Intro Prefs & Voting Electoral comp. Voter Turnout Agency GIP. Political Economics. Ludwig-Maximilians University Munich. Summer term / 59

Transcription

1 1 / 59 Political Economics Ludwig-Maximilians University Munich Summer term 2010

2 4 / 59 Table of contents 1 Introduction(MG) 2 Preferences and voting (MG) 3 Voter turnout (MG) 4 Electoral competition (SÜ) 5 Agency problems (SÜ) 6 General-interest politics (SÜ) 7 Special-interest politics (MG) 8 Applications

3 1 / General-interest politics

4 2 / 59 Efficiency explanation of social security Safety net for the unlucky -> social insurance Altruism Egoism Why political economics? Efficiency explanations do not explain the increase in social spending. They rely on normative social planner approaches while decisions in democracies are made by voting. A positive rather than normative approach is therefore useful.

5 3 / One-dimens. heterogeneity (income/productivity): Consider the following one-period model where as before, variables without a superscript denote average variables: Static economy producing a single commodity Individuals work and consume and vote on a redistributive programm (proportional tax and lump-sum transfers) The preferences of the i th individual are w i = c i + V (x i ), (73) with c: cons., x: leisure, V (.) well-behaved, concave NOTE: Quasi-linear preferences -> consumption absorbs all income effects

6 4 / 59 The budget constraint is given by c i = (1 t)l i + f (74) with t: income tax rate, l: individual labor supply, f : lump-sum transfer and real wage normalized to unity Continuum of individuals differing in one dimension (productivity): being more productive is equivalent to having more effective time. The individual time constraint thus is with e i : individual productivity 1 + e i = x i + l i (75)

7 5 / 59 If we solve the problem for the optimal labor supply with (73), (74), (75), we find l i = L(t) + (e i e) (76) with L(t) = 1 + e Vx 1 (1 t) with L(t) t = 1/V xx (x i ) < 0 NOTE: The tax is distortive and the distorting effect on the labor-leisure decision is the same for all individuals irrespective of their productivity. The government budget constraint then is f = tl = tl(t). (77) Electoral setting: Policy is set in a median-voter framework where two political candidates compete for office.

8 Equilibrium redistribution Taking all constraints from above into account, voter i s indirect utility function is given by W i (t) = (1 t)(l(t)+(e i e))+tl(t)+v (1 L(t)+e) (78) The tax rate t i preferred by voter i is implicitly given by W i t (t) = (e i e) + tl t (t) = 0 (79) taking into account that dli dt = 0 by the envelop theorem. This conditions captures first term: marginal benefit of higher tax rates cum redistribution (positive for poorer voters (e i e) < 0, negative for richer ones (e i e) > 0) second term: marginal cost due to higher distortions and a smaller tax base (negative as L t (t) < 0) 6 / 59

9 7 / 59 This implicitly defines the preferred tax rate of i t i = (e i e)/(l t (t i )) (80) Thus, a poorer voter prefers t > 0 which is the larger the poorer he is whereas a richer voter prefers an income subsidy (t < 0) financed by lump-sum taxes. Individual preferences are thus monotonic in e i. Formally, (78) satisfies the single-crossing condition which ensures that a Condorcet winner exists: there is only one political equilibrium both candidates commit to the policy preferred by the median voter t m : t m = (e m e)/(l t (t m )) (81)

10 8 / 59 Evaluation The model thus predicts that the size of general redistribution reflects preferences of the median voter, that the larger is the deadweight loss as captured by the absolute value of L t (t), the smaller is equilibrium redistribution, and, in particular, that taxes are higher the higher inequality as measured by the distance between median and mean income. It follows that concentration of more income at the top makes redistribution more attractive for the median voter more extreme poverty, however, has the opposite effect

11 9 / 59 Can this simple model explain the evolution of the size of redistributive programmes? Early growth - yes: 1 The extension of the suffrage to poorer voters affecting the median voter s relative income 2 Economic progress and institutional change reducing the transaction costs of collecting taxes and hence the distortions But what about the continuing increase after the late 1960s and 1970s - even though income inequality declined? And what about the 1980s and 1990s when inequality turned again upwards, whereas transfers rose less quickly?

12 10 / 59 One reason for this might be that something is still missing. The fastest growing transfers in the last decades were old-age and health-related transfers. As these transfers mostly benefit older individuals, the simple-median voter model must be modified to allow for heterogeneity in age. This will also allow us to analyze the effect of ageing on redistribution.

