< Return to Video

How Is Econometrics Changing? (Josh Angrist, Guido Imbens, Isaiah Andrews)

  • 0:00 - 0:03
    ♪ [music] ♪
  • 0:04 - 0:06
    - [Narrator] Welcome
    to Nobel Conversations.
  • 0:07 - 0:08
    In this episode,
  • 0:08 - 0:12
    Josh Angrist and Guido Imbens
    sit down with Isaiah Andrews
  • 0:12 - 0:15
    to discuss how the field
    of econometrics is evolving.
  • 0:16 - 0:19
    - [Isaiah] So, Guido and Josh,
    you're both pioneers
  • 0:19 - 0:22
    in developing tools for
    empirical research in economics.
  • 0:22 - 0:23
    And so I'd like to explore
  • 0:23 - 0:25
    where you feel like
    the field is heading --
  • 0:26 - 0:28
    economics, econometrics,
    the whole thing.
  • 0:29 - 0:31
    To start, I'd be interested to hear
  • 0:32 - 0:35
    about whether you feel
    the way in which
  • 0:35 - 0:39
    the local average treatment
    effects framework took hold
  • 0:39 - 0:42
    has any lessons for how
    new empirical methods in economics
  • 0:42 - 0:44
    develop and spread
    or how they should.
  • 0:45 - 0:46
    - [Josh] That's a good question.
  • 0:47 - 0:48
    You go first.
  • 0:48 - 0:49
    [laughter]
  • 0:50 - 0:53
    - [Guido] Yeah, so I think
    the important thing
  • 0:53 - 0:59
    is to come up
    with good convincing cases
  • 0:59 - 1:02
    where the questions are clear
  • 1:02 - 1:06
    and where the methods
    apply in general.
  • 1:06 - 1:08
    One thing I --
  • 1:08 - 1:12
    looking back
    at the subsequent literature.
  • 1:12 - 1:17
    So I really like the regression
    discontinuity literature
  • 1:17 - 1:20
    where there were clearly a bunch
    of really convincing examples
  • 1:20 - 1:23
    and that allowed people
    to think more clearly,
  • 1:23 - 1:27
    look harder
    at the methodological questions.
  • 1:27 - 1:29
    Do clear applications
  • 1:29 - 1:31
    that then allow you
    to kind of think about,
  • 1:31 - 1:34
    "Wow, does this type of assumption
    seem reasonable here?
  • 1:34 - 1:38
    What kind of things do we not like
    in the early papers?
  • 1:38 - 1:40
    How can we improve things?"
  • 1:40 - 1:44
    So having clear applications
    motivating these literatures
  • 1:44 - 1:46
    I think is very helpful.
  • 1:47 - 1:48
    - I'm glad you mentioned
  • 1:48 - 1:49
    the regression
    discontinuity, Guido.
  • 1:49 - 1:53
    I think there's a lot of
    complementarity between IV and RD,
  • 1:55 - 1:57
    Instrumental Variables
    and Regression Discontinuity.
  • 2:01 - 2:03
    A lot of the econometric
    applications
  • 2:03 - 2:05
    of regression discontinuity
  • 2:05 - 2:07
    are what used to be called
    "fuzzy" RD,
  • 2:07 - 2:12
    where, you know, it's not discrete
    or deterministic at the cutoff
  • 2:12 - 2:15
    but just the change
    in rates or intensity.
  • 2:15 - 2:18
    And the late framework
    helps us understand
  • 2:18 - 2:19
    those applications
  • 2:19 - 2:21
    and gives us a clear interpretation
  • 2:21 - 2:25
    for something like
    in my paper with Victor Lavy,
  • 2:25 - 2:28
    where we use Maimonides'
    rule, the class size cutoffs,
  • 2:28 - 2:30
    and what are you getting there?
