YT YouTube2Text

YouTube2Text → Transcripts → YouTube transcript (DKWVDvk1QYo)

YouTube transcript (DKWVDvk1QYo) — Transcript

9,948 words · 1,766 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00welcome you to meetings
0:01uh into this particular session i'm
0:04impressed to draw up this early good to
0:06come out for the session and uh
0:09glad glad to have all of you here we had
0:11talked for a little while
0:13on this subject but you know my plan is
0:15to leave plenty of time
0:17at the end for
0:19for a q a so if you have any any
0:21thoughts along the way floor them up and
0:23uh i'm sure they won't be
0:26unique um
0:29so the this topic is what i think that
0:32those of you presumably all of you since
0:34you're here who do work in in the area
0:37of corporate finance uh encounter
0:40problems associated with endogeneity and
0:42identification in your empirical work so
0:45i thought
0:46i would start like i do sometimes in
0:49classrooms that you start with a
0:50question
0:51and say uh
0:53among all of you who have submitted
0:55papers and
0:57finance how many of you
0:59have had a referee or an editor say
1:03i don't buy your identification strategy
1:05or something to that effect
1:10that's all really does that mean the
1:12rest of you having some good papers yet
1:14too
1:17i think you're hard-pressed
1:20in corporate finance to
1:22uh to do work without
1:25that
1:26endogenous problem coming up without
1:29your identification strategy being
1:31criticized in some way and so
1:34it begs a lot of questions i think in
1:36terms of you know how how can we do this
1:39research in corporate finance in a
1:41meaningful way while while still
1:45dealing with this
1:46endogenous problem a lot of people
1:48complain about how they're treated
1:51by referees and editors on this issue
1:54and
1:54and refer to uh so-called identification
1:57of beliefs so an alternative title this
2:00talk might be eluding the long arm of
2:02the identification police
2:05but the my idea today is not really to
2:08to talk about econometrics i mean
2:10there's been
2:12several very good tutorials at the fma
2:14over the years on that subject i'm more
2:17interested in talking about
2:19sort of research design issues in
2:22empirical corporate finance with a
2:24particular eye towards expanding the the
2:27types of studies that we do because i
2:30one of the concerns that i have
2:32is that this identification problem has
2:35has become
2:36so pervasive in our work and sort of
2:38narrowed down the the styles of papers
2:41that we think are acceptable to the
2:44journals now and as a result we're
2:46potentially limiting the scope
2:48of questions that we're asking in the
2:50field and i don't think that would be a
2:52good development if that's the case so
2:55uh what i'm going to do is is is the
2:57following just sort of talk
2:59a little bit about start about the
3:01basics of what the endogenous problem is
3:04i think everyone's pretty familiar with
3:06that as well as
3:07some of the econometric responses that
3:10exist in the literature and when i talk
3:13about those again this isn't a talk
3:15about econometrics but more just to
3:17point out that
3:19none of these are our perfect solutions
3:21by any stretch they all have
3:24some limitations to them we're never
3:26completely solving the problem
3:30but what we have done is sort of narrow
3:32down the the
3:34the set of papers i think that that we
3:36tend to do in empirical course of
3:38finance and that that raises these these
3:41concerns that i rooted to
3:43uh a couple minutes ago
3:45and then what i'm going to do is just
3:46sort of apply some of these ideas in the
3:49field of research that i've been doing
3:51more work on lately which is capital
3:53structure research so
3:54i'm going to to talk about alternative
3:57different types of research designs all
4:01of which i think are useful in terms of
4:04what we're really trying to do in the
4:06profession which is expand the set of
4:08knowledge that we have on a particular
4:10topic
4:12none of them are perfect many of them
4:14will suffer from identification problems
4:17but that doesn't mean they
4:19they aren't useful and so i i think this
4:22is this is where i'm going to try to
4:24convince you that we need to re-expand
4:26the types of studies that we're doing if
4:29we really want to get at some of the big
4:31questions that we encountered in
4:33corporate finance and i'll run through
4:36several examples from from the recent
4:38capital structure literature that i
4:40think
4:41illustrate the points that i'm trying to
4:43make
4:43both in terms of what we can learn from
4:46those types of studies as well as what
4:48we can't learn and how you can
4:51think about doing studies like that
4:54in a way that uh
4:56will be
4:57viewed as useful by the journals and
5:00therefore uh publishable in uh in the
5:02literature and then we'll try to
5:04draw some conclusions from that
5:07okay so
5:08the basic endogenicity problem i think
5:10everyone is pretty familiar with a lot
5:12of the research
5:14that we do in corporate finance is
5:16really geared towards trying to
5:18establish some causal connection between
5:21some variable x1 in this case and and
5:24some
5:25dependent variable um that lists this as
5:29y
5:29in this particular case so you think of
5:32any regression framework doesn't have to
5:34be a linear regression like this but
5:36it's easier to represent
5:37this way
5:39we have the we have the association
5:41between y and x1 while controlling so
5:44for some vector
5:46of x variables that we think are our
5:50other possible determinants of why we're
5:52trying to establish better access
5:56on online so
5:58common examples in the corporate finance
6:01literature you know does the question
6:03might be does debt constrain investment
6:06somewhere so we can we can test the
6:09association between debt levels and
6:11levels of investment
6:13do taxes affect capital structure so
6:17it's the association between some
6:18measure of the of the
6:20of the tax obligation that the firm is
6:23facing and the leverage choice that they
6:24made
6:26or does
6:27does governance affect value in mind but
6:30a
6:31notable example that has a connection
6:33between ownership
6:35and some measure of value like like
6:37total execute right so that's these are
6:39typical tests all of which
6:42suffer from this this identification
6:44problem because we don't know
6:47whether there's a causal connection
6:48between this variable x x sub 1 and the
6:52dependent variable or whether we've got
6:54some correlated variables so ultimately
6:57the issue is whether
6:58there is some some correlation between
7:00this x1 variable and the error term
7:03of this model now obviously
7:06if we had some idea what t-submitted
7:08variables were we put them in
7:11and i know we could take care of the
7:12problem but the issue is we don't know
7:15what they are
7:16and
7:17it's very easy for a referee or an
7:21editor to say well there's something out
7:22there
7:23that's that's correlated both with your
7:26x1 variable and this dependent variable
7:29that you're trying to test for
7:31and so in the end i just don't buy your
7:33identification strategy i don't buy that
7:35there is a
7:36causal connection between you so for
7:38example
7:40you know if we run we run regressions of
7:42investment on lab rates we will find
7:44that the higher is the leverage the
7:47lower is the level of investment
7:49right now it could be
7:51because debt constrains investment
7:53all right or it could just be that firms
7:56that have poor growth opportunities
8:00tend to have higher leverage so we would
8:02predict they should be in capital
8:03structure
8:05and we just simply can't perfectly
8:07control for growth opportunities so even
8:09if we have some proxies for growth
8:11opportunities on the right hand side
8:13if we can't perfectly control for them
8:16the omitted partisans embedded in the
