| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Family Studies, University of Kentucky, Lexington.
2 College of Social Work, The Ohio State University, Columbus.
Address correspondence to Hyungsoo Kim, Department of Family Studies, University of Kentucky, 315 Funkhouser Building, Lexington, KY 40506. E-Mail: hkim3{at}uky.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. The data were from the 1998, 2000, and 2002 Health and Retirement Study and the 2003 Health and Retirement Study Consumption and Activities Mail Survey. We conducted Tobit regression analyses on five consumption categories of basic needs (such as food) and higher order needs (such as trips and dining out).
Results. Consumption and driving status were significantly associated, showing that driving cessation was related to a 46% to 63% reduction in spending on trips, tickets, and dining out. Another significant relationship emerged between consumption and having never driven. Driving cessation was minimally related to consumption of basic needs (such as food and clothing) and was more strongly associated with higher order needs (such as trips).
Discussion. The findings demonstrate the association between older people's driving status and consumption, specifically higher order activities. Older persons who drive and, presumably, have more opportunities to go to stores, restaurants, and other outside events, spend more on food, tickets, and dining out than those who cease driving or have never driven. Although the direction of causality remains unclear, these findings have implications for those concerned with alternative transportation resources for older adults.
DRIVING is the primary source of transportation for most older people (AARP, 2005
). About 90% of the trips among people older than age 65 are in a private vehicle, either as a driver or passenger, and about 87% of the trips of those older than age 85 are in automobiles (Rosenbloom, 2003). Although people now drive later in life than they did in the past, and this trend most likely will continue, experts also expect that 15% to 52% of the older population will be nondrivers by 2020 (Koffman, Raphael, & Weiner, 2004). About 21% (6.5 million) of seniors aged 65 and older do not drive, and the percentages are even higher among older minority populations (U.S. Government Accountability Office, 2004). More than 600,000 people aged 70 and older stop driving every year and need alternative transportation to maintain their lives (Foley, Heimovitz, Guralnik, & Brock, 2002).
Older people stop driving for many reasons. The most common reasons are medical and age-related problems, such as vision problems (Dellinger, Sehgal, Sleet, & Barrett-Connor, 2001). Older women, who typically quit driving sooner and maintain more transportation alternatives than older men do, are more likely to cite licensing problems and costs of maintaining an automobile as reasons that they no longer drive (Dellinger et al., 2001; Stutts, Wilkins, Reinfurt, Rodgman, & Van Heusen-Causey, 2001). Older women are three times more likely to stop driving than older men despite similarities in health and functional status (Foley et al., 2002; Stutts et al., 2001). Although most drivers voluntarily cease driving, older people with early-stage dementia often remain unaware of their impaired driving skills (Dobbs, Carr, & Morris, 2002; Gil et al., 2001). The most influential risk factors for driving cessation are physical mobility limitations (especially problems associated with walking), strokes, vision problems, and cognitive impairments (MacLeod, Geyer, Satariano, & Ragland, 2004). Nonmedical problems—such as household income, fear of crime, and being in an accident—also discourage older people from driving (Ragland, Satariano, & MacLeod, 2004).
Driving cessation affects older people and their spouses in several ways. Many older widowers become socially isolated after they stop driving (Richardson & Balaswamy, 2004). Older nondrivers have higher levels of depression; engage in fewer out-of-home activities; and spend less time shopping, attending movies, dining out, or taking day or overnight trips than do older drivers (Marottoli et al., 1997; Marottoli et al., 2000). Driving status affects older people's access to grocery stores, banks, physicians, pharmacies, and entertainment, and it can affect their health if it prevents them from purchasing medicines or visiting their doctors (Carp, 1988).
Public and private transportation alternatives can mitigate the disadvantages of driving cessation. Nondriving older men face about 6 years of reliance on alternative transportation, whereas nondriving older women face about 10 years (Foley et al., 2002). However, few older adults avail themselves of these alternatives because they are unsafe, expensive, or simply unavailable (Stutts et al., 2001; U.S. Government Accountability Office, 2004). Many older people worry about their safety on public buses, on trains, and in taxis; and they prefer riding with friends, neighbors, or family members (U.S. Government Accountability Office, 2004). The use of public transit or special transit services by older people dropped by 50% between 1995 and 2001. Strict eligibility criteria often deter older drivers from using special transportation services, and few of those services allow eligible passengers to carry packages or travel to multiple destinations or to cultural events (Rosenbloom, 2003; U.S. Government Accountability Office, 2004). These transportation alternatives are almost nonexistent in rural areas. Supplemental Transportation Programs for Seniors, which rely heavily on volunteer drivers, are emerging in many communities for the frailest older adults as the number of older nondrivers increases (Beverly Foundation, 2004).
