Public Finance II

Transcription

1 Public Finance II Topic VIII: Social Security Amy Finkelstein Spring 2018 Finkelstein () PF Slides Spring / 45

2 Outline 1 Background / Institutional details - see Section + Gruber (Chapter 13) 2 Reasons why we have social security (conceptual) 3 Social Security and Savings 4 Social Security and Labor Supply Finkelstein () PF Slides Spring / 45

3 Social Security Institutions (in <=1 slide!) Pay payroll taxes while working (capped/ regressive) Receive benefits when retired Benefit formula a function of taxes paid (progressive) Benefits paid as an annuity (survival contingent stream of payments) Recall welfare-enhancing potential of annuities Finkelstein () PF Slides Spring / 45

4 What are the economic elements? 1 Forced Savings (make you pay taxes when young; get back as retirement benefits) 2 Redistribution 3 Social insurance annuity component Finkelstein () PF Slides Spring / 45

5 Potential rationales for SS Correct market failures due to adverse selection in annuity markets Redistribution On lifetime earnings basis rather than annual earnings basis (an Akerlovian "tag") Paternalism Mandate savings in retirement Do individuals save "enough" for retirement? Samaritan s dillemma (govt will feel compelled to care for impoverished elderly) Do they optimally annuitize / draw down savings in retirement? Productive effi ciency Admin costs v low (vs. fees in mutual funds etc) Chile... Side comment: in thinking about SS reform, key question is often what you think purpose of SS is... Finkelstein () PF Slides Spring / 45

6 (Some) empirical questions about SS potential benefits Potential benefit I: Redistribution through SS? Potential benefit II: Ameliorating adverse selection in annuity markets through mandates. Empirical questions: Is there adverse selection in annuity markets? What are the welfare consequences of mandates? Potential benefit III: Increasing savings / improving consumption smoothing between working life and retirement. Empirical questions: Do people not save adequately for retirement? Do people save well (portfolio allocation questions) Does SS increase savings / improve consumption smoothing Finkelstein () PF Slides Spring / 45

7 (Some) empirical questions about SS potential costs Potential Cost I: Crowd out of private savings / reduce capital stock Feldstein (and others) view that central welfare questions involve impact of PayGo system on capital stock (and hence output and growth) Diamond (and others): program was designed as windfall / net transfer to Depression generation... Potential cost II: labor supply distortions Focus is mainly on timing of retirement Also an important question about labor supply while young (benefits tax issue; Summers 89) Finkelstein () PF Slides Spring / 45

8 Social Security and Savings SS as insurance against consumption changes in retirement? But retirement is predictable (and controllable). How is this insurance? Insurance behind veil of ignorance (I don t know if I m going to be the type who saves adequately for retirement). Or... paternalism. Key feature of SS / pensions lit: no opportunity for learning (vs e.g. health insurance, driving etc) Finkelstein () PF Slides Spring / 45

9 Social security and savings: some empirical questions Would individuals save optimally for retirement in the absence of Social security? i.e. is there a rationale for SS based on suboptimal savings? How much does Social Security increase consumption in retirement vs. crowd out savings that individuals would otherwise have done for their retirement? i.e. how much "consumption smoothing" (vs crowd out) does SS provdie? Finkelstein () PF Slides Spring / 45

10 Are individuals saving optimally? Two main approaches Look for evidence of anomalies / non rational behavior Calibrated life cycle utilty models Estimate optimal savings and compare to observed savings Hard problem. Key issues: What is optimal behavior? depends crucially on many hard to estimate parameters (recall prior lecture...) Also literature is asking whether behavior is optimal given existnece of SS. That is not the same as whether behavior would be optimal absent SS. Finkelstein () PF Slides Spring / 45

11 Approach 1: Evidence of non rational behavior Defaults: Madrian and Shea (QJE 2001) Default option chosen by firm has large impact on participation and contribution rate in 401k Variation in wealth among households with similar characteristics (e.g. earnings). Bernheim et al. AER 2001 Argue patterns not consistent with rational life cycle behavior e.g. no evidence that those who accumulate more wealth have higher rate of consumption growth (either prior to retirement or after retirement) as would expect based on a "rate of time preference" explanation Drops in consumption at retirement Long established fact. e.g. Bernheim et al. 30% drop in consumption at retirement Life cycle model predicts no change in consumption at (foreseeable) retirement assumption no change in MU of consumption Aguiar and Hurst (JPE 2005): drop in food expenditures, not food consumption (substitution to non- market activities) Finkelstein () PF Slides Spring / 45

