Handcuffs for the Grabbing Hand? Media Capture and Government Accountability by Timothy Besley and Andrea Prat (2006) Group Hicks: Dena, Marjorie, Sabina, Shehryar To the press alone, checkered as it is with abuses, the world is indebted for all the triumphs which have been gained by reason and humanity over error and oppression. James Madison, Report on the Virginia Resolutions, 1798 Better scrutiny leads to reduced corruption and stronger democracies in all society s institutions, including government, corporations and other organizations. A healthy, vibrant and inquisitive journalistic media plays a vital role in achieving these goals. Wikileaks, on the importance of a free media
Introduction Media plays a very important role in any kind of government, especially in democracies. Examples from various countries that move from an autocracy towards democracy offer support. One of the most important roles of the media is to act as a check on corrupt politicians, businessmen, bureaucrats, and generally on governments as a whole. But media itself might be influenced by these actors and might not always divulge all the information it has. The way this influence is exerted might take various forms; from direct cash bribes to favorable legislation affecting the media owners. The main purpose of this paper is to give economic clarification of when a government has the incentive to capture media and the circumstances under which it becomes possible and its political consequences. In the given model, the media can make profits from either its commercial (audience- related) operations or from collusion with the government. It only reports verifiable, negative information and therefore, the probability of an incumbent s reelection decreases with news reported. The incumbent has the incentive to remain in power to extract rents from his political authority. The probability of negative information being discovered by the media is an increasing function of the amount of rent extracted and media independence. Therefore, independent media reporting truthful news: 1) influences the outcome of elections through altering voter opinion (by increasing the possibility of turn- over) 2) forces incumbents to extract less rent This implies, as mentioned in the paper that scandal free countries have either an extremely independent media industry or extremely pliable one. Thus, incumbents have the incentive to silence media. However the possibility of media capture is restricted by the amount of rent that can be obtained. Media can accept 2 types of kickbacks: Direct money transfers and favorable policies. The paper proves that: 1) A larger number of media outlets reduce the risk of capture. 2) Independence of media also decreases the risk of capture. Here, independence of the media is measured by the difficulty of transferring resources to media. 3) Media capture impacts the political process by reducing turnover and increasing rent seeking behavior.
I. The Baseline Model We are faced with a two period retrospective voting model with three groups of players: Media We have! identical active media outlets that look for verifiable information to report They have two sources of income: o Audience related revenues that include sales, subscriptions, etc.; we will refer to the maximum revenues of this kind as!. o Collusion related revenues that include direct bribes, favorable policies generating revenues such as in the FIAT group example (FIAT group controlled two of the top three Italian newspapers until 2003 and benefited from restrictions on car imports from Asia). If!! firms report a piece of informative news and share audience related revenues such that each firm gets!. Voters prefer informative news, and we normalize the revenues from! uninformative news to 0. Media observe no signal if an incumbent is good. For bad incumbents media observe a verifiable signal with probability! where! is dependant on technology, institutions, censorship levels, libel laws, and privacy protection regulations. Politicians Incumbents we meet in the first round are exogenously in power and can be good- g or bad- b (assume nothing in between). That is, a politician is described by! such that!" {!,!} and the!! =! =!. Incumbents are allowed to manipulate news and this is modeled as a bargaining game. News is only printed if it is both bad and verifiable so our incumbents only strategy is to hide bad news by offering media a transfer in exchange for their silence. Note that since all media receive the same information in the basic model the incumbent must successfully silence all media and thus will either pay off all outlets or none. Transfers cost the incumbent!! per media outlet and yield!! /! for each outlet where!! > 0. The!! range from direct monetary transfers in the form of cash bribes to more indirect benefits (favorable policies) and the! represent transaction costs ranging from 0 when its easy to make transfers to when it is impossible to make transfers. o Media types play a role in transaction costs. State owned media likely have the lowest transaction costs, privately owned media outlets are more likely to receive benefits if their owners have homogeneous interests, and cross ownership situations lend themselves nicely to favorable policy arrangements. o Legislative constraints are an example of transaction costs. Payoffs: If re- elected:!!!!! ; if not re- elected!!!!.
