WEBVTT

00:00:00.400 --> 00:00:05.700
The path from cause to effect
is dark and dangerous,

00:00:05.700 --> 00:00:08.900
but the weapons
of econometrics are strong,

00:00:09.700 --> 00:00:13.700
wield differences-in-differences
when witnessing parallel trends.

00:00:14.450 --> 00:00:17.000
♪ [music] ♪

00:00:20.100 --> 00:00:21.423
Masters of metrics

00:00:21.423 --> 00:00:24.800
look for convincing
ceteris paribus comparisons.

00:00:25.100 --> 00:00:29.419
The ideal comparison contrasts
treatment and control groups

00:00:29.419 --> 00:00:30.600
that look similar.

00:00:30.600 --> 00:00:34.700
But sometimes this sort
of comparability is elusive.

00:00:34.700 --> 00:00:36.805
When treatment and control groups

00:00:36.805 --> 00:00:40.100
evolve similarly
in the absence of treatment,

00:00:40.100 --> 00:00:42.307
even if from different
starting points,

00:00:42.307 --> 00:00:44.900
there's hope for causal inference.

00:00:45.600 --> 00:00:48.400
The weapon that exploits
parallel evolution,

00:00:48.664 --> 00:00:50.886
masters say parallel trends,

00:00:50.886 --> 00:00:53.363
is called differences-in-differences...

00:00:53.363 --> 00:00:54.300
(voice whispering)
Differences-in-differences

00:00:54.400 --> 00:00:56.900
- ...or DD for short.
- Alright. Nice.

00:00:56.900 --> 00:00:59.987
Let's see how DD
can help us understand

00:00:59.987 --> 00:01:03.058
one of the most important
economic events

00:01:03.058 --> 00:01:04.370
in US history.

00:01:05.300 --> 00:01:08.300
Look back with me now
at the Great Depression--

00:01:08.800 --> 00:01:12.200
the worst economic catastrophe,
our country has ever known.

00:01:13.100 --> 00:01:16.200
Unemployment hit 25% in 1933--

00:01:16.600 --> 00:01:19.507
a level not seen before or since.

00:01:19.507 --> 00:01:22.100
Millions lost their homes
or their land.

00:01:22.600 --> 00:01:24.737
Suicide spiked, and hungry families

00:01:24.737 --> 00:01:26.766
relied on soup kitchens
and bread lines

00:01:26.766 --> 00:01:28.155
to keep from starving.

00:01:29.400 --> 00:01:34.000
Economists argue fiercely over
the causes of the Great Depression.

00:01:34.000 --> 00:01:36.983
Most agree, however,
that a key piece of the puzzle

00:01:36.983 --> 00:01:39.620
is an epidemic of bank failures.

00:01:39.800 --> 00:01:41.900
This was before deposit insurance.

00:01:42.100 --> 00:01:46.490
So if your bank went bankrupt,
your savings disappeared with it,

00:01:52.600 --> 00:01:56.100
Faced with a banking crisis,
the central bank has a choice:

00:01:56.400 --> 00:01:58.524
lend freely to troubled banks

00:01:58.524 --> 00:02:01.100
or stand aside and refuse to lend.

00:02:01.500 --> 00:02:05.440
Lending freely to banks in trouble
is called easy money.

00:02:05.440 --> 00:02:08.100
Refusing to lend is called tight money.

00:02:10.200 --> 00:02:12.872
Monetarist masters Milton Friedman
and Anna Schwartz

00:02:12.872 --> 00:02:14.611
famously called
the Great Depression

00:02:14.611 --> 00:02:16.350
the "Great Contraction,"

00:02:16.800 --> 00:02:18.367
accusing the Federal Reserve

00:02:18.367 --> 00:02:21.200
of inflicting a misguided policy
of tight money

00:02:21.200 --> 00:02:24.000
on the nation's teetering
financial institutions.

00:02:24.400 --> 00:02:25.873
They argued that easy money

00:02:25.873 --> 00:02:27.985
would have kept
many banks in business,

00:02:27.985 --> 00:02:29.700
shortening the Great Depression,

00:02:30.400 --> 00:02:32.239
But others disagree!

00:02:32.239 --> 00:02:33.769
If banks are insolvent

00:02:33.769 --> 00:02:35.954
because of unwise
lending decisions,

00:02:35.954 --> 00:02:38.900
then bailouts just encourage
more foolishness.

