FINSER_31-2_Complete_Issue Financial planning time horizon and end-of-life mortality expectations Zhikun Liua,*, Russell James IIIb,*, Qi Sunc aEmployee Benefit Research Institute, 901 D Street, S.W. Suite 802 Washington, DC 20024, USA bPersonal Financial Planning Department, Texas Tech University, Box 41210, Lubbock, TX 79409-1210, USA cPacific Life Insurance Company, 840 Newport Center Drive, Newport Beach, CA 92660, USA Abstract Previous studies demonstrate that individuals’ financial planning time horizons significantly impact spending, saving, charitable giving, and bequest decisions. Using longitudinal and cross-sec- tional Health and Retirement Study data, the analyses in this paper reveal that older American adults’ financial planning horizons are strongly determined by their self-perceived life expectancy. Over time, the changes in self-perceived life expectancy, marital and retirement status, health condi- tions, and wealth level will cause individuals to shift their financial planning horizons. The insight gained in this study helps financial planners to better understand the factors driving changes in client financial planning horizons and consequent financial decision-making. © 2023 Academy of Financial Services. All rights reserved. JEL classifications: G4; D12 Keywords: Behavioral finance; Financial planning horizon; Self-perceived life expectancy; Financial decision making 1. Introduction and literature review Financial planning time horizon is an essential area of economic decision-making for both individuals and households (Dow & Jin, 2013). Various studies have used financial planning horizon as an independent variable to predict various outcomes of interest (Hong & Hanna, 2014). *Corresponding authors. Tel.: +1-903-245-9598; fax: 1-202-775-6360; E-mail address: liuzhikun6@gmail.com (Z. Liu); Tel.: +1-806-834-5130, fax: 1-806-742-5033; E-mail address: russell.james@ttu.edu (R. James) 1057-0810/23/$ – see front matter © 2023 Academy of Financial Services. All rights reserved. Financial Services Review 31 (2023) 107–120 Using the Survey of Consumer Finances data, Fisher and Montalto (2010) identify that households with longer financial planning horizons are more likely to meet saving guide- lines. Rutherford and Devaney (2009) find that households with more than five-year plan- ning horizons are more likely to plan their consumption based on their income and use credit cards for convenience rather than revolving debt vehicles. A longer financial planning horizon also influences individuals’ risk preferences (Castro-González et al., 2020). For example, He and Hu (2007) point out that households with longer financial planning hori- zons tend to hold relatively more stocks in their portfolios. When studying the medical cost risk, Ayyagari and He (2017) find that when facing increased medical expenditure risk, people who have a planning horizon longer than five years are willing to take more equity exposure. Liu et al. (2021) study how the financial planning horizon is associated with investors’ stock market return expectations. They con- clude that having a financial planning horizon of one year or less is related to higher expecta- tions of a 20% loss in the next year’s stock market. Financial planning horizon also impacts the actual investment behavior. Asebedo and Browning (2020) found that a longer financial planning horizon is associated with a lower retirement portfolio withdrawal rate. Munnell et al. (2001) conclude that employees with short financial planning horizons have a lower taste of saving and a smaller probability of participating in a pension plan. Using the Health and Retirement Study (HRS) data, Liu and James (2017) find that individuals with longer financial planning horizons are more likely to have valid estate planning documents. Liu and James (2020) find that such individuals are also more likely to make substantial gifts to charity. Past studies found that the financial planning horizon is related to psychological and self- perceived financial well-being. For example, Malroutu and Xiao (1995) conclude that pre- retirees who plan to save for the next five years are more likely to perceive having adequate retirement income versus those who do not have saving plans. Choung et al. (2022) found that people with major depression tend to have a shorter financial planning horizon. While most studies treat people’s financial planning horizon as a measure of time prefer- ence (Khwaja, Sloan, & Salm, 2006). Hong and Hanna (2014) point out that it