Dynamic Incentives and the Optimal Delegation of Political Power

Eric Le Borgne; Gauti B. Eggertsson

doi:10.5089/9781451866551.001.A001

Jump to Content Jump to Main Navigation

Dynamic Incentives and the Optimal Delegation of Political Power

Author:

Mr. Eric Le Borgne

0000000404811396

https://isni.org/isni/0000000404811396 International Monetary Fund

Search for other papers by Mr. Eric Le Borgne in
Current site
Google Scholar

and

Mr. Gauti B. Eggertsson

0000000404811396

https://isni.org/isni/0000000404811396 International Monetary Fund

Search for other papers by Mr. Gauti B. Eggertsson in
Current site
Google Scholar

Contributor Notes

Author’s E-Mail Address: Gauti.Eggertsson@ny.frb.org and ELeBorgne@imf.org

Publication Date:: 01 Apr 2007

eISBN:: 9781451866551

Language:: English

Keywords:: WP; mover accent

Download PDF (442.1 KB)

This paper proposes a theory to explain why a politician delegates policy tasks to a technocrat in an independent institution. It formalizes the rationales for delegation highlighted by Hamilton (1788) and by Blinder (1998). Delegation trades-off the cost of having a possibly incompetent technocrat with a long-term job contract against the benefit of having a technocrat who (i) invests more effort into the specialized policy task and (ii) is better insulated from the whims of public opinion. One natural application of the theory is in the field of monetary policy where the model provides a new theory of central bank independence.

Abstract

“It is a general principle of human nature, that a man will be interested in whatever he possesses, in proportion to the firmness or precariousness of the tenure by which he holds it; will be less attached to what he holds by a momentary or uncertain title, than to what he enjoys by a durable or certain title; and, of course, will be willing to risk more for the sake of the one, than for the sake of the other.”
Alexander Hamilton (1788), Federalist Paper 71: The Duration in Office of the Executive

“Many governments wisely try to depoliticize monetary policy by, for example, putting it in the hands of unelected technocrats with long terms of office and insulation from the hurly-burly of politics.” (emphasis added)
Alan S. Blinder (1998), former Vice-Chairman of the Federal Reserve, pp. 56—57.

I. Introduction

In most countries, some important and prestigious public policy tasks such as interpreting the constitution or conducting monetary policy, are delegated to public officials who are insulated from both job insecurity and political interferences thanks to long-term employment contracts. The judges at the U.S. Supreme Court, for example, are appointed for life and are independent of the elected executive. Similarly, many governments delegate monetary policy to an independent institution that is ruled by career public officials whose terms of office are much longer than the average elected official’s.

In some cases, delegation of political power is constitutionally mandated as in the case of the Supreme Court in the U.S., while in others, politicians find it in their own interest to delegate power even if they are not constitutionally required to do so. A voluntary transfer of power has been particularly noticeable in the case of monetary policy in recent decades. Granting the Bank of England independence, for example, was one of the first acts of the Labor government when it took power in the United Kingdom in 1997. That politicians give away power, on their own accord, may be considered somewhat of a puzzle. Most people and many political economy models, assume that politicians are in the game of accumulating power, rather than giving it away.

This paper tackles the following questions. First, why do politicians voluntarily give away some power and delegate critical public policy tasks to technocrats in independent institutions? Second, do politicians delegate enough tasks—i.e., can delegation be optimal from the perspective of society while not being in the interest of the politician, and if so, under what conditions?

These questions are analyzed by modelling the incentives that technocrats and politicians face. The model assumes a representative democracy where all agents have the same preferences.¹ The difference between a politician and a technocrat is that the politician is subject to elections while the technocrat is not. This difference, captured by short-term election pressures, drives all the results. Election pressures create a rich dynamic incentive structure in the model through the interaction of private and public signals and unobservable effort levels by the policymakers.

The framework proposed underlines a basic cost-benefit trade-off for delegation. The cost of delegation is that the technocrat cannot be dismissed from his job, even if he is incompetent. In the case of the U.S. Federal Reserve, this cost is not trivial since each governor is appointed by the president for a 14-year term (with the exception of the chairman).² In the U.S. Supreme Court, this cost is even starker because justices are appointed for life. Delegation, of course, is more costly if the ability of job candidates cannot be ascertained perfectly prior to hiring (a realistic feature of any hiring decision). Delegation is also more costly if the technocrat is not responsive to changing preferences of the electorate. On the benefit side, two benefits of delegation are identified. They formalize Alexander Hamilton’s argument for long-term contracts for the executive and Alan Blinder’s case for having a technocrat running monetary policy (both cited above).

Hamilton’s rationale for delegation to a technocrat with a long-term employment contract is that he will be ready to “risk more”/work harder for a policy task that he holds, with certainty, for a considerable amount of time. In the model a long-term employment contract enables the policymaker to have a long-term horizon which may improve his performance. In particular, it is shown that a longer employment contract gives the long-term appointee an incentive to invest more effort into his job specific decision making, due to a learning effect, thereby increasing the quality of his decisions. This is labeled the learning-by-doing effect (Arrow, 1962).

Alan Blinder’s argument for delegation is the second benefit being modeled. Blinder suggests that monetary policy should be put in the hands of unelected technocrats with long terms of office and insulation from the hurly-burly of politics. Blinder’s idea is formalized by showing that a politician has an incentive to follow the whims of public opinion. In particular, it is shown that when there is a public signal about the policy problem, the politician has a perverse incentive to follow public opinion even if he has superior private information that contradicts it. The reason is that the politician is reluctant to go against public opinion since if he is wrong (and the public is correct) this would signal low ability and lack of good judgement, and prevent the politician from being re-elected. In contrast, by following public opinion the politician reveals nothing about his ability so the public has no incentive to throw him out of office. A technocrat with a long-term job contract, however, does not face this dilemma since he is insulated from the electorate. This is labeled the insulation effect.

Several lessons emerge from the model, besides identifying and formalizing the basic trade-off outlined above. One is that the more complex the policy task, the more desirable it is to delegate it to a technocrat. Another is that the larger the degree of private benefit from political office (which may include outright corruption), the lower the politician’s incentive to delegate. Uncertainty about the competency of officeholder works in the same direction. It is also shown that delegation is always socially optimal when the policymaker bases his decisions on public opinion, even if the politician may not always find it privately optimal to delegate power in these circumstances. Furthermore, if delegation occurs, it is always socially optimal in the model. The converse, however, is not true; i.e., some tasks are not delegated in equilibrium even though their delegation would improve social welfare.

The model of dynamic incentives in public policy that is developed builds on the seminal career concerns model of Holmström (1982/1999). Holmström shows that, within a firm, a manager of unknown ability can be induced to supply effort that is not directly observed by relying on the manager’s concerns about his future career; i.e., implicit incentives motivate the manager. We extend Holmström’s model in three ways. First, the manager/policymaker is disciplined by elections, whereas in Holmström’s model incentives are shaped by the expectation of higher future wages. Second, learning is introduced in the model to formalize Hamilton’s learning-by-doing effect. Third, a public signal about the desirability of the policy decisions is introduced so as to formalize Blinder’s insulation effect. The first extension was first introduced by Persson and Tabellini (2000).³ The second two are new to the literature and are necessary to model the benefits of delegation that are emphasized. A further difference to Persson and Tabellini (2000) is that the model develops a signal extraction game which may be helpful for many applications, such as studying monetary policy issues, as illustrated in an example in Section V.

The paper provides a formalization of the literature on constitutionalism, whereby certain issues are taken “off the table” in everyday policymaking in order to insulate policymakers from short-term pressures of public opinion. One criteria for insulation in this literature is that there should be a broad agreement on the (long-term) goals of policies that are taken “off the table” (see, e.g., Elster (1995) and Drazen (2002)), a criteria that the model satisfies, since all agents have the same preferences.

Maskin and Tirole (2004) address questions similar to this paper’s. They investigate which policy decisions should be made by politicians and which ones should be decided by judges.⁴ A similarity between this paper and theirs is that both illustrate that elections give a politician perverse incentives to follow public opinion rather than taking advantage of superior private information. In Maskin and Tirole’s two period model, a politician may have a pandering motive because he has a hidden agenda. A politician who is a corporate crony, for example, pretends to be consumer friendly to hide his true preferences for corporate pork, in order to ensure an election in the second period. In the second period he is free to act on his corporate cronyism because he will not be subject to elections at that time. In contrast to Maskin and Tirole’s pandering effect, the insulation motive developed in this paper is not related to hidden agendas. Instead the politician might follow public opinion to eliminate signals about the quality of his judgement. A politician may choose to support a popular war, for example, not because he wishes to hide that he is a “dove” (so he can act upon his true dovish motives in period 2), but because if the war turns out to be successful he does not want to be the only one who voted against it since this would reveal lack of good judgement. A politician in the model is tempted to follow public opinion because even if the war turns out badly he is not likely to be thrown out of office if everyone else made the same error in judgement. The insulation motive therefore relates to a different set of policy tasks than Maskin and Tirole’s hidden agenda model. It arises because elections give rise to perverse dynamic incentives even if all agents have the same political agenda.

Alesina and Tabellini (2006) also address questions similar to this paper’s, i.e., when should policy tasks be delegated to technocrats? A key difference is that Alesina and Tabellini assume that the trade-off between technocrats and politicians stem from the different intrinsic preferences of these agents: some agents are born politicians while others are born technocrats. In contrast, the model developed in this paper assumes a representative democracy where everyone has ex ante the same preferences. To some extent one may interpret the result obtained as giving foundations for different reduced form utility functions of technocrats versus politicians by modelling how elections change the dynamic incentives of policymakers. Another difference is that while this paper’s model illustrates the role of insulation and learning, Alesina and Tabellini mainly focus on other reasons for delegation such as issues related to dynamic inconsistency of public policy, which is an important consideration that is abstracted from in this paper.

II. The Framework

In this section the basic political economy setup is developed. The economy is populated by a large number of citizens and evolves over two time periods, t = 1, 2. There is a political office that can be occupied by only one citizen, the “officeholder.” In this representative democracy, the officeholder is entrusted with (and held accountable for) policy decisions. The game begins with the random selection/election from the population of a citizen to become the elected officeholder (called “politician” thereafter). The politician can decide whether to make the policy decision himself or to delegate it to an appointed agent, which we call a “technocrat,” who is not subject to re-election (i.e., he has a longer term in office).