8:18error term and we have this positive
8:19correlation between uh debt and and the
8:23error term
8:24and we're stuck
8:26so what do we do about that so where i
8:29think the profession has really made
8:31some some big strides in in the last 10
8:34to 15 years is in
8:35understanding econometrics and in the
8:38application of these econometric methods
8:41two problems associated with with
8:44identification and so
8:47it means fairly standard now
8:50to see channel data techniques use
8:52firm fixed effects industry fixed
8:54effects year fixed effects
8:57those are useful to a point
8:59and we've got some
9:01omitted variable that is firm specific
9:04and doesn't vary through time
9:07burn fixed effects are taking care of
9:09that
9:10we're good to go if that's the case but
9:12the problem is we don't know that that's
9:14that's the nature of the authentic
9:16variable so if there's some firm
9:17specific
9:19a time varying
9:21variable that is that is the source of
9:23this exogeneity
9:25because panel data techniques aren't
9:27really going to do it for us and again
9:29unless we really know
9:31what the element variable is which we
9:33can't by definition
9:35it's
9:36we're still going to be subject at times
9:38to this this criticism now
9:40what panel data techniques are
9:43are useful for is that they're at least
9:45limiting the scope
9:47of the endogenating
9:48problem because you can't be criticized
9:51for omitting something that is just
9:54some factor it isn't going to bear with
9:56your time with an affirmative you have
9:58control for that so it's a bit of a
10:00solution in that sense but it can never
10:03completely take care of the problem
10:04unless we really know precisely the the
10:07nature of the problem
10:10another possibility is some sort of
10:11regression discontinuity
10:14so here here the idea would be that we
10:18have some sort of threshold event
10:21that when we cross this threshold we can
10:23test where there's a difference in
10:25behavior someone so an example would be
10:28if you're trying to test
10:30whether creditor control has some impact
10:34on
10:35real investment decisions let's say
10:36something like java and robertson in
10:38their paper so you could you can observe
10:41a covenant violation
10:43right and the question is when you cross
10:45this threshold and violate the covenant
10:48i do observe a difference in behavior
10:50now the the nice advantage of the
10:53regression discontinuity technique is
10:55that right around that threshold
10:58you can plausibly say that firms on just
11:00on either side of that threshold are
11:03roughly the same
11:05their characteristics are roughly the
11:07same and if you can make that claim i
11:09think this is a pretty solid way to try
11:12to deal with the adagi criticism the
11:15limitation here is that
11:17you don't come across that many clean
11:20threshold types of events that you can
11:22really use in in the context that we've
11:25studied finance if you can find them
11:27it's great
11:29by all means use them
11:31but
11:34my experiences are difficult to find
11:37and therefore
11:38not often do you come across a
11:41situations where that's really going to
11:42help you out
11:44with your identification problem
11:46there are possibilities to surrender
11:48some sort of matching mile some
11:49propensity score
11:51type of approach
11:53and again that's great if you can if you
11:55can identify
11:56what you think are the most plausible
11:58set of factors so you can match up firms
12:01and do a difference in difference sort
12:03of approach but again almost by
12:05definition these can't be perfect
12:07because we're talking about
12:09correlated omitted variables that we
12:11don't really know what they are you know
12:13what they are sure we could perfectly
12:15match on them and run this type of test
12:18but
12:19not knowing what they are leaves us a
12:21little bit
12:22in limbo uh instrumental variables not
12:25another possibility again it has some
12:28limitations we need some some variable
12:30that's going to satisfy both the
12:32relevance criteria and the exclusion
12:36criteria
12:37relevance is pretty easy to demonstrate
12:39most of the time with with an iv but
12:42it's also it's really easy to criticize
12:44instrumental variables on the on the
12:47exclusion part and so
12:50they're difficult to find good
12:51instruments are really difficult to come
12:53across they've got to be based on an
12:56underlying economics but
12:58you know there's no great test
13:01for that instrument and so you're always
13:03open to this criticism that you just got
13:05that extreme
13:07and so
13:09all these techniques are useful
13:11all of them should be done
13:13in different contexts but
13:15we can never use these techniques and
13:18think okay we solve
13:20the endogenous problem and we don't you
13:22know we don't solve it using accounting
13:24metrics
13:25so so where the profession is kind of
13:28headed more towards is is the fifth one
13:31which is natural experiments
13:34right what's great about natural
13:35experiments is that
13:37you can identify some plausibly
13:40exogenous event that's truly exogenous
13:43and you can do sort of a different dip
13:46so a different different analysis and
13:49and claim causality in a pretty credible
13:52fashion
13:53and because you can do that
13:56and defend it much more easily than some
13:58of the others
13:59i this has become sort of the holy grail
14:02in a sense in terms of doing empirical
14:05corporate finance uh but it it's not
14:08without problems i think this these are
14:10the concerns
14:12that i was alluding to before
14:15i think the way we start the head
14:18is to have this general belief that
14:20every study has got to have some clean
14:22natural experiment or some some perfect
14:25instrument otherwise the studies
14:28just a quick check on ssrn that i did
14:31you know
14:32put in natural experiments in the
14:34abstract papers i get over 1600 entries
14:37of recent papers
14:39uh in the field mostly in corporate
14:42finance where you see that
14:44now
14:45i want to
14:46make sure i'm clear that i'm not at all
14:48criticized in natural experiments but
14:51they're great
14:52if you've got them and they and they fit
14:55the question you're asking but my
14:56concern is sort of twofold is that what
14:59you tend to see
15:01a little bit more often now than
15:03than you saw before is that authors are
15:05kind of starting with the experiment
15:08rather than with question and that's not
15:12quite the way you want to go about doing
15:14research i'll say more about that in in
15:16a minute in a way it it's kind of
15:19similar to
15:20uh
15:21concerns that have come up over the
15:23years as
15:25sources of data have changed you know
15:27back in the dark ages when i was getting
15:30my degree
15:31there wasn't a whole lot of
15:33machine-readable data
15:35now most data is machine-readable
15:40and as a result
15:42you saw
15:43over time or at least my perception
15:45you saw authors start to shy away from
15:48hand collection day and saying they
15:49wouldn't do the study unless they could
15:51identify a really readily available data
15:54source and then even to the point where
15:57they just be looking for data all the
15:59time see the data in good machine
16:01readable form and say well let me figure
16:03out something to do
16:05with this data and that's that's not
16:07really the way you want to conduct your
16:09search you want you want to start
16:11um
16:12with the question
16:14um so it
16:15alluded to sometimes uh as you know are
16:18we reaching the point where we have sort
16:20of
16:21unnatural obsession with natural
16:23experiments i think we've got to have
16:24these things
16:26all the time i that's one concern
16:28that it's sort of changing the way that
16:30people go about doing research a related
16:33concern though which i think is the
16:34bigger one in terms of of
