Popular in Social Sciences

Books

City Time: On Being Sentenced to Rikers Island

Independent Scholar David A. Campbell

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

Economics

Gender and Economics

University of California Santa Barbara Shelly Lundberg

American Politics

Keeping the Republic

Boston College Marc Landy

Banks and Policy Organisations

Greening the financial sector

European Investment Bank Matteo Ferrazzi

British Politics

Entrepreneurial thinking: the politics and practice of policy impact

University of Sheffield Matthew Flinders

Culture and Society

Carceral Entrapments: Views from the Prison/Street Interface in India

Indian Institute of Technology Delhi Mahuya Bandyopadhya

Measuring Deterrence Motives in Dynamic Oligopoly Games

View more insights in Social Sciences - Economics - Microeconomics

View more figures from Cornell University

View the page of Nathan Chi-Chung Yang

Read the Study

Nathan Chi-Chung Yang presents a novel decomposition approach for measuring deterrence motives in dynamic oligopoly games.

Image courtesy of interviewee. January 31, 2024

Generating Workspace

Export as PDF

This video video discusses the phenomenon of market saturation, where multiple stores of the same brand, such as Starbucks or 7-Eleven, are found in close proximity. It poses the question of whether this business strategy, often seen as cannibalization of profits, can have economic benefits. The speaker, drawing from an economics background, explores the idea that having multiple locations deters competitors from entering the market, providing a strategic advantage.

An extensive review of literature over several years revealed that while it is challenging to quantify deterrence motives without direct data (like internal company discussions), existing economic models, particularly a dynamic oligopoly model, can be adapted to analyze this behavior. The research involves developing a framework to understand the incentives for firms to expand strategically, considering both competitive dynamics and forward-looking behavior.

The study also highlights a movement toward more standardized measurement of preemptive market strategies. The findings suggest that a notable percentage of Starbucks' expansion motivations can be attributed to preemptive considerations, with estimates ranging from 5% to 30%. This video video research aims to provide valuable insights for future studies and practical applications in fields like antitrust and corporate finance, as understanding these motivations could help in predicting firm behavior and assessing risks related to expansion strategies. This video video concludes by linking preemptive motives to broader discussions of retail success and failure, suggesting that overexpansion—especially when financed through debt—can contribute to the struggles faced by many retail chains.

Language:

Export as PDF

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.565 --> 00:02.165
Probably many of us have encountered,

00:02.665 --> 00:04.965
and in fact, I've seen like comedians kind

00:04.965 --> 00:08.645
of quip about this, in their acts where, you know,

00:08.665 --> 00:10.445
you drive down along a street

00:10.625 --> 00:14.685
or you walk, you know, down a, in downtown.

00:14.785 --> 00:17.725
And you know, where, what you see is you see a Starbucks

00:17.745 --> 00:21.565
across the street from another Starbucks, I think in,

00:21.805 --> 00:23.645
a subway station in Tokyo.

00:23.985 --> 00:26.965
You know, there are four, seven elevens in the same station.

00:27.985 --> 00:30.885
And, you know, coming from an economics background,

00:31.755 --> 00:34.885
this type of observation, what seemed a little,

00:35.065 --> 00:36.285
counterintuitive to me,

00:36.285 --> 00:39.445
because, you know, there's this notion of cannibalization

00:39.655 --> 00:42.165
where, you know, if you set up a bunch of shops

00:42.185 --> 00:43.365
and close proximity,

00:43.365 --> 00:46.405
they're gonna eat each other's kind of profits, if you will.

00:47.105 --> 00:51.285
And so that kind of led to kinda a, a rabbit hole,

00:51.505 --> 00:55.205
if you will, of me kind of spending the past decade

00:55.345 --> 00:59.045
or so, kind of looking several years looking into this

00:59.365 --> 01:01.165
phenomena across various papers.

01:02.345 --> 01:05.845
And, and so the phenomena I was particularly interested in

01:06.025 --> 01:09.845
is whether there's actually an economic benefit of them kind

01:09.845 --> 01:10.965
of saturating a market.

01:11.345 --> 01:14.405
And in fact, there could be, if,

01:14.455 --> 01:17.325
there is a benefit associated with deterrence where,

01:17.665 --> 01:19.205
you know, you take up that spot,

01:19.505 --> 01:21.165
if McDonald's takes up a spot,

01:21.275 --> 01:23.965
that means Burger King can't, can't do that.

01:24.025 --> 01:26.325
And in fact, you know, if you read the autobiography

01:27.025 --> 01:31.285
of I think Walmart, the, the book by one of the Waltons,

01:31.585 --> 01:34.005
you know, they, they, they even allude to the fact

01:34.005 --> 01:36.085
that they will kind of, they, they want to,

01:36.085 --> 01:38.525
they set up a Walmart in close proximity

01:39.025 --> 01:41.125
of another Walmart just to keep away

01:41.855 --> 01:44.565
their competitor at the time, which was Kmart.

01:44.945 --> 01:47.685
Um, even, even though they knew this would,

01:47.685 --> 01:50.085
potentially eat away at their existing store sales.

01:50.145 --> 01:53.935
And so this, this kind of behavior, the strategic behavior,

01:54.205 --> 01:56.255
this economic benefit is, is,

01:56.355 --> 01:59.375
has been particularly interesting, to me.

