1. Worker characteristics that influence the choice pension distributions

Retiring pension participants typically have a choice among a series of payout options which include a straight-life annuity and a series of options that provide lifetime income protections to named beneficiaries. The choice to accept a lower monthly benefit associated with the J&S annuity is driven by the desire to provide a continuing monthly income for the chosen beneficiary. It is important to remember that the retiree can designate any individual as a beneficiary; however, the most frequently named individual is a spouse. Thus, we would expect that married participants would be more likely to select the J&S annuity. If this lifetime income protection is a normal good, we would expect families with higher household income would choose a J&S distribution. The ages of the retiree and their spouse are also key factors in this decision. Unfortunately, the data that PBGC was able to provide did not include important information on the retirees that likely affected their choice of an annuity such as marital status, age of a spouse, the work history of the spouse, and other household income. Employing the data that we received from the PBGC, we have the following testable hypotheses.

The monthly benefit from a single-life annuity for a PGBC participant is based on the benefit formula in the retirement plan that previously covered the individual. When determining the benefit for other distribution options, plan actuaries consider the age of the member and the age of the designated beneficiary. These calculations use a unisex mortality so the monthly benefit for a J&S annuity does not reflect sex differences in life expectancy. As a result, the lower benefit associated with a J&S annuity relative to the single-life annuity is the same for a male and female retiree holding the ages of the member and the beneficiary constant. Since women have longer life expectancies compared to men, the similar reduction of benefits tends to result in a greater reduction in the J&S benefit for women than if the cost was determined by a sex-specific lifetable. Thus, we would expect that on average, women will be more likely to accept the single-life annuity and less likely to request a J&S distribution.

The preference for the single-life annuity by women may be enhanced if they are more likely to be married to men who have higher earnings and who are more likely to be covered by their own retirement plan. The retiring female with these circumstances may feel a lower need to select a J&S annuity. In addition, traditional marriage patterns indicate that women are more likely to be married to men who are somewhat older and therefore, have a higher probability of dying before the retired female. One would expect retirees to be more likely to choose a J&S annuity if they have fewer years of life expectancy but are married to a younger spouse with more years of life expectancy and less likely to select a J&S annuity if their spouse is older. Finally, since the proportion of women who are currently married is lower compared to men, women will have a lower demand for providing a beneficiary benefit. Based on these factors, we anticipate that women will be much less likely to choose a J&S distribution from a PBGC covered pension plan.

We anticipate that providing income protection for a spouse is a normal good. Therefore, the probability of selecting a J&S annuity is expected to increase as annual income rises. The PBGC data do not include information on final earnings; however, it does include the monthly pension benefit for most retirees. We use the monthly benefit from the PBGC as a proxy for household income and expect that individuals with higher monthly benefits will be more likely to select a J&S annuity.

Other personal and economic factors influence the decision to select a J&S annuity such as the price of the J&S option and the probability that one's spouse will die before the retiree. The reduction in benefits associated with the J&S annuity is based on the age of the spouse relative to the age of the retiree. As typical of administrative data, we have no information on the spouse's age, work history, or pension coverage. However, we do know the age at retirement of the individual requesting a pension distribution. Holding other factors constant, individuals who request a pension payout at older ages have a shorter remaining life expectancy and a higher probability of dying in each successive year. As a result, we expect that older retirees will be more likely to desire income protection for their spouses. Age at retirement is included in the regression analysis to capture this effect.

The variation in distributional choices over time should capture some of the effect of interest rate changes during the sample period as well as other changes in the economic conditions. We use the two-year Treasury rate interest rates as an indicator of how interest rates might affect annuity choices. This rate started increasing at the end of the Great Recession at the beginning of our sample period, then decreased through 2019, fell to almost zero during the COVID pandemic, and then started increasing toward the end of 2021 (see Figure 1). Thus, the price of selecting a J&S annuity in terms of the lower monthly benefit varied over the sample period. The change in the price of a J&S annuity is associated with these changes in interest rates. We include individual year effects to proxy these changes.

Figure 1. U.S. Treasury Rate at 2-Year Constant Maturity.

