Policy uncertainty: trying to estimate the
uncertainty impact of Brexit
Scott Baker (Northwestern), Nick Bloom (Stanford) and
Steve Davis (Chicago)
September 2nd 2016
1) Measuring policy uncertainty
2) Evaluating our measure
3) Estimating the uncertainty impact of Brexit
Our proxy for Economic Policy Uncertainty (EPU)
comes from computer searches of newspapers
• For 10 major US papers get monthly counts of articles with:
E {economic or economy}, and
P {regulation or deficit or federal reserve or congress or legislation or white house}, and
U {uncertain or uncertainty}
• Divide the count for each month by the count of all articles
• Normalize and sum 10 papers to get the U.S monthly index
50
10
015
020
025
030
0
1985 1990 1995 2000 2005 2010 2015
Gulf
War I
9/11
Clinton
Election
Gulf
War IIBush
Election
Stimulus
Debate
Lehman
and TARP
Russian
Crisis/LTCM
Debt Ceiling
Debate
Black
Monday
Fis
cal C
liff
Shutd
ow
n
Po
licy U
ncert
ain
ty I
nd
ex
US News-based economic policy uncertainty index
Source: “Measuring Economic Policy Uncertainty” by Scott R. Baker, Nicholas Bloom and Steven J. Davis, all data at
www.policyuncertainty.com. Monthly data normalized to 100 prior to 2010. Data to July 2016
Bre
xit
Can focus on narrower areas or economic policy
uncertainty, e.g. : Defense and Healthcare0
10
02
00
30
04
00
1985 1990 1995 2000 2005 2010 2015Year
Defense Uncertainty
Health Uncertainty
Cate
go
rical
Po
licy U
ncert
ain
ty I
nd
exes
Gulf War I
Gulf War II
9/11Clinton healthcare
initiative
Affordable
care act
5Source: “Measuring Economic Policy Uncertainty” by Scott R. Baker, Nicholas Bloom and Steven J. Davis, all
data at www.policyuncertainty.com. Data normalized to 100 prior to 2010.
0
20
040
060
080
010
00
1998 2000 2002 2004 2006 2008 2010 2012 2014 2016
UK Policy Uncertainty Index – striking Brexit spike
Source: www.policyuncertainty.com. Monthly data from 2 UK newspapers (Times and Financial
Times). Data to July 2016
Po
licy U
ncert
ain
ty I
nd
ex
Treaty of
Accession/
Gulf War II
Russian
Crisis/LTCM
Northern
Rock &
Global
Financial
Crisis
General
Election
Euro Crisis
Scottis
h R
efe
rendum
9/11
Ele
ctio
n b
uild
-up
Brexit
020
040
060
080
010
00
UK
EP
U inde
x
02
46
810
BoE
unce
rta
inty
in
de
x
2000 2002 2004 2006 2008 2010 2012 2014 2016
News EPU index vs the Bank Uncertainty Index
Source: Data to July 2016. Thanks to Phil Bunn for BoE data
Correlation=0.458 (p-value 0.001)
0.2
.4.6
01
00
20
03
00
1885 1895 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005
Notes: Index of Policy-Related Economic Uncertainty composed of quarterly news articles containing uncertain or uncertainty, economic or
economy, and policy relevant terms (scaled by the smoothed total number of articles) in the Guardian and the Times
UK Policy Uncertainty since 1885 – even against this
Brexit looks like a large uncertainty shock
3-day week
and 2 national
elections
ERM exit,
Major
election
Thatcher
forced to
resign
Falklands warAtlee-
Churchill
election
Pound
devalues
30%
Boer war
Start of
WWI
Start of
WWII
MacMillan
resigns
Irish
war
IMF
crisis
Salisbury
loses NCV
Salisbury
resignation
040
80
12
016
0
0
50
010
00
15
00
20
00
2002 2004 2006 2008 2010 2012 2014 2016Year
UK Daily EPU: large Brexit shock in EPU news index
but no stock-market volatility shock
Source: www.policyuncertainty.com. Data from Access World News database on 650 UK
newspapers. Daily data to August 25th 2016
Daily
New
s P
olic
y U
ncert
ain
ty I
ndex
Daily
Implie
d V
ol In
dex (IV
I)
0
10
020
030
040
0
1998 2000 2002 2004 2006 2008 2010 2012 2014 2016Year
Po
licy
Un
cert
ain
ty I
nd
ex
9/11
Papandreou calls
for referendum,
then resignsLehman
Bros.Greek
Bailout
Request,
Rating
Cuts
Northern
Rock &
Ensuing
Financial
Turmoil
Treaty of
Accession/
Gulf War II
Nice Treaty
Referendum
Source: www.policyuncertainty.com. From 10 papers (El Pais, El Mundo, Corriere della Sera, La
Repubblica, Le Monde, Le Figaro, the Financial Times, Times, Handelsblatt, FAZ.) Data to July 2016