  • 2:30 - 2:32
    Of course, you can
    answer that question
  • 2:32 - 2:34
    with a linear
    constant effects model,
  • 2:34 - 2:36
    but it turns out
    we're not limited to that,
  • 2:36 - 2:40
    and RD is still very powerful
    and illuminating,
  • 2:41 - 2:43
    even when the correlation
  • 2:43 - 2:46
    between the cutoff
    and the variable of interest,
  • 2:46 - 2:49
    in this case class size,
    is partial,
  • 2:49 - 2:51
    maybe even not that strong.
  • 2:52 - 2:55
    So there was definitely kind
    of a parallel development.
  • 2:55 - 2:56
    It's also interesting --
  • 2:57 - 3:00
    nobody talked about
    regression discontinuity designs
  • 3:00 - 3:01
    when we were in graduate school.
  • 3:01 - 3:05
    It was something that other social
    scientists were interested in,
  • 3:06 - 3:10
    and that grew up
    alongside the LATE framework,
  • 3:10 - 3:12
    and we've both done work
  • 3:12 - 3:15
    on both applications
    and methods there,
  • 3:15 - 3:18
    and it's been very exciting
    to see that develop
  • 3:18 - 3:20
    and become so important.
  • 3:20 - 3:22
    It's part of a general evolution,
  • 3:22 - 3:26
    I think, towards credible
    identification strategies,
  • 3:26 - 3:27
    causal effects...
  • 3:29 - 3:31
    making econometrics
  • 3:31 - 3:33
    more about causal questions
    than about models.
  • 3:34 - 3:35
    In terms of the future,
  • 3:35 - 3:38
    I think one thing that LATE
    has helped facilitate
  • 3:38 - 3:42
    is a move towards
    more creative, randomized trials
  • 3:42 - 3:44
    where there's
    something of interest.
  • 3:46 - 3:48
    It's not possible
    or straightforward
  • 3:48 - 3:51
    to simply turn it off or on,
  • 3:51 - 3:55
    but you can encourage it
    or discourage it.
  • 3:55 - 3:58
    So you subsidize schooling
    with financial aid, for example.
  • 3:59 - 4:02
    So now we have a whole
    framework for interpreting that,
  • 4:04 - 4:07
    and it opens the doors
    to randomized trials
  • 4:07 - 4:10
    of things that maybe would
  • 4:10 - 4:12
    not have seemed possible before.
  • 4:14 - 4:18
    We've used that a lot in the work
    we do on schools in our --
  • 4:18 - 4:21
    in the Blueprint Lab at MIT.
  • 4:22 - 4:27
    We're exploiting random assignment
    in very creative ways, I think.
  • 4:28 - 4:31
    - [Isaiah] Related to that,
    do you see particular factors
  • 4:31 - 4:34
    that make for useful research
    in econometrics?
  • 4:34 - 4:38
    You've alluded to
    it having a clear connection
  • 4:38 - 4:40
    to problems
    that are actually coming up,
  • 4:40 - 4:43
    and empirical practice
    is often a good idea.
  • 4:43 - 4:45
    - Isn't it always a good idea?
  • 4:46 - 4:47
    I often find myself sitting
  • 4:47 - 4:50
    in an econometrics
    theory seminar,
  • 4:51 - 4:52
    say the Harvard MIT seminar,
  • 4:53 - 4:56
    and I'm thinking, "What problem
    is this guy solving?
  • 4:56 - 4:58
    Who has this problem?"
  • 4:58 - 5:00
    And, you know...
  • 5:02 - 5:05
    sometimes there's an
    embarrassing silence if I ask
  • 5:05 - 5:08
    or there might be
    a fairly contrived scenario.
  • 5:09 - 5:12
    I want to see
    where the tool is useful.
  • 5:12 - 5:15
    There are some
    purely foundational tools.
  • 5:15 - 5:16
    I do take the point.
  • 5:16 - 5:22
    There are people who are working
    on conceptual foundations of ...
  • 5:23 - 5:25
    it becomes more like
    mathematical statistics.
