Introduction

1. On the role of politics in public economics: Political Economy and Public Choice TheoryIntroduction to Public EconomicsULBProf. A. Estache

2. Introduction • So far, we have assumed that governments are benevolent and simply try to maximize some social welfare function to achieve some combination of efficiency and equity • In the real world, not always easy for the gvt to simply focus on maximizing welfare • Politicians have considerations other than • getting to the socially efficient level or • conducting cost-benefit analyses in order to approve a project. • Instead, such economic decisions are made in the context of a political system. • => unrealistic to assume like most of the old fashion public economics research did, that governments are benevolent • This lecture is about NON-Benevolent governments or at least not totallybenevolent governments…

3. Do you remember this slide from the 1st lecture?Why study Public Economics? The ultimate goal is to: Understand the role of the State in the economy ..which boils down to answering 4 basic questions: WHEN should the State intervene in the economy’? HOW should it INTERVENE? HOW to ASSESS the economic effects of the various possible interventions? WHY do states intervene the way they do?... that’s what we want to discuss now!

4. So what is the exact focus of this lecture? • This lecture thus focuses on the 4th question of public finance: • Why do governments do specifically what they do? • This allows us to switch from a normative to a positive theory of the public sector • What gvt should do vs. what gvt really does!

5. Once more a reminder… So far, we have learned • How to determine the optimal level of a public good or a private good with an externality • Just set the social marginal costs = SM benefits • Thanks to CBA, we know now how to quantify these costs and benefits • …but this is what economists do…and that does not lead to decisions… • Politicians, other technical experts contribute to, and often LEAD, the decisions • This influences the outcome of the decisions…and this is why the rest of this lecture is to discuss how the decisions are taken in practice

6. Quick overview of what we will see • We start by discussing the “best case scenario” • i.e. when gvts appropriately measure and aggregate the preferences of citizens in deciding what projects to undertake. (LINDAHL pricing) • We will then discuss the problems with these idealized preference aggregation models and turn to more realistic decision making processes • To do so, we examine both direct democracy and representative democracy. • i.e. when government approximates the preferences of its citizens in deciding what projects to undertake • Finally, we examine why government failures to aggregate preferences well matter so much to the efficiency and equity of the economy… • i.e. why gvts don’t always have the ability or willingness to appropriately address market failures.

7. A useful benchmark:What if all citizens agreed on the optimal level of a public good to be provided? => Lindahl Pricing • Ideally, the government could provide public goods through unanimous consent of its citizens. • This can be done by Lindhal pricing • Lindahl pricing is a system where • (1) individuals report their willingness-to-pay for each new unit of the public good (PG), and • (2) the government aggregates preferences to form a measure of the social benefit (MSB) of that next unit of the PG • (3) That MSB is then compared to the marginal social cost (MSC) of the extra unit of PG to get the optimum amount of the PG • (4) That PG can then be financed by charging individuals what they were willing to pay

8. A somewhat more precise discussion of Lindahl Pricing • To illustrate Lindahl’s procedure, imagine that the public good in question is fireworks for two people (Ava and Jack). 1. First, the government announces a set of tax prices for the public good, that is, the share of the cost that each individual must bear. • i.e. Gvt announces that Ava and Jack each pay 50cts or that Ava pays 90cts and Jack pays 10cts 2. Each individual announces how much of the PG he or she is willing to buy at those tax prices • If the individual announcements differ, the government raises the tax price for the person who wants more of the good, and lowers it for the person who wants less. 3. The gvt repeats the sets of tax announcements to construct a marginal willingness to pay schedule (i.e. a demand curve) for each individual showing the relationship between WTP and the quantity of the PG desired 4. The gvt adds up individual WPT at each quantity of PG provided to get the overall demand curve for the PG (DA+J) 5. The gvt relates this overall D curve to the MC curve for the PG to solve for the optimal PG quantity • When a tax price is arrived at where both individuals want the same amount of the public good, the government has reached Lindahl equilibrium. 6. The government produces the public good at that level, and finances it by charging each person their tax price

9. Lindahl pricing Figure 1 Willingness to pay $4 $3 DJACK SMB=DAVA+JACK $2 S=SMC $1 $0.75 DAVA $0.25 0 25 50 75 100 Fireworks

10. Lindahl pricing Figure 1 Willingness to pay Adding up their individual demands curves vertically gives the aggregate demand. Jack’s surplus is this area $4 While Ava’s demand curve for fireworks is fairly flat. Jack’s demand curve for fireworks is fairly steep. $3 DJACK SMB=DAVA+JACK This generates a tax price of 75¢ for Jack in Lindahl equilibrium. The socially efficient level is 75 units. And a price of 25¢ for Ava in Lindahl equilibrium. With this marginal cost, it is socially beneficial to produce the good. Ava’s surplus is this area $2 S=SMC $1 $0.75 DAVA $0.25 0 25 50 75 100 Fireworks

11. Why is a production of 75 units a Lindahl Pricing equilibrium? • Fireworks production of 75 units is an equilibrium for two reasons. • both Ava and Jack are happy to pay those tax prices to get that quantity. • the government has covered the marginal cost (MC) of producing the fireworks by charging each person their marginal willingness-to-pay.