is unclear whether financial planning horizons reflect time preferences or situational variables. They indicate that the financial planning horizon is not measuring time preference but a situational variable because it is significantly related to the demographic variables of the respondents using the Survey of Consumer Affairs (SCF) data. They also suggest that time preference is constant, but the financial planning horizon is not constant. Therefore, the financial planning horizon is not measuring time preference. However, as proposed by Becker and Mulligan (1997), time preference can be endogenously determined, and over time, investment in edu- cation and practice in imagining future outcomes will decrease time discounting. This is consistent with rising time preference into middle age (such as 43). Additionally, time preference should be expected to change in later life based primarily upon changes in life expectancy. Baranov and Kohler (2018) find that individuals actively change their investment decisions based on their subjective longevity, even in a low-income environment. Indeed, it would be irrational if an individual maintains a high preference for rewards paid over 30 years, as one’s life expectancy shortens with aging (moved from 40 to 108 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 20 to 10 to five years.) The explanation is consistent with an increasing rate of time discount- ing later in life, and it is the focus of the current study. Some initial evidence appears inconsistent with this argument. However, if the measure- ment for the financial planning horizon does not extend beyond 10 or 20 years, then longev- ity predictions would not be expected to greatly influence this factor until the end of life. In particular, this occurs when subjective life expectancy falls below the highest financial plan- ning measurement categories. It is this end-of-life subjective life expectancy effect that the current paper explores. The important implications of the households’ and individuals’ financial planning hori- zons motivate this study to investigate the different factors that affect people’s planning horizons. In particular, this study is interested in analyzing the relationship between people’s financial planning horizon and their self-perceived life expectancy. Using the HRS data, both cross-sectional and longitudinal regressions demonstrate the ex- istence of such a relationship. This paper provides robust evidence affirming the following hypothesis: Individuals’ self-perceived life expectancies significantly affect their financial planning horizon. This relationship reveals that the financial planning horizon is, at least in part, a measure of rational (but not fixed) time discounting, based on life expectancy. 2. Data This paper uses the HRS survey data to conduct cross-sectional and longitudinal regres- sion analyses. Variables are selected from the RAND (Version P) HRS data, the cross-wave tracker file, as well as the core HRS data sets from 1998 to 2018. Since the HRS survey does not include direct life expectancy variables, this study constructs respondent-level, self-per- ceived life expectancy variables for each of the available HRS survey waves. The underlin- ing principle for constructing such variables is described as follows: An individual’s self- perceived life expectancy is equal to her/his estimated probability of living to a target age multiplied by the difference between the target age and this individual’s current age. Under this principle, the respondents’ self-perceived life expectancy variables are constructed using the formula below (the formula uses the 2018-year wave as an example; life expectancy var- iables are generated similarly for all the available waves): YearstoLive2018 = R14LIV10ð Þ=100 # R14LIV10A$ Age2018ð Þ In this formula, the variable “YearstoLive(wave)” represents the respondent’s self-per- ceived life expectancy at the point of the survey. The variables R(wave)LIV10 and R(wave) LIV10A are selected from the RAND HRS data. Among them, R(wave)LIV10 is the self- reported probability of living to a certain target age, selected according to the HRS guide- lines, where 0 means “absolutely no chance,” and 100 means “absolutely certain.” For respondents whose age is less than 65 at the point of the survey, this target age is set to be 85. For respondents whose age is between 65 and 69, this target is set to be 80. For those between 70 and 74 years old, this target age is set to be 85. For those between 75 and 79, the Z. Liu