Privatisation in Developing Countries

• Author: William L. Megginson,Natalie L. Sutter
• Date: 01 July 2006
• Published date: 01 July 2006
• DOI: http://doi.org/10.1111/j.1467-8683.2006.00505.x

234

CORPORATE GOVERNANCE

© 2006 The Authors

Journal compilation © 2006 Blackwell Publishing Ltd, 9600 Garsington Road,

Oxford, OX4 2DQ, UK and 350 Main St, Malden, MA, 02148, USA

Volume 14 Number 4 July 2006

Blackwell Publishing IncMalden, USA

CORGCorporate Governance: An International

Review0964-8410Blackwell Publishing Ltd. 2006

2006144••••ORIGINAL ARTICLES

PRIVATISATION

IN DEVELOPING COUNTRIESCOPRORATE GOVERNANCE

*Address for correspondence:

Price College of Business, 307

West Brooks, 205A Adams

Hall, The University of Okla-

homa, Noman, OK 73019-4005,

USA. Tel: (405) 325-2058;

Fax: (405) 325-7688. E-mail:

wmegginson@ou.edu

Privatisation in Developing Countries

William L. Megginson* and Natalie L. Sutter

We survey empirical studies examining privatisation’s effects in developing economies. Most

of these studies f‌ind that privatisation yields improvements in the operating and f‌inancial

performance of divested f‌irms, and only a handful document outright performance declines

after privatisation. Almost all studies that examine post-privatisation changes in output,

eff‌iciency, prof‌itability, capital investment spending and leverage document signif‌icant

increases in the f‌irst four measures and signif‌icant declines in leverage. The studies examined

here are far less unanimous regarding the impact of privatisation on employment levels in

privatised f‌irms. Studies that explicitly address the sources of post-privatisation performance

improvement using data from multiple non-transition economies tend to f‌ind stronger

eff‌iciency gains for f‌irms in regulated industries, in f‌irms that restructure operations after

privatisation, and in countries providing greater amounts of shareholder protection.

Keywords: International f‌inancial markets, economics of regulation, political economy

Introduction

ational governments have been privatis-

ing state-owned enterprises (SOEs) for

two decades now. Not surprisingly, academics

have used many pages to investigate the

implementation and aftermath of these priva-

tisation efforts in countries around the world.

This paper surveys the research on privatisa-

tion in developing countries. These studies

employ many different empirical method-

ologies, cover many different regions and

time periods, and vary greatly with respect to

the type and quality of data employed.

Our task, therefore, is to categorise the

studies of privatisation’s effectiveness in a

way that allows us to assess the impact of

privatisation on different countries and on

different economic agents. Even within fairly

homogeneous groupings, it is very likely

that privatisation will be viewed differently by

consumers, by SOE employees and by the

newly created class of shareholders. In other

words, we must assess the distributive effect of

privatisation as well as the effect of privatis-

ation on f‌irm performance. This is especially

N

important because privatisation (and any

attendant regulatory changes) is often the

sole major component of reform processes in

developing countries.

This survey begins with a description of the

key objectives in our organisational strategy

and a discussion of the methodological prob-

lems researchers have been forced to deal

with. The second part then surveys the group

of studies that gives the broadest assess-

ment of privatisation’s impact. These are stud-

ies that examine how privatisation impacts

the f‌inancial and operating performance of

divested f‌irms by comparing the pre- versus

post-privatisation values of several account-

ing and real output measures. Almost by

def‌inition, these studies examine only share

issue privatisations (SIPs), since these are the

only companies that are publicly traded and

independent after privatisation and thus the

only f‌irms generating f‌inancial statements

that are directly comparable to pre-privatisa-

tion f‌inancial values. This should still give a

fairly accurate picture of the overall effect of

privatisation because SIPs account for more

than two-thirds of the US$1.25 trillion of total

PRIVATISATION IN DEVELOPING COUNTRIES

235

© 2006 The Authors

Journal compilation © Blackwell Publishing Ltd. 2006

Volume 14 Number 4 July 2006

revenues raised by privatising governments

since 1977.