01:59.915 --> 02:01.695
And I think that the, the thing

02:01.695 --> 02:05.625
that makes it intellectually, satisfying to study is,

02:05.645 --> 02:09.625
is actually how challenging it is to, quantify this.

02:09.625 --> 02:12.745
Because if you think about things like deterrence motive,

02:13.525 --> 02:16.545
it's hard to quantify this.

02:16.545 --> 02:19.345
There's no kind of observable data,

02:20.225 --> 02:22.545
directly observable data that indicates, you know,

02:22.545 --> 02:23.785
that they are doing this

02:23.785 --> 02:25.825
because they want to keep competitors away.

02:25.825 --> 02:28.625
Unless you, you have a subpoena on these companies

02:28.625 --> 02:30.865
and you actually can wiretap their,

02:31.145 --> 02:32.905
internal conversations, which, you know,

02:32.905 --> 02:35.665
typically people don't have in this paper that,

02:35.805 --> 02:36.865
we're talking about.

02:37.045 --> 02:41.305
Uh, we use a fairly standard, model of

02:41.765 --> 02:45.145
how firms kind of, open, expand.

02:45.415 --> 02:46.505
Basically it's, you could think

02:46.505 --> 02:49.865
of it like a firm investment model where there, there's a,

02:49.865 --> 02:52.825
some costs associated with, opening a store.

02:52.835 --> 02:55.105
There could be a, there could be costs associated

02:55.105 --> 02:56.225
with closing a store.

02:56.645 --> 03:01.485
And so what this model does is that it, it captures the,

03:01.635 --> 03:05.725
what it may me, what the model aims to do is

03:05.725 --> 03:10.245
to internalize, the incentives these firms have

03:10.425 --> 03:12.725
for these strategic investment decisions.

03:12.725 --> 03:16.205
And so there's essentially two important ingredients,

03:16.205 --> 03:18.165
if you will, or dimensions that this,

03:18.165 --> 03:20.605
these models in general need to capture.

03:20.825 --> 03:24.285
If one wants to study, preemption, the first is that,

03:24.585 --> 03:27.285
there has to be a strategic component, right?

03:27.305 --> 03:31.165
And so these firms need to be making decisions knowing

03:31.165 --> 03:34.085
that their decisions have implications on competitors.

03:34.345 --> 03:37.245
And the same thing, you know, they also expect competitors

03:37.245 --> 03:40.965
to be equally or reasonably strategic as well, right?

03:41.225 --> 03:44.085
And so it's, you, you kind of study in, in what we call kind

03:44.085 --> 03:45.605
of a game type of setting.

03:45.905 --> 03:49.885
And the second kind of dimension of, of that importance is,

03:49.905 --> 03:51.805
is this notion of forward looking behavior

03:51.985 --> 03:56.565
or your investment decision today is not only,

03:57.385 --> 04:00.565
aimed at optimizing your benefit today,

04:00.665 --> 04:03.045
but you also are trying to look ahead

04:03.545 --> 04:06.845
and see, okay, you know, if we open a store here, the,

04:07.035 --> 04:09.685
this is the implication it'll have on a market structure,

04:09.865 --> 04:13.125
not just today, but maybe tomorrow and date, you know, year

04:13.285 --> 04:14.325
after, and so on and so forth.

04:14.745 --> 04:18.005
And so, the model essentially needs

04:18.005 --> 04:19.765
to have these dimensions.

04:19.765 --> 04:23.055
And once you kind of combine these dimensions together,

04:23.485 --> 04:26.575
then it, it, it, it kind of offers us this kind

04:26.575 --> 04:28.895
of canonical, dynamic oligo model.

04:28.915 --> 04:32.575
And so the earliest, what among the earliest versions

04:32.575 --> 04:34.815
of these models were kind created in the eighties,

04:34.895 --> 04:36.495
I think the very, kind

04:36.495 --> 04:41.315
of the workhorse model is the Ericsson peku,

04:41.575 --> 04:46.095
dynamic oligopoly model, which, many, many of,

04:46.275 --> 04:48.535
you know, a lot of the modern research today still,

04:48.795 --> 04:50.015
continues to use.

04:50.275 --> 04:53.815
And so, yeah, so we, we were, we're essentially kind of,

04:54.115 --> 04:58.215
building off of, a a lot of the great kind of,

04:58.315 --> 05:00.855
at least analytical infrastructure

05:00.855 --> 05:04.255
that has already been, developed by,

05:04.395 --> 05:05.455
the past scholars.

05:05.995 --> 05:08.175
Our contributions are multi multifold.

05:08.235 --> 05:10.255
So the first is, you know,

05:10.335 --> 05:13.335
a little more theoretical analytical, where a lot

05:13.335 --> 05:16.175
of our results are actually kind of using,

05:16.635 --> 05:17.655
you know, pen and paper.

05:18.035 --> 05:21.295
Uh, not, not literally, you know, most of us now use iPads

05:21.295 --> 05:22.295
and and whatnot.

05:22.315 --> 05:23.655
But you know, we, we, we kind

05:23.655 --> 05:27.295
of build out the results on paper, if you will, so

05:27.375 --> 05:29.585
that we have a fairly

05:30.265 --> 05:34.465
concrete analytical notion about what is the value

05:34.605 --> 05:35.865
of preemption, right?