A. Political Agency Setup

A simple setup where citizens’ utilities are a function of a policy decision is presented. The quality of this policy decision depends both on ability and effort of the officeholder. An individual’s utility is given by

U_{t}^{j} = E \sum_{t = 1}^{2} υ_{t}^{j}

where the index j refers either to the officeholder (“o”) or a citizen (“c”). The period payoff is

\begin{matrix} υ_{t}^{j} = υ_{t}^{c} (i_{t,} {\bar{r}}_{t}) + R^{j} - \frac{α}{2} {(e_{t}^{j})}^{2} & (1) \end{matrix}

where $υ_{t}^{c} (.)$ is a payoff due to a public policy that accrues to all citizens at time t; this payoff depends on the interaction of the policy decision i_t and a shock ${\bar{r}}_{t}$ . The term $\frac{α}{2} {(e_{t}^{j})}^{2}$ represents the cost of effort. The quality of the policy decision (i.e., the choice of i_t) is a function of how well the policymaker can predict ${\bar{r}}_{t}$ . This prediction depends on the effort and ability of the policymaker in a way is described in the next subsection. Since effort is only useful to the officeholder e^j =0 when j = c. R^j is a binary variable which is equal to R if citizen j is the officeholder (j = o), and zero otherwise. R is an “ego rent” from being in office and managing the public good (as in Rogoff and Sibert, 1988), deriving from the prestige associated with managing public affairs. To economize on notation, discounting is abstracted from since it does not add any insight.

To clarify the game, more structure is added to (1) although the more general formulation is used in Section V. In each period t = 1, 2, whoever is the officeholder has to make a binary policy decision by choosing $i_{t}^{H}$ or $i_{t}^{L}$ . This decision can refer to any public policy; one may for example think of it as choosing between whether or not to fight a war, to invest in a public project, or raising or lowering interest rates. For simplicity it is assumed in the baseline illustration that ${\bar{r}}_{t}$ can only take on two values, either ${\bar{r}}^{H}$ or ${\bar{r}}^{L}$ with equal probability, i.e., $Prob ({\bar{r}}_{t} = {\bar{r}}^{H}) = Prob ({\bar{r}}_{t} = {\bar{r}}^{L}) = \frac{1}{2}$ , and it is assumed that the policymaker (and the public) do not know this variable before choosing i_t. The policy problem is thus to select i_t optimally conditional on a forecast about ${\bar{r}}_{t}$ . In the baseline illustration, the payoffs are normalized so that if the policymaker correctly predicts ${\bar{r}}_{t}$ then

\begin{matrix} υ^{c} (i^{H}, {\bar{r}}^{H}) = υ^{c} (i^{L}, {\bar{r}}^{L}) = 0 & (2) \end{matrix}

and if he makes a mistake then

\begin{matrix} υ^{c} (i^{H}, {\bar{r}}^{L}) = υ^{c} (i^{L}, r^{H}) = - 1. & (3) \end{matrix}

1. Private signal

The oﬃceholder receives a private signal σ_t about the realization of the shock ${\bar{r}}_{t}$ . It is assumed that the quality of the private signal depends both on eﬀort and ability of the oﬃceholder. To illustrate the point of the model in the simplest setup possible, it is assumed that two signals can occur that correspond to the two shocks, i.e., σ_t∈{σ^H, σ^L}. If the oﬃceholder bases his decision on his private signal, his optimal policy decision is trivial: he will set i^H if he observes σ^H and i^L if he observes σ^L. The combination of the oﬃceholder’s forecast and the realization of the shock produce a state s. Four possible states can therefore arise in a given period depending on whether the shock (labeled as H or L) has been rightly (R) or wrongly (W) predicted: i.e.,s = rσ∈{HR, HW, LR, LW}. Since we assume symmetric payoﬀs in the baseline example for HR, LR on one hand and HW, LW on the other, the only pay-oﬀ relevant events are whether the oﬃceholder is right or wrong. The probability that a new oﬃceholder receives a right signal σ conditional on a shock $\bar{r}$ , his effort level e, and expected ability θ is:⁵

\begin{array}{l} Prob (Right signal for the officeholder) = Prob (σ^{H} | {\bar{r}}_{1} = {\bar{r}}^{H}, e_{1,} θ) = Prob (σ^{L} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1,} θ) \\ = \frac{1}{2} + θ + e_{1} \end{array}

and the probability of the wrong signal is:

\begin{matrix} \begin{array}{l} Prob (Wrong signal for the officeholder) = Prob (σ^{H} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1,} θ) = Prob (σ^{L} | {\bar{r}}_{1} = {\bar{r}}^{H}, e_{1,} θ) \\ = \frac{1}{2} - θ - e_{1} \end{array} & (4) \end{matrix}

Here θ is the forecasting ability of the officeholder. It is a random draw from a distribution that can take two values “good” or “bad,” i.e., θ_G > θ_B = 0 with equal probabilities $\frac{1}{2}$ . α ∈ {G, B} are referred to as the (ability) types of the citizens. It is assumed that citizens do not know θ but they all know the joint distribution of θ, i.e., there is symmetric incomplete information. This uncertainty can also apply to a subset of the pool of candidates for public office (e.g., citizens with a Ph.D. in economics). All that is needed for the results is that there is some remaining uncertainty among candidates. Of course, the probability needs to be between 0 and 1 which implies bounds on the feasible values of e and θ.

When the first-period officeholder is in office in period two, it is assumed there is a learning-by-doing effect (Arrow, 1962), so that

\begin{matrix} Prob (Right signal for officeholder) = Prob (σ^{H} | {\bar{r}}^{H}, e_{2}, θ) = Prob (σ^{L} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1}, θ) \\ (5) \end{matrix}

\begin{matrix} = \frac{1}{2} + θ + β_{e 1} + e_{2} & (6) \end{matrix}

and then the probability of the wrong signal is

\begin{matrix} Prob (Wrong signal for officeholder) = Prob (σ^{L} | {\bar{r}}_{1} = r^{L}, {\bar{r}}_{1}, θ) = Prob (σ^{H} | {\bar{r}}_{1} = r^{H}, e_{1}, θ) \\ (7) \end{matrix}

\begin{matrix} = \frac{1}{2} - θ - β e_{1} - e_{2} & (8) \end{matrix}

where the coefficient β> 0. This coefficient is interpreted as corresponding to the complexity of the task at hand since it implies that effort put in the first-period translates into better learning for period two forecasting. For a complex task, one would expect job experience and learning on the job to be extremely important while less relevant for a simple or a manual task. Adding the term βe₁in the second period production function is a simple and transparent way of introducing a learning-by-doing effect, i.e., experience in office increases the ability of the officeholder. Thus, effort in the first-period is akin to a sunk cost) investment; if the first-period officeholder is also in office in the second period, his effort invested in learning (say about the functioning of an economy, the structure of the central bank, etc.) means his accumulated (task specific) knowledge gives him an incumbency advantage compared to a new officeholder.

In the case where there is a new officeholder in period 2 he uses the forecasting technology in 4; a period 1 officeholder that is thrown out of office cannot pass on his individual-and-job-specific knowledge to his successor.

2. Public signal

It is also assumed that there is a public signal ρ_t about the realization of the shock ${\bar{r}}_{t}$ that can take the values ρ^H and ρ^L. It is assumed that this signal is created through a process similar to the private signal, i.e.,

\begin{matrix} Prob (Right public signal) = Prob (ρ^{H} | {\bar{r}}_{1} = {\bar{r}}^{H}) = Prob (ρ^{L} | {\bar{r}}_{1} = {\bar{r}}^{L}) \\ = \frac{1}{2} + E_{p} θ = \frac{1}{2} + \frac{1}{2} θ_{G} \end{matrix}

and the probability of the wrong signal is

\begin{matrix} \begin{matrix} Prob (Wrong public signal) = Prob (ρ^{H} | {\bar{r}}_{1} = {\bar{r}}^{L}) = Prob (ρ^{L} | {\bar{r}}_{1} = r^{H}) & (9) \\ = \frac{1}{2} - E_{p} θ = \frac{1}{2} - \frac{1}{2} θ_{G} \end{matrix} \end{matrix}

where E_pθ is the expected ability of the whole population. This captures the idea that public opinion is formed by the average cognitive abilities of the citizens. Because each citizen does not expect to be pivotal in the policy decision, they do not exert any effort to improve the quality of the public signal. Hence the signal tends, on average, to be of worse quality than the private signal of the officeholder as long as he exerts a positive effort level.

3. Institutions

The politician (randomly selected from the set of citizens), once “elected” at the beginning of period t =1 decides to have the public good being produced under one of two possible institutions/regimes:

Delegation

At the beginning of period t =1, the politician appoints an agent (a technocrat) to be officeholder and make the policy decision. Since all citizens are ex ante identical, this agent is randomly selected by the politician from the set of citizens, and is (contractually) in office for both periods. When the politician delegates the policy decision, his utility then becomes the same as that of a (representative) citizen (he is not in charge of policy, so j ≠ o in (1)) except that he receives a small political rent from office R^D. This captures the idea that elected office confers some prestige to the politician, and the fact that voters, who in equilibrium understand the motives for delegation, view favorably a politician who delegates a task (say monetary policy), as citizens know that this is welfare increasing for them (this is shown later in the paper), or the fact that the politician is the principal of the technocrat, a position that confers some “ego rent.” It is assumed that delegation is credible, i.e., that the politician can make a binding promise not to fire the person he delegates powers to for the duration of the game. This is an important—and entirely reasonable—assumption. To remove a governor of the Federal Reserve or a supreme court justice, for example, is exceedingly hard, and while not impossible, it would be prohibitively costly for most elected politicians.⁶

Political control

At the beginning of period t =1, the elected politician decides to make the policy decision himself. The politician is in office during period t =1 but faces an election at the beginning of period 2. At this stage, an opponent is randomly selected from the set of remaining citizens. Citizens then vote on the opponent versus the incumbent, and the winner is the officeholder in period t = 2.

