Mass Media and Political Accountability [620539]

Mass Media and Political Accountability

Timothy Besley, Robin Burgess and Andrea Prat

Contents

Introduction 1

1. Intellectual Fram ework 3
1.1 Political Agency Problem s 3
1.2 Media and Agency 5
1.3 Media Capture 6

2. Empirical E vidence 8
2.1 Evidence from India 8
2.2 Evidence from Cross-Country Data 11

3. Conclusion 15

4. References 16

The support of the Economic and Social Research Council (ESRC) is gratefully acknowledged. The
work was part of the programme of the ESRC Centre for Analysis of Risk and Regulation.

Published by the Centre for Analysis of Risk and Regulation at the
London School of Economics and Political Science
Houghton Street
London WC2A 2AE

© London School of Economics and Political Science, 2002

ISBN 0 7530 1606 0

All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form
or by any means, without the prior permission in wr iting of the publisher, nor be otherwise circulated in
any form of binding or cover other than that in which it is published and without a similar condition
including this condition being imposed on the subsequent purchaser.

Printed and bound by Printflow, October, 2002.

Mass Media and Political
Accountability1

Timothy Besley, Robin Burgess and Andrea Prat

Introductio n

Mass m edia can play a key role in enabli ng citizens to monitor th e actions of
incum bents and to use this inf ormation in their voting decisions. This can lead to
governm ent which is m ore accoun table and resp onsive to its citizens’ needs. In sp ite
of the intuitive plausibility of the proposit ion, there is comparatively little work in
politica l economy literature th at scrutinises the role and ef fectiven ess of the m edia in
fulfilling this function. A litera ture, however, is emerging which focuses attention on
the im portance of the so-called ‘fourth esta te of governm ent’ in the policy process. A
key feature of the app roach taken h ere is to focus on incentives the m edia have to
produce and dissem inate inform ation.

This paper discusses work on political ag ency problem s with a focus o n recen t work
by the authors. Besley and Burgess (2001, 2002) looks at the effect of m edia on
responsiveness to shocks in India, while Besley and Prat (2001) focuses on the
determ inants and consequences of capture d media, with em pirical evidence from
cross-country data. At the heart of these pape rs is the idea that citizens have im perfect
inform ation about the actions of governm ent. Mass m edia can, therefore, enhance
citizens’ abilities to scrutinise governm ent actions. Here, we develop the m ain them es
and assess the em erging em pirical evidence. The latter identif ies an d tests th e
mechanism s through which develop ment of m ass media can enhance accountability. It
also exam ines the determ inants of m edia capture and hence the constraints on its
capacity to perform a wat chdog role.

There is a host of reasons why governm ents are better inform ed than voters and,
hence, act on the basis of privileged info rmation. Politicians know m ore about their
competence than those who vote for them . They have access to m ore policy advice
and scenarios from a variety of sources. For example, if a bridge or dam is being built,
then it is only through m edia scrutiny that citizens can ascer tain whether proper
atten tion has been p aid to the rele vant costs and benefits. Sim ilarly, when natural
disasters strike, an ac tive mass media inc reases the ability of citizens to monitor ho w
much effort their representatives h ave put in to protect the vulnerable. This is
particularly im portant in low incom e countri es where citizens rely so strongly on state
action for social protection. Suppose, for exam ple, that there is a region of a country
contain ing fifty villages where one village is hit by a flood. W ithout m edia, only tho se
directly affected can observe the actions of gove rnment. However, m ass m edia
enables citizens in all fifty villages to obs erve whether the governm ent is responsive.

1 This is a draft chapter for The Right to Know: In stitutions and the Med ia, edited by R oumeen
Islam, to be published by the World Bank.
1

This rais es the incentive of politicia ns to respon d since citiz ens in the other f orty-n ine
villages m ay use this inform ation in their voting decisions.

For inform ation generated by the m edia to be va luable, it needs to elicit an appropriate
collective response. Even in autocratic se ttings, this m ay be possible. However,
clearly this is m ore like ly to happ en when there are d emocratic in stitutions such as
free electio ns. In a democracy, citizens requi re information which the y can use to
select politicians that serve their needs a nd to punish those that do not. Otherwise
formal democracy has no bite.

While m ost countries have m edia of som e description, its m ere existence is no
guarantee that it is an effective vehicle fo r critica l scrutiny of state action s. This
requires that m edia outlets ha ve real inform ation about such actions w hich they are
willing to p rint o r broa dcast. This will de pend on the exte nt to which the m edia is
regulated, captured or repressed. The latt er is affected by a variety of governm ent
actions. These range f rom policy decision s affecting the regulation of entry and
ownership of the m edia, th rough to explicit bribery o r threats. Th ere are m any
countries w hich, while for mally democra tic, have lim ited amounts of press freedom.
In a sam ple of 151 countries for which we have both a press freedom ranking from
Freedom House and which have held elections in the past five years, we find that 36
receive one of the bottom two press freedom scores and only 59 are in the highest two
catego ries.2 The ‘dem ocracies’ with low press freedom scores also ten d to be low
incom e countries.