12. What does Lindahl Pricing Imply? • Lindahl pricing corresponds to the concept of benefit taxation, which occurs when individuals are being taxed for a public good (PG) according to their valuation of the benefit they receive. • Think of why user fees are so popular! • With Lindahl pricing, the gvt does not need to know the utility functions of individual voters: it gets the voters to reveal their preferences by stating their willingness to pay for different levels of the PG. • NOTE: In the lecture on PG, we max U to obtain each person demand for PG and adding up we got the total social demand …but to do that, we needed to know the preferences of individual voters…so, here, we can get the same outcome with less information! • The equilibrium achieved is also the efficient level of public goods provision

13. Problems with Lindahl Pricing • HMMMM…bad news: • Lindahl pricing is unlikely to work in practice. • Preference revelation problem: Individuals may behave strategically, and pretend their willingness to pay is low in order to get others to bear a larger cost of the public good. • Freer rider problem again! • Preference knowledge problem: It is hard for people to properly value goods they do not shop for on a regular basis. • Preference aggregation problem:Even if people reveal their preferences honestly and know their valuation, aggregating millions of voters’ preferences is difficult in reality.

14. If not Lindahl Pricing, what else??? • Since Lindahl pricing is unlikely to work in practice • …We need more practical solutions to determine the optimum level of PG to provide…would VOTING be a solution? • In that context, 2 key questions: • How can society effectively use voting mechanisms to aggregate individual preferences • How well do the elected representatives carry out the preferences of individual voters?

15. Mechanisms for aggregating individual preferences • Let’s focus on how voting mechanisms can serve to aggregate individual preferences to reach a social decision. • i.e. to decide how much of a PG to produce • We will consider: • Majority voting(direct democracy) • Representative democracy(indirect democracy) • First, let’s focus on direct democracy • i.e. voters directly cast ballots in favor of or in opposition to particular public projects.

16. Direct democracy in the US? Application • 3 ways in which voters participate in direct democracy are through referendum and voter initiative. • A recall election is a special election initiated by citizens with the goal of replacing a sitting official. • Areferendum is a measure placed on the ballot by the government allowing citizens to vote on state laws or constitutional amendments that have already been passed by the state legislature. • A voter initiative is the placement of legislation on the ballot by citizens.

17. Direct democracy in Switzerland? • What are the payoffs of direct democracy for the Swiss? • In Switzerland, new government spending programs must be approved by a referendum of citizens in many Swiss cantons. • This decisionmaking procedure seems like a simple way to address citizen-legislator agency problems • Feld and Matsusaka (2001) estimate spending regressions for Swiss cantons using panel data from 1986 to 1997. • (After controlling for demographics and other determinants of spending)mandatory referendums on new spending are found to reduce the size of the budget by 17% for the median canton.

18. What are the benefits of direct democracy?Look at Switzerland again… Bruno Frey showed a relationship between direct democracy and greater happiness: where there aremore direct democratic initiatives and referendums, residents are happier! (….Switzerland is split into 26 cantons and relatively weak central government. Cantons are powerful but have different forms of democracy).

19. Majority Voting: When It Works (1) • It reduces the very high demands of the Lindahl pricing scheme to get a consensus: • only when citizens were unanimously in agreement did the government achieve equilibriumunder Lindhal pricing • Under majority voting, individual policy options are put to a vote, and the option that receives the majority of votes is chosen. • => Majority voting seems VERY attractive • BUT can majority voting ACTUALLY lead to better outcomes than unanimity?

20. Majority Voting: When It Works (2) • ….well…Majority voting does not always provide a consistent means of aggregating preferences…this is why! • To be consistent, an aggregation mechanism must satisfy 3 goals: • Dominance:If one choice is preferred by all voters, then the aggregation mechanism much be such that this choice is made by society. • Transitivity:Choices must satisfy this mathematical property. • Independence of Irrelevant Alternatives:The introduction of a third choice does not change the ranking of the first two choices.