05:35.885 --> 05:40.225
And so we essentially need a moral framework in order

05:40.765 --> 05:44.625
to kind of decompose the various incentives firms have

05:45.205 --> 05:47.505
for expanding into the new markets.

05:48.245 --> 05:51.945
And by doing so, by kinda writing it out explicitly,

05:51.945 --> 05:53.625
which is not trivial,

05:53.625 --> 05:56.545
because, you know, you have all these different dimensions

05:56.545 --> 05:58.885
of strategy and forward looking behavior.

05:59.225 --> 06:01.725
And so when you're trying to decompose things like bill in

06:01.925 --> 06:04.445
equations, it becomes a fairly thorny task.

06:05.295 --> 06:07.475
And which is why we think there's a, a theoretical

06:07.995 --> 06:10.715
contribution to this paper where, you know, once you kind

06:10.715 --> 06:13.395
of, do this accounting, if you will,

06:13.495 --> 06:16.875
of all these different incentives, you can then say, aha,

06:16.905 --> 06:21.635
here is the specific benefit the firms have from

06:22.385 --> 06:25.915
expanding and as a result of this expansion,

06:26.175 --> 06:28.235
the added profits they get in the future

06:28.265 --> 06:32.035
because their competitors chose not to set up shop

06:32.035 --> 06:33.355
because of them, right?

06:33.375 --> 06:36.475
And so you, you could, you could tell just this, I,

06:36.515 --> 06:39.115
I felt like this is the, the, the, the simplest way

06:39.115 --> 06:42.035
of kind of characterizing this in simple words.

06:42.095 --> 06:45.435
And so you can imagine to translate this into kind

06:45.435 --> 06:48.035
of the analytical components, it's a,

06:48.035 --> 06:49.035
it's a little bit challenging.

06:49.335 --> 06:52.195
But the thing is, once you have, once we, we,

06:52.455 --> 06:54.115
we developed this framework,

06:54.135 --> 06:58.635
and once we had done this, decomposition on paper,

06:59.175 --> 07:02.475
it then makes the kind of empirical exercise a lot simpler

07:02.475 --> 07:05.315
because we, you know, you basically have a formula,

07:05.575 --> 07:08.595
if you will, for how to calculate this.

07:08.935 --> 07:12.795
Uh, it's, it's kind of hidden this intrinsic motivation

07:12.895 --> 07:16.595
for this intrinsic value of preempting the markets,

07:16.645 --> 07:19.515
which you can then kind of put to the data, right?

07:19.515 --> 07:23.955
So once you, once you have this kind of quantity written out

07:24.595 --> 07:27.795
analytically, it's really just a matter of estimating

07:27.855 --> 07:31.115
and fitting a dynamic oligo model with the data,

07:31.405 --> 07:33.715
which is the, in our empirical application,

07:33.745 --> 07:34.955
it's the coffee chain data.

07:35.535 --> 07:38.035
And once you have the model fitted to the data,

07:38.575 --> 07:40.915
you can then just plug in a bunch of values

07:41.255 --> 07:45.675
and just calculate, okay, you know, like 20, 35%

07:45.675 --> 07:49.795
of the motivation for Starbucks is driven by preemptive,

07:50.055 --> 07:51.235
considerations.

07:51.775 --> 07:55.435
And so we, as, I guess as the authors of the paper, we,

07:55.735 --> 07:58.115
our hope was we did the hard work so

07:58.115 --> 08:00.555
that future researchers can just plug it and play

08:00.575 --> 08:04.195
and kind of use a lot of these kind analytical insights in,

08:04.495 --> 08:05.555
you know, subsequent work.

08:05.555 --> 08:08.875
So that, it's, it's a little more concrete.

08:09.435 --> 08:12.715
Starbucks, of course, is perhaps the retailer that we,

08:12.715 --> 08:16.475
that the coffee chain that we notice, has the strongest,

08:16.935 --> 08:20.875
I guess motivation for, preempting the market.

08:20.935 --> 08:22.795
And so the, the beauty about our measures

08:22.795 --> 08:27.115
that we can actually, it's kind of a relative term,

08:27.255 --> 08:29.955
so we can actually say, okay, you know, X percent

08:30.255 --> 08:33.475
of their motivation is driven by preemptive considerations,

08:34.055 --> 08:37.915
and then 95% is just from all other kind of incentives.

08:38.015 --> 08:40.995
And so we, we actually find a bit of a range, you know, I,

08:41.035 --> 08:45.075
I think it, it could range from like around 5% to 30%,

08:45.075 --> 08:46.915
something in that ballpark.

08:47.375 --> 08:51.885
And so, you know, it, it, it, it's, it's, I believe it's

08:51.885 --> 08:54.805
among the first papers out there that actually provides kind

08:54.805 --> 08:57.485
of a percentage, measurement.

08:57.545 --> 08:58.565
In my earlier work.

08:58.785 --> 08:59.925
You know, the way we,

09:00.165 --> 09:03.765
I would've measured this would've been, like in,

09:03.765 --> 09:07.445
in a paper back in 2016 in quantitative economics

09:07.445 --> 09:08.565
with a different co-author.

09:08.905 --> 09:12.685
Um, you know, we had, in order to quantify this incentive,

09:12.745 --> 09:14.645
we did it in a more indirect way

09:14.935 --> 09:16.445
where we said, okay, here's a model.

