Audience Guide

Thomas Roca

Description

• cours magistral - matière : geometry
• mémoire
• cours magistral
• leçon - matière potentielle : from astronomers
• exposé
• expression écrite
• cours magistral - matière potentielle : on poetry

Audience Guide Cast & Characters Exploration of the opera What To Listen For Galileo & Glass Want to Know More? January 26-28 • 8 PM | January 29• 2:30 PM

• powerful voices
• formal key structure
• great admirer
• daughters of a man
• galileo
• medical degree
• dowry
• young girl
• university
• family

Sujets

• Geometry
• Mémoire
• Cours magistral
• Exposé
• Célesta
• Demain.TV

Informations

• Publié par: thaut
• Nombre de lectures: 36
• Langue: English
• Poids de l'ouvrage: 1 Mo

Extrait

Working paper

DT/171/2012

A methodology for Comparing Governance
Database, Institutional Profiles Database
Robustness

An example using corruption data, from simple graph
representation to advanced econometrics

By

Thomas Roca
PhD – Lare-Efi, Groupe d’Economie du Développement – Université Montesquieu-Bordeaux IV















Avenue Léon Duguit - 33608 Pessac (France) - tél : [phone] - fax : [phone]
[e-mail] – [e-mail]
http://lare-efi.u-bordeaux4.fr – http://ged.u-bordeaux4.fr1 A METHODOLOGY FOR COMPARING GOVERNANCE DATABASE, INSTITUTIONAL PROFILES DATABASE ROBUSTNESS



A methodology for Comparing Governance Database, Institutional
Profiles Database Robustness

An example using corruption data, from simple graph representation to advanced
econometrics


Thomas Roca
PhD – Lare-Efi, Groupe d’Economie du Développement – Université Montesquieu-Bordeaux IV
([e-mail])


Abstract
As suggested by Arndt and Oman (2006), Governance indicators have recently blossomed to the extent
that it is no longer hyperbolizing to qualify governance assessment field as a “jungle”.
The question of the choice of relevant indicators thereby arouses for researchers studying institutions. While
governance indicators guides have already been produced [Arndt, C., Oman, C., (2006); UNDP,(2006)], we were
not able to get hold of a comprehensive and actionable methodology to compare rigorously the different
institutional measures currently available.
In this paper, aiming to assess the robustness of AFD ’s Institutional Profiles Database - developed in partnership
with CEPII and the Maastricht Graduate School of Governance - we propose different tools from simple graphic
representation to advance econometrics methods to question any governance indicator relatively to its
counterparts, population’s experience and as far as possible objective data.
We show that IPD’s evaluations, for instance, regarding the extent of corruption, appears much consistent with
Transparency International’s CPI and World Bank’s WGI. However, we also highlight this database specificity,
and report a few outliers. This singularity might result from differentiated perceptions, potentially illustrating a
“French bias” that, conversely, might as well reflect an “Anglo-Saxon bias”, nested in World Bank’s and TI’s
famous indicators.

Résumé
Comme le suggèrent Arndt et Oman (2006), les indicateurs de Gouvernance se sont récemment
multipliés, à tel point qu’il n’est plus désormais exagéré de qualifier le champ de la mesure institutionnelle de
véritable « jungle ». La question du choix des indicateurs les plus pertinents se pose alors pour qui veut étudier
les institutions. Bien qu’il existe quelques guides en la matière [Arndt, C., Oman, C., (2006); UNDP, (2006)],
nous n’avons pas été capable d’identifier une publication détaillant une méthodologie complète et opérationnelle
visant à comparer rigoureusement les différentes mesures institutionnelles actuellement disponibles.
Dans le but d’examiner la robustesse de la base Profils Institutionnels, développée par l’Agence Française de
Développement, en partenariat avec le CEPII2 et la Maastricht Graduate School of Governance, nous
proposons différents outils, de la simple représentation graphique à des techniques économétriques plus
élaborées, afin de comparer n’importe quel indicateur de gouvernance avec d’autres mesures existantes, émanant
d’autres sources (indicateurs concurrents, enquêtes ménages, et dans la mesure du possible, des données
objectives).Nous montrons alors que la base PI, dans le cas de la mesure de la corruption, apparait cohérente
avec les mesures fournies par Transparency International et la Banque mondiale. Néanmoins, nous soulignons
les spécificités de la base de l’AFD et identifions certains outliers. Cette singularité pourrait résulter de
perceptions différenciées illustrant un « biais » français, qui pourrait, inversement, tout aussi bien refléter un
biais anglo-saxon présent dans les données de ces célèbres concurrents.

