Active vs. Passive Decisions and Crowd-out in Retirement Savings Accounts: Evidence from Denmark

Transcription

1 Active vs. Passive Decisions and Crowd-out in Retirement Savings Accounts: Evidence from Denmark Raj Chetty, Harvard and NBER John N. Friedman, Harvard and NBER Soren Leth Petersen, Univ. of Copenhagen and SFI Torben Nielsen, Univ. of Copenhagen and SFI Tore Olsen, Univ. of Copenhagen and CAM December 2012

2 Question Do retirement savings policies e.g., tax subsidies or employer-provided pensions raise total wealth accumulation? Or simply induce shifting across accounts? More generally, do changes in net-of-tax return affect savings rates? Central questions for understanding optimal design of retirement policies Existing evidence mixed [e.g. Hubbard 1984, Skinner and Feenberg 1990, Poterba, Venti, Wise 1996, Engen, Gale, Scholz 1996, Engelhardt and Kumar 2007, Gelber 2010] Largely due to limitations in data and research designs [Bernheim 2002]

3 Overview We estimate crowd-out in retirement savings accounts using Danish tax data 45 million observations on savings from administrative sources We analyze two types of policies Automatic contributions by government or firms to workers retirement savings accounts Price subsidies for retirement savings Main finding: Automatic contributions raise total savings much more than price subsidies Interpret this result through a model of active and passive choice

4 Conceptual Framework: Active vs. Passive Savers Compare impacts of automatic contribs. (M) and subsidies (t) for two types of individuals in a standard two-period model of consumption Active savers choose pension savings (P) and taxable savings (S) to maximize utility as in neoclassical model Passive savers set P = P irrespective of govt. policies t and M We do not specify a positive model of passive choice [Bernheim and Rangel 2009] Examples include fixed costs of attention or present-biased preferences [Carroll et al. 2009] For simplicity, ignore corners: P and S can be negative

5 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Passive Savers Yes Uncertain

6 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Passive Savers Yes Uncertain Prior work (e.g., Madrian and Shea 2001) suggests that automatic contribs. raise total pension contributions

7 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Passive Savers Yes Uncertain But impact of auto. contrib. on total savings unclear for passive agents

8 Impacts of Automatic Contributions on Savings Effect of automatic contribution on total savings depends on how passive agents adjust to reduction in disposable income when M rises Two extreme cases span potential responses

9 Impacts of Automatic Contributions on Savings Effect of automatic contribution on total savings depends on how passive agents adjust to reduction in disposable income Two extreme cases span potential responses 1. Absorb reduction in income by running down bank balance Fixed consumption plan smaller end-of-year bank balance $1 automatic contribution leads to no change in total savings

10 Impacts of Automatic Contributions on Savings Effect of automatic contribution on total savings depends on how passive agents adjust to reduction in disposable income Two extreme cases span potential responses 1. Absorb reduction in income by running down bank balance Fixed consumption plan smaller end-of-year bank balance $1 automatic contribution leads to no change in total savings 2. Absorb reduction in income by reducing consumption Fixed savings target cut consumption to meet savings goal $1 automatic contribution leads to $1 increase in total savings

11 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No

12 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No Prior work on match rates (e.g. Duflo et al. 2006, Engelhardt and Kumar 2007) suggests that subsidies increase pension savings

13 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No But impact of price subsidy on total savings again unclear

14 Impacts of Subsidies on Savings Subsidies only affect pension contributions of active savers Need to be attentive to respond to subsidy Active savers are also cognizant of all substitution possibilities Subsidies could be crowded out significantly more than policies that affect passive savers

15 Outline 1. Test the four predictions

16 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No Data????

