With the entire Baby Boomer cohort expected to reach retirement age by 2030, the ageing population is noticeably increasing the proportion of older workers in the labour force. Concurrently, labour force participation rates among older individuals are also witnessing an upward trajectory. Specifically, the labour force participation rate for those aged 75 and above is projected to increase from 8.9% in 2020 to 11.7% by 2030 (Bureau of Labour Statistics, 2021). This growing trend of increased labour force participation among the older generation, who are traditionally expected to retire, prompts us to consider how workers' expectations and retirees' actual experiences align. An Employee Benefit Research Institute (EBRI) report in 2022 underscores a substantial disparity between workers' retirement expectations and retirees' realities concerning retirement preparedness and income sources.
The EBRI report 2022 reveals that workers who hadn't contemplated working during retirement re-entered the labour force for paid employment. The motivations for seeking paid work in retirement varied, with financial reasons—such as meeting basic needs, fulfilling discretionary spending desires, or mitigating fears of dwindling savings or investments—emerging as significant drivers. An apparent gap in retirement preparedness is evident, with 56% of workers lacking confidence and uncertainty about their retirement date, in contrast to only 21% who express high confidence. This discrepancy underscores considerable unpreparedness in retirement savings among a substantial portion of workers.
Calculating retirement financial needs and employing financial products guaranteeing retirement income can alleviate uncertainties surrounding retirement for workers and prevent retirees from re-entering the labour force due to inadequate savings, excessive withdrawals, or fear of depleting savings. Individuals should possess some form of insured income in retirement to align with their retirement expectations.
Life insurance emerges as a valuable tool aiding families in achieving two pivotal objectives. Firstly, it furnishes financial security in the event of the primary wage earner's demise. Secondly, it facilitates households' long-term savings goals, even amidst uncertainties (Beck & Webb 2003). There's a well-documented link between life insurance and how households allocate their assets (Lin & Grace 2007). Pioneering research by Fortune (1973) suggests that life insurance can act as a substitute for other financial holdings, such as stocks or lower-risk investments. Headen and Lee (1974) further propose a positive correlation between higher net saving rates and life insurance demand, particularly for individuals with fewer assets who view life insurance as a primary alternative. In addition, Lin and Grace (2007) suggest that retirement savings accounts are complementary to total life insurance, especially for younger and middle-aged individuals. These studies highlight the interplay between life insurance and asset allocation strategies across different demographics and financial situations.
Building on the established link between retirement savings accounts and total life insurance ownership, it's worth considering how life insurance itself might influence retirement income expectations. Both term and cash-value life insurance could potentially contribute to greater retirement income satisfaction. Cash-value life insurance serves a dual purpose, acting as both savings and insurance. It can be used to fulfil long-term financial goals like leaving a bequest, covering medical expenses in later life (De Nardi et al. 2010), or meeting other significant financial obligations.
Term life insurance, although purely a form of insurance, can also influence retirement expectations. Workers with strong personal values may be willing to sacrifice current consumption to ensure the future well-being of their loved ones. Furthermore, term life insurance can protect a family's retirement savings from being depleted in the event of a catastrophic loss. As Bernheim et al. (2001) point out, life insurance can ensure that mortgages and other obligations are paid off upon the insured's death.
Hence, insurance is a key part of a comprehensive retirement plan as it is a contingent claim that pays out during periods of higher marginal utility of wealth. Retirement savings involve intertemporal decision-making, where individuals make choices that affect their well-being over time. By incorporating life insurance into their retirement plans, individuals make intertemporal trade-offs to protect their future financial well-being and the well-being of their beneficiaries. Life insurance ownership can also be seen as a risk management strategy to mitigate financial uncertainty in retirement, as it provides financial safety in case of unforeseen events such as premature death. Hence, this study investigates the relationship between life insurance ownership and retirement income rating, with a specific focus on the financial landscape and product offerings prevalent in the United States.
The landscape of employer-sponsored retirement plans has undergone significant transformations in recent decades. Historically, the responsibility of retirement savings rested with both employers and employees, with guaranteed retirement income promised by employers. According to the Congressional Research Service (CRS) report in 2021, the private sector retirement plans experienced a substantial shift from 1975 to 2019, with Defined Benefit (DB) plan participation declining from 27.2 million to 12.6 million, while Defined Contribution (DC) plans witnessed exponential growth. The number of active participants in DC plans surged from 11.2 million to a remarkable 85.5 million, marking a nearly sevenfold increase. Securing a financially secure retirement poses a multifaceted challenge for future retirees, necessitating meticulous calculation of retirement income needs, diligent savings, and navigating the dual threats of market volatility and potential outliving of retirement savings due to the evolving landscape of employer-sponsored retirement plans. Income insurance emerges as a valuable tool in mitigating these risks by offering a guaranteed income stream throughout retirement, irrespective of market fluctuations or longevity.
Households generally take two approaches to prepare for financial uncertainties: accumulating wealth and purchasing life insurance (Lugilde et al. 2019). While accumulating wealth provides a general financial buffer, life insurance serves as a more proactive strategy for managing a specific risk - the loss of a primary income earner. As illustrated by Lin and Grace (2007), household financial insecurity influences the total amount of life insurance purchased. Therefore, it can be safely argued that life insurance serves as a strong indicator of financial foresight, demonstrating a household's proactive approach to risk management and its commitment to long-term financial security.
However, life insurance policies are long-term financial commitments filled with considerable complexity. Households contemplating life insurance purchases may encounter challenges hindering their ability to make informed decisions, including knowledge gaps, affordability concerns, conflicting peer advice, and a lack of readily accessible, comprehensible information resources. Consequently, many households either forego life insurance altogether or fail to secure appropriate coverage.
