Transcript European Commission Directorate General Economic and Financial Affairs Using BCS data for tracking q-o-q GDP growth Andreas Reuter Business and consumer surveys and short-term forecast (ECFIN.

European Commission
Directorate General Economic and Financial Affairs
Using BCS data for tracking
q-o-q GDP growth
Andreas Reuter
Business and consumer surveys and
short-term forecast (ECFIN A4.2)
Outline
1.
Introduction: the Economic Sentiment Indicator (ESI)
2.
Relative weaknesses of the Economic Sentiment Indicator
(ESI)
3.
Refresher on ESI Construction Method
4.
Improving the ESI's tracking performance of q-o-q GDP
growth

step a:
re-constructing the ESI based on
"best-performing" survey questions

step b:
an ESI with amplified changes
2
1. Introduction: the Economic Sentiment
Indicator (ESI)
Purpose of the ESI:
 summarising developments in all 5 sectors covered by DG
ECFIN's Business and Consumer Surveys (BCS):
 industry
 retail trade
 consumers
 services
 construction

tracking GDP growth at Member State, EU and euro-area
level
Added value:


timeliness (complementing delayed
quantitative statistics)
high frequency
3
2. Relative Weaknesses of the Economic
Sentiment Indicator (ESI)
ESI is excellent in tracking GDP growth y-o-y…
120.00
6.00
115.00
4.00
110.00
105.00
2.00
100.00
95.00
0.00
90.00
Correlations:
coincident
0.92
leading 1
0.87
leading 2
0.68
-2.00
85.00
80.00
-4.00
75.00
70.00
-6.00
ESI (quarterly levels)
However…
GDP growth (quarterly, y-o-y, euro area, rhs)
4
2.
Relative Weaknesses of the Economic Sentiment Indicator (ESI)
ESI is less convincing in tracking GDP growth q-o-q…
65
60
55
50
45
40
35
30
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
06/1995
06/1996
06/1997
06/1998
06/1999
06/2000
06/2001
06/2002
06/2003
06/2004
06/2005
06/2006
06/2007
06/2008
06/2009
06/2010
06/2011
06/2012
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
06/1995
06/1996
06/1997
06/1998
06/1999
06/2000
06/2001
06/2002
06/2003
06/2004
06/2005
06/2006
06/2007
06/2008
06/2009
06/2010
06/2011
06/2012
120
115
110
105
100
95
90
85
80
75
70
PMI (quarterly levels)
ESI (quarterly levels)
GDP growth (quarterly, q-o-q, euro area, rhs)
GDP growth (quarterly, q-o-q, euro area, rhs)
downturn signalled with:
2 quarters delay
103
2
1.5
1
0.5
0
-0.5
-1
-1.5
-2
-2.5
-3
102
101
100
99
98
97
06/1995
06/1996
06/1997
06/1998
06/1999
06/2000
06/2001
06/2002
06/2003
06/2004
06/2005
06/2006
06/2007
06/2008
06/2009
06/2010
06/2011
06/2012
96
1 quarter delay
quickness of recovery
underestimated
Correlations:
ESI
PMI
OECD CLI
coincident
0.71
0.87
0.67
leading 1
0.46
0.65
0.38
leading 2
0.19
0.37
0.05
OECD Composite Leading Indicator (quarterly levels)
GDP growth (quarterly, q-o-q, euro area, rhs)
5
3. Refresher on ESI Construction Method
ingredients: balance series of 15 survey questions
The questions are:
 seasonally adjusted
 standardised
% of positive answers minus % of negative answers
Effect:
 comparability of balance series in
terms of mean and volatility
 no series dominates development of
ESI due to a higher amplitude
allocating weights per sector:
Industry: 40% ; Services: 30% ; Consumers: 20% ; Construction: 5% ; Retail Trade: 5%
individual indu question has weight of 13,3% (=40% weight / 3 questions)
calculation of arithmetic mean of weighted balances