Voters Homogeneous electorate that makes their voting decisions using all the information available at the time of the election. Voters do not observe their payoffs at the time of the reelection decision (long term projects etc.) and always vote for the candidate they prefer (consequence of Perfect Bayesian equilibrium where voters use undominated strategies) Payoffs: from bad incumbents 0 and from good incumbents 1. Key Assumptions of the Basic Model Timing Step 1: 1. Only verifiable information can be printed 2. Signals can only be bad 3. All media have the same information (since they are identical and all have the same technology and exist in the same environment) 4. Incumbent knows what signal the media have received and can ask them to verify 5. Incumbent makes offers after signals are realized. (*Underlying assumption that all parties are risk neutral in which case it makes no difference if incumbent makes offers before or after realization). Incumbents type(!) is realized;!! =! =! Step 2 Signals o If good, no signal goes out;!! = = 1 o If bad, we place a probability on whether a signal goes out!! =! =! and! = otherwise Incumbent observes media signal and selects a transfer!! 0 for each outlet! Media outlet! observes transfer!! Step 3 If accepts, reports! = and receives!! /! If rejects, reports the true signal and receives!/! where! is the number of other outlets also reporting the true signal. Voters observe signals and vote for incumbent or a challenger of unknown type
Equilibrium Market Media (Proposition 1) Two cases: a) Captured Media: If! <!/!" each media outlet suppresses its information in exchange for a bribe!! =!". b) Independent Media: If!!/!" each media outlet reports its information truthfully (as the incumbent cannot afford to bribe everyone). With respect to! /!": The maximum available audience related revenues are! and can all go to one outlet if it is the only one to report a piece of informative news. The incumbent only makes offers to outlets if he is assured that he can silence the entire media and can only have such assurance if he offers each informed outlet!! >!. This transfer will cost the incumbent!" and plays a key role in whether he can afford to silence the media. Turnover and Voter Welfare (Proposition 2) Turnover of politicians, voter welfare, and total audience- related revenues are non- decreasing in!,!,! and!. Note that we can rewrite our inequality from above as!"# <! when media is captured. To obtain an independent media, we can increase the number of media outlets!, the difficulty in making transfers!, and/or the commercialization of the media market which would increase!. Alternatively, we could decrease the rents! accessible to politicians. Captured media do not identify bad politicians, and voters have no ability to screen good from bad politicians. When indifferent, our voters choose the incumbent, and thus their expected utility (EVU) is! from the first round and! from the second round. In this case, our turnover rate (probability we replace a challenger) is 0 and the incumbent stays. Independent media will identify bad politicians and our new turnover rate becomes! 1! 0 EVU in round 1!! =!!! +!! =!!! =! 1 + 1! 0 =!!"# in round 2!! =!,!!! +!! =!,!!! +!! =!,!!! +!! =!,!!!! =! 1 + 1! 1! 0 + 1!!! 1 + 1!! 1! 0 =! +! 1!!
where b is a challenger Total EVU or Voter welfare is higher with independent media 2! +!(1!)! as is the turnover rate. With respect to the effect of a change in! on voter welfare and the rate of turnover (which we denote as!), we take first derivatives and find!!! = 1!! > 0!!! = 1! > 0 Recall that we normalized the audience related revenues to 0 for each outlet with non- informative news. Let! be the sum of all audience related revenues so with non- informative news! = 0 and with informative news! =!". To increase turnover, voter welfare, audience related revenues we would need o greater media independence (high!) to decrease chances of media capture o media commercialization (high!) to make bribes less attractive and make it more expensive for politicians to capture media o plurality (high!) for the same reasons o political transparency and efficient news production (high!)
Empirical Data Of note we see that a survey of these 88 countries show state ownership of newspapers is associated with both increased tenure of the chief executive and corruption. Likewise high concentration of ownership is associated with lower turnover and more corruption.