According to a 2024 report by the Life Insurance Marketing and Research Association (LIMRA), the perceived cost of life insurance remains the most significant barrier to its purchase. The report reveals a substantial disconnect between perception and reality, with 72% of respondents overestimating the actual cost of a basic term life insurance policy. This overestimation appears particularly prevalent among younger generations, with over half of Gen Z and Millennials collectively misjudging the cost. Furthermore, the report highlights a troubling lack of informed decision-making. More than half (54%) of respondents base their cost estimates on intuition or rough guesses. This underscores the need for enhanced financial literacy and access to reliable information resources concerning life insurance costs.
The primary determinant of life insurance demand centres on the impact of the insured's death on the remaining household members' future financial well-being. This is particularly concerning for couples with financial obligations or dependents, as the insured's passing could disrupt their consumption patterns. Consequently, a positive correlation exists between financial vulnerability and the amount of life insurance purchased (Lin & Grace 2007). However, households extend their considerations beyond purely financial contributions when making life insurance decisions. Households consider the value of nonmonetary contributions in their insurance purchase, and personal values can play a significant role in life insurance purchase decisions (Bateman et al. 2023). For example, Chui and Kwok (2008) identified a positive correlation between life insurance consumption and societies characterised by higher individualism relative to collectivism.
An increasing number of studies also demonstrate that behavioural factors, such as loss aversion and narrow framing, affect a consumer's insurance purchase decisions. Consumers often make choices that aren't perfectly rational, especially when dealing with risks and probabilities, like those involved with life insurance (Do Hwang 2024; Gottlieb & Mitchell 2020; Huber et al. 2015).
Current research views individuals as boundedly rational. This means they are rational in the sense that they aim to maximise lifetime utility and understand the benefits of planning for uncertain futures. However, they are boundedly rational because gain-loss utility affects their decision. Hence, boundedly rational individuals derive utility not just from spending and leaving bequests, but also by considering the gains and losses associated with life insurance, consistent with Prospect Theory (Do Hwang 2024; Zheng 2020; Barberis et al. 2001).
Term life and cash value life insurance represent the two primary categories of life insurance products. A significant distinction between them lies in their treatment of the savings element and how boundedly rational individuals perceive them in terms of risk (Do Hwang, 2024). Term life primarily functions as pure insurance, providing coverage for a specified term. From the perspective of a boundedly rational individual, term life might be perceived as riskier due to the potential forfeiture of premiums if the insured survives the term. In contrast, cash value life insurance incorporates a savings component that accumulates over time, providing policyholders with access to cash value. Additionally, it guarantees a death benefit payout, making cash value life appear as a safer option for such a group of individuals, as they perceive a return on investment regardless of the timing of a death claim.
Income uncertainty exerts a profound impact on retirement preparedness, as evidenced by the EBRI report highlighting significant percentages of workers and retirees lacking confidence in their retirement, particularly concerning retirement income adequacy. Existing literature explores the association between income uncertainty and retirement savings decisions (Wesslen et al. 2021; Shin & Kim 2018), including the relationship between life insurance and savings in various aspects. For instance, some studies suggest that precautionary savings can partially substitute life insurance (Lugilde et al. 2019). Some suggest the complementary nature of retirement savings accounts and life insurance ownership (Lin & Grace 2007). Furthermore, a body of work examines the link between life insurance and household portfolio allocation (Do Hwang, 2024). However, there appears to be a gap in the literature regarding the association between retirement income expectation and life insurance ownership. This research gap presents a valuable opportunity to explore how life insurance ownership (including both types of term and cash-value life insurance) is associated with retirement income rating. Furthermore, given that the face value amount of life insurance is also associated with household portfolio decisions (Lin & Grace 2007), it will be interesting to explore the association between the face value amount of their life insurance policy and retirement income rating.
While traditional economic models often assume that individuals are perfectly rational agents who maximise expected utility—even when faced with imperfect information and uncertainty (Levin & Milgrom 2004)—a growing body of behavioural economics research challenges this view, especially in the context of complex financial decisions. Concepts like bounded rationality (Simon 1955) and Prospect Theory (Kahneman & Tversky 1979) offer a more realistic perspective on how people make decisions under uncertainty. Bounded rationality suggests that while individuals aim to make rational choices, their decisions are shaped by cognitive limitations, limited information, and time constraints. As a result, they often rely on heuristics rather than exhaustive analysis—a practical strategy in complex situations. Similarly, prospect theory highlights how people evaluate gains and losses relative to a reference point and tend to be loss-averse, feeling the pain of losses more strongly than the pleasure of equivalent gains. This study, while correlational in its empirical approach, seeks to explore associations that are consistent with these behavioural principles, providing insights into the psychological underpinnings of retirement income perceptions among life insurance policyholders.
The hypotheses for this study are as follows:
H1: Cash value life insurance is positively associated with retirement income rating.
Cash value life insurance, commonly referred to as whole or permanent life insurance, combines a death benefit with a savings component. This cash value accumulates over time on a tax-deferred basis, and policyholders can either borrow against it or withdraw from it, thereby providing a potential supplemental source of retirement income. Such features of cash value life insurance policies mitigate perceived losses (e.g., premium forfeiture) through a tangible savings component. This appeals to boundedly rational individuals who may outweigh immediate gains or exhibit loss aversion. In addition, the assurance of a return on investment regardless of the timing of a death claim enhances perceived safety and aligns with a positive gain-loss utility. As a result, individuals with this type of insurance might feel more financially secure regarding their retirement, leading to a higher retirement income rating.
H2: The face value amount of a life insurance policy is positively associated with retirement income rating.