  • 5:26 - 5:28
    I mean, I remember
    an early example of that
  • 5:28 - 5:30
    that I struggled to understand
  • 5:30 - 5:32
    was the idea
    of stochastic equicontinuity,
  • 5:32 - 5:35
    which one of my thesis advisors,
    Whitney Newey,
  • 5:35 - 5:36
    was using to great effect,
  • 5:36 - 5:39
    and I was trying
    to understand that.
  • 5:41 - 5:42
    It's really foundational.
  • 5:42 - 5:45
    it's not an application
    that's driving that --
  • 5:46 - 5:47
    at least not immediately.
  • 5:49 - 5:53
    But most things are not like that,
    and so there should be a problem.
  • 5:54 - 5:59
    And I think it's on the seller
    of that sort of thing,
  • 6:00 - 6:02
    because there's opportunity cost,
  • 6:02 - 6:05
    the time and attention,
    and effort to understand things.
  • 6:06 - 6:07
    It's on the seller to say,
  • 6:07 - 6:09
    "Hey, I'm solving this problem,
  • 6:09 - 6:13
    and here's a set of results
    that show that it's useful,
  • 6:13 - 6:15
    and here's some insight
    that I get."
  • 6:16 - 6:18
    - [Isaiah] As you said, Josh,
    there's been a move
  • 6:18 - 6:21
    in the direction of thinking
    more about causality
  • 6:21 - 6:23
    in economics and empirical
    work in economics.
  • 6:23 - 6:25
    Any consequences of the --
  • 6:25 - 6:27
    the spread of that view
    that surprised you
  • 6:27 - 6:28
    or anything that you view
    as downsides
  • 6:29 - 6:31
    of the way that empirical
    economics has gone?
  • 6:32 - 6:34
    - Sometimes I see
    somebody does IV,
  • 6:34 - 6:38
    and they get a result
    which seems implausibly large.
  • 6:39 - 6:40
    That's the usual case.
  • 6:42 - 6:45
    So it might be, you know,
    an extraordinarily large
  • 6:45 - 6:49
    causal effect of some
    relatively minor intervention,
  • 6:49 - 6:52
    which was randomized
    or for which you could make a case
  • 6:52 - 6:54
    that there's a good design.
  • 6:55 - 6:58
    And then when I see that,
    and, you know, I think
  • 6:59 - 7:00
    it's very hard for me to believe
  • 7:00 - 7:02
    that this relatively
    minor intervention
  • 7:02 - 7:04
    has such a large effect.
  • 7:04 - 7:06
    The author will sometimes resort
  • 7:06 - 7:09
    to the local average
    treatment effects theorem
  • 7:09 - 7:11
    and say, "Well, these compliers,
  • 7:11 - 7:13
    you know, they're
    special in some way."
  • 7:13 - 7:16
    And, you know,
    they just benefit extraordinarily
  • 7:16 - 7:18
    from this intervention.
  • 7:18 - 7:21
    And I'm reluctant
    to take that at face value.
  • 7:21 - 7:24
    I think, you know,
    often when effects are too big,
  • 7:24 - 7:27
    it's because the exclusion
    restriction is failing,
  • 7:27 - 7:29
    so you don't really have the right
    endogenous variable
  • 7:29 - 7:31
    to scale that result.
  • 7:32 - 7:36
    And so I'm not too happy to see
  • 7:36 - 7:39
    you know, just sort of
    a generic heterogeneity
  • 7:39 - 7:42
    argument being used
    to excuse something
  • 7:42 - 7:44
    that I think might be
    a deeper problem.
  • 7:45 - 7:47
    - [Guido] I think it played
    somewhat of an unfortunate role
  • 7:47 - 7:50
    when the discussions kind of
    between reduced form
  • 7:50 - 7:52
    and structural approaches,
  • 7:52 - 7:56
    where I feel
    that wasn't quite right.