09:16.465 --> 09:19.405
And it's a very similar model to what this,

09:19.515 --> 09:22.125
this current paper, this more recent paper uses.

09:22.465 --> 09:24.965
Uh, but in this other paper, what we do is

09:24.965 --> 09:28.645
what we economists call counterfactual based analysis, where

09:28.645 --> 09:32.645
what we, we essentially do is we say, okay, we hypothesized,

09:32.785 --> 09:35.885
we kind of conjecture that this specific part

09:35.885 --> 09:39.405
of the model corresponds to the preemptive motives.

09:39.405 --> 09:42.405
And so we just kind of shut this off, you know,

09:42.945 --> 09:45.965
assign some arbitrary value to this so

09:45.965 --> 09:47.645
that it's no longer apparent.

09:48.265 --> 09:51.445
Uh, we can effectively shut down this.

09:51.445 --> 09:54.325
So-called incentive behind a preemption,

09:54.505 --> 09:58.045
and then we just kind of simulate the expansion rates for,

09:58.585 --> 10:02.325
these retail chains, with and without this incentive

10:02.385 --> 10:05.365
and just compare the growth rates of the, the firms.

10:05.505 --> 10:08.085
And so in that 2016 study, we showed that

10:08.935 --> 10:12.605
preemptive motives are actually associated with,

10:12.845 --> 10:15.325
a faster growth of, in

10:15.325 --> 10:17.445
that case it was looking at McDonald's like

10:17.445 --> 10:18.525
the fast food chains.

10:18.585 --> 10:22.245
The main difference is, is is actually more in terms

10:22.265 --> 10:23.645
of, implementation.

10:23.825 --> 10:26.605
And so when we think about counterfactual based approaches,

10:27.465 --> 10:29.925
you, you could, you could probably sense,

10:29.955 --> 10:32.925
just when I was talking about the 2016 paper,

10:33.155 --> 10:36.165
there's a little bit of a judgment call in terms of, okay,

10:36.165 --> 10:39.365
you know, we, that's why I used the term conjecture.

10:39.365 --> 10:41.845
You know, I, I said, you know, our, my, my, in

10:41.845 --> 10:44.925
that 2016 study, we kind of, we had this intuition

10:45.135 --> 10:47.685
where we said, okay, you know, this part of the model seems

10:47.745 --> 10:51.125
to correspond to the preemptive motives.

10:51.385 --> 10:55.005
And so we, once you have kind of made that decision,

10:55.395 --> 10:58.355
then it, then it becomes easy, right?

10:58.375 --> 11:01.035
But the, the, the, the challenge is that, you know,

11:01.095 --> 11:04.195
our study did this counterfactual in this way,

11:04.255 --> 11:05.355
but then, you know, there,

11:05.355 --> 11:08.235
there were certainly a few subsequent papers not written

11:08.235 --> 11:09.955
by me, but by other researchers,

11:10.415 --> 11:13.755
and they can use different, you know, assumptions

11:13.755 --> 11:17.955
or different conjectures as to what is their belief,

11:17.955 --> 11:19.715
of the mechanism behind this.

11:19.715 --> 11:20.755
And so they, they will kind

11:20.755 --> 11:22.995
of implement the counteract differently,

11:22.995 --> 11:24.555
even if the model is more

11:24.555 --> 11:26.475
or less very similar, you know,

11:26.755 --> 11:28.475
albeit maybe different empirical settings.

11:28.475 --> 11:30.795
But you know, the model is all these models

11:30.855 --> 11:32.075
are more or less the same.

11:32.815 --> 11:35.115
And so, you know, you'll have different studies

11:35.295 --> 11:36.755
and the, these different studies will

11:36.755 --> 11:37.875
have different implementations.

11:37.905 --> 11:41.645
Some might even, are,

11:41.705 --> 11:44.525
use different equilibrium concepts, you know, in order

11:44.545 --> 11:46.525
to kind of run the counterfactuals.

11:46.525 --> 11:50.245
And at that point, you're, you're, the, the challenge is, is

11:50.245 --> 11:51.925
that it becomes hard to compare

11:52.585 --> 11:54.885
the measurements from one study to another.

11:55.025 --> 11:58.685
So I, I wouldn't say this means that you know, you know,

11:58.775 --> 12:02.445
study A is superior to study B or, or, or,

12:02.465 --> 12:03.725
or vice versa, simply

12:03.725 --> 12:05.765
because of the implementation assumptions.

12:06.325 --> 12:09.525
I think the bigger issue with, with kind

12:09.525 --> 12:12.485
of the counter factor based approaches is that it,

12:12.505 --> 12:15.405
it makes it hard for us to actually say, okay,

12:15.405 --> 12:18.285
there is a study in this particular industry

12:19.025 --> 12:21.045
and they found preemptive motives to be

12:21.605 --> 12:22.645
quantified at this value.

12:22.905 --> 12:25.565
And there's another study that showed a different industry

12:26.405 --> 12:28.685
quantify the preemptive motives at this other value,

12:28.865 --> 12:30.725
double the value or, or whatever.

12:30.945 --> 12:33.885
Uh, but you, you, you can't, it's hard to kind

12:33.885 --> 12:35.045
of compare these numbers.