The face value of a life insurance policy represents the sum that beneficiaries receive upon the death of the insured. A higher face value can indicate substantial financial foresight and planning. Individuals with larger face values on their policies may have a heightened sense of financial security, knowing that they have taken steps to provide for their beneficiaries. The psychological comfort that comes from preventing significant financial losses for beneficiaries is a powerful motivator, especially given the principle of loss aversion. This means that a larger policy value suggests a greater buffer against potential financial catastrophe, which then leads to a heightened feeling of security and, as a result, a more favourable retirement income rating.
We use data from the 2016, 2019 and 2022 Survey of Consumer Finances (SCF) provided by the Federal Reserve Board in cooperation with the Department of the Treasury. The SCF provides nationally representative information for households of all ages in the United States and contains detailed household information like income, assets and debts, pensions, and several demographic characteristics. The survey also collects information about life insurance policies and attitudinal questions about households' retirement income rating, which is the focus of this research. To deal with missing observations, SCF uses a multiple imputation technique to replace each missing value with a set of plausible values and generates five replicates. When working with imputed data, it is essential to correct for imputation error as it may overestimate the reliability of the analysis. Thus, after correcting for the imputation error, the final analysis sample contains 16,620 observations.
The outcome variable in this analysis is the retirement income rating. The SCF contains a Likert measurement of retirement income rating. Respondents were asked the following question: “How would you rate the retirement income you receive (or expect to receive) from all sources?” Possible responses ranged from 1 (totally inadequate) to 5 (very satisfactory). Based on this, the retirement income rating is an ordered, qualitative variable. Because the survey respondents were asked how they would rate the retirement income they are receiving or expect to receive, both retirees and pre-retirees are included in the analysis sample, but retired status is controlled for in the analysis.
Types of life insurance policies and the face value amount for each type are the main explanatory variables in this study. The SCF measured ownership of term life insurance policies by asking the question, “Are any of your (family's) policies term insurance?” In this analysis, term life ownership is a dichotomous variable, with a value of 1 if the household owns term life insurance, and a 0 otherwise. The SCF measured ownership of cash value life insurance policies by asking the question, “Do you have any policies that build up a cash value or that you can borrow on?” In this analysis, cash value life insurance ownership is treated as a dichotomous variable, with a value of 1 if the household owns cash value life insurance, and a 0 otherwise.
For the face value amount for term life, the SCF asks, “What is the current face value of all the term life policies that you (and your family living here) have?” For the face value amount for cash value life insurance, the SCF asks, “What is the current face value of all the policies that build up a cash value?” Based on the responses, the face value amount for both term life and cash value life is measured as a continuous variable. In addition, the natural logarithm of these variables is applied after converting the zero-valued observations to one. Doing so helped retain zero-valued observations and also perform the transformation on other observations.
Other explanatory variables used in this study are risk-taking attitude, retired, education, age, female, married, kids, race/ethnicity, income, net worth, and year. Risk-taking attitude encompasses four categories: unwillingness to take any financial risk, taking average financial risk, taking above-average financial risk, and taking substantial financial risk. The reference category is not willing to take any risk.
Furthermore, the level of education includes four categories: high school or less, some college but no degree, associate degree, and college or more. The reference category is high school or less. Retired is a dichotomous variable with a value of 1 if the respondent is retired and a 0 if not retired. Income and net worth are measured as continuous variables and transformed using inverse hyperbolic sine to deal with the skewness of the data, retain zero-valued (and negative-valued) observations and account for income and wealth's unique properties.
Financial literacy is measured using responses to three standard SCF questions assessing basic financial concepts: interest compounding, inflation, and risk diversification. Each correct answer is coded as 1, and the total score ranges from 0 to 3, with higher values indicating greater financial literacy. Likewise, self-assessed personal finance knowledge is a Likert-scale variable. Respondents were asked to rate their knowledge of personal finances on a scale ranging from 1 (“not at all knowledgeable”) to 7 (“very knowledgeable”).
Age and number of kids are also measured as continuous variables. Female is a dichotomous variable with a value of 1 if female and a 0 otherwise. Variable married has a value of 1 if the respondent is married and a 0 otherwise. Race/ethnicity includes five categories: white, Black, Hispanic, Asian and other race/ethnicity. Each category of race/ethnicity is also a dichotomous variable with a value of 1 if chosen and a 0 otherwise. The reference category for race/ethnicity is white. Lastly, the year includes three years: 2016, 2019 and 2022 to control for the time effect.
Table 1 presents the summary statistics for the outcome and explanatory variables used in this analysis, their weighted means, and their standard errors (shown in parentheses). As seen in Table 1, 19% of the respondents expect to have inadequate income in their retirement, and about the same percentage, i.e., 17% of the respondents expect to have satisfactory and very satisfactory income in their retirement. The percentage of respondents who own a term life insurance policy is 48%, which is greater than 19% of respondents who own cash value life insurance. Respondents have a higher face value amount for term life compared to cash value life, which is reasonable given the affordable premium for term life.