13 11 / Two-dimens. heterogeneity (income/productivity+age): Introducing heterogeneity in age in addition to income enables us to focus especially on pensions which are the most rapidly growing gov. transfer systems (beside health-related transfers). Classification of social security + private pension systems Voluntary or compulsory Defining features of compulsory systems: 1 Beveridgean or Bismarckian benefits systems 2 Pay-as-you-go (PAYG), i.e. unfunded, or funded systems 3 Degree of actuarial fairness 4 Universal system or occupational schemes

14 12 / 59 Here, the focus is on PAYG pension schemes with interand intra-generational redistribution. This two-dimensional redistribution is important for the political equilibrium. We will see how a median-voter equilibrium with multidimensional heterogeneity can be computed as long as the policy space is one-dimension.

15 13 / 59 Model Overlapping-generations economy Each generation lives for three periods. Individuals work in the first two periods and are retired in the third period. Within each generation, labor income is heterogeneous. Individuals can invest their saving in a worldwide capital market at a given rate of return. Population growth is (for now) constant. The pension system is organized such that each working individual pays a proportional income tax. This then leads to a non-negative lump-sum transfer to each old individual. System redistributes within + across generations.

16 A young individual i maximizes ω iy = U(c iy )+ U(ciM ) (1 + υ) + cio (1 + υ) 2 +V(x iy )+ V (x im ) (1 + υ) (82) with υ: subjective discount rate, Y, M, O: period of life (young, middle-aged, old) NOTE: linearity of old-age consumption absorbs all income effects... subject to the budget constraint c iy + cim (1 + υ) + cio (1 + υ) 2 = liy (1 t) + lim (1 t) (1 + ρ) + f (1 ρ) 2 (83) with ρ: world interest rate (ass. ρ = υ), f : pension benefits 14 / 59

17 15 / 59 With the time constraint (75), (76) still gives labor supply of the young and planned labor supply of the middle-aged. Consumption when young and planned consumption when middle-aged is given by c = U 1 c (1). A middle-aged behaves in a similar fashion except that all variables from young age are given. An old consumes his pension plus his assets.

18 16 / 59 With n as the exogenous rate of pop. growth, the gov. budget constraint in per-retiree form is - using (76) f = tl M (1 + n) + tl Y (1 + n) 2 (84) = tl(t)(1 + n)(2 + n) Three effects of the pension system 1 it redistributes across generations 2 it redistributes within generations 3 the taxes needed to finance it distort the labor-leisure decision (however here no general equilibrium effects via endogenous factor prices)

19 17 / 59 Voter preferences: Vote over t (or equivalently f ) which then remains forever Old: Revenue-maximizing tax rate as they internalize only benefits and no costs (taxes) Young and middle-aged: Policy preferences according to age and income -> Poorer and older individuals generally with a preference for higher public pensions

20 18 / 59 Voter preferences: Consider a young voter of type (income) i. The indirect utility function defined by his optimal choices and the government budget constraint is affected by a marginal change in t in the following way (envelop theorem): Wt iy (t) = [l iy + lim (1 + ρ) ] + 1 df (1 + ρ) 2 dt (2 + ρ) = (1 + ρ) [L(t) + ei e] + (1 + n)(2 + n) (1 + ρ) 2 [tl t (t) + L(t)] (85) with (76) and (84).

21 19 / 59 The interpretation of (85) is straight-forward: A higher tax entails a higher benefit when old (last term) and a cost when young and middle-old (first term). The benefit is the same for all young voters, while the costs differ with respect to income (i.e. e i ) What about n and ρ? Individual preferences are thus again monotonic in e i.

22 20 / 59 Consider as a special case, the "golden-rule" case ρ = n. Setting (85) equal to zero, we get a condition identical to (79): The average young (e iy = e i = e) gains nothing from the pension system, but suffers from the distortionary taxes > Preference for t = 0. Young voters poorer than the average (e i < e) gain from intra-generational redistribution > Preference for t > 0.

23 21 / 59 (85) can be analogously derived for a middle-aged individual to get Wt im (t) [85a]. Marginal benefits of a tax increase are higher because they are closer in time and marginal costs are lower as taxes apply for one period only. Thus, a voter with the same relative income position (e iy = e im = e i ) prefers a higher tax rate when middle-aged than when young.

24 22 / 59 It is possible to identify a pair of young and middle-aged individuals who always vote alike e im = e iy + (1 + n)(2 + n) [tl t (t) + L(t)] (86) (2 + ρ) As can be easily seen, the middle-aged voter is richer than his young counterpart (e im > e iy ) Summary: Preferences are monotonic in income and age; older and poorer voters always prefer a higher value of t.

