27
Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach Author(s): Gerald Pech Source: FinanzArchiv / Public Finance Analysis, New Series, Bd. 54, H. 4 (1997), pp. 537-562 Published by: Mohr Siebeck GmbH & Co. KG Stable URL: http://www.jstor.org/stable/40912796 . Accessed: 16/06/2014 07:06 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Mohr Siebeck GmbH & Co. KG is collaborating with JSTOR to digitize, preserve and extend access to FinanzArchiv / Public Finance Analysis. http://www.jstor.org This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AM All use subject to JSTOR Terms and Conditions

Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Embed Size (px)

Citation preview

Page 1: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium BudgetCycle ApproachAuthor(s): Gerald PechSource: FinanzArchiv / Public Finance Analysis, New Series, Bd. 54, H. 4 (1997), pp. 537-562Published by: Mohr Siebeck GmbH & Co. KGStable URL: http://www.jstor.org/stable/40912796 .

Accessed: 16/06/2014 07:06

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Mohr Siebeck GmbH & Co. KG is collaborating with JSTOR to digitize, preserve and extend access toFinanzArchiv / Public Finance Analysis.

http://www.jstor.org

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 2: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium

Budget Cycle Approach

by Gerald Pech*

1. Introduction

There are two major stylised facts related to the phenomenon of the po- litical business cycle: first, voting responds to recent economic performance

1 . Second, the time path of policy variables seems to follow a business cycle with a time pattern which corresponds to the timing of elections. One of the earliest contributions in the empirical literature is Tufte (1978), who reports cycles in transfers and taxes. Bizer and Durlauf ( 1 990) and Hess ( 1 992) found that the dynamics of taxes in the U.S. show cyclical behaviour, suggesting the existence of a political budget cycle. Alesina, Cohen and Roubini (1992) report indications of political budget cycles for a sample of OECD countries, although they were not able to find strong electoral effects.

The earlier theoretical literature on the political business2 cycle consid- ered the mechanistic voting behaviour stated in the first observation to be an exogenous device which can be exploited by the politicians. The politicians' discretion in setting policy variables then explains the second observation. But the rational expectations revolution in macroeconomics cast doubt on this view. Following the rational expectations argument, voters would not let themselves be cheated repeatedly3. Even though they may be imperfectly informed, it is not clear, why they should have biased beliefs about the ef-

* This paper is based on a chapter of my doctoral thesis. Part of this paper was done while I was supported by the Graduiertenkolleg „Allokationstheorie, Wirtschaftspolitik und kollektive Entscheidungen" at Bochum and Dortmund. I wish to thank the editor, two anonymous referees, R. Dawid, B. Felderer, C. Folkers, K. J. B. Neumaerker, J. Poterba, W. Rippin, K. Rogoff and participants of seminars at Bochum and Bern for helpful com- ments at various stages of the study.

See, for example, the survey of Kiewiet and Rivers (1985). 2 See Nordhaus (1989) for a survey. 3 See Alesina (1989).

Finanzarchiv N.F., Bd. 54 (1997) S. 537-562 © 1998 Mohr Siebeck Verlag - ISSN 0015-2218

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 3: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

538 Gerald Pech

fects of policies4. So at first sight and contrary to the empirical findings a political business cycle will not persist with rational voters. Furthermore, the forward looking concept of rational expectations seems to rule out any mo- tivation for retrospective voting decisions.

Several reasons have been given in the theoretical literature as to why ret- rospective voting may still be in accordance with rational behaviour on be- half of the voters: voters agree to penalise politicians who do not act in ac- cordance with voters' interests (Ferejohn, 1986) or economic performance reveals some information of interest to the voters. Examples are information about the efficiency of different political instruments which are employed by politicians according to their ideological beliefs (Harrington, 1993), about the true ideological objectives of politicians (Hess, 1991) or about the abil- ity of politicians (Rogoff and Sibert, 1988; Rogoff, 1990). All these ap- proaches highlight the incentives for politicians to manipulate policy. Fur- thermore, the last three contributions emphasize the emergence of a politi- cal business cycle.

This paper extends the arguments in Rogoff (1990) in order to deal with the intertemporal budget constraint explicitly. While Rogoff restricted him- self to a model with a public investment opportunity where the true size of investment cannot be observed before the election takes place, we incorpo- rate public debt explicitly into our analysis. The common starting point is a game of asymmetric information, where the ability of politicians is private information and one policy variable - public debt or public investment - is not directly observable and displays its economic effects only in the post- election period. Pre-election ability is a predictor of post-election ability. Politicians are welfare-oriented and office-motivated. Under these assump- tions, they face incentives to engage in signalling in order to get re-elected. One can show that an equilibrium exists which separates the incumbent with low from the incumbent with high ability. In this equilibrium, the in- cumbent of high productivity distorts economic instruments in order to show his competence. Incumbents who know that they will not be re-elected can do no better than to accept their fate and act according to their welfare ob- jectives5.

Besides considering public debt instead of investment, we also deviate from Rogoff s assumptions in the following respects: Rogoff assumes that taxes are lump-sum and that different abilities express themselves in different lump-sum capacities to spend. Here taxes are distortionary and dif-

4 See Wittman (1995) and the discussion in Coates and Morris (1995). What prevents the prevalance of pooling-equilibria in our context is the fact that the

incumbent with high ability is more efficient in sending the favourable signal than the other.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 4: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 539

ferent abilities express themselves in different marginal productivities of spending activities. The first assumption is necessary in order to have a well defined path of public debt (see Barro, 1 979). As a consequence, in the unique undominated equilibrium, the politician of high ability induces both an inter- temporal and an intratemporal distortion. The second assumption is respon- sible for the main policy implication of this contribution which contradicts an important result of Rogoff 's paper. While he shows that there is no way to improve upon the outcome of the signalling equilibrium by introducing a balanced budget rule, we argue that there generally is. The negative finding emerges because signalling efforts will also prevail under the additional con- straint, but the cost of signalling will generally be higher with the constraint than without. The reasoning behind his result is that, even under the balanced budget rule, politicians of different abilities cannot be readily discerned un- less they employ distorting policies to signal their abilities. Furthermore, as politicians are welfare motivated, they will always choose the cheapest way to send a given signal. Imposing additional restrictions on their signalling behaviour, therefore, can never improve on the outcome.

With different marginal productivities, however, politicians of different ability face budget lines that do not coincide. So if the politician acts under the balanced budget rule, his true ability can be observed without any sig- nalling efforts. Under the rule, ex post intertemporal efficiency is violated for at least one of the potential types. It can be shown, however, that if the political income and, therefore, the incentives to manipulate are sufficiently high, then introduction of a balanced budget rule is justified by an ex ante welfare criterion. In an extension we discuss the introduction of a public in- vestment opportunity in addition to current public expenditures. Public in- vestment may be financed - on the whole or partially - with debt without destroying our previous results.