21. Majority Voting: When It Works (3) • It turns out that with these 3 conditions, majority voting can only produce a consistent aggregation of individual preferences if preferences are restricted to take a certain form. • This is best illustrated with examples. • Consider a community deciding between alternatives to generate additional school funding

22. Majority voting when it works(4) • Schools are an impure PG • Assume they are mostly financed by property taxes => more funding requires more taxes • So community needs to chose between • H: high level of funding • M: medium level • L: low level • Imagine 3 types of voters • Parents: who prefer the following ranking for funding: H, M, L • Elders (no kids=> don’t care about schools): L, M, H • Young couple with no kids: M,L,H • Table 1 illustrates the case.

23. Table 1a

24. A town is deciding on education taxes (and spending). There are 3 possibilities: high, medium, and low spending. There are also 3 groups, represented in equal proportions. Table 1b Consider pair-wise voting: High vs Low, High vs Medium, and Medium vs Low. High vs Low: Parents vote for H, Elderly & Young vote for L. L wins 2-1. High vs Medium: Parents vote for H, Elderly & Young vote for M. M wins 2-1. Medium vs Low: Parents and Young vote for M, Elderly vote for L. M wins 2-1. Since M has beaten both H and L, M is the overall winner in this case. Finally, the “young couples” do not have kids and do not want to pay high taxes right now. Their preferences are for medium spending, then low, then high. While the preferences of the elderly are exactly opposite. The preferences of parents are for high spending, then medium spending, then low spending. In this example, the option chosen by majority voting will be the Medium level of funding, the choice of the median voter (the young couple!)

25. Majority Voting: When It Doesn’t Work (1) Table 2 shows a different scenario, where majority voting does not work. It deals with a case in which private parents who want to send their kids to private schools replace the elderly and have a different preference than public parents who are happy to send their kids to public school

26. Table 2a

27. A town is again deciding on education taxes (and spending). The elderly have been replaced with “private parents.” The other 2 groups are the same as before. Private parents, first and foremost, want low taxes so they can afford to sent their kids to private schools. Assuming that doesn’t happen, however, private parents want high quality public education. Thus, their ordering is low, then high, then medium. Table 2b Consider pair-wise voting: High vs Low, High vs Medium, and Medium vs Low. High vs Low: Only “public parents” vote for H, L wins 2-1. High vs Medium: Only Young Marrieds vote for M, so H wins 2-1. Medium vs Low: Only private parents for L, so M wins 2-1. Hmmm … There is no clear winner. L is preferred to H. H is preferred to M. M is preferred to L. This violates the transitivity assumption and leads to cycling. In this example, there is no consistent outcome from majority voting

28. Majority Voting: When It Doesn’t Work (2) • This set of outcomes is problematic because there is no clear winner and these results violate the principle of transitivity (i.e. resulting in cycling) • Dfn of cycling? when majority voting does not deliver a consistent aggregation of individual preferences. • Note that the failure to get a consistent winner from majority voting does not reflect a failure on the part of individuals–each group has a sensible set of preferences. • The problem is aggregation–we are unable to use voting to aggregate these individual preferences into a consistent social outcome.

29. Majority Voting: When It Doesn’t Work (3) • Our example shows the risks of having an the agenda setter, • Dfn? the person who decides the sequencing of the votes. • In the second situation, the new players can affect the outcome. • For low spending to win, for example, first set up a vote between H and M. H wins. Then a vote between L and H means L will win. • Any outcome can win with appropriate ordering.

30. Arrow’s Impossibility Theorem • In fact, there is no voting system that will produce a consistent outcome • Arrow’s Impossibility Theoremstates that: • there is no social decision (voting) rule • that converts individual preferences into a consistent aggregate function • without either restricting preferences • or imposing a dictatorship.

31. Restricting Preferences to Solve the Impossibility Theorem (1) • One way to solve this problem is to restrict preferences to “single-peaked” preferences. • A “peak” in preferences is a point that is preferred to all its immediate neighbors. Utility falls in any direction away from this point. • Multi-peaked preferences means that utility may first rise, then fall, then rise again. • If preferences are single peaked, majority voting will yield a consistent outcome. • We can visualize our earlier examples. See Figure 2.