Previous investigators have found that driving cessation influences older people's well-being, access to health care, and out-of-home interactions, but no researchers have considered how this life transition is related to older adults' consumption patterns (specifically their spending). In this article we examine the association between nondriving and older people's spending for basic and recreational needs. We distinguish between life maintenance and higher order spending in order to evaluate older adults' consumption. We also contribute to the growing data on driving cessation among older adults by distinguishing older adults who have stopped driving from their counterparts who have never driven and thus have no similar effects on their consumption patterns.
The life cycle hypothesis of consumption has been a theoretical framework to explain individuals' consumption behaviors. Proponents of the life cycle model of consumption assume that most people save in early and middle adulthood in order to accumulate the money that they need for food, housing, and other expenses in retirement (Modigliani & Brumberg, 1954). Retired older adults spend less as work-related consumption requirements change and they substitute home activities for market purchases (Aguiar & Hurst, 2005; Banks, Blundell, & Tanner, 1998; Hurd & Rohwedder, 2004). Older adults spend less than younger adults, in part because their health declines (Browning & Crossley, 2001; Deaton, 1992; Viscusi & Evans, 1990). As people age and their health resources decrease, they experience more mobility constraints, which deter spending in certain areas, especially consumption on out-of-home entertainment (Borsch-Supan & Stahl, 1991; Browning & Crossley, 2001; Viscusi & Evans, 1990). Unless alternative transportation options or purchases from home are readily available, people who used to drive gradually consume less as their access to stores and recreational pursuits decreases. People who no longer drive must prioritize their spending, especially when transportation alternatives are scarce. They focus more on meeting what Carp (1988) referred to as "life maintenance" needs (including food, clothing, banking, and health care) than on higher order consumption that provides fun and relaxation. Some experts call these "essential" or "life-sustaining trips," as opposed to "quality of life" or "life-enhancing trips" (U.S. Government Accountability Office, 2004). Cutler and Katz (1991) and Fisher, Johnson, March, Smeeding, and Torrey (2005) argued that by studying broader measures of consumption expenditures instead of only food consumption, researchers will have more insight into older persons' consumption patterns and well-being. Accordingly, we proposed that (a) nondriving would be negatively associated with consumption among older people, and (b) higher order needs would be more adversely related to nondriving than would life maintenance purchases.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The 2003 CAMS is the second wave of the 2001 CAMS, a supplement to the HRS, which included 5,000 individuals (one from each household) from among the households in the 2000 HRS (13,214 households). In the 2003 CAMS, 3,254 individuals completed interviews. The 2003 CAMS provides data on household consumption patterns, such as expenditures for foods, clothing, and trips.
We combined the 2003 CAMS data (including expenditures) with the 1998–2002 HRS data (with driving status), which allowed us to investigate the association between nondriving and consumption expenses among older people. In the 1998–2002 HRS data, only people older than 65 were asked about their driving status. Thus, among the 3,254 respondents, we used 1,287 people older than age 65.
Table 1 shows a description of respondents and their household characteristics. The average age was 76.9 years old. A majority of respondents were female (68.8%), and 38.2% were married. More than one quarter of the respondents rated their health as fair or poor. On average, respondents indicated that they had about two chronic illnesses (2.1 out of 8) and 0.3 (out of 6) activity of daily living (ADL) limitations. Of 1,287 respondents, 84.8% (1,102) were drivers and 14.2% (185) were currently nondrivers. Among nondrivers, 3.3% (38) had never driven, 6.7% (81) had stopped driving before 1998, and 5.2% (66) had stopped driving after 1998. The average annual income was $36,000.
|
Independent variables
The independent variable was driving status, which we measured by creating a categorical variable where 1 = never drove, 2 = stopped driving before 1998, and 3 = stopped driving after 1998 (between 1998 and 2002). We used those currently driving in 2002 as the reference group. This variable allowed us to examine the transition from driving to nondriving and to take into account older adults who had never driven.
