
therefore some areas in the country, might
have lower or higher elasticity. So that’s the whole range
of elasticities for each occupation. And you see the maximum
and the minimum are at the extremes. So what you see is that
in every, in every of the most common occupations,
the median elasticity is below the critical
elasticity, implying that this market definition
is plausible by the hypothetical monopsonist
test for all of these most common occupations. And because they are
common occupations, that would mean that there
are relatively many jobs because that’s how I
define common, by how many jobs are there in
these occupations. Presumably, for less
common occupations, it might be that the
elasticity is even lower. So furthermore, so okay,
just if you look, you can see that, for example,
the most elastic, so the areas, the kinds of
occupations with more competition are things
like sales representatives and telemarketers and the
least elastic occupations are in trucking, both
light truck and heavy truck have very low
elasticities, meaning that there is relatively
little competition in those specific types
of occupations. Furthermore, we also look
at is there a difference between high- and
low-skilled occupations and we don’t find a
systematic difference. Depending on how you look
at the data, sometimes the low-skilled or the
high-skilled occupations have high elasticities. So the upshot is that from
our data, there is no systematic difference in
competition between low- and high-skilled markets. Basically it depends what
market we’re looking at. So it’s important to look
at the particular market. You can’t assume that
obviously low-skilled markets means the workers
have so many opportunities and they can go anywhere,
so the elasticity is really high. That’s not the case. It might be the case
in some markets. But generally speaking,
there’s no systematic pattern right here. That’s important because
it goes against, again, a common intuition that,
as a low-skilled market worker, I can obviously
just go to another job. This data shows that
that’s not the case, at least not systematically. We also show in this paper
that rural areas have lower elasticities. So generally speaking,
if we are comparing less densely populated areas
with more densely populated areas,
systematically less densely populated areas
tend to have lower elasticity, meaning
lower competition. So that is something that
could be of concern for enforcement. So I just established
that an occupation by commuting zone is
a plausible market definition. Now let’s zoom in
on labor market concentration, which is
one of the reasons why competition might be low. So here, from a paper with
Jose Azar and Marshall Steinbaum, who’s here, and
Bledi Taska, we have data on all U.S. vacancies. And we look at the
concentration of U.S. labor markets
by geography. We find that the majority,
or 60 percent of labor markets are highly
concentrated. And furthermore, as we
can see, concentration tends to be higher in
less densely populated areas. Secondly, we also look at
the link between labor market concentration and
wages and we show that higher labor market
concentration is associated with lower
wages in that market. This graph you are seeing
here doesn’t control for anything. But in our analysis, we
looked at many potential confounding factors
and we find that the relationship is solid
to many potential other factors that we
have accounted for. And furthermore, there’s
now a huge literature that I am citing here
confirming this negative relationship in the U.S. between wages and labor
market concentration. And this uses very
different data sources and also different
market definitions. But the fact is there. I think at this point
can call it a stylized fact. Generally speaking,
higher concentration is associated with lower
wages in the labor market. So this is the end of my
remarks, and I’d like to give the floor to
Professor Prager. (Applause.) DR.