32. Voting rules Figure 2 (a) (b) Utility Utility Young marrieds Young marrieds Private parents Public parents Parents Elders School spending School spending H L M H L M

33. Voting rules Figure 2 (a) (b) Utility Utility Young marrieds Young marrieds Private parents Public parents Parents Elders Private parents are different in the second case. Their utility goes in in either direction from M. And young marrieds are single peaked at “M”. The elderly are single peaked at “L”. Parents are single peaked at “H”. School spending School spending H L M H L M

34. Restricting Preferences to Solve the Impossibility Theorem (2) • The failure of these preferences for the “private parents” in this second case is what leads to the inability of majority voting to consistently aggregate preferences. • Fortunately, single-peakedness is a reasonable assumption for most PG • => good news for majority voting!!!! • Works much less well for private goods or when private substitutes exist for PGs

35. Median Voter Theory • When preferences are single-peaked, then majority voting will deliver a consistent aggregation of preferences of the individual voters. • Under the single-peaked preferences assumption, we can make an even stronger statement about the outcome of majority voting across PG options: look at the median welfare theorem for a useful equivalence result • The median voter theorem states that majority voting will yield the outcome preferred by the median voter if preferences are single peaked. • Dfn: The median voter is the voter whose tastes are in the middle of the set of voters, so an equal number of other voters prefer more and prefer less of the public good. • In both examples, the median voters are the young couples as they always have one group of voters on each side of their position

36. The Potential Inefficiency of the Median Voter Outcome (1) • Thus, the median voter theorem suggests that the gvt only needs to find the preferences of the median voter, and implement that level of public goods!!!! • BUT…This does not account for intensity of preferences, => so social marginal benefits are NOT equal the social marginal cost. • For example, if small numbers of individuals derive enormous benefits from the public good, this should be accounted for in the total social marginal benefits.

37. Potential Inefficiency of the Median Voter Outcome (2) • For example, imagine that there are 1,001 voters in a town, who are considering building a monument that costs $40,040 ($40/person). • Assume all 1,001 voters have single-peaked preferences, so the median voter will determine the outcome. • If 500 citizens value the monument at $100 each, and the other 501 value it at $0, then the social marginal benefit is $50,000, far greater than the cost. However, the monument doesn’t get built. • => inefficient outcome because the median voter outcome does not reflect the intensity of preferences • For instance, that many voters were willing to pay >$40 is irrelevant in this model… all that matters is that the pivotal median voter is not willing to pay $40!

38. Summary on Direct democracy • Many decisions in direct democracies are made by majority voting • It turns out that there are many situations in which majority voting will not lead to an efficient outcome!!! • Even if preferences are single peaked! • Indeed, if preferences are single peaked, majority voting will consistently aggregate preferences • Moreover, then, the outcome is the one chosen by the median voter • HOWEVER that outcome will not always be the efficient outcome! • It may simply reflect the intensity of preferences of minorities, not the voice of the majority!!!!

39. NOW: REPRESENTATIVE DEMOCRACY:Vote-Maximizing Politicians • The alternative to direct democracy is representative democracy • In most countries, voters elect representatives • These are supposed to aggregate the voters preferences and take them into account when they vote on the appropriate level of public goods in Congress or in the Senate • But can we model how politicians behave in a representative democracy???? • The most common theory of this behavior is a version of the median voter theory discussed earlier for direct democracy: • If politicians care about maximizing the number of votes they get, they choose the outcome preferred by the median voter. • In this section, we review this theory for the case of representative democracies, discussing the assumptions and reviewing the evidence

40. Intuition • The median voter theory in the representative democracy context rests on a single key assumption: • All politicians care about is maximizing the number of votes they get! • If this is true, THEN elected politicians will choose the outcome preferred by the median voter (if preferences are single peaked) • => Downs (1957) shows that with vote maximizing politicians, the theory we used to explain direct democracy can also be used to explain representative democracy!

41. Vote-Maximizing Politicians: an example Consider the case an election in which two politicians, Elio and Bart, are running for office and vying to maximize their votes. Eliowants to appeal to those who don’t want to spend on defense, Bart wants the votes of those who want to spend a lot on defense So the key question is what fraction of the budget (0% to 50%) should be spent on defense? Voters are uniformly distributed on this continuum. Figure 3 illustrates how this plays out.

