Cukierman and M eltzer (1986) offer in their paper an explanation of the policy maker’s preference for discretionary policy, which is based upon two major elements, the government’s desire to maximise its re-election prospects, and the existence of an asymmetric information structure^^. They put forward a positive theory of
The Case for Central Bank Independence
discretionary policy and investigate the welfare implications of retaining this policy. Cukierman and M eltzer’s analyses shall not be discussed in great detail, but the following paragraphs will give an overview of the major findings and implications of the analyses carried out.
The analysis is based on the assumption that politicians are ‘political entrepreneurs’ who try to maximise the likelihood of getting re-elected. This assumption is in accordance with a w idely-held view concerning the behavioural patterns in dem ocratic institutions in many W estern countries. Governm ents use their instruments to satisfy the socio-economic demands of their members in a game which culminates in the attempt of political parties to maximise their votes in the next national election.
The controversy between rules and discretion suggests two fundamentally different ways of conducting policy. If one assumes that the policy m aker’s actions are exclusively focused on the maximisation of social welfare, either strategy would lead to a socially optimal outcome. Whereas a policy maker who is primarily interested in the maximisation of support on election day would produce under discretion a sub- optimal outcome for the society. It is im portant to note that governments with discretionary authority differ in their ability to interpret events and in forecasting the future which results in a qualitatively different outcome. A government with good forecasting ability is likely to produce a better social outcome.
Considering the existence of incomplete information on the part of the private sector, the experienced level of welfare under a particular government acts as an indicator of its forecasting ability and future performance. The policy maker’s reaction to the voter’s way of tracing knowledge from observations makes the incum bent government act in a way that enhances its re-election prospects, namely by increasing output at the end of its term of office at the expense of future welfare. The resulting
welfare loss is “directly traceable to the existence o f periodic elections, so we call it 'the cost o f democracy'. It is larger the greater the frequency o f elections These costs will arise as long as information asymmetry prevails. There are two possible ways of eliminating these costs, as suggested by Cukierman and Meltzer, either to disclose unreservedly all information on the side of the government to have a fully- informed public, or to enforce a constitutional commitment to a socially optimal contingent choice of policy instruments. The latter solution would still provide a sufficient degree of flexibility, necessary to perform a stabilisation role in the case of disturbances. Nevertheless, such a com m itm ent involves serious enforcem ent problems which shall not be discussed here.
Cukierman and Meltzer base their analyses on the following set-up. An economy is frequently subject to disturbances which lie beyond the control of the government. The random state variable Xt, which expresses the existence of these disturbances, is introduced into the analysis. This random state variable as well as the settings of the policy instruments, described as at, affect directly social welfare. A government comes into power as a result of democratic elections. Elections are held at the end of each office term. The likelihood of re-election increases for the incum bent government proportional to the level of welfare experienced by the public under this government. Welfare in the current period is a result of the employed set of policy instruments in the past and current period. A crucial element in the analyses is that past policies affect past as well as present welfare. The welfare function is expressed by
Lt = (at-i - xt)2 + (at - Xt)2.
The real value of the disturbance is only observable at the end of period t. The beginning of period t is the latest time at which policy instruments can be set, if the policy maker wishes to affect the public’s behaviour in the current period t. The
The Case for Central Bank Independence
government chooses its set of policy instruments on the basis of obtained information about the future. It has to make a choice between two alternatives with different im plications for the conduct of policy. If the policy maker wishes to set his instruments before the beginning of period t, he chooses the path of pre-commitment, which involves a lesser degree of uncertainty. Whereas a government which opts for waiting, specifically until the beginning of period t, has the chance to obtain additional information. Cukierman and Meltzer refer to these additional information as ‘noisy indicators’ for the random state variables Xi and Xt+i which are obtainable through observations on the variables
yt^ = Xt + êt° and yt^=Xt+i + êt^
where êt® and êt^ are normally distributed white noise processes with zero mean, variance a higher quantity of available information increases the accuracy of the
forecast and enables the government to choose a more optimal set of policy | instrument. The government has no incentive to disclose precise information about yt. ; In fact, complete revelation of all available information on the side of the government ! would preclude the use of economic policy as a tool to improve re-election prospects. t The public has gone through changes in welfare and knows the level of welfare, Lt, | experienced in the past periods. It also knows the variance of states of nature, dfr-. But [ the public lacks information in the sense that it cannot determine the relative i
contributions of nature and governmental action to its welfare. Each government is
characterised by a different value of noise variance, which reflects its forecasting i
ability. This variance is unknown to the public but it tries to make inferences about i the noise variance on the basis of the level of welfare experienced during the i
incumbency period.
Cukierman and Meltzer use an apolitical benevolent government (social planner) as a i standard in their model, in order to observe how re-election prospects maximising I governments behave. The analysis shows that the loss of welfare for the society !
diminishes, when the social planner delays its decision to the latest possible time, which is the start of period t. This outcome contrasts with the result obtained for the case where a politically motivated government, facing a partially informed public, is entrusted with the well-being of the public. It turns out that the instruments are set optimally in the first period, t-1.
The model by Cukierman and Meltzer shows that eveiy government which maximises
its re-election prospects and possesses private information will choose not to maximise social welfare. The costs of democracy will already emerge whenever a substantial part of the public lacks information. Consequently, the combination of a government with discretionary power with uninformed voters will lead to a sub-
optimal outcome, due to the government’s behaviour prior to an election. The public
expects an increase in welfare before the election and makes the decision whom to vote for on that basis. This leads to a temporary increase in welfare before the election, at the cost of diminished welfare afterwards. The implementation of constitutional rules, as guidelines for the conduct of policy, is regarded by Cukierman and Meltzer as a possibility to eliminate the costs of democracy. The social outcome under discretion, proves unsatisfactory since it is sub-optimal, unless an apolitical social planner is in charge of generating welfare for the society. Therefore, the enforcement of a rule appears to be more agreeable as a possible solution. Various problems arise by implementing such a rule: One major problem being the authority which enforces the constitutional rule must have the same information as the government in order to be able to perform its function properly. The required full transfer of information is, first of all, far from easy and secondly, it may not be
desirable in all cases to disclose all available information. As a consequence “the implementation o f a constitutional rule rests ultimately with the ability o f the public and the incentives on the government to release all available information
The Case for Central Bank Independence
violation of the rule. The government will still be tempted to violate the rule whenever the costs of this violation will occur at a later date to the government than the costs of discretion to the public. This temptation is again directly due to the existence of information asymmetry.
Cukierman and Meltzer conclude from their study that the demonstrated principles
“apply to any area in which current governmental decisions affect welfare beyond the current period and in which there is asymmetric inform ation." Furthermore,
“divergence o f views within society, differences in information and the electoral advantage to the government from withholding information appear to be sufficiently common to explain why most government policy remains discretionary."^^