Social Security Literacy and Retirement Well-Being Hugo Benítez-Silva (SUNY-Stony Brook) Berna Demiralp (Old Dominion University) Zhen Liu (SUNY-Buffalo) UM09-11 Introduction and Motivation Financial Literacy has become an even more important issue with the events of the last year and a half Social Security Literacy is still a relatively unexplored topic, while a high level of knowledge by individuals is an underlying assumption of most retirement models Anecdotal evidence indicates the level of knowledge is not very high Contribution We provide evidence of the level of basic knowledge about the Social Security Retirement System using two waves of a Survey funded by Stony Brook and MRRC The second part of this project, assesses the effect on Well-Being of the lack of knowledge about the system using a Dynamic Life-Cycle Model of Retirement We are also able to impute the cost of acquiring knowledge SB 2007-2008 Social Security Study Two telephone surveys conducted at Stony Brook’s Center for Survey Research during August 2007 (funded with a Seed Grant from the V.P. for Research at SB), and December 2008 (funded by MRRC) Samples of 500 and 507 observations, nationally representative. 179 observations of the 2008 survey were re-interviews What is the youngest age at which an eligible worker can apply for his or her own Social Security retirement benefits? (All respondents 2007 and 2008) Response Less than 50 51 to 61 Exactly 62 63 to 64 Exactly 65 Over 66 Don't know Total Frequency Percent 52 108 298 15 83 37 28 621 8.37 17.39 47.99 2.42 13.37 5.96 4.51 100 Cummulative percent 8.37 25.76 73.75 76.17 89.53 95.49 100 What is the youngest age at which an eligible worker can apply for his or her own Social Security retirement benefits? Sample Response 2007 2008 new respondents respondents Less than 50 51 to 61 Exactly 62 63 to 64 Exactly 65 Over 66 Don’t know # Obs. 9.57% 19.14% 41.98% 2.47% 15.43% 5.56% 5.86% 324 7.5% 15.5% 54% 3% 10% 6% 4% 200 2008 reinterviewed respondents 6.19% 15.46% 55.67% 1.03% 13.4% 7.22% 1.03% 97 Total 8.37% 17.39% 47.99% 2.42% 13.37% 5.96% 4.51% 621 What is the earliest age of retirement at which Social Security would pay you full, unreduced benefits? 80 % of respondents 70 65.74 69 69.07 60 50 40 30 20 19.44 19 18.56 10.31 8.33 10 9 6.48 3 2.06 0 Below 65 65 to 67 Over 67 Don't know Responses 2007 respondents 2008 new respondents 2008 re-interviewed respondents What is the maximum age at which you can claim Social Security retirement benefits so that Social Security will adjust your benefits upward [because of the delay in claiming benefits]? 50 45.36 45 % of respondents 40 35 32.5 30.86 30 22.8423.5 25 21.6 20.5 20 16.5 14.2 15 10.49 10 21.65 14.43 12.37 7 6.19 5 0 Below 65 65 to 69 Exactly 70 Over 70 Don't know Responses 2007 respondents 2008 new respondents 2008 re-interviewed respondents What is the minimum number of working years that qualify you to receive Social Security Retirement Benefits? 40 34.5 35 32.99 % of respondents 29.9 30 25.62 25 20 23.15 21.5 19.44 19 17.28 14.51 14.43 12.5 15 12.5 11.34 11.34 10 5 0 Less than 10 Exactly 10 11 to 20 More than 20 Don't know Responses 2007 respondents 2008 new respondents 2008 re-interviewed respondents If you earned $10,000 in a given year from working after you began receiving Social Security Retirement Benefits at age 62, do you think your Social Security benefits would be reduced? 60 52.16 50.5 % of respondents 50 50.52 48.45 45 39.2 40 30 20 8.64 10 4.5 1.03 0 Yes No Don't know Responses 2007 respondents 2008 new respondents 2008 re-interviewed respondents What about if you earned $20,000 in a given year from working after you began receiving Social Security Retirement Benefits at age 62, do you think your Social Security benefits would be reduced? 90 77.32 80 70.99 % of respondents 70 66 60 50 40 30 26 19.75 20 18.56 9.26 10 8 4.12 0 Yes No Don't know Responses 2007 respondents 2008 new respondents 2008 re-interviewed respondents Knowledge of the Early Retirement Age by age categories .8 1 1.00 0.72 .6 0.65 0.53 0.46 0.43 0.32 0.17 0.20 0 .2 .4 0.45 18-34 35-44 45-54 2007 55-64 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) 65 or more Knowledge of the Early Retirement Age by educational levels .6 0.62 0.49 0.48 0.47 0.36 0.36 0.19 0 .2 .4 0.44 Less than HS HS graduates 2007 Some college Bachelor's and above 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) Knowledge of Maximum Retirement Age by age categories .6 0.60 .4 0.40 0.38 0.36 0.33 0.20 0.18 0.17 0 .2 0.23 0.20 18-34 35-44 45-54 2007 55-64 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) 65 or more Knowledge of Maximum Retirement Age by