In an NCHS Issue Brief, Tejada-Vera and Kramarow (Reference Tejada-Vera and Kramarow2022) reported that COVID was the underlying cause of death for over 350,000 deaths in the United States in 2020. They stated that while COVID affected people of all ages, ‘older adults were especially impacted during the first year of the pandemic: 81% of COVID-19 deaths in 2020 occurred among those aged 65 and over’. The impact of COVID on the health and mortality rates of older Americans affected their life expectancy and thus, the demand for J&S payouts.Footnote ³ As a result, individuals making decisions on pension distributions during the COVID years likely made different choices compared to comparable retirees in the earlier years in our sample. To capture this one time, change in life expectancy of older Americans, we include interaction terms between the age of the retiree and the COVID years in all regressions in order to capture the potential changes in life expectancies. Given the shortness of the pandemic (only a few years), it is unlikely that the lower life expectancy during these years was reflected in the pricing of J&S annuities.

2. The PBGC and the benefit protection it provides

The PBGC was established by the Employee Retirement Income Security Act of 1974. The goal of the PBGC is to protect the retirement incomes of Americans who participate in private sector DB pension plans if the firm defaults on its pension obligations (PBGC, 2023a).Footnote ⁴ The PBGC covers both single-employer and multiemployer DB plans; however, the two types of plans are managed in separate programs within the PBGC. The program covering single employers assumes responsibility for terminated plans of failing firms and pays benefits directly to retirees of these plans. Instead of making direct payments to retirees in multiemployer plans, the PBGC provides financial assistance to the plans that are failing. In this paper, we examine only the payout options chosen by participants in single-employer plans.

The single-employer program covers most DB plans that are sponsored by an individual employer in the private sector of the US economy. Employers who offer a DB retirement plan must pay the PBGC a flat-rate insurance premium based on the number of participants. Underfunded single-employer plans pay an additional variable-rate premium based on the amount of unfunded vested benefits. When a firm terminates an underfunded DB plan, the plan's assets and liabilities are transferred to the PBGC, which has the responsibility to pay vested benefits of participants in the plan up to legal limits.Footnote ⁵ The transfer of plans to the PBGC typically occurs when an employer sponsoring an underfunded plan liquidates in bankruptcy, ceases operation, or can no longer afford to maintain the plan.Footnote ⁶ At this point, the PBGC takes over the plan's assets, administration, and payment of benefits. The retirement benefit paid by the PBGC is determined by the provisions of the individual pension plan in which the employee participated. For the worker's benefit to be fully insured, the benefit must have been vested before the plan terminated as the PBGC provisions guarantee only not forfeitable benefits.

In FY 2022, the PBGC provided insurance coverage to 22.3 million workers in 23,800 single-employer DB plans, along with 11.2 million employees covered by 1,360 multiemployer plans (PBGC, 2023b). In FY 2022, the PBGC assumed control of 32 plans that covered nearly 8,000 current and future retirees. In total, the PBGC paid over $7.0 billion in benefits to more than 960,000 retirees during FY 2022, (PBGC, 2022).

The number of PBGC insured active participants as a percent of private sector wage and salary workers has steadily decreased since 1980. The percentage of private sector workers covered by the single-employer PBGC covered plan was 28.8 percent in 1980; however, by 2000, coverage by the PBGC had declined to 16.4 percent of such employees. Coverage fell further to 7.3 percent of employees in 2018. This decline reflects the decline in the incidence of DB plans among private employers (PBGC, 2019a, Table S-33). In 1985, the total number of insured plans was 112,208, compared to 35,375 in 2000 and 23,477 in 2020 (PBGC, 2019a, Table S31). The majority of plans, 61 percent, insured by the PBGC are in the manufacturing industry, while 18 percent are in transportation and public utilities (PBGC, 2019a, Table S-29).

When the PBGC assumes responsibility for a plan, some participants have already retired from their employer and are receiving a pension benefit under the provisions of the terminated plan. The remainder of the participants in the plan have yet to start receiving benefits and will only claim a benefit after the PBGC has assumed responsibility for paying retirement benefits to plan participants. Participants in a failed plan who have not yet claimed a retirement benefit have the option of selecting a distribution method from a menu of options available through the PBGC and those offered by the terminated plan. These benefits are subject to a maximum benefit amount. The maximum monthly guarantee in 1990 was $2,167, $3,222 in 2000, and $6,034 in 2021 (PBGC, 2019a, Table S-53). The benefit cap affects only a small proportion of participants. A study by the PBGC found that 84 percent of participants receive 100 percent of the vested benefits (PBGC, 2019b).Footnote ⁷

Unlike many DB plans, the PBGC does not allow retirees to select lump sum distributions of their pension benefits.Footnote ⁸ We presume, this restriction reflects the preference of policymakers that annuities are preferred to lump sum distributions of pension wealth.