Summary Statistics
| Variables | Mean (Std. Error) |
|---|---|
| Outcome Variable: Retirement Income Rating | |
| Totally Inadequate | 0.1925 (0.0038) |
| Somewhat Inadequate | 0.1233 (0.0032) |
| Enough to Maintain Living Standard | 0.3448 (0.0046) |
| Satisfactory | 0.1678 (0.0036) |
| Very Satisfactory | 0.1715 (0.0036) |
| Explanatory Variables: | |
| Term Life Insurance | 0.4765 (0.0048) |
| Cash Value Life Insurance | 0.1906 (0.0037) |
| Face Value Amount for Term Life | 174539.7000 (3861.3520) |
| Face Value Amount for Cash Value Life | 44397.4400 (1848.2120) |
| Risk-Taking Attitude (Reference: No risk) | |
| No risk | 0.3917 (0.0047) |
| Average risk | 0.3883 (0.0047) |
| Above average risk | 0.1803 (0.0037) |
| Substantial risk | 0.0397 (0.0018) |
| Retired | 0.2296 (0.0042) |
| Education (Reference: High school or less) | |
| High School or less | 0.1078 (0.0029) |
| Some college no degree | 0.2469 (0.0042) |
| Associate degree | 0.2764 (0.0044) |
| Variables | Mean (Std. Error) |
| College or more | 0.3689 (0.0046) |
| Age | 51.9563 (0.1792) |
| Female | 0.2744 (0.0043) |
| Married | 0.5675 (0.0048) |
| Kids | 0.7320 (0.0104) |
| Race/Ethnicity (Reference: White) | |
| White | 0.6864 (0.0043) |
| Black | 0.1482 (0.0032) |
| Hispanic | 0.1111 (0.0029) |
| Asian | 0.0152 (0.0012) |
| Other race/Ethnicity | 0.0390 (0.0018) |
| Income | 130336.5000 (1991.2970) |
| Net worth | 927890.8000 (16805.2800) |
| Financial Literacy | 2.1915 (0.0081) |
| Self-assessed Personal Finance Knowledge | 7.2099 (0.0211) |
| Year (Reference: 2016) | |
| 2016 | 0.3264 (0.0042) |
| 2019 | 0.3333 (0.0044) |
| 2022 | 0.3402 (0.0050) |
Data are from the 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
Likewise, 23% of the respondents are retired, 37% have a college degree or more, and the average age of respondents is 52 years. Furthermore, most of the respondents are male, married, have kids, and are white. The mean income for the sample is $130,337, which shows that the households in the SCF are relatively wealthy by U.S. standards. These relatively wealthy households reflect the structure of American retirement and insurance systems.
This study includes two ordered-probit models. The first model estimates the association between two different life insurance policies—term life insurance and cash value life insurance—and retirement income rating, and the second model estimates the association between the face value amount for each type of life insurance policy and retirement income rating.
RIRi* = β0 + β Xi + ɛ
RIR = 1 if RIR* ≤ μ1 (Totally inadequate)
RIR = 2 if μ1 < RIR* ≤ μ2
RIR = 3 if μ2 < RIR* ≤ μ3
RIR = 4 if μ3 < RIR* ≤ μ4
RIR = 5 if μ4 < RIR* (Very satisfactory)
RIR* is a latent measure of the retirement income rating of an individual i, and RIR is the observed retirement income rating. The unknown threshold μ1, μ2, μ3, μ4 In the model, the coefficients are estimated and associated with responses to the question “How do you rate the retirement income you receive (or expect to receive) from all sources?” β0 is the intercept, while β is a vector of coefficients showing the association of the independent variables with the dependent variable. Xi is a matrix of independent variables used in the models, such as term life insurance, cash value life insurance, term life face value amount, cash value face value amount, risk-taking attitude, retired, education, age, female, married, kids, race/ethnicity, income, net worth, financial literacy, self-assessed personal finance knowledge, and year. Marginal effects are calculated to determine the average effect on the observed dependent variable (RIR). ɛ is the error term that follows the normal distribution.
To address potential endogeneity concerning life insurance choices, an instrumental variable (IV) approach is employed for the study's key explanatory variables: term life insurance ownership, cash value life insurance ownership, and their respective face value amounts. Due to data limitations, this analysis utilises a single instrument: whether the respondent used a financial professional (e.g., a lawyer, accountant, or financial planner) for investment and savings decisions.
Each of the primary models includes two potentially endogenous variables. Given the availability of only one instrument, it was not possible to instrument both variables simultaneously. Therefore, the strategy was to estimate two separate IV models for each primary model. In each case, one life insurance variable was instrumented using a two-stage least squares (2SLS) regression, while the other was treated as exogenous. The primary goal of this analysis is to assess the robustness of the main findings; consequently, the sign and significance of the coefficients are interpreted rather than directly comparing their magnitudes across models.
The findings of this study are most representative of the regulatory, pension, and financial behaviour patterns that are typical of the United States because the study uses the U.S. household survey data. Table 2 reports the marginal effects and the standard errors from the first ordered-probit model that estimates the association between term life and retirement income rating, as well as cash value life insurance and retirement income rating. The dependent variable is retirement income rating measured on a scale of 1 to 5, with 1 representing inadequate retirement income and 5 representing very satisfactory retirement income. The results show that having a term life insurance policy and having a cash value life insurance policy are each associated positively with a higher retirement income rating. Households owning a term life insurance policy have a 0.01 higher probability of rating their retirement income as very satisfied compared to those who do not own a term life insurance policy. Likewise, households owning a cash value life insurance policy have a 0.04 higher probability of rating their retirement income as very satisfied compared to those who do not own such a policy.