16:36the field or the profession in general
16:38is that
16:40it can be very limiting in terms of the
16:42scope for the questions that you ask
16:45right there are some big questions
16:47in our field
16:49that should be addressed
16:51right when these big questions come up
16:54maybe there isn't a perfect natural
16:57experiment maybe there isn't a perfect
16:59instrumental variable that you can use
17:01and if people think well unless i have
17:03that i can't do this study and those
17:05studies don't get done but
17:07we're giving up a lot if that's the way
17:09we approach research i think in the
17:12sense of
17:13really important questions that we could
17:16shed some light on and just
17:17incrementally move towards a richer
17:20knowledge of somebody's questions so i'm
17:23going to talk about the capital
17:24structure research and i'm going to try
17:26to
17:28highlight a little bit more what i mean
17:30by that
17:31so
17:32so let me compare those concerns
17:35with what i think most of us in the room
17:38um would would say this is how we would
17:40advise phd students as to how to go
17:43about
17:44doing research but first and foremost
17:47say well you want to identify an
17:48interesting question
17:50if students come and talk to me that's
17:52frequently the first thing i'm saying
17:54well what's the question
17:55that you're asking people
17:57and and i've noticed this even in terms
18:00of um
18:02interviews uh
18:03at the meetings fma or fa meetings or
18:06interviewing new candidates and you
18:08asked the candidate to
18:10launch into their steel about about
18:12their research and the first thing they
18:14do is they go straight to their
18:16experiment this is an experiment i'm
18:19running
18:20that's not really what most people want
18:22to hear i don't want to go into what's
18:24what's your question what's the question
18:26you're really trying to get at and and
18:29this is the way i think as i said most
18:30of us would probably advise students to
18:33do research by first
18:35identifying an interesting question
18:38right once you've got that
18:41all right then we should be thinking
18:42about developing hypotheses
18:45from first principles from the ground up
18:48and develop a plausible
18:50testable set of hypotheses
18:53then and only then i think we're going
18:55to reach that point
18:57then you start thinking about your
18:59experimental design well what's the best
19:01way
19:02that i could go about trying to shed
19:04some light on on this particular set of
19:09all hypotheses
19:09and this this is the part you know just
19:11to reiterate i said a couple of minutes
19:12later like this is the part that at
19:15times looks like it's getting flipped
19:17and professional so too many studies
19:21are starting with dick spencer
19:24and then
19:25and then moving on
19:27uh from there right so once we identify
19:30this experimental design that we think
19:32is the best for shedding some light on
19:34you know then then we eventually get
19:36some results we've got to let the data
19:39speak but we've got to let the data
19:40speak in in a way that we're well aware
19:43of what the limitations of what we're
19:45doing
19:46and this is
19:48i guess this is more of a criticism of
19:51referees than it is of authors at this
19:53point is that you
19:55you tend to see a lot of referees
19:58who are unwilling to accept limitations
20:01and say well if you don't have the
20:03perfect identification strategy i'm
20:05going to reject your paper that's sort
20:07of leading people to think well i have a
20:09perfect identification strategy
20:11otherwise this paper is not worth doing
20:14i'm not sure that's the best
20:16develop for our profession because there
20:18are a lot of things that we can learn
20:21from studies that are imperfect
20:23because let's be honest every study
20:26is imperfect
20:28in some way
20:29and if the particular way is you can't
20:32necessarily identify causation in your
20:35study it doesn't necessarily mean that
20:38the study isn't worth doing but it does
20:40mean that if you do the study
20:42and you fail to note those limitations
20:45you probably deserve to be rejected at
20:47this point
20:48you can you can provide evidence
20:51you can interpret that evidence you can
20:53understand that maybe you can't
20:56imply causation from that set of
20:58evidence if all that is kind of laid out
21:01for the reader and the reader can decide
21:03in the end well given all these
21:05limitations do i feel like i learned
21:07something from reading the study and if
21:09the answer is yes and
21:11you learned enough
21:13then the journal should be wrong to
21:15accept that paper and they do
21:17i think they do and that's
21:19uh that's important
21:21um
21:22let me just back up a second there's one
21:25one more
21:26uh issue that i wanted to note on this
21:28particular slide
21:31using the example of business
21:33combination loss now
21:35so so this is a natural experiment i
21:37think that has had a lot of success
21:39in the profession and so let me be clear
21:41up front i'm not criticizing this this
21:44natural experiment it's a good one and
21:46the way it's useful
21:48uh is is in the sense that uh
21:51if you think about types of studies in
21:53corporate finance one big class of
21:55studies is asking whether governance has
21:59some causal impact on
22:01value or
22:03managerial actions and you know these
22:06are interesting and big questions
22:08i think but they're tough to get at uh
22:10from an identification
22:12standpoint so
22:14a clever natural experiment
22:17is the introduction of these of these
22:19business combination laws and
22:21these laws are
22:22uh are sort of staged over time in
22:26different states so it makes for a
22:27really nice
22:29exogenous shock to governance that we
22:31can then connect with some sort of
22:33outcome there
22:35it's got a lot of the attributes
22:37that you're looking for
22:39in the actual experiment now it's also
22:41got
22:42it also raises some of the concerns that
22:44you'll always see in natural experiments
22:46as well because none of these what we
22:49call natural experiments and social
22:51science are never really pure natural
22:54experiments because somebody had to
22:56decide
22:57that we're going to change these laws
22:59the laws are not purely exogenous
23:01they're endogenous in some way
23:03and maybe what's driving the changing
23:06laws is correlated with your outcome
23:08variable but these authors who have done
23:11these studies with business combination
23:12laws have been well aware of those
23:14limitations and made a plausible case
23:17that
23:18uh they're they're not interfering with
23:20the
23:21interpretation of
23:23causation right so so it's a pretty nice
23:26clean natural experiment um but it's
23:29probably one that i've seen repeated in
23:32most times of any sort of natural
23:35experiment it's been repeated in some
23:37good ways
23:38right but you start it starts to run out
23:40of juice after a while because you know
23:43you see some some big questions like
23:45does governance affect values governance
23:47affect investment does governance effect
23:50innovation these are all
23:52very useful applications of this natural
23:54experiment but but what seems to happen
23:56in our profession
23:58is that everyone says well you need a
24:01clean natural experiment
24:03i can see from these publications that
24:05the profession seems to accept this one
24:08as that's a useful
24:10natural experiment so what can i do with
24:12this one again you're starting with the
24:14experiment then
24:16rather than the question and we start to
24:18get into some really marginal questions
24:20at this point you know you just keep
24:22following this path you eventually start
24:24seeing studies that are like you know
24:26what's what's the impact of
24:29governance or true business combination
24:31laws and what the ceo had for lunch on
24:33tuesday well