00:02:39.600 --> 00:02:42.965
Economists call this problem
"moral hazard."

00:02:42.965 --> 00:02:46.100
The debate over bailouts
in moral hazard continues today.

00:02:46.500 --> 00:02:48.599
Should financial behemoth
Lehman Brothers

00:02:48.599 --> 00:02:51.500
had been allowed to fail
on the eve of the Great Recession,

00:02:52.000 --> 00:02:54.703
in an ideal world,
we'd answer this question

00:02:54.703 --> 00:02:58.400
applying different Fed policies
to randomly selected regions.

00:02:59.000 --> 00:03:00.250
But we can still learn a lot

00:03:00.250 --> 00:03:02.119
by using differences-in-differences

00:03:02.119 --> 00:03:06.300
to compare trends across areas
with different monetary policies.

00:03:10.900 --> 00:03:12.522
How's that even possible?

00:03:12.522 --> 00:03:15.779
Don't the same Fed policies
apply to all banks in the US?

00:03:15.779 --> 00:03:17.400
- Yeah.
- Good question.

00:03:17.700 --> 00:03:21.484
The Federal Reserve System
is divided into 12 districts,

00:03:21.484 --> 00:03:24.301
each headed by a regional bank.

00:03:24.301 --> 00:03:27.467
Today, Fed policy is set
at the national level.

00:03:27.467 --> 00:03:32.200
But in the 1930s, regional Feds
could do pretty much as they liked.

00:03:32.200 --> 00:03:33.276
Ah, interesting.

00:03:33.276 --> 00:03:35.500
And here's what's
so awesome about that.

00:03:35.500 --> 00:03:39.302
In 1930, the Atlanta Fed,
running the 6th District,

00:03:39.302 --> 00:03:41.473
followed an easy money policy,

00:03:41.473 --> 00:03:45.400
sending wheelbarrows of cash
to rescue insolvent institutions,

00:03:45.900 --> 00:03:48.816
The St. Louis Fed,
running the 8th District,

00:03:48.816 --> 00:03:50.668
followed a tight money policy.

00:03:51.100 --> 00:03:53.900
"Let fail the foolish!"
they said in St. Louis.

00:03:54.300 --> 00:03:58.701
And so a natural experiment
in monetary policy was born.

00:03:58.701 --> 00:04:02.000
Even better, this is
a within-state experiment.

00:04:02.000 --> 00:04:04.347
The border between the 6th
and the 8th districts

00:04:04.347 --> 00:04:06.762
ran smack through
the middle of Mississippi.

00:04:07.300 --> 00:04:09.321
So northern Mississippi
had tight money,

00:04:09.321 --> 00:04:11.870
while southern Mississippi
had easy money,

00:04:11.870 --> 00:04:15.200
but under the same state laws
in banking regulations in both.

00:04:15.770 --> 00:04:16.853
The treatment group

00:04:16.853 --> 00:04:20.151
is the District 6th part
of Mississippi,

00:04:20.151 --> 00:04:23.100
which had access to easy money
during the crisis.

00:04:23.800 --> 00:04:25.091
The control group

00:04:25.091 --> 00:04:27.800
is the District 8th part
of Mississippi,

00:04:27.800 --> 00:04:30.225
which had tight money
during the crisis.

00:04:31.300 --> 00:04:34.040
The key year
in our natural experiment

00:04:34.040 --> 00:04:35.400
was 1930,

00:04:35.800 --> 00:04:37.563
Caldwell & Company,

00:04:37.563 --> 00:04:40.377
a massive financial empire
in the South

00:04:40.377 --> 00:04:41.987
came crashing down.

00:04:42.700 --> 00:04:46.000
Banking is a business
built on confidence and trust.

00:04:46.500 --> 00:04:49.151
The Caldwell meltdown
caused a panic

00:04:49.151 --> 00:04:53.049
that led to a widespread
bank run all at once.

00:04:53.049 --> 00:04:55.114
Depositors wanted their money back,

00:04:55.114 --> 00:04:58.200
causing banks to go bankrupt
and shut their doors.