How the governm ent treats the m edia industr y affects the developm ent of news m edia
and the qua ntity and qu ality of news genera ted. What the r aw data sug gests it tha t
there are hu ge variation s in access to the media across th e world. Data from the W orld
Developm ent Indicators for 1997, show a va riation in circulation between 0.008 daily
newspaper circulation per 1000 population in St Vincent and the Grenadines, to 792
in Hong Kong. There are sim ilarly broad vari ations in television ownership. The sa me
data source shows a variation in televi sion ownership per 1000 population, varying
from 0.1 in Rwanda to 850 in the United States. Not surp risingly, there are strong
links between m edia developm ent and other developm ent indicato rs such as income
per c apita a nd liter acy. Af ter contro lling f or inc ome per capita and regional dummy
variab les, th ere is also evidence that news paper circulation and television ownership
is lower in countries that ha ve a larger fraction of state- owned m edia using the data of
Djankov et al (2001).

Whether causal or not, there is als o a strongly positive co rrelation between m edia
penetration and m easures of press freedom from Freedom House. Media penetration
seem s also to go hand-in-glove with indices of form al and real dem ocracy. Using data
from the Polity IV datab ase,3 countries that are rated as more dem ocratic have higher
levels of news m edia penetration, as measured by newspaper circulation and
television ownership. Of course, the direc tion of causation is unclear. There is a
similar positive correlation between m edia penetration, and the weaker form al notion
of de mocracy, m easured as whether a country has held an electi on in the past five
years.

2 The ranking is on a six point scale.
3 See h ttp://weber.u csd.edu /˜kgledits/Po lity.ht ml

2

What these raw corr elations te ll us is tha t there m ay be significa nt costs associated
with having an underdeveloped m edia. Mo reover, underdevelopm ent of the m edia is
often the result of decisions by governm ents to insulate them selves from scrutiny and
criticism . Often this takes the form of governm ent owner ship, barriers to entry for
priva te media com panies, and anti-de famation laws. W hile this m ay be in the inter ests
of government officials, it is often le ss clear how it serves the public good.
Deregulation of the m edia, th erefore, stands out as a possible powerful policy lever
that cou ld be used to prom ote acco untabi lity in the underdeveloped world. How such
deregulation could be im plemented in the face of governm ent opposition is a
challenge th at needs to b e faced.

This paper is organised as follows. In the next section, we discuss how the m edia can
mitigate political ag ency problem s in theo ry. W e first dis cuss how th e media can
reduce or improve the efficiency of the political system , assum ing that the m edia
operates freely. W e then discuss the theory of m edia capture. Section two reviews the
evidence both from within-count ry and cross-countr y studies. Section three concludes.

1 Intellectual Framew ork

This se ction lays out th e political a gency framework which we view as a useful
organising device for discussing the role of th e media in de mocratic settings. W e first
discuss th e forces that in fluence the effec tiveness of free and independent m edia. We
then discuss what determines th e likelihood that media is captured.

1.1 Political Agency Problems

A good fram ework in which to think about th e role of the m edia is one in which
citizens are im perfectly inform ed about th e actions of governm ent and the track
records of their leaders. To the extent th at we believe that politicians m ay behave
opportunistically and serve their own private agendas, ahead of that of the public at
large, then p olitics is a kind of principal- agent p roblem . The principals are the citizen s
of the polity who finance governm ent activ ities through taxes and are subject to
various regulations, and the agents are the elected officials and bureaucrats who
determ ine policy ou tcomes.4

There are a couple of features of political agency which distinguish it from other
agency relationships that econom ists have studied. First, the in centive schem es on
offer are typically very crude. For exam ple, with politic ians, the o nly sanctio n
typically av ailab le (except for cases of gro ss malfeasance) is not to re-e lect. Mone tary
or other m ore nuanced incentive contract s are alm ost never observed. This has the
consequence that incentives ar e mostly im plicit, with politicians having to guess what
the voters would like them to do, rather than the latter posting perf ormance criteria in
advance. Even in the case of lobbying, it is very hard to im agine a very com plete
contract which specifies the details of what the principa ls (the lobbyists) desire.

4 This approach has a lon g tradition in bo th po litical scien ce an d econo mics beginning with Barro
(1973) and Ferejon (1986). For rece nt reviews see , for example, the contributions in Przeworski,
Stokes and Man in (1999) and Persso n and Tab ellini (2000, chapter 4).
3

The second distin ctive featur e is m ultiplicity of principals . There are many citizen s
and other actors (such as corporations) who diffe r in countless ways. Thus even if the
incentives could be m ade explic it, it m ight be very difficult for the princip als to ag ree
on the ince ntives that they will su bject th eir a gents to. Th us, we could easily f ind
principals with diam etrically opposed inte rests wishing to pull the actions of t he
agents in different directions.