et al. / Financial Services Review 31 (2023) 107–120 109 target is 90. For respondents in the 80-84 age range, the target age is 95. Finally, for respondents whose age is between 85 and 89, the target age is set to be 100. The variable R (wave)LIV10A gives the specific age used in the questionnaire for each respondent during the survey, which ranges from 80 to 100. The summary statistics for the constructed self-perceived life expectancy variables (YearstoLive(wave)) from the year 2000 to 2018 are reported in Table 1. Note that the RAND HRS dataset does not contain the R(wave)LIV10 and R(wave)LIV10A variables before the 2000 wave. Therefore, this summary statistics table uses the 2000-2018 wave range to demonstrate the respondents’ self-perceived life expectancies. The following figure is plotted using the 2018 wave data with individual-level weight applied. This graph indicates that the constructed life expectancy variable has an inverse relationship with the respondent’s age at the survey. This relationship is verified with OLS regression: with high levels of significance, a one-year increase in respondents’ age will, on average, decrease their self-perceived life expectancy by 0.295 years. The control variables of this weighted OLS regression include marital status, presence of children, years of educa- tion, wealth level, retirement status, and various health indicators. Next, this study explores the relationship between the financial planning horizon variable and the constructed life expectancy variable. For the 2018 wave, two types of cross-sectional regressions (OLS and ordered probit) are conducted. The summary statistics for the cross- sectional regression variables are reported in the following table: 3. Model Becker and Mulligan (1997) provide both theoretical and empirical evidence to conclude that time preference varies across individuals and wealth causes patience. Trostel and Taylor (2001) argue that future discounting occurs because the expected marginal utility of con- sumption is declining. In other words, an individual’s ability to enjoy future consumption is expected to be lower. If the financial planning horizon measures people’s time discounting preference, it should be strongly correlated with rational life expectancy changes as well as Table 1 Summary statistics for the imputed life expectancy variables Variables Number of observations Mean (weighted) Standard deviation (weighted) YearstoLive2000 15,554 8.934495 0.0989592 YearstoLive2002 14,420 8.322585 0.0827116 YearstoLive2004 16,780 9.37001 0.0876633 YearstoLive2006 15,467 8.93313 0.099443 YearstoLive2008 14,450 9.078261 0.0825122 YearstoLive2010 18,911 9.804211 0.1208808 YearstoLive2012 17,925 9.176823 0.1041344 YearstoLive2014 16,452 8.87597 0.1146203 YearstoLive2016 12,429 9.110994 0.1268747 YearstoLive2018 12,500 8.509271 0.1353772 Note. Respondent-level weights are applied to each wave. 110 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 wealth changes. The following model and the empirical analysis presented in the next sec- tions of this paper provide both theoretical and empirical evidence for this hypothesis. In the general form of Becker’s model of patience formation, a consumer is assumed to live a finite number of periods. Therefore, the consumer maximizes V = o T i=0 b Sð Þi % fi cið Þ (1) where T represents the length of the consumer’s life span, ci’s are the consumption levels at period i, the functions fi(·) map the consumptions at period i into pleasures at that period, aka utility function. Future utilities are discounted based on the discount function b (·), which is less than 1. Becker and Mulligan define S to be the effort people make to increase their appre- ciation of future utility. S is determined by the time and effort spent in the appreciation of pleasures of the future, by spending on certain goods that distract one’s attention away from current pleasure, and toward future ones by schooling, saving, and so forth. In other words, S is the “patient” factor (Becker & Mulligan, 1997). Applying the patience formation model to this paper, which analyzes the determinants of financial planning horizons, S could represent the effort and time spent by the respondents to make long-term financial plans. Assuming the present values of all assets and earnings are calculated into an initial endowment of wealth A0, then the intertemporal budget constraint can be described as o T i=0 Rici þ pS=A0 (2) where Ri’s represent the interest factors and p stands for the price of S. The