E-Sourcing: Impact of Non-Price Attributes Strategic...

Uploaded by GoogleTechTalks on 08.10.2007

>> So welcome back for you who stayed. This is the second part in our two-part presentation
by Professor D.J. Wu from Georgia Tech. This presentation is on E-Sourcing Impact of Non-price
Attributes Strategic Bidding and Learning. So without further ado, D.J. Wu.
>> WU: Thank you. Okay. Yeah, this will be more fun. Because the last time was really--the
talk wasn't really fine, it's so hard, you know, its math. But this is a little [INDISTINCT]
especially I see my friend, you know, where we got our data sources, he's here. And--yeah,
but I'm also a little upset here because, you know, we have this real expert here. So,
the talk is E-sourcing and we're trying to start the impact of non-price attributes,
okay, like Quality Delivery, okay, relationships and the strategic bidding behavior of the
suppliers, and also the learning behavior of the suppliers. This is really the work
of Vivian Zhong, she is my PhD student. This is really her dissertation. She has been working
of her theory for three years and also the clients of procuring. Yeah, yeah, so, let
me thank procure first. Yeah. So, therefore Vivian is on the job market. So, if Google
is recruiting, you know, check her out. You know, she's extremely smart. Okay. Yeah, she
did almost all the heavy lifting here of the work. So, what's the motivation here? It's--so,
we know a lot of B2C auctions, you know, so actually early on Vivian wrote a paper with
me and other professors, Professor Sinan and Professor Eunjin. It's on consumer auctions.
Okay, the bidding behavior of consumer auctions believe it or not you guys really provide
a lot of research for us to do, okay. Folks like Google like E-bay all those and say,
new firms and a lot of people there actually download data from E-bay which is public.
Okay. So, then you can download, you can start the bidding behavior for consumers. Okay.
But our question is how about firms? So, the question is does B2B auctions different than
B2C auctions. So, we're trying to ask this question, "What's the difference between B2B
versus B2C?" Since we--we thought we know a lot about B2C but we don't know, okay, anything
about B2B until actually Vivian's dissertation. So, this is for the first time we find out,
you know, what's going on. And, oh, what's unique about B2B auctions, well, it turns
out they are--in the literature they have been talking about three things. Okay. The
non-price attribute is going to be different. Okay. Whether you are my income and supplier
make a difference. And--but in E-bay auctions you--anybody can bid, right? A dog can bid,
right? You don't need a name, like an actual name. But for this procurement auctions you
have to be a pre-qualified supplier and definitely you're going to tell the buyer who are you,
right? So, this is different. Second things is, hey, winning the auctions does not mean
winning the deal. Okay. You can bid really low. You thought you're going to win, there's
no way. This procurement auction enterprise is going down, okay. And if you're trying
to bid in, say for a resort like Hilton Head, you know, they're right there in South Carolina
and of course, you know, the higher the price you go, you know, you're going to get it,
right? So, this is different, okay, than consumer auctions. Finally this kind of auction it's
repeated. Because, you know, yeah, I don't know. See, you know, see, in Georgia Tech
we have to buy the pencils, you know, I say that every year, right? So, we have to run
this auctions, you know, over and over again. Okay. Again let me use it again my favorite
case, the Dell Case example is, you know, Dell need to, you know, buy memory chips,
you know, every year, okay, once [INDISTINCT]. So, these auctions they are repeated which
is different to the B2C auctions, right? I just need one iPod, you know. Yeah, I may
even bid for a used car but I just need one used car. As a professor that's all I can
afford, right? So, those are different things, okay, in theory. Okay. So, we're on--but nobody
has actually opened the actual data to detect what's going on. So, that's why, you know,
so we really appreciate the relationship we have and made this company now--everybody
knows it as procure, okay. And actually other firms actually give us the data, the B2B bidding
data. And also dynamics is important. Okay. That has been shown again to B2C auctions,
you know, say how do you bid? Okay. Yeah, how do you bid for iPod in E-bay? Okay. Different
person have different strategies. So, I think smart people at Google you may use what is
called a Snide Strategy. You don't bid, you watch. Okay. And then at the last minute you're
going to bid, all right? See, in all, yeah, there are also other people, you know, you
could call on naīve bidders. You know, they just lack the participated and can't even
bidding or simply using the proxy provided by E-bay, right? So, we know. Okay. We know
something, you know. Actually that's the earlier paper by Vivian and also other researchers,
you know, see about the bidding behavior. Okay. How do people bid. But we don't know--how
do firms bid. So, these are actually deals, right? The contracts, right? Those are billion
dollar contracts, you know, how do firms bid. So, those are the second thing, you know,
that motivates us is different. The last one is, you know, right here at Stanford [INDISTINCT]
for Google Professor Paul Milgram, yeah, the very famous and world class in economics.