Under both delegation and political control the individual rationality condition that the officeholder prefers to be in office than not is imposed (which is simply the condition that the net rent from office is positive). The timing of events in the game is shown in Figure 1.

III. Hamilton’s Hypothesis: Learning

This section reveals the basic trade-off between delegation and political control (while abstracting from the public signal). This is shown by illustrating the learning effect that arises due to delegation (this formalizes Hamilton’s hypothesis). In Section 4, a public signal is added to the model; this generates an insulation incentive.

A. Equilibrium

In order for the politician to decide whether to make the policy decision himself or to delegate it, he compares his utility under both regimes. This, in turn, is a function of the equilibrium level of effort and ability. First, the effort level chosen in period 2 is derived. This effort level is the same whether or not there is delegation, since the officeholder faces the same maximization problem in either case. Then the effort choice in period 1 is calculated, which depends on the policy regime.

1. Period 2 effort

The officeholder’s decision problem is solved by backward induction. In the last period (t = 2) the game is a static one. The maximization problem of the officeholder (politician or technocrat) is to select e₂ to maximize utility in period 2 that is given by

U_{2} {R - \frac{1}{2} α e_{2}^{2}} - {\frac{1}{2} - E_{a} θ - e_{2}}

The first bracket is the net rent from office. The second bracket is the probability of the officeholder choosing the wrong policy decision. This probability is decreasing in effort and expected ability of the officeholder, denoted E_aθ. If the officeholder was not in office in the previous period then. $E_{a} θ = \frac{1}{2} θ G$ . If the officeholder was in office in the previous period then $E_{a} θ = \frac{1}{2} θ_{G} + β e_{1}$ because of the learning-by-doing effect explained above. Recall that it is assumed, for simplicity, that $υ^{c} (i^{H}, {\bar{r}}^{H}) = υ_{t}^{c} (i^{L}, {\bar{r}}^{L}) = 0$ so only the probability of making the wrong decision appears explicitly in this utility function. Also recall that $υ^{c} (i^{H}, {\bar{r}}^{L}) = υ^{c} (i^{L}, {\bar{r}}^{H}) = - 1$ is normalized so that there is nothing that scales the probability in the utility function. The first order condition of this maximization problem is αe₂ – 1 = 0 so that equilibrium effort, denoted e* is:

\begin{matrix} e^{*} = α^{- 1} > 0 & (10) \end{matrix}

The effort choice does not depend on E_aθ, thus it is independent of the officeholder’s ability and his tenure.

2. Period 1 effort under delegation

When the politician delegates the policy decision, the first-period maximization problem of the technocrat (superscript “T”) is to select e₁ (taking the optimal choice of e₂ as given) to maximize expected utility in period 1 that is given by

\begin{matrix} U_{1}^{T} = {2 R - \frac{1}{2} α e_{1}^{2} - \frac{1}{2} α e_{2}^{2}} - {\frac{1}{2} - \frac{1}{2} θ_{G} - e_{1}} - {\frac{1}{2} - \frac{1}{2} θ_{G} - β e_{1} - e_{2}} & (11) \end{matrix}

The first bracket captures, again, the net rent from office of the technocrat. The second (third) bracket is the probability of taking the wrong policy decision in period one (two). The probability of making a mistake in period 2 is also a function of effort in period 1 because of the learning-by-doing effect. This leads to first order condition $α e_{1}^{T} - (1 + β) = 0$ so that a technocrat’s first-period equilibrium level of effort is

\begin{matrix} e_{1}^{T} = \frac{1}{α} (1 + β) > 0 & (12) \end{matrix}

The first-period effort level is composed of two terms; the first (1) is the first-period gain from increasing effort, while the second (β) represents the marginal gain due to the value of effort in period 1 on the quality of the policy decision in period 2; this term is called the learning-by-doing effect. This arises because effort in period one has a long-lasting effect on the officeholder’s forecasting ability. This implies that effort is greater in period 1 than the period 2 effort level, i.e.,

\begin{matrix} e_{1}^{T} = \frac{1 + β}{α} > e^{*} = \frac{1}{α} > 0 & (e1A) \end{matrix}

3. Period 1 effort without delegation

The analysis is more complex without delegation (i.e., political control) because it is now a game of incomplete information. The perfect Bayesian equilibria (PBE) of this game is characterized. The effort choice of the officeholder in period 2, whether he is new in office or not, is e* = α⁻¹. So, the expected utility from electing a new politician in office is

U_{2}^{N P} = - {\frac{1}{2} - \frac{1}{2} θ_{G} - e^{*}}

where NP stands for electing a “new politician.” In deciding between a new officeholder and the incumbent voters compare the expected utility of retaining the incumbent politician which is given by:

U_{2}^{I P} = - {\frac{1}{2} - q (s_{1}) θ_{G} - β e_{1} - e^{*}}

where IP stands for re-electing the “incumbent politician.” The variable q(s₁) is indexed by the realization of the state in period 1. It is the probability voters put on whether the policymaker is a good type. Note that this probability depends on the realization of the policy decision and the shock ${\bar{r}}_{1}$ both of which are observed before voting. The politician is thrown out of office if $U_{2}^{I P} < U_{2}^{N P}$ i.e.,

\begin{matrix} q (s_{1}) θ_{G} + β e_{1} < \frac{1}{2} θ_{G} & (13) \end{matrix}

Note that the learning-by-doing effect, i.e., βe₁, gives the politician an incumbency advantage. The probability q(s_i) is computed by using Bayes rule. In the case where the politician makes a wrong forecast, for example, it can be calculated by computing the probability that the shock is low when the signal is high (this probability is the same, given the above assumption, as the one when the shock is high but the signal is low), i.e.,

\begin{array}{l} q (s_{1} \in Wrong signal) = qw (θ_{G}, e_{1}) = Pro b (θ_{G} | {\bar{r}}_{1} = {\bar{r}}^{L}, σ^{H}, e_{1}) \\ = \frac{\frac{1}{2} Prob (σ^{H} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1,} θ_{G})}{\frac{1}{2} Prob (σ^{H} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1}, θ_{G}) + \frac{1}{2} Prob (σ^{H} | {\bar{r}}_{1} = {\bar{r}}^{L}, e_{1}, θ_{B})} \\ = \frac{\frac{1}{2} - θ_{G} - e_{1}}{2 (\frac{1}{2} - \frac{1}{2} θ_{G} - e_{1})} \end{array}

Substituting this into (13), the officeholder is thrown out of office if and only if the incumbency effect is small enough, i.e.,

\begin{matrix} β e_{1} < \frac{\frac{1}{2} θ_{G}^{2}}{1 - θ_{G} - 2 e_{1}} & (14) \end{matrix}

In order to focus on election versus appointment of officeholder, it is assumed that this condition is satisfied, since if it were violated the incumbent would always be elected.⁷ β can always be chosen small enough so that this condition is satisfied (since in the limit, as β = 0, there is no incumbency advantage).

Using Bayes rule, the probability of the policymaker being a good type conditional on receiving the right signal is given by

\begin{matrix} q (s_{1} \in Right signal) = qR (e_{1}, θ_{G}) = \frac{\frac{1}{2} + θ G + e_{1}}{2 (\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1})} & (15) \end{matrix}

The choice of first-period policies and effort of the incumbent politician is the e₁ (taking e₂ as given) that maximizes expected utility in period 1 which is given by

\begin{matrix} \begin{array}{l} U_{1}^{P} = {R - \frac{1}{2} α e_{1}^{2}} + {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}} (R - \frac{1}{2} α e_{2}^{2}) - {\frac{1}{2} - \frac{1}{2} θ_{G} - e_{1}} & (16) \\ - {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}} {\frac{1}{2} - qR (e_{1}, θ_{G}) θ_{G} - β e_{1} - e_{2}} - {\frac{1}{2} - \frac{1}{2} θ_{G} - e_{1}} {\frac{1}{2} - \frac{1}{2} θ_{G} - e *} \end{array} \end{matrix}

The difference between this expression and the utility of the technocrat $(U_{1}^{T})$ is that the politician is uncertain whether or not he will stay in office in period 2. This is the reason why the net rent from office in period 2, i.e $(R - \frac{α}{2} e_{2}^{2})$ , is multiplied by ${\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}}$ which is the probability of the officeholder being right in period 1. If he is wrong, he loses the election in the beginning of period 2. The third curly bracket corresponds, as in the delegation case, to the probability of making a wrong decision in period 1. The first term of the second line corresponds to the loss of utility associated with making a wrong forecast in period 2 and this term is weighted by the probability of the politician still being in office. Note that the probability of the politician still being in office needs to be weighted by the probability of being a good type conditional on the politician being re-elected (i.e., he made the right decision in period 1) using the Bayes rule. Finally, the last term is the utility of the politician in period 2 if he gets the policy wrong in period 1, weighted by the probability of that happening. In this case, he will receive the same utility as the average citizen and the outcome does not depend on his effort in period 1 or 2.

The first order condition with respect to effort is:

\begin{matrix} α e_{1} = (R - \frac{1}{2 α}) + 1 + β (\frac{1}{2} + \frac{1}{2} θ_{G} + 2 e_{1}) + θ_{G} {q_{R} - \frac{1}{2}} + \frac{\partial q_{R}}{\partial e_{1}} θ_{G} {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}} & (17) \end{matrix}

Using (15) one observes that both last two terms equal $\frac{1}{4} \frac{θ_{G}^{2}}{(\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1})}$ but with the opposite sign and thus cancel out. Assuming that α ≠ 2β to ensure that a solution exists the effort level that solves (17), denoted $e_{1}^{P}$ , is

\begin{matrix} e_{1}^{P} = \frac{{1 + β (\frac{1}{2} + \frac{1}{2} θ_{G})} + {R - \frac{1}{2 α}}}{α - 2 β} > 0 & (18) \end{matrix}

It is shown in the appendix that the denominator of this expression has to be positive. There are two effects that influence the politician’s effort, the learning-by-doing effect and the career concerns effect. As already discussed, the former is related to both the complexity of the task and the expected first-period ability of the officeholder, while the latter is an increasing function of the net rent from office. Under the delegation regime the career concerns effect was not present since the officeholder is in office in period two with probability one (so that his net rent is guaranteed). The career concerns effect (extending Holmström’s terminology to the political/administration context) captures the incentive the first-period incumbent has to increase effort above its myopic level so as to raise the probability of staying in office and receiving the net (private) rent from office on top of the utility that each (representative) citizen obtains. The career concerns effect could also be coined a tournament effect (Lazear and Rosen (1981); Green and Stokey (1983)), since it induces extra effort through the reward of a prize (the private rent) which is only available to the winner of the election.