Our primary objective is to examine the distribution choices of participants in terminated DB plans that have been taken over by the PBGC and to determine how these choices differ across retiree economic and demographic characteristics. We also examine whether the distribution choices from PBGC plans have changed over the past decade.

3. Determinates of distribution choices

In DB plans, benefits are usually specified as a straight-life annuity where monthly benefits are determined by a formula that generally is based on years of service, final average salary, and a generosity factor. Plans have a normal retirement age, which is the age when the worker can retire and receive an unreduced benefit. Most DB plans also offer a reduced benefit that can begin at an earlier age. Plans are required by federal regulations to offer a J&S option as the default distribution for married workers. The J&S benefit is based on the present value of the straight-life annuity with the monthly J&S benefit being determined using an appropriate interest rate and mortality tables. The retiree can select the straight-life benefit if their spouse agrees to waive the J&S benefit.Footnote ⁹ A majority of DB plans also offer a lump sum option where the retiree is paid the present value of the straight-life annuity.

4. PBGC administrative data and distribution options

In response to a Freedom of Information Request, the PBGC provided us with relevant information on all individuals who initiated a benefit from the PBGC between 2012 and 2021. The data file was received in November 2022 and included all PBGC Trusted Plan participants who had an actual retirement date in this time period, a total of 250,032 participants. The specific information on PBGC participants included the distribution option selected, individual information on age at retirement, amount of benefit, months of service, and gender. In the following analysis, we use all of the information on retirees provided by the PBGC. Other information that was requested was not included in the data file provided to us due to concerns that the additional information might allow someone to identify individuals in the sample.

The number of new claimants per year varies between 22,000 and 26,000. Of those retirees, 36 percent were women and 64 percent were men. Overall, 20 percent of the sample was aged 55–59, 37 percent were 60–64, and 43 percent were 65 and over. The proportion of new retirees by sex remains relatively stable over the decade, while the proportion of new retirees in the oldest age group increases over time (see Table 1).

Table 1. PBGC sample composition means

Source: Administrative records from the PBGC and our calculations.

5. Benefit options chosen by PBGC claimants

During our sample period, slightly over half of all claimants requested a J&S distribution while 42 percent chose to receive a straight-life annuity and less than 10 percent chose a month certain distribution (Table 2). The proportion of new retirees selecting these options remains relatively constant across the 10 years in our sample varying by only one or two percentage points from year to year. The most interesting observation from these data is that 58 percent of men select one of the J&S options while only 38 percent of women request a J&S benefit.

Table 2. Distribution options selected from PBGC by sex and year

Source: Administrative records from the PBGC and our calculations.

The main reason for selecting a J&S annuity is to provide a continuing benefit to one's spouse after the death of the retiree. Unfortunately, the PBGC does not retain information on marital status. However, national data on marital status by age indicate that women in these age ranges are less likely to be married compared to men of similar ages. During our sample period, men aged 55–64 were 6–8 percentage points more likely to be married and therefore, a larger percentage of men would be more likely to consider a J&S annuity. An even larger difference in marriage rates in the population was observed for individuals aged 65 and over where men were over 20 percentage points more likely to be currently married (Table 3). Thus, differences in marriage rates likely explain much of the difference in the sex differential in the choice of a J&S benefit. In considering these differences in marriage rates, one should remember that the national rates do not necessarily reflect the marriage rates of individuals in the PBGC sample. In addition, the mean age of retirement for the PBGC sample who are over age 65 is much younger than the mean age of the US population over age 65.

Table 3. Percent married by sex and age: US population

^a The Census Bureau did not release its standard 2020 ACS 1-year estimates because of the impacts of the COVID-19 pandemic. The 5-year estimated data are used for our table for 2020. All other data are from ACS 1-year estimates subject tables. Retrieved March 4, 2023.