12:35.145 --> 12:38.685
And so we hope that with this newest study, we hope

12:38.685 --> 12:42.805
that this is as close as one can get to the final word on,

12:42.865 --> 12:44.805
on this topic, topic where, you know,

12:44.805 --> 12:47.285
if there's a little bit of more standardization.

12:47.625 --> 12:51.765
So how we kind of measure these things, and you know, as,

12:51.955 --> 12:54.565
because industrial organization research

12:54.565 --> 12:55.805
tends to be case-based.

12:56.065 --> 12:58.765
You know, you, you, you know, in that paper, you know,

12:59.765 --> 13:01.365
normally we focus on one industry

13:01.625 --> 13:03.765
or even one sub-industry for each paper.

13:04.385 --> 13:07.885
And so I think there could be, there could be some merit

13:07.905 --> 13:09.485
of being able to do some type

13:09.485 --> 13:11.045
of meta-analysis in the future.

13:11.865 --> 13:14.005
You know, as more and more researchers study this,

13:14.025 --> 13:16.525
and there's more and more measurements across different

13:16.525 --> 13:19.125
empirical settings, across different, time periods,

13:19.405 --> 13:21.085
I think is valuable to be able to say, okay,

13:21.085 --> 13:23.725
here's this reasonably standardized measure,

13:23.745 --> 13:26.885
at least under the context of the standard model

13:26.915 --> 13:30.685
that most people use and, you know, applying this measure,

13:30.865 --> 13:32.205
but to different settings.

13:32.205 --> 13:34.965
So I think there could be some value from a kind

13:34.965 --> 13:36.605
of a broader perspective and,

13:36.625 --> 13:39.565
and perhaps in, in a practical sense, you know, if,

13:39.565 --> 13:42.885
if antitrust wants to, you know, come up with measures

13:43.025 --> 13:45.725
to determine whether they need to subpoena a company

13:45.825 --> 13:50.245
or not, measurement of, of, of deterrent modus,

13:50.345 --> 13:53.205
at least from a circumstantial evidence perspective,

13:53.305 --> 13:56.965
has always been a huge challenge for, antitrust.

13:56.965 --> 13:59.485
And so, not surprisingly, if you look in the, in,

13:59.505 --> 14:02.125
in the past, antitrust cases, there's very few

14:03.155 --> 14:06.565
that have, you know, been successful at, you know,

14:07.245 --> 14:10.045
prosecuting firms for, entry deterrents.

14:10.045 --> 14:13.725
It's, it's a, it's a, it's a difficult thing to, to show,

14:13.945 --> 14:16.325
like of, of course if you can get a subpoena

14:16.325 --> 14:17.645
and you can wiretap companies,

14:17.645 --> 14:19.285
then you can a lot more details,

14:19.345 --> 14:22.245
and that's probably where you can actually get a conviction.

14:22.305 --> 14:24.045
But in general, it's hard to build up

14:24.045 --> 14:25.805
that circumstantial evidence, right?

14:25.825 --> 14:29.205
And so I think the immediate implication is, you know,

14:29.205 --> 14:32.045
thinking about our policy, I think another

14:32.925 --> 14:34.845
interesting implication, and,

14:34.845 --> 14:37.525
and this is something, I'm exploring with the, the,

14:37.545 --> 14:40.485
the same co-author of this study we're talking about,

14:41.305 --> 14:45.605
is an implication on how we think about firm value, right?

14:45.705 --> 14:49.725
And so, if you look at corporate finance, a branch

14:49.725 --> 14:54.125
of corporate finance interested in looking at how

14:54.185 --> 14:57.045
to predict whether firms fail or not, right?

14:57.045 --> 14:59.965
And so you, you, you see a lot of retailers now are not,

15:00.275 --> 15:01.845
some retailers don't do very well.

15:02.675 --> 15:05.405
They close stores on mass, you know, and,

15:05.405 --> 15:08.645
and so if, if there's a way to develop some sort

15:08.645 --> 15:13.365
of prediction as to, whether their kind

15:13.365 --> 15:14.925
of entry strategies

15:14.945 --> 15:17.525
or expansion strategies are considered to be risky,

15:17.945 --> 15:20.365
if there's a way to kind of predict this,

15:20.375 --> 15:22.285
there could be some value for,

15:22.545 --> 15:25.005
for people in corporate finance, especially those

15:25.075 --> 15:28.445
that are trying to come up with kind of valuations of firms

15:28.475 --> 15:30.325
that they want to invest in.

15:30.505 --> 15:33.325
And so in this kind of follow up work, that li

15:33.425 --> 15:36.445
and I are kinda working on, we're trying

15:36.445 --> 15:40.365
to find a connection between this preemptive

15:41.035 --> 15:44.325
measure, you know, how incentivized they are to

15:45.385 --> 15:47.805
expand aggressively for the purpose

15:47.805 --> 15:49.125
of keeping competitors away.

15:49.625 --> 15:52.325
And, and in using this reasonably standardized measure,

15:52.345 --> 15:54.525
we want to see is there a connection between

15:55.055 --> 15:56.765
these preemptive motives

15:57.225 --> 16:00.765
and ex post failure of firms, right?

16:00.765 --> 16:02.605
So are they doing this?