24:35yeah i guess you can establish causation
24:37with this natural experiment but now
24:39you're asking the question nobody cares
24:41about it
24:42so you have to worry about that a lot
24:44and you run into this problem precisely
24:46because
24:48you're starting with the experiment that
24:50you think will pass muster with journals
24:52instead of
24:53starting with uh with what is the
24:56interesting and
24:57big question
24:58all right so now let me
25:00uh talk a bit about capital structure
25:03research as an example of some of these
25:05these themes that i'm talking about so
25:09um there's nothing particularly unique
25:12about capital structure research with
25:14respect to these identification problems
25:16i'm choosing simply because i've been
25:18doing papers in this area and so i know
25:20this area a little bit more and you'll
25:22see in some of the examples i'm using in
25:24my own papers
25:26it's not because i think these are the
25:28best papers in that field but i know
25:30these papers better and then they're
25:32representative of some of the points
25:34that that i'm trying to make but there's
25:37some really fundamental questions in
25:40capital structure research that
25:42amazingly enough after all these years
25:44we haven't done a great job of answering
25:47that
25:48we don't really have
25:50a great idea of what are the primary
25:53determinants of capital structure
25:55decisions we have some pretty good
25:57theory
25:58about what certain market frictions
26:00often matter
26:02taxes bankruptcy costs
26:04agency costs
26:06asymmetric information
26:08but it's it's difficult it's it's proven
26:11difficult to establish what factors are
26:14really first order determinants of our
26:16capital structure decisions and part of
26:18the issue has always been
26:20proper identification now another whole
26:22segment of this literature is just sort
26:24of getting out of static models
26:26versus dynamic models i'm not really
26:28going to talk that much about that
26:31it's directly relevant to
26:33the topics today but that's also a
26:35challenging aspect
26:37in this literature
26:40but what you see
26:41i mean what i want to talk about in the
26:43rest of the time are a variety
26:46of different sort of research approaches
26:49in this field that i would argue are all
26:51very useful and they all
26:54give us some information in different
26:56aspects of the set of information that
27:00we need to ultimately draw some
27:02conclusions about about capital
27:04structure
27:05and only one of them
27:07falls in this category of natural
27:10experiments
27:11what's important i think to to note is
27:14that even though all these are
27:16potentially useful
27:18they all have different limitations too
27:21so you're getting different pieces of
27:23information from them so
27:25you you can't view them as as
27:28being able to draw the same sort of
27:30inferences in terms of causation
27:33but you can
27:34do them sometimes as telling you a
27:36little bit more about what's first order
27:38and what's not first order and i'll i'll
27:40be a little bit more precise about that
27:42in a couple of minutes so
27:44so the classic approach of course is
27:46just doing all this type of regression
27:48you've got leverage on the left hand
27:50side
27:51you've got a set of your testable
27:53determinants on the right hand side and
27:55other controlled variables you might
27:57also
27:58estimate this using an even panel
28:00regression techniques uh to try to
28:03control for basic facts of some sort
28:06and literature's got tons of studies
28:09that that fit into this
28:11category but there's three gonna be two
28:13main problems i mean one is
28:15is the identification problem that we're
28:17we're primarily focused on but as a
28:20practical matter too the second problem
28:22has been that it hasn't proven that
28:24approval you know we
28:26end up identifying
28:28some relatively vague factors that seem
28:31to matter you can say that leverage is
28:33associated with profitability
28:36growth opportunities as measured by
28:38market book ratios usually
28:40firm size
28:42that's great in a way but when we think
28:44about relating that to capital structure
28:47theories we find ourselves in this trap
28:50where it's consistent with multiple
28:52theories
28:53uh even multiple classes of fears
28:56between static mode trade-off models and
28:58pecking order models
29:00they can be reconciled
29:02with this this set of observed
29:04determinants and so
29:06in the end
29:07we feel like we haven't learned that
29:09much at times with with this literature
29:12right so
29:13the way literature has progressed i
29:15think and this is this is a useful
29:17progression is is to think about
29:19different types of empirical approaches
29:22that can shed some additional light now
29:24one one
29:25type would be the natural experiment
29:28and can we can we observe some exogenous
29:31shock
29:32to one of these hypothesized
29:33determinants
29:35while the others become constant
29:37and see whether that shock seems to
29:39elicit a reaction confirms an inner
29:42choice of capital structure
29:44that's one way
29:45to shed something
29:47on a specific fact
29:50another possibility
29:52is to conduct what's really more of
29:54descriptive data analysis type of
29:57studies now
29:59these next two descriptive data and
30:01longitudinal types of studies are kind
30:03of similar
30:04in the sense that
30:06you would probably characterize them
30:08both but one certainly by the name but
30:10both of them are sort of descriptive
30:12work
30:14and by that i mean they're not generally
30:18set up as being specific hypotheses that
30:21are tested into a traditional scientific
30:24way
30:25and and sometimes these kind of studies
30:27get a bad name you see people use
30:29descriptive
30:31as a pejorative term that studies it's
30:34descriptive
30:35but i would argue
30:37that this has a place
30:40in the set of research approaches
30:42uh and in particular in capital
30:45structure i think it's been useful
30:47i'll give you a couple of examples of
30:49how they have been useful longitudinal
30:52studies are
30:53are similar in the sense if we look over
30:56a long period of time
30:58whether capital structure has changed
31:00over that period of time or whether we
31:02can somehow connect that
31:04with changes in either firm or macro
31:06economic characteristics in a way that
31:09is consistent or inconsistent with with
31:12what we think of as as the theory
31:13underlying this so
31:15the longitudinal models are a little bit
31:18like
31:19natural experiments and that you you can
31:21now go over a long period of time shocks
31:23to multiple
31:25possible determinants of capital
31:27structure in a sense think of it as sort
31:29of a horse race
31:30among these different factors as to what
31:32seems to be first order and what's not
31:36another approach might be structural
31:38models
31:39right where we really try to get into
31:41the dynamics of
31:44capital structure policy
31:46over time can we specify some sort of
31:49objective function that the manager is
31:51maximizing specify a set of exogenous
31:55determinants as well as a set of
31:57endogenous variables that would be
31:59connected with
32:00uh with leverage and so in a sense
32:02specify the nature
32:04of the endogeneity problem and then take
32:07this to the data through some structural
32:09estimation to see whether the data seem
32:11consistent with that
32:13the predictions from that structural
32:15model lots of good examples of that
32:18in the recent literature as well
32:21and last and certainly not least is that
32:23i think there's room
32:25in this literature for clinical or case
32:28study sort of approaches now it's
32:30difficult i think to do a stand-alone
32:34clinical study
32:36years ago the jfp published lots of
32:38those that sort of backed away
32:40from that over time but
