Transcript Use of surveys

Use of surveys: the case of
inventories
Presentation to a DG-ECFIN Workshop
12/13 October 2009
Philippe de Rougemont,
Euro Area Macroeconomic Development (EAM)
Directorate General Economics
The views expressed are solely those of the author, and do not
necessarily reflect those of the ECB
Menu
• Part 1 Use of surveys for conjunctural
analysis
• Part 2 Inventories in this recession
• Part 3 Surveys on inventories
• PMI
• DG-ECFIN
• Comparison, joint use
• Part 4 Inventories in national accounts
• Link with surveys
2
Part 1
Use of surveys for
conjunctural analysis
3
Use of surveys (1)
• Traditionally:
– “Good”: (1) timeliness, (2) reliability, (3) frequency, (4)
flexibility
– “Bad”: (1) coverage, (2) not integrated, (3)
subjective/qualitative
• But:
– (a) Surveys can provide “objective” information
– (b) Surveys can be seen as integrated
• “One respondent”
– (c) Surveys provide a competitive view on economic
development
• Ex. Puzzle: volatility of hard data on production versus surveys
4
Use of surveys (2)
• Subjective/qualitative
– Nature of the question
•
•
•
•
Forecast (“What is your expected production?”)
Sentiment (“Are you optimistic?”)
Assessment (“How is your order book compared to normal?”)
Report of an observation (“Did your past production change?”)
=> objective
» Stocks versus flows
• Other issues: seasonality, working days, concepts
– Response practices of respondents
•
•
•
•
Modality of response: discrete rather than continuous information
Band corresponding to no change
The genuine “reporting” period considered
Other issues (same): seasonality, working days, concepts
Work of the CBI on respondents’ practices very useful
– Users’ responsibility
• Often linear link presumed (between a survey balance and a change
of an indicator)
5
Use of surveys (3)
• Two uses of surveys:
– Independent narrative
• Briefings / Monthly Bulletin / Annual Report
– Input to “Mechanical tools” for “backcasts, nowcasts, forecasts” of
GDP. See Monthly Bulletin April 2008
• Importance of the horizon of estimate
• Bridge equations
– ECB WP 276, 622, 925, 975
• Factor models: substantial response on survey releases
– ECB WP 894, 949, 953
• Two methods:
– “Accounting approach”: methodological perspective; focusing on the
survey questions directly relevant (GDP nowcast  questions on
past production)
• Frequency transformation; natural average (1,2,3,2,1)
• A priori lags; but respondents’ practices
– “Business cycle approach”: empirical perspective; all survey
questions can legitimately be used: common factor
• Factor models; no a priori lags
6
Use of surveys (4)
• Avoiding misuses and communication mistakes (e.g.):
– Frequency transformation: erroneous to average monthly
surveys over the quarter without justification
– Avoid picturing indicator’s year-on-year growth with monthly
surveys
– Consider the lagging/smoothing effect of stock indicators and
of the relevant “reporting” period
• Avoid PMI/DG-ECFIN comparisons without explanations!
• A conceptual issue
– Are “month on month” survey questions superior to “3 month
on 3 month”?
– Superior information content:
• e.g.: mid November, 6/9 of the information for Q4 growth is
available
– Extraction of signal problem: (1,2,3,2,1) polynomial cannot be
inverted
7
Use of surveys (5)
• In the recent recession:
– The slowdown and then collapse in activity
were apparent in surveys, reasonably early
– However, the amplitude of the collapse was
not well captured by surveys
• Non linear effects
– Some econometric evidence
– Conceptually perhaps ambiguous
• Smoothing effects
– Some econometric evidence too
– Now, an upturn (?)