Inform ation availability is at the heart of this agency theoretic view of governm ent
and politics. W hen the principals try to in fluence policy either via the ballot box or by
lobbying, then they do so with limited inform ation about th e agents whom they are
lobbying. There are bo th problem s of hidden action (m oral hazard) – w hen an agent
has discretion to m ake or take a bribe unbeknown to the citizens, and hidden type
(advers e se lection ) – when the m otivations and/or com petence of the agent i s
unknown. Effective incentives would ideally punish incum bent politicians for bribe-
taking and/or incom petency. But if this ca nnot be widely observ ed, then doing so is
difficult.5

There are good reasons for believing that private inform ation gathering exercises by
the principals are unlikely to provide sufficien t infor mation. This insi ght is at least as
old as Downs (1957) w ho argued that voter s would be “rationa lly ignorant” about
politics because th ey hav e sign ificant costs of being inform ed and ne gligible ben efits –
they being only one out of a m ass of voter s. This rational ignorance comes on top of
the usual free-riding problem in voting its elf, when a si ngle voter m ay not find the
benefits of voting exceeding the cost. Th is might suggest great pessim ism in the
ability to f ind solution s to politica l agency problem s.

However, there are at least a coup le of reasons to be m ore optim istic. First, th ere c an
be significant private gains to becom ing info rmed about policy. Consider the case of
old-age pensions. Any rationa l individual planning for thei r retirem ent would have a
significant private benefit from understanding the evolving public policy debate in
this area. Second, the power of forces such as a civic duty m ay make private benefit a
poor guide to what happens in practice.

The third re ason f or potentia l optim ism is the prim ary f ocus here – the role of the
media. Mass m edia can be a powerful source of inform ation generation at low cost to
citizens. By being bundled with other activities (s uch as sport or entertainm ent) m any
individuals m ay be induced to regard acquisi tion of infor mation not as a chore, but as
a pleasure, thus raising the general level of awareness abou t policy and public affairs.
This, at lea st, is the ro sy-eyed vie w. Here, we will us e the f ramework of politic al
agency to g ive a tigh ter specification of the issues. W e will s uggest th at discuss ions o f
the effectiv eness of th e media can usefully be decom posed into two parts: (i) the
forces that enable a free and independent m edia to induce governm ents to better serve
the pub lic interes t; (ii) the forces that lead the governm ent to successf ully intervene to
silen ce the m edia.

5 Besley an d Case (1 995) sug gests th at there is ev idence for th e em pirical sig nifican ce of politica l
agency m odels applied to U.S. Governors , some of whom periodically face a ter m limit whic h bars
them from running for office again . Incentives for governors to ac quire reputations vary at such
points – those with re-election incentives have stron ger incentives th an th ose withou t. Th ey find
that there are distin ct policy differen ces between Governors in their in itial an d term -limited term s.
4

1.2 Media and Agency

There are three m ain routes by which non-captured media can affect political
outcom es: sorting, discipline and policy sa lience. W e will discuss each of these
possibilities in turn, and then look at which features of politics and m edia m arkets are
likely to give the largest incentive to the m edia to perform this function.

Sorting refers to the process by which poli ticians are selected to hold office. The
motivation o f politic ians is a potentia lly im portan t issue f or citizen s – some politicians
(albeit rarely) such as Gandhi and Nelson Mandela enjoy an alm ost saintly status,
while others are reviled. Th e practical ground is typically interm ediate between these
two. However, the kind of infor mation pr ovided in the m edia can potentially be
important to voters who are deciding who to put in charge. This would include
inform ation about the previous track record of potential office-holders. Actions that
they perform in office m ay also be an im portant source of inform ation about their
underlying motivation or com petence. By prin ting news stories that responsibly cast
light on this, the m edia can be a pow erful force. To the exten t that sorting is effective,
ex-post incentives are less necess ary.

The role of the m edia in achiev ing discipline is most relev ant in situations of hidden
action. Suppose that a politician is thinking of taking or m aking a bribe and that the
probability that this will becom e public depe nds on the ef ficacy of the m edia in both
discovering it and broadcasting it widely. Then the m arginal cost of that action will
tend to be higher, potentially deterring th e individual from doing so. Thus, we would
expect m edia development to be negatively correlated with such action s leading to
bette r incum bent dis cipline.

The m edia can also have an effect on which issues are salien t to voters. Besley and
Burgess (2002) consider the case of a vul nerable population in a developing country
subject to shocks such as droughts and floods. Such populations are dependent on
state ac tion to m itigate the im pact of these shock s but ne ed politica l clout to ge t their
interests on the political agenda. One way of achieving this is for politicia ns to f ind it
worthwhile to develop reputations for be ing responsive in th e event of shocks
occurring. T his requ ires that these groups are infor med about act ions of politician s
and that, if they are infor med, then the issue is salient am ong the m any other
charac teristics of politic ians tha t they care abo ut. Assum ing that respo nsiveness of
governm ent to droughts or floods is a signifi cant enough issue to these citizens, then
more inform ation enhances the salience of this issue at the ba llot box and, thereby,
creates ince ntives f or politicians to build reputations for being responsive. Mass
media can thus play a central role in enhancing responsiveness by providing
inform ation that citizens can use to decide whom to vote for.