32:42but i think clinical studies are very
32:45useful in conjunction with with some of
32:48the others and i'll give you some
32:49examples of that how you might combine
32:52sort of a descriptive data analysis or
32:54longitudinal study with with a clinical
32:57analysis of the subset of your data in a
32:59way
33:00that sheds a lot of light on what's on
33:02what's really going on it can really
33:04help you tease out the identification
33:07issues as well
33:09and so so let me give you
33:12a few different examples
33:14from the literature that highlight the
33:16use of these different approaches so i
33:19want to ultimately make two points one
33:21is that each of them
33:22are useful in shedding light on an
33:25aspect of the capital structure problem
33:29and then secondly i make sure we
33:31understand that each of them also has
33:33limitations
33:34so you have to be careful when you do
33:36these studies is to understand what you
33:38can and cannot conclude from the type of
33:41study that you've chosen to do
33:44so one approach as we mentioned is a
33:46natural experiment
33:48a good example that
33:50are these studies that have looked at
33:53staggered changes and income tax rates
33:56either across states within the u.s like
33:59the hydrogen lung fist paper or
34:01across different countries throughout
34:03the world like fatio and schum have done
34:07so it's sort of a classic natural
34:10experiment diff and diff kind of
34:12approach where you've got the change in
34:14in data changing leverage on the left
34:16hand side
34:18as a function of the change in the tax
34:20rate that's t variable set of
34:23firm specific variables and denoting
34:26with x and i's
34:30and
34:31industry specific variables as well
34:34right so you're trying to control for
34:35all these other possible determinants of
34:38leverage you're seeing shock
34:40to tax rates
34:42at that point in time
34:44do you see firms respond
34:47with changes in language right now
34:50as with most natural experiments as i
34:52mentioned before they're not pure
34:54natural experiments so there's there's
34:56still a little bit of endogenous that
34:59you worry about in this and so you might
35:01worry that somehow there's still some
35:03systematic differences that you haven't
35:05control for
35:07between
35:08firms say within states that have the
35:10tax rate change versus those that didn't
35:12so you'll see clever clever ways to try
35:16to get at that by looking right around
35:18the borders between states look at
35:20counties that are right along the border
35:22with the idea being that the economic
35:24conditions of companies on either side
35:27of the border got to be roughly
35:29identical or orders of countries in case
35:32of a focus
35:35and so you so you do your best
35:38to hold constant all the other possible
35:41economic determinants of this leverage
35:43but one group has to shock tax rates the
35:46other group does not have that shock to
35:48tax rates do we see a difference
35:50in leverage as a result of that of that
35:53shock to library so
35:55if done
35:57in in a
35:58in a correct manner this type of study
36:01is really useful
36:02for getting at this causal connection
36:04between this factor taxes
36:07and capital structure change
36:11what it can do
36:13is tell you something about whether
36:15taxes are really a first order
36:16consideration or not because
36:19you by definition if you've done study
36:21right you held all these other factors
36:24constant
36:25and you isolate in on the tax rate now
36:28again that's not a criticism of this
36:29type of study
36:31i it's just a
36:33a point that in doing this sort of study
36:36you've got to be clear on what it is
36:38your goal is
36:39if you're going into the study saying
36:41well my question is i want to
36:43identify what determines capital
36:45structure this isn't really doing that
36:50if your question though is do taxes have
36:53an impact on capital structure at the
36:55margin yeah this is the way you want to
36:58do it here you're able to isolate in on
37:01the causal impact of taxes on leverage
37:04decisions it's very usefulness
37:07but it's a subset of the information
37:09that we're trying to put together in
37:11order to get a bigger picture of what's
37:14really happening with
37:16corporate capital structure decisions
37:17all right so that that's one
37:20one type of stuff all right so the other
37:23one of the other types of studies i
37:24talked about is descriptive data
37:26analysis
37:27now
37:29in some respects this is
37:31a tricky way to go about
37:34doing empirical research because
37:36as i said before you're not necessarily
37:38starting with very firm
37:40hypotheses but rather it's a little bit
37:42more exploratory in nature so
37:45the goal ultimately in a study like this
37:48is to provide
37:50enough information to the reader that
37:52they that they somehow will think
37:54differently about capital structure than
37:56they did before
37:57and if they don't feel that way
38:00then you don't have anything you just
38:02described a bunch of data that everyone
38:04shrugs their shoulders and says well so
38:06what
38:07cares
38:09so i'm going to give you a couple of
38:11examples of this type of work one of
38:13which is my own was stephen kian who was
38:15in the rfs a couple years ago and so
38:18here the approach we took in this in
38:20this study
38:21is to say let's get away from
38:25thinking about it in being sort of a
38:27standard regression framework where
38:28we're looking at a bunch of right-hand
38:30side variables
38:31determining what's on the left-hand side
38:33which is leverage that's that's instead
38:36let's start with the left-hand side
38:39all right let's let's try to observe
38:42some major discontinuity in the
38:44financing behavior of the firm and then
38:46back out
38:48what seems to be driving that that
38:50change in financing behavior so the way
38:52we went about doing this is to say okay
38:54let's let's first identify firms
38:57have what we call proactive change in
39:00leverage and by proactive we simply mean
39:02that they actually issued debt
39:05in order to change their leverage as
39:07opposed to they bought back shares or
39:09had some other action it just ended up
39:12causing their leverage to be different
39:13we want them to be proactively issuing
39:15debt in a way that changes their
39:17leverage ratio and the second
39:20aspect of that was that we were
39:22requiring that the resulting leverage
39:24ratio
39:26was at least 10 percent above
39:28some estimate of their target where the
39:30target is being estimated using sort of
39:33standard empirical models of
39:36capital structure
39:38right so we put in the second
39:40requirement because
39:42we thought well this is this is sort of
39:44a unique and challenging situation for
39:47the literature in the sense that we've
39:49now got firms that are deliberately
39:52taking action that pushes them well away
39:55from what we think of as their target
39:57refrigeration
39:59right and so we're just asking two very
40:01simple questions subsequently why did
40:04they do it
40:06what looks like is the underlying
40:07motivation
40:09for undertaking this stuff exchanged and
40:11then secondly
40:12how do we see the leverage ratio evolve
40:15sub subsequently
40:16does that tell us something about
40:18whether they they seem to be uh
40:20targeting a particular
40:24right so
40:25so what do we find so first of all what
40:28we tend to observe is that they seem to
40:30be doing this
40:32uh for reasons of financing
40:35specific investment sometimes it's
40:37capital expenditures or r d sometimes
40:40it's more
40:41working capital considerations right but
40:43these don't appear
40:45that they're borrowing
40:47in order to say for example buyback
40:49shares
40:50it also doesn't appear
40:52that they're borrowing so that they can
40:54move to some new target leverage ratio