Association Between Term Life Insurance, Cash Value Life Insurance Ownership, and Retirement Income Rating: Ordered-Probit Model
| Independent Variables | Retirement Income Rating | ||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Term Life Insurance | −0.0083* (0.0037) | −0.0027* (0.0012) | −0.0020* (0.0009) | 0.0023* (0.0010) | 0.0107* (0.0048) |
| Cash Value Life Insurance | −0.0276*** (0.0044) | −0.0090*** (0.0014) | 0.0067*** (0.0011) | 0.0077*** (0.0012) | 0.0356*** (0.0056) |
| Risk-Taking Attitude (Reference: No risk) | |||||
| Average risk | −0.0642*** (0.0049) | −0.0207*** (0.0017) | −0.0116*** (0.0010) | 0.0210*** (0.0018) | 0.0755*** (0.0056) |
| Above average risk | −0.0798*** (0.0055) | −0.0270*** (0.0021) | −0.0181*** (0.0017) | 0.0254*** (0.0019) | 0.0995*** (0.0071) |
| Substantial risk | −0.0939*** (0.0076) | −0.0332*** (0.0033) | −0.0258*** (0.0039) | 0.0288*** (0.0022) | 0.1240*** (0.0124) |
| Retired | −0.0733*** (0.0054) | −0.0239*** (0.0018) | −0.0179*** (0.0015) | 0.0205*** (0.0016) | 0.0946*** (0.0070) |
| Education (Reference: High school or less) | |||||
| Some college no degree | −0.0229** (0.0081) | −0.0067** (0.0023) | −0.0023** (0.0007) | 0.0078** (0.0028) | 0.0240** (0.0082) |
| Associate degree | −0.0242** (0.0083) | −0.0071** (0.0024) | −0.0025** (0.0007) | 0.0083** (0.0029) | 0.0255** (0.0084) |
| College or more | −0.0768*** (0.0081) | −0.0264*** (0.0026) | −0.0187*** (0.0015) | 0.0245*** (0.0029) | 0.0975*** (0.0088) |
| Age | −0.0005** (0.0002) | −0.0002** (0.0000) | −0.0001** (0.0000) | 0.0001** (0.0000) | 0.0006** (0.0002) |
| Female | 0.0174** (0.0056) | 0.0057** (0.0018) | 0.0042** (0.0014) | −0.0049** (0.0016) | −0.0224** (0.0072) |
| Married | −0.0014 (0.0052) | −0.0004 (0.0017) | −0.0003 (0.0013) | 0.0004 (0.0014) | 0.0018 (0.0067) |
| Kids | 0.0101*** (0.0017) | 0.0033*** (0.0006) | 0.0025*** (0.0004) | −0.0028*** (0.0005) | −0.013*** (0.0022) |
| Independent Variables | Retirement Income Rating | ||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Race/Ethnicity (Reference: White) | |||||
| Black | −0.0198*** (0.0051) | −0.0068*** (0.0018) | −0.0059*** (0.0017) | 0.0053*** (0.0013) | 0.0272*** (0.0072) |
| Hispanic | 0.0169** (0.0064) | 0.0052** (0.0019) | 0.0034** (0.0011) | −0.0049* (0.0019) | −0.0207** (0.0076) |
| Asian | 0.0310* (0.0135) | 0.0092* (0.0037) | 0.0052* (0.0015) | −0.0091* (0.0041) | −0.0363* (0.0145) |
| Other race/Ethnicity | 0.0100 (0.0090) | 0.0032 (0.0028) | 0.0022 (0.0018) | −0.0028 (0.0026) | −0.0125 (0.0109) |
| Income | −0.0300*** (0.0012) | −0.0098*** (0.0004) | −0.0073*** (0.0004) | 0.0084*** (0.0004) | 0.0387*** (0.0014) |
| Net worth | −0.0054*** (0.0003) | −0.0018*** (0.0001) | −0.0013*** (0.0001) | 0.0015*** (0.0001) | 0.0070*** (0.0004) |
| Financial Literacy | −0.0300*** (0.0012) | −0.0098*** (0.0004) | −0.0073*** (0.0004) | 0.0084*** (0.0004) | 0.0387*** (0.0014) |
| Self-assessed Personal Finance Knowledge | −0.0054*** (0.0003) | −0.0018*** (0.0001) | −0.0013*** (0.0001) | 0.0015*** (0.0001) | 0.0070*** (0.0004) |
| Year (Reference: 2016) | |||||
| 2019 | −0.0174*** (0.0040) | −0.0057*** (0.0013) | −0.0042*** (0.0010) | 0.0049*** (0.0011) | 0.0224*** (0.0052) |
| 2022 | −0.0108 (0.0045) | −0.0034 (0.0015) | −0.0024 (0.0010) | 0.0031 (0.0013) | 0.0136 (0.0057) |
Data are from the 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
Furthermore, several demographic and financial characteristics are also significantly associated with retirement income rating. Respondents with a college degree or higher have a 0.10 greater probability of rating their retirement income as very satisfied compared to those with a high school education or less. In contrast, female respondents and those with children are slightly less likely to rate their retirement income as very satisfied, with marginal effects of −0.02 and −0.01, respectively. Additionally, higher household income (0.04), greater net worth (0.01), and higher self-assessed knowledge of personal finance (0.02) are all positively associated with a greater probability of reporting very satisfied retirement income.
Moreover, considering the variations in the magnitude of marginal effects between two distinct categories of policyholders from Table 2, an additional test was performed to investigate whether one set of policyholders is more inclined to positively rate their retirement income in comparison to the other set, depending on their policy type. The findings of this analysis are displayed in Table 3, indicating that individuals with cash value policies are more likely to positively rate their retirement income compared to those with term life policies. All differences in marginal effects are statistically significant at the 0.1% level, underscoring the strength of the observed relationships.
Comparing Marginal Effects Between Cash Value Life Insurance and Term Life Insurance
| Retirement Income Rating | |||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Difference in marginal effects (cash value life - term life) | −0.0199*** (0.0055) | −0.0065*** (0.0018) | −0.0048*** (0.0013) | 0.0056*** (0.0015) | 0.0256*** (0.0071) |
Data are from the 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
Table 4 reports the marginal effects and the standard errors from the second ordered-probit model that estimates the association between the face value amount for each type of life insurance policy and retirement income rating. The results presented in Table 4 show a positive association between face value amount for both types of term life and cash value life insurance and the retirement income rating. While these results have a strong statistically significant effect, it is important to note that the magnitude of the marginal effects indicates that this association is relatively modest.
