Jump to ContentJump to Main Navigation
The Challenges of PrivatizationAn International Analysis$

Bernardo Bortolotti and Domenico Siniscalco

Print publication date: 2004

Print ISBN-13: 9780199249343

Published to Oxford Scholarship Online: April 2004

DOI: 10.1093/0199249342.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 October 2019

(p.121) Appendix 1: Data and Methodology

(p.121) Appendix 1: Data and Methodology

Source:
The Challenges of Privatization
Publisher:
Oxford University Press

The empirical analyses are based on a series of international data sets which have been constructed at the Fondazione Eni Enrico Mattei, and contain several privatization, economic, financial, political, and institutional data, mainly for the period 1977–2001.

In this appendix we will describe how the variables have been constructed, and their sources.

1. Privatization Variables

The primary sources of our privatization data are Privatization International and Securities Data Corporation,1 reporting globally 3,535 transactions worth over $127 billion (current as of 2001) in 140 countries.

Our sources report information at the transaction level about the type of privatization (PO or PS). The data for each privatization carried out by private sale or public offer include the date of the deal, the company's industrial sector and country, the total value of the transaction (in current US$ million), the per cent for sale, and other qualitative information. In the sample group, the maximum value recorded for a single operation is around $40 billion (current), while the minimum is $100,000 for an average of $306 million.

2. Political Variables

In order to test the political theories, we need data about the partisan dimension of privatization. In particular, we want to identify the political orientation of privatizing governments over time.

In this direction, we have retrieved the political history of the forty-nine countries in the La Porta et al. (1998) sample from the Banks et al. 1997 edition of the Political Handbook of the World. This source reports election dates, dates of appointment of the cabinets, and a description of political systems around the world up to 1997. We updated this information for the years 1998–99 by use of Internet sources mentioned in the detailed definitions of the political variables.

We then used Wilfried Derksen's Electoral Web Sites and classification system to label incumbent governments, considering the platform and ideological orientation of the supporting parties. Four possible categories are identified: (i) democratic conservative (right wing); (ii) centrist and Christian-democratic; (iii) democratic left wing; (iv) non-democratic.

(p.122)

Table A1.1. Privatization variables

Variable

Definition

Source

ABROAD

Percentage of privatized stock placed on non-domestic financial markets (flag Rule 144a included). The variable refers to each single PO.

Privatisation International, and Securities Data Corporation.

ELREVENUES

Aggregate revenues from total operations in electricity generation per country 1977–97 (US$1996mil).

Privatisation International, and Securities Data Corporation.

ELSALES

Total number of operations by Public Offer (PO) and Private Sales (PS) in electricity generation per country 1977–97.

Privatisation International, and Securities Data Corporation.

ELSTOCK

Cumulative stake sold in electricity generators at the firm level.

Privatisation International, and Securities Data Corporation.

ENERGY

Dummy taking the value 1 when the privatized company belongs to the following sectors: electricity (generation), oil and gas production.

Privatisation International, and Securities Data Corporation.

FINANCE

Dummy taking the value 1 when the privatized company belongs to the following sectors: banking, financial intermediation, insurance.

Privatisation International, and Securities Data Corporation.

IPO

Dummy taking the value 1 when the Share Issue Privatization (SIP) is an Initial Public Offer (IPO).

Privatisation International, and Securities Data Corporation.

PO/DEALS

Ratio of the number of Public Offers (PO) to the total number of privatizations implemented in the period 1977–2001.

Privatisation International, and Securities Data Corporation.

PO/SALES

Ratio of the number of Public Offers (PO) to the total number of sales implemented in the period 1977–1996.

PRIVAMV

Market value of privatized firms.

Elaboration on Datastream.

PRIVATRADE

Value of trades of privatized firms.

Elaboration on Datastream.

REV/GDP

Ratio of total revenue (in US$1995mil) cumulated in the period to 2000 GDP (in US$1995mil).

Securities Data Corporation and World Bank, World Development Indicators (2002).