25 23 / 59 Political Equilibrium Who is the pivotal voter? For this, we focus on the monotonicity of individual preferences in income and age. All old prefer the revenue-maximizing tax rate. All young with income above the average prefer t = 0. All other prefer an interior solution. Thus, if neither the poor young nor the old constitute the majority, the equilibrium is interior: the median voter is a pair of voters - a poor young and a richer middled-aged voter - who both prefer the same tax rate.

26 24 / 59 Let e m be the middle-aged median voter and t m his preferred policy. The relation between e m and t m is obtained by setting W im t (t) = 0 (85a) and solving for e i = e m. Furthermore, with (86), we can identify the young voter who prefers t = t m. With e i distributed in the population with c.d.f. F(.), in equilibrium, the number of voters in favor of t > t m must be equal to the number of voters in favor of t < t m.

27 25 / 59 Thus, we find 1 + (1 + n)f(e m ) +(1 + n) 2 F[e m (1 + n)(2 + n) [t m L t (t m ) + L(t m )]] 2 + ρ 1 + (1 + n) + (1 + n)2 = (87) 2 [Coalition of the old and a fraction of the middle-aged and the young making up exactly half of the electorate.] If we plug in e m from (85a) and (86), (87) implicitly defines a unique equilibrium tax rate t m.

28 26 / 59 Further Considerations So far, we have focused on a pension system with a proportional income tax as contributions and lump-sum benefits which redistributes across, but also within generations. How do the results vary if the system is less redistributive? There is empirical evidence that countries with earningsrelated ("Bismarckian") public pension systems have higher contribution rates than flat-rate ("Beveridgean") systems - the average effective contribution rate in 1995 was 35% in the former and 19% in the latter.

29 27 / 59 Koethenbuerger, Poutvaara and Profeta (2008) analyze to what extent this can be reconciled by an "efficiencyredistribution" trade-off. They in particular ask whether a median voter model is able to explain the positive correlation between the size of the social security system and the degree to which pension benefits depend on past contributions.

30 28 / 59 Koethenbuerger, Poutvaara, Profeta: Model Individuals differ in two dimensions, age and productivity. Labor supply is endogenous. Pension benefits encompass an earnings-related and a flat-rate component. As earnings-related benefits cause smaller labor supply distortions, the efficiency costs of financing are lower the smaller the flat-rate part.

31 29 / 59 Koethenbuerger, Poutvaara, Profeta: Simulation The political-equilibrium contribution rates are computed for eight EU countries. Country-specific values are used for the pension system, the tax system,... For this, each population is decomposed into 3 age groups and 5 income groups. For each country, the contribution rate preferred by each age/income group is calculated. The median voter is identified.

32 Koethenbuerger, Poutvaara, Profeta: Simulation Figure 23: Data ξ: annual pop. growth rate, g: growth rate, θ: Bismarckian index, τ w : general wage tax rate, τ eff : eff. contribution rate. 30 / 59

33 Koethenbuerger, Poutvaara, Profeta: Results Figure 24: Simulation Results with γ: elasticity of mar. disutility with respect to labor supply 31 / 59

34 32 / 59 Koethenbuerger, Poutvaara, Profeta: Results Positive relation between θ and τ of about The median voter s preferred contribution rate is most likely increasing in θ. Reduced intra-generational redistribution makes social security more attractive for high-income middle-aged individuals, while at the same time reducing its efficiency costs. This numerical analysis suggests that the median voter model is able to explain the stylized fact that intra-generationally more redistributive systems are smaller.

35 33 / 59 (Possible) exam question (Median voter theorem) We have discussed some papers in class which test the empirical relevance of the median voter model among them Gerber and Lewis (2004) "Beyond the Median: Voter Preferences, District Heterogeneity, and Political Representation" and Koethenbuerger, Poutvaara and Profeta (2008) "Why are More Redistributive Social Security Systems Smaller? A Median Voter Approach". Choose ONE of the two papers and elaborate on it answering the following questions: 1 How does the paper address the question as to the relevance of the median voter theorem? 2 What is the chosen methodology? 3 What are the main results?