Yeah, he's auction theorist. Okay. He's an authority of auction theories. But he's calling
for the new theory about facts. A procure really shocked us, you know, we took a look
at all the auctions going on with procure, we're trying to fit with all the literature--sorry,
no fit. That's it. See, yeah, none of this textbook auction, you know, say models. In
a fix what they're using in business practice. So, that's why--so, we want to understand
what's going on in business practice. Okay. Find other facts and see if we can deal with
some theories. So, those are the motivations. Okay. Any questions? Okay. So, after Vivian's
dissertation actually--yeah, this is only two chapters of her dissertation, two papers.
Okay. The first paper, okay, I'm going to talk a little bit more today is trying to
find out in B2B auctions, you know, what are the bidding dynamics, you know, how do they
bid? Okay. Also, what's the impact of a known price attributes. See, if I know I am an incumbent,
okay, supplier. Okay. Do I bid differently? Versus, I'm a new guy, okay, I'm a new entrant.
Okay. So, that's the question if, you know, whether supplier type, okay, well play a role,
okay, in terms of their bidding. Okay. The second point is how do--yeah, that's the second
point actually, known price attribute like supply type, okay, affects bidding behavior
and also the auction outcome. Remember, winning the auction. That's not usually been winning
the deal, the contract, right? So, but, what's the relationship, right? Between, okay, A,
my type, okay, my supply type. B is my bidding strategy. Okay. And--yeah, and also the enterprise,
what's going on. And the second question is, what--when you have repeated auctions, you
know, see, do suppliers after they learn? Right, because I'm bidding, right, again and
again. Say, some auctions market, you know, we do it daily like electricity, you know,
bidding, right, daily. Okay. Now, procure is busy, you know, there's some firms they're
running this events more and more infrequently, right? So, then see--so, this is the same
thing, right? So, I'm bidding trying to get this deal from Dell to sell Dell memory chips,
you know. Do I learn something, right, overtime? Okay. So, those are what we're trying to do.
Okay. This is the literature of what I'm saying. Okay. This is what we're doing basically.
So, yeah, I think it is--I'll meet the first--work that actually show you the evidence, you know.
So, we have lots of theory but those theory does not fit precisely, okay, what the industry
people they are doing. Okay. But the first thing we're going to do, we have to find out
what's doing on in industry. So, this is the first--I believe this first work that shows
the empirical evidence of what's going on. So, this is actually maybe--you know him this
is actually my friend Norbert Ore from Georgia Pacific. Yeah, this is the most recent lecture
he actually gave in my class at Georgia Tech. He was also a frequent speaker for procure
empower. Yeah, he was there when you ran in that bucket. And this year actually--and he
did, you know, video show. Okay. So he's really authority about strategic sourcing and it
turns out they have different stages, so reverse auction is not running the auction only. So
there are a lot of things before the auction and after the auction. So this is very important
to understand e-sourcing. So you have this information gathering stages. You have this
implantation stages. You also have to manage. Okay. Yeah. I believe procurement software
have all these, you know, modules, functions, you know, to satisfy the needs of different
stages. Okay. So the whole point is I want to borrow--his slides is there as a process.
Okay. It's not just running the auction and then this is the model. Okay. So this is--we
tract our major buyer in the hi-tech company. Okay. It's a big one, you know, major buyer.