16:03.505 --> 16:06.885
Um, so one of the assumptions we, we will make is,

16:06.945 --> 16:10.685
of course, firms are acting rationally and optimally,

16:11.025 --> 16:15.125
but as a, let's say you're an outside investor,

16:15.185 --> 16:18.525
you might still care, you know, to see, at least

16:18.525 --> 16:20.725
for risk analysis, you might want to see

16:21.065 --> 16:23.685
and assess, you know, whether, okay, you know,

16:24.145 --> 16:27.605
we can assume firms are all kinda being as strategic

16:27.705 --> 16:29.285
and as optimal as possible,

16:29.425 --> 16:32.645
but perhaps there's even a way to rank, you know,

16:32.645 --> 16:35.565
equally optimal firms in terms of their strategies

16:35.865 --> 16:38.605
and rank them based on their risk profiles as to, okay,

16:38.605 --> 16:41.325
is this entry strategy going to lead

16:41.345 --> 16:43.405
to a new bed bath and beyond?

16:43.405 --> 16:45.285
Which we know, you know, you know, a lot

16:45.285 --> 16:47.285
of these mall brands are just, you know, closing.

16:47.865 --> 16:50.765
And, and I think it, from a substative perspective,

16:50.985 --> 16:52.125
it could also be interesting

16:52.125 --> 16:55.805
because, most of us, especially in the marketing side

16:55.805 --> 16:59.045
of things, has often thought of digital commerce has kind

16:59.045 --> 17:01.285
of the thing that kills off a lot of brick

17:01.285 --> 17:02.965
and mortar retail chains.

17:03.385 --> 17:06.605
Uh, but you know, there's a smaller branch

17:06.985 --> 17:08.805
or a smaller number of people,

17:08.865 --> 17:11.685
or in particular very few people, perhaps just a, a handful

17:11.745 --> 17:15.445
of us who think that actually a lot of the, the,

17:15.445 --> 17:17.725
the issues is more about over expansion,

17:17.775 --> 17:21.765
especially if this expansion is fueled by debt, right?

17:21.765 --> 17:25.525
Which you've seen a lot of these retailers, of,

17:26.425 --> 17:29.045
you know, suffer from, expos, you know, when it,

17:29.105 --> 17:31.165
it could be okay, you know, when the times are good,

17:31.185 --> 17:34.525
but when the times are bad, it, it really amplifies,

17:34.525 --> 17:36.485
the pain for, for some of these retailers.

17:36.545 --> 17:40.645
And so I think down the road, I, I, I hope to be able

17:40.645 --> 17:42.965
to kind of integrate some of these insights more closely.

17:43.475 --> 17:45.005
Also to, you know, a lot

17:45.005 --> 17:47.165
of the discussions we have about retail

17:47.255 --> 17:48.685
apocalypse, although I think

17:48.705 --> 17:50.975
The discussions haven't really been

17:50.975 --> 17:53.535
as active recently about retail apocalypse,

17:53.535 --> 17:54.815
but you know, it's always there, you know,

17:54.815 --> 17:56.735
certainly if you follow retail

17:57.435 --> 18:01.535
expert gurus on the LinkedIn kind of crowds, I.

00:00 - Probably many of us have encountered, and in fact, I've seen like comedians kind of quip about this, in their acts where, you know, you drive down along a street or you walk, you know, down a, in downtown. And you know, where, what you see is you see a Starbucks

00:17 - across the street from another Starbucks, I think in, a subway station in Tokyo. You know, there are four, seven elevens in the same station. And, you know, coming from an economics background, this type of observation, what seemed a little, counterintuitive to me, because, you know, there's this notion of cannibalization where, you know, if you set up a bunch of shops and close proximity, they're gonna eat each other's kind of profits, if you will.

00:47 - And so that kind of led to kinda a, a rabbit hole, if you will, of me kind of spending the past decade or so, kind of looking several years looking into this phenomena across various papers. And, and so the phenomena I was particularly interested in is whether there's actually an economic benefit of them kind of saturating a market. And in fact, there could be, if, there is a benefit associated with deterrence where, you know, you take up that spot,

01:19 - if McDonald's takes up a spot, that means Burger King can't, can't do that. And in fact, you know, if you read the autobiography of I think Walmart, the, the book by one of the Waltons, you know, they, they, they even allude to the fact that they will kind of, they, they want to, they set up a Walmart in close proximity of another Walmart just to keep away their competitor at the time, which was Kmart. even, even though they knew this would,

01:47 - potentially eat away at their existing store sales. And so this, this kind of behavior, the strategic behavior, this economic benefit is, is, has been particularly interesting, to me. And I think that the, the thing that makes it intellectually, satisfying to study is, is actually how challenging it is to, quantify this. Because if you think about things like deterrence motive, it's hard to quantify this. There's no kind of observable data,

02:20 - directly observable data that indicates, you know, that they are doing this because they want to keep competitors away. Unless you, you have a subpoena on these companies and you actually can wiretap their, internal conversations, which, you know, typically people don't have in this paper that, we're talking about. we use a fairly standard, model of how firms kind of, open, expand.