PRAGER: Great. Thanks, Ioana. Good morning, everyone. Before I get started,
I want to thank the Department of Justice, and
especially Doha Mekki for organizing
today’s workshop. I’m really delighted that
the DOJ and FTC are both taking issues of labor
market concentration quite seriously. So let me just pick up
right where Ioana left off. Ioana has already laid out
for us the evidence for a link between employer
concentration and worker pay. And this last plot that
she has put up is really a great example of what a
wide variety of papers in the academic literature
have recently established, which is that
labor markets with high employer concentration
tend to have low wages. The evidence on this
point, as Ioana said, is pretty compelling. But it doesn’t necessarily
have to imply that limited competition in
labor markets is itself the cause of lower wages. This is an important
point because if labor market concentration and
low wages tend to go together for some other
reason that isn’t necessarily causal, then
the agency’s attempt to shore up labor competition
may not result in the desired salutary effects
on labor itself. So let me give you one
very simplified example from actually Ioana’s map. And I think I had
another label on this. There we go. Okay. So this green low
concentration market that’s on the left of the
map where you’re looking is the San Francisco
commuting zone. Now, San Francisco, as you
know, is a dense urban area. It’s got many employers,
which means that labor concentration is
going to be quite low. No single employer or
small group of employers can by themselves
dominate the labor market. Wages in San Francisco
are also quite high by national standards,
especially if we compare them to a place like
Northern Maine. So that’s my other little
red-highlighted market at the top right of the map. This part of Maine
is quite rural, with relatively few employers
and therefore a very highly concentrated labor
market by the definition that’s been used
in the literature. Wages here are a lot lower
than they are in San Francisco, about
30 percent for hospital-employed nursing
administrators and pharmacists. But the cost of living is
also substantially lower. In fact, it’s about 40
percent lower than it is in San Francisco. So it’s actually not
clear that the workers in Northern Maine are the
ones getting the raw deal. Now, economists do control
for these kinds of patterns. And in fact, Ioana’s work
and some of the other papers that she has
mentioned also control of these kinds of issues. But I think this is still
a good illustrative example because it
highlights some of the issues that might come up. So in particular, if
you’ve got things that co-move with employer
concentration and wages that are a little bit
more complicated than cost of living, and especially
if they’re changing within a single labor market
over time, then that can contaminate the results. So what should be clear
hopefully from this example is that high
concentration going hand in hand with low wages is
not necessarily a sign that lack of labor market
competition is really to blame. And since the rest of
today’s workshop will be spent talking about
anticompetitive behavior and possible antitrust
remedies, I want to spend the remainder of my
remarks on why we should not dismiss these issues
out of hand, why, despite the fact that there are
other mechanisms that might explain at least
some of the patterns that Ioana has mentioned,
there’s reason to think that some of these
findings are indeed explained by high
concentration causing low wages. So before I do that, let
me first walk through another way of thinking
about the conceptual explanations for how
employer market power can result in low wages. So when employers are
concentrated and possess market power, that market
power can result in lower wages in a couple
of different ways. The first one that you’ll
often hear economists talk about is one that I refer
to as classical monopsony. Classical monopsony is
the theoretical construct that directly mirrors the
type of market power that a monopolist seller has
when it is the only seller of a good. So this is something Ioana
already mentioned, but let me give you sort
of the more typical definition, which is that
a classical monopsonist is an employer that is the
only game in town, meaning workers can either work
for that employer or be out of a job entirely. Now, for every worker,
as Ioana has already explained, there’s a
certain wage that the employer has to pay in
order to keep that worker there rather than
preferring to be unemployed. So for instance that
could be a wage that’s just enough to cover the
cost of the commute and childcare, as
a bare minimum. Normally employers have
to compete for a worker. And one way they do that
is by bidding up wages. But a monopsonist employer
can just decide, you know what, I’m going to pay
workers less than what the going wage would be
if the labor market were perfectly competitive and
I’m going to accept that that means I’m going to
get fewer employers, or sorry, employees. Some workers will not be
willing to work at that wage. So I’m going to have fewer
employees producing my good and generating
revenue. But on the other hand,
I’ll be able to pay them less and so I might
actually end up more profitable in
the long run. And that’s exactly this
hypothetical monopsonist test that was discussed. So classical monopsony
where the employer really has a great deal of labor
market power is one possible mechanism for
concentration to drive down wages. But this classical
monopsony picture that I’ve just described is
really more of a useful abstraction than a
description of how actual labor markets
work in practice. Very few employers have a
truly captive audience of potential workers, right? If conditions get bad
enough, some workers might pick up and move
to a new city. Others might invest
in retraining for a different career in a
different industry. So the other issue here
is that the classical monopsony mechanism really
only works for reducing wages if every worker in
the company that’s doing the same job is