12 Approach 2: Calibrated Life Cycle Model Scholz et al. (2006) Are Americans saving optimally for retirement? Calibrated stochastic life cycle model of optimal savings. Standard model enriched to allow for: Precautionary savings; Buffer stock behavior; Asset tested public transfers Uncertainty regarding mortality date and medical expenditures Taxation etc. HRS panel ( birth cohort) with data on wealth and entire past earnings history Start at in 1992 and follow through in 2004 calculate household specific optimal wealth targets given observed earnings history, survival probabilities, distribution of out of pocket medical expenditures etc. Conclude: Over 80% of HRS hh have accumulated more wealth than their optimal targets For 20% that are undersaving, wealth deficit is generally small Nice that they look not just at means but at whole distribution Finkelstein () PF Slides Spring / 45

13 (Some) comments Paper is about current elderly" - not saying anything about today s 40 year olds.. Part of what is driving their result (how rabbit comes out of hat): utility discounted by survival probabilities so optimal plan is one of steadily falling consumption. But why discount utility by survival probability? Only because annuity market not functioning?! More generally, as with Brown and Finkelstein (2008), paper highlights key parameters on which we need more work How is housing wealth treated? state dependent utility, equivalence scales, etc etc. Finkelstein () PF Slides Spring / 45

14 SS and Savings II: Crowd out Does Social Security crowd out private savings (vs. provide consumption smoothing benefits across work and retirement?) Note: retirement is predictable so presumably consumption smoothing across work and retirement needed due to mypoia / inadequate savings? Why do we care about crowd out? Cost of public funds Reduced benefits: the larger the crowd out, the less consumption-smoothing benefit across work and retirement of SS from mandatory savings Implications for capital stock NB: Welfare implications of crowd out unclear Crowding out of precautionary savings by providing an annuity could be effi cient Key point: emprical evidence on crowd out effect per se not suffi cient to answer welfare question... more work needed! Finkelstein () PF Slides Spring / 45

15 SS and Savings Crowd out: Evidence Evidence that SS associated w improved living standards of elderly suggest not 100% crowd out of private savings Time series: elderly poverty rate has fallen from ~30% in 1956 to ~10% in 2000 (poverty rates for kids and working population essentially stable) Costa (JPubEc 1999): introduction of Old Age Assistance associated with increased independent living of older non married women (mostly widows) Using state variation in program as well as program changes between 1940 and 1950 Note this is a need-based program that was made available to states (later turned into SSI) Also very interesteing in terms of incidence (under-studied in general) Finkelstein () PF Slides Spring / 45

16 SS and Savings Crowd out: Evidence (con t) Estimates of savings crowd out Attanasio (2003 QJE; 2003 AER) Italy pension reform: reduced pension benefits for younger workers in public sector (less so for older workers and private sector workers). DD estimats of change in savings for young public sector workers relative to older public sector workers and private sector works finds percent of the reduction in SS wealth offset by higher savings UK pension reform.(similar type natural experiment and similar results) Finkelstein () PF Slides Spring / 45

17 SS and Labor Supply Will focus on two questions: SS as a "benefits-tax"? (Summers 1989) Impact of SS on retirement behavior Finkelstein () PF Slides Spring / 45

18 Social Security as a benefits tax Payroll tax to finance social security benefits If people recognize that taxes paid today go for benefits tomorrow (and there is no redistribution and benefits are fully valued relative to cost) then there would be no distortionary effect of payroll tax One key question: are workers aware of benefits-tax linkage in Social security? If not, payroll tax functions like a pure distortionary tax (no outward shift of labor supply) One argument for "indivdiuals accounts" is to make the tax-benefit linkage more salient Finkelstein () PF Slides Spring / 45

19 Social Security as a benefits tax Relatively little work on effect of payroll tax on prime age LFP Incredibly important for effi ciency cost of SS (Summers 89) Gruber (1997 JOLE) on large reduction in payroll tax in Chile Finds full pass through to wages, no effect on employment Saez et al. (QJE 2012) on large increase in payroll tax in Greece applied only to net entrants (those who started working on or after January 1, 1993) worker wages compensated for increase in employer payroll tax, but not employee payroll tax therefore new workers have higher labor costs, same posted earnings (excludes employer but includes employee payroll tax), lower net earnings perhaps because can t pay similarly workers differently due to e.g. fairness, bargaining etc? Finkelstein () PF Slides Spring / 45