Source American Community Survey.

Among the 127,503 retirees selecting a J&S distribution, 48 percent selected the 50 percent payout option to the survivor and 45 percent chose the 100 percent survivor benefit. Once again, we observe a sizable difference in the distributional preferences by sex as men are more likely to request a 100 percent payout compared to women. Among men, 49 percent requested a 100 percent J&S while 43 percent chose the 50 percent J&S. In contrast, among female claimants, only 32 percent chose the 100 percent options while 62 percent selected the 50 percent option (Table 4).

Table 4. Size of J&S survivor benefit selected by sex

Source: Administrative records from the PBGC and our calculations.

The third major payout option is a life annuity for the retiree that also provides a certain number of guaranteed monthly payments; that is, if the retiree dies before the guaranteed number of payments have been received, the beneficiary will continue receiving a monthly check for the duration of the guarantee, the certain-and-continuous annuity described earlier. Overall, less than 10 percent of PBGC retirees selected this option. Among those that did choose the month certain annuity, 53.8 percent chose the 15-year guarantee, while 27.8 selected the 10-year and 17.1 percent the five-year guarantee. Examining the length of guarantee by sex shows that 57.1 percent of men chose the 15-year guarantee while only 45.5 percent of the women selected this longer guaranteed payout period (Table 5).

Table 5. Number of guaranteed months of benefit payments

Source: Administrative records from the PBGC and our calculations.

The PBGC did not provide information on the default straight-life annuity amount for retirees who selected a J&S or a certain-and-continuous annuity. However, we do have the monthly benefit for each retiree given the payout option selected. The real and nominal mean benefits by option are shown by year in Table 6. Interestingly, the monthly benefits for the J&S annuity are greater than those for retirees who requested a straight-life annuity. This implies that the average straight-life benefit for these retirees would have been higher than the monthly benefit for those who actually selected the straight-life annuity. Thus, within the PBGC system, retirees with higher default straight-life annuity are more likely to choose a J&S benefit.

Table 6. Average monthly benefit by payout option

Monthly benefit values have been adjusted for inflation (real dollars in 2021).

Source: Administrative records from the PBGC and our calculations.

6. Determinants of choice of distribution

In this section, we analyze the different factors that can explain the distribution choices of retirees from PBGC administration plans. We focus on the three main categories: straight-life annuity, J&S annuity, and certain-and-continuous annuity. First, we examine the choice among the three basic payout options. We model the choice decision as a multinomial logit and estimate the parameters by maximum likelihood. We consider the following explanatory variables: dummy variables for the year of retirement (2013–2021, thus omitting 2012 and setting it as the base year), a dummy variable equal to one if the individual is male and zero otherwise, two dummy variables for age at retirement (less than 60, between 60 and 64, with 65 and older being the base case), and interactions between the male variable and the age dummy variables. Since the impact of COVID affected the health and life expectancy of people differently based on their age and sex, we also include interactions between the year dummies 2020 and 2021 with the male and age dummies in the regressions.

It would be useful to know the final salary of those claiming benefits from the PBGC or the straight-life annuity value for all claimants so that one could better understand whether distribution choices vary by income; however, these data were not available. The PBGC did provide information on the number of months the participant had worked at the time the distribution was requested. In general, longer tenure would be associated with greater benefits. Unfortunately, data on months worked were included for only 63.8 percent of the retirees. Based on the missing data, we chose not to include months worked in the regression model discussed in the following analysis; however, regressions examining the choice of distribution option that include months worked are shown in Appendix Tables A2–A4. Including months worked and limiting the sample does not influence the qualitative results presented in this section.

Leaving out all individuals who chose a distribution other than these three main choices, we are left with 248,709 observations. In this sample, 51 percent of the individuals chose J&S and 42 percent chose straight-life. In Table 7, we report the marginal effects of the explanatory variables evaluated at the mean of the sample. Standard errors are also reported in the table.