  • 7:56 - 7:59
    The instrumental
    variables assumptions
  • 7:59 - 8:04
    are at the core - structural
    assumptions about behavior -
  • 8:04 - 8:05
    they were coming from economic --
  • 8:07 - 8:10
    thinking about the economic
    behavior of agents,
  • 8:10 - 8:15
    and somehow it got
    pushed in a direction
  • 8:15 - 8:18
    that I think wasn't
    really very helpful.
  • 8:19 - 8:22
    The way I think, initially the --
  • 8:23 - 8:26
    we wrote things up, it was
    describing what was happening,
  • 8:26 - 8:29
    there were a set of methods
    people were using,
  • 8:29 - 8:32
    we clarified what
    those methods were doing
  • 8:33 - 8:38
    and in a way that I think
    contain a fair amount of insight,
  • 8:39 - 8:42
    but it somehow
    it got pushed into a corner
  • 8:42 - 8:45
    that I don't think
    was necessarily very helpful.
  • 8:45 - 8:47
    - [Isaiah] I mean,
    just the language
  • 8:47 - 8:49
    of reduced form versus structural
  • 8:49 - 8:51
    I find kind of funny
    in the sense that,
  • 8:51 - 8:53
    right, the local average treatment
    effect model, right,
  • 8:53 - 8:56
    the potential outcomes model is
    a nonparametric structural model,
  • 8:56 - 8:59
    if you want to think about it,
    as you sort of suggested, Guido.
  • 8:59 - 9:01
    So there's something a little funny
  • 9:01 - 9:03
    about putting these
    two things in oposition when --
  • 9:03 - 9:05
    - [Guido] Yes.
    - [Josh] Well, that language,
  • 9:05 - 9:08
    of course, comes from
    the [inaudible] equations framework
  • 9:08 - 9:10
    that we inherited.
  • 9:10 - 9:12
    It has the advantage
  • 9:12 - 9:14
    that people seem to know
    what you mean when you use it,
  • 9:14 - 9:16
    but might be that people
    are hearing different,
  • 9:16 - 9:18
    different people
    are hearing different things.
  • 9:18 - 9:21
    - [Guido] Yeah. I think [inaudible]
    has sort of become --
  • 9:21 - 9:23
    used in a little bit
    of the pejorative way, yeah?
  • 9:23 - 9:25
    - [Josh] Sometimes.
    - [Guido] [The word].
  • 9:25 - 9:28
    Which is not really quite what
    it was originally intended for.
  • 9:30 - 9:33
    - [Isaiah] I guess something else
    that strikes me in thinking about
  • 9:33 - 9:36
    the effects of the local average
    treatment effect framework
  • 9:36 - 9:38
    is that often folks will appeal to
  • 9:38 - 9:42
    a local average treatment effects
    intuition for settings well beyond
  • 9:42 - 9:44
    ones where any sort
    of formal result
  • 9:44 - 9:45
    has actually been established.
  • 9:45 - 9:50
    And I'm curious, given all the work
    that you guys did to, you know,
  • 9:50 - 9:52
    establish late results
    in different settings,
  • 9:52 - 9:54
    I'm curious, any thoughts on that?
  • 9:55 - 9:57
    - [Guido] I think there's
    going to be a lot of cases
  • 9:57 - 10:02
    where the intuition
    does get you some distance,
  • 10:03 - 10:05
    but it's going to be
    somewhat limited
  • 10:05 - 10:08
    and establishing
    formal results there
  • 10:08 - 10:09
    may be a little tricky
  • 10:09 - 10:13
    and then maybe only work
    in special circumstances,
  • 10:15 - 10:17
    and you end up
    with a lot of formality
  • 10:17 - 10:20
    that may not quite
    capture the intuition.
