The ability of retirees to optimally manage their assets and debts influences their income security and well-being in retirement. This may be especially true for low income households who can quickly exhaust their wealth if they poorly manage their portfolios or if they had not adequately saved for retirement. An alarmingly high fraction of elderly individuals in the general population are documented to have low level of debt literacy and experience financial distress. In partnership with North Carolina State University, the Institute of Consumer Money Management is funding research into the financial well-being of state and local retirees in North Carolina.
This paper provides novel evidence regarding wealth and debt management among recent retirees from state and local employment, who are heavily insured in retirement. All retirees in the sample have a base retirement income from a defined benefit pension and Social Security and many have contributed to the employer-provided retirement saving plans offered by the state.
The primary goal of the research is to enhance our understanding of how low-income retirees manage and utilize their retirement savings in the drawdown phase of life. We are particularly interested in the role of financial literacy on asset management and financial distress. If lower income households have lower levels of financial literacy, then disparities in financial wellness in retirement may stem both from inadequate savings while working and from poor asset management in retirement.
Timeline
- September – Mid-October 2019: Creating a preliminary version of the HRS dataset.
- Mid-October – End of October 2019: Finalize the HRS dataset.
- November – Mid-December 2019: Learning the method by review the literature on applications of neural networks and look for programming packages in R that will be suitable for this project.
- Mid-December – End of December 2019: Producing a report on the models studied, with information on the advantages and disadvantages of each model and how to implement them in R. Based on the report, make a detailed plan with to-do list and timeline for the next two phases.
- January 2019: First attempt at application. Researchers will begin by doing unsupervised learning analysis on the HRS dataset.
- February – April 2020: Implementing the models. Researchers will train the unsupervised ANN models and look for patterns of correlation between demographics, work type and status, household income, and credit card debt.
- May – June 2020: Adapting the ANN model to work with the NC datasets.
- July – August 2020: Comparing results both between datasets and between ANN and conventional regression models
- September – October 2020: Write up the results into a research paper.