02:45 - Basically it's, you could think of it like a firm investment model where there, there's a, some costs associated with, opening a store. There could be a, there could be costs associated with closing a store. And so what this model does is that it, it captures the, what it may me, what the model aims to do is to internalize, the incentives these firms have for these strategic investment decisions. And so there's essentially two important ingredients,

03:16 - if you will, or dimensions that this, these models in general need to capture. If one wants to study, preemption, the first is that, there has to be a strategic component, right? And so these firms need to be making decisions knowing that their decisions have implications on competitors. And the same thing, you know, they also expect competitors to be equally or reasonably strategic as well, right? And so it's, you, you kind of study in, in what we call kind of a game type of setting.

03:45 - And the second kind of dimension of, of that importance is, is this notion of forward looking behavior or your investment decision today is not only, aimed at optimizing your benefit today, but you also are trying to look ahead and see, okay, you know, if we open a store here, the, this is the implication it'll have on a market structure, not just today, but maybe tomorrow and date, you know, year after, and so on and so forth. And so, the model essentially needs

04:18 - to have these dimensions. And once you kind of combine these dimensions together, then it, it, it, it kind of offers us this kind of canonical, dynamic oligo model. And so the earliest, what among the earliest versions of these models were kind created in the eighties, I think the very, kind of the workhorse model is the Ericsson peku, dynamic oligopoly model, which, many, many of, you know, a lot of the modern research today still,

04:48 - continues to use. And so, yeah, so we, we were, we're essentially kind of, building off of, a a lot of the great kind of, at least analytical infrastructure that has already been, developed by, the past scholars. Our contributions are multi multifold. So the first is, you know, a little more theoretical analytical, where a lot of our results are actually kind of using,

05:16 - you know, pen and paper. not, not literally, you know, most of us now use iPads and and whatnot. But you know, we, we, we kind of build out the results on paper, if you will, so that we have a fairly concrete analytical notion about what is the value of preemption, right? And so we essentially need a moral framework in order to kind of decompose the various incentives firms have

05:45 - for expanding into the new markets. And by doing so, by kinda writing it out explicitly, which is not trivial, because, you know, you have all these different dimensions of strategy and forward looking behavior. And so when you're trying to decompose things like bill in equations, it becomes a fairly thorny task. And which is why we think there's a, a theoretical contribution to this paper where, you know, once you kind of, do this accounting, if you will,

06:13 - of all these different incentives, you can then say, aha, here is the specific benefit the firms have from expanding and as a result of this expansion, the added profits they get in the future because their competitors chose not to set up shop because of them, right? And so you, you could, you could tell just this, I, I felt like this is the, the, the, the simplest way of kind of characterizing this in simple words. And so you can imagine to translate this into kind

06:45 - of the analytical components, it's a, it's a little bit challenging. But the thing is, once you have, once we, we, we developed this framework, and once we had done this, decomposition on paper, it then makes the kind of empirical exercise a lot simpler because we, you know, you basically have a formula, if you will, for how to calculate this. it's, it's kind of hidden this intrinsic motivation for this intrinsic value of preempting the markets,

07:16 - which you can then kind of put to the data, right? So once you, once you have this kind of quantity written out analytically, it's really just a matter of estimating and fitting a dynamic oligo model with the data, which is the, in our empirical application, it's the coffee chain data. And once you have the model fitted to the data, you can then just plug in a bunch of values and just calculate, okay, you know, like 20, 35% of the motivation for Starbucks is driven by preemptive,

07:50 - considerations. And so we, as, I guess as the authors of the paper, we, our hope was we did the hard work so that future researchers can just plug it and play and kind of use a lot of these kind analytical insights in, you know, subsequent work. So that, it's, it's a little more concrete. Starbucks, of course, is perhaps the retailer that we, that the coffee chain that we notice, has the strongest, I guess motivation for, preempting the market.

08:20 - And so the, the beauty about our measures that we can actually, it's kind of a relative term, so we can actually say, okay, you know, X percent of their motivation is driven by preemptive considerations, and then 95% is just from all other kind of incentives. And so we, we actually find a bit of a range, you know, I, I think it, it could range from like around 5% to 30%, something in that ballpark. And so, you know, it, it, it, it's, it's, I believe it's among the first papers out there that actually provides kind

08:54 - of a percentage, measurement. In my earlier work. You know, the way we, I would've measured this would've been, like in, in a paper back in 2016 in quantitative economics with a different co-author. you know, we had, in order to quantify this incentive, we did it in a more indirect way where we said, okay, here's a model. And it's a very similar model to what this,

09:19 - this current paper, this more recent paper uses. but in this other paper, what we do is what we economists call counterfactual based analysis, where what we, we essentially do is we say, okay, we hypothesized, we kind of conjecture that this specific part of the model corresponds to the preemptive motives. And so we just kind of shut this off, you know, assign some arbitrary value to this so that it's no longer apparent. we can effectively shut down this.

09:51 - So-called incentive behind a preemption, and then we just kind of simulate the expansion rates for, these retail chains, with and without this incentive and just compare the growth rates of the, the firms. And so in that 2016 study, we showed that preemptive motives are actually associated with, a faster growth of, in that case it was looking at McDonald's like the fast food chains. The main difference is, is is actually more in terms

10:22 - of, implementation. And so when we think about counterfactual based approaches, you, you could, you could probably sense, just when I was talking about the 2016 paper, there's a little bit of a judgment call in terms of, okay, you know, we, that's why I used the term conjecture. You know, I, I said, you know, our, my, my, in that 2016 study, we kind of, we had this intuition where we said, okay, you know, this part of the model seems to correspond to the preemptive motives.

