Workshop on European Composite Indicators

1
OECD composite indicators pros and cons,
weighting, sensitivity, updating issues

• Workshop on European Composite Indicators
• Brussels, 24th-25th October
• Paul Conway

2
The plan

• The indicators of product market regulation (PMR)
• Economy wide
• Rich coverage of regulation data but limited time
  dimension
• Random weights technique
• The indicators of regulatory reform (regref) and
  regulation impact
• Sectoral
• More limited coverage of regulation data but runs
  from 1975 to 2003

3
The pros and cons of CIs

• Trade off between the richness of information and
  ease of communication.
• Effective tool for communicating with
  policymakers and the public.
• Also essential in empirical work on the linkages
  between policy and performance.
• Combining information from diverse areas risks
  becoming meaningless.

4
The indicators of Product Market Regulation (PMR)

• Measure the degree to which policies promote or
  inhibit competition
• Policy focused ie, not based on opinion surveys
  or market outcomes
• Economy wide indicators
• Summarise information on 139 economy-wide or
  industry specific regulatory provisions
• Calculated for 1998 and 2003

5
The PMR indicator system
6
The PMR indicators the weighting technique

• Low-level indicators simple or weighted
  averages. Occasionally weights are derived from
  other indicators.
• Mid- and upper-level indicators are calculated
  using factor analysis.
• Data revisions and updating is problematic given
  that the weights and indicator groupings change.

7
PMR in 1998 and 2003
8
Uncertainty

• No matter what technique is used weights are
  effectively value judgements.
• Stochastic weights
• Technique developed by Anders Hoffmann ex-OECD
  and a co-author of the Handbook on Constructing
  Composite Indicators
• Generate 10000 sets of random weights (each of
  which sums to 1)
• For each country calculate 10000 PMR indicators
  using random weights
• Look at the distribution of indicator values and
  country rankings

9
The PMR indicators random weights
10
The PMR indicators random weights
11
The PMR indicators random weights
12
Interpreting confidence interval comparisons

• Should compare indicators calculated using the
  same weights
• Comparisons assume confidence intervals proxy all
  uncertainty
• Data collection
• Missing data treatment
• Indicator construction etc. etc.

13
The PMR indicators probabilistic rankings
14
The PMR indicators probabilistic rankings
15
Indicators of PMR heterogeneity
16
The regref indicator system
17
Regref and PMR 1998
18
Regref and PMR 2003
19
Regref by country
EURO-zone
20
Pace and scope of reform in different industries
21
The regulation impact indicators

• Given the increasing share of services,
  regulation in this sector has a substantial
  impact on other sectors
• The size of that impact is also a function of the
  linkages between the service sector and the rest
  of the economy
• These linkages can be quantified using
  input-output tables and the Leontief transform
• Reg impact indicators can be constructed using
  these data (Faini et al 2004)

22
The reg impact indicators
Sector j one of the sectors included in the
regref indicators w jk total dependence of k on
sector j
23
Box plot of reg impact by sector, 2003
24
Box plot of reg impact by country, 2003
25
Regulation impact in ICT-using, ICT-producing,
and non-ICT-intensive sectors, 2003
26
Summary of results on PMR and labour
productivity linkages

• To varying degrees, anti-competitive regulation
  lowers labour productivity growth by
• slowing the diffusion of productivity shocks
  across countries
• Curbing ICT investment
• Reducing the penetration of foreign affiliates
• Slowing firm entry