The assumption that the true public deficit is directly unobservable to the public is essential for our results. For a model with only three instruments - debt, taxes, and government spending - the deficit as defined in the national account would also be the one required to assess the budget constraint in the model, so reporting the deficit would remove the asymmetric information from the model. We believe that in reality it is much more complicated for the public to calculate the deficit, which is necessary to solve her decision problem. In a context as the one at hand, this would mean calculating the change in the public sector's net worth which is hard to assess from the re- ported deficit (see, e.g., Buiter, 1983). Even if the government were obliged to report a deficit measure based on public sector's net worth instead of a sim- ple cash flow measure, it would have to evaluate assets and liabilities which would open the door to manipulation. Then the public will distrust any infor- mation as long as there is no trustful auditor of the government's account.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 5: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

540 Gerald Pech

Compared to the realisation of full information, the imposition of a bal- anced budget constraint in chapter 5 is only a second best device. The rele- vance of this second best case could be considered as doubtful because if anybody can attain the information necessary to verify the obedience of a balanced budget rule, he will also be able to transmit the same information to the public, i.e. by writing the measured deficit into the national account, thereby undoing the asymmetric information problem. The only unobserv- able factor left would be the identity of the incumbent, but this could be de- duced immediately from his current account. Although in a strict sense we cannot escape from this kind of reasoning, we can try to argue why the im- position of a balanced budget constraint may be feasible, whereas an im- provement of the information function of the national account in isolation is not. Ultimately, the argument necessarily hinges on whether or not informa- tion can be credibly transmitted to the citizen. We feel that the existence of a balanced budget constraint in itself could ensure that the information trans- mitted is also credible. Suppose that some auditing agency were charged with the task of collecting information on the true deficit. This information can either be used exclusively to be transmitted to the citizens or it can also be used to verify whether the government has obeyed the balanced budget rule. In the case of disobedience, let the budget law be nullified. In both cases, the citizen has no means of verifying whether the agency works correctly. In the former case, the agency's reputation is the only basis for trust. With a balanced budget rule, this would be different: if the agency's behaviour has a direct impact on the actions of the government, any of its decisions will be subject to judicial verification. It is exactly the institutional surrounding nec- essary to make a balanced budget rule work6, which gives credibility to in- formation emerging from the budget process. No improvement of account- ing standards alone could accomplish this, unless the accounting agency ex- pects to get sued for distributing incorrect information.

Another point with respect to why imposing a balanced budget constraint makes a difference are the informational problems involved in the decom- position of some given deficit figure. Such a decomposition is necessary in order to derive information with respect to any specific aspect7. Imposing a balanced budget rule will once and for all do away with the question of which deficit measure contains the information the voter needs for his voting decision. This problem only has to be solved once when the budgetary in- stitution is designed. Note that we do not address the question of whether the balanced budget rule is the best of all attainable rules here nor do we

6 See, e.g., Inman, 1996, pp. 9-10. ' See, e.g., Eisner, 1984.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 6: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 541

discuss what is the most sensible rule in a growing economy. So far we only show that a rule such as the balanced budget rule will definitely become desirable, if the gains from manipulating the economy become large enough.

We proceed as follows: in section 2, we set up the model. In section 3, we derive the policies for the different kinds of politicians under full informa- tion. In section 4, the game between politician and voter under asymmetric information is analysed. We show that there exists a separating equilibrium of the signalling game where the highly productive type distorts in order to signal his true competence. Section 5 contains various extensions and ap- plications of our model. In section 5.1, we highlight the implications of ex- tending the time horizon. In section 5.2, the consequences of introducing a balanced budget rule are analysed. Our results concerning the benefits of a balanced budget rule are compared to the results of Rogoff (1990) in section 5.3. In 5.4, we discuss the implications of having public investment and pub- lic debt at the same time. A summary and discussion of our proposed bud- get rules concludes the article.

2. Setting up the Model

We consider a small open economy. Initially we restrict the time horizon to two periods. Call the last period T and call the present period T- 1 . This denomination is chosen to make the feedback structure obvious and to facil- itate the argument when extending the model over more than two periods. The economy is inhabited by agents who enjoy the same utility from public and private consumption, defined as

(1) WT_X = v(gT_x) + u(cT_x) + ßET_x[v{gT) + u(cT)l

gt and ct are public and private consumption, respectively. ET_X is the ex- pectation operator in time 7-1, ß is the discount factor, and v and u are con- cave, well-behaved, subutility functions which satisfy the Inada conditions to ensure the existence of an interior solution. The private agent observes current taxes TT_X and infers future taxes rT. He allocates his expected net wealth to present and future consumption subject to the intertemporal bud- get constraint

(2) cT_l+ßET_icT=(l+r)kT_2-f(TT_l)-(l+r)-lET_if(TT) where initial private capital claims (including human capital) kT_2 are given by history. Taxes are distortionary. /(r,) is the total loss of private wealth from tax policy in period i, which is composed of the direct tax payment xt and an excess burden h (t,). Then/(rf) = T, + h (T,) where h t(t,) > 0, /itt(t,)>0

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 7: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

542 Gerald Pech

to provide an argument for tax smoothing8. The agent faces an international capital market with the interest rate r where he can intertemporally smooth his consumption stream. We assume that ß ( 1 + r)"1 = 1 . Then the optimal con- sumption policy is given implicitly by

(3) Vi^TM^r

where Xt - uc(ct) denotes the marginal evaluation of private wealth by the private sector.

If the agent is a voter, then his full utility is

r't is a stochastic taste shock, which influences the voter's decision and conceivably encompasses the effect of election-eve debates9. We assume that Ef) = 0. If the agent is a politician, then his utility is10

n^ = wr-i+yr-i +PyT-

yt is the discounted political income. It is weighted by the re-election prob- ability P. We assume that the functional form of both W(R) and W(P) is com- mon knowledge.

The incumbent maximises W(R) subject to the reaction function of the pri- vate sector and the budget constraint. The private sector reacts in two ways to government policies. First, agents adjust their consumption policy accord- ing to (3). Second, they decide on whether or not to re-elect the government. This latter decision is trivial in the case of full information, but requires so- phistication in the case of asymmetric information, which we discuss in sec- tion 4. The government budget constraint is given by

(4) (1+r) bt_x + q(£t) gt- T, = bn

where t e {T-'' T}. bt refers to the net indebtedness of the public sector in period t. Net indebtedness bt = Bt- Ht is the difference between nominal

8 For a micro-foundation of this functional form see Bohn (1986). 9 Rogoff(1990, p. 25). The implied welfare orientation can be rationalised on the grounds that any politi-

cian is elected from among the public and, therefore, has the same preferences. Note that in election games with a final period and no commitment any move of the politician in the first period can be sequentially rational, if he is purely opportunistic, see, for exam- ple, Greenberg (1990, example 8.4.) or Harrington (1993, pp. 35-36). Empirical obser- vations suggest that politicians behave as though they had some ideological motivation which can be interpreted as a special form of welfare orientation (see Alesina and Rosen- thai, 1995). We assume that political income accrues in the form of "Ego rents", see Rogoff (1990, pp. 24-25). If running for the presidency is costly, an equilibrium in the political market demands the existence of "rents" that can be secured by getting into office, see, e.g., Barro (1973).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 8: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 543

public debt outstanding, B, and public assets, H. Both B and H earn an inter- est r for the creditor1 1. We set HT = BT = 0, i.e. all assets are sold off and the public debt outstanding is paid back in the final period. Note that if Ht is un- known, as we assume, then there is no way to directly assess net indebted- ness from the nominal debt figure Bt alone. For any government in t, Bt_x and Ht_x are given exogenously. q (£,) is the producer price of one unit of the public consumption good in terms of the private consumption good and de- pends on whether the ability of the politician in charge of the government, £„ is high (¿*) or low (e1). The producer price is below average for a poli- tician with high ability [i.e. q? = q (£, = é*) = q° - i] and above average for a politician with low ability [i.e. q' - q (£, = £*0 = q° + i] by some small amount i. This expresses our assumption that overall efficiency in the pub- lic sector is higher with a politician of high ability and thus productivity in the supply of public goods is higher. In what follows, we will, therefore, use the terms "productive" and "able" interchangeably to address the character of a politician. Furthermore, we assume that time (r-l)-ability is a perfect predictor of time T-ability, i.e. for any incumbent ëp = ép_x, i e {L, H] 12.