II. Extensions of the Baseline Model A. Moral Hazard In this extension, an incumbent chooses how much rent to extract, and the probability he is caught is increasing in the rent consumed as well as media activity. The media can directly influence rent extraction; however, screening good candidates from bad is more difficult. The rent extracted is denoted y [ 0,1]. The remainder goes to voters. Politicians, as in the baseline model, are either good or bad. If a politician is good, they derive no benefit from rents and choose y=0. On the other hand, a bad politician derives a linear benefit from rent. The paper assumes that politicians will only serve two terms. If bad politicians make it to a second term, they will extract y=1, the total rent available. Bad incumbents have no reelection motive besides the possibility of rent consumption in the second term. The probability of detection given y is denoted Ψ( y)q with Ψ 0, Ψ > 0,Ψ( 0)= 0, Ψ ( 0)= 0,Ψ(1) = 1, and lim Ψ ( y)=. As before, q [ 0,1], but for illustration it is assumed that at least one media outlet is active and that there is no media capture. Good incumbents will choose y=0 both terms, and bad incumbents appropriate y=1 in the second term. So, what remains is to derive optimal rent extraction of bad incumbents during the first term, and to ensure reelection. During the first term, given q, the bad incumbent solves max y { y + 1 Ψ( y)q}. This yields an optimal rent appropriation level ŷ, which satisfies Ψ ( y)q = 1, which allows us to show: y 1 Meaning that greater media activism reduces the rent appropriated by politicians. In this scenario, political turnover is given by p(q) =! (ŷ(q))q(1!" ). Intuitively, the right hand side is the probability a bad politician was in office originally multiplied by the probability the bad politician gets caught extracting rent. There will be two competing effects on turnover as q increases. With a fixed y, media outlets are more likely to detect rent extraction; however, this increase in q encourages incumbents to reduce rent extraction. This makes impact of active media on turnover ambiguous. To see this mathematically, the sign of " p'(q) =! (ŷ) +! (ŷ)q dŷ % # $ dq& ' (1(" )
is indeterminate without specific functional forms. Although the effect on turnover is ambiguous, voter welfare is unambiguously increases from media activism. Expected voter welfare is given by W (q) = 2! + (1!! )[1! ŷ +" (ŷ)q! ] where the first term is the situation where a good incumbent is selected in the first term and returns to power with certainty. The second term is the situation where a bad incumbent is selected, extracts rent in the first period but is caught and replaced by a good incumbent. Notice that: The sign of the derivative with respect to q is unambiguously positive. Voters may want bad politicians to consume some positive level of rent because rent seeking is the only mechanism for screening politicians. Equilibrium rent seeking will always exceed the level desired by voters, meaning greater media activism is valuable on the margin since: Voters marginal benefit from rent is! '(ŷ)q" while the marginal cost is 1. Incumbent sets! '(ŷ)q = 1. As a result the marginal cost is always greater than the marginal benefit. The argument for why voter welfare is increasing in q also implies that expected rents are decreasing in q. Expected rents can be written as: Notice that: R(q) = (1!! )[1+ ŷ!" (ŷ)q! ]. This expression refers to total rents extracted rather than those extracted by a particular politician Incumbents always choose rent levels below the expected rent- maximizing level because they care about maximizing their own rents rather than the total rents extracted. This makes incumbents more cautious in rent seeking than rent maximization implies. An increase in q will lower rent extraction through the discipline effect (politicians are more cautious and extract less rent) and a screening effect (more bad politicians are caught and removed).
This entire discussion summarizes Proposition 3. B. Endogenous Media Entry This scenario is the same as the baseline case except that the number of media outlets is endogenous. There are a large number of potential media who can become active by paying a fixed cost c. This adds a Step 0 to the baseline game, during which, potential media outlets choose whether to enter. This is simultaneous and noncooperative. The outlets that have paid c receive an informative signal with probability q. Combining Step 0 with the baseline model yields Proposition 4: Two kinds of equilibria in media market o (a) If mod( qa / c)> r / τa, media is independent and number of outlets is m = mod(qa / c) o (b) if mod(qa / c) < r / τa, media is captured and number of outlets is ( ) mod r / τa Whether or not media is free is determined by comparing r / τa (the maximum number of outlets the government is willing to pay off) and qa / c (the equilibrium number of media entrants if media is independent). We can see that the higher fixed costs (c), the greater the chance that there will be media capture. So, from proposition 2, it is obvious that greater barriers to entry reduce political turnover and voter welfare. C. Endogenous Monitoring The baseline model assumed exogenous information gathering technology for all firms. This extension alters the baseline model by making media quality - the ability of outlets to detect signals dependent on investment decisions. This means that higher resources devoted (such as talented personnel, sophisticated equipment, insider knowledge) increase the success of a journalistic investigation. Here, media will be differentiated vertically in terms of quality and capture will depend on how much each media outlet invests in monitoring. This investment will cost c(q), where c(.) is increasing, convex, and twice differentiable and q is the same verifiable signal of the incumbents quality as in the baseline model. Investment decisions are made simultaneously and non- cooperatively. Detection of the incumbent here also depends on the realization of the variable v where v is uniformly distributed between [0, 1]. There is verifiable information that an incumbent is bad if v q i
