Leviathan at bay: constitutional versus political controls on government.

Sutter, Daniel

A dependence on the people is, no doubt, the primary control on government; but experience has taught mankind the necessity of auxiliary precautions.

- James Madison, Federalist No. 51

I. INTRODUCTION

Citizens possess two main methods of controlling politicians, elections and constitutional constraints. The British system of parliamentary supremacy and America's written constitution with judicial review by the Supreme Court reflect the contrasting approaches. Political theorists continue to debate the relative effectiveness of constitutional and electoral controls on politicians. These controls seemingly are substitutes: if elections effectively control politicians, constitutional constraints become redundant and can only lower the effectiveness of government.

Public choice theory separately examines elections and constitutions. An extensive literature beginning with Barro [1973] analyzes electoral controls, but has not considered constitutional controls. Constitutional political economy examines various constraints on politicians (Buchanan [1987], Brennan and Buchanan [1980]), yet assumes elections insufficiently control politicians. Recent studies of term limits (i.e., Dick and Lott [1993]) represent one effort to analyze constitutional and electoral controls together.

I integrate constitutional limits on government into a model of electoral control of politicians. My model features two types of politicians, knaves who abuse discretionary powers at their disposal, and angels who never misuse power. Citizens delegate a variable amount of discretionary power to government, which provides benefits if properly employed but lowers welfare if abused. The interaction between constitutional limits and elections is complicated; constraints complement electoral controls based on politician payoffs (selection and incentive effects) while the screening and sorting effects substitute for constraints.

I proceed in this paper as follows. Section II describes the citizen's constitutional delegation of authority to government and incorporates electoral screening of candidates. Optimal delegation of power increases with the effectiveness of screening, but unless screening is perfect, constitutional limits help keep Leviathan at bay. Section III endogenizes the candidate pool and examines the selection effect constitutional controls. The composition of the candidate pool depends on delegation to government, which has implications for the propensity of politicians to pursue their personal self-interest. While constitutional political economy relies heavily on the threat from knavish politicians, many political observers contend that most politicians sincerely pursue the public interest as they perceive it. I show that politicians' observed honesty does not invalidate the argument for chains.

Section IV considers the ability of repeated elections to sort knaves out of office. The sorting effect increases optimal delegation of power to government, but constraints can still play a useful role if knaves try for office and electoral review is not continuous. Section V allows knaves to act opportunistically and considers the incentive effect of retrospective voting and post-political incentives. Constitutional constraints complement electoral incentives by limiting opportunistic knaves' gain from shirking. The final section concludes.

II. CONSTITUTIONAL CHOICE IN A MODEL OF ANGELS AND KNAVES

I adopt a principal-agent view of government. A representative citizen selects an agent to exercise the powers of government on her behalf. The representative citizen framework abstracts from differences amongst the population about the actions the agent should undertake. The control of politicians becomes difficult in the absence of a "common interest," both positively and normatively. Consequently, models of electoral control typically employ this assumption.(1)

The agency model assumes a conflict of interest between the principal and agent. The agent can take some action (shirking) which advances his interest at the principal's expense. Political misbehavior in a representative citizen model might involve seeking personal financial gain, allocating public projects on the basis of bribes or friendship instead of efficiency, or violating the rights of political opponents. I assume two types of politicians exist so political institutions can have screening, sorting, and selection effects. A politician's type is private information. Politicians differ in their willingness to misuse the powers of office (shirk): angels never misuse power, while knaves will use the powers of government for their own purposes.(2) Angels are internally constrained; guilt might result in a bribe lowering their utility. Knaves lack the angels' internal constraint, so misappropriated funds raise their utility. While none of us may be above considerations of self-interest, casual empiricism suggests people differ in their willingness to pursue their self-interest at the expense of others. My angels are not totally selfless; their decision to pursue office in section III depends on the expected payoff of public service. Knaves always abuse power, except in section V where they behave opportunistically.