REV/GDP

Total revenues from privatization to Gross Domestic Product in country i in year t. Total revenues are revenues in current US$ from total privatization deals (PO and PS). Gross Domestic Product is expressed in current US$.

Privatisation International Database, IFR Thomson Database, World Development Indicators.

REVENUES

Aggregate revenues from privatizations during the period 1977–2001, in US$1995mil per country.

Privatisation International, and Securities Data Corporation.

SIZE/CAP

Ratio of the implied market value of the company ( SIZE : obtained by dividing total revenues from the SIP by the percentage of capital privatized, multiplied by 100) in current US$ to the market capitalization in the year of the SIP.

Privatisation International, IFC Emerging Stock Markets Factbook 1999, Federation International des Bourse des Valeurs (FIBV).

STOCK

Average percentage of capital sold by company over the period 1977–2001 per country.

Privatisation International, and Securities Data Corporation.

TLC

Dummy taking the value 1 when the privatized company belongs to the telecommunications sector.

Privatisation International, and Securities Data Corporation.

TRANSACTIONS OR DEALS

Total number of privatizations by Public Offer (PO) and Private Sale (PS) (tranches or complete sales) implemented in the period 1977–2001.

Privatisation International, and Securities Data Corporation.

UTILITY

Dummy taking the value 1 when the privatized company belongs to the following sectors: airline, airport, electricity distribution, gas distribution, rail services, rail track, water and sewage.

Privatisation International, and Securities Data Corporation.

WSTOCK

Weighted average percentage of capital sold over all firms, where the weights are given by the ratios between the revenues from privatization, by PO and PS, and total revenues in country i in year t.

Privatisation International, and Securities Data Corporation.

(p.123)

(p.124)

Table A1.2. Economic and financial variables

Variable

Definition

Source

AVDEFICIT

Country average of the public sector Deficit as a percentage of GDP in the three years before each SIP.

World Development Indicators, International Financial Statistics.

AVGROWTH

Average annual rate of growth of GDP per capita for the period 1977–1996.

World Development Indicators (1995).

CAP

Stock market capitalization to Gross Domestic Product in country i at year t. Stock market capitalization in year t is calculated as the average between the end-of-year market capitalization deflated by the end-of-year Consumer Price Index in year t and t−1. Stock market capitalization refers to a country's main stock exchange.

Beck, Demirgüç-Kunt, and Levine (1999), updated using data from IFC, Emerging Stock Markets Factbook, and FIBV.

CONSUMPTION

Average consumption of electricity in KwH 1977–96.

International Energy Agency.

DEBT

Total debt as a percentage of Gross Domestic Product of country i in year t. Total debt is expressed as the whole stock of direct, government, fixed term contractual obligations to others outstanding at a particular date. It includes domestic debt (such as debt held by monetary authorities, deposit money banks, non-financial public enterprises, and households) and foreign debt (such as debt to international development institutions and foreign governments).

International Financial Statistics.

DEFICIT

Average deficit of central government for the three years prior to the first privatization.

World Bank (1995).

FLOAT

Total value of trades on the major stock exchange/GDP.

World Development Indicators.

GDP

The logarithm of GDP US$1995.

World Bank, World Development Indicators (2002).

GDP PER CAPITA

Ratio of Gross Domestic Product in constant US$ 1996 to population in country i in year t. Total population counts all residents regardless of legal status or citizenship.

World Development Indicators, World Bank, International Financial Statistics.

GROWTH

Annual percentage growth rate of Gross Domestic Product at market prices based on constant local currency in country i in year t. Aggregates are based on constant 1995 US$.

World Development Indicators, and http://www.worldbank.org.

LOG OF GNP

Log of the average Gross National Product (1977–1996).

World Bank, World Development Indicators (1995).

MV

Monthly total market capitalization.

Elaboration on Datastream

SOE

Degree of importance of the public company in a state's economy in the year before the first privatization. Average of: (i) ratio between the value-added of the SOE and GDP; (ii) SOE employment as a percentage of the total work force; (iii) SOE gross investment on total investment, (where available).

