Conference on next steps for the Japanese SNA

Tokyo, 25 March 2005

Measuring Productivity
Paul Schreyer, OECD

1
 Contents of presentation
1. Why is productivity important?
2. Concepts: labour and multi-factor productivity
3. Measuring output
4. Measuring labour input
5. Measuring capital input
6. OECD Productivity database
7. OECD Factbook
8. Conclusions
2
“Productivity isn’t everything
but in the long run it is almost
everything”
Paul Krugman The Age of
Diminishing Expectations

3
 1. Why is productivity important?
• Basis for improvements in real incomes and
economic well-being:
• Monetary policy (inflationary pressures)
• Fiscal policy (financing of health, education,
welfare)
• Slow productivity growth = conflicting demands
for distribution of income more likely

4
 2. Concepts: labour productivity

• Every measure of productivity is a ratio between output

and input(s)
• Simplest and most frequently-encountered measure: labour
productivity:

Q
Labour productivity (LP) =
L
• Indicates how efficiently labour is used in production
• Not necessarily an indicator of effort per worker

5
 Table of productivity measures

Type of input measure

Capital, labour &
Type of output intermediate inputs
Labour Capital Capital & labour
measure: (energy, materials,
services)
Labour productivity Capital productivity Capital - labour MFP
KLEMS multi-factor
(based on gross (based on gross (based on gross
Gross output productivity
output) output) output)

Labour productivity Capital productivity Capital – labour

Value-added (based on value- (based on value- MFP (based on -
added) added) value-added)

Single factor productivity measures Multi-factor productivity (MFP) measures

6
 2. Concepts: multi-factor productivity

• Ratio of output and combined capital and labour input

Q
Multi − factor productivity (MFP ) =
F(L, K )

• Indicates how efficiently combined labour and capital are

used in production
Cost share
• When measured as rate of change: of capital
Cost share
of labour

 Qt   Lt   Kt   At 
ln t −1  = s L ln t −1  + s K ln t −1  + ln t −1 
Q  L  K  A 
% change % change % change % change
in output in labour in capital in MFP
input input 7
 MFP and labour productivity

• Useful tool – decomposition of labour productivity

 Qt   Lt    Kt   Lt    At 
ln t −1  − ln t −1  = (1 − s L ) ln t −1  − ln t −1   + ln t −1 
Q  L   K   L  A 

% change labour Effect of capital

Effect of MFP
productivity intensity

8
 MFP and labour productivity

• Decomposition of labour productivity important:

• Is labour productivity driven by investment in
‘traditional’ capital (--> rise in capital intensity)?
• Is labour productivity driven by investment in
‘intangible’ capital, i.e., innovation, organisational
change, R&D (--> change in MFP)
• Example: Canada

9
Decomposition of labour productivity, Canada

10
MFP – framework

11
 MFP framework: measurement issues

Tricky areas:
-Implicit pricing
-Non-market
services
-Quality change

12
 Measuring output: 3 tricky areas
1) Output measures for industries where there is
implicit pricing so that even nominal output
measures are debated:
• Banking (FISIM, how to deal with new
financial instruments and risk?)
• Insurance (how to deal with catastrophes?)
Î OECD Taskforce recommends to measure
insurance output as premiums minus expected
claims, not actual claims
Î Otherwise sharp drop in output when there is a
13
catastrophe
 Measuring output: 3 tricky areas
2) Output measures for industries where technology
moves fast and where rapid quality change makes
price measurement difficult
• Example ICT: computer production,
communication services
3) Output measures for industries with non-market
producers
• Traditionally from input side
• But increasingly demand for output-based
measures Î Atkinson Report UK
14
 Measuring output: 3 tricky areas
Chart 5. Real value added per employed person in the health and
social work industry
Indices, 1982=100

France Italy Japan United States

140

130

120

110

100

90

80

70
1982 1987 1992 1997
15
Source: OECD.
 Multi-factor productivity framework:
measurement issues

Comparability of
GDP growth
rates:

-Non-market Measuring user

services costs
-High-tech
industries Splitting mixed
income
16
 Multi-factor productivity framework:
measurement issues

-Hours actually
Comparability of worked
GDP growth
rates: -Hours worked
by
-Non-market -Measuring user occupation/skill
services costs
-High-tech -Capital
industries -Splitting mixed measures by
income industry
17
 4. Measuring labour input
• Ideal measure: total number of hours worked,
broken down by type of labour input
• Number of persons employed or number of jobs
can generate biased measures of productivity if
hours per person change or if there are multiple
job holdings
• Total hours should comprise the hours of
employees, the hours of self-employed and the
hours of unpaid workers (e.g. family members in
agriculture)
18
 Example: impact of different measures of labour
input on labour productivity, France
Industry (mining, manufacturing and construction) Market services