Association Between Face Value Amounts of Term Life Insurance, and Cash Value Life Insurance and Retirement Income Rating: Ordered-Probit Model
| Independent Variables | Retirement Income Rating | ||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Face Value for Term Life Insurance | −0.0010** (0.0003) | −0.0003** (0.0001) | −0.0002** (0.0001) | 0.0003** (0.0001) | 0.0013** (0.0004) |
| Face Value for Cash Value Life Insurance | −0.0027*** −0.0003 | −0.0009*** −0.0001 | −0.0007*** −0.0001 | 0.0008*** −0.0001 | 0.0035*** −0.0004 |
| Risk-Taking Attitude (Reference: No risk) | |||||
| Average risk | −0.0635*** (0.0049) | −0.0204*** (0.0017) | −0.0114*** (0.0010) | 0.0207*** (0.0018) | 0.0745*** (0.0056) |
| Above average risk | −0.0789*** (0.0055) | −0.0266*** (0.0021) | −0.0177*** (0.0017) | 0.0251*** (0.0019) | 0.0982*** (0.0071) |
| Substantial risk | −0.0930*** (0.0076) | −0.0329*** (0.0033) | −0.0253*** (0.0039) | 0.0286*** (0.0022) | 0.1227*** (0.0124) |
| Retired | −0.0751*** (0.0055) | −0.0244*** (0.0018) | −0.0181*** (0.0015) | 0.0210*** (0.0016) | 0.0966*** (0.0070) |
| Education (Reference: High school or less) | |||||
| Some college no degree | −0.0224** (0.0081) | −0.0065** (0.0023) | −0.0022** (0.0007) | 0.0076** (0.0028) | 0.0235** (0.0082) |
| Associate degree | −0.0234** (0.0082) | −0.0069** (0.0023) | −0.0024** (0.0007) | 0.0080** (0.0028) | 0.0247** (0.0084) |
| College or more | −0.0753*** (0.0080) | −0.0259*** (0.0026) | −0.0182*** (0.0015) | 0.0240*** (0.0029) | 0.0954*** (0.0088) |
| Age | −0.0004** (0.0002) | −0.0001** (0.0000) | −0.0001** (0.0000) | 0.0001** (0.0000) | 0.0006** (0.0002) |
| Female | 0.0175** (0.0056) | 0.0057** (0.0018) | 0.0042** (0.0014) | −0.0049** (0.0016) | −0.0225** (0.0072) |
| Married | −0.0001 (0.0052) | −0.0000 (0.0017) | −0.0000 (0.0012) | 0.0000 (0.0014) | 0.0001 (0.0067) |
| Kids | 0.0104*** (0.0017) | 0.0034*** (0.0006) | 0.0025*** (0.0004) | −0.0029*** (0.0005) | −0.0134*** (0.0022) |
| Independent Variables | Retirement Income Rating | ||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Race/Ethnicity (Reference: White) | |||||
| Black | −0.0195*** (0.0051) | −0.0067*** (0.0018) | −0.0057** (0.0017) | 0.0052*** (0.0013) | 0.0267*** (0.0072) |
| Hispanic | 0.0161* (0.0064) | 0.0050* (0.0019) | 0.0032** (0.0011) | −0.0046* (0.0019) | −0.0197** (0.0076) |
| Asian | 0.0303* (0.0135) | 0.0090* (0.0037) | 0.0051*** (0.0015) | −0.0089* (0.0041) | −0.0355* (0.0145) |
| Other race/Ethnicity | 0.0096 (0.0090) | 0.0030 (0.0028) | 0.0021 (0.0017) | −0.0027 (0.0026) | −0.0120 (0.0109) |
| Income | −0.0294*** (0.0012) | −0.0096*** (0.0004) | −0.0071*** (0.0004) | 0.0082*** (0.0004) | 0.0379*** (0.0014) |
| Net worth | −0.0053*** (0.0003) | −0.0017*** (0.0001) | −0.0013*** (0.0001) | 0.0015*** (0.0001) | 0.0069*** (0.0004) |
| Financial Literacy | −0.0009 (0.0023) | −0.0003 (0.0008) | −0.0002 (0.0006) | 0.0002 (0.0007) | 0.0011 (0.0030) |
| Self-assessed Personal Finance Knowledge | −0.0134*** (0.0009) | −0.0043*** (0.0003) | −0.0032*** (0.0002) | 0.0037*** (0.0003) | 0.0172*** (0.0011) |
| Year (Reference: 2016) | |||||
| 2019 | −0.0175*** (0.0040) | −0.0057*** (0.0013) | −0.0042*** (0.0010) | 0.0049*** (0.0011) | 0.0224*** (0.0052) |
| 2022 | −0.0110* (0.0045) | −0.0035* (0.0015) | −0.0024* (0.0010) | 0.0031* (0.0013) | 0.0138* (0.0057) |
Data are from 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
Consistent with the findings presented in Table 2, the results for other explanatory variables— such as having a college degree or higher (0.10), higher income (0.04), and greater self-assessed knowledge of personal finance (0.02)—also exhibit strong statistical significance and, in some cases, larger marginal effects. Importantly, while the face value of life insurance policies shows only modest marginal effects, their statistical significance is comparable to that of other key predictors, suggesting that these variables remain robust correlates of retirement income rating.