Table 7. Distribution choices of retirees from PBGC administered plans: marginal effects

The number of observations is 248,709. The numbers in parentheses are standard errors. The Year variables are dummy variables equal to one for the given year, equal to zero otherwise (the year left out is 2012). The variable Male is a dummy variable equal to one if the individual is identified as male, zero otherwise. The two Age variables are dummy variables equal to 1 if the individual is less than 60 years old (Age<60), zero otherwise; equal to 1 if the individual is between the age of 60 and 65 years old (60≤Age<65), zero otherwise. The degree of statistical significance is shown by 10% (*), 5% (**), and 1% (***).

Estimation results indicate that multiple explanatory variables have economically and statistically significant impacts on the distribution choices. The sex of the individual has a very large impact, and the magnitude of this effect differs by age at retirement. Before year 2020, for individuals 65 years old and older, being male raises the probability of choosing J&S by about 26.9 percentage points relative to female, while lowering straight-life by about 26.3 percentage points. As shown by the age/sex interaction terms, the sex differences are reduced for individuals who claim benefits at younger ages. For those younger than 60 years old, being a male increases the probability of choosing J&S by about 14.4 percentage points (26.896–12.521) over females, while for those between the ages of 60 and 65, males are 18.9 percentage points (26.896–7.996) more likely to select a J&S annuity.Footnote ¹³ Table 8, panel A more clearly illustrates the age and sex differentials in the annuity choices by showing the predicted probability of selecting a straight-life annuity and a J&S annuity by age and sex. As mentioned earlier, the difference in the choice of a J&S annuity probably reflects a lower marriage rate among women along with other factors such as women tend to be married to older men who also are likely to have had higher earnings and may also have earned their own pension benefit.Footnote ¹⁴

Table 8. Predicted probability of distribution choices by age and sex

There are statistically significant changes in the choice of distribution over time, primarily between distribution choices before and after 2016. For females older than 65 years old, compared to the first year of the observation period, 2012, the probability that an individual will choose J&S increases beginning in 2016 and has a maximum impact of 5.4 percentage points occurring in 2020. Correspondingly, the probability of the other two options tends to decrease over time, with a maximum impact of −1.5 percentage points in 2020 for certain-and-continuous and −3.9 percentage points in 2020 for straight-life annuities.

For the COVID period, looking at the interactions with the age dummy variables, there is a significant impact only in 2021. Individuals younger than 60 years old are 5.6 percentage points more likely to choose a straight-life annuity and 3.6 percentage points less likely to choose a J&S annuity, compared to individuals older than 65 years old. For individuals between 60 and 65, the strongest impact is in 2021, with a reduction in the probability of choosing a certain-and-continuous annuity of 1.8 percentage points. A statistically significant difference over sex appears in year 2020, where everything else equal, being a male increases the probability of choosing a straight-life annuity by 2.2 percentage points and decreases the probability of choosing J&S by 2.5 percentage points.

Retirees selecting a straight-life annuity receive the maximum monthly benefit for the rest of their life; however, this option does not provide any continuing benefit for spouses. Individuals with this annuity have no other choices to make concerning their pension benefits. In contrast, retirees who select a J&S or a month's certain annuity must select the magnitude of the survivor benefit or the length of the guaranteed payments. We now examine each of these options.

6.1 Magnitude of J&S benefit

We now study the determinants of the benefits selected by the individuals who opted for J&S. We focus on the three most common options: 50, 75, and 100 percent. These three options cover about 98 percent of all the individuals who chose J&S, giving us 124,648 observations of individuals who selected a J&S annuity. The 50 and 100 percent benefit options represent 50 and 46 percent of the observations, respectively. As in the previous section, we model these three options with a multinomial logit using the same set of explanatory variables (year dummies, male and age dummies, plus interactions) and estimate by maximum likelihood. The partial effects evaluated at the mean of the sample and their standard errors are reported in Table 9.

Table 9. Size of survivor payments: marginal effects

The number of observations is 124,648. The numbers between parentheses are standard errors. The Year variables are dummy variables equal to one for the given year, equal to zero otherwise (the year left out is 2012). The variable Male is a dummy variable equal to one if the individual is identified as male, zero otherwise. The two Age variables are dummy variables equal to one if the individual is less than 60 years old (Age<60), zero otherwise; equal to one if the individual is between the age of 60 and 65 years old (60≤Age<65), zero otherwise. The degree of statistical significance is shown by 10% (*), 5% (**), and 1% (***).