The timing of the game between incumbent and the public is as follows:

(1) In the beginning of period 7*-l, the incumbent observes his actual abil- ity, sets taxes rT_{, government expenditures gT_x and finances the bud- get by issuing debt bT_{.

(2) After observing the gT_x and rT_ly the private sector updates its expec- tations on future tax policies and chooses its consumption cT_v Expec- tations on future tax policies depend on who the consumer expects to run the government in the next term. Under asymmetric information, the public cannot directly observe the level of debt outstanding nor the true ability of the incumbent. As the equilibrium of the game is separating, however, the private sector can always infer both values.

(3) Under asymmetric information, the electorate, after observing rT_ly gT_h calculates some a posteriori probability P(gT-i> Tt-i) that the incumbent is the high ability type. As the equilibrium is separating, P is either 1 or 0.

(4) At the end of period 7-1, voters observe 7] and elections are held. The electorate can choose between staying with the incumbent or drawing some candidate at random from the set of challengers, who are distrib- uted over {eHt £*-}.

1 1 The implications of having an investment opportunity which yields an endogenous rate of return are discussed in chapter 5. " Here our model differs from that of Rogoff (1990) and Rogoff and Sibert (1988), where ability follows a serially correlated stochastic process. Our assumption simplifies ex- position in the two-period case and allows the determination of the reference case for the high ability incumbent - that is, his full information policy - as a policy of constant debt.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 9: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

544 Gerald Pech

3. The Optimisation Problem under Full Information

In order to set up the problem under full information, we assume that the 77-shocks are sufficiently small, so that an incumbent is never elected if the public knows that he has a low ability13. Ruling out the impression shocks, it is evident that the public does better by choosing a candidate at random from the set of challengers which contains at least some //-ability types, than by staying with a politician who is unable with certainty. If his ability is com- mon knowledge, the incumbent cannot influence his re-election prospects, so his expected gain from winning the election, P yT, is exogenous. There- fore, in T-Ì he maximises his pay off W^' by maximising the common pay oïïWT_x.

It is then easy to find the solution for the problem of the government in T-Ì. This is accomplished by backward induction, so we consider the pol- icy in period T first. With all (r-l)-instruments fixed and the consumption plan of the private sector executed except for the last period, an optimal pol- icy in period T requires

(5) ^(£7-)"1vg(gr) = ̂ /^(^). This means that the government has to observe its budget constraint

and set taxes and public expenditures so that the marginal utility of pub- lic consumption per dollar expended equals the marginal costs of public funds.

In order to analyse time (r-l)-decisions of the government, let ET_xVT(bT_x, *r-i) denote the value function ET_xVT=ET_x{ß v[gT(bT_x)] + i/r_1[cr_1(r:r_1, bT_x), cT^T-'^T-')'y £t) where UT_X is comprised of the M-terms in (1). It is straightforward to show that the reaction functions of the time ̂-government are dgT/dbT_x < 0, dXjldbT_x > 0. This, together with our assumptions about u, v and/, proves concavity of VT(bT_x, xT_x' £T). Maxi- mising v (gT_x) + ET_xVT(bT_X9 rT_x; e) subject to the government budget con- straint and the reaction function of the private sector, yields the first-order conditions

(6) q(eT_xrlvg(gT_x) = ET_xVZ(bT_x, xT_x' eT),

(7) Vi/rfc) = ET_xvZ(bT_x, xT_Xi eT) .

13 The only purpose of the 77-shocks is to generate some noise in order to estab- lish strong monotonicity of the re-election probability P in the a priori probability of facing a type H, A In the full information case, however, we may set P(P = l)=l and P(P=0)=0.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 10: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 545

Xj_x is determined in equilibrium together with the optimal government pol- icy14. For the government, the marginal utility of public consumption per dollar expended should equal the marginal cost of public funds. Addition- ally, the marginal cost of debt finance must equal the marginal cost of taxa- tion.

From (6) and (7), we immediately obtain a condition which governs the efficient employment of the current period variables xT_x and gT_x'

(8) ^(eT1v^(^_1) = Ar_1/T(rr_1).

Note that the tax/spending policies which satisfy conditions (6) and (7) dif- fer for incumbents with low and high ability.

Now we are going to determine the value of the policy variables which a high or low ability incumbent will choose. The following lemma lc-d hold under the additional assumption that the price elasticity of demand for the public good is smaller than 1 , which implies that a higher price of the pub- lic good goes along with higher current nominal expenditures G = q g. In what follows, the index H or L refers to the policy which a type H or type L will choose, respectively. Outcomes which prevail in equilibrium have one asterix. Outcomes which refer to some out-of-equilibrium situation are marked with two asterixes.

Lemma la: If a type H is re-elected with certainty, he chooses a stationary path of public good supply, taxes and nominal government expenditures, i.e. 8?>8?: t£W and G& = G?'

Proof: Appendix la.

Lemma lb: A type L who is re-elected with certainty chooses gT*-*'-gT**> Tf_! = Xj and Gj_x = Gj .

Proof: Appendix la.

In both cases, re-election prospects are exogenously determined and gov- ernments in both periods are identical. Policy making is driven by the wel- fare motivation of the government. Consequently, it enacts a policy of tax

14 Working backwards from (5) to (7) which are interconnected through (2b) and the government budget constraint, we can determine the sequence T,(Ar_i), t=T-', Tup to the parameter Ar_j. From (2a) we have a mapping of (efficient) tax policy levels onto pri- vate net wealth v, which is related to an initial value kj^iy). Finally, there is a unique value of kT_i which the private sector would choose and which in turn fully determines the sequence T,(Ar_i), see on this Kydland and Prescott (1980).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 11: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

546 Gerald Pech

and real expenditure smoothing. Stationarity of the solution is implied by having the same cost- and utility-subfunctions in both periods and the iden- tity of objective and subjective discount rates. An intertemporal differenti- ation is not beneficial because the marginal evaluation of funds in both pe- riods are the same.

Lemma lc: If a type L knows that he will not be re-elected and if the abso- lute value of the price elasticity of demand for the public good is smaller than one, the incumbent realises a deficit after the first period. We can also show that he realises a lower supply of public goods and higher taxes than the high ability incumbent of lemma la in the first period, that is, g£!i < gr-i and if *j > t£*! .

Proof: Appendix la.