17 Outline 1. Test the four predictions 2. Estimate fraction of active vs. passive savers Passive Data x Active 0 1 response to M and t 3. Analyze characteristics of active vs. passive savers

18 Institutional Background Denmark has three major retirement savings policies: 1. Tax-deferred individual savings accounts 2. Tax-deferred employer savings accounts 3. Defined benefit public pension system Fixed benefit of DKr 140,000 (US $23,000) with earnings test based on non-retirement savings

19 Data and Sample Definition Universe of Danish income tax returns, Sample restrictions: (1) not self employed and (2) age million unique individuals 45 million observations on savings Data on total financial savings from 3 rd party reports to tax authority

20 Data on Savings 1. Flows into retirement savings accounts from tax records 2. Third-party reports by financial companies on non-retirement account bals. Used for wealth tax until 1996 and then to monitor financial firms 3. Third-party reports on unsecured debt and mortgage interest from 1997 We find no impacts of the policies we study on liabilities

21 Summary Statistics Variable Full Sample Top Bracket Sample (1) (2) Gross Labor Income 199, ,618 Gross Taxable Income 217, ,607 Net Capital Income -14,549-20,541 Assets (not incl. home equity) 51,602 60,495 Assets >10% of labor income 47% 42% Assets/Gross Labor Inc. Ratio Total Savings 23,904 32,752 Saving Rate 18.92% 15.32% Liabilities (not incl. home mortgage) 76,539 95,374 Change in Liabilities 5,681 5,529 Net Savings Rate 4.06% 9.68% Pension Flows Fraction with Indiv. Pension 27% 36% Indiv. Pension 3,143 4,081 Indiv. Pension Contribution Rate 1.18% 1.25% Indiv. Capital Pension 1,859 2,643 Indiv. Annuity Pension 1,284 1,438 Fraction with Employer Pension 59% 83% Employer Pension 15,542 21,717 Employer Pension Contribution Rate 5.67% 6.98% Fraction with Any Pension 66% 90%

22 Summary Statistics Variable Full Sample Top Bracket Sample (1) (2) Demographics Age Female 52% 44% Married 48% 58% Has Partner 62% 73% Homeowner 51% 68% College Degree 41% 59% Some Economics Training 4% 4% Number of Observations 45,428,846 17,712,370

23 Part 1 Impacts of Automatic Contributions

24 Impact of Automatic Contributions Do automatic contributions to retirement accounts affect individuals total saving? Ideal experiment: randomize automatic contributions holding fixed total compensation ($100 increase in pension + $100 reduction in earnings) Two quasi-experimental research designs: 1. Variation in employer-provided pensions 2. Government mandatory savings plan Both designs yield similar results; focus today on employer changes

25 Impacts Employer-Provided Pensions Research design: event studies around job changes Compare impacts of sharp increases or decreases in employer pension contributions at the time of job change Two potential concerns with this design Job switches may be endogenous Total compensation changing as well; need to net out income effect We address both concerns after presenting baseline results

26 Contribution or Savings Rate (% of income) Event Study around Switches to Firm with >3% Increase in Employer Pension Rate for Individuals with Positive Pension Contributions or Savings Prior to Switch Δ Employer Pensions = 5.65 Δ Total Pensions = Year Relative to Firm Switch Employer Pensions Total Pensions

27 Contribution or Savings Rate (% of income) Event Study around Switches to Firm with >3% Increase in Employer Pension Rate for Individuals with Positive Pension Contributions or Savings Prior to Switch Δ Employer Pensions = 5.65 Δ Total Pensions = Year Relative to Firm Switch Employer Pensions Total Pensions

28 Contribution or Savings Rate (% of income) Event Study around Switches to Firm with >3% Increase in Employer Pension Rate for Individuals with Positive Pension Contributions or Savings Prior to Switch Δ Employer Pensions = 5.65 Δ Total Savings = Year Relative to Firm Switch Employer Pensions Total Savings

29 Percent at Corner Fraction at Corner around Switches to Firm with >3% Increase in Employer Pension Rate Δ Zero Pension Contrib.= 1.4% Year Relative to Firm Switch Individual Pensions

30 Percent at Corner Fraction at Corner around Switches to Firm with >3% Increase in Employer Pension Rate Δ Zero Pension Contrib.= 1.4% Predicted = 28.4% Year Relative to Firm Switch Individual Pensions Predicted with Full Crowd-Out

31 Percent at Corner Event Study of Fraction at 0 Corner around Switches to Firm with >3% Increase in Employer Pension Rate Δ Zero Pension Contrib.= 1.4% Predicted = 28.4% Δ Corner in Savings = 1.0% Predicted = 28.4% Year Relative to Firm Switch Individual Pensions Total Savings Predicted with Full Crowd-Out Predicted with Full Crowd-Out

32 Percent of Individuals Changes in Individual Pension Contributions in Year of Firm Switch For Individuals Contributing to Private Pensions in Prior Year Change Required to Offset 3% Increase D Private Pension Contributions as a Percentage of Income