  • 10:20 - 10:22
    Sometimes I'm somewhat
    uneasy with them
  • 10:22 - 10:24
    and they are not necessarily
    the papers I would want to write,
  • 10:25 - 10:27
    but I do think something --
  • 10:27 - 10:31
    intuition often does capture
    part of the problem.
  • 10:33 - 10:36
    I think, in some sense we were
    kind of very fortunate there
  • 10:37 - 10:39
    in the way that the late paper
    got handled at the journal,
  • 10:39 - 10:42
    is that, actually, the editor,
    made it much shorter
  • 10:42 - 10:46
    and that then allowed us to kind of
    focus on very clear, crisp results.
  • 10:47 - 10:52
    Where if you -- you know, this
    somewhat unfortunate tendency
  • 10:52 - 10:53
    in the econometrics literature
  • 10:53 - 10:55
    of having the papers
    get longer and longer.
  • 10:55 - 10:57
    - [Josh] Well, you should
    be able to fix that, man.
  • 10:57 - 10:59
    - [Guido] I'm trying to fix that.
  • 10:59 - 11:02
    But I think this is an example
    where it's sort of very clear
  • 11:02 - 11:03
    that having it be short
    is actually --
  • 11:03 - 11:05
    - [Josh] You should impose
    that no paper
  • 11:05 - 11:07
    can be longer than the late paper.
  • 11:07 - 11:08
    - [Guido] That, wow.
  • 11:08 - 11:10
    That may be great.
  • 11:10 - 11:12
    - [Josh] At least no theory,
    no theory paper.
  • 11:12 - 11:14
    - [Guido] Yeah,
    and I think, I think...
  • 11:14 - 11:17
    I'm trying very hard to get
    the papers to be shorter.
  • 11:17 - 11:19
    And I think there's a lot of value
  • 11:19 - 11:21
    today because it's often
    the second part of the paper
  • 11:21 - 11:26
    that doesn't actually get you much
    further in understanding things
  • 11:27 - 11:30
    and it does make things
    much harder to read
  • 11:31 - 11:33
    and, you know,
    it sort of goes back
  • 11:33 - 11:36
    to how I think econometrics
    should be done,
  • 11:36 - 11:38
    you should focus on --
  • 11:39 - 11:41
    It should be reasonably
    close to empirical problems.
  • 11:42 - 11:44
    They should be very clear problems.
  • 11:45 - 11:49
    But then often the theory
    doesn't need to be quite so long.
  • 11:49 - 11:50
    - [Josh] Yeah.
  • 11:51 - 11:55
    - [Guido] I think things have gone
    a little off track.
  • 11:56 - 11:58
    - [Isaiah] A new relatively
    recent change
  • 11:58 - 12:00
    has been a seeming big
    increase in demand
  • 12:00 - 12:02
    for people with sort of
    econometrics,
  • 12:02 - 12:05
    causal effect estimation skills
    in the tech sector.
  • 12:05 - 12:07
    I'm interested,
    do either of you have thoughts
  • 12:07 - 12:10
    on sort of how
    that's going to interact
  • 12:10 - 12:12
    with the development
    of empirical methods,
  • 12:12 - 12:14
    or empirical research
    in economics going forward?
  • 12:15 - 12:17
    - [Josh] Well, there's
    sort of a meta point,
  • 12:17 - 12:21
    which is, there's this new
    kind of employer,
  • 12:22 - 12:28
    the Amazons and the Uber,
    and, you know, TripAdvisor world,
  • 12:28 - 12:29
    and I think that's great.
  • 12:29 - 12:33
    And I like to tell my students
    about that, you know, especially --
  • 12:33 - 12:36
    at MIT we have a lot of
    computer science majors.
  • 12:36 - 12:37
    That's our biggest major.
  • 12:37 - 12:42
    And I try to seduce some of those
    folks into economics by saying,
  • 12:42 - 12:46
    you know, you can go
    work for these,
  • 12:46 - 12:49
    you know, companies that people
    are very keen to work for
  • 12:49 - 12:51
    because the work seems exciting,
  • 12:52 - 12:54
    you know, that the skills
    that you get in econometrics
  • 12:54 - 12:56
    are as good or better
  • 12:56 - 13:00
    than any competing
    discipline has to offer.
