Introduction
Relatively little is known about why people do not successfully plan for retirement and whether planning and information costs might affect retirement planning behaviors. This paper reports on the study of financial literacy and its effect on individuals' retirement planning behaviors. I show that financial literacy is widespread among an array of individuals, most nearing retirement age. They use several methods including: retirement calculators, financial advising experts, and retirement meetings or seminars. Many elderly Americans are unsure about their efforts to adequately prepare for retirement; one-third of adults nearing their 50s have yet to develop a retirement plan of any kind (Lusardi 1999, 2003; Yakoboski and Dickemper, 1997). The question is raised as to why. What explains such a low level of planning for retirement?
In Section II, there is an analysis of past literature pertaining to related studies. There are comparisons with those authors' finding and the findings of this paper. Section III contains descriptions of the sample, the explanatory variable, and the outcome variable. Section IV shows the regression equation of the key variables and control. It also holds the interpretations made from the analysis. Section V includes the conclusion and review of analysis. Section VI displays graphs and tables relative to the study. These sections are followed by a works sited page.
Section II
There is a plethora of research on retirement planning behaviors and factors that impact individuals' said behaviors. One work by Oliva Mitchell and Gary Fields (1984) is "The Economics of Retirement Behavior" which serves the purpose of investigating economic factors and the impact they have on determining retirement behavior – most specifically pension plans. Mitchell and Fields use a subsample from the Benefit Amount Survey, which was developed by the US Department of Labor. They used this data set primarily because most data sets include individuals that are not yet at retirement age; their data included only individuals that had reached the mandatory age to receive their pensions. Because of this, their results were free of "censored spells" problems that are faced by other labor-force modelers. The data consisted of 8,733 males who retied between the ages of 60 and 68. They observed that retirement behavior is due in part to worker preferences and differences in income opportunities; because of this, Mitchell and Fields concluded that retirement age is highly dependent on pension rules and income opportunities that come with those rules. An argument that I make concerning this paper is that they factor worker preferences into their results. A person's individual preferences regarding how long they continue to work and what causes them to stop working can include factors that are unobservable, making it their conclusion questionable.
Robert Clark, along with Robert Hammond, Christelle Khalaf, and Melinda Sandler Morrill wrote about the preferences specifically, "Planning for Retirement? The importance of time preference." (2017). They explore how worker's characteristics and preferences are associated with planning and saving for retirement. In doing so, the authors assess the quality of retirement planning and how consistent the behavior of observed individuals follows the plan created for them. The study explores different measures of planning, supplemental plan participation, contribution levels, planned retirement age, and plans for working after retirement. Clark et al use a data set derived from a survey of public sector workers in North Carolina, merged with corresponding administrative records containing information from each employee – including earnings, job information, agency type, years of service, and age – from the North Carolina Retirement Systems Division. The respondents were asked questions that assessed each individuals' financial literacy objectively, in addition to questions that pertained to self-assessed measures of financial knowledge. The authors concluded that the way individuals save and prepare for retirement is associated with the extent of their retirement planning. Individuals who actively participate in retirement planning are more prepared to meet their intended retirement goals as they reach the end of their time at their career job. The authors also take in to account whether planners have a college degree or not and found that individuals with a degree are more likely to plan for retirement. Clark et al found through their research that 73% of their sample planned to work after retirement – whether it be out of necessity or pure want, but the average age of planned retirement was 63. For preference elicitation, they established the predictive power of risk and time preference in explaining the full range of economic behavior observed. On risk preference, respondents were given the option of changing to one of two jobs – one with constant income and one with an income that is 100% higher or 20% lower. It was shown that 67.5% would chose the safer, more risk averse, job. Next, respondents were asked two questions to assess their level of patience. The respondents were given the option of $1,000/month from Social Security or $500/month plus an up-front lump sum of $85,000; 34.2% chose the larger benefit (more patient), while 49.7% chose the up-front payment (less patient). Because of this, combined with the measure of risk, Clark et al concluded that 62.7% of respondents were more patient; this conclusion led them to their next: planners are slightly more risk averse, but the small difference was statistically insignificant. They also conclude that behavioral aspects of decision making tell us the importance of how individuals prepare for retirement. I find this paper to be informative and well-devised; it follows closely with the factors that I consider in both the explanatory and outcome variables to find a similar conclusion. Individuals that are more education on finances tend to plan for retirement more than the individuals who are less informed or less educated in general.