B. Delegation of Political Power

1. Welfare and the Existence of Trade-offs Between Regimes

One can now analyze whether a newly elected politician has any incentive to delegate the policy decision. Recall that, by assumption, if a politician delegates the policy decision but still remains in office he gets a rent R^D > 0 (as the principal of the technocrat) that is strictly less than the net rent from office that the technocrat obtains (and the politician supplies not effort). The politician’s utility when he delegates is called U^D, and recall that his utility is denoted U^P if he does not delegate. Thus the politician delegates if and only if

\begin{matrix} U^{D} > U^{P} & (19) \end{matrix}

where, after substituting the equilibrium effort values into the above equation, we get

\begin{matrix} U^{D} = 2 R^{D} - 1 + θ_{G} + \frac{{(1 + β)}^{2} + 1}{α} & (20) \end{matrix}

which is only a function of the structural parameters R^D, θ_G, β and α. Note that this utility is an increasing function of θ_G with a slope of 1. The utility of the politician if he does not delegate can also be written in terms of the structural parameters by substituting (10) and (18) into (16). We can express U^P in terms of a quadratic function of θ_G so that

\begin{matrix} U^{P} = γ_{1} + γ_{2} θ_{G} + γ_{3} θ_{G}^{2} & (21) \end{matrix}

where

γ_{3} = (\frac{1}{4} \frac{β^{2}}{α - 2 β} + \frac{1}{4} \frac{β^{3}}{{(α - 2 β)}^{2}} - \frac{1}{8} \frac{β^{2}}{{(α - 2 β)}^{2}} + \frac{1}{4}) > 0

and it is shown in the appendix that γ₃ must be positive for all permissible parameters in the model. Therefore, utility is increasing in θ_G in the positive quadrant (U^D, θ_G) space.

The generic form of U^P and U^D as a function of θ_G (or, in the general case where θ_B ≠ 0: θ_G –θ_B) is shown in Figure 2. The curve when the politician chooses not to delegate is denoted “politician” while the curve when he chooses to delegate is denoted “technocrat.”

As can be seen, there are, in general, two possible intersections for these curves. However, only one of them is an admissible equilibrium of the model, namely point B. To see this suppose that point A is an equilibrium. Note that the slope of the U^P curve is smaller than that of U^D. Consider now the utility under Delegation. Using (20) one has

\frac{\partial U^{D}}{\partial θ_{G}} = 1

and, using the envelope theorem, the slope of U^P can be calculated by taking a partial derivative of (16) to obtain⁸:

\frac{\partial U^{P}}{\partial θ_{G}} = 1 + \frac{1}{2} (R - c (e_{2})) + \frac{1}{2} θ_{G} β e_{1} + \frac{1}{2} (qR - \frac{1}{2}) + \frac{\partial q_{R}}{\partial θ_{G}} {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}} > \frac{\partial U^{D}}{\partial θ_{G}}

but we know that this cannot be the case in point A (because the slope of U^P line in Figure 2 is smaller than the technocrat line). So this cannot be an equilibrium for the parameter values assumed (i.e., positive rents and positive effort).

The discussion above indicates that, for a given set of parameters, equilibrium will be on the right hand side of point A in Figure 2. If the ability spread is in the range between A and B delegation dominates political control, while political control dominates delegation to the right hand side of B. To have an interesting theory of endogenous delegation one must show that an equilibrium exists on both sides of point B, depending on the parameters. If this can be established, one can discuss how the equilibrium depends on parameter values and whether or not delegation occurs in equilibrium. Thus the question of most economic interest is whether there exists a trade-off between delegation and political control, i.e., are there configurations of the parameters on both sides of B that satisfy all the restrictions of the model?

In addition to the welfare functions derived in equations (20) and (16) a candidate solution has to satisfy the conditions that (i) the implied probabilities of every event are between 0 and 1; (ii) effort is positive; and (iii) the individual rationality constraints of the politician and technocrat are satisfied. The proof of existence of this policy trade-off is trivial. One only need to establish a numerical example that shows that for one set of parameters delegation dominates and in another it does not. This example is given below. With existence of solution established one can discuss how the welfare trade-off depends on the different parameters of the model. That discussion does not require any of parameter values in Example 1.

Example 1.

Suppose that the coefficients of the model are (θ_G, R, R^D, α, β) = (0.2, 0.2, 0.15,100,1). This implies that $(e_{1}^{T}, e_{1}^{P}, e *, U^{D}, U^{P})$ = (0.02, 0.018, 0.01, –0.45, –0.46), i.e., U^D > U^P, the necessary conditions are satisfied, and delegation dominates. If it is assumed instead that θ_G = 0.4 (and the same values for the other parameters) then $(e_{1}^{T}, e_{1}^{P}, e *, U^{D}, U^{P})$ = (0.02, 0.0193, 0.01, –0.25, –0.2402), i.e., U^D < U^P, the necessary conditions are satisfied, and delegation is dominated so that the politician retains control.

2. The Welfare Trade-off

The last section established that endogenous delegation can occur and that, whether or not it happens depends on the parameters of the model. In other words, the nature of the policy task has an effect on whether or not delegation takes place. It was established earlier that it can be individually rational for a self-interested politician to delegate policy decisions. In this case the officeholder becomes the agent of the elected politician, who himself is the representative of citizens at large. It is also clear that the individual rationality constraint of an appointed officeholder is satisfied since he also gets a strictly positive private rent from office so that he is better off than the rest of the citizenry. It was also showed that the utility of the officeholder is increasing in θ_G whether or not he delegates power but it increases more if he retains power, i.e., the slope of U^P is greater than U^D.

The relationship between the utility of the politician with and without delegation is shown in Figure 3 below. This figure corresponds to the right hand side of point A in Figure 2. For small values of the ability parameter, delegation to a technocrat dominates. As θ_G increases (or, more generally, the spread between θ_G and θ_B) the utility of keeping power increases until the two lines cross so that the politician does not delegate any more. The intuition is straightforward and captures—albeit in a crude fashion—the main trade-off between delegation and democratically accountable power: if the politician appoints an independent technocrat he cannot fire him (even if he turns out not to be as competent as expected)! If there is significant uncertainty about the ability of the technocrat, the politician will be increasingly reluctant to give away power that extends beyond his election term because society can be stuck with an incompetent technocrat that cannot be fired.

The slope of the two curves, therefore, indicates that the politician will be more willing to delegate power to technocrats if he can easily identify whether or not the candidates are qualified to handle it, since this implies a lower spread between θ_G and θ_B. In the case of a Governor of the Federal Reserve, for example, one may argue that it is fairly easy to identify the necessary qualification of a candidate (e.g., experience in banking or academia, a Ph.D., etc.). The nature of the task at the Federal Reserve is also fairly uncontroversial and well agreed upon (i.e., pursue “maximum employment” and “stable prices”). Similarly, before appointing a judge to the Supreme Court, the politician can observe the track record of the technocrat (i.e., the judge) and the objective of the judge is fairly well agreed upon (interpreting the constitution). Arguably the ability of technocrats to perform some other tasks may be subject to a greater uncertainty. Some public offices, fiscal policy for example, depend on interpreting the wishes and needs of the electorate which may frequently shift over time. In this case, technical qualification may not be sufficient for determining whether a candidate is suitable for appointment or not, which is an argument for the politician to maintain power in his own hands.

There are other parameters, however, that can lead to endogenous delegation, namely the complexity of the task and rents from office.

3. Comparative Statics

Having established the existence of a trade-off between delegating power and retaining it, and having showed how it depends on the expected ability of the officeholder one can now turn to analyzing the effect of the learning-by-doing effect, the cost of effort, and rents.

Learning-by-doing (β) What is the effect at the margin (i.e., around the cutoff point of the two curves in Figure 3) of increasing the learning-by-doing effect in the forecasting task? To find this, one takes the partial derivative of U^D, i.e.,

\frac{\partial U^{D}}{\partial β} = e_{1} = 2 \frac{1 + β}{α}

The envelope theorem can be used to calculate the derivative of U^P with respect to β, i.e.,

\frac{\partial U^{P}}{\partial β} = e_{1} (\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1})

Consider the point at which the two curves intersect. At this point, one knows that the effort level expanded by the technocrat must be higher than that of the politician, so that $e_{1}^{P} < e_{1}^{T},$ , since the technocrat needs to compensate the politician for the loss of rent. This implies that

\frac{\partial U^{P}}{\partial β} = e_{1}^{P} (\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}^{P}) < \frac{1 + β}{α} (\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}^{P}) < \frac{\partial U^{D}}{\partial β}

and the inequality follows because it is known that $(\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}) < 1 .$

So, the more learning-by-doing on the job, the more desirable it is to delegate power as shown in Figure 4. First, observe that increasing β shifts both curves up so that it increases welfare in both regimes. The reason is that in both cases the officeholder will inherit his effort in period 1 (in the case of the politician, this only increases welfare if he is reelected) and the higher β the more useful this will be for policy. The result above illustrates, however, that utility under the delegation regime (the solid curve) increases by more than utility if the politician retains power. The reason for this is that the technocrat has a higher stake in being in office (because he cannot be fired) so that his effort in period 1 is more sensitive to the learning parameter. In contrast, the envelope theorem implies that if the politician conducts the policy task, then $\partial e_{1}^{P} / \partial β = 0$ (when this derivative is evaluated at the optimal $e_{1}^{P}$ ). Since the utility of the politician (and society in general) is increasing in the effort of the technocrat in period one, this increase in the effort by the technocrat raises the politician’s utility.