Again, re-election is exogenous. However, the identity of the successor is random. The result again is implied by tax and real expenditure smoothing considerations. The unable politician knows that his own contribution to wel- fare is below average, so he runs a deficit in order to carry out a policy of tax and real expenditure smoothing. We summarise our results with the fol- lowing proposition:

Proposition 1: In the case of 77 =0, Lemma la and Lemma lc fully describe the policy sequences under full monitoring.

Proof: If 77=0, under full information, type H is always re-elected and a type L is always dismissed. D

In the game with asymmetric information, the type H might find himself in a situation where the public does not recognise his true identity. In order to know what he is prepared to do to make himself discernible, we have to specify what his policy choice would be, if he were not recognised. The re- spective policy choice in this state - which, as we are going to show is never realised in equilibrium - is stated in the following Lemma Id:

Lemma Id: Under the same conditions as in lemma lc, if a type H knows that he will be succeeded by a candidate who is chosen at random from the distribution of challengers, he chooses a surplus and a realisation of first- period expenditures and taxes (g?** >*?**) suc^ that 8t*' < 8?-' < 8t-i and T7*-l> TT-' >T-'-

Proof: Analogous to the proof of lemma lc.D

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 12: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 547

Figure 1 Undominated Separating Equilibrium

T II

V *H *

g

In figure 1 we have depicted the choices (g, t) in T-' under full moni- toring as (gL*, 1^*) for typeL and (#"*, r***) for type// politicians. We also have plotted the tax and real expenditure configurations which satisfy the inner-period optimality condition (8) for the incumbent: <P''(S) = {T(S) 'q(érx v8(g) = A/T(T)}, ¿ = L, H. In other words, to fulfill inner-period efficiency for any given marginal utility of g, the tax price per unit of g must be the same for either type. Therefore, the cost of taxation per unit of revenue must be higher for a high ability type who faces a lower price per unit, so taxes have to be higher. The SL-line represents those points which are attainable for an L-regime if there is a balanced budget rule, i.e. S'(g)= {T(g)| q(e')g = i:}9 i=L,H. This line from the origin has a slope of q(eL). The E^-line represents the balanced budget line for a type H in- cumbent. It has a slope of q(eH). As we have shown, (gH*, T*7*) is on the SH-'inc. (gL*, r*'*) is situated to the north-west of (gH*, t"*) and below the sMine, i.e. under full monitoring, the type L realises less real expenditures and higher taxes than a type H and still runs a deficit. Both (gH*, T^*) and

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 13: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

548 Gerald Pech

(gL*, TL*) are situated on the tfMine of the respective incumbent, which means that, with full monitoring, nobody can gain anything by violating the opti- mality condition at the intratemporal margin (8). Note that S does not refer to an optimality condition, but to an imaginary balanced budget constraint which only coincides with the intertemporal optimum for H. It is easy to understand that, with a change in the production costs of g, we have to change (g, t) to satisfy the intratemporal optimality condition. Furthermore, O has a negative slope and 0H is situated to the north-east of &L.

4. The Problem under Asymmetric Information

Now we assume that voters can observe the realisations of r and g, but cannot directly observe the incumbent's type or bT_x. This turns the government's problem into finding a strategy for a signalling game. Having observed g and r, the voters update their beliefs as to which kind of incum- bent they face. Let P be the probability that the incumbent is a type H. This probability may depend on g and T. Inserting the expected realisations of the policy variables into (1), we see that he will be re-elected if

(9) ß [P VT(e?) + ( 1-P) VT(efr - E VT(ë)] + rç^ - £ f? > 0.

(9), together with the distribution of 7], implicitly defines the re-election prob- ability P [(g, t)]: after adjusting for the impression shock rj, the expected pay off from staying with the incumbent has to be at least as high as the ex- pected pay off with a candidate, who is drawn at random from the distribu- tion of challengers.

4.1. Sequential Equilibrium We analyse a sequential equilibrium in pure strategies in the signalling

game between an incumbent in 7-1 and the public. In this game, the voter's response is re-election, which occurs if (9) is fulfilled, or dismissal. The in- cumbent chooses gT_{ and xT_x in order to maximise his objective function WT~l (/?), observing the voters' response15. In equilibrium, the type L chooses gL, ih and the type H chooses gH, if*. The public's a posteriori probability of facing a high ability type is Bayes-consistent, i.e. for (gH, t*1) * (gL, t^) they set P(gH, rH) = l and P(gL, rL) = 0. Then (g' t'' ie(H,L), and P [P (g' T1)] constitute a sequential equilibrium.

15 See equation (11) below.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 14: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 549

4.2. Separating Equilibria

In order to construct an equilibrium for this game, we start off with the observation that an incumbent of either type would prefer to implement his full information policy, if he were recognised. Yet the low ability type in- cumbent could be tempted to compromise on his welfare objective in order to convince the public that he is indeed a high ability type. This in turn could induce the high ability type, to strive for recognition of his true type. If the high ability type is successful in making himself discernible, we say that the equilibrium is separating.

Since the incumbent in T does not face a re-election restraint, he chooses the same policy as under full monitoring. Observing that in 7-1 re-election probability depends on rand g by P [P (g, r)], his expected utility from con- sumption is given implicitly by

(10) ZT-l®=WT-l{gT-u ¿r-i^r-i, bT_x), P [P(TT_Xi gr.Ol ¿T-ih

where the budget constraint of the government is used to eliminate bT_x and we substitute the subutility from the optimal consumption choice, which satisfies (2) and (3). For ease of notation, we drop the time indices and ab- breviate the expression in (10) as Z[g, r, P(t, g), e1]. The incumbent's over- all utility is ZT~l+P [PíTt-.j, £t*_i)] y. The gain of an incumbent of high abil- ity, if he is recognised, consists not only of the additional political income [P(l) - P(0)] yrbut also of the higher realisations of g and c in the last pe- riod - evaluated as [P(l) - P(0)] [VT(e?) -EVT(ë)]. In contrast, the low ability incumbent, if he would be re-elected, evaluates his political income, but loses utility from the economic damage he causes, [P(l) - ^(0)] [VT{£j) - E VT(ë)] < 0, which he has to set against his additional political income16. We will assume that the difference is positive, at least at gL*, rL*, which im- plies that the type L would like to be re-elected, at least if he can follow his most preferred policy at the current stage. Suppose that the equilibrium is separating, i.e. we have (gL, t^) * (gH, r"). Then a type L is recognised and P (gL> T1) = 0, so L can do nothing but follow the same policy as under full monitoring, (gL*, TL*).

It remains to show that the equilibrium is indeed separating17. First, we define the non-convex set A, which contains all realisations of g and T, which a type L would never realise, even if he could induce the public to set P = 1 by doing so. These are the policy options from which a type H can choose

16 This implies that the high ability type is always prepared to compromise more than the low ability type in order to insure his re-election.

My argument here closely follows the discussion in Rogoff (1990).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 15: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

550 Gerald Pech

in order to make himself discernible. More formally,

(11) ÍA:(í>t)|Z(^,TL,l>£L)+i8[P(l)-P(0)]yr-Zr-V*.T£'*>0,eL)>0}.