Alan Gustman and Thomas Steinmeier have several papers on retirement behavior. One is "Retirement and Wealth" (2001). Gustman and Steinmeier observe reduced retirement and wealth equations to analyze the relationship between them. They primarily focus on how individuals perceive the future value of pension and social security income and the effect it has on retirement behavior. Using data from the first four waves of the Longitudinal Health and Retirement Study for individuals born from 1931 to 1941(approximately 5,600 individuals after eliminating all respondents that did not meet their set criteria), Gustman and Steinmeier sketched a theoretical structure that generated relationships between retirement and wealth income in accordance with the correlation between leisure and time preference. They then conducted empirical tests to determine whether the parameters obtained in reduced form retirement equations were likely to be useful for behavior analysis. Gustman and Steinmeier also considered whether exogenous factors effect retirement and wealth equations. From their studies they concluded that heterogeneity in time preference, combined with liquidity constraints that bind some individuals implies that individuals with high time preferences and imperfect ability to borrow money may value future income from pensions and social security much less than the amount calculated from market interest rates. They also concluded that the reason many fail at successfully planning for retirement pertains to the fact that many individuals value the future benefits to be far less than expected.
Another paper by Gustman and Steinmeier, "Employer Provided Health Insurance and Retirement Behavior" (1993) analyzes the effects employer provided health benefits on retirement. They look at the value placed on retiree health coverage benefits and how long employees wait to retire. By focusing on three main questions throughout the paper – What is the size of the effect of retiree health insurance on retirement outcomes at different ages? What is the effect of having omitted retiree health insurance from retirement models on the parameters that have been estimated? What are the likely effects of changes in the availability or value of retiree health benefits on retirement behavior in the future? – they are able to conclude that retiree health benefits delay retirement until the age of eligibility, and then it is accelerated. In this paper, Gustman and Steinmeier use data from employer provided health insurance retrieved from the 1969-1979 Retirement History Study, which is the only available nationally representative longitudinal data set which contains the required information for a structural retirement analysis. Gustman and Steinmeier found that 78% of current full-time employed males over 40 had employer provided health insurance; of females in the same situation, 16% less received employer provided health insurance. For part-time employees, the percentage was almost 50% lower. Of the males who currently received health benefits, 51% expected to continue to receive said benefits during retirement; of the females who currently received health benefits, only 42% expected to continue to receive them in retirement. They also concluded that employees with no retiree health coverage had accelerated retirement ages when presented with the option to receive retiree health insurance. Also, they found that incentives created by retiree health insurance, as opposed to health insurance for active employees are responsible for most of the effects of health insurance on retirement outcomes. Because of the size of the value health insurance creates relative to the wages received, and because there are opposing effects of retiree health insurance before and after the age of eligibility for benefits, the overall effects on retirement incentives were modest. Gustman and Steinmeier found that due to the observation, the resulting effects on retirement behavior were small.
Section III. Data
The sample used in this study is a cross-sectional sample from the National Longitudinal Surveys (NLS). These consist of surveys collected by the United States Department of Labor, more specifically, the Bureau of Labor Statistics. It is an accumulation of surveys that track the labor market and other important life experiences of men and women in America. For this analysis, the National Longitudinal Survey of Youth 1979 (NLSY79) is used. The NLSY79 contains a sample of approximately 12,700 individuals born between 1957 and 1964; these individuals were between the ages of 14 and 22 when the surveys were first conducted in 1979, and they have been surveyed annually since. The format of these surveys is such that confidentiality is installed and participants feel less inclined to report false information. This data set accurately represents the United States population as random sampling was assigned. To perform an analysis of financial literacy's effect on retirement planning, the time interval of 2006-2014 is pooled, generating blank observations, or blank males and blank females approximately blank to blank years old. This is ideal for this study as the parameters of interest require laborers to be nearing the typical age of retirement action. It contains data on current and past retirement planning habits, lack thereof, and expected retirement outcomes.