first-order con- ditions with respect to consumption at each period are as follows: b 0 Sð Þ o T i=0 i % b Sð Þ ! "i$1 % fi cið Þ " # = k0 = f 0 0 c0ð Þ (3) where l 0 denotes the marginal utility of wealth. The marginal benefit of S depends on the length of the consumer’s life (T), the level of future utilities, and the level of discount rate.1 From the first order conditions depicted by Eq. (3), if an increase in life expectancy (from T to T + Dt) is accompanied by an increase of lifetime earnings’ that maintains the marginal utility of wealth, l 0, then this change (Dt) will increase the marginal benefit from investing in future-oriented capital – the patient factor S.2 Therefore, longer lifetime, or self-perceived life expectancy increase, will directly motivate consumers to plan further into the future. Hence, this model predicts a positive relationship between the consumers’ life expectancy and their financial planning horizons. 4. Results To examine which factors and to what extent each of these factors drive the change in people’s financial planning horizons, this study compares the OLS and ordered probit Z. Liu et al. / Financial Services Review 31 (2023) 107–120 111 regression results cross-sectionally. This paper also conducts longitudinal robustness checks to examine whether intrapersonal changes in these factors cause changes in the financial planning horizon over time. Table 2 reports the weighted means for the variables of interest in the 2018 wave. Figure 1 demonstrates how the “years to live” variable measuring self- perceived life expectancy varies with age. Table 3 reports the ordinary least squares regres- sion results with the 2018 wave respondent level weight applied. The results in Table 3 demonstrate significant relationships between the financial planning horizon variable and the independent variables such as self-perceived longevity, race, mari- tal status, retirement status, and wealth levels. Note that the age variable becomes significant after omitting the life expectancy variable from the OLS regression. This finding indicates that the effect of aging on financial planning horizon is captured by the self-perceived life expectancy variable. Similarly, the heart condition variable becomes marginally significant after omitting the life expectancy variable from the OLS regression. This suggests that part of the effect of a heart condition diagnosis on financial planning horizon is captured by the self-perceived life expectancy variable. Table 4 reports a more detailed analysis, showing the average marginal effect of the or- dered probit regression for the 2018 wave data. This regression also takes the 2018 respond- ent-level sample weight into consideration. Based on the results of both OLS and the ordered probit regression, there is a strong relationship between the respondent’s self-per- ceived life expectancy and their financial planning horizon. Table 2 Summary statistics for the 2018 wave cross-sectional analysis Variables Mean Standard deviation Financial planning horizon (next few months ¼ 1; next year ¼ 2; next few years ¼ 3; next 5–10 years ¼ 4; longer than 10 years ¼ 5) 3.345433 1.171501 Years to Live (constructed variable measuring self-perceived life expectancy) 8.118878 5.552278 Male (male ¼ 1; female ¼ 0) 0.4724504 0.4992765 Age 68.72538 7.4852 Black race (White ¼ 0; Black ¼ 1; else ¼ 0) 0.0805977 0.2722359 Other race (White ¼ 0; Black ¼ 0; else ¼ 1) 0.0591926 0.2360019 Married (married ¼ 1; else ¼ 0) 0.6936006 0.4610309 Presence of child (have children ¼ 1; no children ¼ 0) 0.904557 0.2938471 Retired (retired ¼ 1; else ¼ 0) 0.5474733 0.5326426 Years of education 13.83241 2.736728 Cancer (have cancer (excluding skin) ¼ 1; else ¼ 0) 0.1679238 0.3738256 Heart condition (diagnosed with heart condition ¼ 1; else ¼ 0) 0.241735 0.4281654 Wealth (natural logarithm of total wealth) 12.43344 1.98501 Note. Respondent-level sample weight of the 2018 wave HRS data is applied to the summary statistics. 