If the type H refuses to be re-elected and chooses some point (gH, xH) e ' A, then the public would set P =0 whenever the equilibrium is separating. Con- fronted with P =0, the type H is succeeded by a candidate which is drawn by random, so from Lemma Id he chooses (gH**t r^**) 18. Now we can con- struct the set B, consisting of all allocations which are weakly preferred by type H incumbent over his reservation value given in (gH**, T^**) with

(12) {B:(g,T)'Z(gH,TH,l,eH)+ß[P(l)-P(O)]y -Z'-I(i"",tH-*,0,^0}.

It is straightforward to show that A n B is not empty. The proof, which is omitted, rests on the observation, that the gains which a type H can secure by his re-election always exceed the gains which a type L could expect (see Pech, 1996). So, if the points (gL' r1*) and (r"**, gL**) are situated close to each other, the type H is always ready to compromise more than the type L in order to insure his re-election.

4.3. Undominated Separating Equilibrium In signalling games, there may exist a large number of separating equilib-

ria, as the second party can hold any belief in response to out-of-equilibrium messages. We can reduce this number to a single point by requiring that the private sector sets P =1 not only for (gH*> r"*) but for any g, r e An B19. This is justified because a type L can never win anything by realising a point inside of AnB. Then the type H can choose any point within the set An B. The point within the set that allows the highest utility is the point where his indifference curve is tangential to the curve which marks the boundary of the set A20. The boundary curve is implicitly defined as the curve where in- equality (11) becomes binding. These are the points that L would just not choose.

In appendix 2, we show that the tangential point in figure 1 is always sit- uated above &H. Because of this, we can see that efficient signalling not only causes an intertemporal distortion through debt policy, but also distorts at the intratemporal margin. The incumbent of high ability extends the sup-

18 One can show that whenever (gH**, r"**) e ' A, then (gH*, rH*) e ' A. Iy See proposition 2, Rogoff (1990, p. 29). i ne argument in appendix I establishes surhciency.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 16: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 551

ply of public goods beyond the point where the marginal cost of public funds are equal the marginal utility from public consumption per dollar expended. Because incumbents of different ability face different expenditures per unit of the public good, the type H can offer a more favorable tax/expenditure mix. In order to signal his true character, he overemphasizes this capacity.

Now define the income expansion path r(v) = *F[g (v)] which denotes points (g, r) which solve the optimisation problem of an incumbent of type H for different parameter values of the political income y 2 1 . If (gH*, T^*) e A, the type H is unrestricted and will always choose that point. The exposition in appendix 2 shows that the higher the political income, the lower is the util- ity from consumption which the incumbent shares with the private agents. This is also intuitive because higher stakes will make both types ready to compromise on the welfare maximising policies they would apply under full monitoring.

4.4. Ruling out all Pooling Equilibria

Pooling equilibria might still prevail if the a priori P is sufficiently high, so that after observing some policy (gp, tO the private sector holds on to its beliefs and sets the a posteriori P (gp, T7*) = P, if the type H concedes to such an equilibrium. However, we can show that, in any pooling equilibrium (gp, r7*), there exists a deviation of the type H which improves his utility and cannot be matched by the type L. Refining the equilibrium concept by mak- ing use of the intuitive criterion, we can rule out such equilibria22.

We summarise our results in the following proposition:

Proposition 2: There exists a unique, separating, intuitive equilibrium (gL*> ^' (*", g' P(%ge AnB = 1); P(r, g e 'A = 0) for the signalling game under asymmetric information.

5. Extensions and Applications 5.7. Extending the Time-horizon of the Game

Up to this point we have analysed a two period model. It is obvious that we can find incentives for inefficient debt policies in a two period context but nothing like a business cycle. In order to extend the model, we assume that the stochastic evolution of the ability shocks to some politician j is as

21 y(g) is defined through / in appendix 2 as { r= *F(g) | [r(y), g(y)] e argmax l(y)}. Ll See Cho and Kreps (1987, pp. 195-196, 202). The proof parallels the proof of prop- osition 3 in Rogoff (1990) and is omitted.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 17: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

552 Gerald Pech

follows: ̂U) = er"10); ̂ ~1U) = ^~3U) et cetera. Elections are held at the end of T-s for odd values of s. Such a configuration implies that the public can learn nothing from looking back more than one period when casting a vote. This assumption is in line with the empirical finding that voting beha- viour is only influenced by current economic performance23. It is obvious, therefore, that in r-5-even there is never any signalling. I.e. in 7-2 we have the problem

Max {WT-2U = v(gT_2)+ßET_2VT-l(TT_2, bT_2, V2, e)h ie(H,L)

s.t. bT_2 = ( 1 + r) bT_3 + q(ë) gT_2 - TT_2,

^-2=rr_2{(l+r)[(*r_3 + ̂ ^

where kT_3 > 0 and bT_3 are exogenously given by history. Following the same argumentation as in lemma 1, we can show that the type L realises a deficit and the type H realises a budget surplus. Going back one period further, i.e. in the pre-election period 7-3, the type H would again face incentives to dis- tort his behavior to using his instrument mix and choose a lower surplus or even a deficit in comparison to period 7-2. With this sequence, a political business cycle emerges.

5.2. The Advantages of a Balanced Budget Rule

Implementing a balanced budget rule is frequently promoted as a means of coping with obviously welfare reducing debt policies. Yet our results sug- gest that we should be careful with such an argument. This is because sig- nalling not only implies a cost, but is also beneficial in that it reveals infor- mation. So a priori it is not clear whether there is any way to improve upon the signalling equilibrium in a world with private information. Nevertheless, it is easy to see that the benefits of the signalling behaviour can also be re- tained under the balanced budget rule, i.e. the incumbent of high ability can be discerned under the balanced budget rule. On the other hand, having a balanced budget rule is not completely costless. There is at least an ineffi- ciency cost from the inflexibility of government instruments in the case of a low ability incumbent who, under discretion, would like to follow a wel- fare maximising policy. Therefore, we have to compare efficiency costs under both policy regimes.

The returns from signalling can be maintained under the balanced budget rule because the //-type will have to choose a position on EH while the

23 See e.g. Fair (1978).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 18: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 553

L-type will be constrained to choose a point on EL. For both incumbents, the optimal point is the intersection of their respective ¿z-line with the 0-line be- cause nobody could gain anything by violating intratemporal efficiency. If the incumbent in 7-1 is of type //, he is simply bound to run a policy that is welfare maximising. On the other hand, the L-type is restrained from run- ning that welfare maximising policy - which would imply running a deficit - although he would choose such a policy in the signalling game above. For election periods other than 7-1, the balanced budget rule obstructs the tax and real expenditure smoothing policy of both the L- and the //-type.

Naturally we want to address the question of whether the balanced bud- get rule furthers the interest of the citizen, so we evaluate the effects on the utility from consumption alone. However, as the original equilibrium was separating, one can even argue that the balanced budget rule is also benefi- cial to the politician whenever it is beneficial to the citizen. Both the ab- solute cost of signalling and the cost from having inflexible instruments depend on the form of the production and utility functions. Yet we know that the cost of signalling increases with the political income y, whereas the inflexibility cost does not depend on y. Therefore, we can conclude that there is always some threshold value y such that, for y>y, the expected utility under the balanced budget rule is higher.

Proposition 3: For the 2-period game, there always exists y>y so that the expected utility of the game under a balanced budget rule is higher than the expected utility in the original signalling game.