educational levels .4 0.39 0.32 .2 .3 0.30 0.31 0.18 0.15 .1 0.14 0 0.05 Less than HS HS graduates 2007 Some college Bachelor's and above 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) Knowledge of the Early Retirement Age by income levels .6 0.62 0.55 0.54 0.53 0.49 0.45 0.42 0.19 0 .2 .4 0.43 0.49 Less than $35K $60K to $100K $35K to $60K Refused Over $100K 2007 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) Knowledge of Maximum Retirement Age .4 by income levels 0.37 0.31 .3 0.29 0.26 0.27 0.25 0.24 0.17 0 .1 .2 0.21 0.18 Less than $35K $60K to $100K $35K to $60K Refused Over $100K 2007 2008 Source: SB 2007/2008 Social Security Study (all respondents from each year) The Dynamic Model (I) We solve and simulate a Dynamic Life Cycle Model of Retirement Behavior under uncertainty Individuals face life uncertainty, income uncertainty, health uncertainty, and employment uncertainty, and optimally choose how much to consume, save, work, and when to claim retirement benefits We model in detail the U.S. retirement system and add unemployment insurance The Dynamic Model (II) The benchmark model does an excellent job at replicating the actual data on the proportion of individuals who claim benefits at each age. This model assumes individuals know and understand the complex incentive structure of the retirement system in the United States. We have shown this is hardly a realistic assumption, but can we assess whether the pervasive lack of knowledge is costing us much in terms of welfare. Actual vs. Simulated Retirement Claiming Age Actual (in %. 2006) Simulated (%) 62 53.8 52.93 63 8.56 15.84 64 10.4 11.16 65 22.3 17.82 66+ 2.7 2.22 The Dynamic Model (III) One of the great advantages of utility based lifecycle models is that we can use them to perform welfare calculations, in which we compare the well-being of agents under different regimes We can, for example, use the model to compare the well-being under two different incentive structures, or as in this research under two different informational structures We can compare the benchmark model with a variety of possible informational structures The Dynamic Model (IV) We have re-solved and re-simulate the model assuming that individuals do not know the details of the retirement system, and behave mimicking the people around them. This means, for example, what when an individual reaches age 62, he does not make any complicated calculation to decide whether she should claim benefits, but behaves like the average person, who claims 52% of the times at that age. We know this behavior is suboptimal, but the key is to compute the welfare gain from information, and with that impute the cost that the individual is willing to bear to access that better level of information Results of the Model We find that a large proportion of individuals would gain from full-information. This proportion varies by age. Around 95% of those who are 60 years old would sharply benefit from better information. But only 28% of those who are 40. The welfare gain is computed in terms of willingness to pay out of the current level of wealth. Technically, we find that an average 60 year old who knows the rules of the system would have to be compensated with more than 50% of his current wealth in order to accept the suboptimal behavior we have explained. The result hinges on the particular informational structure we have assumed, but many others can be studied like for example assuming the consequences of thinking that the ERA is 63 or 64. A ranking of informational structures we are considering Complete ignorance (e.g. “I believe the youngest age cannot be anything other than 65”) Ignorance with awareness about ignorance (e.g. “I think it is 65, but I could be wrong.”) Ignorance with a prior (e.g. “I guess it is 65 with 50% chance being correct”) Accurate knowledge (e.g. “I am sure it is 62 and I have double checked”) How might incentives work with various informational structures Copying the majority around As in the current model Optimizing with given information Using given knowledge as constraints Searching for extra information Needs further assumptions on search costs Hiring professionals Needs further assumptions on agency costs Conclusions We find Americans have a limited knowledge of some of the basic rules of the Retirement System We find that a large proportion of Americans would strongly benefit from access to good information compared with suboptimal “me too” strategies which do not take into account the individual’s particular circumstances Informational structures can be key to assess the appropriateness and consequences of the policy changes analyzed in our models