The representative citizen's constitutional choice is the delegation of authority to a single governmental agent.(3) Government requires authority to provide benefits to citizens; but the misuse of power by a knave harms the citizen. Both the benefit of proper use and the cost of misuse of power increase with the authority government possesses. A well-intentioned politician in my model faces no difficulty in serving the representative citizen.(4)

Let x be the amount of authority delegated to government, normalized so that x [element of] [0, 1]. If x = 0 government lacks authority, while x = 1 is unlimited government. I do not consider potential problems enforcing a constraint on government; that is, the choice of x binds. The elements of a constitution captured in x include allowable tax bases and rates, the conditions for taking private property, and procedural rights for individuals in legal disputes with the government (due process). The [TABULAR DATA FOR TABLE I OMITTED] citizen's net benefit when authority is used properly to produce public goods is B(x), with B(0) = O, B[prime] [greater than] 0, and B[double prime] [less than] 0. Let T(x) be the citizen's cost if power is misused, with T(0) = 0, T[prime] [greater than] 0, and T[double prime] [greater than] 0. Further, assume B[prime] (0) [greater than] T[prime] (0) = 0. Table I summarizes notation.

I initially assume an exogenous candidate pool consisting of n angels and m knaves. Let [Pi] be the equilibrium probability an angel holds office. Elections initially select a candidate at random from the pool to take office, so [Pi] = n/(m + n). The selection of several candidates from the pool does not alter n/(n + m).

The representative citizen chooses x to maximize

[Pi] [multiplied by] B(x) - (1 - [Pi]) [multiplied by] T(x),

The first order condition for an interior solution is

(1) [Pi] [multiplied by] B[prime](x) - (1 - [Pi]) [multiplied by] T[prime](x),

Let [x.sub.0] be the optimal amount of discretion. Figure 1 illustrates the constitutional choice. Optimal delegation equates the marginal benefit of proper use of authority with the marginal cost of misuse. B[double prime] [less than] 0 and T[double prime] [greater than] 0 ensure satisfaction of the second order condition. The candidate pool must contain some angels to rule out a no authority corner solution; B[prime](0) [greater than] T[prime](0) = 0 and [Pi] [greater than] 0 together imply [x.sub.0] [greater than] 0. A no constraint corner solution ([x.sub.0] = 1) occurs if B[prime](1) [greater than or equal to] (1 - [Pi]) [multiplied by] T[prime](1)/[Pi]. A vanishing marginal benefit of delegation or unbounded marginal cost of abuse, B[prime](1) = 0 or T[prime](1) = [infinity], rules [x.sub.0] = 1 out. I wish to ensure a nontrivial constitutional choice problem. Discretion becomes worthless so limiting government is costless if B[prime](1) = 0. But [x.sub.0] = 1 when elections perform no meaningful control function means constitutional constraints can never be optimal. So I impose B[prime](1) [less than] (1 - [Pi]) [multiplied by] T[prime](1) to initially rule out unlimited government.

Electoral campaigns might reveal a candidate's knavishness. Suppose a candidate chosen at random out of the pool is screened and if identified as an angel, assumes office. Selection continues until a candidate passes screening. Assume the campaign produces a noisy information signal regarding a candidate's type. Let [s.sub.k] ([s.sub.a]) be the probability of incorrectly identifying a knave (angel). An angel takes office unless mistyped while a knave passes screening only if mistyped. The probability an angel holds office with screening is

(2) [[Pi].sub.s] = n [center dot] (1 - [s.sub.a])/[n [center dot] (1 - [s.sub.a]) + m [center dot] [s.sub.k]].