8
Use of surveys (6)
PMI Manufacturing "World", US, e uro are a, Japan and UK- up to Septem ber
2009
60
O utput PMI "World", US, euro area, Japan and UK - up to September
2009
70
65
55
60
55
50
50
45
45
40
40
35
World
35
30
UK
Japan
25
Euro area
30
US
20
Ja
n
Ju
l-0
9
O
ct
-0
9
Ju
l-0
8
O
ct
-0
8
Ja
n09
A
pr
-0
9
Ju
l-0
7
O
ct
-0
7
Ja
n08
A
pr
-0
8
06
Ju
l-0
6
O
ct
-0
6
Ja
n07
A
pr
-0
7
-0
6
pr
A
-0
6
Ap
r-0
6
Ju
l-0
6
Oc
t-0
6
Ja
n07
Ap
r-0
7
Ju
l-0
7
Oc
t-0
7
Ja
n08
Ap
r-0
8
Ju
l-0
8
Oc
t-0
8
Ja
n09
Ap
r-0
9
Ju
l-0
9
Oc
t-0
9
15
25
Ja
n
World
UK
Japan
Euro area
US
Sources: Markit, ISM
9
Use of surveys (7)
Euro area industrial production and PMI
2.0
63
Industrial production
PMI output
58
1.0
53
0.0
-0
8
A
pr
-0
8
Ju
n08
A
ug
-0
8
O
ct
-0
8
D
ec
-0
8
Fe
b09
A
pr
-0
9
Ju
n09
A
ug
-0
9
-1.0
Fe
b
D
ec
-0
7
48
43
38
-2.0
33
-3.0
28
-4.0
23
Sources: Eurostat, Markit
Last observation in September 2009 for PMI and in July 2009 for industrial production
10
Part 2
Inventories in this
recession
11
PMI diffusion indexes on inventories (euro area)
(1) destocking movement from summer 2008 in retail and in manufacturing inputs; (2)
and from December 2008 in manufacturing finished goods; (3) destocking speed
maximum in March-June 2009; (4) slowing since May-August 2009
PMI - Inventories in manufacturing (euro area)
PMI - Inventories in manufacturing and retail (euro area)
54
58
52
56
50
54
48
52
Total manufacturing
Retail
Total
50
46
48
44
46
42
44
40
36
Input
Finished goods
Total
Au
g97
Au
g98
Au
g99
Au
g00
Au
g01
Au
g02
Au
g03
Au
g04
Au
g05
Au
g06
Au
g07
Au
g08
Au
g09
34
Last observation in September 2009
Source: Markit
40
38
36
Au
g97
Au
g98
Au
g99
Au
g00
Au
g01
Au
g02
Au
g03
Au
g04
Au
g05
Au
g06
Au
g07
Au
g08
Au
g09
38
42
12
Inventories in an economy (I)
Producer A
Retailer R
Finished
goods
Input
goods for
resale
Producer B
Wholesaler W
Finished
goods for
goods
resale
Input
Consumers
Imports
Exports
13
Inventories in an economy: scramble for cash (II)
Deepening of the financial crisis
Producer A
Retailer R
+
Finished
goods
-
-
Input
goods for
resale
Producer B
Wholesaler W
+++
Finished
goods for
goods
resale
-
-
Input
Consumers
Imports
Exports
14
Inventories in an economy: cutbacks of inventories
by cutting production (III)
Producer A
Retailer R
-
Finished
goods
-
-
Input
goods for
resale
Producer B
--
Wholesaler W
+
-
Finished
goods for
goods
resale
-
Input
Consumers
Imports
Exports
15
Levels of inventories – World, US, Japan, UK,
euro area
Inve n tori e s (of pu rch ase ) PMI "W orld", US, e u ro are a, Japan an d UK up to
Sept 2009
55
50
45
40
World
UK
Japan
Euro area
US
35
Last observation in September 2009
Sources: Markit, ISM
ct
-0
9
O
-0
9
Ju
l
09
pr
A
-0
9
Ja
n
O
ct
-0
8
-0
8
Ju
l
08
pr
A
-0
8
Ja
n
O
ct
-0
7
-0
7
Ju
l
07
pr
A
-0
7
Ja
n
ct
-0
6
O
-0
6
Ju
l
06
pr
A
Ja
n
-0
6
30
16
Destocking is a worldwide movement
• Worldwide destocking
=> Nature of the negative contributions of
net trade to growth in 2008Q3, 2008Q4 and
2009Q1 for the euro area
1. Is it merely destocking by partners; or genuine
weakening of their final demand?
2. What is the impact of the goods in transit?
• A kind of destocking at sea
• Net trade contribution to GDP must sum to zero
worldwide (please)
17
Part 3 Surveys on inventories
• PMI
• DG-ECFIN
• Comparison, joint use
18
PMI – the question on inventories
• PMI respondents are being asked about:
“The level of inventory (in units, not
money) this month compared with the
situation one month ago.”