Thus, the more learning-by-doing, the more important it is to give it to a technocrat with a long-term contract because the long-term benefit of being in office gives him the incentive to invest in learning (supplying effort). This in effect is reminiscent of Hamilton’s claim in the Federalist Paper 71 about the “tenure of the executive” cited in the introduction. Hamilton states “it is a general principle of human nature, that a man will be interested in whatever he possesses, in proportion to the firmness or precariousness of the tenure by which he holds it; will be less attached to what he holds by a momentary or uncertain title, than to what he enjoys by a durable or certain title; and, of course, will be willing to risk more for the sake of the one, than for the sake of the other.” In the case of monetary policy there are probably few policy decisions that require more specialized knowledge at the decision level than monetary policy. The Federal Reserve in the U.S., for example, is the single largest employer of Ph.D. economists in the world. The discussion at FOMC meetings is highly technical where district bank presidents and staff economists may bring to bear a wealth of statistical and economic concepts before policy decisions are made. It is not uncommon, for example, that policymakers discuss results from sophisticated statistical models with statical error bands, use anecdotes from different Federal Reserve district, and discuss different economic theories to interpret several different data series. There are few, if any, public policy tasks that involve technical discussion of this kind. In the case of fiscal policy, for example, it is rare that economic or statical theory is discussed in the context of everyday decision making at the U.S. Treasury. It is not always the case, even, that the Secretary of the Treasury is familiar with the technical language required.

One possible interpretation of β is that it reflects “complexity” of the task. It seems reasonable to assume that more learning on the job is important in tasks that are complex than those that are relatively manual and simple.

Private information (1/α). Next comparative static with respect to the parameter α⁻¹ which indexes the cost of effort are investigated. The higher α⁻¹ the more effort both the technocrat and the politician will exert and thus the better informed they will be to make the best choice. As α⁻¹ reaches the limit of feasible range for the model parameters, the policymaker will exert enough effort to be perfectly informed and reach the best possible decision. The interest is in knowing the effect of increasing 1/α at the margin. The partial derivative of U^D is

\frac{\partial U^{D}}{\partial α^{- 1}} = {(1 + β)}^{2} + 1

Using the envelope theorem to calculate the derivative of U^P with respect to α⁻¹:

\frac{\partial U^{P}}{\partial α^{- 1}} = \frac{1}{2} α^{2} {(e_{1}^{P})}^{2} + {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}^{P}} \frac{1}{2}

Consider now the point at which the two curves U^P and U^D intersect. At this point, it is known that $e_{1}^{P} < e_{1}^{T}$ by the same argument as in the last section, so that

\frac{\partial U^{P}}{\partial α^{- 1}} = \frac{1}{2} α^{2} {(e_{1}^{P})}^{2} + {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}^{P}} \frac{1}{2} < \frac{1}{2} {(1 + β)}^{2} + {\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1}^{P}} \frac{1}{2} < \frac{\partial U^{D}}{\partial α^{- 1}}

which suggests that the comparative static of α⁻¹ is the same as that of β. Thus, the analysis suggests that the higher α⁻¹ the higher the incentive to delegate policy to a technocrat and the graphical analysis in Figure 4 applies once again. IT was suggested in the last subsection that the parameter β was a measure of “complexity” of the policy task. A similar interpretation of α⁻¹ is also suggested because it measures the extent of “private information” the policymaker has relative to the public. Further discussion of this interpretation is deferred to the next section where it is discussed in the context of the public signal when this interpretation is natural.

Rents and Corruption. Finally, one can also see that increasing R will shift the “politician curve” up. This makes delegation less likely. The intuition is immediate: delegation is a way for the politician to increase his utility (and that of a representative citizen) because it depends on the common good. However, delegation comes at the expense of forsaking the net private rent associated with managing the public good (this private benefit can range from an ego rent to outright corruption). To the extent that this private rent is large, the concern of the politician for the public good will be smaller so that he is less likely to delegate. This is shown in Figure 5.

So, to summarize:

Proposition 1.

(Endogenous delegation of political power).

In the equilibrium of the model the set of variables $(e_{1}^{T}, e_{1}^{P}, e^{*}, U^{D}, U^{P})$ depends on the set of variables (θ_G, R, R^d,α, β). Delegation occurs if U^D > U^P. Both delegation and non-delegation equilibrium exist and which arises depends on the value of the parameters (θ_G, R, R^d,α, β). For the policy task, the followingholds at the margin:

i) An increase in the complexity of the task (measured by α⁻¹or β) increases the incentive to delegate;

ii) An increase in the rent from office (R) reduces the incentive to delegate;

iii) An increase in the uncertainty of the ability of the officeholder (θ_G– θ_B) reduces the incentive to delegate.

The proof of existence of both equilibria follows from Example 1 and the rest from the discussion in the sections above.

C. Social Welfare and Delegation

Proposition 1 establishes a theory of delegation. It shows that the decision to delegate depends on the properties of the policy task and how it affects political rents. The question now is whether or not delegation is optimal for society as a whole (if it occurs) rather than from the perspective of the politician. To be more precise, the question is: under what circumstances $U_{c}^{D} > U_{c}^{P}$ where D denotes delegation, P that the politician retains power and c indicates that utility refers to that of the representative citizen? Observe that the politician’s utility is the sum of his private rent from office and the citizens utility $U_{c}^{D}$ or $U_{c}^{P}$ . Assuming that the policymakers delegate, it is then straightforward to show that this is socially optimal if

\begin{matrix} 2 R^{D} < 2 R - c (e_{1}^{P}) - c (e_{2}^{P}) & (22) \end{matrix}

This condition implies that delegation will always be socially optimal as long as the “rent” the politician obtains by delegating power is smaller than the net rent he extracts when he performs the public task himself. This is a sensible condition to impose. The case when delegation occurs because the politician can extract higher private rents by delegating is theoretically uninteresting because in that case one could always engineer delegation by making R^D arbitrarily high. It is also implausible that a politician would receive higher private rents from the public office by giving it to someone else. Indeed imposing condition (22) makes this theory more interesting—and plausible—because it implies delegation can be optimal for the politician even if it reduces his net private rents.

So, to summarize:

Proposition 2.

Assume condition (22) holds. When delegation occurs, it is socially optimal.

Even if delegation is always socially optimal under condition (22), it may well be the case that the politician does not want to delegate. In other words, the following can occur: $U_{c}^{D} > U_{c}^{P}$ but U^D < U^P. This can arise because the politician may receive high rents from the policy task that more than compensate the (social) gains from delegating. In this case, it would be optimal for the society to “force” the politician to delegate by constitutional rules.

IV. Blinder’s Hypothesis: Insulation

One now turns to another key rationale for delegating policy to a technocrat: the fact that a technocrat is insulated from the whims and pressure of the citizenry while politicians may become “slaves of public opinion.” It is indeed well known that politicians have a hard time contradicting or confronting the electorate prior to elections. To be more precise, they have an incentive to ignore the superior private information they have about the policy choice and instead base policy on public opinion. This is always sub-optimal in this model because citizens all share the same preferences and the politician is not maximizing these preferences by following public opinion but instead sacrificing social welfare to be re-elected.

The setup is the same as in Section 3.1 except for the following changes. A public forecast (ρ_t, t =1, 2) of the shock is now introduced (it is observable at the beginning of every period).⁹ As discussed in section II., this forecast, if formed in accordance to the same forecasting technology as the officeholder’s forecast, except that, since only the politician is compensated for supplying effort in producing the forecast, citizens’ forecasts are based only on average expected ability (i.e., citizens’ effort is zero¹⁰). Recall that the officeholder’s forecast is his private information so that he can either reveal it (as in the previous section) by basing his decision on it, or hide it by basing his decision on the public signal. Note that if the politician follows public opinion, his actions will reveal nothing about his type. As a consequence it will be shown that, in equilibrium, he will always be re-elected since he has an incumbency advantage. This assumption captures the idea that a politician can always claim that a shock turned out to have been exceptional and that he cannot be blamed for it since the whole population agreed about the forecast.

When the politician can follow public opinion, three equilibria are possible: (1) the politician delegates power to a technocrat (who is, by design, insulated from public opinion); (2) the politician does not delegate but uses his (superior, in expectations) private information to make policy choices; and (3) the politician does not delegate but mimics public opinion (public opinion equilibrium). The first and the second equilibria are exactly the same as those analyzed in the last section, since in these cases the public signal is ignored. One can therefore turn directly to the public opinion equilibrium.

A. Public Opinion Equilibrium

In case the politician decides to follow public opinion in period 1 and base his decision on the public signal, his maximization problem becomes:¹¹

\begin{matrix} \max_{e_{1}} U_{1}^{PO} = {2 R - \frac{1}{2} α e_{1}^{2} - \frac{1}{2} α e_{2}^{2}} - {\frac{1}{2} - \frac{1}{2} θ_{G}} - {\frac{1}{2} - \frac{1}{2} θ_{G} - β e_{1} - e_{2}} & (23) \end{matrix}

Throughout, the “PO” superscript refers to the politician following “public opinion.” The first bracket captures, again, the net rents from office of the politician. The second bracket is the probability of taking the wrong policy decision in period one. When the politician follows public opinion, this probability does not depend on his effort in period 1 since he will not base his decision on his private signal in that period. The last bracket is the probability of making the wrong policy decision in period two. Note that the probability of making a mistake in period 2 is also a function of effort in period 1 because of the learning-by-doing effect. The above problem leads to the following equilibrium effort level

\begin{matrix} e_{1}^{PO} = \frac{β}{α} > 0 & (24) \end{matrix}

which, depending on whether β > 1 or not, is higher (or not) than e* = α⁻¹, the period 2 effort level. Due to the learning-by-doing effect and a positive effort in period 1 it is easy to verify that the politician will always be re-elected. The reason is that, although his first-period action does not reveal any information to the public (because it is conditioned on the public signal), since the politician puts in (strictly) positive effort in period 1, he is expected to do better than anyone that runs for office against him (an incumbency advantage). To summarize:

Proposition 3.

If the politician follows public opinion he is always re-elected.