And, yeah, maybe you look around and he's right here. But I'm not--I'm not going to
tell you the name. Okay. You see. It turns out--you know, in these companies e-sourcing
business process the first to pre-qualification. Okay. They qualify suppliers. Okay. They invite
suppliers. So this business is invitation only. Okay. They send all these of X, right,
to request for information or request for proposals, okay, then do the auction. Okay.
Then after the auction they also have the stage called the contract warning. And this
particular company is not automated. It's not automated. It's actually it's a manual
selection. Okay. Who is going to win? Okay. When they select a winner actually they take
in consideration of this non-price attributes. Okay. What's unique about this research is
we have--we are fortunate to have two different data sets. Okay. The first date set is the
auction data set. Okay. We also have the second data set. It's who actually got the deal.
Okay, the winner data set. And we also know the supplier type information. Okay. Whether
you are incumbent, okay, long-term supplier or you are a new-entry guys. Okay. So it's
very unique. Yeah. Let me you a story with one picture. I know where we found it, okay.
The winner is the incumbent. Yeah. This is very different than what the media has been
saying. The previous work, okay, because people worried. Okay. Reverse auction is going to
piss off my long-term suppliers. Okay. It's not going to be sustainable. Okay, there are
a lot of cry. It's not the case. Okay. You know they also made the pricing firms and
like a--yeah, like for procure, you know, they wont be able to sustain them all because
they don't have--you know, that's not true. Okay. Now, so data issue is not actually now
more and more firms. Okay. And they are using reverse auction using e-sourcing. Okay. This
is actually the evidence. So we have to show this paper which is Mark Camarillo with many
people. They gave us actual comments. You know this paper--yeah, commercialize was nominated
for the best paper, for the ICS conference in--yeah, this is Vivian's paper, I was you
know. And, yeah, we got a lot of comments, you know, from those really industry experts
and--like Ricky Hartch, Scientific Atlanta, Mark Camarillo and Norbert Ore. Let me tell
you what we found. Okay. So in front here these are actual contract, actual contract
for three years, for major buyer in a hi-tech industry. And the Y axis's is--if I'm going
to award the deal, okay, to these new guys. So let's say in China, the savings. Okay.
Yes, suppose I'm going to award the deal, okay, to the lowest of bidder. So those are
the potentials savings. Okay. And the X axis's, you know--you know, if you travel east right
there is the potential or--is the actual savings if you're going to award this deal to your
long-term supplier. So are you with me? So you know, so this picture we're not just trying
to say, okay, supposed the deal if I'm going to give okay to my--let's say supposed there's
a point right over here, okay, in this white triangle area. Okay. That means, so the savings
I'm going to get, okay, compare it to the new guy versus my long-term supplier the same
is bigger, okay, if you get the deal. I'll award deal to my incumbent, okay, my long-term
supplier. Okay. And in that shaded area right there is if you're going to award the deal,
okay, to the new guy, okay, and then your savings will be huge, bigger. Okay. Compare
if your want to give the deal to the same supplier, your long-term supplier or the incumbent.
Okay. So this is 45 degree line. So that's--that degree line separates, right. So those shaded
area is suppose you're only going for the lowest of prices. Okay. Then in that shaded
area you should award the deal to this new entry guy. Okay. And if that in this area--in
this area, okay, you should award the deal to this long-term supplier. Okay. Guess what
happened? This is the actual data, okay, actual contracts. So what showing right there--so
this so--the circle okay, white dot is what actually happened. It's this new guy or we
called non-preferred supplier actually was awarded the deal. And that square is the preferred
supplier I've got the deal. Okay. Take a look at this picture. You know, it's almost all
the dots are squares. Okay. That means, okay, is the incumbent that got most of the contract.
Okay. And take a look right there in this shaded area, okay. That's a strong evidence
that shows the influence of non-attributes, right? You see why you got the deal. You offer
to look a lot of deals right over there, right? And why offer those deals, you know, to this--to
your incumbent. You know, you're not saving a lot. You could save a lot more if you give
to this to new-entry but they did not. Okay. So that's evidence of the non-attributes,
the influence of non-attributes. And there also roughly--you know, you see, I don't know.