Keywords: Corruption, Institutional Profiles Database, AFD, Global Corruption Barometer, Governance, CPI,
Transparency International, Corruption measurement, Perception indicators, Econometrics, Panel Data.

JEL classification: O11, O17, O19 2 WORKING PAPER N° 171
Content

I. Introduction ...................................................................................................................................................... 2
II. Graphic representations, towards a dashboard .......................... 4
A. Data clustered .............. 4
B. Comparisons using Gap estimators ............................................................................................................. 5
C. Towards more accuracy: representing every single country, fostering readability ..... 9
D. Combined Plot Box ................................... 13

III. Using Econometrics ....................................................................................................... 14
A. Pairwise correlations.................................. 14
B. Factorial map using Principal Components Analysis ................................................................................ 15
C. Using multivariate analysis ....................... 17
D. Testing French and Anglo-Saxon bias, Confidence and media bias ......................... 18

IV. Introducing Panel data ................................................................................................................................. 22
A. Methodology .............................................................................................................. 22
B. Corruption evaluations determinants, panel data ....................... 24
C. Testing biases, explaining measurement gaps ........................... 25

V. Concluding remarks ...................................................................................................................................... 27
VI. References ....................................................................................................................................................... 28
VII. Appendix ......................... 30




I. Introduction
As suggested by Arndt and Oman (2006), Governance indicators have recently blossomed to the extent
that it is no longer hyperbolizing to qualify governance assessment field as a “jungle”.
The question of the choice of relevant indicators thereby arouses for researchers studying institutions. Although
governance indicator's guides have already been produced [Arndt, C., Oman, C., (2006); UNDP,(2007a,b)], we
were not able to get hold of a comprehensive and actionable methodology to compare rigorously the different
institutional measures currently available.
In this paper, we will provide actionable examples for comparing governance indicators provided by different
sources. Governance databases such as the World Bank’s (WB) Worldwide Governance Indicators (WGI)
gather many indicators illustrating the fact that the governance field remains tremendously wide.

We decided to showcase corruption evaluations as this field attracted many research studies and measurement
attempts. Hence, we will compare corruption data gathered from three different sources depicting expert’s
evaluations of the corruption amount across countries and time.
Our three corruption assessment providers will be Transparency International (Corruption Perception Index -
CPI), the World Bank (WGI: Control of Corruption) and the French Development Agency’s Institutional
Profiles Database (Corruption control).
3 A METHODOLOGY FOR COMPARING GOVERNANCE DATABASE, INSTITUTIONAL PROFILES DATABASE ROBUSTNESS

The first question our reader may ask would be the following: why comparing governance assessments?
First of all, we need to stress that most of the governance indicators are based on perceptions (expert’s or
population’s). Furthermore, these measures are mainly constructed by Think tanks, National or International
Institutions, following their own objectives, embodying their own ideology. Thus, comparing governance
indicators and investigating their differences and similarities make sense for who wants to use them.

In this paper, we will try to provide a methodology for comparing governance indicators on different levels,
using differentiated techniques, from simple graphic representation to more elaborated econometrics. We
suggest that three levels of analysis may be performed:

- Internal comparison: comparing the different databases between each other’s;
- External comparison: comparing indicators’ efficiency to evaluate a common phenomenon;
- Testing their sensibility to already identified biases.
Nevertheless,