The magnitude of the marginal effects for the face value of cash value life insurance is higher than for term, as shown in Table 4. Therefore, similar to Table 3, an additional test was performed to examine whether one set of policyholders is more inclined to positively rate their retirement income in comparison to the other set, depending on the face value of their respective policy types. The results, presented in Table 5, align with the initial findings from Table 3, indicating that while a higher face value is a positive predictor for both policy types, the association between face value and retirement income rating is substantially more robust for holders of cash value life insurance. The statistical significance of this difference (p < 0.001) provides strong evidence that policy type plays a key role in how a policy's face value is associated with retirement income rating.
Comparing Marginal Effects Between Face Value of Cash Value Life Insurance and Face Value of Term Life Insurance
| Retirement Income Rating | |||||
|---|---|---|---|---|---|
| 1 (Totally Inadequate) | 2 | 3 | 4 | 5 (Very Satisfied) | |
| Difference in Marginal Effects (face value, cash value, life – face value term life) | −0.0017*** (0.0004) | −0.0005*** (0.0001) | −0.0004*** (0.0001) | 0.0005*** (0.0001) | 0.0022*** (0.0006) |
Data are from 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
The first-stage regression results from the 2SLS models confirm the instrument's validity. The use of a financial professional is positively and statistically significantly associated with both the ownership and face value amounts of term and cash value life insurance. For all specifications, the F-statistic for the weak instrument test substantially exceeds the Stock-Yogo (2005) critical value thresholds, indicating that the chosen instrument is strong and relevant (see Appendix).
The second-stage results reveal a consistent pattern. After instrumenting endogeneity, the estimated causal effects of the life insurance variables on the retirement income rating remain positive while remaining statistically significant.
In the ownership model (Table 6), when either term life or cash value life is instrumented, its positive effect on retirement income rating is enhanced and remains significant.
Similarly, in the face value model (Table 7), instrumenting for the face value of either term life or cash value life results in a positive, statistically significant positive effect on the retirement income rating.
Instrumental Variable Analysis for the Association Between Term Life Insurance, and Cash Value Life Insurance Ownership and Retirement Income Rating
| Independent Variables | IV (Term life instrumented) | IV (Cash value life instrumented) |
|---|---|---|
| Term Life Insurance | 5.5014*** (IV) (0.5655) | 0.3432*** (0.0216) |
| Cash Value Life Insurance | 0.7413*** (0.0663) | 3.4783***(IV) (0.2189) |
| Risk-Taking Attitude (Reference: No risk) | ||
| Average risk | 0.0106 (0.0434) | 0.2067*** (0.0186) |
| Above average risk | 0.1916*** (0.0392) | 0.2955*** (0.0211) |
| Substantial risk | 0.7458*** (0.0556) | 0.3730*** (0.0308) |
| Retired | 1.1620*** (0.0854) | 0.4857*** (0.0199) |
| Education (Reference: High school or less) | ||
| Some college no degree | −0.3665*** (0.0639) | −0.0537* (0.0272) |
| Associate degree | −0.5719*** (0.0829) | −0.0636* (0.0280) |
| College or more | −0.3855*** (0.0931) | 0.1990*** (0.0292) |
| Age | 0.0069*** (0.0010) | −0.0112*** (0.0010) |
| Female | −0.3071*** (0.0388) | −0.1700*** (0.0207) |
| Married | −0.8174*** (0.0914) | −0.2379*** (0.0247) |
| Kids | −0.1852*** (0.0171) | −0.0624*** (0.0062) |
| Race/Ethnicity (Reference: White) | ||
| Black | −0.2755*** (0.0508) | −0.1813*** (0.0273) |
| Hispanic | 0.5053*** (0.0717) | 0.1686*** (0.0279) |
| Asian | 0.0247 (0.0717) | 0.0330 (0.0452) |
| Independent Variables | IV (Term life instrumented) | IV (Cash value life instrumented) |
| Other race/Ethnicity | 0.4188*** (0.0693) | 0.0533 (0.0317) |
| Income | 0.0567*** (0.0105) | 0.0809*** (0.0054) |
| Net worth | 0.0066* (0.0029) | 0.0124*** (0.0015) |
| Financial Literacy | −0.0974*** (0.0174) | −0.0249*** (0.0087) |
| Self-assessed Personal Finance Knowledge | 0.0253*** (0.0067) | 0.0456*** (0.0034) |
| Year (Reference: 2016) | ||
| 2019 | 0.2003*** (0.0258) | 0.1358*** (0.0148) |
| 2022 | 0.4135*** (0.0452) | 0.2078*** (0.0188) |
Data are from 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
Instrumental Variable Analysis for the Association Between Face Value Amounts of Term Life Insurance, and Cash Value Life Insurance and Retirement Income Rating
| Independent Variables | IV (Face Value Term life instrumented) | IV (Face Value Cash value life instrumented) |
|---|---|---|
| Face Value for Term Life | 0.3482***(IV) (0.0305) | 0.0250*** (0.0014) |
| Face Value for Cash Value Life | 0.0503*** (0.0037) | 0.2341***(IV) (0.0135) |
| Risk-Taking Attitude (Reference: No risk) | ||
| Average risk | 0.0382 (0.0348) | 0.2178*** (0.0166) |
| Above average risk | 0.1299*** (0.0367) | 0.2831*** (0.0195) |
| Substantial risk | 0.5657*** (0.0411) | 0.3328*** (0.0287) |
| Retired | 1.1487*** (0.0713) | 0.5381*** (0.0192) |
| Education (Reference: High school or less) | ||
| Some college no degree | −0.2555*** (0.0468) | −0.0069 (0.0238) |
| Associate degree | −0.4270*** (0.0595) | −0.0176 (0.0245) |
| College or more | −0.3320*** (0.0747) | 0.2096*** (0.0263) |
| Age | 0.0069*** (0.0008) | −0.0088*** (0.0008) |
| Female | −0.2225*** (0.0291) | −0.1525*** (0.0187) |
| Married | −0.6959*** (0.0676) | −0.2148*** (0.0216) |
| Kids | −0.2095*** (0.0163) | −0.0638*** (0.0056) |
| Race/Ethnicity (Reference: White) | ||
| Black | −0.1205*** (0.0333) | −0.1086*** (0.0223) |
| Hispanic | 0.3976*** (0.0527) | 0.1167*** (0.0238) |
| Asian | 0.0308 (0.0604) | 0.0247 (0.0412) |
| Other race/Ethnicity | 0.3314*** (0.0535) | 0.0243 (0.0287) |
| Income | 0.0253* (0.0111) | 0.0595*** (0.0058) |
| Net worth | 0.0052* (0.0026) | 0.0133*** (0.0014) |
| Financial Literacy | −0.1134*** (0.0157) | −0.0253** (0.0079) |
| Self-assessed Personal Finance Knowledge | 0.0279*** (0.0055) | 0.0474*** (0.0031) |
| Year (Reference: 2016) | ||
| 2019 | 0.1715*** (0.0207) | 0.1175*** (0.0133) |
| 2022 | 0.3076*** (0.0311) | 0.1580*** (0.0158) |
Data are from 2016, 2019 and 2022 Survey of Consumer Finances. Number observations = 16,620. Standard errors are in parentheses.