Using this proposition one can then express the utility of the politician if he follows public opinion. It is obtained after substituting the equilibrium effort values (10) and (24) into (23), i.e.,

\begin{matrix} U^{PO} = 2 R - 1 + θ_{G} + \frac{1 + β^{2}}{2 α} & (25) \end{matrix}

When the politician follows the whims of public opinion, utility takes exactly the same form as welfare under delegation, i.e., the line has a slope 1 in (U, θ_G) space. To investigate whether a politician will find it individually rational to follow public opinion and ignore his private information, one can perform the same analysis as when analyzing whether or not the politician delegates (Section 3.2). By drawing up a diagram in (U, θ_G) space, one obtains the same picture as shown in Figure 3 with the line for the technocrat being replaced by the non-insulated politician. Thus, the politician may or may not choose to follow public opinion, depending on parameter values. It is easy to verify the proposition below following the steps shown in the previous section. In Proposition 4 one denote the equilibrium when the politician bases his decisions on his private information rather than public information by “P”, and it is the same equilibrium as analyzed in the last section where “P” stood for “politician”:

Proposition 4.

In the equilibrium of the model, the set of variables $(e_{1}^{PO}, e_{1}^{P}, e^{*}, U^{PO}, U^{P})$ depends on the set of variables (θ_G, R, α, β). The politician always follows public opinion if U^PO > U^P. Both equilibria exist, and which arises depends on the value of the parameters (θ_G, R, α, β). For the policy task, the following hold at the margin:

i) An increase in the complexity of the task (β) reduces the incentive to follow public opinion;

ii) An increase in the rent from office (R) increases the incentive to follow public opinion;

iii) An increase in the uncertainty of the ability of the officeholder (θ_G –θ_B) reduces the incentive to follow public opinion.

It is now possible to investigate whether a politician that always follows public opinion would ever choose to delegate policy to a technocrat so as to insulate the policy decisions from the whims of public perceptions.

B. Welfare and The Existence of Trade-offs Between Regimes

Consider a range of parameters (θ_G, R, α, β) so that the politician prefers to follow public opinion (i.e., U^D > U^PO). Note that this inequality does not depend on the value of θ_G. This is because the politician will never be fired. The key trade-off in this case comes through variations in the parameter α. The reason for this is that, for low values of a, the technocrat invests more effort into policy making and thus has superior information about the policy task compared to the public. The politician that follows public opinion, however, invests little effort since he does not take advantage of his private information. This means that the cost of following public opinion is higher, the lower is α. Figure 6 highlights the basic trade-offs facing a politician that follows public opinion. His welfare increases more slowly in α⁻¹—which is a measure of his private information—if he retains the policy task than if he delegates it to the technocrat. This can be seen by taking a partial derivative of U^D with respect to α⁻¹, i.e.,

\frac{\partial U^{D}}{\partial α^{- 1}} = 1 + β^{2} + 1 + 2 β > \frac{\partial U^{PO}}{\partial α^{- 1}} = \frac{1 + β^{2}}{2}

Furthermore, the higher the value of β the higher the incentive of a politician that follows public opinion to delegate to an insulated technocrat. To see this, note that

\frac{\partial U^{D}}{\partial β} = e_{1}^{T} = \frac{1 + β}{α} > \frac{\partial U^{PO}}{\partial β} = e_{1}^{PO} = \frac{β}{α}

The intuition for why a higher β increases the incentive for delegation is straight forward. Since the insulated technocrat supplies a higher effort in period 1 than the politician (because he uses this effort to retrieve a private signal for policy making), β, the learning-by-doing parameter, has a stronger impact on utility under delegation.

Interestingly, one can interpret both these parameters (i.e., α and β) as the degree of complexity of a task. Presumably the more complex the task is, the more important the learning effect is. One may also argue that the more important private information is for a given task—namely information that is only available to the technocrat—the more complex and “specialized” that task is. Thus, these two effects reinforce each other: the more complex the task, the more important it is to delegate it.

Proposition 5.

The more complex the policy task (as measured by an increase in β and α⁻¹), the more likely a politician that is a slave of public opinion is to delegate it so that the policymaker is insulated from the whims of public opinion.

One implication of our analysis which should be clear from the above discussion, is that if an equilibrium takes place in which the politician ignores his own information in favor of public opinion, it is always socially optimal to delegate to an insulated technocrat. To see this, note that when the politician follows the public opinion he does not reveal anything about his type; this indicates that there is no advantage of elections. In contrast, there is always a benefit to delegation, since under delegation the technocrat is insulated from public opinion and, therefore, has an incentive to use his (superior) private information rather than the public signal. It is not guaranteed, however, that a politician who would follow public opinion would want to delegate even if it were socially optimal to do so. This depends on the private rents he extracts from office and the other parameters of the model. To summarize:

Proposition 6.

If the politician ignores his private information and follows public opinion in equilibrium, it is always socially optimal to delegate power to an insulated technocrat. However, it may not be in the interest of a politician to delegate.

Another implication of the analysis is that the public signal makes it less likely that a politician chooses to delegate. This should be obvious because the public signal can only make a politician that chooses to retain power better off (since he is always free to ignore the signal). For a public signal to increase the incentive to delegate, one would need to introduce some altruistic motives for politicians (e.g., a legacy motive) or a belief that delegation today can commit future politicians to leave the policy task in the hands of technocrats. These extensions seem relatively straight-forward and are left to future research.

V. Example: Central Bank Independence

In this section, the simple political economy model of Section 2.1 is embedded into a small but standard macro model to give an example of how the more general theory can be applied in practice. This is achieved through the use of the standard “Phelps problem”¹² whereby the government minimizes deviation of inflation and the output gap from zero subject to an expectation-augmented Phillips curve, as in Kydland and Prescott (1977) and Barro and Gordon (1983). Unlike these authors, in the model below, it is not assumed that there is any inflation bias. The inflation bias is one of the traditional motivation for central bank independence. The Nobel prize committee, for example, cited central bank independence as one of the main practical implications of Kydland and Prescott’s (1977) derivation of the inflation bias in its prize announcement in 2004. Here The aim here is to illustrate how the model’s mechanism can be considered as a complementary rationale to the inflation bias for central bank independence. The theory of delegation developed, however, can be used for many different applications than monetary policy.

The period utility function of a (representative) citizen is given by Kydland-Prescott preferences:

\begin{matrix} υ_{t}^{c} = - \frac{1}{2} π_{t}^{2} - \frac{1}{2} λ x_{t}^{2} & (26) \end{matrix}

where π_t is inflation and x_t is the output gap, i.e., the deviation of output from the natural rate of output. The supply side is characterized by an expectation-augmented Phillips curve, as in Kydland and Prescott (1977),

\begin{matrix} π_{t} = k x_{t} + π_{t}^{e}, t = 1, 2 & (27) \end{matrix}

where $π_{t}^{e}$ denotes expectations of inflation that are formed before monetary policy is set in each period. This extends the Kydland-Prescott example by adding a demand side that relates the output gap to monetary policy

\begin{matrix} x_{t} = - ε (i_{t} - {\bar{r}}_{t}), t = 1, 2 & (28) \end{matrix}

where i_t is the nominal interest rate and ${\bar{r}}_{t}$ is the natural rate of interest (Kydland and Prescott, in contrast, assume that the central bank controls inflation directly). Like the loss function and the Phillips curve, a demand equation of this kind can be derived from microfoundations, as illustrated for example in Eggertsson (2003). The natural rate of interest is only a function of exogenous shocks. It is the real interest rate that is consistent with output being at the natural rate of output at all times. It is assumed that the central bank chooses i_t in each period before the shock is realized so that the policy problem is exactly the same as discussed in Section II. (but here the loss function is endogenous since it is a function π_t of and x_t). As before, it is assumed that ${\bar{r}}_{t}$ is equal to r^H and r^L with equal probability.

If the central bank could perfectly forecast the natural rate of interest, it would set $i_{t} = {\bar{r}}_{t}$ , resulting in zero inflation and zero output gap. This minimizes the bank’s loss function. If the central bank cannot perfectly forecast the natural rate of output, this equilibrium may not be feasible. For example, if the central bank misses the target for the natural rate of interest so that $i_{t} < {\bar{r}}_{t},$ there will be excessive inflation and output will be above its natural level. By contrast, if $i_{t} > {\bar{r}}_{t},$ there will be deflation and output slump. Since the central bank sets the nominal interest rate before observing ${\bar{r}}_{t},$ its problem is to predict the future value of ${\bar{r}}_{t},$ in order to minimize its losses.

In the last (second) period, the maximization problem of the officeholder is to

\begin{matrix} \max_{{e, i, π}} E [υ^{o}] & (29) \end{matrix}

Optimal policy under discretion is now analyzed. The government treats expectations as constants and this reduces the problem to a one-period model.

From (27) and (28), one gets

\begin{matrix} x = - ε (i - \bar{r}) & (30) \end{matrix}

\begin{matrix} π = - ε k (i - \bar{r}) + π^{e} & (31) \end{matrix}

\begin{matrix} x = - ε (i - \bar{r}) & (32) \end{matrix}

where the government treats π_e as a constant. After substitution of (30) and (32) into (29), the maximization problem of the officeholder becomes

\begin{matrix} \max_{i} \frac{1}{2} {(- ε k (i^{σ} - {\bar{r}}^{s}) + π^{e})}^{2} + \frac{1}{2} λ {(- ε (i^{σ} - {\bar{r}}^{s}))}^{2} & (33) \end{matrix}

and the subscript sσ ∈ {HR, LR, HW, LW}refers to the combination of the shock s ∈ {H, L} and the forecast σ ∈ {H, L} i.e., whether σ turns out to be a correct (R) forecast about S or not (W).