You see you can see that. Yes? >> How many of it?
>> WU: I don't know, maybe 30% at the time. You only see...
>> Of procurement? >> WU: ...and interesting behavior. It's--actually
you see that most of the time suppliers--the long-term suppliers, the incumbent suppliers
they bid the same, almost the same, right? It's all around that 45 degree line. It's
the actual matches. They matches these new-entries bid. Do you see that? And there also real
cases, actually this long-term supplier they bid lower, even lower. So they are willing
to give the buyer more. You see that? You'll see all those dots right there. So these areas
that means the suppliers is bidding lower, okay, the long-term supplier, the incumbent
is bidding lower than new-entry guys. So, you know, they actually willing to give them
more. Well, a possible explanation is, you know, they say they have an advantage. You
know, see--you now see--yeah. They have, you know, basically they can afford, okay, to
give more. So there's also evidence is this long-term supplier they are willing to maintain
the long-term relationship, right? But actually give you a better deal. You--so either I match
the deal of this new-entrant guy or actually I--or even lower. Yeah, and basically give
you even more. And in exchange you also find evidences--evidence. It's in some real cases,
okay, the buyer is actually too willing to pay more. So normally this previous auction
this is the price I paid by last year than I need to--you know, its cost reduction right
over the years. That case is actually I'm paying more than I paid last year? Okay. In
those real cases all those years where awarded exclusively to incumbent. You see that? You
see the evidence. You know they are trying to maintain of this relationship. Okay. And
take consideration of non-price attribute. So this is basically my talk and now you can
go. Okay. So this is the actual reverse auction. I--and, you know I won't have time to go over.
This is the actual auction we started. Okay. This is only one auction platform. Procurement
has all kind of choices, you know. I want to say the buyer you can actually--you set
up your own auction. See different auction rules you can choose. You know its rally a
platform. But for this particular and let's say a hi-tech buyer, okay, they use a fixed
one, a fixed, okay, platform. That's what we did for this paper. Okay. Once the information
is being revealed this is important. Okay. What information they're going to reveal.
Okay. There are firms who messed up. Okay. How do they mess up? They actually reveal
to the actual bid price of the suppliers and that's the endures. Then I'm not going to--I'm
not going to participate. Well, you can imagine if I'm going to participate my final bid is
really very close to my actual cost, right? You can imagine, right? It's, you know, close.
Okay. Yeah. If you reveal my actual bid, okay, that means all my rivals they actual--you
know, they have a guess. They should know my price in range, right? So therefore, you
know, it's hurting me. But this company is something very smart. You see, they did not
reveal the actual bid prices. What they reveal is the rank. Okay. And say where are you right
now? Okay. See. You own bids, of course. You can't see it, right? Okay. And also the kind
of lowest bid, you know. And also your rank, where you are, okay, in terms of your ranking,
okay, with the other rival suppliers. And you can see this information as long you bid.
Okay. You have to bid. You cannot just go there to watch, all right. So, you have to
bid. Okay. So, we have these two data sets. This is unique. And the first is the Auction
Data Set, okay. The other one is the Cost of Savings Data Set. And we actually know
the supply type. We know the actual price and we know the price before and after the
auctions. So, we can actually count the savings so, we actually have a model, okay. This is
what makes Georgia Tech unique, right? We always try to build new theories. And as I
butt in there we started with business practice. So, we actually--we're trying to make friends
with those--yeah, those industry leaders like a procure or Google. We love--but we also
wanted to make a contribution. So, once the new theory, we can build. Okay. The new theory
we want is--the basic idea here is the supplier type and number of additional preferred items.
Okay. This particular supply can provide and the number of previous auction experiences.
Okay. Those three things, we used this as a proxity--proxy for non-price attributes.