Proof: Expected utility in the game for two periods under the balanced bud- get rule is PZ7""1 (*£*!, t£_;, l.e^Hl-^Z7""'^", rf",O, ety where P is the a priori probability of drawing an //-type. Point (#£!*, t£!*) is lo- cated at the intersection of &L and EL. Expected utility in the original game is P ZT-l(g%-i, Tt-i> 1. £?-i) + 0-J*) ZT~'gT-' rt' 0, éf_x). The only ex- pression which depends on y is ZT~ì(gT_ìì T^, 1, £r_i), which is strictly de- creasing in y as shown in appendix 224. Therefore, there is always some y >y for which utility in the original game is lower than in the balanced budget game. D

Note that the public need not be able to verify directly whether the bud- get rule is observed or not. It is sufficient that there is an auditing agency which can be trusted to enforce obedience to the budget law. But if the aud-

24 H trades Z for y at the price (p[P(')-P{0)] where q»0 and -^>0. dy

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 19: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

554 Gerald Pech

iting agency could transmit the required information credibly, then the in- formation asymmetry would vanish. Removing the asymmetric information problem is always superior to imposing a balanced budget law because it re- fers to a first best economic outcome. However, there are reasons to suspect that it is possible to impose the rule where it is much more difficult to im- prove the information technology in isolation. Clearly, imposing a balanced budget rule successfully means not only improving accounting standards, but also implies setting up a system of checks and balances. Most notably, the auditing agency would itself become subject to ex post control. This in turn enforces its ex ante credibility and turns the figures it produces into "hard facts". So the institutions which are necessary to make a balanced bud- get rule work would also give credibility to the information which emerges from that process. It seems hard to imagine, how any attempt to improve the information function in isolation of the budget process could arrive at the same credibility standards.

5.3. A Comparison with Rogoff's Results

Rogoff analyses the effect of introducing a balanced budget rule in a sim- ilar model in the presence of costly signalling efforts. He argues that soci- ety is better off in the unconstrained signalling equilibrium than in the sig- nalling equilibrium under the budget rule. The reasoning behind his result is that high-ability political incumbents who are at least partially welfare-mo- tivated will choose the cheapest way to signal their true identity. So impos- ing any additional restraint on them will lower welfare. Low ability incum- bents do not engage in signalling because the equilibrium is separating. What makes the difference between our model and Rogoff's model is that, in the latter, the budget sets of both types of politicians intersect in the space of visible instruments (t, g). Even under the balanced budget rule, there are tax/expenditure combinations which they both can choose. Therefore, costly signalling will prevail in the equilibrium under the balanced budget rule.

In Rogoff (1990), there is a distortion in the choice of current expenditure versus public investment. There is no debt in the model. In what follows, we briefly sketch Rogoff's argument as far as it is related to the discussion of the balanced budget constraint.

In pre-election years, under an //-type, there is systematic under-invest- ment in the public sector. Most importantly, ability does not change the pro- ducer price of public expenditures, but increases the budget by some lump- sum amount25. The budget constraint which an incumbent of type i faces is

25 Consequently, &H and &L coincide.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 20: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 555

given by

where Ak?^ is net investment in the public sector and MlT_x is the lump-sum amount which determines the budgetary capacity of a low and a high ability type, respectively.

Imagine a budget rule demanding

which does not depend on MlT_x. Of course, this is a rather artificial way to model a balanced budget rule. Ultimately it would require public investment and public expenditure to be in a predetermined relation. Recall that there is no debt in Rogoff 's model.

The game that Rogoff analyses has a similar graphical representation to our game. For ease of argument, we can assume that there is some point (t1*, gL*) which represents the policy which the L-type chooses and some other point (r77, gH) which represents the policy that the //-type chooses. Both points are located on the income expansion path *F. The restrictions from the signalling game have similar contours, so we have sets A, B, and AnB with an equivalent interpretation as in our game. Now, consider the situation with an additional constraint E which can be depicted as some ar- bitrary line in figure 1 .

First suppose that S passes through (r7'*, gL*) but does not pass through (t77, gH). Then, signalling is taking place under an additional restriction and the //-type selects the point where he can make himself discernible at low- est cost. However, any point that he can select is worse than (r77, g77), which is the best signalling point without additional constraint. Thus the policy out- come is the same under the L-type and worse under the //-type, so expected welfare has been lowered.

Now suppose that S does not even pass through (tl*, gL*). This even dis- torts the choice of the L-type. The set of possible compromises 'A for the L-type now changes to 'A' and the //-type can choose among allocations in A' to make himself discernible. However, one can show that A' is a subset of A, which means that the L-type is prepared to distort even more when he knows that his best alternative is the constitutionally constrained point in- stead of (T7-*, gL*)26.

In order to accommodate Rogoff 's and our approach, note that our results would also occur in Rogoff 's model if we were to replace the assumption that ability of political agents shows in total capacity with the assumption

26 See Rogoff (1990, p. 31).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 21: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

556 Gerald Pech

that ability shows in different productivities. On the other hand, there is no meaningful way to incorporate public debt into a model of policy-making unless we introduce distortionary taxation.

5.4. Public Debt and Public Investment

Now we analyse the case in which government instruments incorporate debt and public investment. In the full information case, efficiency demands that we equate the marginal proceeds from investment with the marginal util- ity from current expenditures. Assuming that the fixed capital can be sold off in T 27, there might be a deficit or a surplus in 7-1, depending on the marginal productivity of the investment opportunity compared to the mar- ket interest rate.

Having public debt and investment to manipulate in the asymmetric in- formation case, the government will use both instruments in a distortionary way. A given surplus of current expenditures over taxes is financed by ne- glected investment and by debt issue. In equilibrium, the marginal cost of having higher taxes in T, ET_X Ar/T(rr) is equated to the marginal return on investment. This restates that signalling will occur in such a way that a given result is reached at minimum cost. The absolute incurred cost of signalling will again depend on the political income parameter.

Now suppose that a budget rule demands that the net investment be fi- nanced by raising net government liabilities, i.e. AbT_i=AkY-' and, at the same time, q gT_x s TT_{. Now the problem of inefficient signalling vanishes and the high-ability incumbent is still discernible from low-ability. With the current budget separated, both debt and public investment are irrelevant for the purpose of signalling, so their level is determined by welfare criteria alone28. This implies that the return on investment is equated to the present cost of higher taxes. In our model, not much is changed if we prefer a less strict budgetary rule, i.e. one demanding AbT_x = Y^kr-i » 7 < 1 • Any govern- ment will invest additional tax receipts in the investment opportunity as long as it is beneficial from a welfare point of view. This is because, under such a policy regime, each voter knows that no government will levy taxes unless investing is sensible. For some governments, the welfare cost of inflexible debt policies under the rule might even be reduced. Note that the case /< 1 might encompass a pay-as-you-use finance scheme, at least if the parameter y is an exogenous parameter and not subject to political manipulation. On

27 Note that with longer time horizons, additional debt has to be restricted by net investment, see Folkers (1983, p. 37) and Buiter (1983, p. 56).

For a similar argument within the context of collective decision-making, see Buchanan (1957, p. 167).

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 22: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 557

the other hand, choosing y> 1 brings us back to our original problem. Ad- ditional invisible sources of finance will again open the race for the most pleasing current balance sheet.