After rearranging the first-order conditions (w.r.t. e, i^H, i^L, π_HR, π_HW, π_LR, π_LW), the equilibrium values of, i^H, i^L, π_HR, π_HW, π_LR, π_LW are:

\begin{matrix} i^{H} = (\frac{1}{2} + E_{2} [θ] e) {\bar{r}}^{H} + (\frac{1}{2} - E_{2} [θ] - e) {\bar{r}}^{L} & (34) \end{matrix}

\begin{matrix} i^{L} = (\frac{1}{2} - E_{2} [θ] e) {\bar{r}}^{H} + (\frac{1}{2} + E_{2} [θ] + e) {\bar{r}}^{L} & (35) \end{matrix}

\begin{matrix} π_{HR} = ε k (\frac{1}{2} - E_{2} [θ] - e) ({\bar{r}}^{H} - {\bar{r}}^{L}) & (36) \end{matrix}

\begin{matrix} π_{HW} = - ε k (\frac{1}{2} + E_{2} [θ] + e) ({\bar{r}}^{H} - {\bar{r}}^{L}) & (37) \end{matrix}

\begin{matrix} π_{LR} = - ε k (\frac{1}{2} + E_{2} [θ] - e) ({\bar{r}}^{H} - {\bar{r}}^{L}) & (38) \end{matrix}

\begin{matrix} π_{LW} = - ε k (\frac{1}{2} + E_{2} [θ] + e) ({\bar{r}}^{H} - {\bar{r}}^{L}) & (39) \end{matrix}

\begin{matrix} x_{s} = k^{- 1} π_{s}; & s \in {HR, LR, HW, LW} & (40) \end{matrix}

where E₂[θ]=qθ_G+(1–q)θ_B+e₁ Note here that the notation E₂θ is defined as the expected ability that includes the effort in the previous period.¹³ It can be noticed that, with an endogenous loss function, all the equilibrium values are now functions of the expected ability of the officeholder, i.e., $e = e^{*} (q) > 0, i^{H} = i^{H *} (q), i^{L} = i^{L *} (q), π_{s} = π_{s}^{*} (q)$ and $x_{s} = x_{s}^{*} (q)$ . To find the value for e, one again obtains condition (10) and. one can then substitute the endogenous values of π_s and x_s into the loss functions and obtain the solution for e₂, i.e.,

\begin{matrix} e_{2}^{s_{1}} = \frac{ϕ}{α - ϕ} E_{2}^{s_{1}} θ, & s_{1} \in {HR, LR, HW, LW} \end{matrix}

where it is assumed that the cost function is $c (e) = \frac{1}{2} α e^{2}$ and defined the following coefficient $ϕ = 4 (1 + \frac{λ}{k^{2}}) ε^{2} k^{2} {(r^{H} - r^{L})}^{2}$ . Note that one needs α > ϕ for an equilibrium with positive effort to exist. The effort choice depends on the realization of s₁ in period 1 since this has an effect on the expected ability of the officeholder. Note that this implies that in the macroeconomic setup the effort choice can be higher under delegation in both period 1 and 2, whereas in the previous section the effort choice was the same in period 2 across the two regimes.

The first-period allocation, as far as the macroeconomic variables is concerned (i.e., i^H, i^L, π_s, x_s) will be the same as those derived in (34)-(40) except that E₂[θ] is replaced with E₁[θ] and in this case e₁ does not appear in the expression for E₁[θ]. The derivation of the first-period equilibrium effort level follows directly from the analysis of Section A. with the added analytical complication that the loss function is endogenous. Note that the derivations (34)-(40) imply that, for given ability, a higher level of effort reduces output and inflation variability. Since delegation implies higher effort, this implies that if a politician delegates power then central bank independence leads to a reduction in both output and inflation variability.

This example makes some progress along a dimension that some have considered as a limitation of the literature on central bank independence. Much of this literature implies (see e.g., Rogoff, 1985) that society delegates policy to achieve lower inflation variability, but at the cost of inducing higher variability in output (for a notable exception see Alesina and Gatti, 1995). Some argue that this prediction is not consistent with the data and that output and inflation variability usually go hand in hand (e.g., Alesina and Summers, 1993). In the above example, delegation is not aimed at lowering inflation volatility at the expense of higher output volatility. Instead, it occurs to insulate monetary policy making from public opinion and facilitate learning. As a consequence the theory developed is consistent with a decline in output volatility going hand in hand with lower variability in inflation.

The model implies that average inflation is zero across regimes. This can be seen from equations (34)-(40). The model can be extended to imply that central bank independence also implies lower average inflation. Assume that society’s loss function given by (26) is modified to:

\begin{matrix} υ_{t}^{c} = - μ_{t} [\frac{1}{2} π_{t}^{2} + \frac{1}{2} λ x_{t}^{2}] & (41) \end{matrix}

This is a generalization of the previous loss function since it now allows the loss function to depend on the state of the economy so that μ_t (s) can take four values,i.e s∈{HR, LR, HW, LW}. This loss function allows for asymmetry across states of the economy. In particular, one can consider the possibility that society attaches higher losses to a recession (i.e., a negative output gap and deflation) than to an expansion (i.e., a positive output gap and inflation), i.e. μ^HR < μ^LR and μ^HW < μ^LW.

For a given level of effort, one can now solve for the optimal level of inflation, output and interest rate following the same steps as in the previous section. The only difference is that now π^e ≠ 0. The resulting expressions for π_s and x_s are somewhat more complicated. What they illustrate is that a higher level of effort does not only reduce inflation and output variability but also the average level of inflation. When the central bank weighs recessions and booms differently, average inflation is different from zero. Since the central bank puts a lower weight on an economic boom than on a recession, it sets the nominal interest rate as if it is giving a higher probability on the recessionary state relative to the solution derived in the previous section. This causes an average inflation bias that is unrelated to the standard inflation bias.¹⁴ What is immediately clear, however, is that a higher level of effort decreases average inflation since a higher level of effort improves the accuracy of the central bank’s forecast. Since central bank independence, when it occurs, is associated with higher effort, central bank independence leads not only to lower variability in output and inflation, but it also reduces the average level of inflation.

The standard argument for an independent central banker is that such a delegation can reduce or even eliminate an inflation bias (e.g., Rogoff, 1985). The theory of central bank independence that emerges from our simple example, however, does not rely on dynamic inconsistency, as is the case with existing theories of independent central banks. To the extent that preferences are heterogeneous, therefore, it will generally be better that the preferences of the policymaker reflect that of the underlying population. If one defines goal independence as the freedom for the central banker to choose its own goals, one can show from the model that while instrument independence is socially optimal, goal independence is not.

While the theory of central bank independence that is derived from this simple example does not rely on a dynamic inconsistency of monetary policy, as standard theories of central bank independence do, the theory developed in this paper does not contradict the conventional story. In this respect, the rationale for central bank independence is a complement, rather than a substitute, for existing theories of central bank independence.

VI. Conclusion

This paper studied the costs and benefits of delegating public policy to an independent agency outside the political spectrum (i.e., free from election pressures). To get at the heart of the issue—the role of election pressures—the most simple framework possible was considered: a two period model with the possibility of elections at the end of period 1. While this framework is extremely simple, it is hoped that it gets at some basic issues that are likely to remain important in a more general setting. In particular, the dynamic incentive structure that arises because of elections enables to formalize both Hamilton’s (1788) and Blinder’s (1998) rationales for delegation. Delegation trades-off the cost of having a possibly incompetent technocrat with a long-term job contract against the benefit of having a technocrat who (i) invests more effort into the specialized policy task and (ii) is better insulated from the whims of public opinion. The relevance of the model in the context of monetary policy was then emphasized and new insights and conditions regarding central bank independence were obtained. To conclude, some limitations and suggestions for further extending the framework are underlined.

Perhaps the most important issue that is abstracted from is the role of dynamic inconsistency. This is done—not because of a belief that dynamic inconsistency is unimportant—but because that issue has been extensively studied before (the role of dynamic inconsistency in delegation of public tasks is analyzed by Alesina and Tabellini, 2006) and goal of this paper is to focus instead on the role of election pressures on the incentives of the policymaker. Another important abstraction is the assumption that all agents have the same preferences. By abstracting from heterogenous preferences, the possibility that the policymaker may have a hidden agenda is excluded. As shown by Maskin and Tirole (2004) hidden agendas can give the policymaker an incentive to pander to public opinion, a consideration that is not captured in the model developed in this paper but would tend to tilt the results even further towards delegation.

Perhaps the most important limitation of the model is that it has only two “periods.” An important extension would include several periods—this should be relatively straightforward though. The conjecture is that this paper’s results would generalize to this setting, i.e., Blinder’s and Hamilton’s hypothesis—that are formalized in this paper—would still be of great importance to understand the basic trade-offs for delegation. This extension, however, is also important for other reasons than to illustrate the robustness of the highlighted trade-offs. An extension of this kind could answer several questions in political economy that are quite unrelated to the issue of optimal delegation of political power, because it would allow us to analyze the optimal degree of election pressures over time for a given policymaker. In contrast, one has assumed that this is an either/or question because in this paper’s model the policymaker either faces elections at the end of period 1 or not. There is no middle ground. While this seems appropriate to analyze the choice between a politician and technocrat, it rules several interesting possibilities that lie between those two extremes, i.e. how frequently should elections occur for the elected politician? What is the optimal term length? An extension to multiple periods—that allows for intermediate cases with varying degrees of election pressures—could answer questions such as: (1) Why did almost all U.S. states in the period 1910—1970 lengthen the term of state governors from 2 years to 4 years? An analysis along the lines of our model would suggest that a 2-year term made the governor more susceptible to short-term pressures and also did not give the officeholder enough incentive to invest effort into learning on the job. These channels would be two reasons for the increase in state governors’ term length. The fact that terms were increased at different times in different states could give some empirical content to the various determinants of delegation that are analyzed in the model. (2) What tasks should be allocated to the Supreme Court in the United States (where members are appointed for life), which tasks should be allocated to the Senate (where each senator has a term-length of 6 years), which ones to the President (with a 4-year term), and which policy tasks should be allocated to the House of Representatives (each congressman has a term of 2 years). The basic framework developed would suggest that the role of learning on the job, resistance to short-term political pressure and the danger of ending up with incompetent policymakers would all be important to understanding the optimal task allocation. It is hoped that the present framework will be helpful in addressing these important questions in future work.

