00:00:00
Well, we certainly know that Amazon is
00:00:02
dominating the retail segment right now
00:00:04
thanks to the growth of e-commerce, but
00:00:06
what do Amazon fulfillment centers mean
00:00:09
for local economies in those towns or
00:00:13
townships where those facilities are
00:00:15
located? Some research done uh earlier
00:00:18
this year by our guest Sergey Netine who
00:00:21
is a professor of operations,
00:00:22
information, and decisions here at the
00:00:24
Wharton School. He's also senior vice
00:00:26
dean for innovation and global
00:00:27
initiatives. He was part of the research
00:00:29
team looking into this and he joins us
00:00:31
right now. Sergey, great to catch up
00:00:33
with you again. How are you, sir?
00:00:35
>> Very good. Thank you for having me, Dan.
00:00:37
>> Thank you. I think this is an
00:00:39
interesting topic because we all I think
00:00:42
most of us have communities where we've
00:00:45
seen a business come in, a big business,
00:00:48
a well-known business, and you kind of
00:00:50
wonder it's you think, "Wow, that's got
00:00:52
to be great for the community."
00:00:55
But you you dive into the data as to
00:00:58
whether or not or how much it is a great
00:01:01
thing for these local communities.
00:01:03
First, I guess give us the backstory on
00:01:05
why this was such an important or an
00:01:08
interesting piece for you and your team
00:01:10
to look at.
00:01:11
>> Certainly, Dan. So uh the backstory uh
00:01:15
is that uh there are lots of studies
00:01:18
done previously on Walmarts and uh what
00:01:21
we know is that when Walmart comes into
00:01:23
town um
00:01:26
lots of bad things happen. First of all,
00:01:29
uh prices go down which kills retail
00:01:32
around Walmart and many mom and pop
00:01:34
stores um go bankrupt and then their
00:01:38
employees get hired by Walmart at a much
00:01:41
lower wage. And so there are several
00:01:44
papers over the years uh some back in
00:01:47
the '9s that shows that essentially an
00:01:49
effect of Walmart coming into town if at
00:01:52
best it's it's zero and and probably a
00:01:55
little bit of a negative. Um then there
00:01:58
are some studies done on uh big
00:02:01
factories um that employ thousands of
00:02:03
people and and we know that the effect
00:02:05
is typically positive. Um beyond that we
00:02:09
don't really know much about big
00:02:12
facilities coming into town and what's
00:02:14
interesting about Amazon is that it has
00:02:17
many of those facilities. There are
00:02:19
literally hundreds and hundreds of uh
00:02:21
warehouses which are popping up all
00:02:23
around the world in the last decade or
00:02:25
so. um and they employ significant
00:02:28
number of people. So that was kind of
00:02:30
you know the the academic interest in
00:02:32
me.
00:02:33
>> But usually and I maybe this is a little
00:02:35
bit of the policy side of it. When a
00:02:38
company like Amazon or UPS or whoever it
00:02:40
is or or an automaker is looking to
00:02:43
build a facility in a certain area, the
00:02:46
local government looks to bend over
00:02:49
backwards to try and give that company
00:02:51
tax breaks, benefits, whatever, so that
00:02:54
it makes it more palatable for the
00:02:56
company to come into that uh into that
00:02:58
town and and uh and put up a facility.
00:03:02
uh that is absolutely true and uh in our
00:03:06
case we don't know how much local
00:03:09
governments uh contributed it it's it's
00:03:11
a very complicated formula there are
00:03:13
some tax breaks there are some you know
00:03:16
land subsidies and so on um in certain
00:03:19
cases we see some local news about how
00:03:21
much the government tried to attract the
00:03:24
facility certainly [clears throat] um
00:03:26
there was lots of publicity about Amazon
00:03:28
HQ2 when it uh when it was open in in
00:03:32
Virginia. Uh so we see bits and pieces.
00:03:35
Um but that's not the data that we had
00:03:37
unfortunately.
00:03:38
>> Right. And and so in terms of what you
00:03:41
were looking at, I guess employment is
00:03:43
part of it. Uh median income is probably
00:03:45
part of it and and I read that probably
00:03:48
poverty looking at the poverty side of
00:03:50
this is also included as well. Exactly.