20 Social Security as a benefits tax (con t) Liebman, Luttmer, Seif (JPubEc 2009) look at labor supply response for older workers to sharp changes in marginal return to work Trying to look at whether individuals understand benefit-tax linkage in SS Examine various discontintuities in effective marginal tax rate created by SS system Findings e.g. benefits are max{own,50%spousal} so marginal return to work drops discontinuously at point where spousal social security benefit is double own individuals approaching retirement response to these incentives (e.g. break in tax-benefit link) on extenisve (retirement) margin. mixed evidence on intensive margin Finkelstein () PF Slides Spring / 45

21 Social Security and Retirement Focus on this motivated in large part by large decline in LFP of eldery (recall DI debate) Why would SS affect retirement Lots of different effects to consider See e.g. discussion in Crawford and Lilien (QJE 1981) or Diamond and Gruber (1999) Proceed in 3 steps (useful template for thinking through problems - e.g. for generals and beyond...) Benchmark case: fully rational actors and perfect markets [always useful to start here!] With supply side market failures (imperfections in credit and insurance markets) With demand side failures / departures from full rationality Finkelstein () PF Slides Spring / 45

22 Benchmark: Perfect markets, full rationality Consider: Income and substitution effects SS will have no effect on retirement if 2 conditions are met No income effects: no redistribution within or across generations No substitution effects: Actuarially fair treatment of retirement for each person NB: much stronger than actuarially fair on aveage. V unlikely to hold (annuity benefits not priced on individual-specific basis. Not even adjusted for gender!) Finkelstein () PF Slides Spring / 45

23 Income effects (redistribution w/in and across generations) If SS represents a net transfer (increase in wealth) to an individual should > earlier retirement (leisure a normal good) Taxes regressive; benefit formula progressive; mortality differentials by SES undo some of benefit progressivity (bc of annuity) Lots of redistribution through the system (e.g. Diamond and Gruber 1999) Transfers to older generation (windfall to elderly during great depression) Along income distribution From men to women From single to married Within couples from more equal to less equal earnings Finkelstein () PF Slides Spring / 45

24 Substitution effects Financial return to another year of work once eligible for benefits If adjustment of benefits is not person-specifi actuarially fair at every age, will tend to find net tax or subsidy on additional work Diamond and Gruber: Implicit tax from working another year: net SSW =EPDV(benefits)-Payroll taxes during continued work Implicit tax = newssw /earnings (Note: discounting by preferences and mortality) Compute net SSW if retire today. Compute net SSW if retire next year. Compare Implicit tax rate from social security from working another year: net change in SSW implied by that year of work divided by potential earning from that year of work Positive tax > SS causes disincentives to work another year through foregone SSW Finkelstein () PF Slides Spring / 45

25 What determines net change in SSW? Pay an extra year of taxes if work an extra year (lowers net SSW) Additional year of worked used to recompute SS benefits which tends to raise net SSW If haven t worked 35 years, replace a zero year w positive earnings Even if have worked 35 years, age-earnings profile > likely to be replacing a previously low earnings year. Additional year of work for work at ages 62 and beyond delay in claiming benefits (lowers net SSW) actuarial adjustment to benefits (designed to be fair on average) Finkelstein () PF Slides Spring / 45

26 What do implicit taxes look like? Diamond and Gruber (1997) System roughly actuarially fair on average for married men at age 62 No implicit tax on additional work. Remember this when get to empirical evidence! System penalized work after 65 Delayed retirement credit is less than actuarially fair given foregone year of SS benefits Finkelstein () PF Slides Spring / 45

27 Why would SS affect retirement: Supply side failures Imperfections in credit and insurance markets Liquidity contraints > ambiguous effect Some young workers are liquidity constrained (diffi cult to forrw against future). SS payroll tax therefore forces them to consume less whne young > as a result have more wealth when retire > income effect produces earlier retirement Inability to tap this source of wealth until early retirement age may > work until retirement age when otherwise might not have Imperfect annuity markets SS provides real annuities that are not available in market Ambiguous effect on retirement from annuities through SS: Greater effi ciency in planning lifetime consumption works like income effect (therefore leads to earlier retirement) Link between work and size of annuities > additional incentive to work more (only way to get access to annuities) Want to retire earlier but can t access annuity till 62 > delay retirement Finkelstein () PF Slides Spring / 45