The sex of the individuals has a very strong impact on the size of the survivor benefit chosen, with men being more likely to select the 100 percent J&S annuity. One should remember that a higher percentage of the retiree benefit that is available to the survivor requires a lower benefit during the life of the retiree. During the pre-COVID period, for individuals 65 years old and older, males are much more likely to opt for the 100 percent benefit option (probability increased by 17.9 percentage points) and less likely to opt for 50 percent benefit (probability decreased by 19.4 percentage points).

As for the impact of the age of the individuals, unsurprisingly, younger individuals are more likely to choose the larger survivor benefits. For example, before 2020, we see that females who are less than 60 years old when retiring are 6 percentage points more likely to choose 100 percent benefit and 4.9 percentage points less likely to choose 50 percent benefit compared to women over the age of 65 at retirement. The impact of the male variable may reflect the age differences between the spouses, with wives typically being younger than their husbands. There are some variations in the probability of choosing the different benefit levels for some years but no noticeable pattern.Footnote ¹⁵ The predicted probabilities of these choices by age and sex are shown in Table 8, panel B.

Similar to the results in Table 7 for the analysis of the distribution choice, there are significant differences in 2021. Looking at the year dummy 2021, we see that holding everything else constant (females older than 65 years old), compared to 2012, individuals are more likely to choose 50 percent benefit (by 3.3 percentage points) and less likely to choose 100 percent benefit (by 3 percentage points). The impact of age is amplified in 2021, compared to years before 2020. For example, in 2021, being less than 60 years old instead of older than 65 reduces the probability of choosing 50 percent benefit by an additional 4.1 percentage points. The impact of sex is also amplified, with an additional decrease in the probability of choosing 50 percent benefit for males in 2021 compared to 2012.

6.2 Years of benefits for certain-and-continuous annuity

Individuals opting for the certain-and-continuous option need to decide the length of time of guaranteed payments. Remember, in this option, benefits will be paid as long as the retiree survives; however, if the retiree dies before the number of guaranteed payments, the beneficiary will continue to receive the same monthly payment for any additional months. We report in Table 5 the number of individuals who choose the eight most popular options in terms of months of benefits (54 individuals are left out). Table 5, panel A contains all the observations, while panels B and C break down the numbers between males and females. The great majority of the observations cluster over 60, 120, and 180 months, with 180 months being the most popular (53 percent of the observations). Looking at panels B and C, we see a meaningful difference between males and females, the latter choosing less frequently the longest option (180 months) and opting relatively more frequently for shorter options like 60 and 120 months.

We next model the selection of the length of time chosen by the individuals who opted for the certain-and-continuous option. For this, we focus on the three most popular options (60, 120, 180 months), which represent the bulk of the observations. This gives us 16,395 observations, with 17, 28, and 54 percent choosing, respectively, 5, 10, and 15 years. We again use a multinomial logit model. We use the same set of explanatory variables as described earlier (year dummies, male and age dummies, plus interactions) and estimate the choice of the length of the guaranty by maximum likelihood. The resulting partial effects evaluated at the mean of the sample and their standard errors are reported in Table 10.

Table 10. Length of certain-and-continuous payouts: marginal effects

The number of observations is 16,395. The numbers between parentheses are standard errors. The Year variables are dummy variables equal to one for the given year, equal to zero otherwise (the year left out is 2012). The variable Male is a dummy variable equal to one if the individual is identified as male, zero otherwise. The two Age variables are dummy variables equal to one if the individual is less than 60 years old (Age<60), zero otherwise; equal to one if the individual is between the age of 60 and 65 years old (60≤Age<65), zero otherwise. The degree of statistical significance is shown by 10% (*), 5% (**), and 1% (***).