More generally, by publicising politicians’ st ances on particular issues, the m edia can
change th e structure of salient issu es in elections. This may m ean that th e media
wields considerable influence. In som e countries this can be less than benign,
depending on the m otives of newspaper owne rs – news stories can, for exam ple,
public ise ev ents tha t enhance the sa lienc e of ethnic tens ions leading to victim isatio n
of these groups.

5

All three of these effects rely on the m edia providing reliable inform ation. The
quality of the news on offer to citizens depends on a number of fac tors. First, the
transparency of the political system is im portant. In count ries without traditions of
free inf ormation f lows, inf ormation will tend to be hard to o btain. The e xtent to which
the m edia will prin t more speculative news will also be dependent on the lega l
environm ent in which the news m edia operate. F or exam ple, the United States allows
a public f igure def ence in libe l law and theref ore makes it easier f or news m edia to
print stories without recourse to libel suits. This contra sts with the UK where libel
laws m ake publishing speculative news stories more difficult.

Also im portant to news quality will be th e methods and traditi ons of investigative
journalism . This m ay depend on the training of journalists and the extent to which
news editors reward inve stiga tive a ctivity and a llow journalis ts to devote tim e to it. It
may also rest on som e perception of how keen the citizens of a particular country are
to be informed, and hence the comm ercial value of breaking important news stories.

1.3 Media Capture

As we have just argued, to keep govern ment accountable to the electorate a country
needs effective m edia. In this section we study one of the m ain obstacles to m edia
effectivenes s: the po ssibility of political c apture. W hat motivates m edia? Firs t, the y
want to reach a large audience. This is true both for newspapers, because sale and
advertising revenues are linked to circu lation, and for television statio ns, because of
advertising and, where applicab le, cable fees. Com petitio n for audience interest
pushes th e media to lo ok for interesting news and to estab lish a reputation for
reliability.

While the desire to increase m arket share is com mon to m ost industries, there is also a
second m otivation that derives from the speci al role of media as political m onitors.
Anytim e som eone is monitored, they m ay be tempted to ingra tiate or threa ten th e
monitor in order to get bette r coverage. In the case of gove rnment and the m edia, this
takes a variety of form s. Som e are very si mple, like cash bribes offered to individual
journalists, violent th reats, or cens orship. Some are more subtle and typically not
illegal. The governm ent can pass regulation th at benefits the ultim ate owner of a
particular m edia outlet. For instance, if a newspaper is owned by an industrial
conglom erate which also owns a car com pany, beneficial regulation can take the form
of a tariff on car im ports.

To ascertain which of the two m otivations we should expect to prevail, Besley and
Prat (2001) build a sim ple model of m edia cap ture. There are three classes of players:
voters, politicians and m edia. Voters are rational and, for sim plicity, they have
homogeneous preferences. Their problem is they canno t monitor th eir politicia ns
direc tly. All the inf ormation they receive com es from the m edia.

The political side of the model is repres ented by a standard two-period accountability
problem . In the first period, a politician (the incu mbent) is ex ogenously p ut in power.
The type of the incum bent (good or bad) is not directly observable. At the end of the
first period, an election is held in which voters can re-elect the incum bent or replace
him/her with a challenger of random type . In the second period, the candidate who
wins the election is in power.
6

The m edia industry is made up of n a prior i iden tical outlets. With som e probability,
they r eceive verif iable news on the type of the p olitician, wh ich th ey can repor t to th e
public (f or simplicity we assum e that they are either all inf ormed or all uninf ormed).
We assum e that a m edia outlet cann ot fabricate news and that an outlet that reports
inform ative news has a higher audience than one that reports no news. Moreover, the
audience sh are of an outlet that reports ne ws is decreasing in the number of other
outle ts that report news. The best case f or an outle t is to be the only one to brea k
news.

We also assum e that news can only be bad, that is, one m ay have verifiable
inform ation that a politician is bad but not that s/he is good. Th is assumption is not
restrictive for our purposes because the governm ent would never want to suppress
positiv e news. The im portan t assu mption is th at news can not be f abricated. Allo wing
for that – while keep ing the assum ption th at voters are rationa l – would m ake the
analys is very difficult. The credib ility of media would aris e as a result of a com plex
signalling gam e.

The revenue of a m edia outlet has two components that correspond to the tw o
motivations discussed above. The first is an increasing function of audience. T he
second is a transfer from the gove rnment (whi ch should be interpreted loosely as
favourable regulation). The cost to the politic ian of m aking a transfer of a given value
to the ou tlet depends on transaction costs. Th is is because som e forms of transfer m ay
be illeg al or politica lly costly, while others can b e disguised as norm al policy -making.
More about transaction cost s will be said later.

The tim ing of the gam e is as follows: (1) Th e media outlets receive or do not receive
verifiable infor mation about the in cumbent; (2) The incum bent knows what
inform ation the m edia got and m akes them transfer offers; (3) Each outlet chooses
whether to accept o r reject the offer; (4 ) The outlets that accept th e offer suppress
their inf ormation ; the ones tha t reject it re port their inf ormation to vo ters; (5) Vote rs
re-elect the incum bent or repl ace him /her with a challenger.

