Faculty-Student Team Explores Social Security Benefit Valuation and Risk
Social Security benefits continue to make up a substantial portion of most Americans’ retirement portfolios. With many people depending on this program during their golden years, understanding the associated value and risks of future Social Security payments is of the utmost importance. But just what are the associated risks for recipients based on their health profile and wage history?
This is the question an NJIT faculty-student team, led by Assistant Professor of Finance Steve Taylor and Assistant Professor of Accounting Ming Fang Taylor, set out to answer. The students on board represented an interdisciplinary and interschool collaboration and included Yassmin Ali, a computer science major; Xun Wang, a business data science Ph.D. student; and Pablo Sota ’19, now an applied mathematics/applied physics Honors alumnus with an interest in finance and statistics. The group was supported by a seed grant from the Henry J. and Erna D. Leir Research Institute for Business, Technology, and Society at Martin Tuchman School of Management, and published a paper about its findings, “Social Security Benefit Valuation, Risk, and Optimal Retirement,” in the journal Risks. Late last year, Taylor and Sota presented the paper at the CMStatistics 2019 conference at University of London in England, alongside such notable schools as Columbia, Harvard, Massachusetts Institute of Technology and Johns Hopkins.
A key step in the team’s research involved principal component analysis (PCA) of the yield curve on interest rates that is used to discount future cash flows. PCA is a data analysis tool that reduces the dimensionality of data by identifying the most relevant aspects of a data set. In this case, out of 50 interest-rate-swap time series, where fixed interest rates are exchanged for floating rates, the two-year, 10-year and 30-year swap rates were particularly important because, as Taylor pointed out, they are frequently traded swap tenors (or lengths of a swap) that constitute the backbone of the yield curve.
The team also developed a Python program that included present-day value of future Social Security cash flows, life expectancy forecasting and associated-risk methods, and worked together through a shared GitHub repository, a web-based version-control platform. Additionally, students learned to use PuTTY/SSH in a Linux environment, a tool that enabled them to securely access a remote workstation in Taylor’s office; and cron, a software utility to schedule routine tasks.
One significant finding concerns the impact of a major-disease diagnosis on the value of Social Security benefits. “If one acquires a disease which results in a 10-year life expectancy from diagnosis, this results in nearly an 80% loss … in value of future Social Security benefits. A disease with very high mortality rates and short life expectancy, like pancreatic cancer, results in a near total loss,” Taylor explained. “Generally speaking, we found that it typically makes sense to initiate Social Security benefits at age 62 in the event one is diagnosed with even a relatively benign disease that has an average life expectancy less than 15 years.”
Still, while there are instances where an earlier start is justified, some two-thirds of the population begins receiving Social Security benefits at age 62 or 63 as opposed to age 70 when payments are at their highest, the team discovered. The move, said the team, often stems from poor financial advice and/or decision-making.
Overall, “we find that the average American, in terms of both wage history and health profile, has a total benefit value of approximately $360,000, whereas numerous sources have reports that inflation-adjusted individual and employer contributions are only approximately $250,000, which is not a good sign for the long-term health of the program,” noted Taylor.
Looking ahead, the team would like to create a public calculator and dashboard visualization that enables estimating the value of future Social Security benefits for asset allocation purposes. These tools would integrate survival curves for major diseases with high mortality rates as well, to better advise on the best time to start receiving Social Security benefits. Also on the to-do list is extending the team’s research methods to investigate those troubling solvency issues.
“This project provided a means for us to branch out from our typical, more theoretical quantitative finance research into an applied problem that is of interest to tens of millions of Americans. We want to continue working on understanding the risks inherent in entitlement-related programs and especially how they may be improved,” Taylor said.
Wang, who appreciated the opportunity to participate in the project, and along with the other students gained practical technology skills useful for employment in the tech industry, said she learned a lot — from collecting, cleaning and sorting data, to developing skills in Python, to collaborating with others. “As a Ph.D. student … [it is all] very valuable and meaningful for my future project and dissertation.”