33 Change in Total Pension Contribs. (% of income) Changes in Total Pension Contributions vs. Changes in Employer Pension Rates for Firm Switchers Contributing Pensions Prior to Switch Total Pensions Pass-Through Rate: b = 95% (0.2%) Change in Employer Pension Contributions (% of income)

34 Change in Total Savings Rate (% of income) Changes in Total Savings Rates vs. Changes in Employer Pension Rates for Firm Switchers with Positive Savings Prior to Switch Total Savings Pass-Through Rate: b = 90% (0.9%) Change in Employer Pension Contributions (% of income)

35 Change in Total Savings Rate (% of income) Changes in Total Savings Rates vs. Changes in Wage Rates for Firm Switchers with Positive Savings Prior to Switch Marginal Propensity to Save: 10.2% (0.1%) Percentage Change in Wage Rate

36 Employer Pensions: Pass-Through Estimates Sample: All Firm Switches All Firm Switches All Firm Switches Mass Layoff Large Changes First Switch Switch Ages Dep. Var.: Δ Emp. Pens. Contrib. Rate Δ Wages Δ Pension Rate Δ Savings Rate Δ Savings Rate Δ Savings Rate Δ Savings Rate Δ Savings Rate Δ Retirement Balance (1) (2) (3) (4) (5) (6) (7) (0.002) (0.009) (0.009) (0.001) (0.110) (0.011) (0.018) (0.380) No. of Obs. 910,866 2,078,612 2,078,612 36, , ,273 55,608

37 Employer Pensions: Pass-Through Estimates Robustness Checks Sample: All Firm Switches Renters All Firm Switches Single Individuals Dep. Var.: Δ Net Savings Δ Savings Rate Δ Household Savings Δ Savings Rate (1) (2) (3) (4) Δ Emp. Pens. Contrib. Rate (0.014) (0.013) (0.010) (0.015) No. of Obs. 1,858, ,450 2,024, ,188

38 Pass-Through of Employer Pensions to Total Savings Pass-Through of Employer Pension to Total Savings by Years Since Firm Switch Years Since Firm Switch

39 Total Saving From Switch to Age 60 (% of Earnings before Switch) Total Wealth Accrued from Switch to Retirement (Age 60) vs. Changes in Employer Pension Rate at Switch Δ Retirement Balance / Δ Emp. Pension = 5.81 (0.38) Change in Employer Pension Contributions (% of Income)

40 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+_S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No Data Yes Yes??

41 Impacts of Mandates Employer pensions provide good identifying variation but may not be identical to impacts of government policies Ex: workers may be more willing to change consumption plans when switching firms Now directly study a mandate imposed by Danish government in 1998 Required individuals above eligibility threshold of DKr 35,000 to contribute 1% of savings to mandatory savings account Estimate impacts using RD and DD designs Main finding: $1 contribution to mandatory savings plan ~ $1 increase in pensions and total savings

42 Mandated Savings (DKr) Mandated Savings (M) Around Eligibility Threshold in Income (DKr 1000s)

43 Effect on Mandate on Total (Non-Employer) Savings: Threshold Approach Percent with Non-Employer Savings > 1962 DKr Income (DKr 1000s) Empirical

44 Percent with Non-Employer Savings > 1962 DKr Effect on Mandate on Private Savings: Threshold Approach Private Savings Pass-Through Rate: b = 117% (27%) Income (DKr 1000s) Empirical Predicted with 100% Pass-Through

45 Mandated Savings Plan: Pass-Through Estimates Dep. Var.: Δ Total Pensions % with Total Pensions > Mean Δ Total Ind. Savings % with Total Ind. Savings > Mean Δ Total Savings % with Total Savings > Mean (1) (2) (3) (4) (5) (6) Pass-Through Estimate (0.251) (0.172) (14.692) (0.271) (1.744) (0.290) Research Design RD RD RD RD RD RD No. of Obs 37, ,001 92, ,157 92, ,157

46 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No Data Yes Yes??