112 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 Both the OLS and the ordered probit regression results show that a respondent’s financial planning horizon is strongly correlated with her or his self-perceived life expectancy, race, marital status, retirement status, and wealth level. Ceteris paribus, individuals with longer self-perceived life expectancy, on average, tend to report longer financial planning horizons. This result provides direct evidence for the main hypothesis of this study and agrees with the prediction of Becker and Mulligan’s (1997) model of patience formation. Married couples, on average, are more likely to report longer financial planning horizons. Wealth level is also associated with the respondents’ financial planning horizon, indicated by both OLS and the ordered probit results. Everything else equal, people who have more wealth are more likely to plan relatively longer into the future. Retirement status also stands out as a potential factor affecting financial planning horizon. Interestingly, age is only a marginal significant determinant of people’s financial planning horizons. This result agrees with the findings of Trostel and Taylor (2001). The following graph (Fig. 2) captures the average financial planning horizon in each age category. As one can see from the illustration above, the average financial planning horizons in dif- ferent age groups do not follow a declining pattern with age increases. In other words, age is not closely related to self-reported financial planning horizon among these older American adults. Dow and Jin (2013) also point out that age is not a primary driver of an individual’s financial planning horizon, which agrees with the finding of this paper. However, this study does not support Dow and Jin’s (2013) argument, which indicates that “life expectancy is a poor predictor of financial planning horizon.” Instead, robust evidence has been provided by this study supporting the hypothesis that older American adults’ self-perceived life expect- ancy affects their financial planning horizon positively and significantly. Based on the model of patience formation from Becker and Mulligan (1997), the patience factor, which measures the respondents’ efforts and time spent to make long-term financial plans in this case, should decrease later in life based on life expectancy after peaking around the middle age. Therefore, self-perceived life expectancy should affect financial planning time horizon at an older age when the life expectancy is closer to the financial planning horizon measurement Fig. 1. Average life expectancy across age spectrum. Z. Liu et al. / Financial Services Review 31 (2023) 107–120 113 zones. Since the respondents in the HRS data set are older American adults, one should see a significant impact of subjective longevity on the financial planning horizon. 5. Longitudinal robustness check After confirming the cross-sectional relationship between the people’s financial planning horizon and their self-perceived life expectancy, the next step is to check whether this rela- tionship persists with the same individual over time. As Fig. 3 indicates, the average values of the respondent’s self-perceived life expectancies do not vary significantly across different HRS sample waves from the year 2000 to the year 2018. This figure demonstrates the con- sistency of our different sample waves. Table 4 shows the fixed-effect longitudinal analyses and Table 5 shows random-effect longitudinal analyses from the year 2000 to 2018. Notice that the control variables that do not vary over time for our respondent sample (such as race, gender, years of education, etc.) are excluded from the longitudinal regression analyses. Table 3 OLS regression results on financial planning horizon Variables Coefficients Coefficients (“Years to Live” variable omitted) Years to live (constructed variable measuring self-perceived life expectancy) 0.0195*** (0.0040) Male 0.0228 0.0005 (male ¼ 1; female ¼ 0) (0.0358) (0.0357) Age $0.0047 $0.0115*** (0.0025) (0.0023) Black race (White ¼ 0; Black ¼ 1; else ¼ 0) $0.2500*** (0.0709) $0.2488** (0.0730) Other race (White ¼ 0; Black ¼ 0; else ¼ 1) $0.0794 (0.0744) $0.1288 (0.0725) Married (married ¼ 1; else ¼ 0) 0.1264** (0.0399) 0.1437*** (0.0389) Presence of child (have children ¼ 1; no children ¼ 0) $0.0752 (0.0679) $0.0634 (0.0637) Retired (retired ¼ 1; else ¼ 0) 0.1063** (0.0326) 0.1005** (0.0327) Years of education 0.0004 (0.0063) 0.0069 (0.0061) Cancer (have cancer (excluding skin) ¼ 1; else ¼ 0) $0.0045 (0.0463) $0.0286 (0.0450) Heart condition (diagnosed with heart condition ¼ 1; else ¼ 0) $0.0585 (0.0390) $0.0828* (0.0403) Wealth (natural logarithm of total wealth) 0.1033*** (0.0152) 0.1083*** (0.0144) Notes. The number of observations is 11,433 for the OLS regression, including all the variables listed above. The number of observations is 11,826 for the OLS regression without the “Years to Live” variable. Respondent- level ample weights are applied. Standard errors are reported in parentheses. ***Statistically significant at 0.1- percent level. **Statistically significant at 1- percent level. *Statistically sig- nificant at 5- percent level. 