World Bank (1995).

TRADE

Monthly total trading value.

Elaboration on Datastream.

TURNOVER

Stock market total value traded to total market capitalization in a country in year t. Total market value in year t is deflated by the Consumer Price Index in year t. Market capitalization in year t is calculated as the average between the end-of-year market capitalization deflated by the end-of-year Consumer Price Index in year t and t−1. Trading value and market capitalization refer to a country's main stock exchange.

IFC Emerging Stock Markets Factbook 1999, Federation International des Bourse des Valeurs (FIBV).

(p.125)

(p.126)

Table A1.3. Political variables

Variable

Definition

Source

CENTRE

Dummy variable taking the value 1 when the incumbent executive in country i in year t was supported by ‘centrist’ parties, and 0 otherwise. This label includes parties which are in the centre of the political spectrum without officially adhering to free market values, Christian- democratic parties, and wide coalitional governments without a clearly discernible orientation.

Banks et al. (1997), Wilfried Derksen's Electoral Web Sites (www.agora.stm.it/elections), Zarate's World Political Leaders 1945–2001 (www.terra.es/ personal2/monolith), Library of Congress Country Studies (http:// lcweb2.loc.gov/frd/cs/cshome.html).

DISPR

Disproportionality index. Sum of absolute differences between electoral votes share and seats share, for all the parties. Such divergence usually means overrepresentation of major parties and partial or complete exclusion of minor ones. Thus, increasing values of the index accord to the majoritarian rule, lower values to the proportional one.

Original dataset from Lijphart, updated using Electoral Studies, various years; Banks et al. (1997); Elections around the World; Parties and Elections in Europe; Political Reference Almanac.

ELECTION

Dummy variable taking the value 1 on the year of a country's elections, and zero otherwise. In presidential systems, presidential elections are considered. In parliamentary systems, general elections are considered.

Banks et al. (1997), Wilfried Derksen's Electoral Web Sites (Persson and Tabellini, 2001).

ENP

Concentration index computed over parties seats shares in the legislative chamber.

Original dataset from Lijphart, updated using Electoral Studies, various years; Banks et al. (1997); Elections around the World; Parties and Elections in Europe; Political Reference Almanac.

LEFT

Dummy variable taking the value 1 when the incumbent executive in country i in year t was supported by ‘left-wing parties’ and 0 otherwise. Left-wing parties include labour, socialist, social-democratic, and communist parties.

Banks et al. (1997), Wilfried Derksen's Electoral Web Sites (www.agora.stm.it/elections), Zarate's World Political Leaders 1945–2001 (www.terra.es/personal2/monolith), Library of Congress Country Studies (http://lcweb2.loc.gov/frd/cs/cshome.html).

MWOP

Discrete measure which accounts for the type of government in office: one party, minimal winning, minimal winning--one party, or neither of them.

Original dataset from Lijphart, updated using Electoral Studies, various years; Banks et al. (1997); Elections Around the World (www.electionworld.org); Parties and Elections in Europe (www.parties-and-elections.de/indexe.html), Political Reference Almanac (http://www.polisci.com/almanac/nations.htm).

NONDEM

Dummy taking the value 1 when the privatization was implemented by a dictatorial, military, or authoritarian ruler.

Wilfried Derksen's Electoral Web Site.

PARTISAN

Indicator for the government's partisanship. It is computed as the weighted average of the score attached to parties forming the government coalition, according to Huber and Inglehart (1995) and it ranges from 0 to 10. Weight i-th equal the number of seats held by party i-th in the legislative chamber over the total held by the government coalition. Null weight is assigned to parties whose seats are not essential for the government coalition to hold the absolute majority.

Electoral Studies, various years, Banks et al. (1997), Zarate's World Political Leaders since 1945 (www.terra.es/personal2/monolith), Library of Congress Country Studies (http://lcweb2.loc.gov/frd/cs/cshome.html), Administration and Cost of Elections (www.aceproject.org), Elections Around the World (www.electionworld.org) Parties and Elections in Europe (www.parties-and-elections.de/indexe.html), Political Reference Almanac (http://www.polisci.com/almanac/nations.htm).