indicates significance at the 5% level.
indicates significance at the 1% level.
indicates significance at the 0.1% level.
Finally, post-estimation Durbin and Wu-Hausman tests were conducted to test for endogeneity formally. The results presented in the Appendix reject the null hypothesis of exogeneity. This provides strong statistical evidence that endogeneity is present, confirming that the IV approach is both appropriate and necessary for generating more reliable estimates.
One of the goals of the SCF is to estimate household wealth in the United States; therefore, the sample is disproportionately chosen, and households are relatively wealthy (Hanna et al. 2018). Furthermore, there are different types of cash value life insurance policies, and the benefits of one type of policy may outweigh the benefits of another. For example, the benefits of a universal life insurance policy may outweigh the benefits of a whole life insurance policy for a household. The study cannot go into detail about other types of cash value life insurance due to data limitations. The study's findings could be built upon by looking into the association between different types of cash value life insurance and retirement income ratings in the future. Additionally, a significant limitation of this study is its exclusive focus on the United States financial market. The findings regarding the relationship between cash value life insurance and retirement income rating may not be directly generalizable to countries where such products are not widely available or where different retirement savings mechanisms dominate.
This study reveals a statistically significant association between life insurance ownership and a more favourable retirement income rating, within the U.S. financial and regulatory context. Our findings indicate that individuals holding either a term life or a cash value life insurance policy are more inclined to have a higher retirement income rating compared to those without such policies. This suggests that life insurance provides a sense of security against future uncertainties, directly influencing households' expectations about their financial future. Individuals with life insurance may report greater confidence in their retirement, either through the supplemental income potential of cash value policies or the assurance of survival-contingent income for their loved ones provided by term policies.
The observed positive associations between life insurance ownership (both term and cash value) and a more favourable retirement income rating are notably consistent with insights from behavioural economics, even within our correlational framework. Rather than solely reflecting purely rational, utility-maximising choices, these findings suggest the influence of concepts like bounded rationality and Prospect Theory. For individuals navigating complex financial landscapes, life insurance, especially cash value policies, may be perceived as a tangible and simpler mechanism for mitigating future financial anxieties and providing a sense of security against potential losses, such as the depletion of savings or income uncertainty.
The stronger association observed with cash value policies can be further explained by narrow framing. The explicit savings component of cash value is more readily interpreted as a direct contribution to retirement income, offering a clearer “gain” reference point compared to the contingent nature of term life benefits. This perceived reduction in potential financial “losses” contributes to a heightened subjective sense of financial well-being and satisfaction with retirement preparedness, aligning with the principles of loss aversion.
Beyond the direct financial aspects, ownership of term life insurance is also associated with a more favourable retirement income rating. A significant advantage of holding such a policy may be the psychological comfort it imparts. Owners of term life policies may experience diminished financial apprehension because they have diversified beyond conventional retirement income streams, providing a safeguard to help their loved ones prevent potential financial vulnerabilities. For instance, consider a young couple diligently saving for retirement; the untimely death of one partner in the absence of a term life policy could severely disrupt the surviving partner's retirement preparations. This example highlights how term life insurance can bolster an individual's conviction in the adequacy of their retirement income by mitigating a significant potential “loss” scenario.
However, it is also crucial to acknowledge alternative explanations for these observed associations. For instance, the positive association, particularly for term life insurance ownership, might not exclusively stem from the direct behavioural impact of the product, but could rather serve as a proxy for more comprehensive financial planning activity. Households that acquire term life often do so due to the presence of dependents or significant liabilities like a mortgage, indicating a proactive and holistic approach to financial management. In this light, the positive association could reflect the broader effects of such disciplined planning behaviours on retirement income perceptions, rather than solely the direct effect of the insurance product itself. Future research could further disentangle these overlapping factors.
Overall, the observed associations align with the consumption smoothing aspects of life insurance products. Despite common critiques regarding the advantages of cash value insurance, this study provides empirical evidence supporting its significant value in retirement planning, particularly given its perceived role as a direct contribution to future income. Both term and cash value policies appear to enhance subjective financial well-being by providing a buffer against perceived future financial risks, influencing individuals' satisfaction with their retirement preparedness.