Topic 3: Policy Design: Social Security

Transcription

1 Topic 3: Policy Design: Social Security Johannes Spinnewijn London School of Economics Lecture Notes for Ec426 1 / 33

2 Outline 1 Why social security? Institutional background Design & Incentives Sustainability & Reform 2 Social Security and Savings 3 Social Security and Annuities 2 / 33

3 Social Security Program that taxes workers to provide income support for the elderly In many countries largest social insurance program, and fast growing - Krugman: "Loosely speaking, the post-cold-war government is a big pension fund that also happens to have an army" Facing major fiscal imbalance. 3 / 33

4 Social Security around the World Three Pillars: Government-provided pension benefits Employer-provided accounts Individual retirement accounts Pillar 1 in UK: Basic State Pension (involuntary, guarantees basic income) Additional Pension (depends on contributions through national insurance scheme) Pillar 1 in US: Social Security benefits are most important source of income for retired people 60% get more than 50% of their income from SS 30% get more than 90% of their income from SS Paid by payroll taxes 4 / 33

5 Social Security and the Government s Role Consumption smoothing: 1 save when working at young age for when retired at older age 2 insure against the risk (?) of disability at older age 3 insure against longevity But why provided by the government? 1 people don t save suffi ciently! 2 insurance/annuity market failures 3 redistribution & poverty relief 5 / 33

6 Institutional Background; how does it work? How is it financed? How are benefits calculated? Why is the reform of social security a central issue? 6 / 33

7 Fully Funded vs PAYG Fully funded schemes savings used to pay future benefits contributions are invested in assets returns are credited to the scheme s fund Pay As You Go schemes payment collected from today s workers to pay benefits for today s retirees contractarian in nature, run by the state Important difference PAYG allows for redistribution/insurance across generations PAYG allows every generation to receive more than it contributes if rate of growth > interest rate (Samuelson 1958) PAYG entails a Legacy Debt: first generation receives benefits without contributing 7 / 33

8 SS & Labor Supply Incentives at Prime-Age Relation between pension benefits and worker s contributions (tax-benefit link) affects the labor supply of prime-age workers. Defined Contribution schemes accumulate contributions and returns in account account used to finance/calculate pension retirees bear risks: real rates of return to assets, future pricing of annuities Defined Benefit schemes wage history used to calculate pension include set of wages, length of service sponsor ( = tax payers) bears some risks 8 / 33

9 SS & Labor Supply Incentives at Old-Age Relation between pension benefits and retirement age determines the labor supply for old-age workers. Adjustment for early/late retirement can be actuarially fair, if not implies an implicit tax rate on labor consider total difference in social security wealth SSW from working additional year SSW = PDV (benefits) PDV (contributions) disability insurance aspect suggests that implicit tax rate should be positive people who work longer contribute to the benefits of those who retire earlier 9 / 33

10 SS & Labor Supply Incentives: Empirical Evidence Introduction of Early Entitlement Age to receiving Pension 10 / 33

11 SS & Labor Supply Incentives: Empirical Evidence Cross-country comparison 11 / 33

12 Fiscal Imbalance 12 / 33

13 Fiscal Imbalance Life expectancy has increased, birth rate has decreased, wage growth has decreased increased dependence of retirees on workers Over the next 75 years, the US Social Security has promised $ 3.7 trillion more in benefit payments than it plans to collect in SS taxes Reform? tax base, benefit age, benefits (Diamond & Orszag 2005) personal retirement accounts (Feldstein 2005) Important confounds not all issues are inherent to PAYG / defined benefit system important to look at transition as well, rather than comparing steady states 13 / 33

14 Outline 1 Why social security? Institutional background Design and Labor Incentives Sustainability and Reform 2 Social Security and Savings 3 Social Security and Annuities 14 / 33

15 Social Security and Savings Mandated savings per se only justified if people would save too little otherwise (or if government has access to better investments) Claim in 80 s: people don t save suffi ciently for retirement. Observed patterns seem inconsistent with rational life-cycle model. Controversial puzzle is the massive drop in consumption after retirement hard to explain this within life-cycle model; behavioral explanations seem natural however, preferences may change. Also, drop in consumption is smaller than drop in expenses 15 / 33