Figure 1 illustrates the effect of screening; the marginal benefit curve shifts up and the marginal cost curve shifts down. Let [x.sub.s] be optimal delegation with screening. Screening increases the probability an angel holds office as long as 1 - [s.sub.a] [greater than] [s.sub.k], and hence [x.sub.s] [greater than] [x.sub.0]. The screening function of elections substitutes for constitutional constraint. Since [[Pi].sub.s] [greater than] [Pi], for some values of B[prime](1) and T[prime](1) a corner solution may be optimal with screening. But even the most ardent democrats stress that elections do more than randomly select among candidates. Consequently an interior solution without screening must be robust to a wide range of parameter values. A high marginal cost of abuse for close to unlimited government, T[prime](1) [approximately equal to] [infinity], accomplishes this. But T[prime](1) [approximately equal to] implies that [[Pi].sub.s] must equal 1 for x [x.sub.s] = 1. And [[Pi].sub.s] = 1 requires [s.sub.k] = 0, so as long as knaves slip through electoral screening, some constitutional constraint on government remains optimal.

III. DELEGATION AND ELECTIONS WITH AN ENDOGENOUS CANDIDATE POOL

Electoral screening and constitutional delegation affect the incentives for politicians to pursue office, undermining the assumption of an exogenous candidate pool. I now endogenously determine the candidate pool, which requires specification of the payoffs from political and private sector employment for knaves and angels. Limiting delegation helps induce knaves to select out of politics.

Candidates forego private sector earnings when running for office. Let E be the opportunity cost of candidacy, which I set equal for each type politician. Angels and knaves receive payoffs in office proportional to the benefit of public goods production or the social cost of abuse, u [center dot]B(x) and v [center dot] T(x) respectively.(5) A mixed candidate pool requires payoff heterogeneity amongst at least one type of politicians. I assume that all angels have the same coefficient u while v is distributed over [0, V] with K(v) knaves having a coefficient greater than or equal to v.

Candidate selection is as in section II. Potential politicians decide to enter the pool. A politician not selected out of the pool or identified by screening as a knave receives a payoff of zero. Let q be the probability of selection for screening for a member of the pool (more than one candidate may be screened). The expected payoffs from candidacy for each type of politician are

Angels: q [center dot] (1 - [s.sub.a]) [center dot] u [center dot] B(x)

Knaves: q [center dot] [s.sub.k] [center dot] v [center dot] T(x).

The politicians with the highest expected payoffs join the pool. Of importance is the value of v which equates knaves' expected payoff of candidacy with angels', [v.sup.*], which is

(3) [v.sup.*] = (1 - [s.sub.a]) [center dot] u [center dot] (B(x)/[[s.sub.k] [center dot] T(x)]

Only knaves with v [greater than or equal to] [v.sup.*] might become candidates. Three cases are possible. If [v.sup.*] [greater than] V no knave becomes a candidate, so all politicians are angels and [Pi] = 1. If [v.sup.*] [less than or equal to] V then K([v.sup.*]) knaves would become candidates before any angels join the pool. K([v.sup.*]) may be sufficiently large that q [center dot] (1 - [s.sub.a]) [center dot] u [center dot] B(x) [less than] E, in which case the candidate pool is all knaves and [Pi] = 0. If K([v.sup.*]) is not too large, angels enter until q [center dot] (1 - [s.sub.a]) [center dot] u [center dot] B(x) = E and we have a mixed pool.

The probability an angel holds office is now a function of authority delegated to government, [Pi](x). Suppose we begin with a mixed pool as soon as x is large enough to attract candidates;(6) this parallels the exogenous mixed pool in section II. If an increase in x increases [v.sup.*], fewer knaves become candidates, while if [v.sup.*] falls fewer angels campaign for office. For fixed values of s, and [s.sub.k], [Pi] then changes with [v.sup.*]. The derivative of [v.sup.*] with respect to x is

(4) d[v.sup.*] / dx = {(1 - [s.sub.a]) [center dot] u [center dot] [T(x) [center dot] B[prime]x)

- B(x) [center dot] T[prime](x)]}/{[s.sub.k] [center dot] [[T(x)][Xi].sup.2]]}

The sign of d[v.sup.*]/dx is the same as the sign of the bracketed term in the numerator, which is ambiguous. The increase in benefits from proper use of power may initially make public office relatively more attractive to angels. But declining marginal benefit of public goods provision and increasing marginal cost of misuse imply that increased discretion eventually has an adverse effect on the composition of the pool. Figure 2 illustrates the likely relationship between delegation and [Pi].