So, non-price attributes, you know, one is the type whether you are my preferred supplier
or not. Also, you know, the experiences, you know, they participated also--yeah. The supplier
may not just supply one item. So, many--they can supply in this particular--all items,
you know. So, you are the hi-tech buyer, what do you need? Memory chips? But also I can
supply other stuff. I can supply, you know, see--a bunch of items. Okay. So, therefore--but
within your portfolio of items, your products, you know, and also, you know, how many prefer
you have. Okay. That's a measure, you know. You have--you need all preferred items, you
know, for the buyer. So, then--yeah, those are important. Okay. Yeah. That you can imagine
that's a long--is a measure of the relationship, right. You are dealing with these guys more
frequently. Okay. Well, imagine that way is having an impact. Okay. Those are--that's
to the bidding strategies. Okay. And also to the out--auction outcomes. How do we measure
auction outcomes is to find a bids and also the cost of savings and the cost award in
probability. Okay. And how likely you were actually get the deal. So, are you with me?
Okay. So, we--so, again, we have this IRE supply type. E--I stands for whether you are
incumbent or not. R stands for relationship. Okay. E stands for experience, auction experiences.
Oh, boy, let me see. The first thing we do here is, you know, we have this auction data.
We're going to do a classic cluster analysis. We want to know how do firms bid. Remember,
nobody knows. Okay. Of course, you're going talk with from my friend, Mark. He can tell
you how he's going to bid, but that's just a one firm, right? And let's see. I want to
know like this big data set. Okay. And how do firms do bid. Okay. So, you can do cluster
analysis. You can do statistic analysis. And how do I measure? So, in--let's focus on the
item level that thinking about memory chips in a [INDISTINCT] okay. And we measure the
time, okay, this firm get in. Okay. Now, we measure the firm stop bidding. Okay. They
cannot really quit, you know, but they actually they, you know, they stop bidding. That's
the time we measure. Now, how frequently this firm bid? Okay. And also--yeah. When they
actually--they change their bids, okay. What's the magnitude? Okay. They reduce their bids.
So, are you with me? Those are the four measures. Okay. Okay. There are theories but I'm not
going to explain. But anyway--so, we reveal those four measures. Take a look what's going
on. Yes. There are five clusters. Okay. You know, it's a three-year data set. Okay. And
these are actually how firms bid for the first time. Okay. So, we believe our research team
is the first who discover, you know, how do firms bid for industrial contracts. Okay.
Take a look of those pink one. Okay. Those we call early evaluators. In this firm that
get in, okay, they bid high then they bid a little bit than no--not bidding. The reason
is, you want--you don't even realize that suppliers, they don't even know the market
condition. So, these guys, they come in. They bid. They get a sense up on the market. Okay.
But they are not that competitive, right? They cannot bid anymore because, you know,
you bid more than making a loss, right? But that's, you know, that's the best I can give
to you, Mark, you know, right there. You know, I stop bidding. Okay. And take a look of--so,
there are also mid-evaluators. There are guys that actually get in the middle of the auction.
This is roughly doing our auctions. Okay. Then do a few bids then stop bidding. Okay.
These are mid-evaluators. Okay. You also have this that we call it opportunist. You know,
these firms, they watch. Until the end of the auction, they get in. Okay. They bid,
okay. Sort of like [INDISTINCT] you know, in the e-Bay are auctions. Okay. And we also
have guys called the participators. Those firms, they just love to participate, you
know. They all--they already--they are human being, you know. They are going to play with
you until the end of the auction. Okay. Why this is interesting? First of all, they are
patterns. And all the stuff I just told you, it's new. Because in previous auction theory,
you know, you should not have this kind of behavior. Okay. Yeah. If you follow previous
auction theory and open the textbook, you know, firm should all bids, know strategically.
Okay. Or what we called straightforward bidding. Okay. And basically, you know, the--what they're
going to produce, yeah, it's--yeah. That's why this is new. So, in the--for the first
time, we uncover--see, they, of course, they're all firms right there. So, this master strategic
guy is--they are keeping bidding, you know. It's also called, see, a ratchet strategy.
So, you know, when you, you know, a roller skater, you know, ratcheting. Okay, or called
the pedestrian strategy, you know, you work like this, right. Yeah. So, basically, those
firms, they came in bidding, okay. And so, if those the blue lines, so that's the theory
predicts, you know. Actually, you shouldn't bid like that. And that is--yeah, it's optimal.