42. Figure 3 a Voters for Elio Voters for Bart (a) Defense spending E1 B1 50% 0% 25% Voters for Bart Voters for Elio (b) Defense spending E2 B1 50% 0% 25% Voters for Elio Voters for Bart (c) Defense spending E2 B2 50% 0% 25% Voters for Elio Voters for Bart (d) Defense spending 50% 0% E3= B3 = 25%

43. Figure 3 b In this case, the candidates split the vote equally. Elio choose E1 while Bart chooses a much higher level of defense, B1. Voters for Elio Voters for Bart (a) Defense spending By doing so he gets the majority of the votes. But now imagine that Elio changes his position to E2. E1 B1 50% 0% 25% Voters for Bart Voters for Elio (b) Defense spending E2 B1 50% 0% 25% In response, Bart lowers his position to B2. Voters for Elio Voters for Bart (c) Defense spending This process will continue until the median voter’s preferences are arrived at. E2 B2 50% 0% 25% By doing so, Bart now gets a majority of the votes. Voters for Elio Voters for Bart (d) Defense spending 50% 0% E3= B3 = 25%

44. Conclusion on the example • If the politicians are indeed purely voters maximizers, the changes in political view will continue until both candidates support the median voter (25% of budget on defense) • Any candidate with a position different to that of the median voter…will lose votes • ..so all politicians need to know under a representative majority regime is the position of the median voter since the median voter model is a model of demand aggregation under majority rule. • Thinks of it this way: • Economists find the market demand for private goods by horizontally summing the demand curves of all individuals in that market; • For Political scientists, similarly, individual demands can be aggregated through majority rule voting and in that case, the demand of the entire group is the demand of the median voter.

45. Who is the medium voter in Europe on Social Security Issues? • Paper by Koethenbuerger, Poutvaara and Profeta (Oxford Economics Papers, 2007) • Using data on eight Europeancountries, they find that the median voter is typically middle-agedwith high income. • For these voters, earnings-related social security systemsare more attractive than flat-rate systems • This is because the earnings related system generates • Less intragenerational redistribution • Lower distortions in labor supply. • In this context, the median voter modelis also able to account for a considerable degree of cross-countryvariation in contribution rates.

46. ..but not all empirical studies find the median voter model useful: …look at tax policy • Consider the income distribution… a key policy issue…but complex…(based on Alesina and Glaeser: Fighting Poverty in the US and in Europe) • Assume that the median voter is poorer than the average voter (very realistic throughout the world) • => Since the median voter is poorer than the voter paying the average tax, the median voter will favor progressive taxation • In fact the higher the support for high taxes and high transfers from the richer to the poor • There should thus be more taxation and redistribution in Europe if pre-tax inequality is higher than in the US • However, in Europe pre-tax inequality is lower than in the US • …yet taxation is more progressive! • So what’s wrong with the median voter story here? • The poor may not have enough political voice… • it is not 1 person= 1 vote but 1 $= 1 vote! • It ignores the fact that the median voter may have a dynamic vision of the world (hoping for upward mobility!)..and this is truer in the US than in EU! • (Often in Europe it is proportionality voting that prevails, not majority voting)

47. To understand its limits, let’s look at the assumptions of the Median Voter Model • Why do we find situations in which politicians move away from the position of the median voter? • The problems with the median voter model come from its assumptions: • Single-dimensional voting: Voters only care about one issue. • In reality, politicians are elected on a bundle of issues and voters may lie a polar positions on different issues…so would only work is all issues are highly correlated • Only two candidates: With a 3rd candidate, there is no stable equilibrium since every candidate move will impact an election • No ideology or influence:Assumes politicians only care about votes, not ideological positions…not true! • No selective voting: All citizens actually vote….not always true! • No money as a tool of influence: ignores crucial role of campaign financing. • Perfect information along three dimensions: voter knowledge of the issues, politician knowledge of the issues, and politician knowledge of voter preferences...clearly unrealistic!

48. So…Evidence on the Median Voter Model… • While the median voter model is a potentially powerful tool, the evidence on its predictive power is mixed • It helps explains the broad picture • Where the voter is more conservative, politicians will take conservative decisions • But…it certainly does not completely explain legislator behavior. • Not totally surprising in view of the assumptions it builds on… • strategic voting, agenda manipulation, cyclical majorities, abstention of voters (e.g. alienation) and a strong influence of interest groups occur in decision making processes (which is very likely) could all “kill” the median outcome . • In particular, there is strong evidence that legislators consider their own ideology, and that of their core constituency, when they vote on policies.

49. Cycling in Representative Democracies The CYCLING issue that arises in direct democracy can also arise in representative democracy if legislator preferences are not single peaked.

50. The Political Business Cycle (1) • Another issue that arises in representative democracy is the “political business cycle.” Pocketbook issues matter. • Bill Clinton’s mantra – “It’s the economy, stupid!” • Some models relate the fraction of the vote for the incumbent to the economy’s growth rate and inflation rate. • On average, predicted share of the vote for the incumbent party is within 2.6 percentage points of the actual vote received. • For the US, the models have done a pretty good job at predicting winners in presidential elections