114 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 T ab le 4 A v er ag e m ar g in al ef fe ct o f th e o rd er ed p ro b it re g re ss io n re su lt s V ar ia b le s A v er ag e m ar g in al ef fe ct s F in an ci al p la n n in g h o ri zo n N ex t fe w m o n th s N ex t y ea r N ex t fe w y ea rs N ex t 5 – 1 0 y ea rs L o n g er th an 1 0 y ea rs T h re sh o ld p ar am et er s 1 2 3 4 5 Y ea rs to li v e $ 0 .0 0 33 * * * $ 0 .0 0 2 1 * * * $ 0 .0 0 2 1 * * * 0 .0 0 2 9 * ** 0 .0 0 4 5 * * * (0 .0 0 06 ) (0 .0 0 04 ) (0 .0 0 0 4 ) (0 .0 0 0 6 ) (0 .0 0 09 ) M al e $ 0 .0 0 30 $ 0 .0 0 1 9 $ 0 .0 0 1 9 0 .0 0 2 7 0 .0 0 4 2 (0 .0 0 58 ) (0 .0 0 37 ) (0 .0 0 3 6 ) (0 .0 0 5 1 ) (0 .0 0 79 ) A g e 0 .0 0 10 * 0 .0 0 0 6 * 0 .0 0 0 6 * $ 0 .0 0 0 9 * $ 0 .0 0 1 3 * (0 .0 0 04 ) (0 .0 0 03 ) (0 .0 0 0 3 ) (0 .0 0 0 3 ) (0 .0 0 06 ) B la ck ra ce 0 .0 3 35 * * 0 .0 2 1 6 * * 0 .0 2 1 0 * * $ 0 .0 2 9 9 * * $ 0 .0 4 6 2 * * (0 .0 1 11 ) (0 .0 0 74 ) (0 .0 0 7 3 ) (0 .0 1 0 1 ) (0 .0 1 57 ) O th er ra ce 0 .0 0 90 0 .0 0 5 8 0 .0 0 5 6 $ 0 .0 0 8 0 $ 0 .0 1 2 4 (0 .0 1 17 ) (0 .0 0 75 ) (0 .0 0 7 3 ) (0 .0 1 0 4 ) (0 .0 1 62 ) M ar ri ed $ 0 .0 1 83 * * $ 0 .0 1 1 8 * * $ 0 .0 1 1 4 * * 0 .0 1 6 3 * * 0 .0 2 5 2 * * (0 .0 0 63 ) (0 .0 0 41 ) (0 .0 0 4 0 ) (0 .0 0 5 7 ) (0 .0 0 86 ) P re se n ce o f ch il d 0 .0 1 27 0 .0 0 8 2 0 .0 0 8 0 $ 0 .0 1 1 3 $ 0 .0 1 7 6 0 .0 1 09 0 .0 0 6 9 0 .0 0 6 7 0 .0 0 9 7 0 .0 1 4 9 R et ir ed $ 0 .0 2 14 * * * $ 0 .0 1 3 8 * * * $ 0 .0 1 3 4 * * * 0 .0 1 9 1 * ** 0 .0 2 9 6 * * * (0 .0 0 52 ) (0 .0 0 35 ) (0 .0 0 3 5 ) (0 .0 0 4 6 ) (0 .0 0 75 ) Y ea rs o f ed u ca ti o n 0 .0 0 00 0 .0 0 0 0 0 .0 0 0 0 0 .0 0 0 0 0 .0 0 0 0 (0 .0 0 10 ) (0 .0 0 07 ) (0 .0 0 0 6 ) (0 .0 0 0 9 ) (0 .0 0 14 ) C an ce r 0 .0 0 03 0 .0 0 0 2 0 .0 0 0 2 $ 0 .0 0 0 2 $ 0 .0 0 0 4 (0 .0 0 72 ) (0 .0 0 47 ) (0 .0 0 4 5 ) (0 .0 0 6 4 ) (0 .0 1 00 ) H ea rt co n di ti o n 0 .0 0 84 0 .0 0 5 4 0 .0 0 5 3 $ 0 .0 0 7 5 $ 0 .0 1 1 6 (0 .0 0 60 ) (0 .0 0 39 ) (0 .0 0 3 7 ) (0 .0 0 5 4 ) (0 .0 0 82 ) W ea lt h $ 0 .0 1 55 * * * $ 0 .0 1 0 0 * * * $ 0 .0 0 9 7 * * * 0 .0 1 3 8 * ** 0 .0 2 1 4 * * * (0 .0 0 24 ) (0 .0 0 15 ) (0 .0 0 1 4 ) (0 .0 0 2 1 ) (0 .0 0 33 ) N o te s. N u m b er o f o b se rv at io n s is 1 1 ,4 3 3 . R es p o n d en t- le v el am p le w ei g h ts ar e ap p li ed . S ta n d ar d er ro rs ar e re p o rt ed in p ar en th es es . * * * S ta ti st ic al ly si g n ifi ca n t at 0 .1 - p er ce n t le v el . * * S ta ti st ic al ly si g n ifi ca n t at 1 - p er ce n t le v el . * S ta ti st ic al ly si g n ifi ca n t at 5 - p er ce n t le v el . Z. Liu et al. / Financial Services Review 31 (2023) 107–120 115 The results from the longitudinal OLS regression confirm that the respondents’ financial planning horizons are strongly correlated with their self-perceived life expectancy. Moreover, changes in self-perceived longevity result in changes to financial planning horizons. The marginal effects of the longitudinal ordered probit regression in Table 6 confirm that, ceteris paribus, respondents with longer self-perceived life expectancy tend to have longer Fig. 2. Average financial planning horizon in different age categories (2018). Note: The 2018 HRS respondent- level sample weight is applied to this summary graph. Fig. 3. Average life expectancy summary and shift-trend (2000-2018). 116 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 financial planning horizons, controlling their marital and retirement status, presence of chil- dren, wealth levels, and health conditions. The longitudinal analysis results from the tables above affirm the main conclusion of this study: The respondent’s financial planning horizon is strongly correlated with rational life expectancy changes as well as wealth changes. Hence, financial planning horizon is a partial measure of rational (not fixed) time discounting. Overall, multivariate regression results show evidence that people’s life expectancy influences their planning horizon, even after controlling for their financial and health conditions. 