paid the same wage. Otherwise, it’s not
really a classical monopsonist. So in practice, what that
ends up meaning is that a much larger concern about
employer concentration is what it does to workers’
so-called bargaining leverage. When an employer grows
bigger, whether that’s organically or via merger
and acquisition activity, that means workers in that
industry now have fewer employers competing to
hire them, fewer employers trying to win them over
by offering pay raises or improvements in working
conditions and frankly just fewer new job
openings or vacancies, in Ioana’s terminology,
outside of their current employer where
they already work. This means that a
juggernaut employer can put downward pressure on
wages and that can push down wages that are the
going market wage and affect workers who are
working throughout that labor market, not just at
the large employer itself. So basically when an
employer concentrates or consolidates, that’s going
to lead to worst outside options for workers and
that will eventually affect equilibrium wages
in that labor market. And unlike classical
monopsony, this bargaining leverage is
something that can work even when workers doing
the same job aren’t necessarily all
paid the same. So it’s much more
applicable to real- world labor markets than a
classical monopsony mechanism might
appear to be. But whether it’s classical
monopsony or bargaining leverage, the end
result is the same. It’s that higher employer
concentration is going to put downward pressure on
wages and that can indeed be a causal mechanism. Now, thinking about
employer concentration leading to employers
having market power over workers requires a little
bit of mental gymnastics if you’re coming from
a standard antitrust background. In most antitrust
applications, the entity that has the market power
is the producer or seller of a good or service and
the entities that can be harmed are the
purchasers of that good. When we’re talking about
labor market power, we have to turn this
intuition on its head. In other words, the
entity that has the market power is now the
purchaser of workers’ labor and the entities
that can suffer harm are the workers that sell
their services to the employer. But the good news is that
the intuition that you already possess from
applications where the seller has monopoly power
translates pretty easily into applications where
the buyer has monopsony power. You just have to
invert a few things. So we’ve inverted who
has the market power. It’s now the buyer of
the labor instead of the seller of the good. And we’ve got to invert
what the effect is. The powerful buyer wants
lower prices of labor instead of a powerful
seller wanting higher prices of the good
that it is selling. And so, the effect is to
drive wages down instead of driving
product prices up. With that inverted
intuition in place, we can also think about extending
many of the familiar tests from antitrust and many of
the familiar tools to the labor market
power context. The hypothetical
monopsonist — or sorry, the hypothetical
monopolist test, which we would normally use to
ask whether a monopolist seller could, in
principle, increase prices by a small but significant
and non-transitory amount becomes the hypothetical
monopsonist test which then asks whether a
monopsonist employer could in principle decrease
wages by a small but significant and non-
transitory amount. Similarly, the upward
pricing pressure index, or UPP, which normally
measures the relative price markup that a
merger can generate can become the downward wage
pressure index to measure the relative mark down in
wages that an employer merger might generate. The bottom line is that
although we’re used to thinking of market power
in the hands of sellers, the tools that antitrust
practitioners have also developed can also equally
be used to analyze market power in the hands
of employers. So let’s say we have the
tools to analyze employer market power. The question is does that
mean the DOJ or FTC could actually act to enforce
competition in labor markets when competition
is threatened. Well, we’ve gotten some
examples of that in the introductory
remarks already. And here again, I’d like
to draw some parallels to what we’re used to
thinking about from enforcement in
product markets. So for example,
existing high levels of concentration in market
power are not actionable in and of themselves. So even if there were
only a handful of key employers in Northern
Maine, the agencies can’t just break up those
employers without good reason. And a good reason
might be something like anticompetitive conduct
by those employers. So if no-poaching
agreements in place in which they agree
not to hire each other’s workers away by offering
them a raise, that’s essentially analogous to
price-fixing by sellers or if those employers have
punitive, unnecessarily punitive non-compete
clauses that prevent their workers from being able
to leave and work at a different job, that’s
something analogous to a predatory contract
termination fee. So these are the types
of cases that might be actionable, and they’ll be
discussed in a lot more detail in the
second panel. So I won’t get
into the weeds now. But what I do want to draw
your attention to is a second potential area for
enforcement, which is mergers of employers. The horizontal merger
guidelines, of course, outline thresholds such
that if a proposed merger of two sellers increases
product market concentration by more than
that threshold amount, then the merger is
automatically subject to some scrutiny. Analogously, if a
merger of two employers increases labor market
concentration, then there’s an argument to be
made that that should or could also be subject to
similar types of scrutiny from the agencies. And crucially here,
economic theory is informative about when we
might expect an employer merger to generate such
downward wage pressure. If workers can’t easily
move to a different job, whether that’s because
they’re geographically tied to their area
because they prefer things that are close to home
or because it would take years of training to