• Answer around mid month
=> flow indicator, objectively depicting the
mid month to mid month change in the
volume of inventories
19
PMI-based euro area inventories – but showed in
levels (detrended)
PMI - Inventories in manufacturing and retail (euro area)
Estimated levels
PMI - Inventories in manufacturing (euro area)
Estimated levels
50
60
40
40
Input
Finished goods
Total
Total manufacturing
Retail
Total
30
20
20
10
09
08
Au
g-
07
Au
g-
06
Au
g-
05
Au
g-
04
Au
g-
03
Au
g-
02
Au
g-
01
Au
g-
00
Au
g-
99
Au
g-
98
Au
g-
97
-10
Au
g-
-20
0
Au
g-
Au
g-
97
Au
g98
Au
g99
Au
g00
Au
g01
Au
g02
Au
g03
Au
g04
Au
g05
Au
g06
Au
g07
Au
g08
Au
g09
0
-20
-40
-60
-30
-40
Last observation in September 2009
Sources: Markit and author’s calculations
Note: author’s calculations cumulating demeaned PMI results and then detrending and demeaning
20
PMI inventories – levels, changes and implied
contributions to growth (change in change)
Implied contributions of inventories to growth
based on PMI
(euro area)
Index and change in index – no unit
10
Level of
index
Change in
Change in change in
index
index*
5
3
20
09
Q
1
2
20
09
Q
20
09
Q
4
20
08
Q
3
20
08
Q
2
20
08
Q
4
1
20
08
Q
20
07
Q
3
0
20
07
Q
2007Q3
37
2007Q4
45
8
2008Q1
54
9
1
2008Q2
61
7
-2
2008Q3
62
2
-5
2008Q4
57
-5
-7
2009Q1
42
-15
-10
2009Q2
17
-24
-9
2009Q3
-3
-20
5
* The change in change in index is consistent with a
-5
contribution to growth
-10
Euro area – Last observation in September 2009
Sources: Markit and author’s calculations
The index, change in index, and change in change in index have no unit
Thus, the change in change in index needs to be rescaled to obtain a contribution of inventories to GDP growth
21
Implied contributions to growth – PMI based
calculations (no unit)
Total inventories
Contribution of inventories to growth
EA
DE
FR
IT
2008Q2
2008Q3
2008Q4
2009Q1
2009Q2
2009Q3
-2.3
-5.3
-6.9
-9.8
-9.2
4.5
-1.5
-4.4
-8.0
-10.0
-12.5
9.1
-1.1
-8.1
-12.0
-7.1
-7.8
8.2
-3.2
-0.3
-0.8
-11.9
-7.3
2.9
Manufacturing inventories
Contribution of inventories to growth
EA
DE
FR
IT
2008Q2
2008Q3
2008Q4
2009Q1
2009Q2
2009Q3
-0.8
0.7
-9.8
-12.2
-10.3
6.7
-0.4
3.5
-13.5
-15.1
-16.6
9.7
2.7
0.8
-19.1
-10.4
-4.0
8.9
-2.1
1.4
3.8
-8.5
-11.1
-0.7
ES
NL
-2.4
-3.8
-5.6
-11.4
-5.4
2.8
AT
-2.4
-4.3
-9.8
-14.3
-0.6
9.2
-4.9
-2.4
-2.4
-18.3
-10.2
9.6
Euro area – Last observation in September 2009
Sources: Markit and author’s calculations
The implied contribution to growth has no unit, and thus, needs to be rescaled to obtain a contribution of inventories to GDP growth
22
DG-ECFIN – the question on inventories
• Respondents are asked about: “Do you
consider your current stock of finished
products to be …?
+ too large (above normal)
+ adequate (normal for the season)
- too small (below normal)”
• Stock indicator depicting an assessment
• For retail: “Do you consider the volume of
stock you currently hold…”
23
DG-ECFIN survey on inventories (assessment of
stock levels): finished goods and retail
DG-ECFIN - Inve ntorie s of finishe d goods
(e uro are a)
DG-ECFIN - Inventories in retail
(euro area)
24
25
22
20
20
18
15
16
14
10
12
5
10
-0
9
Ja
n
-0
7
-0
5
Ja
n
-0
3
Ja
n
Ja
n
-0
1
-9
9
Ja
n
-9
7
Ja
n
Ja
n
-9
5
-9
3
Ja
n
-9
1
Ja
n
Ja
n
-8
9
Ja
n
-8
5
Ja
n
Ja
n
Jan-09
Jan-07
Jan-05
Jan-03
Jan-01
Jan-99
Jan-97
Jan-95
Jan-93
Jan-91
Jan-89
Jan-87
Jan-85
-8
7
8
0
Euro area – Last observation in September 2009
Source: DG-ECFIN
24
Comparing DG-ECFIN survey with PMI
-
Stocks (DG-ECFIN) versus flows (PMI)
Assessment compared to normal (DG-ECFIN) versus objective measure (PMI)
Detrended levels of inventories in retail (euro area)
using DGECFIN or PMI
ECB estimation
Detrended / demeaned levels of inventories of finished goods
in manufacturing (euro area) using DG-ECFIN or PMI
ECB estimation
20
10
10
5
50
25
40
PMI cumulation (lh)
DG ECFIN detrended (rh)
23
30
19
09
08
21
20
ug
A
ug
-
07
A
ug
-
06
A
05
ug
A
ug
-
04
A
ug
-
03
A
ug
-
02
A
01
ug
A
ug
-