40:57right but also the second main result is
40:59that when you look at what happens after
41:01this initial jump in leverage
41:04right it looks like they're starting to
41:07move back towards what we think is the
41:09target reference ratio
41:11but they're in no hurry at all to do it
41:14it's a really slow
41:16adjustment back
41:19towards the target and they even though
41:21they have opportunities to be a little
41:23bit more proactive about it and they
41:26could have moved back to the target
41:28quicker in some cases they don't
41:30you sort of see it drifting back down
41:33and what's kind of interesting too is
41:34that when we see subsequent situations
41:37where they seem like they have a need
41:38for funds they go out and borrow again
41:41even though they're still well above
41:42their target they get even further above
41:45their target later on
41:47right so
41:49what does all that mean how is how is
41:51this
41:52then a useful study in capital circles
41:54some of you might think it isn't useful
41:56to study in the catholic church why do
41:58we think it's it's a useful
42:01study to do well first of all if you
42:04think about it
42:05what this seems to imply that
42:09this managing towards some sort of
42:11stationary or static target leverage
42:13ratio doesn't seem to be a first order
42:15concern for managers they don't seem to
42:17be behaving
42:20well that's kind of interesting in the
42:22sense of
42:23what we think of in terms of capital
42:25structure now many of them are sort of
42:27stationary static target sorts of models
42:31and this is kind of saying maybe we
42:33shouldn't be digging along those those
42:35lines in terms of theoretical capital
42:37structures now the second aspect of it
42:40is that it seems to be pointing
42:43towards the financial investment needs
42:45and then the subsequent evolution of
42:47cash flows
42:48as being more first-order drivers of the
42:51dynamics of leverage through time
42:56now let's be clear that you know when
42:58you do a study like this
43:01we in no way can claim some causation
43:04and anything that i just described to
43:06you and we have to be really careful as
43:09we're as we're laying out these results
43:11that we're not at all making any such
43:13claims
43:14about causation right but why might it
43:17still be useful in a sense what we're
43:19trying to do is we're sort of
43:22reducing the the scope of the models
43:24that are really plausible models for how
43:27firms really behave with respect to
43:29capital structure and if we can do that
43:32and if we can do that sufficiently
43:36then we feel like well that's a
43:37contribution that is useful to the
43:39capital structure literature and
43:41unfortunately for us of referendum
43:44agreed that
43:45was useful enough
43:47now
43:48a similar sort of example
43:51is this reason they provide
43:57it's forthcoming
43:59in general finance so
44:01what they do at a very basic level is is
44:04fairly simple they're really just
44:06describing
44:08the variation in leverage ratios
44:11within individual firms
44:13through time now
44:15why should you do that why would you
44:17start a study like that
44:21if you think you're going to contribute
44:22to the capital structure of literature
44:24well the idea
44:26is that
44:28most of the models and most of our most
44:30of the thinking i think within the
44:31profession through time has been that
44:34capital structure within firms is
44:37reasonably stable
44:38through time
44:40right where you really get a lot of
44:41variations in the cross section
44:44right and that's kind of reflected
44:46in the type of work that you see in that
44:48structure most of the models and the
44:51most empirical work are designed to try
44:53to explain that cross-section
44:56why the cross-sectional variation well
44:58that's not necessarily a good approach
45:01if what's really going on is that a lot
45:03of variation is happening within the
45:05firm over time if that's just as
45:07important as the cross-section then
45:09we've sort of got to rethink
45:11uh what our models are really saying how
45:13we're approaching the empirical works
45:15that's sort of where the where the study
45:17is tell you from the outset again the
45:20idea is it's descriptive data analysis
45:23let's try and describe the data in a way
45:25it's going to cause people to rethink
45:28how they're approaching capital
45:30structure and if they're something
45:31that's unique enough
45:33that does cause people to do this
45:35rethinking of
45:37how capital structure is determined this
45:40is a useful way to to approach the
45:42problem right so what do they find well
45:43first of all they do fine find
45:45substantial
45:47instability in the leverage ratios of
45:50individual firms
45:52through time
45:53i have a lot of variability surprisingly
45:56i think to a lot of people a ton of
45:58variability in individual firms they do
46:01find episodes
46:03in which firms look like they have
46:05relatively stable leverage ratios but
46:08those are pretty limited
46:10to time periods in which leverage
46:13is really quite low oftentimes zero
46:16so you see a lot of stability and some
46:18firms have no debt and they keep no debt
46:20for a while or really load that they
46:22keep that for a while
46:24and that's when you tend to see
46:25stability when leverage is higher it
46:27tends to be bouncing all over the place
46:29now why is the bounce all over the place
46:31well i find like like we did the
46:33patriots steve mckean and i
46:35they find it seems to be connected with
46:38episodes in which there are strong
46:40investment needs right and this is where
46:42they sort of
46:43they do a really nice job
46:46of combining some clinical or case-based
46:48evidence to complement
46:50what they're doing on a large-scale
46:52basis in terms
46:54the of documenting
46:55think they dig into the data
46:58and identify well what seems to be the
47:01underlying cause for these individual
47:03firms for
47:04for increasing their leverage when they
47:06see it increasing or decreasing it go
47:09down and they find pretty strong
47:11evidence that it seems to be associated
47:13first with the company expansions
47:16and due to changing economic
47:18circumstances for those firms and then
47:20subsequently
47:22a reduction in leverage in those time
47:24periods in which it it's actually
47:27uh either excess cash flows that get
47:29produced even though they're actually
47:31retransitioned
47:33right so again i think this is
47:35very useful type of evidence in this
47:37approach even though it looks like a
47:38very simple
47:40data description approach at the outset
47:43it's carefully done in a way that
47:45ultimately when you're done with it
47:46you're thinking well i gotta rethink the
47:48way i was thinking about capital
47:50structure as a result of this so it's
47:52very informative to subsequent theory
47:55because now any theory that comes along
47:58that purports to be
48:00modeling how capital structure is
48:02actually chosen has to deal with the
48:04sort of facts that are out there to
48:06stylize facts that let's say sufferings
48:09actually
48:14okay that's sort of uh that's sort of
48:16what i'm summarizing here's an
48:18incredible
48:19theories of capital structure that have
48:21to deal with this time series variation
48:24and not just focus on cross-section
48:26anymore
48:27it also is implying that
48:30sort of contrary to probably what most
48:32of us teach in the classroom with
48:34respect to capital structure that the
48:37leverage per se is really kind of a
48:39second order importance
48:41to valuation
48:42the firm's behavior is that they don't
48:44really care that much
48:46what their leverage ratio is at any
48:48point in time
48:50if you're like me
48:51you're teaching in the classroom you're
48:52oftentimes teaching sort of a
48:55static trade-off sort of model where
48:57you're optimizing at this point
48:59that