The incumbent will capture either all or none of the informed media outlets. The cost of buying off everyone = m*!*a, where m is the number of informed media outlets. The incumbent will bribe everyone if r m*!*a The maximum number of outlets that can be captured (M) = mod (! ). Nobody is captured if!! informed media exceeds M. The incumbent bribes only if v > q M+1, where v is the minimum necessary effort to detect an incumbent s type. Depending on how v is realized, there are 3 possible outcomes: i. v > q 1 Nobody is informed and the incumbent gets reelected ii. v q M+1 The number of informed outlets is greater than the number of outlets the incumbent can afford to buy off, so the incumbent is removed from office iii. q M+1 < v q 1 The informed media is captured and incumbent gets reelected Equilibrium choices of q are given by Lemma 5 Let M = mod(r/!") and let q (k) be the unique q such that c (q) = a/k. In equilibrium, q 1 = = q M = q (1) and, for every i M+1, q i = q (i). A bad incumbent is removed from office with probability q (M+1). Media outlets invest in monitoring up to a point where the marginal cost equals the marginal revenue. Marginal revenues are different for firms depending on which category of media outlet they fall into. If q > q M+1, it equals a, but if q q M+1, it equals a/i. Improved media efficiency can be modeled as a decrease in the cost of investing in monitoring. C(q) = κc(q), whereκ > 0 and C(.) has the same properties discussed earlier. κ can be interpreted as the degree of transparency in the operation of government. Voter welfare and turnover are now determined by the probability that a bad incumbent is not reelected, ˆq(M +1). By Lemma 5, ˆq is implicitly defined by κ C (q) = a. r mod τa + 1 This indicates C'( ˆq(M +1)) is nondecreasing in a andτ, and nonincreasing in!. Thus, an increase in a has 2 effects: it encourages media to invest more in monitoring, and it increases the cost for the incumbent to buy off the media. This discussion combined with the fact that C (q) is increasing in q leads to Proposition 6.
Suppose the media choose their monitoring levels endogenously. Turnover and voter welfare are nondecreasing in! and!, and non increasing in! This extension lets us compare two seemingly identical countries but where one has an institutional environment where capture is more likely (lower a or! or a higher!). The media industry in this country will be less differentiated and this is a sign of media freedom. Examples of media differentiation from the US, UK, and Italy offer support to this conclusion. D. Bribing as Access Apart from giving bribes and introducing favorable legislation, there can be other ways of influencing the media and give them an opportunity to raise their revenues. For example, a politician could give an exclusive interview to a media outlet that is friendly to him. So if the government cannot make direct monetary transfers but can promise interviews (I i ) to specific media outlets where!! is between 0 and 1, then!! = 1 means that the interview is granted if i keeps quiet about embarrassing news.!! = 0 means that there is no interview granted. The additional profit from granting interviews, π (m ), where m is the number of outlets given selective interviews, is decreasing in m. This is because for an exclusive interview, commercial value is really high as everybody is tuned in to that specific channel (or buying that newspaper) to follow the interview. Proposition 7: Suppose that the government bribes the media by offering access only to favored media outlets. Then the media industry is captured if and only if π (n) > a. This means that the only case in which media could be silenced is if profits from an exclusive interview exceed what the media outlet could earn by breaking some negative news. As the number of media outlets goes up, risk of capture decreases. With media diversity, if you are one of the many firms airing an exclusive interview, you would rather break some bad news. Conclusions: The paper gave insights into the political economy of media capture. The baseline model and extensions highlight the relationship between features of the media, media independence, and political outcomes. In particular, the discussion made clear the extra dimensions that should be considered when regulating the media. Above and beyond standard
competition, it is important to consider how any changes affect welfare both directly (prices, advertising rates) and indirectly through political accountability. Media capture is increasingly possible when expected rent of incumbent is high enough and number of media outlets is small. So media plurality is an important factor to ensure media independence. Higher transaction costs expressed in terms of time, energy, financial means and reputation lost decreases the possibility of media capture. Policy changes, such as establishing minimum technical qualifications for entering media and changes in law of media ownership, may have an affect on media capture. Any policy changes should be evaluated to see is they potentially limit the number of media outlets and/or lower transaction costs (making it easier for government to make transfers to the media. A potential innovation in the media industry must also be evaluated on the basis of its effect on the ability of the government to influence news production to its advantage. A merger between two big media firms, privatization of state media, or a major foreign investment in local media, amongst many other things, could have serious consequences for the media- voter- government relationship. Future research may build upon the models presented by relaxing the assumption restricting media to report only verifiable information Finally, the authors note that media bias can come from other sources: Reputational concern of the media Confirmatory cognitive bias Ideological bias of reporters and an increase in the number of independent media may actually increase the level of bias in the media.