In sear ching for the e quilib rium of this game, the m ain question is whether th e
incum bent finds it profitable to buy off the media industry or not. If an outlet thinks
that a ll the other ou tlets a re going to be q uiet, then its incen tive to re ject the
incum bent’s offer goes up because it would be the only one to break news to voters
and it would gain a large audience. This m eans that, in an equilibrium in which all
media sell o ut, the incu mbent m ust pay each outlet as if it were the only one wh o
could break news. Even if we keep the to tal industry potential revenues constant,
increasing the num ber of m edia outlets m akes it more expen sive for the incum bent to
buy their silence. This is the sense in which m edia pluralism is good for m edia
independence.

Besides the num ber of outlets , the other param eters that determ ine whether m edia are
captured or not in equ ilibrium are tran saction costs and the am ount of audience-
related revenues. Both decrea se the likelihood that the in cumbent m anages to silence
media. Instead, the p robability that m edia are in formed does not affect m edia capture
but, of course, it increases the proba bility tha t voters ar e informed.

7

This m odel can be extended in a num ber of directions. Besides having a ‘type’, the
politician may choose the am ount of rent extraction activity s/he carries out. The
large r the r ent, the hig her the pro bability that the m edia are able to spot him /her.
Then, m edia monitoring does not only weed out bad politicians but has also a
discip line effect because it m akes dishonest po liticians m ore likely to get caught. This
creates a U-shaped relation between media effe ctiven ess an d political turnover. The
probability that the incu mbent is replaced is low, both when the effectiv eness is ve ry
low (because no politician is ever caught), and when the effectiveness is very high
(becaus e no politician dares to ex tract rent). Other poss ible extensions includ e
ideological m edia, vertical di fferentiation, and endogenous entry.6

To summ arise, the m edia captu re model yield s severa l testable im plication s. The
probability of media capture, and hence of bad politica l outcom es such as cor ruptio n,
depends po sitively on the follo wing vari ables: m edia industry concentratio n,
transaction costs, and au dience-related revenues.

Transaction costs are particularly interest ing. One should expect that they depend on
media ownership. If the outlet is stat e-owned, the governm ent may appoint the
managem ent and con trol the resou rces. If it is owned by a f amily or has a con trolling
shareholder, the governm ent can pa ss regula tion favouring the interest of the owne rs.
If it is a w idely-held cor poration, then the governm ent cannot benefit the owners
directly but has to do som e direct tran sfer to the outle t. Finally, if the owner is a
foreign entity, transfers m ay becom e very difficult.

2 Empirica l Evidence

Our analytical fram ework ha s served to identif y a num ber of channels through which
mass media can influence policy-m aking. Info rmation provided via the m edia can be
used in voting decisions. This can both increas e the salience of par ticular issues and of
selec ting po liticians which act in th e public in terest. A f ree press can a lso serve as a
direct check on the excesses of politician s. Therefore, we might expect som e effects
of factors like corruption. In general, it is not e asy to find reliable evid ence to match
with the richness of the theor etical possibilities. In th is section we review the lim ited,
available evidence. W hat we know com es mainly from fairly reduced-for m cross-
country evidence. This is a notoriously diffi cult context to study, with directions of
causality being near impossible to discer n and highly im perfect m easures of m ost
variab les. A more prom ising app roach is to explo it data from within coun tries tha t for
one reason or another exhibit variation in media activity. A case in point is India
where state -level m edia institutions vary s ignificantly. We will d iscuss the f indings
that a re available from this and o ther con texts. Given its grea ter re liability, we
discuss the within-country evidence first and follow this by an assessm ent of what can
be learned from cross-country studies.

2.1 Evidence from India

The trad ition of free and independent press has perm eated som ewhat into the
developing world. A prom inent example is India which ha s a newspaper industry that

6 See Besley and Prat (2001) for details.

8

is distinguished from the bulk of other low incom e countries by being both free and
independent (Ram , 1991). Sen (1 984) has attributed a m ajor role to this f actor in
explaining why India has not experienced any m ajor fa mines in the post-
Independence era. He observes that, “India has not had a fa mine since Independence,
and given the nature of Indian politics and so ciety, it is not likely that India can have a
famine even in years of great food problem s. The governm ent cannot afford to fail to
take p rompt action when large-s cale star vation threatens. Ne wspapers play an
important part in this, in m aking the fact s known and forcing th e challeng e to be
faced.” (Sen, 1984: 84). In contrast, the lack of de mocracy and of freedom of
inform ation have been pointed to as r easons behind why China experienced a m ajor
famine between 1958 and 1961, with excess m ortality figures ranging b etween 16.5
and 29.5 m illion. Representative democracy an d the m edia have also b een identified
as factors in African countries which have been successful in preventing fam ines (see
Dreze and S en, 1989). As the Sen statem ent m akes clear, m edia increases the salien ce
of government perform ance in famine si tuations by providing information on the
action s of politicians, which citiz ens can use in their voting decisions.