00:03:53
Yeah. So we looked at those three
00:03:55
variables and we looked at the county
00:03:56
level. You know there are different
00:03:58
levels you can look at at state probably
00:04:00
at the state level uh it's hard to see
00:04:02
an impact because states are big and you
00:04:05
know if you go all the way down to say
00:04:07
zip code level or um you know census
00:04:10
block level there's not much data
00:04:12
available. So typically these kind of
00:04:14
negotiations with the local uh
00:04:17
government they happen at the county
00:04:18
level and and that's where we find uh a
00:04:22
significant increase in employment and
00:04:25
we find a significant decrease in
00:04:27
poverty and we find a significant
00:04:30
increase in uh median household incomes.
00:04:33
is the expectation that these I guess it
00:04:36
has to be that there's a long-term
00:04:39
element to having a facility like this.
00:04:41
Like if Amazon's going to make an
00:04:43
investment to put a big fulfillment
00:04:45
center in your area, the expectation is
00:04:47
they're going to keep it there for a
00:04:49
while. And so that level of employment
00:04:51
is probably somewhat constant as well
00:04:54
that it's going to be there for a while.
00:04:56
>> Absolutely. and and the numbers that we
00:04:58
find they're average over the first
00:05:00
three years but certainly uh they
00:05:03
continue they persist for a long time
00:05:06
period. uh I think what is the most
00:05:09
interesting thing that we find is that
00:05:11
there is direct employment by Amazon say
00:05:14
you open a facility you have 3,000
00:05:16
people say working there some are
00:05:18
smaller some are bigger but what we find
00:05:21
is that there is a multiplier effect of
00:05:23
about two so for every direct job that
00:05:26
Amazon creates by employing people
00:05:28
directly uh there is there are more jobs
00:05:31
created around this facility and as you
00:05:33
can imagine many of those jobs are
00:05:35
probably in deliveries right so Someone
00:05:37
has to hire these people UPS, USPS. Then
00:05:41
uh then there are companies popping up
00:05:43
that are delivery service providers for
00:05:45
Amazon. Then there is Amazon Flex which
00:05:48
are gig workers uh also working and
00:05:50
delivering packages. And then of course
00:05:52
there is retail. You have to feed people
00:05:54
who work at the warehouse for a while.
00:05:57
There is some construction going on.
00:05:59
Then someone has to do maintenance and
00:06:00
so on. So that's kind of an important um
00:06:03
important uh thing to know that actually
00:06:06
more people get jobs than what directly
00:06:09
uh is offered by Aza and so are you able
00:06:12
to tell as well not only what happens to
00:06:14
the incomes of the workers who are there
00:06:17
obviously you know they're going to get
00:06:18
paid for the work they're doing but that
00:06:21
that other component that outside
00:06:22
component like is is it somewhat of a
00:06:25
rising tide raises all boats philosophy?
00:06:28
>> Uh exactly. So the median household
00:06:30
income is for the entire county, right?
00:06:33
And so um and and we show that this
00:06:36
increase is about 2.3%
00:06:39
for the entire county which of course
00:06:41
means for people who got employed
00:06:43
directly or indirectly the increase is
00:06:44
probably much much higher. Um but uh
00:06:48
overall when you even aggregated for a
00:06:50
much bigger county you still observe
00:06:53
increase in uh in median. But the the
00:06:56
poverty question of this I think is is
00:06:58
very important. It's very interesting as
00:07:00
well that if you have these components
00:07:02
of rising employment and an element of
00:07:04
of rising income, you would expect that
00:07:08
there's going to be a a downside impact
00:07:11
on the poverty level in a particular
00:07:13
area.
00:07:15
>> Exactly.
00:07:17
other other work that we've done at
00:07:19
Amazon where um uh we like looked at and
00:07:23
you know I have to say that I I am I am
00:07:26
working at Amazon one day a week and
00:07:28
that's where the data comes from data
00:07:30
come directly from Amazon so kind of a
00:07:33
full disclosure there we do find that a
00:07:37
lot of people join Amazon this is as a
00:07:40
first job so they did not have a job
00:07:42
before they did not have income before
00:07:44
and so for them it's kind of a drastic
00:07:46
increase uh in income. Uh others join
00:07:50
from uh fast food and and services and
00:07:53
other kind of retail and because Amazon
00:07:57
uh has substantially higher pay than
00:07:59
many of those places. So you know Amazon
00:08:01
established $15 minimal pay across the
00:08:04
company uh for a while since 2018 for
00:08:07
example. uh that also leads to this
00:08:10
increase in uh median householding. A
00:08:14
>> and so does data like this and and maybe
00:08:16
this is part of the process that Amazon
00:08:19
had when they started to think about
00:08:20
these facilities, but does it impact the
00:08:24
decisions of where they put these
00:08:26
fulfillment centers so that they can
00:08:28
have the maximum benefit for the people
00:08:31
in a particular area?