6. Conclusion Using both cross-sectional and longitudinal analyses with the HRS data, this paper exam- ines the potential factors that drive both differences and changes in American adults’ (age 50+) financial planning time horizons and reviews the significance of people’s financial planning horizon shifts. The empirical analysis in this study provides direct evidence to sup- port the hypothesis that an individual’s financial planning horizon is strongly correlated with their self-perceived life expectancy, race, marital status, retirement status, and wealth level. Ceteris paribus, people with longer self-perceived longevity tend to have longer financial planning horizons. Over time, changes in self-perceived life expectancy, marital and retire- ment status, health conditions, and wealth level will cause individuals to shift their financial planning horizons. The findings of this paper can help financial planners to identify the factors, such as self- perceived longevity and wealth changes, that may cause potential shifts to the financial plan- ning time horizon of their clients. Practitioners can also benefit from this study in terms of educating their clients about the importance of their decisions on the financial planning Table 5 Longitudinal OLS regressions on financial planning horizon Variables Fixed-effect Random-effect Years to live 0.00479*** 0.00847*** ($0.00121) ($0.000765) Married $0.0614* 0.0692*** ($0.0265) ($0.0146) Presence of child 0.0796 $0.0505* ($0.0715) ($0.0243) Retired 0.0780*** 0.0340** ($0.0171) ($0.0125) Cancer $0.0498 $0.016 ($0.0348) ($0.0195) Heart condition $0.0541* $0.0519** ($0.0275) ($0.0164) Wealth 0.0294*** 0.103*** ($0.00704) ($0.00329) Notes. The number of observations is 49,690. ***Statistically significant at 0.1- percent level. **Statistically significant at 1- percent level. *Statistically sig- nificant at 5- percent level. Z. Liu et al. / Financial Services Review 31 (2023) 107–120 117 horizon, which is one of the most important areas of economic decision-making for both individuals and households. The discoveries in this paper demonstrate that financial planning time horizon is not stable among older American adults and can be expected to change in later life, both rationally and empirically. Therefore, practitioners should take into considera- tion that clients’ self-perceived life expectancy has a significant impact on their financial planning time horizon later in life. If the practitioners want to encourage longer financial planning horizon at the later life stages, non-consumption issues, such as bequest motives, may become more important. 7. Future research This paper presents a new way to construct the variable that depicts the respondents’ self- perceived life expectancy using the Health and Retirement Study data. This life expectancy variable has an inverse relationship with the respondent’s age and this inverse relationship is verified with OLS regression analysis. More robust tests can be performed on this self-per- ceived life expectancy variable to examine whether it is a good measurement of the respond- ents’ longevity expectation and whether it can be used to construct an indicator for their mortality salience. Once these tests are performed and approved positive, this variable can then be widely used as a measure of self-perceived longevity (or mortality salience) for stud- ies in different financial planning fields such as bequest motives, investment choices, long- term care insurance adoptions, and estate planning decisions. Table 6 Marginal effects of longitudinal ordered probit regression Variables Average marginal effects Financial planning horizon Next few months Next year Next few years Next 5-10 years Longer than 10 years Threshold parameters 1 2 3 4 5 Years to live $0.0019*** $0.0009*** $0.0006*** 0.0014*** 0.0019*** (0.0001) (0.0001) (0.0000) (0.0001) (0.0001) Married $0.0135*** $0.0062*** $0.0042*** 0.0104*** 0.0135*** (0.0027) (0.0012) (0.0009) (0.0021) (0.0027) Presence of child 0.0129** 0.0059** 0.0040** $0.0100** $0.0129** (0.0045) (0.0021) (0.0014) (0.0035) (0.0045) Retired $0.0083*** $0.0038*** $0.0026*** 0.0064*** 0.0083*** (0.0024) (0.0011) (0.0007) (0.0018) (0.0023) Cancer 0.0031 0.0014 0.0010 $0.0024 $0.0031 (0.0036) (0.0017) (0.0011) (0.0028) (0.0036) Heart condition 0.0102*** 0.0047*** 0.0032*** $0.0079*** $0.0102*** (0.0031) (0.0014) (0.0010) (0.0024) (0.0030) Wealth $0.0190*** $0.0087*** $0.0059*** 0.0147*** 0.0190*** (0.0006) (0.0003) (0.0002) (0.0005) (0.0006) Notes. The number of observations is 49,690. Respondent-level sample weights are applied. Standard errors are reported in parentheses. ***Statistically significant at 0.1- percent level. **Statistically significant at 1- percent level. *Statistically sig- nificant at 5- percent level. 