An endogenous pool complicates the citizen's constitutional choice problem. The first order condition for optimal delegation becomes

(5) [Pi](x) [center dot] B[prime](x) = [1 - [Pi](x)] [center dot] T[prime](x) - [Pi][prime](x) [center dot] [B(x) + T(x)].

The final term in equation (5) reflects the impact of delegation on the composition of the pool, this is an added benefit if [Pi][prime] [greater than] 0 but an extra cost if [Pi][prime] [less than] 0. Optimal delegation with an endogenous pool, [x.sub.e], likely involves [Pi][prime]([x.sub.e]) [less than] 0. Although [Pi][prime]([x.sub.e]) [greater than] 0 cannot be ruled out at this level of generality, equation (5) implies this requires a high marginal cost of misusing power. But equation (4) implies that a large value of T[prime] results in an increasingly knavish candidate pool. Constitutional constraints can strengthen selection effects by limiting a knave's reason to pursue office.(7)

The potential for harm if the discretionary powers of the state fall into the "wrong" hands is a major argument for constitutional constraints. As Hume [1964, III, 117-8] wrote, "in contriving any system of government, and fixing the several checks and controls of the constitution, every man ought to be supposed a knave, and to have no other end, in all his actions, than private interest." Brennan and Buchanan [1985, 55] defend the Leviathan model of government on worst-case grounds. Constitutional political economy seems dependent on the empirical assumption of a knavish candidate pool. Many political scientists and close observers of politics, however, challenge the validity of this assumption. For instance, Mikva notes

The politicians and other people I have known in public life just do not fit the "rent-seeking" egoist model that the public choice theorists offer. ... Not even my five terms in the Illinois state legislature ... nor my five terms in the United States Congress, prepared me for the villains of the public choice literature. [1988, 167](8)

Selection effects helps reconcile these views. The delegation of authority to government affects the composition of the pool of politicians. We cannot infer the character of politicians without chains from the average character of politicians under chains. Indeed, the candidate pool with optimal delegation can consist entirely of angels if [v.sup.*] [greater than] V, making constitutional constraints appear inconsistent with the observed character of politicians.(9) The evil knaves do motivates constitutional constraints, yet in equilibrium only angels may be chained.

IV. REPEATED ELECTIONS: SORTING KNAVES OUT OF OFFICE

Citizens can vote unsatisfactory politicians out of office. I turn now to the interaction between electoral sorting and constitutional limits.(10) A tension exists between the screening and sorting functions of elections: voters must accept the challenger if they want to dump the incumbent. Since my interest is the interaction between electoral and constitutional controls, I simply assume elections cannot simultaneously sort and screen.

Sorting requires repeated elections. Let a politician's time horizon be h [greater than] 1 periods with an election each period. Repeated elections here serve a pure sorting purpose; section V allows elections an incentive effect by letting knaves behave opportunistically. The remainder of this paper assumes an exogenous candidate pool containing m knaves and n angels. Citizens directly and perfectly observe a politician's behavior, so first period performance perfectly signals type.(11) Citizens return any incumbent who has not abused power and vote out of office any politician who has.

Candidates in the first period of their political careers fill all open seats. The fraction of terms an angel holds office with sorting, [[Pi].sub.r] is

(6) [[Pi].sub.r] = n [center dot] h/(n [center dot] h + m).