Russian
Crisis/LTCM French &
Dutch Voters
Reject
European
Constitution
Ongoing
Eurozone
Stresses
European Economic Policy Uncertainty IndexBrexit
1) Measuring policy uncertainty
2) Evaluating our policy uncertainty measure
3) Estimating the uncertainty impact of Brexit
A) Evaluation of Policy Uncertainty Index: Market Use
Market use suggests some information in our EPU data
I) Numerous users including: Goldman Sachs, Citibank, JP
Morgan, Blackrock, Wells Fargo, IMF, Fed, ECB etc
II) This has led Bloomberg, FRED, Reuters and Haver to stream
our data for their financial and policy users
B) Evaluation: comparison to stock volatility (e.g. VIX)
10
20
30
40
50
60
50
10
01
50
20
02
50
30
0
1990 1995 2000 2005 2010 2015
US
VIX
(re
d)
US
Ec
on
om
ic P
olic
y U
nc
ert
ain
ty In
de
x (
Blu
e -
--) 9/11
WorldCom
Fraud
Gulf
War II
Lehman
Failure, TARP
Asian
Financial
Crisis
Gulf
War I
Euro-
zone
Crisis
Debt Ceiling
Dispute
Russian
Crisis,
LTCM
DefaultStimulus
DebateClinton
Election
Notes: The figure shows the U.S. EPU Index from Figure 1 and the monthly average of daily values for the 30-day VIX.
Bush
Election
Fis
ca
l C
liff
Go
ve
rnm
en
t
Sh
utd
ow
n
C) Running Detailed Human Audits
10 undergraduates read ≈ 10,000 newspaper articles to date using a 63-page audit guide to code articles if they discuss “economic uncertainty” and “economic policy uncertainty”
14
Find humans and computers give similar results in
large samples (in fact both make mistakes)
Po
lic
y U
nc
ert
ain
ty I
nd
ex
01
00
20
03
00
1985 1990 1995 2000 2005 2010
Computer
Human
Correlation=0.76
1) Measuring policy uncertainty
2) Evaluating our measure
3) Estimating the uncertainty impact of Brexit
Best guess using VARs based on the Baker,
Bloom and Davis (2016) numbers
• BBD (2016) estimated impact of Great Recession (GR) shock
• GR raised global EPU by 100 points (2011-12 vs 2005-06)
• Brexit raised UK EPU probably by 800 points at maximum, but probably 200 points on average (Q3 2016 to Q2 2017)?
• Both shocks have possibly similar persistence, but Brexit is national (GR was global) so maybe scale down Brexit impact?
• So our guess is uncertainty impact Brexit GR impact, but very hard to tell – large, unprecedented uncertainty shock
Ind
ustr
ial
Pro
du
cti
on
, (%
)
Months after the economics policy uncertainty shock
Notes: Plots the impulse
response function for
Industrial Production and
employment to an increase
in the policy-related
uncertainty index from the
2005-2006 average value to
the 2011-2012 average
value. The central (black)
solid line is the mean
estimate while the dashed
(red) outer lines are the 90%
confidence bands.
Estimated using a monthly
Cholesky Vector Auto
Regression (VAR) with 3
lags on the EPU index,
log(S&P 500 index),
unemployment rate, and log
industrial production, plus a
full set of country, year and
month fixed-effects. Country
data weighted by the
number of newspapers used
to make the EPU series. Fit
to monthly data from
1985M1 to 2012M12 where
available. Estimated on data
from Canada, China,
France, Germany, India,
Italy, Japan, Korea, Russia,
Spain, UK and the USA.Un
em
plo
ym
en
t Im
pac
t, (
%)
0.1
.2.3
.4
0 6 12 18 24 30 36
-1.5
-1-.
50
0 6 12 18 24 30 36
12 country VAR for Great Recession (from BBD 2016)
Assuming Brexit similar impact predicts drop
of up to 1% (vs trend) in Q42016 & Q1 2017
Ind
ustr
ial
Pro
du
cti
on
, (%
)
Months after the economics policy uncertainty shock
Em
plo
ym
en
t Im
pact,
(%
)UK VAR for Great Recession – but small sample so
large standard errors, so I prefer 12 country sample-4
-3-2
-10
0 6 12 18 24 30 36month
0.2
.4.6
.8
0 6 12 18 24 30 36month
Notes: Plots the impulse
response function for
Industrial Production and
employment to an increase
in the policy-related
uncertainty index from the
2005-2006 average value to
the 2011-2012 average
value. The central (black)
solid line is the mean
estimate while the dashed
(red) outer lines are the 90%
confidence bands.