47 Part 2 Impacts of Subsidies for Retirement Saving

48 Impact of Subsidies Denmark has two types of tax-deductible savings accounts: Capital pensions: paid as a lump sum Annuity pensions: paid as annuity Subsidy for capital pensions for individuals in top income tax bracket was reduced in 1999

49 Taxation of Capital and Annuity Pensions for Individuals in the Top Tax Bracket Capital Pensions Annuity Pensions Pre-1999 Post-1999 Pre-1999 Post-1999 Tax deduction 59% 45% 59% 59% Capital gains tax 15% 15% 15% 15% Tax on payout 40% 40% Taxed as income Taxed as income

50 Flow Subsidy for Capital Pension Contribs. Reduction in Subsidy for Capital Pensions in 1999 Control group DSubsidy = -14% Treated group Note: $1 6 DKr Gross Income Prior to Pension Contribution (DKr 1000s)

51 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr) 1996

52 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr)

53 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr)

54 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr)

55 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr)

56 Impact of 1999 Capital Pension Subsidy Reduction On Capital Pension Contribs Capital Pension Contribution (DKr) Income Relative to Top Tax Cutoff (DKr)

57 Capital Pension Contribution (DKr) Impact of Subsidy Reduction On Individual Capital Pension Contribs Year DD Impact Estimate: β = (97.65) Subsidy for Capital Pension Reduced K Below Top Tax Cutoff 25-75K Above Top Tax Cutoff

58 Percent of Individuals Impact of 1999 Capital Pension Subsidy Reduction on Distribution of Capital Pension Contributions for Prior Contributors Percentage Change in Capital Pension Contributions (P t P t-1 )/ P t to 1998

59 Percent of Individuals Impact of 1999 Capital Pension Subsidy Reduction on Distribution of Capital Pension Contributions for Prior Contributors Percentage Change in Capital Pension Contributions (P t P t-1 )/ P t to to 1999

60 Fraction of Individuals Respond to Subsidy Active savers in interior always change pension contributions when subsidy changes Can therefore estimate fraction of active savers based on fraction of individuals who reoptimize because of reform By analyzing distribution of changes, we estimate that only 17% of prior contributors reoptimize contribution levels because of subsidy change Nearly all of the response comes from the 16% of individuals who exit capital pensions completely in 1999 (extensive margin)

61 Percent Still Contributing to Capital Pension Effect of 1999 Reform on Fraction of Capital Pension Contributors by Year for Individuals Contributing Prior to Reform Years Since 1999 Reform

62 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers (Neoclassical) No No Yes Uncertain Passive Savers Yes Uncertain No No Data Yes Yes Yes?

63 Crowd-out of Pension Savings Two crowd-out parameters of interest 1. Shifting across pension accounts What happens to each $1 taken out of capital pensions? Relevant parameter for impacts of a policy that targets one type of retirement account (e.g. firm match for 401k)

64 Crowd-out of Pension Savings Two crowd-out parameters of interest 1. Shifting across pension accounts What happens to each $1 taken out of capital pensions? Relevant parameter for impacts of a policy that targets one type of retirement account (e.g. firm match for 401k) 2. Shifting from pension accounts to taxable savings accounts What happens to each $1 taken out of pension savings? Relevant parameter for determining overall impact of retirement savings subsidies on total savings Corresponds to estimates reported in prior work

65 Annuity Pension Contribution (DKr) Impact of Capital Pension Subsidy Reduction On Annuity Pension Contributions Subsidy for Capital Pension Reduced Annuity Pension Offset: β = 56% (4.7%) Year 25-75K Below Top Tax Cutoff 25-75K Above Top Tax Cutoff

66 Total Pension Contribution (DKr) Impact of Capital Pension Subsidy Reduction On Total Pension Contributions Subsidy for Capital Pension Reduced Year 25-75K Below Top Tax Cutoff 25-75K Above Top Tax Cutoff

67 Total Pension Contribution (DKr) Impact of Capital Pension Subsidy Reduction On Total Pension Contributions Subsidy for Capital Pension Reduced Total Pensions Pass-Through Rate: β = 44% (4.7%) Change in Capital Pensions Year 25-75K Below Top Tax Cutoff 25-75K Above Top Tax Cutoff

68 Shifting from Pensions to Taxable Savings Now analyze shifting from retirement accounts to taxable accounts Use change in capital pension subsidy as an instrument for total pension contributions $1 reduction in capital pensions 44 cent reduction in total pensions Does this 44 cents go into additional consumption or taxable savings? Difference-in-differences design too noisy to detect impacts on taxable savings Differential shocks across income groups in time series