Sorting increases the proportion of terms served by angels, [[Pi].sub.r] [greater than] [[Pi].sub.0], because angels serve longer than knaves. The increase in [[Pi].sub.r] increases optimal delegation, [x.sub.r] [greater than] [x.sub.0]; thus electoral sorting substitutes for constitutional constraints. The ex post nature of sorting (it identifies knaves when they shirk) limits its effectiveness. Only constitutional constraints prevent entirely the possibly substantial cost of abuse of (nearly) unlimited discretion. The condition which rules out a corner solution without sorting, T[prime](1) [approximately equal to] [infinity], also ensures [x.sub.r] = 1 is not optimal with sorting.

Increasing the frequency of elections or lengthening political careers improves the efficiency of sorting; equation (6) shows that [[Pi].sub.r] [right arrow] 1 as h [right arrow] [infinity]. But continuous electoral review is not optimal since elections are costly (Barro [1973]).(12) As long as knaves remain in the candidate pool and electoral review is not instantaneous, [[Pi].sub.r] [less than] 1, which ensures [x.sub.r] [less than] 1 if T[prime](1) [approximately equal to] [infinity].

V. INCENTIVE EFFECTS OF ELECTIONS AND CONSTITUTIONAL CONSTRAINTS

Repeated elections make tenure dependent on behavior. Assuming knaves always abuse power renders reelection incapable of deterring shirking. I now consider opportunistic knaves, who abuse power only when this raises expected utility. Early models of electoral control focused primarily on the incentive effect of retrospective voting (Barro [1973], Ferejohn [1986]).

Politicians now have a two period time horizon. Citizens vote retrospectively, first period politicians fill open seats, and electoral competition does not screen candidates for office. Since an opportunistic knave may refrain from misusing power to secure a second term in office, satisfactory first period performance no longer necessarily identifies an angel. All knaves shirk in their second term in office. The last period problem complicates the control of shirking through retrospective voting, but is not crucial in the present model because angels always behave. The probability of shirking in period two when incumbents are reelected equals the fraction of knaves in the pool, [Pi] = m/(m + n).

Interpret the payoffs specified in section III now as per period payoffs. Let R be a knave's payoff in office when not shirking and [Beta] [less than] 1 be the utility discount factor. A knave who shirks in period one returns to private sector employment at a wage of E in period two. A knave refrains from shirking in the first period if

(7) R [greater than or equal to] (1 - [Beta]) [center dot] v [center dot] T(x) + [Beta] [center dot] E.

Holding office must provide a higher payoff than private sector employment, R [greater than] E, for a knave to behave. The coefficient v is distributed over the interval [0, V]. Reelection is an ineffectual incentive for the knaves who value shirking the most.

Allowing knaves to respond to incentive effects increases the proportion of terms in which power is used as intended, [[Pi].sub.i] [greater than] [[Pi].sub.r], which, everything else equal, allows a loosening of constitutional constraints, [x.sub.i] [greater than] [x.sub.r]. The incentive effect of elections seems to substitute for constitutional constraints. But an increase in discretion increases the gain from shirking. With x sufficiently large, equation (7) is violated and knaves misuse power when first elected to office, as previously assumed. Constraints complement the incentive effect of elections by limiting the knave's gain from misusing power.

Post-political incentives can control last term shirking by politicians: making a pension or criminal prosecution contingent on performance can supplement electoral defeat as a deterrent. Lott [1990] demonstrates that post-political affect how often final term representatives vote, the form of last period shirking identified by Lott [1987].(13) But the benefit from shirking increases with x, while several factors limit the extent to which the punishment can increase with x. First, an upper bound on punishment exists, after which the marginal penalty must be zero. Second, timing could limit the maximum punishment; life imprisonment is less of a penalty for an elderly former politician. Third, the penalty is discounted while the benefits of shirking occur now. Impatient knaves, the ones least likely to respond to electoral incentives, are similarly unlikely to respond to penalties. Finally, a politician might avoid prosecution by going into exile. Unlimited delegation tends again to overwhelm the incentive for knaves to behave.