49:00is the capital structure that maximizes
49:02the value this kind of implies that
49:04either that that nice uh
49:07inverted u-shaped curve that we draw up
49:09on on the whiteboard
49:11for capital structure is either really
49:13flat
49:15right or it's total nonsense
49:17one of the two
49:19um
49:20and there again obviously then what
49:22seems to be true then is that the main
49:24determinants of leverage ratios
49:27must be factors that are a little
49:28different than what we traditionally
49:30think of as
49:31standard measures of taxes or distress
49:34costs that are traded off
49:38okay so let's move on to another another
49:40different type of study so those last
49:43two are kind of descriptive data
49:45analyses
49:46that you know hopefully i've convinced
49:48you that they shed some useful life
49:51on the capital structure puzzle another
49:53way to go about this
49:55uh is to do a longitudinal type of study
49:59again the
50:00the potential merits of a longitudinal
50:02study is that you've got this long
50:04period of time where lots of things
50:06might be changing lots of the
50:08underlying factors that could determine
50:10capital structure can change over a long
50:13period of time we can see whether
50:15capital structures tend to change in a
50:17direction that we think they should
50:19based on those those changes in uh
50:22underlying fundamentals so this study by
50:24graham leary roberts was coming out
50:27in the jfd is a good example
50:30they're really looking at
50:32leverage ratios over nearly a century
50:35within the u.s so you've got a period of
50:37time in spanning
50:39world war ii
50:41the
50:43macroeconomic expansion that takes place
50:46after world war ii in the u.s high
50:48inflation period in the 70s in the u.s
50:51the most recent financial crisis all
50:53this
50:54gets spanned
50:55uh in their study and they can ask then
50:58some fairly simple questions you know
51:00first have capital structures change so
51:03on average over that period of time
51:06secondly
51:07can our existing models
51:09account for those changes are they
51:11consistent with those those changes
51:14and then third if not the answer does
51:16seem to be no in this case
51:19what underlying forces seem to be
51:21driving this this variation so
51:23so again when you when you
51:25approach the question
51:27in this way the the result that you're
51:29going to get
51:31um you know
51:32maybe you find that everyone's
51:33consistent with what we thought you
51:35usually don't find that kind of thing in
51:37reality so if you don't
51:39i your your end result is to present a
51:42set of facts that are a challenge to the
51:45existing theory they're going to inform
51:47theory it's going to be useful in that
51:49sense right it's also going to
51:52narrow the set of of theories that are
51:54truly credible
51:56any any theory that reports explain
51:58capital structure
52:00is going to have to account for this
52:01this time series variation leverage
52:04ratios it doesn't seem to be accounted
52:06for by our standard factors
52:09so what do they do right
52:11what they find is first of all there's
52:13been a fairly large increase in leverage
52:16over time if you go back to sort of the
52:18post
52:19world war one period and look at average
52:22leverage leverage ratios back then
52:25i think i'm trying to remember the
52:27magnitude but they're nearly doubled now
52:29what they were
52:30back then right
52:32coincident with that
52:34you see a fairly large decline in cash
52:37holders
52:38the phone size for this separate
52:40increase in leverage now that may be
52:42surprising to some because a lot of
52:44what you read in the press and even some
52:46of
52:47our studies in corporate finance in
52:48recent years talk a lot about
52:51these huge cash balances that us firms
52:54have now
52:55that's a more recent phenomenon and
52:57these cash balances actually were
52:59enormous back in the early part of the
53:02of the last century
53:04and they declined quite a bit over a
53:06period of time with these leverage
53:08ratios that
53:09have gone up it's only recently that
53:12they've served cash balances have turned
53:14back up
53:15right what's interesting though is that
53:18firm characteristics that we normally
53:20associate
53:21with capital structure
53:23on average don't really change
53:26significantly in a way that's consistent
53:28with
53:29these firms having greater debt capacity
53:32so think of our traditional determinants
53:35of leverage they don't account for the
53:37fact that leverage ratios are so much
53:39higher now
53:40than they were
53:41you know back in 1920 or so
53:44all right what's also interesting i
53:46think is that they do find
53:48a negative association between corporate
53:51borrowing
53:53and government borrowing
53:54just
53:55u
53:57hints
53:58at the possibility that there's there's
54:00there's some crowding out that's taking
54:02place the government is borrowing a lot
54:05that's crowding out of the corporate
54:07sector from from borrowing it has
54:09reasonable rates
54:11so again once once they're done with
54:14this have they established any causation
54:16between variables and leverage no
54:20they're not claiming any such causation
54:22even with their last result with the
54:24government crowding out
54:26the most they're saying is that you know
54:28there's some hints of that in the data
54:30all right which i think is useful in
54:32terms of subsequent studies that that
54:34could potentially examine that question
54:36in more detail so
54:39these aren't this is again not a study
54:41that has
54:43great identification of anything
54:46but i would argue is very useful to us
54:48as scholars who are interested in
54:51capital structure because it's
54:53it's again narrowing the range of
54:55plausible explanations for what's going
54:58on you finish this paper you learn
55:00something
55:01that you didn't already know
55:03about capital structure
55:05that's the definition of a contribution
55:10now the last study i'll talk about is
55:11again
55:12one of my own and i apologize for the
55:15answer for
55:16talking about another study of mine
55:18again i'm not representing these as the
55:20best studies i'm just
55:22representing them as as representative
55:24studies of a particular approaches so
55:26this is this is another longitudinal
55:29type of study that i've been working on
55:31with co-authors lance barr's
55:33kent lane
55:35in which we're studying capital
55:36structure decisions of u.s firms over a
55:39period of time actually predates the
55:41grandeur in robert's paper our our
55:44period of time is 1905
55:46to nineteen twenty four now why
55:49study capital structure back in
55:51the dark ages like that especially if
55:54if you've ever looked at data back then
55:57it's a little challenging to
55:59uh to look at financial data for a
56:01period that predates the sec the
56:04standard
56:05standardization of data is not
56:07particularly good
56:09back at that time but
56:11we thought well this is a very useful
56:13time period to take a look at for
56:15capital structure theories for two
56:17reasons one is that it stands the period
56:20of time
56:21in which the u.s first introduced
56:23corporate and personal income tax rates
56:26so prior to 1909 i believe it was there
56:29were no corporate income taxes
56:31there were no personal income taxes
56:33either in the u.s
56:35great days but
56:37not anything many of us have experienced
56:40so if taxes are really important then we
56:42ought to see it when taxes are
56:44introduced so sort of the logic of these
56:47at the same time
56:49the spans this period of time
56:51in which the us enters world war one
56:55which we show in the paper is associated
56:57with very large
56:59but short-term shock to investment
57:01opportunities for the u.s firms
57:04right now what's what's
57:06neat we think from a from an empirical