28 Why would SS affect retirement: "Behavioral" Departures from full rationality (e.g. myopia). Ambiguous effects: By forcing people to save more than they would myopically choose > greater welath at retirement > retire earlier (income effect) Myopic individuals might retire too soon (not realized have not saved enough) but for limited eligibility until early retirement age Comment: Different mechanisms have v different implications for welfare / policy (e.g. income vs. substitution effects) Ideally want to know now only what (net) effect of SS is on retirement but why Finkelstein () PF Slides Spring / 45

29 Empirical work: time series motivation Source: Gruber textbook Finkelstein () PF Slides Spring / 45

30 Some empirical work "Natural experiments" Notch babies (Krueger and Pischke 1992) - end of double indexing in the notch unexpectedly reduces SS benefits Earnings test (Friedberg 2000). Finds bunching at kink of earnings test threshold before it was aboloshed Liebman, Luttmer and Seif (2009) retirement response to discontinuous changes in effective SS tax rate (e.g. when get to 35 years of earnings) Fetter and Lockwood (forthcoming) on Old Age Assistance Finkelstein () PF Slides Spring / 45

31 Old Age Assistance (Fetter and Lockwood) OAA means-tested program created along w SS in the 1930s (later became SSI) large program for individuals 65+: about one-quarter of 65+ receive OAA in 1940 larger than social security until 1950 state administered (variation across states in eligibility and benefit levels) Two reasons OAA is expected to reduce late-life labor supply Income effect: reduces marginal utility of consumption in retirement that is gained from working more Substitution effect: earnings limit in order to receive = tax on work after 65 relative to work at younger ages Finkelstein () PF Slides Spring / 45

32 Finkelstein Figure 2: () Labor force participation in 1940, PF by Slides age and state OAA payments per person Spring / 45 Time series patterns Tables and Figures Figure 1: Government old-age support and retirement over the mid-20th century (a) Aggregate trends (b) Age profiles, Labor force participation rate, men OAA+OASI payments per person 65+, 2010 USD Share in the labor force at Census LFP OAA+OASI (includes pre 1935 state programs) Notes: Panel (a) shows labor force participation rate of men 65 and older, from Series D35 of U.S. Bureau of the Census (1975), and payments under Old Age Assistance (OAA) and Old Age and Survivors Insurance (OASI) per person 65 and older, in 2010 US dollars. OAA payments data come from Parker (1936) for 1925 to 1935 and Series Bf621 of Carter et al. (2006) for 1936 onwards. OASI payments data come from Series BF396 of Carter et al. (2006). Panel (b) shows share of men in the labor force at each age, calculated from the Censuses. Rates prior to 1940 are adjusted for comparability: see Appendix Section A.1 for details.

33 (OASI) per person 65 and older, in 2010 US dollars. OAA payments data come from Parker (1936) for 1925 to 1935 and Series Bf621 of Carter et al. (2006) for 1936 onwards. OASI payments data come from Series BF396 of Carter et al. (2006). Panel (b) shows share of men in the labor force at each age, calculated from the Censuses. Rates prior to 1940 are adjusted for comparability: see Appendix Section A.1 for details. Differential time series across states Figure 2: Labor force participation in 1940, by age and state OAA payments per person 65+ Share in the labor force at Census Age in April 1940 Above median states Below median states Notes: Figure shows share of men in the labor force at the time of the 1940 Census, in states with aboveand below-median OAA payments per person 65+ in 1939, for states with an eligibility age of 65 in Finkelstein () PF Slides Spring / 45

34 Main findings OAA can explain more than half of the large decline in late life male labor force participation between 1930 and 1940 Welfare cost to recipients of high tax rates implicit in OAA s earnings test were small magnitude of decline in overall labor force participation rates given OAA eligibility suggests that about half of eligibles changed their labor force participation in response so for about half, OAA was valued at at least a dollar and for other half, looks like much of the "labor force exit" was out of unemployment or employment in work relief programs strucutral model complements this "reduced form" welfare analysis. Finkelstein () PF Slides Spring / 45

35 Some empirical work (con t) Spikes literature compelling rejection of null of no behavioral response In US, EEA at 62 (actuarially fair adjustment on average) and FBA originally at 65 (with penalty to work beyond 65) Modeling dynamic problem [will cover after spikes] Finkelstein () PF Slides Spring / 45

36 Spikes literature (US) Source: Diamond and Gruber (1997, NBER WP) Finkelstein () PF Slides Spring / 45

37 Spikes comments Spike at age 65 Recall implicit tax on work at age 65 Is the spike a big problem for the Card et al. age 65 Medicare RD? Spike at age 62 Recall actuarially fair adjustment (on average) at 62 Maybe people just like to retire at 62 and that s why they set EEA? Finkelstein () PF Slides Spring / 45