We added several control variables, shown in Table 1, that previous researchers had identified as relevant to consumption spending. They were age, gender, race, geographic region (rural vs urban), education, household composition, change in marital status, work status, volunteer status, health status, change in living arrangements, household income, and financial assets.
We measured health status using the following three variables: (a) self-rated health, (b) chronic conditions, and (c) ADLs. Self-rated health captured people's subjective perceptions of health status, whereas ADLs and chronic conditions assessed more objective conditions (Wallace & Herzog, 1995). We measured self-rated health with fair or poor health, where 1 = if self-rated health in 2002 was fair or poor and 0 = otherwise (i.e., excellent, very good, or good self-rated health). The HRS surveys recorded eight types of chronic conditions that commonly occur in late life. They include high blood pressure; arthritis; stroke; cancer; lung disease; diabetes, heart condition; and psychiatric problems, including emotional and nervous conditions. We measured chronic conditions according to the total number of these conditions that the respondent reported having. ADLs measure functional status (i.e., the ability to perform various activities, specifically walking, dressing, bathing, eating, bedding, and toileting; Spector & Fleishman, 1998). We created an ADL index based on the total number of activity limitations that the respondent reported having.
Household composition was a categorical variable that controlled for both household size and type: married couple living with one or more people, married couple only (reference group), single with one or more people, and single only. We measured change in marital status with a binary variable: 1 = became widowed between 1998 and 2002 and 0 = otherwise. Because living arrangements also influence spending, we included a binary variable that represented change in living arrangements between 2000 and 2002 (1) or no change (0). Only 5% of the respondents had changed their living arrangements. We measured work status and volunteer activities as binary variables (i.e., work status = 1 if respondent currently worked and 0 = otherwise, and volunteer activities = 1 if respondent engaged in volunteer work and 0 = no involvement in volunteer work). Financial assets referred to the amount of cash on hand or cash equivalent checking or savings accounts or certificates of deposit that were easily converted to cash. Unlike with nonfinancial assets, many older people can liquidate their financial assets and use the money for consumption.
Analysis
Many of the households had zero expenditures in certain categories, specifically tickets (66.7%), trips (44.4%), dining out (21.5%), and clothing (15.5%). We used Tobit regression, which is most commonly used under these circumstances (McDonald & Moffitt, 1980; Nieswiadomy & Rubin, 1995). The coefficients in Tobit regression must be interpreted differently from those that emerge in linear regressions. In Tobit regression, the expected value of the dependent variable is nonlinear in independent variables and thus differs at each value of the independent variables. Researchers generally use the marginal effect (the impact of change in an independent variable on change in a dependent variable), which is evaluated at the mean values of independent variables, to assess the magnitude of the effects. For a dummy or categorical variable like driving status, one calculates the marginal effect as the discrete change in the expected value of the dependent variable as the driving status moves from one state to another (e.g., change from driving to stopping driving after 1998; Long, 1997; McDonald & Moffitt, 1980). Based on the logarithm of expenditures, we computed the marginal effect using the formula
|
|
is change in log expenditure; y1 and y0 are expenditures if driving status (x) is never driving or stop driving (= 1) and currently driving (= 0), respectively; and X is the mean values of other independent variables. Thus y1 = e x y0, where e implies the level of expenditure at stopping driving or never drove compared to current driving, and then (1 – e ) indicates decreased level of consumption spending. We used 100 x (1 – e ) to explain the magnitude of the association of driving status with consumption spending.
Health and driving status concomitantly are associated with consumption (i.e., older adults in poor health presumably spend and drive less or not at all as vitality and functional capacities decline). We clarified this obfuscation by conducting a hierarchical regression in which we entered health status in step one (Model I) and driving status in step two (Model II). We used the Akaike information criteria (AIC), which examines goodness of fit, to identify the better fitting model while comparing the significance and magnitudes of the health and driving status variables. If the AIC difference between the two models exceeds 1, one can conclude that the model with the lower AIC value fits better than the model with the higher AIC value (Long, 1997).