Using NSLY79, explanatory variables are gathered on individuals' idea of what retirement is considered: working for fun, reduced work hours, stopping work all together, reduced spouse work hours, receiving Social Security or pension income, as well as expected retirement age. Summary statistics of the surveyed individuals' ethnicity and gender can be found on Table 1. Of the number observed, percent are males and percent are females. The ethnicity variable is a dummy variable representing three subsets: White, Black, and Hispanic. Respondent's age is factored continuously over the time interval observed. For this analysis, missing values are not calculated in. The outcome variable (retirement planning) is a dummy variable created based on the actions of survey respondents from 2006 to 2014. Factors in this variable include whether or not the respondent has consulted with a financial planner regarding retirement or used a computer software to plan their retirement, calculated their needed income during retirement, attended retirement meetings, read about retirement, or considered what age they wish to retire.
Section IV. Results
In this paper, we infer a significant correlation between financial literacy and planning for retirement. As described, this analysis hypothesizes the extent of financial literacy and its impact on whether or not individuals plan for retirement. We will take into account the unobservables (U) but assume that it has no effect on the explanatory variables. Since we know that U having a mean independence of x is false, we can conclude that U could have many factors (i.e. time preferences, work ethic); we will not take them into account for the estimator.
E[U/(financial literacy)]=0
From this we can assume a constant variance between U and X.
The estimation begins with an OLS equation using covariates to observe the effects the independent variable (which I have created a dummy for) have on retirement planning. For clarification purposes, I will break down the independent variable into its original variables.
Retirement Planning = β0 + βfinancialliteracy
expands to
Retirement Planning= β0 + βreduceworkhrs + βstopworkcompletely + βreceivepension
+ βreceivessincome + βworkforfun + βspouse_nowork + βspouse_reduceworkhrs
For men and women, we test for statistical significance at α = 0.05. Not surprisingly, we see that there is a very strong positive correlation between being financially literate and planning for retirement. For the full sample, being financially literate gives an 87.8% chance of being more likely to plan for retirement. With men, we can see that that their ethnicity can have a negative correlation with planning for retirement. If the subject is black, he is .820% less likely to plan for retirement. For a hispanic male subject, there is a 1.86% percent less chance that he will plan for retirement. Interestingly, age for both male and female respondents only show a .226% chance of increasing their retirement planning every year – meaning the older respondents get, the less they care about planning for retirement. While women still have a 88.3% chance of planning for retirement, their ethnicity also plays a role in determining their planning habits. Black female respondents are slightly less likely to plan for retirement, while hispanic female respondents are slightly more likely to plan for retirement; this is slightly different than the observation made for men. For this analysis, it can be inferred that ethnicity does not have a significant effect on individuals' retirement planning actions. All of this can be seen in Table 2.
If we look at the factors considered for retirement planning, there are a couple of interesting observations to be made. Hispanic respondents are 10.5% more likely calculate their needed retirement income than black respondents. Females are 1.7% more likely to consult financial planners than males. Females are also 6.4% more likely to attend meetings about retirement.
Observing the factors considered for financial literacy, black respondents are 7.5% more likely to consider retirement a reduction in work hours than hispanic respondents. Male respondents are 0.8% more likely to expect a pension during retirement. There is not a significant difference in the expectation of receiving Social Security income or expecting to work during retirement for fun. These observations can be found in Table 3.
Section V. Conclusion
Based on the nominal figures of the sample, individuals who are financially literate are much more likely to plan for retirement. The regression analysis does not show much correlation between ethnicity, gender, or age of respondents and the effect it has on their retirement planning habits. With a regression that has an r-squared value of 0.766, we can infer that our model does indeed accurately predict whether someone who is financially literate is more likely to plan for retirement. There are habits that appear to be more prominent in particular ethnicities and genders in comparison to others – specifically that hispanic respondents are 10.5% more likely to calculate needed retirement income compared to black respondents, and female respondents are 6.4% more likely to attend meetings regarding retirement than male respondents. In comparison to others who have studied similar cases, the results match almost all past findings.
Section VI