  • 13:00 - 13:01
    So you should at least
  • 13:01 - 13:04
    take some econ, take some
    econometrics, and some econ.
  • 13:05 - 13:07
    I did a fun project with Uber
  • 13:08 - 13:10
    on labor supply of Uber drivers
  • 13:10 - 13:13
    and was very, very exciting
    to be part of that.
  • 13:13 - 13:15
    Plus I got to drive
    for Uber for a while
  • 13:16 - 13:18
    and I thought that was fun too.
  • 13:18 - 13:21
    I did not make enough
    that I was tempted to
  • 13:21 - 13:25
    give up my MIT job,
    but I enjoyed the experience.
  • 13:25 - 13:31
    I see a potential challenge to our
    model of graduate education here,
  • 13:32 - 13:37
    which is, if we're training people
    to go work at Amazon, you know,
  • 13:38 - 13:41
    it's not clear why, you know,
    we should be paying
  • 13:41 - 13:43
    graduate stipends for that.
  • 13:43 - 13:47
    Why should the taxpayer effectively
    be subsidizing that.
  • 13:47 - 13:51
    Our graduate education
    in the US Is generously subsidized,
  • 13:51 - 13:53
    even in private universities,
  • 13:53 - 13:56
    it's ultimately -- there's a lot of
    public money there,
  • 13:56 - 13:59
    and I think the
    traditional rationale for that is,
  • 14:00 - 14:02
    you know, we were training
    educators and scholars,
  • 14:02 - 14:06
    and there's a great externality
    from the work that we do,
  • 14:06 - 14:08
    it's either
    the research externality,
  • 14:08 - 14:10
    or a teaching externality.
  • 14:10 - 14:12
    But, you know,
    if many of our students
  • 14:12 - 14:15
    are going to work
    in the private sector,
  • 14:16 - 14:22
    that's fine, but maybe their
    employers should pay for that.
  • 14:22 - 14:23
    - [Guido] But maybe
    is not so different
  • 14:23 - 14:27
    from people working
    for consulting firms.
  • 14:27 - 14:29
    It's not clear to me
  • 14:29 - 14:33
    that the number of jobs
    in academics has changed.
  • 14:33 - 14:36
    - [Josh] I feel like this
    is a growing sector,
  • 14:36 - 14:39
    whereas consulting --
    you're right to raise that,
  • 14:39 - 14:42
    it might be the same
    for consulting,
  • 14:43 - 14:45
    but this, you know,
  • 14:45 - 14:48
    I'm placing more and more
    students in these businesses.
  • 14:48 - 14:50
    So, it's on my mind
    in a way that I've sort of,
  • 14:51 - 14:54
    you know, not been attentive
    to consulting jobs,
  • 14:54 - 14:57
    you know, consulting
    was always important,
  • 14:57 - 14:59
    and I think also
    there's some movement
  • 14:59 - 15:03
    from consulting back into research,
    it's a little more fluid.
  • 15:04 - 15:08
    A lot of the work in both domains
  • 15:08 - 15:09
    I have to say,
    it's not really different
  • 15:09 - 15:13
    but, you know, people who
    are working in the tech sector
  • 15:13 - 15:15
    are doing things that are
    potentially of scientific interest,
  • 15:15 - 15:17
    but mostly it's hidden.
  • 15:17 - 15:19
    Then you really I have to say,
  • 15:19 - 15:21
    you know, why is the government
    paying for this?
  • 15:22 - 15:24
    Yeah, although, yeah,
    I mean to Guidos point,
  • 15:24 - 15:26
    I guess there's
    a data question here
  • 15:26 - 15:30
    of it has the sort of
    total [no-neck] sort of say
  • 15:31 - 15:35
    for-profit sector employment
    of econ Ph.D. program graduates
  • 15:35 - 15:38
    increased or has it just been
    a substitution from finance
  • 15:38 - 15:40
    and consulting towards tech.