VI. DISCUSSION

Constitutional constraints and elections have been regarded as substitutes in preventing the misuse of power by politicians. My analysis indicates a complicated interaction between constraints and elections. Constraints complement electoral controls which affect politicians' payoffs (selection and incentive effects) while electoral controls preventing undesirable politicians from holding office (screening and sorting effects) substitute for constraints. Constitutional controls are designed to prevent abuse of power, while the sorting and incentive effects require on-the-job failure to remove politicians from office. Preventive measures are most appropriate to prevent abuses of power which result in significant, irreversible costs in a short period of time. Screening and sorting do not reduce the probability of the abuse of power to zero if screening is imperfect or if knaves remain in the candidate pool. If the marginal cost of abuse of unlimited power is very high, constraints on government remain optimal.

The framework of my model is favorable electoral controls. The common interest framework rules out possible disagreements among voters in directing their elected officials. Differences of opinion can lead to the nonexistence of equilibrium under majority voting, creating additional discretion for politicians. The well-intentioned politician causes no harm in this model. Hence constraints impose a very real cost on citizens if elections insure good politicians hold office. And detection of shirking occurs with certainty by the end of one term in office. Relaxing these assumptions strengthens the importance of constitutional constraints. Emergence of a role for constraints in an environment so conducive to electoral control is surprising.(14)

Delegating extensive authority to government indirectly undercuts the effectiveness of elections by increasing a knave's payoff. Delegation could additionally directly undermine electoral controls. Citizens may be unable to enforce any controls on a politician with sufficient power (to expropriate, jail or exile opponents without trial). By subverting democracy politicians can delay (perhaps forever) the imposition of electoral punishment. Ferdinand Marcos declared martial law in 1972 and did not face a competitive election for 14 years Constitutional limits have a clear advantage over elections on this score.

1. See Barro [1973], Lott and Reed [1989], Banks and Sundaram [1993], Reed [1994]. Ferejohn [1986] and Dixit, Grossman and Helpman [1997] examine electoral control in the absence of a common interest. The impact of a super-majority voting rule (Buchanan and Tullock [1962]) and the protection of minorities, two common elements of a constitution, cannot be analyzed within a representative citizen framework.

2. Coate and Morris [1995] and Brennan and Hamlin [1995] employ a similar categorization of politicians. An alternative form of heterogeneity has politicians differ in their cost of producing public goods (Banks and Sundaram [1993], Reed [1994]).

3. A single agent avoids the issue of separation of powers. Sutter and Tabarrok [1997] extend the model to multiple agents.

4. An alternative argument for constitutional constraints is the inability of the best-intentioned public servants to raise welfare.

5. An equal opportunity cost across types ensures that selection effects result from the payoffs of holding office. Perhaps, as Brennan and Hamlin [1995] argue, private sector employers can effectively monitor their employees. Good politicians' payoffs in Coate and Morris [1995] similarly depend on the value of benefits generated.

6. For sufficiently small values of x the candidate pool may be empty since B(0) = T(0) = 0.

7. Lott and Reed [1989] discuss the possible selection effects of electoral sorting. Elections reduce the period over which a power can be misused, while constitutional controls eliminate the opportunity to misuse this power altogether. The difference is important if knavish politicians value highly even a limited opportunity to misuse power.

8. For similar criticisms see Kelman [1987], Etzioni [1988, 51-63], and Kelman [1988, 217-23].

9. Empirical support exists for a largely angelic candidate pool; Dougan and Munger [1989] for instance find no evidence of widespread shirking in their study of Senate voting.

10. Lott and Reed [1989], Banks and Sundaram [1993], and Reed [1994] include a sorting effect of repeated elections.

11. Alternatively citizens might observe an outcome which depends on an unobserved exogenous variable as well as the politician's actions. Imperfect observation of action introduces noise in to the performance signal and reduces the effectiveness of electoral sorting.