Though suggestive, the Sen analysis doe s not establish a robust link betwee n
developm ent of m ass media and governm ent re sponsiveness. Analysis of the role of
media in influencing governm ent policy ha s recently been deepened b y Besley and
Burgess (2002). Using panel data for Indian states for the 1958-1992 period, they look
at two policy response system s. First, public distribution of food as a response to falls
in food production associated with droughts. Second, spending on calam ity relief as a
response to crop dam age caused by floods. They then examine how newspapers and
politics affect how responsive state governm ents in India are to these shocks. They
find that higher newspaper circulation is associated with gove rnment being m ore
responsive to falls in food production a nd flood da mage. The m agnitudes of the
effects that they find is la rge – a ten per cent drop in food production is associated
with a one per cent increase in public food dist ribution in states which are at the
median in term s of news paper circulation per capita; whereas for stat es that are in th e
75 per cent percentile a ten pe r cent drop in food production is associated with a 2.28
per cent increase in public f ood distribution. Interestingly, Besley and Burgess (2002)
find that it is newspapers published in the regional languages that are driving their
results.7 This m akes sense – it is regional langua ge newspapers that report on localised
shocks and are m ore likely to be read by local, vulnerable gr oups and politicians.
These resu lts hold up in the face of a whole array of robustness checks. T hus, we have
strong evidence that even within India, va riation newspaper circulation can explain
how responsive Governm ent is to the needs of its citizenry.

In line with political a gency theor y, it is the interp lay b etween m ass m edia and
political institutions that determ ines govern ment responsiveness. Besley and Burgess
(2002) exam ine how various political fact ors influence governm ent responsiveness.
They find that political turnout increases how responsive state governm ents are to
drought and flood shocks. Greater political com petition is also asso ciated with greater
responsiveness. This m akes sense as hi gher turnout and m ore intense political

7 We ha ve dat a on newspaper circulation broken down into nineteen different languages. Hindi and
English (at least in part ) are national in scope whereas t he rem ainder t end to be st ate-speci fic.
Poor, vulnerable p opulation s will ten d to be con versan t in the state-sp ecific lan guage. In our
regressions, therefore, news paper circulation is broken ou t into Hind i, English and ‘Other’.
9

competition will incr ease the incen tives f or politician s to build reputatio ns for being
responsive to the need s of their citizens.

Table 1 is taken from Besley and Burgess (2001) and gives the rank of a m easure of
state-level responsiveness in public food dist ribution in India, al ong with rankings of
other state characteristics. The responsiv eness m easure tells us how much food, on
average, is publicly distributed as a re sponse to food production changes in a given
state. According to this m easure, Kerala is the most responsive and Bihar the least
responsive state. The striki ng feature of this rank ing is how weakly this
responsiveness m easure relate s to the incom e per capita ranking. However, the rank of
newspaper circulation follows the responsiveness ranking quite closely.

Table 1: Ranking of sixteen Indian states in responsiveness and other
variables

(1 = highest)

State name Responsiveness Per capita income Per capita
newspaper
circula tion
Kerala 1 13 1
Maharashtra 2 3 2
West Bengal 3 5 4
Tamil Nadu 4 8 3
Gujarat 5 4 6
Assam 6 10 15
Uttar Pradesh 7 11 8
Andhra Pradesh 8 9 10
Karnataka 9 6 7
Rajasthan 10 15 9
Punjab 11 1 5
Orissa 12 12 16
Haryana 13 2 13
Jammu & Kashm ir 14 7 11
Madhya Pradesh 15 14 12
Bihar 16 16 14
10

These results, along with those in Besley and Burgess (2002), are strongly supportive
of the notio n that m ass m edia helps to so lve p olitical agen cy problem s and m ake
governm ents m ore acco untable. The intu ition of what is driving the resu lts is sim ple –
by m aking the actions of the politicians m ore transparent the m edia is providing
inform ation to citizen s as to the likelihood th at they will be protected in the future.
Citizen s, in turn, a re using th is in formation in their voting decisions. Politicians
realise this and this creates an incentive for politicians to be responsive to shocks.
Note tha t this incentive is pre sent e ven wh ere politicians have no inherent interest in
protecting citizens and are onl y doing so in the interests of garnering the votes of
vulnerable citizens. Mass m edia thus aff ects responsiveness, both by increasing the
salience of the social protection issu e, and by affecting th e selecti on of politicians via
voting.

The Indian evidence is consistent with ot her recent country studies su ggesting th at
media can have effects on policy-m aking. Fo r exam ple, Yates and Stroup (2000) look
at pesticide decisions by the U.S. Envi ronm ental Protection Agency a nd find that
more draconian standards are set where m ore newspaper articles about safety have
been publish ed histor ically. This f its with th e idea that issue salience can change with
the m edia. Larcinese (2001) finds that m ass m edia both determ ines the political
knowledge of citizens and also driv es turnou t in data from the UK. Strömberg (2001)
relates New Deal spending in county-leve l data for the United States to radio
ownership, finding a pos itive asso ciation betw een the two which s uggests that areas
with higher penetration of ratios were more successful at attracting New Deal
spending. What m akes the Indian evidence par ticular poignant is th e fact that it com es
from a low incom e country.