Again, with the budget rule there might be some inflexibility cost which is independent of political income. This will eventually be balanced out if political income and, therefore, incentives to engage in signalling become sufficiently strong. Introducing classical rules of fiscal behaviour like the balanced budget rule or a pay-as-you-use finance scheme for public invest- ment will be preferable if signalling behaviour is serious enough a problem. Such rules disclose information at a well-determined price while signalling costs can become arbitrarily large, even if each politician has an incentive to send a given signal - i.e. always a signal that his opponent cannot send - at minimum cost.

6. Conclusion

We have developed a model of debt policy determination under asymmet- ric information which shows that, in a rational expectations equilibrium, in- centives persist for politicians to compromise on their welfare motivation and generate a budget cycle in order to get re-elected. In the game between the incumbent and the public, the incumbent's ability and the net value of the government are private information. We showed that there exists a unique separating equilibrium where the public correctly infers the incumbents' character and government's net worth because the incumbent of higher abil- ity distorts his instrument choice in order to signal his true character. A main feature of our model is that incumbents have different productivities, so in an equilibrium induced under a balanced budget rule, both characters are always discernible. This is why we found a balanced budget rule welfare enhancing in our game. This contrasts with results from a similar game an- alysed by Rogoff, where politicians' total capacities diverge but productiv- ities are identical. In such a setting, incentives to signal persist under the bal- anced budget rule and signalling costs even increase. In our setting with diverging productivities, no such problem occurs.

Therefore, we propose that imposing a balanced budget rule supplemented by a pay-as-you-use oriented financing scheme which does not yield to political manipulation is eventually welfare enhancing. Imposing a rule is welfare enhancing if the incentives to engage in signalling are sufficiently strong, i.e. if the political income is sufficiently large. Note that any rule which binds equilibria away from signalling points with lower welfare will eventually become beneficial, if signalling becomes a sufficiently urgent problem. What we do not address in our paper is the issue of which rule to

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 23: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

558 Gerald Pech

apply if there are many rules to choose from. We leave this question for fu- ture research.

Appendix

Appendix la

Proof of Lemma la: From the envelop theorem, we have E Vl(b' eH) = q(£HTl vg[gT{bT_x, q(£H)]}. gr(.) [rr(.)] is an implicit function in bT_x and q(eT) where gT(.) [[rr(.)]] is monotonically decreasing [increasing] in bT_x and q(£T). Furthermore, we can show that gT-'(>) [?t-i] is monotonically in- creasing [decreasing] in bT_x and decreasing [increasing] in q{£T-')- Then from (6) and (7) we immediately obtain proposition la.

Proof of Lemma lb: Analogous to proof of proposition la.D

Proof of Lemma lc: A type L is followed by a politician who is drawn at random from the set of challengers, so that the follower is a type H with probability P and a type L with probability (1-P). Then (6) becomes q(eLr'(gT-i) = EVÏ(b; ë) where

EV¡(b; ë) = (Ì-P) q(eLrlvg{gT[l>T-i, q(eL)]ì + Pq(£HTlvg{gT[bT_x,q(eH)]}.

Initially we examine the extreme cases where P is either 0 or 1 . For P = 0, we have the result from proposition lb. For P = 1, that is, if the representa- tive is followed by a politician of high ability, he runs a deficit if at bT_x = 0

(A.I) q(eLrlvg{gT_l[O,q(eL)])>q(eHrl vg{gT[0, q(eH)]} and the elasticity of demand for public goods is smaller than 1 , as shown in appendix lb.

Next consider the case where 0<P< 1. At bT_x =0 we have:

(1 -P) q{eLTlvg{gT[^ q(EL)]} +Pq(eHrl vg{gT[0, q(eH)]}

<q(eLrlvg{gT-X[0,q(eL)]}- Again, with inflexible expenditures, we have a deficit and g^L'<gT*' and t£i>t£Î.D

Appendix lb

Here we show that intertemporal efficiency in the case of lemma lc de- mands a deficit if the absolute value of the elasticity of demand for public

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 24: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 559

goods is less than one. Suppose there is a small perturbation in the price of the public good at q°, so qH=q°-i and qL = q°+i. Then rewriting (A.I) using (6) and (7) and allowing for the equality sign yields

(A.2) ?iT_jr(GT.i-b) = ^vg(gT_O>^iVg(gT) = XTfr(GT + b).

With 1=0, (A.2) is fulfilled as an equality at b=0. With i>0, b > 0 solves (A.2) at Xt_x = A^, if dGT_x/di > 0 and dGj/di < 0. From dGT_x/di = g°+q° áglàq > 0 this is the case when 1 > -q°/g° dg/dq.

Appendix 2

In this appendix we show that the optimum is the tangential point (t77, gH). This point is situated as indicated in figure 1 in the north-east of a hypothet- ical point C" which is situated at the intersection of &H and the contour of 'A and, therefore, characterised by the absence of an intratemporal distor- tion. We show that in C a variation of taxes and government expenditures in accordance with the budget constraint which is directed to the north-east in figure 1 improves the welfare of type H. Partially differentiating the im- plicit form objective function ZT~l, we derive

^T~l^£-lX'*- = -**-iM*T-i)-V?(l, e>) and

dZT~^T'l'£Í)'-r->- =M*--l> + *<«W0. ¿>-

Type H is restricted to select a point in 'A. Observing that the political in- come accruing to H is exogenous to this problem, once H has decided to sig- nal his true type, the Lagrangean becomes:

L = ZT-'gT_Xi rr_!, 1, e?_0 +<p[Z7M(s£1, t£i, 0, e^)

-ZT-'gT_Xi rr_!, l, e^)- [PO)-P(O)] yTl

(p > 0 is the shadow price of a reduction in expected political income. In- specting (A.3) and (A.4) shows that d(p/dy>0. The first-order conditions with respect to gT_x and rT_{ are:

(A.3) v^r.O + q(eH) V¡(',eH) = tp[vg{gT,x) + q(eL) VTh{',eL)l

(A.4) Vi/rfei) + Vl(h£H) = plViWti) + Vl(h£L)l

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 25: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

560 Gerald Pech

and the Kuhn-Tucker Conditions

(A.5) Z^teKi, T&, 0, e£ü - ZT-'gT_u rT_u 1, e^) -[P(l)-P(0)]yr2>0

<p>0

^Z7"1^, t£i. 0, e£x) - ZT-'gT-x, tr-i> L ̂ r-i) -[P(l)-P(0)]yr] = 0

In C" we have (eH) A^ fx{tT-') = vg(gT-') and (A-5) with equality. In- spection of (A.3) and (A.5) shows that they are not simultaneously satisfied on 0H. Therefore, C" is not an optimum. A variation in public expenditures Õ8 and in taxes 5* in accordance with (A.5) has to satisfy:

(A 6) 5S < h-iMiT-ù + Vb 8r-vg(gT-i) + q(eL)VbT'

In C' we have a deficit, so V^(l, eL) <- q(eL)~l vg. Then both numerator and denominator are negative in (A.6) and we find that in C":

õ*]c<q(eH)' In C a change in taxes dr in accordance with the constraint increases the value of the objective function:

[-^T-if^T-i)-yh^eH))àT

+ [ vÄ (ír-i) + 9(^)^(1,^)]!^ I cdT>0.