10:51 - And so we, once you have kind of made that decision, then it, then it becomes easy, right? But the, the, the, the challenge is that, you know, our study did this counterfactual in this way, but then, you know, there, there were certainly a few subsequent papers not written by me, but by other researchers, and they can use different, you know, assumptions or different conjectures as to what is their belief, of the mechanism behind this.

11:19 - And so they, they will kind of implement the counteract differently, even if the model is more or less very similar, you know, albeit maybe different empirical settings. But you know, the model is all these models are more or less the same. And so, you know, you'll have different studies and the, these different studies will have different implementations.

11:37 - Some might even, are, use different equilibrium concepts, you know, in order to kind of run the counterfactuals. And at that point, you're, you're, the, the challenge is, is that it becomes hard to compare the measurements from one study to another. So I, I wouldn't say this means that you know, you know, study A is superior to study B or, or, or, or vice versa, simply because of the implementation assumptions.

12:06 - I think the bigger issue with, with kind of the counter factor based approaches is that it, it makes it hard for us to actually say, okay, there is a study in this particular industry and they found preemptive motives to be quantified at this value. And there's another study that showed a different industry quantify the preemptive motives at this other value, double the value or, or whatever. but you, you, you can't, it's hard to kind

12:33 - of compare these numbers. And so we hope that with this newest study, we hope that this is as close as one can get to the final word on, on this topic, topic where, you know, if there's a little bit of more standardization. So how we kind of measure these things, and you know, as, because industrial organization research tends to be case-based. You know, you, you, you know, in that paper, you know, normally we focus on one industry

13:01 - or even one sub-industry for each paper. And so I think there could be, there could be some merit of being able to do some type of meta-analysis in the future. You know, as more and more researchers study this, and there's more and more measurements across different empirical settings, across different, time periods, I think is valuable to be able to say, okay, here's this reasonably standardized measure, at least under the context of the standard model

13:26 - that most people use and, you know, applying this measure, but to different settings. So I think there could be some value from a kind of a broader perspective and, and perhaps in, in a practical sense, you know, if, if antitrust wants to, you know, come up with measures to determine whether they need to subpoena a company or not, measurement of, of, of deterrent modus, at least from a circumstantial evidence perspective, has always been a huge challenge for, antitrust.

13:56 - And so, not surprisingly, if you look in the, in, in the past, antitrust cases, there's very few that have, you know, been successful at, you know, prosecuting firms for, entry deterrents. It's, it's a, it's a, it's a difficult thing to, to show, like of, of course if you can get a subpoena and you can wiretap companies, then you can a lot more details, and that's probably where you can actually get a conviction. But in general, it's hard to build up

14:24 - that circumstantial evidence, right? And so I think the immediate implication is, you know, thinking about our policy, I think another interesting implication, and, and this is something, I'm exploring with the, the, the same co-author of this study we're talking about, is an implication on how we think about firm value, right? And so, if you look at corporate finance, a branch of corporate finance interested in looking at how to predict whether firms fail or not, right?

14:57 - And so you, you, you see a lot of retailers now are not, some retailers don't do very well. They close stores on mass, you know, and, and so if, if there's a way to develop some sort of prediction as to, whether their kind of entry strategies or expansion strategies are considered to be risky, if there's a way to kind of predict this, there could be some value for, for people in corporate finance, especially those

15:25 - that are trying to come up with kind of valuations of firms that they want to invest in. And so in this kind of follow up work, that li and I are kinda working on, we're trying to find a connection between this preemptive measure, you know, how incentivized they are to expand aggressively for the purpose of keeping competitors away. And, and in using this reasonably standardized measure, we want to see is there a connection between

15:55 - these preemptive motives and ex post failure of firms, right? So are they doing this? so one of the assumptions we, we will make is, of course, firms are acting rationally and optimally, but as a, let's say you're an outside investor, you might still care, you know, to see, at least for risk analysis, you might want to see and assess, you know, whether, okay, you know, we can assume firms are all kinda being as strategic

16:27 - and as optimal as possible, but perhaps there's even a way to rank, you know, equally optimal firms in terms of their strategies and rank them based on their risk profiles as to, okay, is this entry strategy going to lead to a new bed bath and beyond? Which we know, you know, you know, a lot of these mall brands are just, you know, closing. And, and I think it, from a substative perspective, it could also be interesting

16:52 - because, most of us, especially in the marketing side of things, has often thought of digital commerce has kind of the thing that kills off a lot of brick and mortar retail chains. but you know, there's a smaller branch or a smaller number of people, or in particular very few people, perhaps just a, a handful of us who think that actually a lot of the, the, the issues is more about over expansion, especially if this expansion is fueled by debt, right?

17:21 - Which you've seen a lot of these retailers, of, you know, suffer from, expos, you know, when it, it could be okay, you know, when the times are good, but when the times are bad, it, it really amplifies, the pain for, for some of these retailers. And so I think down the road, I, I, I hope to be able to kind of integrate some of these insights more closely. Also to, you know, a lot of the discussions we have about retail apocalypse, although I think

17:48 - The discussions haven't really been as active recently about retail apocalypse, but you know, it's always there, you know, certainly if you follow retail expert gurus on the LinkedIn kind of crowds, I.

Faculti AI can make mistakes