16 Do Saving Policies increase Total Savings? Do mandates and subsidies increase total savings? If people save too little, a savings mandate like SS is only effective if it increases total savings Mandated or subsidized savings may have a crowd-out effect on private savings SS has high admin costs and does not accommodate heterogeneity in preferences Empirical answer has changed substantially with better variation Feldstein (1974): time series analysis suggests SS decreases private savings by 50% Chetty et al. (2012): quasi-experimental variation with large admin date crowd-out close to 0 for majority of savers 16 / 33

17 Active vs. Passive Savers Chetty et al. (2012) analyze whether retirement savings policies increase total wealth accumulation consider changes tax subsidies, mandates and employer-provided pensions use admin data for entire Danish poplulation Empirical analysis suggests the presence of active (15%) vs. passive (85%) savers. How to identify? mandates have no impact on active savers, but increase total savings of passive savers subsidies do not affect the savings decision of passive savers, but active savers respond by reallocating savings across acounts Presence of passive savers makes automatic contributions more effective in increasing total retirement savings. active savers are better prepared, so targetting the passive savers seems desirable. 17 / 33

18 Event-Study Design: Job Switchers 18 / 33

19 Event-Study Design: Close to Zero Crowd-out 19 / 33

20 Event-Study Design: Almost Full Pass-through 20 / 33

21 Event-Study Design: "Falsification" 21 / 33

22 RDD: Income Threshold for Savings Mandate 22 / 33

23 Dif-in-Dif: Change in Tax-deductibility of Savings 23 / 33

24 Subsidy: Substitution or Zero Response 24 / 33

25 Active vs. Passive: New vs. Old 25 / 33

26 Active vs. Passive: Illustrate Heterogeneity 26 / 33

27 Outline 1 Why social security? Institutional background Design and Labor Incentives Sustainability and Reform 2 Social Security and Savings 3 Social Security and Annuities 27 / 33

28 Theory of Annuities Annuity provides a stream of future cash payments conditional on being still alive against a payment today. Annuity provides insurance against outliving one s wealth Full annuitization of wealth is optimal (Yaari 1965, Davidoff, Brown and Diamond 2005) if no uncertainty other than time of death actuarially fair priced 28 / 33

29 No Annuities. Only Savings. Without annuities, individuals can save at interest rate R for future consumption Life-time utility equals V (W ) = max L s=0 π s δ s u (c s ) FOC s.t. L R s c s = W s=0 π s δ s u s u=log (c s ) = λr c s = π s δ s λr s. 29 / 33

30 Per-Period Annuities With per-period annuities, individuals can buy consumption at time s against price p s Life-time utility equals V A (W ) = max s.t. L s=0 L p s c s = W s=0 π s δ s u (c s ) FOC π s δ s u (c s ) = λp s u=log c s = π s δ s λp s 30 / 33

31 Actuarially Fair Annuities An actuarially-fair annuity has prices p s = π s R s V AF (W ) = max s.t. L s=0 π s δ s u (c s ) L π s R s c s = W s=0 FOC π s δ s u s u=log (c s ) = λπ s R c s = δs λr s 31 / 33

32 Full Annuitization Prices of future consumption are lower when buying annuity than when saving wealth, p AF s = π s R s < R s Fully anuitize your wealth! If δ = R, consumption is independent of s. Value of annuitization: V AF (θw ) = V (W ) θ < 1, but share depends on desired consumption trajectory; annuitization lowers the prices of future consumption and more so the further in the future they are prices don t need to be actuarially fair for full annuitization to be optimal 32 / 33

33 What can explain the low levels of annuitized wealth? Annuity products are irreversible / illiquid; still annuitize the share of wealth that you do not need to be liquid Positive utility from leaving bequests; still annuitize the share of wealth that you do not want to bequeath Private information about private risk Adverse Selection? empirical evidence suggests that people who die earlier do not annuitize less wealth (Poterba and Finkelstein 2004) welfare effects of adverse selection are estimated to be small = 2 percent of annual premiums (Einav et al. 2010) Behavioral constraints are likely to be important again (Brown et al. 2008) 72 percent prefer annuity over a savings account in "consumption frame" 21 percent prefer annuity over a savings account in "investment frame" 33 / 33