57:08standpoint is that that shock to
57:10investment opportunities
57:12is plausibly quite exogenous no one's
57:14really anticipating the outbreak of
57:17world war one
57:19and so we can see how firms react
57:23to this shock to investment
57:24opportunities at the same time as it
57:27were observing how they react to
57:31fairly large changes in personal and
57:33corporate income tax rates because one
57:35of the other things i won't get into the
57:37details but one of the other things that
57:39you see when the war breaks out
57:41is that the u.s imposes very large
57:44excise taxes on
57:45on the firms particularly the firms that
57:47would potentially benefit from from the
57:50outbreak of the war so there's this very
57:52large change
57:54in marginal tax rates it actually gives
57:56a strong incentive to equity financing
57:59during the war so we're able in some
58:01sense to have a bit of a horse race
58:03going on between tax effects and
58:06dynamic investment effects right so what
58:08we find
58:10in the paper is that there's really very
58:12little evidence that
58:14shocks to tax rates have much of an
58:15impact at all
58:17on the leverage ratios
58:19you see these corporate
58:21income taxes initially imposed basically
58:24nothing happens to the average leverage
58:27ratio
58:28right but what we do see which is quite
58:30similar actually to the d'angelo enroll
58:33paper is that even though average
58:35leverage ratios aren't changing very
58:37much
58:38uh individual firm leverage ratios are
58:41bouncing all over the place during this
58:44period of time and we're able to link it
58:46pretty tightly
58:47with the evolution of investment
58:49opportunities and and cash flows in
58:52these firms over this over this period
58:54of time
58:55right so
58:56so our interpretation is kind of similar
58:59to
58:59the d'angelo role argument that you know
59:02it looks like the dynamics of investment
59:04opportunities and cash flows are what's
59:06really driving the leverage ratios
59:09rather than
59:10some of the
59:11traditional determinants of
59:14of leverage right but
59:16you know in trying to interpret this
59:19there are challenges some of which are
59:21closely related to the identification
59:23challenges that that we've been talking
59:25about one is that when you look at
59:28you're doing any sort of longitudinal
59:29study if you're going to focus in on
59:32specific shocks that interest you you
59:34have to be aware of the fact that
59:36they're not the only shocks that are
59:37going on
59:38during this period of time so in our
59:41case there's sort of a a market panic
59:44that takes place in in 1907
59:47the federal reserve
59:49is created in 1913
59:53and after the end of world war one
59:55there's actually a pretty
59:57strong recession that borders on on
1:00:00depression
1:00:02during that period of time all of which
1:00:05could potentially have some impact on
1:00:07observed leverage ratio so even though
1:00:09we can find some evidence that's
1:00:11consistent
1:00:13uh with this investment
1:00:15shock associated with world war one
1:00:17being a causal determinant of the
1:00:20leverage changes
1:00:21we can't necessarily point to that
1:00:23unless we do something more so so where
1:00:26we are in the study now is sort of
1:00:28trying to supplement what we've got with
1:00:30a little bit more of a clinical analysis
1:00:32of individual companies and that's
1:00:35involved actually collecting
1:00:37the annual reports from around 1917 for
1:00:41these 60 or so firms that comprise our
1:00:44sample so
1:00:46as you might expect they're they're not
1:00:48in most libraries so you know you got to
1:00:51go
1:00:51in our case and go to the columbia
1:00:53university library
1:00:55get these all copied and you have this
1:00:57huge stack of annual reports that we've
1:01:00got to go through but it's a very useful
1:01:02way
1:01:03to try to tease out what's really
1:01:05driving these changes in leverage
1:01:07because now when you read these annual
1:01:09reports the narrative that's in the
1:01:11annual reports
1:01:12right you can actually see the firms are
1:01:15saying specifically
1:01:17you know we're borrowing in order to do
1:01:19this
1:01:20and that way you are getting directly at
1:01:24the issue of causation the data by by
1:01:27using this supplementary clinical type
1:01:29of of evidence you don't have a clean
1:01:32natural experiment you don't have
1:01:34instrumental variables right but looking
1:01:37in the clinical sense you are getting
1:01:38some some of the same issues right in
1:01:41fact i think i would encourage you to
1:01:42think about
1:01:44uh sort of clinical sort of analysis in
1:01:47in the same way that you would think of
1:01:48as a natural experiment both of them are
1:01:52identifying sort of a narrow
1:01:54situation in which you can fairly
1:01:57cleanly identify what's really going on
1:02:00the natural experiment does it one way
1:02:03the clinical study does it it doesn't
1:02:04another structural model
1:02:07does it yet another way there's very
1:02:08specific set of conditions this but this
1:02:11is what happens
1:02:12right all these are different ways of
1:02:15of obtaining the same beast
1:02:17in a sense
1:02:19all right so let me let me just kind of
1:02:20conclude at this point you'll open it up
1:02:22to the questions
1:02:24the points i want to make
1:02:25first of all that
1:02:27these identification problems aren't
1:02:29going away
1:02:30they're pervasive in corporate finance
1:02:33researchers
1:02:35much more so in corporate finance
1:02:37research than in some other aspects of
1:02:40finance research they're there they're
1:02:42real problems they're things we have to
1:02:44deal with as scholars in order to be
1:02:46able to tease out some sort of
1:02:49believable conclusion
1:02:51econometrics can help and has
1:02:54really helped i think we've got great
1:02:55advances
1:02:57in our understanding of econometrics
1:02:59like i said over the last 10 to 15 years
1:03:01and they really help us
1:03:03narrow down the scope of this problem
1:03:05they can never solve the problem our
1:03:07metrics doesn't make the indonesian
1:03:09problem go away
1:03:11it just sort of narrows the scope
1:03:14natural experiments are in some ways the
1:03:16most useful econometric solution
1:03:19um but but i worry that an over-reliance
1:03:23on these natural experiments is it can
1:03:26be harmful
1:03:27in our pursuit of knowledge in the sense
1:03:29that i've described before that that
1:03:30really kind of
1:03:32if you think you have to have a natural
1:03:34experiment you are constraining yourself
1:03:36from the types of questions that you can
1:03:38really ask and if the profession as a
1:03:40whole were to do that
1:03:42then we're constraining the body of
1:03:43knowledge that that we're producing
1:03:46about questions in
1:03:48in corporate finance so
1:03:50i'd like to encourage you to think about
1:03:52these alternative methodologies that
1:03:54that we went through as as equally
1:03:56useful useful ways of expanding that
1:04:00body of knowledge
1:04:02even if they're not able to really
1:04:05provide complete identification and they
1:04:07frequently don't
1:04:09but i would argue that's okay as long as
1:04:11you understand
1:04:13you can explain the limitations of of
1:04:15what you're doing
1:04:16you can still produce papers that are
1:04:19very useful in terms of our
1:04:21understanding of the problems
1:04:23so let me stop there and open up any
1:04:26questions
1:04:29let me correct that music is the
1:04:31simultaneous model but that doesn't mean
1:04:35there isn't some other omitted variables
1:04:50okay well thanks for listening

About this transcript

This page contains the full transcript of YouTube transcript (DKWVDvk1QYo) , generated from the public captions YouTube serves with the video. The transcript has 9,948 words across 1,766 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.