The 2003 CAMS provided a household composite weight variable to correct for initial selection probabilities, different attrition rates, and different participation rates (University of Michigan, 2005). Researchers often use sampling weights to create population estimates to correct for sampling biases (Heeringa & Connor, 1995). However, DuMouchel and Duncan (1983) recommended using a standard F test to evaluate significant differences between weighted and unweighted estimates. If the F test is not significant, Winship and Radbill (1994) suggested using the unweighted estimates because they are more efficient; when this test is significant, they advised using the weighted variables to avoid sampling biases. Although we used the weighted variables for our descriptive analyses and when analyzing the food and clothing consumption categories, we found that the unweighted estimates worked better for trips, tickets, and dining out based on the F test. We present the findings from these analyses in the Results section.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Hurd & Rohwedder, 2004; Nieswiadomy & Rubin, 1995). Although investigators already have shown that people spend less as they age, we suggest that driving status, and, specifically, driving cessation, also is associated with opportunities to consume (and thus consumption itself, especially in areas associated with life satisfaction and lifestyle).
We found almost no significant relationships between gender and consumption, but we did observe important differences between those who had never driven (who were all women) and those who had once driven. We were unable to examine interaction effects between gender and driving status due to the multicollinearity between gender and the interaction term of never driving and gender. Nondrivers who had never driven spent less on food and dining out than nondrivers who previously had driven. Researchers rarely have differentiated among nondrivers, although only 80% of women older than age 65 were licensed drivers in 1997. The reasons for lower food consumption among these older women are unclear, but presumably they have adapted to nondriving by more frequently cooking and eating at home than those who recently stopped driving. These differences most likely will diminish as future cohorts of older women include more drivers (Federal Highway Administration, 2004; Mohammadian, 2005; Rosenbloom, 2003). Minority and marital status also were related to consumption, but in different ways. Older people from minority backgrounds spent less on food, clothing, and dining out than did older White people, although we observed no significant differences with respect to tickets or trips. However, widowed and married people had different spending patterns. Older widowed people dined out more often and bought more tickets than did older married people. These findings were consistent with those of Utz, Carr, Nesse, and Wortman (2002), which showed increased activity among widows and widowers. Utz and colleagues also observed less activity among older African Americans than among older White widows and widowers, which might have been a result of lifestyle and economic influences. The significant associations between consumption and minority and marital status remained even when we controlled for changes in respondents' health.
The significant relationship between consumption and driving status suggests the need for transportation alternatives for older nondrivers, especially so that they can participate in life-enhancing activities. With easier access to out-of-home activities, older nondrivers probably would spend more on cultural and other recreational events. Although many communities recognize the needs of nondriving seniors and offer transportation services, older people often avoid these programs because they are inflexible and expensive (U.S. Government Accountability Office, 2004). The Beverly Foundation, experts on senior transportation options, recommends that these services will appeal more to seniors if administrators address the "5As": availability, accessibility, acceptability, affordability, and adaptability. In addition to creating more transportation options, experts need to offer better training, consulting, and screening (including on-the-road assessments) in order to smooth older people's transition from driving to nondriving (Dobbs et al., 2002).
We recognize several limitations of this research. First, we focused on expenditures, not on real consumption, as did previous researchers who have linked consumption with expenditures. Consumption is an output from both market goods or services and time. Older adults who stop driving may spend less but may stabilize their consumption by increasing their home activities, as Hurd and Rohwedder (2004) observed. Second, our cross-sectional analysis limits our conclusions about causality. We need more research to examine the causal associations between driving and consumption. Third, we were unable to control for spouses' driving status because the data were incomplete, but investigators in future studies should consider how and if the driving spouses of nondrivers influence consumption patterns. Fourth, although we examined the association between driving cessation and health care spending (including prescription drugs) and found no association between them (results not shown), we did not focus on substitution effects, such as the possibility that older people decreased spending on food in order to pay for increasing medical costs. Future researchers need alternative conceptual models if they are to shed light on these and other dynamics that are related to consumption in late life.
|
Footnotes |
|---|
Received for publication August 18, 2005. Accepted for publication February 20, 2006.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. Mezuk and G. W. Rebok Social Integration and Social Support Among Older Adults Following Driving Cessation J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2008; 63(5): S298 - S303. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