standardisation of the ESI and:
Effect:
addition of 100
 values >100 indicate above-average
economic sentiment
multiplication by 10
 2/3 of observations will be in the
interval [90 ; 110]
(assuming normality)
6
4. Improving the ESI's tracking performance
of q-o-q GDP growth
step a:
1.
re-constructing the ESI based on
"best-performing" survey questions
Correlation of all individual survey questions with i)
reference series, ii) q-o-q GDP growth:
quarterly
for industry:
averages
Gross Value Added in Manufacturing
of
for services:
balance
Gross Value Added in Services
series
for consumers:
Household and NPISH final consumption expenditure
for construction:
Gross Value Added in Construction
for retail trade:
Household and NPISH final consumption expenditure
7
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step a: reconstructing the ESI based on "best-performing" survey questions
2.
Construction of 3 new sector-specific Confidence
Indicators (CIs):
 CIs summarise overall perceptions / expectations at
individual sector level
 calculation: arithmetic mean of (seasonally adjusted)
balances for specific questions
 questions included in sectoral CIs are also the ones used
to construct the ESI
1. CI based on the 2 best performing questions (reg. correlation with
reference series & GDP q-o-q)
2. CI based on the 3 best performing questions (reg. correlation with
reference series & GDP q-o-q)
3. CI based on all forward-looking questions of the respective sector
3.
For each sector: selection of the best CI (reg. correlation
with reference series & GDP q-o-q)
8
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step a: reconstructing the ESI based on "best-performing" survey questions
Intermediate Results:
Industry
Services
Consumers
Retail Trade
Construction
order books
- currently
business
- last 3 months
household's fin.
position
- next 12 months
business activity
(sales)
- last 3 months
order books
- currently
stock of (finished)
products
- currently
demand for firm's
services
- last 3 months
econ. situa-tion in
MS
- next 12 months
volume of stock
- currently
firm's employment
- next 3 months
production
- next 3 months
demand for firm's
services
- next 3 months
unemploy-ment in
MS
- next 12 months
business activity
(sales)
- next 3 months
building activity
- last 3 months
likelihood of
saving money
- next 12 months
expected orders
with suppliers
- next 3 months
production
- last 3 months
expected level of
major purchases
- next 12 months
Correlation with GDP q-o-q (with reference series):
0.61 (0.54)
0.65 (0.71)
0.64 (0.58)
0.46 (0.22)
0.31 (0.32)
0.73 (0.68)
0.67 (0.73)
0.64 (0.69)
0.54 (0.27)
0.32 (0.35)
9
Improving the ESI's tracking performance of q-o-q GDP growth – step a: reconstructing the ESI based on "best-performing" survey questions
4.
Re-construction of the ESI, using the set of questions of
the new CIs:
Slight improvements…
120
02
01
110
Turning points:
modified ESI records 0-change in
quarter where GDP-downturn
starts, while the ESI still signals a
rise
01
100
00
-01
90
-01
80
-02
-02
70
-03
60
Amplitude:
modified ESI records steeper
downward slope than ESI, being
more in line with GDP-growth
-03
03/1996
11/1996
07/1997
03/1998
11/1998
07/1999
03/2000
11/2000
07/2001
03/2002
11/2002
07/2003
03/2004
11/2004
07/2005
03/2006
11/2006
07/2007
03/2008
11/2008
07/2009
03/2010
11/2010
07/2011
03/2012
4.
ESI with new questions (quarterly levels)
ESI (quarterly levels)
Correlations:
ESI
modified ESI
improvement
coincident
0.71
0.77
8%
leading 1
0.46
0.51
12%
leading 2
0.19
0.25
30%
GDP growth (quarterly, q-o-q, euro area, rhs)
10
4.
Improving the ESI's tracking performance of q-o-q GDP growth
step b:
an ESI with amplified changes
Intuition of the approach:
Comparable changes in the ESI should be taken more "seriously",
when reflected by many survey questions.
>>>> change in ESI should be multiplied, if a
critical amount of questions changes in the same direction
We propose: 8 (out of 11) questions
Example:
Instead:
change in ESI
should be
-2*x (with x > 1)
We propose:
multiplication by 3
standard deviation of balance series
0.5
0
-0.5
-1
-1.5
-2
Change in modified ESI: -1.8
change in 1999Q1 (compared to previous quarter)
Change in modified ESI: -2
change in 2000Q4 (compared to previous quarter)
11
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth
Calculation of the new method:
1. Sum all standardised weighted questions per quarter:
Variable is called:
2.
ZNEW
Calculate the (modified) ESI:
12
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth
4.
3.
Calculate the absolute change of ZNEW per quarter:
Variable is called:
ZNEW change
sum of weighted standardised questions (per quarter)
4.
Calculate for each quarter a variable taking value 1 if >=8
questions go up / go down (trigger variable):
5.
Re-calculate "ZNEW change" mutiplying it by 3 (only in
case the "trigger variable" has value 1):
Variable is called:
ZNEW change (amplified)
13
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth
6.
Re-calculate ZNEW, adding "ZNEW change (amplified)" of
quarter t to "ZNEW" of quarter t-1:
sum of weighted standardised questions (per quarter)
Variable is called:
ZNEW (amplified)
7.
Standardise "ZNEW (amplified)" and thus obtain a new ESI
with amplified change:
Variable is called:
ESINEW (amplified change)
14
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth
4.
improvements:
compared to
1.5 current ESI:
Results:
120
1
 "micro"0.5
volatility of
GDP better
0
captured
110
100
compared to
PMI:
 "micro"volatility of
GDP better
captured
-0.5
90
 2008downturn
-1.5
announced
by steeper
-2
slope (more
-2.5
in line with
GDP)
-1
80
70
-3
03/1996
09/1996
03/1997
09/1997
03/1998
09/1998
03/1999
09/1999
03/2000
09/2000
03/2001
09/2001
03/2002
09/2002
03/2003
09/2003
03/2004
09/2004
03/2005
09/2005
03/2006
09/2006
03/2007
09/2007
03/2008
09/2008
03/2009
09/2009
03/2010
09/2010
03/2011
09/2011
03/2012
09/2012
60
 better
leading
properties
current ESI (quarterly levels)
ESINEW - amplified change (quarterly levels)
PMI (quarterly levels, rescaled to long-term mean of 100)
GDP growth (quarterly, q-o-q, euro area, rhs)
 2009upswing
reflected
with steeper
slope (in line
with GDP)
time-period
current ESI
ESINEW (ampl.)
PMI
increase
compared to
current ESI
98Q3 – 02Q1
0.58 (0.28)
0.76 (0.52)
0.74 (0.54)
31% (85%)
4% (-4%)
02Q2 – 07Q1
0.79 (0.64)
0.85 (0.69)
0.86 (0.63)
8% (8%)
-1% (10%)
07Q2 – 12Q2
0.74 (0.39)
0.91 (0.73)
0.89 (0.61)
23% (87%)
2% (18%)
98Q3 – 07Q1
0.64 (0.41)
0.79 (0.58)
0.79 (0.58)
25% (41%)
0% (0%)
98Q3 – 12Q2
0.75 (0.49)
0.90 (0.73)
0.87 (0.65)
20% (51%)
3% (12%)
Correlation with GDP growth q-o-q (in brackets: leading 1 correlations)
increase
compared to
PMI
15
Conclusion:

BCS data can be used to construct indicator tracking q-o-q
GDP growth satisfactorily

key of the approach: consider not only the (average) values
of the balance series, but also the amount of series moving
up/down

approach is still in its infancy and needs further testing
16
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth
Shortcomings of the approach
Calculations are done with quarterly averages of BCS questions
>>>> indicator could only be published once a quarter
Is it possible to apply the same construction method to monthly
data?
technically yes !
Will quarterly averages of the resulting monthly ESI-series
remain well-correlated with q-o-q GDP growth?
yes:  even slightly higher
correlations with GDP q-o-q

However…
correlation of the two
quarterly ESI series is at
0.97
17
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth – Shortcomings of the approach
Problem of constructing ESI with amplified changes for
monthly data:
Sep 2009: in upswing-period, ESINEW with
amplified changes drops by 10 points
(=1 standard deviation)
too high volatility
125
115
105
95
85
75
65
55
ESINEW - amplified changes (monthly levels)
current ESI (monthly levels)
18
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth – Shortcomings of the approach
Source of volatility (note: amplifying changes does not only increase the
amplitude of the series, but also its volatility):
 If amplification is applied in t-1, but not in t, ESI will usually suggest a
drop in sentiment in t (also in case the underlying data continues the
upward/downward trend of t-1)
105
104
103
102
101
100
99
98
97
96
95
+2*3 = +6
unamplified ESI
+2
Jan


+2
Feb
ESI (amplified change): assuming
8 questions go up in Feb
+2
Mar
Apr
This additional volatility improves the fit of our quarterly series, but
renders the monthly series TOO volatile.
Main reason for this difference: criterion for amplification is more
restrictive in case of quarterly set-up:
 for quarterly: >= 8 questions must have gone up/down over 3
months-period (amplification in 63% of quarters)
 for monthly: >= 8 questions must have gone up just one month
19
(amplification in 73% of the months)
4.
Improving the ESI's tracking performance of q-o-q GDP growth – step b: an ESI with amplified growth – Shortcomings of the approach
Solution:
 When multiplying change of t (compared to t-1), the
resulting amplified change should be added to ESI for
month t, but also ESI of t+1 and t+2 (1/3 of the change
respectively should be added).
Approach smoothens the monthly curve substantially…
125
115
105
95
85
75
65
55
ESINEW - amplified changes (monthly levels)
ESINEW - amplified changes distributed over 3 months (monthly levels)
When constructing quarterly averages, correlations
with q-o-q GDP growth remain high.
20