00:08:33
>> Um absolutely. So there is a lot of
00:08:37
thought that goes into locating a
00:08:39
warehouse and and there are all kinds of
00:08:41
considerations going. One is of course
00:08:43
being close to demand. Um so essentially
00:08:47
what we see is uh there are very few
00:08:49
kind of uh uh counties with very small
00:08:52
population where warehouses would be
00:08:54
located and that makes sense. There is
00:08:55
just no one there to shop around. Um
00:08:58
another one is locatedating close to
00:09:00
supply. So you you really want some
00:09:03
people uh to work at the warehouse and
00:09:05
if you locate in the middle of the
00:09:06
desert there's no one going to be there
00:09:08
to work for. Um and so we had to account
00:09:12
for this because it's not
00:09:13
straightforward. It's not like warehouse
00:09:15
gets selected randomly. So if you just
00:09:17
look at the impact of Amazon warehouse,
00:09:20
you might get into this issue that
00:09:21
warehouses allocated within patch,
00:09:24
>> right? And it's not like they're going
00:09:25
to put a, as you kind of alluded to,
00:09:27
they're not going to put a facility in a
00:09:29
desert and then all of a sudden the town
00:09:31
is going to pop up around it.
00:09:34
>> No, no, you also have to have other
00:09:36
things, you know, some roadexes and, you
00:09:38
know, some uh because trucks are going
00:09:41
to be going back and forth and so on. Um
00:09:43
and of course if you look at big
00:09:45
counties um all of the big counties
00:09:47
which which would be uh million plus in
00:09:50
population they all have Amazon
00:09:52
warehouse because that's demand that's
00:09:54
supply right um and so we've been
00:09:57
looking mostly at mediumsize uh counties
00:10:00
because the way we look at this this
00:10:02
problem we look at the county that gets
00:10:05
an Amazon warehouse and then we compare
00:10:07
its development to a county that did not
00:10:10
get an Amazon warehouse and then we see
00:10:12
the differences. So that's kind of a
00:10:14
methodological design.
00:10:16
>> Anything stand out that that really
00:10:18
grabbed you from doing this research?
00:10:21
>> Um I I think what's remarkable is how
00:10:24
strong the the effects are and you know
00:10:27
we used several methodologies. We also
00:10:30
used some public data uh to uh to test
00:10:34
robustness of our results and and
00:10:36
results were very very strong. Um there
00:10:38
was some methodological challenges like
00:10:41
because of the COVID we had to stop our
00:10:44
timeline in 2017 because CO is kind of
00:10:47
weird. You lots of things went
00:10:49
differently and so we could not look at
00:10:51
the most recent years and and and
00:10:54
frankly the the biggest impact is
00:10:56
probably more recently because Amazon
00:10:58
increased its uh minimum wage to $15 in
00:11:01
2018. So I think if anything uh our
00:11:04
results are are very kind of a
00:11:07
conservative um on the conservative
00:11:09
side. Uh so that I think is one thing.
00:11:12
Um and then of course the multiplier. So
00:11:14
we found that multiplier is two. So for
00:11:16
every person that Amazon hires another
00:11:18
job is created elsewhere in the county.
00:11:21
So that is remarkable and this kind of
00:11:24
results in the past academics found uh
00:11:26
for big factories. Um but nothing like
00:11:30
that has ever been observed say for
00:11:32
Walmart or you know for other uh retail
00:11:34
facilities.
00:11:38
Thanks very much for your time today.
00:11:40
>> Thank you. Thank you Dan for having me.
00:11:42
Great pleasure as always.
00:11:43
>> Yes sir. Sergey Netine who is a
00:11:45
professor of operations information and
00:11:47
decisions here at the Wharton School.