POLINST

Standardized mean of the three measures DISPR, ENP , and MWOP . The standardization is performed over the whole sample.

RIGHT

Dummy variable taking the value 1 when the incumbent executive in country i in year t was supported by ‘democratic-conservative parties’, and 0 otherwise. Democratic conservative parties are defined as parties adhering to traditional values in combination with free-market ideology and law-and-order positions.

Banks et al. (1997), Wilfried Derksen's Electoral Web Sites (www.agora.stm.it/elections), Zarate's World Political Leaders 1945–2001 (www.terra.es/personal2/monolith), Library of Congress Country Studies (http://lcweb2.loc.gov/frd/cs/cshome.html).

RIGHTGOV

Dummy variable taking value 1 for Scandinavian civil law countries, and 0 otherwise.

La Porta et al. (1998).

(p.127)

(p.128)

Table A1.4. Institutional variables

Variable

Definition

Source

AGENCY

Dummy taking the value 1 when an independent agency as regulatory institution is present.

Lewington (1997).

ANTIDIRECTOR

Index that measures the legal protection that a country's company law provides against the risk of expropriation by managers. The variable takes into account the existence by law of (i) proxy voting by mail, (ii) cumulative voting for directors, (iii) oppressed minority mechanisms, (iv) requirements about the deposit of shares prior to general share holders meeting, (v) minimum percentage of shares to call for an extraordinary meeting at 10 per cent or below, and (vi) the pre-emptive rights that can be waived only by a shareholder's vote. It ranges from 0 to 6.

La Porta et al. (1998).

COMMON LAW

Variable that takes the value of 1 if a country belongs to the common law legal tradition and 0 otherwise. It never explicitly appears in estimates together with other legal traditions as this would cause problems of co-linearity with the constant. It is, therefore, the implicit benchmark with which the influence of other legal traditions is compared. In our sample group, the common law countries are: the United Kingdom, the United States, Australia, Canada, Hong Kong, India, Ireland, Israel, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Sri Lanka, Thailand, Zimbabwe.

La Porta et al. (1998).

CREDIBILITY

Average grade in terms of risk of contract repudiation and risk of expropriation in country i in year t. It ranges from 0 to 10.

International Country Risk Guide.

FRENCH LAW

Variable that takes the value of 1 if a country belongs to the French Civil Law tradition and 0 otherwise. In our sample group, the countries belonging to this tradition are: France, Argentina, Belgium, Brazil, Chile, Columbia, Ecuador, Egypt, Greece, Indonesia, Italy, Jordan, Mexico, Holland, Peru, the Philippines, Portugal, Spain, Turkey, Uruguay, Venezuela.

La Porta et al. (1998).

GERMAN–SCANDINAVIAN LAW

Variable that takes the value of 1 if a country belongs to the German–Scandinavian Civil Law tradition and 0 otherwise. In our sample group, the countries belonging to this tradition are: Austria, Germany, Switzerland, Japan, South Korea, Taiwan, Denmark, Finland, Norway, Sweden.

La Porta et al. (1998).

POOL

Dummy taking the value 1 when a wholesale electricity market (‘pool’) is operational.

Lewington (1997).

REGULATION

Regulatory index taking the value 3 when AGENCY, POOL , and TPA dummies are present.

Lewington (1997).

TPA

Dummy taking the value 1 when Regulated Third Party Access is granted by law.

Lewington (1997).

VINT

Dummy taking the value 1 when a vertically integrated electric system is present.

Lewington (1997).

(p.129) When the ideological orientation of a government remained unclear (due to frequent party changes and mergers in countries such as Turkey, Peru, Pakistan, and South Korea), we referred to the description of the political settings and institutions by the Federal Research Division of the Library of Congress of the United States. This source also allowed us to classify the most controversial cases.