118 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 Literature review suggests that, everything else equal, individuals with longer financial planning time horizons tend to enjoy good implications of their financial decision making. For instance, long-term planners are more likely to meet certain saving guidelines, use credit cards for convenience instead of revolving debts, hold relatively more stocks in their portfo- lio, and have valid estate planning documents. Future studies can further explore different behavioral strategies on how to educate clients about the importance of longer-term planning and promote this type of financial planning based on each client’s unique situation. Notes 1 In the special two-period case, that is, T¼1, equation (2) can be simplified as b 0 Sð Þ f1 c1ð Þ ! " = k0 = f 0 0 c0ð Þ (4) 2 If the discount rate is unchanged, Becker and Mulligan (1997) suggest that con- sumption in the first T years would not be affected. The consumption during Dt would be positive. By Eq. (3), this will cause an increase of the marginal benefit from time preference investing. Thus, the equilibrium accumulation of future-ori- ented capital is increased. References Asebedo, S. D., & Browning, C. M. (2020). The psychology of portfolio withdrawal rates. Psychology and Aging, 35, 78–90. Available at: https://doi.org/10.1037/pag0000424 Ayyagari, P., & He, D. (2017). The role of medical expenditure risk in portfolio allocation decisions. Health Economics, 26, 1447–1458. Available at: https://doi.org/10.1002/hec.3437 Baranov, V., & Kohler, H. P. (2018). The impact of AIDS treatment on savings and human capital investment in Malawi. American Economic Journal: Applied Economics, 10, 266–306. Available at: https://doi.org/10.1257/ app.20150369 Becker, G. S., & Mulligan, C. B. (1997). The endogenous determination of time preference. The Quarterly Journal of Economics, 112, 729–758. Available at: https://doi.org/10.1162/003355397555334 Castro-González, S., Fernández-López, S., Rey-Ares, L., & Rodeiro-Pazos, D. (2020). The influence of attitude to money on individuals’ financial well-being. Social Indicators Research, 148, 747–764. Available at: https://doi. org/10.1007/s11205-019-02219-4 Choung, Y., Chatterjee, S., & Pak, T. Y. (2022). Depression and financial planning horizon. Journal of Behavioral and Experimental Economics, 98, 101877. Available at: https://doi.org/10.1016/j.socec.2022.101877 Dow, J. P., Jr., & Jin, Y. (2013). The determination of individual financial planning horizons. Southwestern Economic Review, 40, 137–149. Fisher, P. J., & Montalto, C. P. (2010). Effect of saving motives and horizon on saving behaviors. Journal of Economic Psychology, 31, 92–105. Available at: https://doi.org/10.1016/j.joep.2009.11.002 He, P., & Hu, X. (2007). Household investment – The horizon effect. Available at SSRN. Available at: https:// ssrn.com/abstract¼798431 or http://dx.doi.org/10.2139/ssrn.798431 Hong, E. O., & Hanna, S. D. (2014). Financial planning horizon: A measure of time preference or a situational factor? Journal of Financial Counseling and Planning, 25, 184–196. Khwaja, A., Sloan, F., & Salm, M. (2006). Evidence on preferences and subjective beliefs of risk takers: The case of smokers. International Journal of Industrial Organization, 24, 667–682. Available at: https://doi.org/ 10.1016/j.ijindorg.2005.10.001 Z. Liu et al. / Financial Services Review 31 (2023) 107–120 119 Liu, Y., Guillemette, M., & Gray, B. (2021). A behavioral understanding of stock market return expectations: From evaluation frequency and financial planning horizon perspective. Available at SSRN 3906757. Liu, Z., & James, R. N. III., (2017). Behavioral and non-behavioral factors affecting will and trust ownership. Available at SSRN https://ssrn.com/abstract¼3037748 Liu, Z., & James, R. N. III., (2020). Finding the next major donor: The relationship between financial planning horizon and charitable giving. Journal of Personal Finance, 19, 49–65. Malroutu, Y. L., & Xiao, J. J. (1995). Perceived adequacy of retirement income. Financial Counseling and Planning, 6, 17–23. Munnell, A. H., Sunden, A., & Taylor, C. (2001). What determines 401(k) participation and contributions. Social Security Bulletin, 64, 64–75. Rutherford, L. G., & DeVaney, S. A. (2009). Utilizing the theory of planned behavior to understand convenience use of credit cards. Journal of Financial Counseling and Planning, 20, 48–63. Trostel, P. A., & Taylor, G. A. (2001). A theory of time preference. Economic Inquiry, 39, 379–395. Available at: https://doi.org/10.1093/ei/39.3.379 120 Z. Liu et al. / Financial Services Review 31 (2023) 107–120