2.2 Evidence from Cross-C ountry Data

A num ber of corruption m easures have r ecently becom e available for different
countries. It is interesting to see whether m easures of press freedom are correlated
with these m easures. In accordance with the theoretical discussion ab ove, we would
expect that greater press sc rutiny w ould be associated w ith lower corruption. Ahrend
(2001) and Brunetti and W eder (1999) have ca rried out exercises along these lines and
show that there is a n egativ e correlati on between press freedom measures and
corruption in cross-country data. It is di fficult, however, to draw caus al influen ces
from such fi ndings, if media can be captu red by corrupt governm ents who can then
get away with even greater corruption, then the two would be co-determ ined without
one causing the other. To m ake fur ther progr ess, we need to m easure features of th e
media m arket which may m ake m edia cap ture more or les s likely – proxies for the
transactions costs discussed in Besley and Prat (2001).

One prom ising route takes advantage of th e data on m edia ownership collected in
Djankov et al (2001). It seem s plausible to think that state- ownership of the m edia, in
particular, will lower the cost of capturing the m edia. Sugge stive of this id ea, Djankov
et al (2001) find that higher corruption is found in countri es with fewer state-owned
newspapers. They find no effect for television.

Using the sam e data, Besley and Prat (2001) also look at corruption as an outcom e.
Using three different sources of data on corruption, they find that corruption is
11

negatively correlated with foreign ownership of the m edia. This finding is robust to
including a wide variet y of different control variables. They inte rpret this finding a s
evidence that foreign ownership m ay be corre lated with f actors that m ake the m edia a
more effective inform ation generating device.

The Djankov et al (2001) and Besley and Prat (2001) results t ogether, point to the
need for us to better understand w hat determ ines m edia capture. A rough way of
measuring m edia capture em pirically is to look at whether a c ountry’s press freedom
is given a score of less th an, or equal to, two on Free dom House’s six point scale.8
We can then ask which characteristics of a country’s m edia m arket are significantly
correlated with m edia capture so m easured. To do this, we exploit the data of Djankov
et al (2001). Specifically, we in clude three variables: the ex tent of foreign ownership,
the extent of state ownership, and a m easure of ownership concentration. In line with
Besley and Prat (2001)’s theore tical predictions, capture is m ore likely if there is m ore
state-ownership of newspapers and there is greater con centration in ownership of
newspapers. The latter suggests that m edia capture is affected by m edia plurality.
Unlike the case of corru ption as an outcom e, there is no sig nificant effect of foreign
ownership on the probability of capture.

Besley and Prat (2001) al so ask whether the observe d correlation between m edia
ownership a nd political outcom es is because p rivate or foreign ownership m akes
media m ore efficient, o r becaus e it m akes them less sus ceptible to po litical captu re.
Using perceived m edia independence as an in strum ental variable, it is possible to run
a test to check whether, conditional on being non-captured, m edia ownership
influences p olitical outcom es. In the case of private ownersh ip, there app ears to be n o
efficiency g ain. The beneficial effect of having private m edia com es only from the
fact tha t they are less likely to be captured . Instead, in the case of foreign m edia, the
over-identification test fails, and one cannot exclude that both th e efficiency and the
non-capture channels are active.

These relationships can be se en graphically in Figures 1 and 2. Figure 1 plots pres s
freedom ( measured on a 1-6 scale) agai nst the degree of state ownership of
newspapers using the data from Djankov et al (2001) and Besley and Prat (2001). A
high press freedom score denotes more fr eedom. The line gives the regression line
between these points emphasising the downwar d-sloping relationshi p. If we define
press capture as having a press freedom score less than, or equal to, three, we can
predic t the probability that press will be cap tured. W e do so of f of three variables :
state ownership of newspapers, foreign ow nership of newspa pers, and concentration
in newspaper ownership. W e then graph the International Country Risk Guide
corruption m easure (also on a 1-6 scale) agai nst our predicted probability of capture.
The result is given in Figure 2. The upward-sl oping pattern in the data should now be
clear – with countries with a higher pr obability of being captured having m ore
corruption. While crude, they illustrate how th e cross -country data can be used to
inform the theoretical discussion.

8 This is cl osely related to th e em pirical analys is of Besley and Prat (2001).
12

(mean) own3 press_fr Linear prediction
0 116

Figure 1

cor_icrg
Pr(capture).044003 .921306

Figure 2

Djankov et al (2001) also casts light on the political salience argum ent since they find
that a very broad array of welfare and policy indicators respond positively to a low er
fraction of state-ownership of newspapers. If one takes ow nership as being correlated
with the q uality of the inf ormation-gene ration process, this would be true; for
exam ple, if profit-m otivated m edia invest m ore in finding out im portant news stories,
then the changing policy prioriti es found in the data are cons istent with the idea that
issue salience is chang ing with developm ent of the m edia.