1

The third section surveys multi-

national, multi-industry studies and evaluates

the evidence presented in single-country or

single-industry studies. Other country and

industry studies are then examined, after

which the f‌ifth section summarises tests of the

economic effectiveness of privatising infra-

structure companies, and emphasises studies

of privatisations of telecommunications com-

panies, electric utilities and water and sewer-

age companies. The f‌inal section concludes the

paper and summarises the overall evidence

of privatisation’s effectiveness in developing

economies.

Strategic approach to classifying and

assessing tests of privatisation’s

effectiveness

Besides diff‌iculties in categorisation, research-

ers must grapple with many challenging

methodological problems in order to analyse

objectively privatisation’s economic impact.

2

An important problem is that of data availa-

bility and consistency. The amount of infor-

mation that must be disclosed is much less in

most countries than in the United States, and

these standards vary from country to country

as well as over time within individual coun-

tries. Furthermore, the possibility of sample

selection bias can arise from several sources,

including the desire of governments to make

privatisation “look good” by privatising the

healthiest f‌irms f‌irst.

There are also many problems in measuring

performance changes that arise from using

accounting or stock data. The problems with

accounting data include determining the cor-

rect measure of operating performance, select-

ing an appropriate benchmark with which to

compare performance, and determining the

appropriate statistical tests to use. The f‌inance

literature has not reached a consensus on the

ways to deal with these problems for US

companies, much less privatised international

f‌irms. Barber and Lyon (1996) argue that test

statistics designed to determine whether there

is abnormal performance using accounting

data are mis-specif‌ied when the sample f‌irms

have performed unusually well or poorly.

They suggest that sample f‌irms must be

matched to control f‌irms with similar pre-

event performance, which is especially diff‌i-

cult in studies of privatised companies due to

the lack of truly comparable f‌irms.

Given these diff‌iculties, the results of each

of the studies we discuss must be kept in per-

spective. We also note that the studies of post-

privatisation performance rarely examine the

welfare effects on consumers. Most important,

few studies control for the possible use of

market power by the privatised f‌irms. That is,

performance improvements could be due to

greater exploitation of monopoly power –

which has harmful effects on allocative eff‌i-

ciency – rather than improved productive eff‌i-

ciency. Many of the studies on performance

changes after privatisation examined the

effects of divestiture on groups such as work-

ers, but few early tests examine the effect of

privatisation on consumers. This is a serious

drawback, since one of the principal reasons

for launching privatisations, particularly of

monopoly utilities, is consumer dissatisfaction

with the service provided by state-owned

f‌irms. Fortunately, more recent tests have

explicitly measured redistributive and market

power effects, though with varying degrees of

success. The vast majority of the studies cited

below report increases in performance associ-

ated with privatisation using at least one, and

usually several performance measures. This

consistency is perhaps the most telling result

we report; privatisation appears to improve

performance measured in many different

ways, in many different countries.

With the above caveats in mind, this survey

evaluates the results of dozens of studies that

employ accounting and/or real output data

to examine the impact of privatisation on

the operating eff‌iciency, ownership structure

and/or f‌inancial performance of former SOEs

in developing economies. Though these

studies are detailed in the accompanying

tables, and most are discussed at least brief‌ly

in the text, we also specify which studies we

think are the most important – and why we

think this is so. A further organisational step

is to present summary information for each of

the studies we examine in a series of tables.

Presenting this information in tabular form

saves us from having to sequentially discuss

each paper’s sample construction method-

ology, estimation procedure and empirical

results in the section’s text. Instead, we can

identify key f‌indings that appear in many dif-

ferent studies, and can discuss methodological

pros and cons for entire groups of studies,

rather than for each paper in turn.

Empirical studies comparing pre-

versus post-privatisation

performance of SIPs

Most of the studies summarised in this sec-

tion examine how privatisation affects f‌irm

performance by comparing pre- versus post-

divestment data for companies privatised via

public share offering. Since the f‌irst study

236

CORPORATE GOVERNANCE

© 2006 The Authors

Journal compilation © Blackwell Publishing Ltd. 2006

Volume 14 Number 4 July 2006

to be published using this methodology is

Megginson, Nash and van Randenborgh

(1994), we will refer to this as the MNR meth-

odology. This approach compares three-year

average post-privatisation f‌inancial and oper-

ating performance ratios to the three-year

pre-privatisation values of f‌irms in different

countries and industries. They test for the

signif‌icance of median changes in ratio values

in the post- versus pre-privatisation periods.