220.0 145.0
Value-added per Value added per
140.0
hour: 3.7% per hour: 1.8% per
200.0
year 135.0 year
180.0 130.0
Value-added per
employee 125.0
Value-added Value-added
160.0 (headcount): 3.3%
per full- 120.0 Value-added per per full-
per year
time person employed time
140.0 equivalent 115.0 (headcount): equivalent
person: 1.3% per year person:
Value-added per 110.0
120.0 3.4% per Value-added per employee 1.5% per
person employed
year 105.0 (headcount): 1.0% per year
(headcount): 3.4%
year
100.0 per year 100.0
78

80

82

84

86

88

90

92

94

96

98

78

80

82

84

86

88

90

92

94

96

98
19

19

19

19

19

19

19

19

19

19

19

19
19

19

19

19

19

19

19

19

19

19
*Output is me a s ure d a s a qua ntity inde x of va lue -a dde d.
Source: INSEE.

19
 Labour input: statistical issues

1) Two main statistical sources for hours worked:

• Labour force surveys (LFS)
• Establishment or firm-based surveys (ES)
• Results are not normally consistent
Î Important role for national accountants:
Bring together these sources and provide labour
input measures by industry (and sector) that are
consistent with output measures
20
 Labour input: statistical issues
2) Splitting mixed income:
• Important for productivity measurement: how
big is labour income?
• To date: crude approximations are made
Î National accounts could help by putting
forward higher quality estimates
3) Information on composition of hours worked and
relative remuneration:
• What is skill composition of hours?
• What is the wage difference between21
skill levels?
 5. Measuring capital input

• Conceptually correct measure: capital services

(Theory see Griliches, Jorgenson, Diewert)
• Capital services = flow of productive services that
capital delivers in production
• Measurement assumption: capital services = fixed
proportion of productive capital stock

22
Productive capital stock of a particular asset type

K t = I t + h1I t −1 + h 2 I t − 2 + ... + h T I t −T

Past Efficiency and

investment retirement parameter

23
Capital services of a particular asset type
Growth of capital services of a particular asset type =
Growth of productive stock of this asset type

 St   Kt 
ln  = ln  because S t = λK t
 St −1   K t −1 

Normally assumed as
constant

24
 Price of capital service = user costs

How much would the owner of a capital good charge

in a competitive market to rent the capital good out for
one period (Diewert 1974)?
u t = (rt + δ t − i t )p t
Rate of asset
price change

Rental price
or user cost Rate of
Required rate depreciation
or return

25
 Price and quantity of capital services

Value of capital services for asset i = u it Sit

User costs and the rate of capital service for each asset
type are brought together to form an index of overall
capital input = the rate of change of total capital
services
 Kt   K t

ln t −1  = ∑iw i ln t −1 
t i

K   Ki 
Average share of each
asset in total value of
capital services
26
Capital services and net capital stock, Australia

27
 Capital services and national accounts

Capital services are different from net and gross capital

stock
But different does not mean unrelated
Desirable: consistent entity of
•Net capital stock and balance sheets
•Productive stock and capital services
•Consumption of fixed capital (=depreciation)
This is one important topic of SNA revision, discussed
next week in Canberra by Canberra II Group 28
 Capital services and national accounts

Some measurement issues:

•Availability and quality of investment series
•Breakdown by type of asset important
•Scope of assets
•Investment deflators

29
 6. OECD Productivity database

•Since March 2004, OECD productivity data available

on the internet:
•Productivity:Statistics Portal
•For total economy:
•Growth of GDP, hours worked and capital services
•Labour productivity growth
•MFP growth
•Labour productivity levels
30
 % change per year

0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
D Ita 5.0
en ly
m
ar
Ja k
p
B e an
lg
Sw ium
e
G de
N er m n
U eth an
ni
te erl y
d a
Ki nd
ng s
do
U Ca m
ni

1990-95
te nad
d
St a
at
F r es
an
1995-01 Po ce
rtu
A u ga
st l
ra
G lia
re
e
Average rates of MFP growth

Fi ce
nl
an
Ir e d
la
nd
31
 7. OECD Factbook

•Since last week: OECD Factbook

•Online version is freely available at
http://new.SourceOECD.org/factbook and enables the
downloading of the data underlying the tables and
graphs
•First compilation of economic, social and
environment indicators across OECD countries
•Chapter on macro-economic indicators includes
productivity comparisons, in particular MFP growth
rates
32
 8. Conclusions
•Productivity is a key indicator for analysis of
economic growth --> significant demand from policy
makers
•Simple ratio but sometimes difficult to measure
•Desirable: close integration of productivity measures
and their ingredients with the national accounts
•Release of ‘official’ productivity statistics by
statistical agencies

33
Thank you for your attention!

34