Conclusion

Worldwide there has been a drive to m ake governm ents m ore accountable to the
needs of citizens. The g overnance agenda has been pushed by a range of actors fr om
domestic and international NGOs through to the internatio nal financial organisations.
Though there is widesp read c onsensus on the n eed to improve accoun tability, it is
much less clear what m echanism s might be us ed to achieve this. This paper m arshals
evidence w hich suggests that a free and independent press working in conjunction
with dem ocratic ins titutions can m ake governments m ore responsive to the needs to
citizens. Political agency m odels have proved to be a useful foundation for thinking
about the role of the m edia, since they put the im portance of inform ation in the
political pro cess at cen tre-stage. We have discussed the various chains of influence
via which the media can have an im portan t input into the po licy process, accord ing to
such m odels. W e then argued that the m odels had som e empirical m ileage, with
evidence tying the electoral process, the media, and governm ent actions together.
Whilst the empirical literature is in its infa ncy, the data certainly suggests that there
are som e links to the facts that the media sect or has a role to play in political agency.
Greater understanding of what m akes governm ents better servants of the people, and
of how strengthening institu tions supports th is role, defines a rich agenda for future
work in political economy.

Our central conclusion m ust be that a free and indepe ndent m edia should not be
viewed as a luxury that only rich countries can afford. In stead, our analysis suggests
that it should be viewed as a requisite and integral part of representative dem ocracy. A
key question then is what dire ctions of reform can stren gthen th e role of media in
prom oting accountability? The question of m edia regulation, theref ore, is a central
question here. In past d ecades, m ost countri es had heavy, industry-specific ru les fo r
the press and for broadcasting. Instead, in the 1990s, many developed countries
(including the US and those in the EU) saw a dram atic shift in the philosophy behind
media regulation. The new paradigm is that media should be ruled by the sam e basic
competition policy p rinciples th at are applied to the other industries, whic h in turn are
mostly based on laissez -faire, excep t when it can be prov en that consum er welfare is
hurt. This has led to th e elim inatio n of media-specific rules, such as restrictions o n
ownership, concentration and pricing. Howeve r, it should by now be clear that m edia
is a different industry because of its role as po litical watch dog. Consum er welfare,
defined in term s of the inte rest of custom ers (viewers and a dvertisers ), is a lim iting
notion in that it does not take into account the effect of the industry on voters’
welfare. Future research in the area should re-evaluate ex isting regulatory regim es in
this new lig ht.

15

References

Ahrend, R. (2001) ‘Press Freedom , Hum an Capital and Corruption,’ typescript.

Barro, R. (1973) ‘The control of politici ans: An econom ic model,’ Public Choice, 14,
19-42.

Besley, T. and Burgess, R. (2001) ‘Polit ical Agency, Governm ent Responsiveness
and the Role of the Media’ , European Economic Review , 45(4-6), 629-640.

Besley, T. and Burgess, R. (2002) ‘The Political Econom y of Government
Responsiveness: Theory and Evidence from India,’ Quarterly Journal of Economics,
(forthcom ing).

Besley, T. and Case, A. (1995) ‘Does Political Accountability Affect Econom ic
Policy Choices? Evidence From Gubernatoria l Term Li mits, Quarterly Journal of
Econom ics, 110(3), 769-98.

Besley, T. and Prat, A. (2001) ‘Handcu ffs for the Grabbing Hand? : Media Capture
and Political Accountability, ’ typescript, LSE.

Brunetti, A. and W eder, B. (1999) ‘A Free P ress is Bad News for Corruption,’
typescript, University of Basel.

Downs, A. ( 1957) An Economic Theory of De mocracy . New York: Harpe r Collins.

Djankov, S., McLeish, C., Nenova, T. and Shleifer, A. (2001) ‘W ho Owns the
Media?’ typ escrip t.

Dreze, J. an d Sen, A. (1989) Hunger and Public Action. Oxford: Clarendon Press.

Ferejohn, J. (1986) ‘Incumbent perform ance and electoral contro l,’ Public Choice, 50,
5-25.

Larcinese, V. (2001) ‘Inform ation acquisi tion, ideology and turnout: theory and
evidence fro m Britain’, typescrip t, LSE.

Persson, T. and Tab ellini, G. (20 00) Political Econom ics: Explainin g Econom ic
Policy . Cambridge: MIT Press.

Przeworski, A., Stokes, S.C. and Manin, B. (19 99) Democracy, Accoun tability, and
Representation . Cambridge: Cam bridge University Press.

Ram, N. (1991) ‘An Independent Press and Anti-Hunger Strategies: The Indian
Experience’ in Dreze, J. and Sen. A. The Political Econo my of Hunger Volum e 1.
Oxford: Oxford Univers ity Press.

Sen, A. (1981) Poverty and Famines. Oxf ord: Oxf ord University P ress.

16

Sen, A. (1984) ‘Food Battles: con flicts in the access to food,’ Food and Nutrition,
10(1), 81-89.

Ström berg, D. (2000) ‘Radio’s Im pact on th e New Deal,’ typescri pt, Departm ent of
Econom ics, IES, Stockholm .

Yates, A. J. and Stroup, R.L. (2000) ‘Media Coverage and EPA Pesticide Decisions,’
Public Choice , 102, 297-312.

17