36 34 / Ageing Basic considerations Let us now consider what changes if we look at ageing, i.e. a decreasing n. Browning (1975) (see also Sjoblom, 1985): How does the preferred contribution rate increase with the age of the individual? Intuition (repetition): in calculating the marginal cost of a higher benefit level, past contributions are sunk costs which do not influence the trade-off between marginal costs of future contributions and marginal benefits of future pension benefits The older the politically decisive individual, i.e. the median voter, the higher is thus the realized contribution rate (relatively short period of contributing and relatively long retirement period)

37 35 / Ageing BUT: As can then be seen by the analysis above of t m with (85a) and (86), (87), a smaller population growth rate n has ambiguous effects on t m : A smaller n shifts the median voter s identity toward someone more favorable to pensions (political effect = median effect) A smaller n, however, makes pensions also less attractive for all young and middle-aged (economic effect = rate of return effect)

38 6.2 Ageing Conclusion 1 When the population decrease is not too large: Identity of the median voter (group) unchanged Rate of return effect dominates -> Ageing: lower contribution rates 2 When the population decline is large: Identity of the median voter changed Median effect possibly dominates -> Ageing: possibly higher contr. rates 36 / 59

39 37 / Ageing Voting with ageing: 2- Generation/Period OLG model Model: Casamatta, Cremer, Pestieau (2001) OLG-model with two generations: one working and one retired generation Individuals with two dimensional heterogeneity: age and productivity inter- and intra-generational redistribution Labor supply completely inelastic Interest rate and average wage given Borrowing constraints Pensions with Beveridgean and Bismarckian component (shares fixed)

40 38 / Ageing Expected pension benefits of an individual with earning w with pt+1 e (w) = (1 + n t)t t aw + }{{} pt+1 o }{{} Bismarckian part Beveridgean part pt+1 o (w) = (1 + n t+1)t t+1 w }{{} (1 + n t )t t a w }{{} Av. revenues in t + 1 Av. Bismarckian part (88) (89) w: gross wages with wɛ[w, w + ], mean w and median w m n: population growth rate t: contribution rate a: Bismarckian factor

41 39 / Ageing Assumption here: Bismarckian part: based on labor market agreements and insurance principles Beveridgean part: component of the social assistance program inspired by considerations of justice and altruism

42 40 / Ageing Effect of an unexpected demographic shock (ageing): Workers in the majority (if n > 0) If the pension system become too unattractive for workers, the system may be abolished. Retirees may suffer: They have paid contributions as workers, but majority voting could deny them their benefits. With ageing: Median age/median generation unchanged (= working generation) -> no political effect, only economic effect (contrib. rate )

43 41 / Ageing Voting with Ageing: More-Generation/Period OLG Models a) Reference: Galasso and Profeta (2004) The paper investigates the financial sustainability of unfunded pension systems under population ageing in 6 OECD countries: France, Germany, Italy, Spain, Uk, US Unless productivity rises enough, either contribution rates must rise and/or pension benefits decrease. For a reform, the conflicting interests of the different generations have to be balanced in a political process to develop a "politically sustainable" social security system. For this, the authors aim at evaluating quantitatively how political constraints shape social security systems.

44 42 / Ageing The focus is to measure the magnitude of the change in the size of the current systems (=the contribution rate) between 2000 and 2050 which is needed to retain the political sustainability. Again, the economic effect and the political effect of ageing are important. Which of the two effects will dominate is influenced by the economic and political decision of the individuals. To disentangle both effect, the model is simulated for 2050 with keeping the medians voter s age at the level of 2000.

45 43 / Ageing Results The simulations show that aging will largely raise social security contributions rates: the political effect is thus much stronger than the economic effect. For Germany and Spain, the large political effect (increasing of contribution rate) is reduced by a big economic effect. In Italy, the economic effect is quite small. For France, UK and US the economic effect surprisingly increases the contribution rate as well (reason: only small change in contribution rate) The generosity of the system - given by the replacement rat - increases only for UK and US.

46 44 / Ageing When individual labor market behavior (endogenous labor supply) is introduced to the economic environment, the political effect is still very relevant, but less than in the previous scenario. An increase in the effective retirement age in all countries, however, moderates the hike in the contribution rates: On average an increase in the effective retirement age of one year is associated with a reduction in the steady state political equilibrium contribution rate of 1.5 percentage points. The higher retirement age furthermore would make pension replacement rates more generous.

47 45 / Ageing b) Reference: Razin, Sadka, Swagel (2002) The focus is on the relationship between ageing and the tax rates and the size of the welfare state for the United States and 12 western European countries for The aim is to examine the relationship between the dependency ratio and two measures of the size of the welfare state: the tax rate on labor income and the generosity of social transfers. As a further independent variable (beside the dependency ratio and other variables) income skewness is included, i.e. the ratio of the income share of the top quintile to the share of the middle three quintiles ("rich versus middle"). This is supposed to correspond to the ratio of mean to median income as suggested by standard theory.

48 46 / Ageing Figure 25: Estimation results

49 47 / Ageing Information about the estimation OLS regressions with country-fixed effects; social transfers are in logs. Note dependency ratio here defined as one minus the labor force as a share of the population!