00:05:01.000 --> 00:05:03.321
We'll use differences-in-differences

00:05:03.321 --> 00:05:06.614
to measure the effect
of contrasting monetary policies

00:05:06.614 --> 00:05:09.164
in response to the Caldwell crisis.

00:05:12.500 --> 00:05:16.439
This figure plots the number
of banks in Mississippi by year,

00:05:16.439 --> 00:05:19.273
for the 8th and 6th districts.

00:05:19.273 --> 00:05:24.200
Let's start in 1929 - a year
before the Caldwell crash.

00:05:24.200 --> 00:05:27.646
There are 169 banks
open in the 8th,

00:05:27.646 --> 00:05:31.366
and 141 banks open in the 6th.

00:05:31.366 --> 00:05:32.784
Over the next year,

00:05:32.784 --> 00:05:37.000
we see a similar handful
of banks fail, in both districts.

00:05:37.400 --> 00:05:40.135
The change in the number
of banks in operation

00:05:40.135 --> 00:05:42.168
is remarkably similar.

00:05:42.168 --> 00:05:44.600
That's what parallel trends look like.

00:05:45.500 --> 00:05:48.783
In November 1930, Caldwell crashes,

00:05:48.783 --> 00:05:50.400
and the panic begins.

00:05:51.200 --> 00:05:53.250
Banks fail frequently in the 8th
district,

00:05:53.250 --> 00:05:55.300
which had Tight money,

00:05:55.800 --> 00:06:00.100
but the decline is slower in the
6th District which had easy money.

00:06:00.800 --> 00:06:03.000
The diverging Trends in this period,

00:06:03.000 --> 00:06:06.300
might be attributable to
easy versus Tight money

00:06:06.600 --> 00:06:11.800
in July, of 1931, the 8th
District, abandons type money.

00:06:11.900 --> 00:06:14.200
So now both districts are easy.

00:06:14.700 --> 00:06:16.900
Parallel Trends are restored

00:06:17.300 --> 00:06:22.200
in a counterfactual world where the sixth
district follows a Tight money policy.

00:06:22.500 --> 00:06:23.800
What might have happened

00:06:24.300 --> 00:06:28.400
if we extrapolate the trend of
the 8th District to the sixth,

00:06:28.500 --> 00:06:29.900
it would look like this.

00:06:30.400 --> 00:06:30.700
So

00:06:30.800 --> 00:06:35.900
The treatment effective easy money is
how much the 6th District deviated From

00:06:35.900 --> 00:06:38.900
the Path implied by
the 8th District trend.

00:06:41.100 --> 00:06:44.000
How many banks did the
Easy Money treatment save

00:06:44.000 --> 00:06:49.800
this table reports data for the treatment
group District Six in the first row

00:06:49.800 --> 00:06:54.000
and data for the control group
District 8 in the second row.

00:06:54.000 --> 00:06:57.400
The First Column shows the
number of banks in business,

00:06:57.400 --> 00:07:03.400
before The Crisis began in 1930.
The second column shows 1931.

00:07:03.400 --> 00:07:08.900
This is the key period when each district
had differing monetary policies during

00:07:08.900 --> 00:07:10.900
the crisis the rightmost.

00:07:11.200 --> 00:07:11.600
Column

00:07:11.700 --> 00:07:18.000
reports changes within the district
district, 6 lost 14 Banks while District

00:07:18.100 --> 00:07:23.800
8 lost 33, the mathematical formula
for the treatment effect is simple.

00:07:24.400 --> 00:07:28.700
We subtract the change in banks
in operation, in the 8th District

00:07:28.900 --> 00:07:32.200
from the change in banks
in operation in the sixth.

00:07:32.800 --> 00:07:35.700
Hence. The name
differences in differences,

00:07:37.000 --> 00:07:40.900
negative 14, minus negative, 33 equals 9.

00:07:41.100 --> 00:07:41.600
19.

00:07:42.500 --> 00:07:46.800
We estimate that 19 Banks
were saved by easy money

00:07:47.300 --> 00:07:51.700
in practice tables and figures like
those shown here are the beginning

00:07:51.800 --> 00:07:54.300
rather than the end of a DD analysis,

00:07:55.400 --> 00:08:00.000
the problem of how to gauge the
statistical, significance of DD, estimates

00:08:00.100 --> 00:08:02.300
turns out to be exceedingly, tricky

00:08:02.400 --> 00:08:05.600
and a regression is typically
part of the solution,

00:08:09.200 --> 00:08:10.700
the key assumption behind a

00:08:11.100 --> 00:08:17.700
DD analysis is that of parallel Trends
recall, the principle of ceteris paribus,

00:08:17.900 --> 00:08:21.700
our ideal comparison would have
the two districts experienced,

00:08:21.700 --> 00:08:27.800
an identical business environment, except
for one factor easy or tight money.