The first implication of this is that in C a move towards the north-east increases H 's objective function. In order to see the second implication, note that the restriction is also an indifference curve - prescribed to type L - and a tangential point is always a local extreme point. If, furthermore, the pref- erence ordering is complete and indifference curves are continous, convex and cross at most twice in {g+,T+} - defined as the set of all points which H strictly prefers to C - then such a tangential point in the direction of the move must exist and is also the unique maximum.

References

Alesina, A., 1989, Comments and Discussion on "Alternative Approaches to the Politi- cal Business Cycle" by W. D. Nordhaus, Brooking Papers on Economic Activity, 2, pp. 50-56.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 26: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

Public Debt and Classical Rules in Public Finance 561

Alesina, A., G. D. Cohen and N. Roubini, Macroeconomic Policy and Elections in OECD Democracies, Economics and Politics, 8, pp. 55-98.

Alesina, A., and H. Rosenthal, 1995, Partisan Politics, Divided Government, and the Econ- omy, Cambridge.

Barro, R., 19/3, ine Control ot Politicians - An bconomic Model, Public Choice, 14, pp. 19-42.

-, 1979, On the Determination of Public Debt, Journal of Political Economy, 87, pp. 940-947.

Bizer,D. S.,andS. N.Durlauf, 1990, Testing the Positi ve Theory of Government Finance, Journal of Monetary Economics, 26, pp. 123-141.

Bohn, H., 1986, Financial Markets and the Macroeconomy, Ph.D. Thesis, Stanford Uni- versity.

Buchanan, J., 1957, Public Principles ot Public Debt, Homewood. Buiter, W., 1983, The Theory of Optimum Deficits and Debt, Federal Reserve Bank of

Boston Conference Series, 27, pp. 46-69. Cho, I.-K., and D. M. Kreps, 1987, Signalling Games and Stable Equilibria, Quarterly

Journal of Economics, 102, pp. 179-221. Coates, S., and S. Morris, 1995, On the Form of Transfers to Special Interests, Journal of

Political Economy, 103, pp. 1210-1235. Eisner, R., 1984, Which Budget Deficit? Some Issues of Measurement and their Implica-

tions, American Economic Review, 74, pp. 138-143. Fair, R. C, 1978, The Effect of Economic Events on Votes for President, Review of

Economics and Statistics, 60, pp. 159-173. Ferejohn, 1986, Incumbent Performance and Electoral Control, Public Choice, 50,

pp. 5-25. Folkers, C, 1983, Begrenzungen von Steuern und Staats ausgaben in den USA. Eine

Untersuchung über Formen, Ursachen und Wirkungen vorgeschlagener und realisierter fiskalischer Restriktionen, Baden-Baden.

Greenberg, J. H., 1990, The theory of social situations. An alternative game theoretic approach, Cambridge.

Harrington, J. E., 1993, Economic Policy, Economic Performance, and Elections, Amer- ican Economic Review, 83, pp. 27-42.

Hess, G., 1991, Voting and the Intertemporal Selection of Tax Rates in a Macroeconomy, Economics and Politics, 3, dd. 41-62.

-, 1992, Are Tax Rates too Volatile?, Southern Economic Journal, 60, pp. 72-88. Inman, R. P., 1996, Do Balanced Budget Rules Work? U. S. Experience and Possible Les-

sons for the EMU, NBER Working Paper 5838, Cambridge/Mass. Kiewiet, D. R., and D. Rivers, 1985, A Retrospective on Retrospective Voting, in:

H. Eulau and M. S. Lewis-Beck (eds.), Economic Conditions and Electoral Outcomes - The United States and Western Europe, New York.

Kydland, F. E., and E. C. Prescott, 1980: Dynamic Optimal Taxation, Rational Expec- tations, and Optimal Control, Journal of Economic Dynamics and Control, 2, 1980, pp. 79-91.

Nordhaus, W. D., 1989, Alternative Approaches to the Political Business Cycle, Brook- ing Papers on Economic Activity, 2, pp. 1-68.

Pech, G., 1996, Besteuerung und Staatsverschuldung in der Demokratie, Frankfurt/M. Rogoff, K., 1990, Equilibrium Political Budget Cycles, American Economic Review, 80,

pp. 21-36. Rogo it, K., and A. Albert, i^öö, ülections and Macroeconomic Policy Cycles, Review or

Economic Studies, 55, pp. 1-16. Tufte, E. R., 1978, Political Control of the Economy, Princeton. Wittman, D. A., 1995, The Myth of Democratic Failure. Why Political Institutions are

Efficient, Chicago.

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions

Page 27: Public Debt and Classical Rules in Public Finance: Some Lessons from the Equilibrium Budget Cycle Approach

562 Gerald Pech

Abstract

In this paper, we prove the existence of a separating equilibrium in a signalling game with public expenditures, distortionary taxation and public debt where voters cannot ob- serve the ability of politicians or the true size of public debt. We show that policies under a balanced budget rule are information-revealing if ability shows in a difference in pro- ductivity in the supply of public goods. The policy under the budget rule dominates the policy in the signalling equilibrium from a welfare point of view if incentives to engage in signalling are sufficiently strong. In the case of a public investment opportunity, the balanced budget rule for current expenditures can be supplemented by a rule stating that public investments be completely or partially debt financed. Therefore, a rule like the pay-as-you-use principal is recommendable under the aspect of coping with incentives to use inefficient signalling methods, if the same principle goes along with a balanced bud- get rule for current expenditures.

Kurzfassung

In dieser Arbeit wird die Existenz eines separierenden Gleichgewichts in einem Signalspiel mit öffentlichen Ausgaben, verzerrenden Steuern und Staatsverschuldung ge- zeigt, in dem Wähler weder die Fähigkeit der Politiker noch die wahre Höhe der Staats- verschuldung direkt beobachten können. Wenn sich unterschiedliche Fähigkeiten in un- terschiedlichen Produktivitäten bei der Bereitstellung öffentlicher Güter niederschlagen, offenbart die Politik unter einer Regel des materiellen Haushaltsausgleichs die verbor- gene Information. Unter einem Wohlfahrtsaspekt dominiert die Politik das Gleichgewicht des Signalspiels, wenn die Anreize zu signalisieren hinreichend stark sind. Im Falle einer Investitionsalternative in einem öffentlichen Projekt kann die Haushaltsausgleichs- regel auf den öffentlichen Konsum beschränkt und durch eine Regel ergänzt werden, nach der öffentliche Investitionen auch durch Verschuldung finanziert werden können. Die An- wendung einer Regel nach dem Pay-as-you-use-Prinzip ist daher unter dem Aspekt, An- reizen zu ineffizientem Signalisierungsverhalten zu begegnen, empfehlenswert, wenn ebendieses Prinzip mit einer materiellen Haushaltsausgleichsregel für staatlichen Kon- sum einhergeht.

Dr. Gerald Pech Lehrstuhl für Finanzwissenschaft Fakultät für Wirtschaftswissenschaft Ruhr-Universität Bochum D-44780 Bochum E-Mail: [email protected]

This content downloaded from 195.34.79.174 on Mon, 16 Jun 2014 07:06:07 AMAll use subject to JSTOR Terms and Conditions