Recent Data Work at the Groningen Growth and Development Centre

1
Recent Data Work at the Groningen Growth and
Development Centre

• Complementing PWT with capital (global, 1980)
• 10-Sector database (global, 10 sector, 1950)
• Historical national accounts (global, 10 sector,
  1870-1950)
• EU KLEMS Growth Accounts (OECD, 30 sector, 1970)
• EU KLEMS Level Accounts (OECD, 30 sector, 1997)
• World Input-Output Database (global, 30 sector,
  1995)
• International trade flows
• National Supply and Use Tables
• Environmental accounts
• Socio-economic accounts

2
INDUSTRY PRODUCTIVITY LEVELS by Robert
Inklaarand Marcel P. Timmer Groningen Growth
and Development CentreUniversity of Groningen
The EU KLEMS project is funded by the European
Commission, Research Directorate General as part
of the 6th Framework Programme, Priority 8,
"Policy Support and Anticipating Scientific and
Technological Needs".
3
Motivation

• Complementing EU KLEMS Growth and Productivity
  Accounts
• Industry perspective
• Why are levels interesting?
• Explaining differences in output (per hour)
  across countries
• Competitiveness (unit labour costs)
• Measure of technology distance to the frontier
• Differences in input use (ICT, skills, energy,
  business services)

4
Coverage

• For benchmark year 1997
• 28 industries plus aggregates as in EU KLEMS
• 30 countries as in EU KLEMS
• Extrapolation possible with EU KLEMS

5
WHATS NEW?

• Long tradition GGDC and NIESR (single deflated
  labour productivity mostly). This data set
  provides
• Multilateral KLEMS type comparisons
• More countries and industries covering whole
  economy
• Mix of production and expenditure PPPs
• Methodology for multilateral MFP from Jorgenson
  and Nishimizu (1978) and CCD (1982).
• Primal approach to retain consistent breakdown of
  real GDP
• Add sectoral output and intermediate inputs
  measures which allow aggregation across
  industries

6
Methodology

• Labour productivity
• Multi factor productivity

7
Nominal Data

• National Supply and Use tables, transformed into
  industry-by-industry IOT (45 industries) at basic
  prices
• Normalized on EU KLEMS value added and gross
  output
• Breakdown of intermediate inputs into domestic
  E,M,S and imports
• Breakdown of value added into
• labour compensation (30 types)
• capital compensation (8 types) and profit

8
Figure 1 Example of Aggregation with and without
integration in car manufacturing
9
Price Data

• Two alternative sets for output PPPs
• OECD/Eurostat PPPs for expenditure categories
• Expenditure prices, incl margins / net taxes /
  imports, excl exports
• Based on specified prices
• GGDC PPPs from production side
• Producer output prices, incl exports
• Based on unit value ratios
• Mix of both is ideal
• PPP-O agriculture / mining / utilities /
  manufacturing (partly) / communication /
  transport / trade
• PPP-E manu (partly) / construction / finance /
  business services / personal services

10
Price Data (2)

• Intermediate input PPPs based on output PPPs
  weighted with shares in total intermediate input.
• Labour PPPs based on relative wages (5x2x3 types)
• Capital PPPs based on relative rental prices (8
  types)
• Non-market services Implicit output PPP based on
  input PPPs such that MFP1

11
Some Results

• FIG 1 Value added per hour worked, 2005, US1
• FIG 2 Sensitivity of LP
• Single versus double deflation of value added
• Set of output PPPs (expenditure PPPs versus mix)
• FIG 3 ICT and Non-ICT capital services per hour
  worked, 1997
• FIG 4 MFP, 2005, US1

12
FIG 1 Value added per hour worked, 2005, US1
13
FIG 1 Value added per hour worked, 2005, US1
14
FIG 2 Sensitivity of LP
15
FIG 2 Sensitivity of LP
16
FIG 3 ICT and Non-ICT capital services per hour
worked, 1997, market economy
17
FIG 4 MFP value added based, 2005, US1
18
Concluding remarks

• Industry level accounts are feasible
• Importance of sectoral output measures
• Further research
• DATA
• Hours worked
• Intangibles
• Recent benchmark year
• METHOD
• Ex-ante measures of capital and loss of duality
• Time-space consistency with more benchmarks
• CCD and heterogeneity role for MST?
• More data, more noise?