1
00:00:00,000 --> 00:00:06,094
I. When I first talked about why we model,
in this class, one of the reasons I gave

2
00:00:06,094 --> 00:00:10,745
was to design better institutions. In this
set of lectures we're going to talk about

3
00:00:10,745 --> 00:00:15,231
something known as mechanism design. And
the idea behind mechanism design is that

4
00:00:15,231 --> 00:00:19,660
you formalize an institution. You do so as
follows, you basically decide, here are

5
00:00:19,660 --> 00:00:24,146
the set of actions that people can take,
and here are the payoffs, or the outcomes,

6
00:00:24,146 --> 00:00:28,409
associated with those actions. So in a
sense, you're defining the rules of the

7
00:00:28,409 --> 00:00:32,521
game, the actions people can take, and
what the payoffs are gonna be. Now in

8
00:00:32,521 --> 00:00:36,994
constructing these mechanisms, there's two
features that often arise in social

9
00:00:36,994 --> 00:00:41,524
context that you try to overcome through
the proper writing of a mechanism. The

10
00:00:41,524 --> 00:00:45,825
first one is, hidden actions, so often
times, especially this is true in work

11
00:00:45,825 --> 00:00:50,241
environments, you don't see the actions
people take, you only see the outcomes.

12
00:00:50,241 --> 00:00:54,771
What you'd like to do is you'd like to
write employment contracts and sentence

13
00:00:54,771 --> 00:00:59,531
structures in such a way, that people take
the actions that you want. So even though

14
00:00:59,531 --> 00:01:04,127
the actions are hidden you?d like to write
some sort of mechanism so that the actions

15
00:01:04,127 --> 00:01:08,185
people are going to take, the optimal
actions for people to do are going to be

16
00:01:08,185 --> 00:01:12,054
actions you want. Now there's gonna be
other cases where's it's not actions

17
00:01:12,054 --> 00:01:16,170
you're worried about but its information.
So you may not know someone's type. So,

18
00:01:16,170 --> 00:01:20,338
for example, someone could be high ability
or low ability. Someone could be a safe

19
00:01:20,338 --> 00:01:24,403
driver or a risky driver. And what you'd
like to do is you'd like to write some

20
00:01:24,403 --> 00:01:28,468
sort of contract so that these people
separate out. So you can identify who the

21
00:01:28,468 --> 00:01:32,687
high type people are and who the low type
people are or, who the risky drivers are

22
00:01:32,687 --> 00:01:36,379
and who the safe drivers are. So we
construct these mechanisms. Two of the

23
00:01:36,379 --> 00:01:40,477
problems we want to sort of overcome are
hidden actions, because we can't see what

24
00:01:40,477 --> 00:01:44,376
people are doing, and hidden information.
We can't figure out what type?s people

25
00:01:44,376 --> 00:01:48,274
are, what they like, how risky they are,
how much ability they have, that sort of

26
00:01:48,274 --> 00:01:52,423
thing. So what we'd like to do is figure
out how do we run institutions to overcome

27
00:01:52,423 --> 00:01:56,322
these two problems. So once we get that
core idea in mind, about sort of how we

28
00:01:56,322 --> 00:02:00,320
can use institutions to figure things out
that are hidden, we'll take them, we'll

29
00:02:00,320 --> 00:02:04,289
apply them to some real world examples.
So, the first thing we'll apply it to is

30
00:02:04,289 --> 00:02:08,176
auctions. When we think about auctions,
there's a lot of different ways we can

31
00:02:08,176 --> 00:02:12,164
auction things off. And we'll show how we
can think of these different rules as

32
00:02:12,164 --> 00:02:16,077
mechanisms. Then what we'll do is we'll
show a theorem that shows which of these

33
00:02:16,077 --> 00:02:19,873
rules are best. So do we have an ascending
bid auction, do we have a sealed bid

34
00:02:19,873 --> 00:02:23,571
auction, or do we have something called a
second price auction? So we can use

35
00:02:23,571 --> 00:02:27,561
theory, we can use models to figure out
which of those particular auction rules is

36
00:02:27,561 --> 00:02:31,562
gonna work best. After we do auctions
we'll look at something called public

37
00:02:31,562 --> 00:02:35,877
goods. So public goods are sort of like a
collective action problem, there goods

38
00:02:35,877 --> 00:02:40,301
that everybody benefits the more we have
of, but you'd like everybody else to pay

39
00:02:40,301 --> 00:02:44,671
for them. So things like roads, clean air,
clean beaches, those sorts of things, we'd

40
00:02:44,671 --> 00:02:49,200
all like those sorts of things. Exist, but
we prefer that other people pay for them.

41
00:02:49,200 --> 00:02:53,035
So we'll talk about what a public good is
and we'll talk about how you can use

42
00:02:53,035 --> 00:02:56,920
mechanism design to come up with solutions
to public good problems. Now in doing

43
00:02:56,920 --> 00:03:00,513
this, I'm gonna do it a little bit
untraditionally. So, why don't I refer to

44
00:03:00,513 --> 00:03:04,349
a quote here, Marie Galmont once said,
imagine how physic, difficult physics

45
00:03:04,349 --> 00:03:08,039
would be if electrons could think. So
remember when we talked about modeling

46
00:03:08,039 --> 00:03:11,730
people, how difficult it was. And we had
three basic models that are rational

47
00:03:11,730 --> 00:03:15,469
choice model, we got a psychological
model, and we have rule based models. What

48
00:03:15,469 --> 00:03:19,353
we're gonna do here is we're gonna do
mechanism design the standard way, and the

49
00:03:19,353 --> 00:03:23,627
standard way is to assume that people are
rational. So we're gonna sort of lay out

50
00:03:23,627 --> 00:03:28,091
the basics of mechanism design assuming
rational agents. But after we do that

51
00:03:28,091 --> 00:03:33,078
we'll talk through what if people suffered
from physiological biases or were slightly

52
00:03:33,078 --> 00:03:37,600
irrational Or what if it were the case
that people just followed simple rules

53
00:03:37,600 --> 00:03:42,007
would our same results still follow. So
we're gonna to have a dialogue using

54
00:03:42,007 --> 00:03:46,530
mechanisms design as a framework for
thinking again about how we model people.

55
00:03:46,530 --> 00:03:50,762
Okay, so that's an outline of what we're
going to do. What we're going to do first

56
00:03:50,762 --> 00:03:54,734
is just add some of the basics of
mechanism design, talk about hidden action

57
00:03:54,734 --> 00:03:58,914
information, move on to options, and then
conclude with some discussion of public

58
00:03:58,914 --> 00:04:00,953
good games. All right, let's get started.