Estimated using a monthly
Cholesky Vector Auto
Regression (VAR) with 3
lags on the EPU index,
log(S&P 500 index),
unemployment rate, and log
industrial production, plus a
full set of country, year and
month fixed-effects. Country
data weighted by the
number of newspapers used
to make the EPU series. Fit
to monthly data from
1985M1 to 2012M12 where
available. Estimated on
data from the UK only
Assuming Brexit similar impact predicts drop
of up to 2% (vs trend) in Q42016 & Q1 2017
-8-6
-4-2
0
0 6 12Quarters
-2-1
.5-1
-.5
0
0 6 12Quarters
US VAR for Impact on GDP and Investment (quarterly)G
DP
Re
sp
on
se
, %
Notes: VAR-estimated
impulse response
functions for GDP and
Gross Fixed
investment to an EPU
innovation equal to the
increase in the EPU
index from its 2005-
2006 to its 2011-2012
average value, with 90
percent confidence
bands. Identification
based on three lags
and a Cholesky
decomposition with the
following ordering:
EPU index, log(S&P
500 index), federal
reserve funds rate, log
gross investment, log
gross domestic
product). Fit to data
from 195 to 2014.
Investm
en
t R
esp
on
se,
%
Assuming Brexit similar impact predicts GDP
drop of 1% (vs trend) in Q42016 & Q1 2017
-6-4
-20
0 6 12Quarters
Investment is the part of GDP that reacts by far the
most - I think because investment is dominated by
larger firms, which are forward lookingNotes: VAR-
estimated impulse
response functions
for GDP,
Consumption and
Gross Fixed
investment to an
EPU innovation
equal to the
increase in the EPU
index from its 2005-
2006 to its 2011-
2012 average
value, with 90
percent confidence
bands. Identification
based on three lags
and a Cholesky
decomposition with
the following
ordering: EPU
index, log(S&P 500
index), federal
reserve funds rate,
log gross
investment, log
gross domestic
product). Fit to data
from 195 to 2014.
Investment
GDP
Consumption
%Im
pa
ct
Conclusions
- Policy uncertainty rose dramatically globally after Brexit,
particularly in the UK
- Estimating the impact is frankly very hard! For the UK maybe
a 1% hit to GDP (would be bigger except offsetting exchange
rate and interest rate movements)
- Likely to be spread out over the next year, with initial impact in
production side of economy followed by household side
Prior Q&A
Q1) What are the best uncertainty measures to look at?
A) Given implied volatility did not rise post Brexit, we are less
keen on this. News feels attractive as focused on current beliefs
Q2) Will uncertainty metrics (news, IVI etc) capture uncertainty
around Brexit?
A) Probably not – this is a unique event and these are only
rough measures. News at least has the ability to search by word
Q3) How should uncertainty metrics evolve over next few years?
A) Hard to tell – except for the Great Recession, prior uncertainty
shocks have been very quick, but Brexit likely to persist, I fear
Prior Q&A
Q4) How quickly will Brexit uncertainty show in the data?
A) Firms likely to respond far faster (although exchange rate
drop at same time). Consumers probably only respond when
employment and wages respond, so with 1 or 2 quarters lag
Q5) How does Brexit interact with credit (is it a credit shock)?
A) Credit conditions likely to worsen as the left tail of outcomes
have grown. Recently working on “The uncertainty-finance
multiplier” & seems real and financial frictions amplify each other
Q6) Will uncertainty impact supply side (e.g. via productivity)
A) Probably, but evidence limited. Uncertainty likely to
particularly reduce most irreversible investment – R&D, training,
management reorganizations – which will would cut TFP growth
Prior Q&A
Q7) What channels will Brexit work through – e.g. housing?
A) My guess is that commercial property will be hit (firms are
forward looking) but private housing hit only once the economy
starts to drop (consumers less forward looking). Also vary
heavily by sector because of large exchange rate cut.
Q8) Is the uncertainty impact different from the direct impact?
A) Historically all uncertainty shocks if the data are bad news, so
have a direct and an uncertainty component. In the case of
Brexit I would think the uncertainty component is particularly
large early on (not much has happened) and direct effects larger
later.