00
A
ug
-
99
A
ug
A
ug
A
ug
A
-10
98
0
97
0
10
-5
17
0
-20
-30
-40
-10
PMI cumulation (lh)
DG ECFIN detrended (rh)
97
-10
ug
A
-15
A
ug
98
A
ug
99
A
ug
00
A
ug
01
A
ug
02
A
ug
03
A
ug
04
A
ug
05
A
ug
06
A
ug
07
A
ug
08
A
ug
09
15
-20
13
-30
11
-20
Euro area – Last observation in September 2009
Sources: Markit and author’s calculations
For manufacturing, the (linear) trend is obtained by minimizing the quadratic differences over August 1997 to September 2008
25
Voluntary or involuntary (de)stocking
PMI and DG-ECFIN surveys can be seen as providing complementary information
Their joint utilisation allows differentiating between voluntary and involuntary
(de)stocking behaviours
DGECFIN balances
+
PMI diffusion + Involuntary (stocking)
index
- Voluntary (destocking)
Voluntary (stocking)
Involuntary (destocking)
26
Stylised inventory cycle: contributions to GDP growth are
of opposite sign to the deviation of inventory levels to trend
Direct interpretation of DG-ECFIN survey for inventory contributions to GDP
growth
Given that:Y = C + Δ S
We have
ΔY /Y = Δ C /Y + Δ Δ S /Y
-
+
27
Part 4
Inventories in national
accounts
Link with surveys
28
Inventories in national accounts (UK example)
•
•
15% of GDP
Three “tiers” (about equal):
1. Manufacturing: input (a), output (b), work in
progress (c)
2. Trade: wholesale (a), retail (b)
3. Other: construction, energy, mining, agriculture,
services (yes)
=> Finished goods in manufacturing = 1/9 of total
•
Survey coverage
– DG-ECFIN covers 1b and 2b
– PMI covers 1a, 1b and 2b
29
Inventories in quarterly national accounts
• Inventories in Q-national accounts often
compiled as residual
• Seasonal and working days adjustments
• Chain-linking volumes
– No additivity of the GDP demand identity
– Absence of release of the change in
inventories in volume (by some statistical
institutes)
30
Changes in inventories and contributions of
inventories to GDP growth up to 2009Q2/Q3
Eurostat and PMI
Mar-08 Jun-08 Sep-08 Dec-08 Mar-09 Jun-09
National accounts data
Contribution of inventories to GDP growth
Changes in inventories - In % of GDP
Changes in inventories - In million euro
0.3
2.8
1681
-0.2
-0.2
-3053
0.4
0.2
4776
0.3
0.5
10082
1
9
-2
7
-5
2
-7
-5
PMI based data (no unit)
Contribution of inventories to growth
Changes in inventories
Sep-09
-0.6
-0.6
0.0
-0.7
-598 -12305
-10
-15
-9
-24
5
-20
Sources: Eurostat, Markit and author’s calculations
Note: Changes in inventories calculated by residual; chained-link volumes
31
PMI-based contributions to growth (direct
calculation) and national accounts measure
15
PMI-based contributions of inventories to growth
(manufacturing and retail no unit - lh) and national accounts
contributions of inventories to GDP growth (million euro - rh)
euro area
15000
-10
-15
-20
Aug-09
Aug-08
Aug-07
Aug-06
Aug-05
Aug-04
Aug-03
-5
Aug-02
0
Aug-01
0
Aug-00
5000
Aug-99
5
Aug-98
10000
Aug-97
10
-5000
-10000
T otal manufacturing
Retail
T otal
National accounts (rh)
-15000
-20000
Last observation in September 2009 and October 2009 extrapolated
Sources: Markit and author’s compilations; Eurostat
32
Surveys and national accounts
• Simple econometrics point at a better fit
using input inventories than finished goods
– Interpretation?
• Use of total inventories not superior
• Too small sample
• Perhaps a working days adjustment
difficulty
33
CONCLUSIONS
• Surveys have the potential to provide
“competitive” information on stocking behaviours
by business
– Good timeliness
– Crucial detail of inventories
– Residual nature of change in inventories in Q-national accounts
• Surveys have to be well understood
–
–
–
–
The importance to discriminate between inventories
The stocks versus flows issue
How to extract a contribution of inventories from PMI
How to link DG-ECFIN and PMI to extract a voluntary/involuntary
pattern (or movements in inventory targets)
=> Wish: surveys to have both flow and stock
information on inventories
34