switch to a different occupation with a similar
level of pay or because the work they do is
highly specific to that employer, like operating
high-tech, custom-built machinery, then the
merger will generate more downward wage pressure
for those workers. So as an example, and to
my knowledge, this is purely hypothetical,
suppose Amazon and Microsoft were to merge. Purely hypothetical. They are currently the
two tech giants in the Seattle area, of course. And economic theory tells
us to expect that tech workers who can easily
pick up and move to Silicon Valley after
Amazon and Microsoft merge are going to be much less
affected than Seattle-area tech workers who are tied
to Seattle and are then probably going to have to
accept lower wages or at least slower wage growth
from the new supergiant “Amaz-Soft” or however
you want to abbreviate the two. Now, since mergers aren’t
really a focus of any of today’s panels but
are presumable are of interest for this
audience, I want to spend the rest of my remarks
elaborating on them a little bit. I discussed earlier why
the fact that wages tend to be lower in places
that have high employer concentration doesn’t
necessarily mean that the lower wages are about the
concentration itself, although the evidence
does seem to point in that direction. It could be, you know,
correlation rather than causation. So for determining
whether labor market concentration can actually
cause low wages, we can use some of the tools
that Ioana described earlier. But there are also other
types of evidence that we might turn to in order
to bolster the case. It’s useful, for example,
to look at a different type of evidence where
concentration is changing in a local labor market. And we can look at what
happens to wages as that concentration changes. Mergers are a very
natural way to do this. And that’s what I did in
a recent academic study with my coauthor,
Matt Schmitt. In the study, we examined
what happens to hospital workers’ pay growth
after a hospital merger. We looked at 10 years
of consummated hospital mergers in the U.S. from 2000 through 2010. And we categorized
hospital labor markets as falling into one of five
bins, depending on their merger activity. Over a given time period,
a labor market might experience no
within-market hospital mergers, in which case
it’s assigned to our empirical control group,
or a labor market might experience hospital
mergers that affect the labor market concentration
to differing degrees. We measure how much
concentration is affected by a merger using the
change in the employment Herfindahl index
induced by the merger. So this is actually the
same measure that shows up on Ioana’s maps. And if employer HHI or
Herfindahl goes up by a lot due to the merger,
then the labor market becomes less green and
more red on one of her lovely maps. We binned our mergers
into four categories, from the smallest to the
largest increases in HHI. So now let me walk you
through sort of the economic hypotheses this
theory might suggest to us. If employer concentration
actually causes downward wage pressure, then what
we would expect to see is that wages grow slower
after the mergers that generate the largest
increases in local labor market HHI. If wages in those markets
grow at the same rate as in other markets that
either have smaller mergers or no merger
activity, then that might imply that increasing
employer concentration isn’t actually what’s
slowing down the wage growth. Similarly, if employer
concentration is in fact the culprit, then we would
expect to see workers that re more specialized
to the hospital industry being more affected by
wage slowdowns following a hospital merger. And in the study, the way
that we look at this is by binning workers
into three categories. We’ve got some blue collar
workers like custodial staff and cafeteria
workers, who are generally not very specialized to
the healthcare context. We’ve got some higher
skilled, but still somewhat specialize and
somewhat generalist workers. These are mostly white
collar stuff, folks like social workers and
insurance claims professionals. And then our third and
most highly specialized category of workers are
nursing administration people and pharmacists,
who are the most healthcare-specific
workers that we can observe in the data. So concentration
actually causes downward wage
pressure, the group we would expect to see
affected the most by hospital mergers are the
nurses and pharmacists. And we would expect to see
those effects most of the largest mergers that move
local employer HHI the most. And in fact, that is
exactly what we see. So for our least
specialized blue collar category, which are the
four bars on the left of the plot, we see no
effects of hospital mergers on subsequent wage
growth, no matter the size of the concentration
increase induced by the merger. For t specialized
worker groups, so our sort of white
collar non-medical staff in the middle and then
the nursing and pharmacy category being the
right-hand set of bars, we actually do see wages
growing slower following the largest HHI increase
in mergers, although not so much when you look at
the smaller mergers that only move HHI
by a little bit. Now in the paper, we’ve
got a bunch of additional analyses that suggest
these estimates are genuinely picking up an
effect of employer labor market power on wages
rather than other mechanisms. But the bottom line that
I want you to take away from this is what we have
corroborating evidence that at least some of the
patterns that Ioana has documented in terms of the
correlation between high concentration and low
wages really can be attributed to a story of
probably causation rather than just correlation. So what that means is that
an antitrust environment that supports competitive
labor markets might actually be effective at
combating downward wage pressure. And that means that the
discussions that’ll take place for the rest of
today are both meaningful and important. So I encourage everyone to
stick around for the rest of the day. Thanks. (Applause.)

Leave a Reply

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