69 Taxable Savings (DKr) Impact of Capital Pension Subsidy Reduction On Taxable Savings Year Subsidy for Capital Pension Reduced Change in Total Pensions K Below Top Tax Cutoff 25-75K Above Top Tax Cutoff

70 Regression-Kink Design Have to compare individuals with similar incomes to have adequate precision on taxable savings outcome Exploit sharp change in marginal incentives at top bracket in 1999 Does marginal propensity to save in taxable accounts rise when individuals cross into top bracket after 1999? Implement using regression-kink design Estimate slope d[saving]/d[income] in DKr 75K window above and below top tax cutoff Plot difference in slopes above vs. below top cutoff by year

71 Total Pensions RK Coefficient Change in Marginal Propensity to Save in Retirement and Non-Retirement Accounts at Top Tax Cutoff by Year Taxable Savings RK Coefficient Year Total Pensions Taxable Savings

72 Total Pensions RK Coefficient Change in Marginal Propensity to Save in Retirement and Non-Retirement Accounts at Top Tax Cutoff by Year Taxable Savings RK Coefficient Year Total Pensions Taxable Savings

73 Change in Total Pensions (DKr) Change in Total Pension Contributions Post-Reform ( ) minus Pre-Reform ( ) Income Relative to Top Tax Cutoff (DKr) Change in Slope at Cutoff = / 1000 (1.2) Total Pensions Pass-Through Rate Δ Tot. Pens. / Δ Cap. Pens.: β = 0.48 (0.05)

74 Change in Taxable Savings (DKr) Change in Taxable Savings Post-Reform ( ) minus Pre-Reform ( ) Change in Slope at Cutoff = 5.9 / 1000 (2.8) Crowd-Out of Pension Contribution ΔTaxable Saving / Δ Pension Contrib.: β = (0.67) Income Relative to Top Tax Cutoff (DKr)

75 Change in % with Taxable Savings Above Median Change in Fraction with Above-Median Savings Post-Reform ( ) minus Pre-Reform ( ) Income Relative to Top Tax Cutoff (DKr) Ch. in Slope at Cutoff = 0.019% / 1000 (0.004%) Crowd-Out of Pension Contribution ΔTaxable Saving / Δ Pension Contrib.: β = (0.22)

76 Crowd-Out Estimates Consider impacts of a DKR 1000 increase in pre-tax income DKR 10.0 less contributed to retirement accounts when subsidy fell MTR of 60% disposable income rises by 0.4 x 10.0 = DKR 4.0 DKR 3.92 of this comes from reduced taxable savings DKR 0.08 from reduced consumption net saving up by DKR % of the increase in pension contributions due to subsidies is financed by offsetting reductions in savings in taxable accounts Based on this estimate, we calculate that each DKr 1 of expenditure on subsidies raises total saving by less than 1 cent

77 Pension Subsidy Pass-Through Estimates Difference-in-Differences Regression Kink Annuity Contrib. Total Pensions Contrib. Taxable Savings Taxable Savings Taxable Savings Threshold (1) (2) (3) (4) (5) Capital Pension Contrib (0.047) (0.047) Total Pension Contrib (2.453) (0.665) (0.222) No. of Obs. 4,697,656 4,697,656 4,697,656 7,011,068 7,011,068

78 Impacts of Government Policies on Savings for Active vs. Passive Savers Automatic Contribution Price Subsidy Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Raises Pension Contribs. M+P? Raises Total Savings M+P+S? Active Savers No No Yes Uncertain Passive Savers Yes Uncertain No No Data Yes Yes Yes No

79 Part 3 Identifying Active vs. Passive Savers

80 Heterogeneity: Active vs. Passive Savers Use pass-through of automatic contribs. and extensive-margin response to subsidy to quantify degree of active response Passive ESP MSP SUBSIDY Active response to M and t Employer Pensions: 1 minus pass-through = 16.8% Mandated Savings Plan: 1 minus pass-through = 13.8% Subsidy Reduction: fraction who reoptimize pension = 17.4%

81 Heterogeneity: Active vs. Passive Savers Are differences between impacts of automatic contributions and subsidies driven by active vs. passive choice? Test the model by analyzing 3 predictions about heterogeneity