Once again, the estimates indicate that the sex of the individual has a large and significant impact on the choices of retirees. Males 65 or older have a 14.8 percentage points higher probability of choosing the longest option (15 years) compared to females, while 8.3 percentage points and 6.5 percentage points lower probability of choosing the 5- and 10-year options, respectively, in years before 2020. Unsurprisingly, age at retirement has an economically very strong impact. For females and years before 2020, holding everything else constant, being less than 60 years old increases the probability of choosing the 15-year option by 25.1 percentage points, while reducing the shorter options by 14 percentage points (10 years) and 11.1 percentage points (5 years). The impact for females between 60 and 65 years old is less but still significant: 15-year up by 10.3 percentage points, 10-year down by 6 percentage points, 5-year down by 4.3 percentage points. Being male instead of female mitigates the impact of age at retirement. For example, for individuals younger than 60 years old before 2020, the probability that a male chooses the 15-year benefit is 9.3 percentage points lower compared to a female. The predicted probabilities of the length of the guarantee by age and sex are shown in Table 8, panel C.

Across time, we see a significant change in the second half of our period, with probabilities of choosing the 15-year option decreasing between 4.9 and 8.2 percentage points, with the biggest increase in probabilities going to the shortest option (5 years). Contrary to the analysis of the distribution choice (Table 7) and the size of the survivor payment (Table 9), the impact of age and sex does not change as much during the COVID years, with none of the partial effects being statistically significant at the 5 percent significance level.

7. Estimated value of monthly benefits

Earlier, we described the payout options available to individuals covered by PBGC-managed plans. In these plans, the benefit is first shown as a straight-life annuity which is determined by the benefit formula of each plan. The PBGC also offers a J&S option and a months-certain annuity. Individuals who select one of these distributions receive a lower monthly benefit in exchange for a survivor's benefit. An interesting observation shown in Table 6 is that the mean J&S benefit was about 10 percent higher than the straight-life annuity for retirees in our sample. This finding suggests that individuals with higher potential straight-life annuities are more likely to select a J&S annuity.

It would be useful to better understand how monthly benefits paid by the PBGC vary by individual characteristics and distributional choices. Using the limited data provided by the PBGC, we estimate the log of real monthly benefits as a function of the age and sex of the retiree. In this model, we use the actual age that benefits were first received along with an indicator variable for being male, with being female as the omitted category. Receiving a straight-life annuity is the base case in this model as the regression includes dummy variables indicating that the retiree selected either a J&S annuity or a months-certain benefit. The sample includes individuals who selected one of these three types of benefits. In addition, we include a series of year variables indicating the year the benefit was first received with 2021 being the base year. The monthly benefits are in real 2021 dollars.

The model is estimated using two samples: the sample we used earlier that includes all individuals who claimed benefits between 2012 and 2021 and a sample that includes only those individuals for whom the PBGC reported their months of services with their employer. As discussed earlier, only about 60 percent of the respondents had a value for months of service. Including this variable reduces the sample to 150,836 individuals. In general, months of service should be positively correlated with the retirement benefit since most DB formulas include length of service in the benefit calculation formula. In the regression, this variable is included in log terms. Thus, the coefficient on months of service provides an estimate of the elasticity between months of service and the monthly benefit. The regression results are shown in Table 11, column 1 that shows the estimated coefficients from the model that includes months of service while column 2 reports the estimates for the complete sample without months of service. The qualitative results are similar for the two models.

Table 11. Estimates of value of monthly benefit (log of monthly benefit)

The numbers between parentheses are standard errors. The variables Months certain and J&S are dummy variables equal to one if the individual selected a months-certain benefits, zero otherwise; equal to one if the individual selected a J&S annuity, zero otherwise. The variable Male is a dummy variable equal to one if the individual is identified as male, zero otherwise. The Year variables are dummy variables equal to one for the given year, equal to zero otherwise (the year left out is 2021). The degree of statistical significance is shown by 10% (*), 5% (**), and 1% (***).

Focusing on the results in column 1, we see an interesting series of results. First, men have an 18 percent higher monthly benefit from the PBGC-managed plans. Second, each additional year of age increases the monthly benefit by 3.8 percent. Third, a 10 percent increase in months of service yields an 8.2 percent increase in the monthly benefit. At the mean values, this implies that 18 additional months of service results in a $48 per month higher benefit. These findings are consistent with our expectations. Introducing these control variables into the analysis helps explain the observations on monthly benefit by type of payout options shown in Table 6. Holding age, sex, and length of service constant, the J&S annuity is now shown to be 2.7 percent smaller than the straight-life annuity. Interestingly, the model in column 2 without months of service shows the J&S coefficient is positive.