12. This conclusion relies on the assumption of perfect observation of shirking by voters. A shorter term reduces the cost to voters of misbehavior, and could produce an offsetting decrease in effort exerted monitoring politicians' performance.

13. Lott [1987] previously showed that a last period did not affect how representatives vote when they vote. Let p be the utility value of the punishment imposed for misusing government power, which is imposed with certainty and only on actual shirkers after completion of the term in office. A knave in his second term in office shirks if u [center dot] T(x) - [Beta] [center dot] p [greater than or equal to] R.

14. On the other hand, I have examined the aspects of electoral control separately. Their combined effect exceeds the effect of any one component. The screening effect of elections, for instance, strengthens the selection effect.

REFERENCES

Banks, Jeffrey S. and Rangarajan K. Sundaram. "Long-Lived Principals, Short-Lived Agents" Manuscript, University of Rochester, 1993.

Barro, Robert "The Control of Politicians: An Economic Model." Public Choice, 14, 1973, 19-42.

Brennan, Geoffrey and James M. Buchanan. The Power to Tax. Cambridge: Cambridge University Press, 1980.

-----. The Reason of Rules. Cambridge: Cambridge University Press, 1985.

Brennan, Geoffrey and Alan Hamlin. "Economizing on Virtue." Constitutional Political Economy, 6(1), 1995, 35-56.

Buchanan, James M. "The Constitution of Economic Policy." American Economic Review, September 1987, 243--50.

Buchanan, James M. and Gordon Tullock. The Calculus of Consent. Ann Arbor: University of Michigan Press, 1962.

Coate, Stephen and Stephen Morris. "On the Form of Transfers to Special Interests" Journal of Political Economy, December 1995, 1,210-35.

Dick, Andrew R. and John R. Lott, Jr. "Reconciling Voters' Behavior with Legislative Term Limits." Journal of Public Economics, 50, 1993, 1-14.

Dixit, Avinash, Gene M. Grossman and Elhanan Helpman. "Common Agency and Coordination: General Theory and Application to Government Policy Making." Journal of Political Economy, August 1997, 752-69

Dougan, William R. and Michael C. Munger. "The Rationality of Ideology." Journal of Law and Economics, April 1989, 119-42.

Etzioni, Amitai. The Moral Dimension: Toward a New Economics. New York: Free Press, 1988.

Ferejohn, John. "Incumbent Performance and Electoral Control." Public Choice, 50, 1986, 5-25.

Hamilton, Alexander, James Madison and John Jay. The Federalist Papers. New York: Mentor Books, 1962.

Hume, David. The Philosophical Works, Volume 3, edited by T. H. Green and T. H. Grose. Darmstadt: Scientia Verlag Aalen, 1964.

Kelman, Mark. "On Democracy-Bashing: A Skeptical Look at the Theoretical and 'Empirical' Practice of the Public Choice Movement." Virginia Law Review, 74(2), 1988, 199-273.

Kelman, Steven. "'Public Choice' and Public Spirit." The Public Interest, 87, 1987, 80-94.

Lott, John R., Jr. "Political Cheating." Public Choice, 52, 1987, 169-86.

-----. "Attendance Rates, Political Shirking, and the Effect of Post-Elective Employment." Economic Inquiry, 28, 1990, 133-50.

Lott, John, R., Jr. and W. Robert Reed. "Shirking and Sorting in a Political Market with Finite-lived Politicians." Public Choice, 61, 1989, 75-96.

Mikva, Abner J. "Foreword." Virginia Law Review, 74(2), 1988, 167-77.

Reed, W. Robert. "A Retrospective Voting Model with Heterogeneous Politicians." Economics and Politics, March 1994, 39-58.

Sutter, Daniel and Alexander Tabarrok. "Separation of Powers in a Model of Angels and Knaves." Manuscript, University of Oklahoma, 1997.