00:08:29.200 --> 00:08:33.500
Both districts would have identical
types of customers who would go bankrupt

00:08:33.500 --> 00:08:35.300
at exactly the same rate.

00:08:35.700 --> 00:08:38.600
The skill of their employees
would be equal and so on

00:08:39.200 --> 00:08:43.400
perfect ceteris, paribus comparisons
would allow us to clearly see

00:08:43.600 --> 00:08:48.700
the causal effect of different fed
policies in this case, that's not possible.

00:08:49.100 --> 00:08:53.600
But the idea of parallel Trends
is based on a similar concept.

00:08:53.600 --> 00:08:57.900
If we see that the two regions
experienced similar Trends in the number

00:08:57.900 --> 00:08:58.800
of banks over.

00:08:59.000 --> 00:09:04.000
Time in the absence of treatment, we
can assume they are good comparisons.

00:09:04.400 --> 00:09:09.800
We see that the two districts move in
parallel both before the crisis. And after,

00:09:10.100 --> 00:09:12.400
when they have the same Fed policy,

00:09:13.100 --> 00:09:18.300
the only time the district's behave differently
is when the Fed policy is different.

00:09:19.400 --> 00:09:26.700
In view of this Fed policy is a likely
cause of diverging Trends from 1930 to 1931.

00:09:27.800 --> 00:09:31.800
But we should also check for other
changes unique to northern, Mississippi.

00:09:32.300 --> 00:09:33.200
What do you mean?

00:09:33.500 --> 00:09:39.100
Imagine that bad tornadoes? Hit Northern,
but not Southern. Mississippi in 1930,

00:09:39.600 --> 00:09:44.300
these tornadoes devastate Farms
causing Farmers to default on loans,

00:09:44.600 --> 00:09:46.800
which drives their Banks out of business.

00:09:47.400 --> 00:09:52.200
Then the sixth and eighth districts
would differ in not one, but two ways

00:09:52.700 --> 00:09:57.200
Fed policy, and whether and
we'd have trouble identifying

00:09:57.600 --> 00:10:02.500
Policy as the causal Factor behind
increased bank failures in the eighth.

00:10:07.200 --> 00:10:10.200
DD credibility lives or
dies with the claim that the

00:10:10.200 --> 00:10:13.600
only reason northern
Mississippi was special in 1930

00:10:13.900 --> 00:10:16.000
is differing. Regional Fed policy.

00:10:16.600 --> 00:10:22.100
We're in DD heaven with strong Visual
Evidence of parallel Trend in general.

00:10:22.200 --> 00:10:24.200
The first step in evaluating whether

00:10:24.200 --> 00:10:30.200
to use DD is usually this type of
visual confirmation of parallel Trends

00:10:30.200 --> 00:10:31.700
outside of the period.

00:10:31.700 --> 00:10:36.900
When we expect to see a treatment
effect, the treatment in our example,

00:10:37.100 --> 00:10:39.700
Easy money in the face of bank failures

00:10:40.500 --> 00:10:45.000
metrics Masters, use DD to
explore effects of many policies

00:10:45.800 --> 00:10:47.900
like the minimum, legal drinking age,

00:10:48.500 --> 00:10:52.200
and environmental changes
like access to clean water.

00:10:52.900 --> 00:10:54.200
In our next video.

00:10:54.500 --> 00:10:59.200
We'll see an example of how regression
is used to implement a DD approach.

00:11:00.800 --> 00:11:06.900
Are you a teacher click to explore ways to
use these videos in class if your learner

00:11:07.000 --> 00:11:11.200
Make sure this video sticks by taking
a few quick practice questions,

00:11:11.600 --> 00:11:14.200
or if you're ready.
Click for the next video.

00:11:14.600 --> 00:11:20.000
You can also check out Mr. Use website for
more courses, teacher resources and more.