Recent developments
in productivity
measurement
Paul Schreyer
OECD Statistics Directorate
Canberra, 20 November 2012
1
Introduction
• Productivity = output/input
• Issues:
– Identifying, measuring and aggregating
inputs and outputs
– Level of measurement (economy, industry,
firm)
• Academic community dealing with
productivity measurement and analysis
• World KLEMS network
• NSOs: no clear trend
2
This presentation
1. Bringing nature into the productivity
picture
2. The firm level: productivity measurement
with micro-data
No claim for comprehensive presentation of
recent developments
3
Bringing nature into the productivity
picture
4
Bringing nature into the picture – input side (1)
• Typical inputs: labour, produced capital,
intermediate inputs
• Often neglected: non-produced natural assets:
– Mineral resources
– Soil/land
– Timber
– Aquatic resources
– Water
5
Bringing nature into the picture – input side (2)
• Why important?
– Assessing contribution of natural assets to
economic growth
– Measuring productivity correctly
– Policy implication: is growth driven by MFP or
by natural assets
– Note: without measurement, direction of bias
unknown
6
Volume index of subsoil asset removals, Australia,
1989=100
250,0
200,0
150,0
100,0
50,0
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
0,0
Source: OECD calculations, based on ABS data.
7
No unambiguous direction
Effect of including natural resource input on
measured productivity growth:
• Traditional MFP > adjusted MFP if :
– natural resource input growth > traditional
input growth
– i.e., total input growth has been understated
– i.e., traditional MFP growth has been
overstated
• And vice versa
8
Norway – Difference between adjusted and
traditional MFP growth
4
3
Difference in percentage points
2
Traditional MFP overstated
1
0
-1
-2
Traditional MFP understated
-3
-4
1985
1990
1995
2000
2005
Time
Source: OECD, work in progroess.
9
Challenge: quality of natural resource input
• Capture changing marginal extraction
costs (which may be increasing)
• Capturing changing quality in the
resource itself eg declining soil quality
 failing to do so will overstate measured
contribution of natural resource to output
and understate MFP
10
Effects on productivity measures:
Australia’s mining industry
• Study by Productivity Commission (Topp,
Soames, Parham, Bloch 2008):
• Similar in spirit except that mining output
is adjusted for declining yields
• Underlying rate of productivity growth is
around 2.5 per cent p. a., compared with
stagnant standard MFP (1974 to 2007)
Natural resource input has grown less
quickly than other inputs, so MFP was
understated by traditional measure
11
Bringing nature into the picture – output side (1)
• Production processes often accompanied
by undesirable outputs, e.g., emissions
• From producer and MFP measurement
perspective:
• Relevant in presence of environmental
policies:
– explicit price (e.g., tax) or
– implicit price (marginal abatement costs due
to regulation)
• Are traditional MFP measures over-or
understated?
12
Again, no unambiguous effect on measured
productivity (1)
Example:
• Given inputs (labour, capital,…)
• Rising traditional output
• Constant emissions
adjusted MFP > traditional MFP
Productivity growth was required to keep
emissions at bay
13
Again, no unambiguous effect on measured
productivity (2)
• But overstatement of traditional MFP if
emissions grow quicker than traditional
output
• For many pollutants (NOx, Sox, CO2,…)
relative decoupling in many OECD
countries
 Understatement of traditional MFP
14
Private and social valuation
• Producer perspective = private
valuation
– marginal abatement cost for producer
• Welfare perspective = social valuation
– marginal cost to society = producer costs +
consumer costs + externalities
• Both perspectives meaningful but should not
be mixed up
• If productivity measurement is based on
producer theory, producer perspective is
called for
15
OECD work in this area…
• As part of green growth indicator work
– MFP adjustment with natural asset inputs
– MFP adjustment with undesirable outputs
– Index of natural resources
16
Important international development:
SEEA
• System of Integrated Environmental and
Economic Accounts
• Adopted at UN level in 2012
• Consistent accounting for environmenteconomy interaction
• Basis for indicator work
• Unifying element: balance sheets
– Stocks, additions, removals
– Physical and monetary valuation
• Major task ahead: implementation
17
The firm level: productivity
measurement with micro-data
18
Firm-level measurement
• Drawbacks
– No prices, capital proxy, employees,
incomplete sector coverage, short time-spans
• Avantages
– Entry, exit, reallocation
– Within-firm cycle/growth
– Understanding/measuring both firm-level
levers and environmental factors driving
growth
19
Stylised facts from micro estimates (1)
• Huge productivity dispersion
– Even within very narrowly defined industries
– Firm size plays an important role
– But how accurately are outputs measured?
20
UK: Labour productivity by firm size
Indices: 2007 = 100
110
Large enterprises (250+
employees)
105
100
95
90
Medium businesses
(50-249 employees)
Small businesses (0-49
employees)
85
80
75
2003
2004
2005
2006
2007
2008
2009
2010
Source: J. Saleheen, Bank of England 2012
21
Stylised facts from micro
estimates (2)
• Reallocation or resources to highproductivity producers important
• Competition—consumers can easily switch
suppliers
• Labor and capital market flexibility
• Summary measure of reallocation:
correlation between productivity and
market share
22
Correlation between Productivity and Market
Share
0.6
0.5
0.4
0.3
0.2
0.1
re
a
Ko
wa
n
Ta
i
S
U
hi
le
C
rtu
ga
l
e
Po
nc
Fr
a
U
K
Ar
ge
nt
in
N
a
et
he
rla
nd
s
G
er
m
an
y
0
Source: Ch. Syverson November 2012
23
Firm-level measurement requires
dealing with…
• Large volumes of data
• Confidentiality issues
– Small countries
– Narrowly defined industries
• No international standards – reduced
comparability
• NSOs have taken up issue
24
Conclusions
25
Conclusions (1)
• Nature of productivity implies
cumulation of measurement challenges
• Quality of source data (national accounts,
firm-level data) key
• Integrating productivity measurement
into official statistics important but not
yet widespread
26
Conclusions (2)
• Tricky output measurement in particular in:
– Financial services
– Health, education, general administration
– Undesirable outputs
• Tricky input measurement:
– Hours worked by industry and by skills
– R&D capital (new in national accounts)
– Natural capital
• Intangibles
27
Thank you!
28