38 EEA introduced in 1963 Source: Gruber textbook Finkelstein () PF Slides Spring / 45

39 Understanding spike at age 62 No implicit tax on average so why spike? Conjectures Pent up demand to retire prior to EEA at 62 but cannot due to Liquidity constraints Imperfect annuity markets / have to wait for legal access Signaling - take EEA as signal of what government considers appropriate Given spike at 62, can t be sure how much spike at 65 due to implicit tax on work at this age vs other potential factors Signaling Other things linked to 65 (Medicare; private pensions often penalize work after 65) Finkelstein () PF Slides Spring / 45

40 Retirement decision a dynamic problem "Implicit tax" is about one-year accrual rate. Not clear this is right approach What really matters in deciding whether to continue to work next year is how my entire pattern of future work will change my benefits Stock and Wise (1990) "option value" model: by working retain option to work more in future key insight: need to consider entire future path of incentives not just the one-year accrual rate See also Coile and Gruber (1997) Finkelstein () PF Slides Spring / 45

41 Manoli and Weber (2011) Austrian employer pension system: workers receive an lump sum cash benefit from their employer at entry into retirement Amount of severence pay depends on tenure in step wise function. Discrete increases in benefits at 10, 15, 20, 25 years of job tenure (note: not age) Study retirement response to these discontinuities Begin with static analysis (spikes / bunching) But need dynamic model... Finkelstein () PF Slides Spring / 45

42 Discrete changes in benefits Fig. 1. Payment Amounts based on Tenure at Retirement Severance Pay Fraction Years of Tenure at Retirement Notes: There are two forms of government-mandated retirement benefits in Austria: (1) government-provided pension benefits and (2) employer-provided severance payments. The employer-provided severance payments are made to private sector employees who have accumulated sufficient years of tenure by the time of their retirement. Tenure is defined as uninterrupted employment time with a given employer and retirement is based on claiming a government-provided pension. The payments must be made within 4 weeks of claiming a pension according to the following schedule. If an employee has accumulated at least 10 years of tenure with her employer by the time of retirement, the employer must pay one third of the worker's last year's salary. This fraction increases from one third to one half, three quarters and one at 15, 20 and 25 years of tenure respectively. Since payments are based on an employee's salary, overtime compensation and other non-salary payments are not included when determining the amounts of the payments. Provisions to make these payments come from funds that employers are mandated to hold based on the total number of employees. Severance payments are also made to individuals who are involuntarily separated (i.e. laid off) from their firms if the individuals have accumulated sufficient years of tenure prior to the separation. The only voluntary separation that leads to a severance payment, however, is retirement. Employment protection rules hinder firms from strategically laying off workers to avoid severance payments and there is no evidence on an increased frequency of layoffs before the severance pay thresholds. Finkelstein () PF Slides Spring / 45

43 Spikes in retirement at thresholds Fig. 3. Distribution of Tenure at Retirement, Full Sample Individuals Years of Tenure at Retirement Notes: This figure plots the distribution of tenure at retirement at a monthly frequency. Each point captures the number of people that retire with tenure greater than the lower number of months, but less than the higher number of months. Tenure at retirement is computed using observed job starting and job ending dates. Since firm-level tenure is only recorded beginning in January 1972, we restrict the sample to individuals with uncensored tenure at retirement (i.e. job starting after January 1972). Finkelstein () PF Slides Spring / 45

44 Need for dynamic model Show spikes in retirement at tenure threshold and dips before Why not apply Saez (2010) to get elasticity of retirement wrt change in incentive? Would miss dynamics: labor supply behavior in response to anticipated benefit changes Set of people "at risk" to bunch at threshold endogeneous / affected by threshold e.g. can see dips but it s not just them... Option value from continuing to work to reach next threshold Dyanmic problem requires dynamic model Key additional parameters to identify include: arrival rate of any shocks that affect retirement decision (e.g. health?) + degree of forward looking behavior / discounting Finkelstein () PF Slides Spring / 45

45 Analog to Part D - response to donut hole Tempting to apply Saez (2010) and related "bunching" papers from labor supply / taxation literature But not a static problem Series of sequential decisions (fill / not fill prescription) as shocks arrive Utilization behavior long before donut hole may be affected by donut hole Key concluding comment: in the end, everything comes back to health care! Finkelstein () PF Slides Spring / 45