Appendix

Proof that α – 2β > 0

First note that, since the probability has to be between 0 and 1, one knows that

\frac{1}{2} + \frac{1}{2} θ_{G} + e_{1} + β e_{2} \leq 1

and that, using the equilibrium values for e₁ and e₂ under Delegation, one has

\frac{1}{2} + \frac{1 + β}{α} + \frac{β}{α} \leq 1

So that

\frac{1 + 2 β}{α} \leq \frac{1}{2}

and

2 (1 + 2 β \leq α)

Which gives

\begin{matrix} α - 2 β \geq 2 (1 + β) > 0 & (42) \end{matrix}

□

Proof that the coefficient γ₃ of $U^{D} = γ_{1} + γ_{2} θ_{G} + γ_{3} θ_{G}^{2}$ is strictly positive

The coefficient γ₃ is strictly positive if

\frac{β^{2}}{α - 2 β} + \frac{β^{3}}{{(α - 2 β)}^{2}} > \frac{1}{2} α \frac{β^{2}}{{(α - 2 β)}^{2}}

i.e., if

(α - 2 β) β^{2} + β^{3} > \frac{1}{2} α β^{2}

α β^{2} - β^{3} > \frac{1}{2} α β^{2}

α - β > \frac{1}{2} α

\frac{1}{2} α - β > 0

α - 2 β > 0

which is known to be strictly positive by (42).□

References

Alesina, A., and R. Gatti, 1995, “Independent Central Banks: Low Inflation at No Cost?” American Economic Review, Vol. 82, pp. 196–200.
- Search Google Scholar
- Export Citation
Alesina, A., and L. Summers, 1993, “Central Bank Independence and Macroeconomic Performance” Journal of Money, Credit, and Banking, Vol. 25, pp. 151–62.
- Search Google Scholar
- Export Citation
Alesina, A., and G. Tabellini, 2006, “Bureaucrats or Politicians?,” mimeo Harvard and Bocconi Universities.
- Search Google Scholar
- Export Citation
Arrow, K., 1962, “The Economic Implications of Learning by Doing,” Review of Economic Studies, Vol. 29, pp. 155–73.
- Search Google Scholar
- Export Citation
Barro, R.J., and D.B. Gordon, 1983, “A Positive Theory of Monetary Policy in a Natural Rate Model,” Journal of Political Economy, Vol. 91, pp. 589–610.
- Search Google Scholar
- Export Citation
Besley, T., and S. Coate, 2003, “Elected Versus Appointed Regulators: Theory and Evidence,” Journal of the European Economics Association, Vol. 1, pp. 1176–1206.
- Search Google Scholar
- Export Citation
Drazen, Allan, 2001, “‘Laying Low’ During Elections: Political Pressure and Monetary Accommodation,” mimeo University of Maryland.
- Search Google Scholar
- Export Citation
Drazen, Allan, 2000, Political Economy in Macroeconomics, Princeton University Press, Princeton, NJ.
- Search Google Scholar
- Export Citation
Elster, J., 1995, “The Impact of Constitutions on Economic Performance,” Proceedings of the Annual World Bank Conference on Economic Development, Washington: The World Bank, pp. 209–226.
- Search Google Scholar
- Export Citation
Eggertsson, G., 2003, “How to Fight Deflation in a Liquidity Trap: Committing to Being Irresponsible,” IMF Working Paper, No. 03/64, Washington DC.
- Search Google Scholar
- Export Citation
Eggertsson, G., and E. Le Borgne, 2003, “A Political Agency Theory of Central Bank Independence” IMF Working Paper, No. 03/144, Washington DC.
- Search Google Scholar
- Export Citation
Green, J., and N. Stokey, 1983, “Tournaments and Contracts,” Journal of Political Economy, Vol. 91, pp. 349–64.
- Search Google Scholar
- Export Citation
Holmström, B., 1999, “Managerial Incentive Problems: A Dynamic Perspective,” Review of Economic Studies, Vol. 66, pp. 169–82. Originally, 1982, in Essays in Economics and Management in Honor of Lars Wahlbeck (Stockholm: Swedish School of Economics).
- Search Google Scholar
- Export Citation
Kydland, F.E., and E.C. Prescott, 1977, “Rules Rather than Discretion: The Inconsistency of Optimal Plans,” Journal of Political Economy,.Vol. 85, pp. 473–91.
- Search Google Scholar
- Export Citation
Lazear, E.P., and S. Rosen, 1981, “Rank-Order Tournaments as Optimum Labor Contracts,” Journal of Political Economy, Vol. 89, pp. 841–64.
- Search Google Scholar
- Export Citation
Le Borgne, E., and B. Lockwood, 2003, “Do Elections Always Motivate Incumbents? Experimentation vs. Career Concerns,” IMF Working Paper No. 57/03, Washington DC, forthcoming in Public Choice.
- Search Google Scholar
- Export Citation
Maskin, E., and J. Tirole, 2004, “The Politician and the Judge: Accountability in Government,” American Economic Review, Vol. 94, pp. 1034–54.
- Search Google Scholar
- Export Citation
Persson, T., and G. Tabellini, 2000, Political Economics: Explaining Economic Policy, MIT Press, Cambridge, MA.
- Search Google Scholar
- Export Citation
Prat, A., 2005, “The Wrong Kind of Transparency,” American Economic Review, Vol. 95, pp. 862–77.
- Search Google Scholar
- Export Citation
Rogoff, K., 1985, “The Optimal Degree of Commitment to an Intermediate Monetary Target,” Quarterly Journal of Economics, Vol. 100, pp. 1169–89.
- Search Google Scholar
- Export Citation
Rogoff, K., and A. Sibert, 1988, “Elections and Macroeconomic Policy Cycles,” Review of Economic Studies, Vol. 55, pp. 1–16.
- Search Google Scholar
- Export Citation
Wilson, J.Q., 1989, Bureaucracy: What Government Agencies Do and Why They Do It, Basic Books, New York.
- Search Google Scholar
- Export Citation

The authors thank Allan Drazen for many useful comments and several colleagues and seminar participants. The views expressed here are those of the authors, and do not necessarily reflect the position of the Federal Reserve Bank of New York, the Federal Reserve System, or the International Monetary Fund.

This paper does not address the issue of dictatorship versus democracy. This paper’s assumption of homogenous agent preferences, which is made so as to illustrate in the simplest way the rationale for delegation, is not appropriate for analyzing dictatorship issues.

Although if a governor does not complete his term, his successor does not get a new 14 year term, but first has to serve the existing 14 year term. Hence many governors have, in practice, a shorter term once appointed.

See also Le Borgne and Lockwood (2003) who study questions that are more similar to the one developed in this paper although they do not model the learning or insulation effects. Instead, they assume that effort and ability are multiplicative which also gives a rationale for delegation (through an “experimentation” channel).

Besley and Coate (2001) contrast direct election with political appointment of regulators in a model where electing regulators produces more pro-consumer regulators.

In an early version of this paper (Eggertsson and Le Borgne, 2003) a richer (but more complex) forecasting technology was presented (whereby ability and effort entered multiplicatively). This gave rise to an experimentation motive.

For example, Franklin Delano Roosevelt, one of the most powerful Presidents in U.S. history, tried to change the Supreme Court in the Spring of 1937 but failed.

The individual rationality to stand for election also needs to be checked for both the incumbent and the opponent, given voters’ cutoff rule. The individual rationality condition requires that this gain be positive. For this to be the case, the weak assumption that $R \geq \frac{α}{2} e *$ is imposed, i.e., that the net rent from office is nonnegative.

One can confirm by (15) that $\frac{\partial q_{R}}{\partial θ_{G}} > 0$ and one knows from Bayes rule that $q_{R} > \frac{1}{2} .$

An “insulation effect” can be obtained in our model through various channels. Aside from the one highlighted in this paper, another one could have been to assumed that public opinion changes stochastically and favors one policy or another depending on the draw, while the optimal policy function is unchanged.

It is assumed that there are many citizens so that each citizen’s opinion will not weigh much in public opinion. In this case, the citizen will have no incentive to invest any effort.

Parameters are assumed such that the politician would never want to resign voluntarily and would never want to follow public opinion in period 2. It can be verified that one can always assume parameters for the rent and β so that this is the case.

This was first coined the Phelps problem by Thomas Sargent.

However, as in ^{Section 3}, under Democracy, if the incumbent politician is not reelected, the e₁ term drops out

Using the word “bias” here may be somewhat misleading, since higher average inflation is optimal for this social loss function (as opposed to the standard inflation bias that is suboptimal and is due to inefficient lack of credibility–the time inconsistency problem in monetary policy).

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/001/2007/091/article-A001-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully

Collapse
Expand

Dynamic Incentives and the Optimal Delegation of Political Power

Author:

Mr. Eric Le Borgne

and

Mr. Gauti B. Eggertsson

Volume/Issue:: Volume 2007: Issue 091

Publisher:: International Monetary Fund

ISBN:: 9781451866551

ISSN:: 1018-5941

Pages:: 35

DOI:: https://doi.org/10.5089/9781451866551.001

View raw image

Figure 1.

Timing

View raw image

Figure 2:

Possible cutoff for welfare

View raw image

Figure 3:

Basic tradeoff in delegation

View raw image

Figure 4:

An increase in complexity of the task increases delegation

View raw image

Figure 5:

A decrease in corruption increases delegation

View raw image

Figure 6:

Basic insulation/mimicking tradeoff

Topics

Countries

Series

About

Resources

Contributor Notes

Abstract

I. Introduction

II. The Framework

A. Political Agency Setup

1. Private signal

2. Public signal

3. Institutions

Delegation

Political control

III. Hamilton’s Hypothesis: Learning

A. Equilibrium

1. Period 2 effort

2. Period 1 effort under delegation

3. Period 1 effort without delegation

B. Delegation of Political Power

1. Welfare and the Existence of Trade-offs Between Regimes

2. The Welfare Trade-off

3. Comparative Statics

C. Social Welfare and Delegation

IV. Blinder’s Hypothesis: Insulation

A. Public Opinion Equilibrium

B. Welfare and The Existence of Trade-offs Between Regimes

V. Example: Central Bank Independence

VI. Conclusion

Appendix

References

Same Series

Other IMF Content

Other Publishers

Asian Development Bank

Inter-American Development Bank

International Labour Organization

The World Bank

Share Link

Table of Contents

Headings

Figures

Metrics

Foresee