50 48 / Ageing Results: Columns 1+2: Dependency ratio with a statistically significant negative effect on the tax rate: a 1-%-point increase in the dependency ratio leads to a nearly 0.4-%-point decrease of the tax rate. Income skewness also with a statistically significant negative effect which is opposite to the standard result that inequality leads to pressure for redistribution. Columns 3+4: Dependency ratio with a statistically significant negative effect on (log) transfers per capita. Income skewness, however, without a significant effect. How can the coefficients for government jobs, trade openness and the unemployment rate be explained?

51 49 / Ageing Evaluation: There is thus evidence for a negative correlation between the dependency ratio and the size of the welfare state. The authors call this a "puzzle". They explain this observation by an effect called the "fiscal leakage" effect which occurs in an unfunded pension system in which current workers are net contributors and the retired are net beneficiaries.

52 50 / Ageing Evaluation: From our analyses above, we know, however, that this points to the fact that the economic effect dominates the political effect. Note: Other as in our earlier analysis, the size of the government or the scope for redistribution does not depend on particular measures of the skewness of the income distribution. Here, the relative size of the economic and political effects is decisive.

53 51 / Ageing Concluding Remarks: Casamatta et al. (2001) suggest that with ageing retirees may suffer. They have paid benefits as workers, but majority voting might deny them their benefits. Ageing may leave the retirees worse off (elimination of pensions). This result is different from the results by Galasso and Profeta (2004) who show that between 2000 and 2050 the contribution rates in all countries will increase.

54 52 / Ageing Concluding Remarks: The reason for this difference is the different number of overlapping generations. Casamatta et al. (2001): only two generations. This implies that the political effect (i.e. an increase of the age of the median voter) cannot work very well. Galasso and Profeta (2004): several generations / age groups (18-95 potentially). This clearly allows for a change of the age of the median voter (political effect) (see also Sinn and Uebelmesser, 2002). With only 2 or 3 generations, an important conflict of interest among workers is assumed away.

55 53 / Ageing Lobbying Basic idea: organized groups can contribute resources to candidates that favor their interests Retirees with an advantage Retirees more single-minded (homogeneous) than workers If lobbying requires both time and resources, this also favors retirees.

56 54 / Ageing Mulligan and Sala-i-Martin (1999) propose a model with time-intensive lobbying There are two overlapping generations, which form two lobbies competing for political influence to extract resources from each other. An interest group s political influence depends on the amount of non-working time of its members. The members of each group would like to work too much from their group s perspective as time worked means less time spent on lobbying (free-riding).

57 55 / Ageing A key mechanism is that the elderly have lower wages and thus smaller opportunity costs of lobbying. Result: Because the elderly lobby more, transfers go from the young to them. Another interpretation: single-mindedness of the elderly

58 56 / Ageing (Positive) Conclusion Both median voter and lobbying model predict that the interests of the elderly are over-represented in the political process. Without altruism, each cohort takes into account the effects of policy change for itself in the current and in the future periods. So each cohort takes into account the effect on retirees, but only the young take into account the effects on the young.

59 57 / 59 References Main references Persson, T. and G. Tabellini (2000), Political Economics - Explaining Economic Policy, MIT Press, Cambridge (Chapter 6).

60 58 / 59 References Casamatta, Georges, Helmuth Cremer and Pierre Pestieau (2001): "Demographic Shock and Social Security: A Political Economy Perspective", International Tax and Public Finance 8, Galasso, Vincenzo and Paola Profeta (2004): "Lessons for an Aging Society: the Political Sustainability of Social Security Systems", Economic Policy, Koethenbuerger, Marko, Panu Poutvaara and Paola Profeta (2008): "Why are more redistributive social security systems smaller? A median voter approach", Oxford Economic Papers 60, Razin, Sadka and Swagel (2002): "The Ageing Population and the size of the welfare state", Journal of Political Economy 110,

61 59 / 59 For Further Reading Browning, Edgar K. (1975): "Why the Social Insurance Budget is too Large in a Democracy", Economic Inquiry 13, Casamatta, G., H. Cremer, and P. Pestieau, (2000), "Political Sustainability and the Design of Social Insurance," Journal of Public Economics 75, Mulligan, Casey B., and Xavier Sala-i-Martin (1999): "Gerontocracy, Retirement, and Social Security", NBER Working Paper Sinn, Hans-Werner and Silke Uebelmesser (2002): "Pensions and the Path to Gerontocracy in Germany", European Journal of Political Economy 19, Sjoblom, Kriss (1985): "Voting for Social Security", Public Choice 45,