Presentation is loading. Please wait.

Presentation is loading. Please wait.

A good measure of productivity Eric Bartelsman Vrije Universiteit Amsterdam and Tinbergen Institute Washington, World Bank, October 31, 2005.

Similar presentations


Presentation on theme: "A good measure of productivity Eric Bartelsman Vrije Universiteit Amsterdam and Tinbergen Institute Washington, World Bank, October 31, 2005."— Presentation transcript:

1 A good measure of productivity Eric Bartelsman Vrije Universiteit Amsterdam and Tinbergen Institute Washington, World Bank, October 31, 2005

2 Overview  Who Cares about Productivity? Why? Which measures?  Sources of Productivity Data  Designing a Research Agenda  Statistical Roadmap  Policy Questions

3 Information for Policy  Appropriate indicators depend on policy problems  System of National Accounts (SNA) developed by Stone, provides information to track Keynesian problem of demand shortfall  Schumpeterian view on growth (and cycles) lacks systematic framework of data collection and presentation

4 Information for Policy  Monetary policy requires:  Indicators of potential output  Supply side: labor, capital services, TFP  Indicators of resource bottlenecks  Unit labor costs  Indicators of state of business cycle  Cyclicality of productivity, prices, wages  Research to understand economic mechanisms  Many uses of productivity information

5 Information for Policy  Structural Economic Policy  Competition, Privatization, Deregulation  Innovation, R&D  Resource markets (unemployment, labor quality, financial markets, imported inputs)  …needs performance indicators  International/sectoral productivity comparisons  Comparisons over time  Proper ‘evaluation’: treatment/control groups

6 Which measure of productivity  This is a practical issue and depends on question at hand and on availability/reliability of data  Some findings are robust across measures, while other depend crucially on measure used

7 Importance of Measurement  Productivity is output/input  Measurement errors do not cancel  Output or input measures: apples and oranges problem (across variables, across units, over time)  Practical problems:  collection of data often not ‘consistent’  Data often not integrated, or integrated in different ways over time or across units

8 Overview  Who Cares about Productivity? Why? Which measures?  Sources of Productivity Data  Designing a Research Agenda  Statistical Roadmap  Policy Questions

9 Data Sources  Level of aggregation  Macro: National Accounts and Labor Force Survey  Micro: firm or establishment census or surveys  Sectoral: Top-down, bottom-up, integrated  Ideally, data would be consistent and integrated at each level of aggregation

10 INPUT-OUTPUT TABLE INDUSTRIESFINALDEMN.GRS.OUT. IND VALADD GDP or GDI

11 Consistent, Integrated Dataset  See Beaulieu and Bartelsman (2005)  Dataset  detailed information on the gross output, value added, final demand expenditures, and use of intermediate inputs by industry  Consistent  dataset where the underlying components are based on the same definitions and industry classifications  Integrated  despite the numerous data sources employed, the estimates conform to the accounting identities linking production, income, and expenditures

12 Comparisons of NA vs micro sources

13

14 Sources of labor data  Labor input not in SNA (sometimes in satellite acct)  Main source: household survey  Employment and earnings surveys (excl small firms, often mfg, mng, util)  Business Statistics (survey or census)  Business register  Social Insurance register

15 Comparisons of sources of labor data Source: Productivity growth in Jamaica 1991-2000, Eric Bartelsman, report for WB

16 Sources of price data  Ideal: deflators for each cell of IO; deflators for outputs and inputs at micro level  Aggregate or sectoral deflators: ‘Schreyer method’  Expenditure PPPs vs Ind-of-origin PPP  For micro level: assume constant markup and use nominal sales

17 Overview  Who Cares about Productivity? Why? Which measures?  Sources of Productivity Data  Designing a Research Agenda  Statistical Roadmap  Policy Questions

18 Statistical Roadmap  For EU countries: EU KLEMS  STAN for OECD, multiple sources for US  Other countries: stimulate similar work, but process directly from original datasource  Try to harmonize types of original sources available, and methods used to create consistent integrated data

19 Provision of metadata. Approval of access. Disclosure analysis of cross-country tables. Disclosure analysis of Publication Researcher Policy Question Research Design Program Code Publication Network Metadata Network members Cross-country Tables NSOs Distributed micro data research

20 Recent Productivity Research: does it answer policy questions?  Labour Productivity Levels  cross-country, sectoral timeseries  Growth Accounting  Contribution of ICT, other capital, TFP  Statistical Analysis  Cross-country convergence (million regressions)  Sectoral and micro-level datasets

21 Convergence in Labor Productivity Levels  What drives convergence process?  Many factors are significant in cross-country growth regressions  Do we really know if 10% above or below is meaningful  international price comparisons  definitions of output and input  Why converge to ‘average’ of best country?  But: we learn who belongs to which ‘convergence clubs’

22 Growth Accounting Source: van Ark & O’Mahony 1995-2000: contribution in pct-points to average annual growth

23 What was the question?  No role for policy environment or spillovers  Only factors that are explicitly purchased can contribute to output in the framework  No indication of market failures  It is assumed that representative firm makes optimizing decisions  No accounting of source of higher quality inputs  Contribution of quality of capital or labor to output can be computed  Does contribution of input X (e.g. ICT, or foreign- owned capital) differ across countries/sectors/time?  And: most of output growth is accounted for

24 Sectoral Comparisons Source: van Ark & O’Mahony 1995-2002: contribution in pct-points to average annual growth

25 Sectoral and micro data  Econometric evidence in panel data setting  R&D  Shifting of R&D based on tax-incentives  Spillover benefits of R&D done abroad  ICT  Firm-level evidence on return to ICT investment  Differences by Ownership status  Corporate Taxes  Cross-country ‘productivity’ response to relative tax- rate shifts  Policy evaluation of ‘innovation vouchers’  Many other studies: trade-openness, competition, training


Download ppt "A good measure of productivity Eric Bartelsman Vrije Universiteit Amsterdam and Tinbergen Institute Washington, World Bank, October 31, 2005."

Similar presentations


Ads by Google