They also employ binomial tests for the per-

centage of f‌irms changing as predicted. This

empirical procedure has several obvious eco-

nomic and econometric drawbacks. Of these,

selection bias probably causes the greatest

concern, since a sample of SIPs will be biased

towards the very largest companies sold dur-

ing any nation’s privatisation programme.

Furthermore, since governments have a na-

tural tendency to privatise the “easiest” f‌irms

f‌irst, those SOEs sold via share offerings (par-

ticularly those sold early in the process) may

well be among the healthiest state-owned

f‌irms.

3

Another drawback of the MNR

methodology is its need to examine only

simple, universally available accounting vari-

ables (such as assets, sales and net income) or

physical units such as number of employees.

Obviously, researchers must be careful when

comparing accounting information generated

at different times in many different countries.

Most of the studies cited here also ignore

(or, at best, imperfectly account for) changes

in the macro-economy or industry over the

seven-year event window during which they

compute pre- versus post-privatisation per-

formance changes. Finally, the studies cannot

account for the impact on privatised f‌irms of

any regulatory or market-opening initiatives

that often are launched simultaneously with

or immediately after major privatisation

programmes.

In spite of these drawbacks, studies employ-

ing the MNR methodology have two key

advantages. First, they are the only studies

that can examine and directly compare large

samples of economically signif‌icant f‌irms,

from different industries, privatised in differ-

ent countries, over different time periods.

Since each f‌irm’s performance is compared to

its own results a few years earlier using

simple, inf‌lation-adjusted sales and income

data (that produce results in simple per-

centages), this methodology allows one to

eff‌iciently aggregate multi-national, multi-

industry results. Second, while focusing on

SIPs yields a selection bias, it also yields

samples that encompass the largest and

most politically inf‌luential privatisations. With

these methodological caveats in mind, we turn

to a summary of the f‌indings of studies using

the MNR technique. These multi-national,

multi-industry studies are detailed in Table 1.

Multi-national, multi-industry

studies

Megginson, Nash and van Randenborgh

(MNR) (1994) compare three-year average

post-privatisation f‌inancial and operating

performance measures with the same three-

year average pre-privatisation performance

measures for 61 companies from 18 countries

and 32 industries that were divested during

1961–1989. Using information obtained from

prospectuses, annual reports and secondary

sources, they examine whether the perfor-

mance of these companies improves after they

are privatised. They document economically

and statistically signif‌icant post-privatisation

increases in output (real sales), operating

eff‌iciency, prof‌itability, capital investment

spending and dividend payments, as well as

signif‌icant decreases in leverage. They f‌ind

no evidence of employment declines after

privatisation, and in fact the median level of

employment actually increases signif‌icantly

(at the 10 percent level). They also f‌ind sig-

nif‌icant changes in the number and identity

of f‌irm directors around the time of initial

privatisation.

MNR f‌ind their basic results are unchanged

when they compare f‌irms operating in com-

petitive versus noncompetitive (regulated

and/or protected) industries, when they ex-

amine privatisations where the government

surrenders control and contrast these with

revenue privatisations where the purpose of

share sales is primarily to raise cash, and

when they compare industrialised (OECD)

and developing country privatisations. When

MNR partition their data based on the

fraction of a f‌irm’s board that is replaced,

however, they document signif‌icantly greater

performance improvements for the group of

f‌irms that experience 50 percent or greater

turnover than for the group of companies

experiencing less dramatic change in direc-

tors after divestment.

Boubakri and Cosset (1998) analyse the

privatisation experience of 79 companies

from 21 developing countries and 32 indus-

tries divested between 1980 and 1992. They

document economically and statistically sig-

nif‌icant post-privatisation increases in output

(real sales), operating eff‌iciency, prof‌itability,

capital investment spending and dividend

payments – as well as signif‌icant decreases

in leverage. They also f‌ind that employ-

ment typically increases, but not signif‌icantly.

The f‌inancial and operating performance