But if you open--of course, the theory has assumptions, has the model setup. But if you
open this like a real--yeah, real, you know, reality--all real with data set, you discover
the other clusters. And these are the major strategies. So, that's why this--the first
time we show the clusters, okay and the firms, they actually--they bids straight forwardly
but they also bids strategically. So, those strategic data is important. This is a new.
Okay. We need to take into consideration. And--yeah. This--yeah. We have similar behavior,
you know, auction level. So, I'm going to skip all that. Oh, boy, you know--yeah, I'm
going to skip all that. So, you know, what I want to tell you is the--yeah, what do we--actually,
they would've find. Okay. Yeah. I'd be happy to share with you a copy of the paper, you
know. I don't want to show actually those data. You know, we actually would get, you
know, those econometric analyses. Yeah. The found is, is we find out, okay, first of all,
bidding strategies is not identical. Okay. We have different firms that have different
bidding strategies. They are heterogeneous. Okay. Yeah. Five clusters at the item level.
Okay. And we got evidence showing that supply type or non-price attributes actually change
the solution. So, whether--so, we show you--yeah. We actually show you the distribution of these
five clusters at the item level. But then for you take a look whether you are incumbent,
you know, all in your new entry, the distribution after changes. So, that means--that's the
first evidence. Supply type actually influence in the old bidding strategy. So, that's the
first takeaway. Second takeaway is actually--it's also changes the auction outcomes. Okay. So,
what happen is, you know, the--there's a significant impact, okay, of about--of the auction outcomes,
you know. As we say that the incumbent guys, okay, they actually bid higher overall in
the final bids. Okay. Because they know I'm incumbent. Okay. I have other things, you
know, so the buyer actually value. Maybe I'm close to you. Maybe my quality, you know,
there are a lot of other reasons. We have to show the evidence. And the buyers know
that and the buyers take that into consideration, okay, they--the deal actually too the incumbent.
But also new entry guys, you know, has benefit. Okay, that's the second lecture, so why those
new entry guys they keep in participating, right. So, you know. You know, and I participating
in this auction I bid--yeah, by the way, Dell really runs auctions, you know, to get memory
chips. Okay, it's just amazing how they are ahead of, you know, the profession. So the
question is, you know, so again, if there's a puzzle, you know, why if I'm not getting
the deal I am keeping--still participating in these reverse auctions, okay. The second
question is that because they have a sharp memory? They forgotten what happen. Those
are actually terms around--from industry. So therefore, we--therefore we need a model
right, this is Georgia Tech. So when it have repeated auctions, you know, the natural way
to do is Markov Chen okay. I would Markov Chen [INDISTINCT] and we also measure, between
the auctions, what kind of information you'll reveal, okay. So you do the first auction,
T minus one then do another one right in between, you know what's the lowest bid from the previous
auction, okay. And you also--you know your rank from the previous auction and you also
know whether you actually got the deal or not, whether you got the contract or not.