In order to identify correctly the political preferences of the incumbent governments, we distinguish presidential and parliamentary systems. In the former, we considered the political orientation of the president's party and his cabinet; in the latter, the political orientation of the parliamentary majority supporting the cabinet. By the same token, in order to identify political switches, we consider presidential elections in presidential systems, and general elections—or simple changes of parliamentary majorities—in parliamentary systems. Determining whether political systems are presidential or not depends on answering a number of questions: following Persson and Tabellini (2001), we choose to check first whether the executive depends on a parliamentary majority; second if the president is elected by direct popular vote or with a de facto similar method of choice (like the US system), and he forms and leads the cabinet appointing and dismissing ministers (including the prime minister, if this office is present); and third (in those few cases where the political system is still uncertain of classification) whether the president is the most important decision-maker, holding the core of the executive power. We considered presidential ballots and parliamentary majorities only in France, a presidential country which (p.130) is customarily considered parliamentary in cases of ‘cohabitation’. ‘Cohabitation’ occurs when the president loses the parliamentary majority support and must abandon the reality of power to the prime minister if a party other than his own ever has a majority in the National Assembly (Aron 1982).

We have to attribute a political label to each country-year. When we observed a change in a government's political orientation after elections or (in parliamentary regimes) during the same legislature, we matched the political data with the dates of privatization sales. We attributed the political label to the government implementing the majority of the sales in the year. For example, a political switch from a centrist to a right-wing majority occurs in Italy in May 1994: five deals out of nine were implemented by the newly elected government in 1994, so we attached the label ‘right wing’ to that year. When a tie occurred, we used the (current) dollar amount of revenue to discriminate. For example, in France after the 1997 elections in June, the newly elected left-wing government implemented the same number of sales (two) of the former right-wing government. The left-wing government raised 93 per cent of the total revenue of that year, so we attached the left-wing label to France in 1997.

This methodology allows us to attach unambiguously one of the political dummies (i.e. right wing, centre, left wing, non dem) to each country-year.

3. The Control Sample

The rules for identifying the matching firms for our sample of privatized companies are as follows.

Table A1.5. Industrial sectors

Sectors

SIC numbers

Petroleum industry

13, 29

Finance/Real Estate industry

60–69

Consumer durables industry

25, 30, 36, 37, 50, 55, 57

Basic industry

10, 12, 14, 24, 26, 28, 33

Food/tobacco industry

1, 9, 20, 21, 54

Construction industry

15–17, 32, 52

Capital goods industry

34, 35, 38

Transportation industry

40–42, 44, 45, 47

Utilities industry

46, 48, 49

Textiles/trade industry

22, 23, 31, 51, 53, 56, 59

Services industry

72, 73, 75, 80, 82, 87, 89

Leisure industry

27, 58, 70, 78, 79

Source: Campbell (1996).

(p.131)

Table A1.6. Control sample

Criteria

Total

Percentage

Number of companies

143

100

Best case

78

54.54

Second-best case

64

44.75

Third-best case

1

0.69

Source: Fondazione Eni Enrico Mattei.

  • Best case. We first match by country. A sample of size greater than or equal to 1 passes this screen. Within this sample, we next match by industry. As to the industry classification, we use the Campbell (1996) system based on two-digits SIC numbers (see Table A1.5). A sub-sample of size greater than or equal to 1 passes this second screen. We sort this sub-sample by market capitalization, and choose the private firm with the market capitalization closest to our privatized firm within the 30 per cent range.

  • Second-best case. If we do not find any match in the country, we first match by industry. A sample of size greater than or equal to 1 passes this screen. Then we pick up an international firm in the same sector with the market capitalization closest to our privatized firm in the 30 per cent range.

  • Third-best case. We do this if we do not obtain the best case or the second-best case. We first match by country. A sample of size greater than or equal to 1 passes this screen. Then we pick up the domestic firm with the market capitalization closest to our privatized firm in the 30 per cent range (Table A1.6).

Notes:

(1) In April 1998, Privatisation International merged with IFR Platinum of Thomson Financial, a leading provider of financial data. From 2001, all the transactions reported in IFR-Platinum are also contained in Securities Data Corporation.