  • 15:40 - 15:42
    - [Josh] I may be reacting
    to something
  • 15:42 - 15:44
    that's not really happening.
  • 15:44 - 15:46
    - [Guido] I've actually
    done some work
  • 15:46 - 15:48
    with some of these tech companies.
  • 15:49 - 15:52
    I don't disagree with Josh's point
    that we need to think
  • 15:52 - 15:54
    a little bit about
    the funding model,
  • 15:54 - 15:56
    who is it in the end paying
    for the graduate education.
  • 15:57 - 15:59
    But from a scientific perspective,
  • 16:00 - 16:03
    not only do these places
    have great data
  • 16:03 - 16:05
    and nowadays they tend to be
    very careful with that
  • 16:05 - 16:07
    for privacy reasons,
  • 16:07 - 16:09
    but they also have great questions.
  • 16:10 - 16:13
    I find it very inspiring
    kind of to listen
  • 16:13 - 16:16
    to the people there and kind of see
    what kind of questions they have,
  • 16:16 - 16:17
    and often they're questions
  • 16:18 - 16:22
    that also come up outside
    of these companies.
  • 16:22 - 16:27
    I have a couple of papers
    with Raj Chetty and Susan Athey,
  • 16:27 - 16:32
    where we look at ways
    of combining experimental data
  • 16:32 - 16:34
    and observational data,
    and kind of their --
  • 16:36 - 16:39
    Raj Chetty was interested
    in what is the effect
  • 16:39 - 16:43
    of early childhood programs
    on outcomes later in life,
  • 16:43 - 16:46
    not just kind on test scores,
    but on earnings and stuff,
  • 16:46 - 16:48
    and we kind of developed methods
  • 16:49 - 16:52
    that would help you shed
    light on that, onto some --
  • 16:53 - 16:57
    in some settings
    and the same problems came up
  • 16:57 - 17:01
    kind of in this
    tech company settings.
  • 17:01 - 17:03
    And so from my perspective,
  • 17:03 - 17:05
    it's the same kind of --
  • 17:05 - 17:08
    I was talking to people
    doing empirical work,
  • 17:08 - 17:10
    I tried to kind of look at these
    specific problems
  • 17:10 - 17:13
    and then try to come up
    with more general problems,
  • 17:15 - 17:18
    we formulated the problems
    at a higher level,
  • 17:18 - 17:23
    so that I can think about solutions
    that work in a range of settings.
  • 17:23 - 17:25
    And so from that perspective,
  • 17:25 - 17:28
    the interactions
    with the tech companies
  • 17:28 - 17:30
    are just very valuable
    and very useful.
  • 17:32 - 17:35
    We do have students now
    doing internships there
  • 17:35 - 17:38
    and then coming back
    and writing more interesting thesis
  • 17:38 - 17:43
    as a result of their
    experiences there.
  • 17:45 - 17:47
    - [Narrator] If you'd like to watch
    more Nobel Conversations,
  • 17:47 - 17:48
    click here,
  • 17:48 - 17:50
    or if you'd like to learn
    more about econometrics,
  • 17:50 - 17:53
    check out Josh's
    "Mastering Econometrics" series.
  • 17:54 - 17:57
    If you'd like to learn more
    about Guido, Josh and Isaiah
  • 17:57 - 17:58
    check out the links
    in the description.
  • 17:59 - 18:01
    ♪ [music] ♪
Title:
How Is Econometrics Changing? (Josh Angrist, Guido Imbens, Isaiah Andrews)
ASR Confidence:
0.80
Description:

more » « less
Video Language:
English
Team:
Marginal Revolution University
Duration:
18:03

English subtitles

Revisions Compare revisions