You don't know who got it but you know--you know whether you got it or not, right. So
this is again--so this a different way to put that Markov Chen. Okay, these are the
statement variables, how do we change? Okay, it's a latest Markov Chen Model, okay. And
this is our--you know, conjectures or hypothesis's. We actually have--it's all supported. You
know, it turns out at the suppliers they actually bid adaptively. They learn, okay. They learn
between auctions. Yeah, this is the first time, okay, you can, you can have a theory,
you conjecture but you don't have evidence. You know, this--her work is the first time
I have to reveal the evidence to show they do bid adaptively. And they also respond to
the rank, you know, to--they are ranked, okay. So this their actual data so if you move from
auction T minus one to auction T, right? So, depends whether your rank is high or rank
is low, also it depends on the strategy you are using. Are you early evaluator, are you
a mid evaluator, are you a participator or are you an optimist. And those are the four
clusters at the auction level if you measure those are the four clusters, okay. Now, you
see that, they're not changing, okay, and it depends on their rank or high. See, that's
the--this is--the probability for those of you who knows Markov Chen, okay, those are
the transition probabilities, okay. It shows that, you know, it's different between different
strategies. Okay, also depend--different between your rank, you know, whether you are high
rank or low rank, okay. Yeah, let's just take one example, okay. Example--so, if I'm an
early evaluator, okay. And then--see, I have a high rank so then you see what's going to
happen, you see I'm going to change. See, next time, you know, very likely, I want to
be a participator. Yeah, you see what I'm saying. Or you take a look at this guy. Suppose
I am a sniper, okay, and my rank is high, so--you know, for the previous one, then the
next time what I'm going to do is, you know, see I am, you know, moved, change my strategy
very likely to be a participator. You see that? You know, there's a change going on,
yeah. So, what's going on here is the falling, okay. You put that thing, you know, into this
really Markov Chen diagram, you kind of see that. You see how those firms, they are keep
changing and how do you make sense of all that if you are--let me tell you one sentence,
you know, so you can summarize all those strategy going on is, firms--actually we have evidence,
they are using this strategy called the wind stay, lose shift, okay. All this if, you know,
new words, okay. This, you know, math, okay, all these data that shows you. See, I'm bidding
with my current strategy, if I'm winning, okay, I'm going to stick with my strategy,
okay. If I don't get it, okay, I'm going to shift. How I'm going to shift, you know, we
actually show you the transition probability. It depends, you know, your rank, how you are.
So therefore, to summarize, you know, see that why they are keep in participating, they
are learning so why they are participating, okay? Of course, another reason is they have--they
have lots of capacity. You know what I'm saying, they are looking for, you know, for deals,
you know, yeah. But actually, they can learn market price because they know the lowest
bid price, they have some idea about market, okay. They also learn how competitive they
are, right because they know their rank information. They also--they learn actually bidding strategies,
okay, over the time. And as it turns, you know, the--yeah, procuring allows firms to
actually score different suppliers before I actually tell you what your score, what
your score. But you guys don't know each other, okay. I tell you whether you might prefer
guys or not, incumbent or not, right. But you have a chance, okay, to actually to move
up, okay. You join into this auction as a new entry so--a new company from China, okay.
So, if I can deliver, if I can perform, say the next auction, hey, your score got raised.
You can also, okay, move into that most favorable nation is, okay. Move into a--the incumbent
list if you can perform, if you can deliver, okay. So, therefore, that's why they are learning,
you know, they are learning. So, they can go back and work it out, okay. So, you know,
see what I need. See, what's my problem, right is cost of delivery of quality, right. So
they actually have a chance to move. So, therefore to summarize the work by Vivian, mainly the
following, this is only two chapters and she is now doing a big thing, okay, using the
big [INDISTINCT] data set. Okay. So far, she has discovered is suppliers do bid strategically,
this is new, okay. And--but why it's--why it's a big deal because how do the bid actually
affects the buyers cost savings? So, Google should care, okay. So they should care about
the bidding strategy of your suppliers because--yeah, the obedient strategy affects the final bids
and then the final bids affects your cost savings. Okay. So, we have evidence. And non-price
attributes, you know, is very influential. In theory, again, Professor Paul Milgrom has
proved, okay, in this bidding auction the decisive role is not the price. It's actually
this non-price attributes, okay. But we show the evidence, okay. Yeah, among the first.
Then we say it, why suppliers keep in bidding, okay, because they are learning, okay. And
all this is in sharply contrast with consumer auctions. So therefore, procurement auctions
is very different than consumer auctions. So that's all I have, you know, ongoing work--Vivian
is trying to start a different auction platforms, procure has them all, right. This is particular
high-tech, they know how to do business, you know, they have fixed, okay, platform and
she's now trying to compare. So, what if this is a company sourcing for lawyers versus a
company sourcing for hotel rooms, a company sourcing semi-conductors and a company who's
sourcing landscaping. And then it's a different information [INDISTINCT] okay, because it's
all there. Okay. Then she's trying to study the impact of this information revelation,
okay. And--to all those stuff we care about it, auction outcomes, final bids, cost of
savings and the contract award probability. So, that's actually her ongoing work is actually
with procuring. Okay. That's all I have okay? Yeah, thank you, thanks for coming.