82 Heterogeneity: Active vs. Passive Savers Are differences between impacts of automatic contributions and subsidies driven by active vs. passive choice? Test the model by analyzing 3 predictions about heterogeneity 1. [State dependence] Individuals should respond more to subsidy when actively changing pensions for other reasons Test if new pension contributors allocate more money to annuity accounts than capital accounts than prior contributors

83 Impact of Subsidy Change: Old vs. New Contributors Dep. Var.: Contributes to Capital Pension (1) (2) Post 1999 Post 1999 x New Saver (0.009) (0.006) (0.006) (0.006) Controls X No. of Obs. 142, ,998

84 Heterogeneity: Active vs. Passive Savers Are differences between impacts of automatic contributions and subsidies driven by active vs. passive choice? Test the model by analyzing 3 predictions about heterogeneity 1. [State dependence] Individuals should respond more to subsidy when actively changing pensions for other reasons 2. [Types] Individuals who optimize actively should respond to subsidy and undo automatic contributions to a greater degree Proxy for active optimization: frequency of changes in pension contributions in other years

85 Percent Responding to Capital Pension Subsidy Change in 1999 by Frequency of Active Changes in Other Years % Exiting Capital Pension and Raising Annuity in b = (0.005) Percentage of Other Years with Change in Individual Pension Contributions

86 Pass-Through of Emp. Pensions to Total Pensions Pass-Through of Employer Pension Changes for Firm-Switchers by Frequency of Active Changes in Other Years b = (0.005) Percentage of Other Years with Change in Individual Pension Contributions

87 Heterogeneity: Active vs. Passive Savers Are differences between impacts of automatic contributions and subsidies driven by active vs. passive choice? Test the model by analyzing 3 predictions about heterogeneity 1. [State dependence] Individuals should respond more to subsidy when actively changing pensions for other reasons 2. [Types] Active savers should respond to subsidy and undo automatic contributions to a greater degree 3. [Observable heterogeneity] Active savers likely to be those who are already saving and planning for retirement Carroll et al. (1999): individuals with high discount rates don t save for retirement and also put off financial planning

88 % Exiting Capital Pension and Raising Annuity in 1999 Heterogeneity in Response to Capital Pension Subsidy by Wealth/Income Ratio b = 7.1 (0.4) Wealth/Income Ratio in 1998

89 Pass-Through of Employer Pensions to Total Savings Heterogeneity in Pass-Through of Employer Pensions by Wealth/Income Ratio b = (4.10) Wealth/Income Ratio in Year Prior to Switch

90 % Extensive Margin Substitution in 1999 Heterogeneity in Responses to Subsidies and Employer Pensions by Age Pass-Through of Employer Pensions to Total Savings s 30s 40s 50s Age in Year of Subsidy Change (1999) or Firm Switch Subsidy Response

91 % Extensive Margin Substitution in 1999 Heterogeneity in Responses to Subsidies and Employer Pensions by Age Pass-Through of Employer Pensions to Total Savings 20s 30s 40s 50s Age in Year of Subsidy Change (1999) or Firm Switch Subsidy Response Employer Pension Pass-Through

92 Observable Heterogeneity in Response to 1999 Subsidy Reduction Dep. Var.: Exits Capital Pension and Increases Annuity Contribution in 1999? (1) (2) (3) (4) (5) (6) W / Y Ratio (0.004) (0.004) (0.004) (0.004) (0.004) Age (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) College (0.004) (0.003) (0.004) Economics Education (0.007) (0.007) Controls X No. of Obs. 62,641 62,641 62,641 62,641 62,641 62,641 Mean of Dependent Variable: 11.6%

93 Conclusion Two important strands of research have developed independently Crowd-out of retirement savings in non-retirement accounts Impacts of active vs. passive policies within retirement accounts These two issues are closely related Degree of crowd-out depends fundamentally on whether savings change is made actively or passively

94 Conclusion Automatic contributions raise savings significantly via passive changes Tax subsidies less effective at raising savings for three reasons 1. Infra-marginal spending on inattentive agents 2. Crowd-out rates high among attentive agents 3. Attentive agents better prepared for retirement to begin with United States flow tax expenditure on tax-deferred savings accounts exceeds $125 billion [JCT 2012] Are these policies the best way to raise retirement savings?