Background of the Study
Dubey, Bhasin, Gupta, and Sharma (2011) pointed out that elderly people are trapped between weakening traditional values and an absence of adequate social security. Significant changes have been taking place in the structure of aging population in developing countries. It is estimated that nearly two-thirds of the total elderly population, defined as those who are sixty years and older, are living in developing countries. The elderly population of Bangladesh, a developing country, is experiencing substantial population growth (Uddin, Islam, & Kabir, 2012). The growth rate of the elderly is higher than the growth rate of the total population. Elderly people represented 5.2 and 5.7 percent of the population in 1961 and 1991, respectively. This rate increased to 6.1 and 7.6 percent in 2001 and 2014, respectively. The current number of elderly people in Bangladesh is more than the total population of Cuba, and more than twice the populations of countries such as Lebanon, Singapore, and Denmark. It is projected that about 11 percent of the current population will become elderly by 2025 (U.S. Census Bureau, 2015). A country is considered to be an elder-populated country if its elderly population is 11 percent or more. By 2050, the total elderly will number more than 50 million, accounting for 20 percent of the population.
These changes in population characteristics have a serious effect on the overall socioeconomic development of the country. In most cases, elderly people are not able to earn their own incomes. Consequently, they must depend on others for their livelihood. With the increase in the elderly population, the demand for home care, either informal or formal, is likely to increase. This demand for home care cannot be met without an adequate supply of informal or formal care providers. In Bangladesh, informal care provisions for the elderly are traditionally made by adult children and spouses. During the last few decades, the availability of home care services has rapidly declined owing to the breakdown of the traditional joint family along with rapid modernization and urbanization (Hossain, 2013; Hossain, Akhtar, & Uddin, 2006; Tostensen, 2004). There is also an absence of a formal care system. Most elderly people lack adequate financial resources owing to either poor pension coverage or inadequate benefit payments by the formal social security systems (Gillion, Turner, & Latulippe, 2000). As a result, the majority of the elderly are dependent on their family members for home care (Bongaarts & Zimmer, 2002). Cain (1991) found that 62 and 50 percent of the elderly lived with a married son in 1976 and 1986, respectively, in the rural areas of Bangladesh.
It is hypothesized that the elderly in Bangladesh do not receive home care from their children or spouse. The question arises as to who will provide care for the elderly? What is the role of friends, relatives, and neighbors in providing home care for the elderly? This study examines the home care that is received by the elderly and identifies where there are further needs. We purposely conducted this study in Sylhet because the area is undergoing rapid urbanization, with the youth moving from the area for work opportunities abroad and leaving their parents at home.
Research Strategy, Data, and Empirical Model
This study employed the household survey method and a structured questionnaire survey technique to collect data from desired respondents sixty years of age or older. Quantitative information such as socioeconomic characteristics, child characteristics, informal and formal care received, and future expectations of home care was collected.
The Sylhet district was purposively selected, including respondents from both urban and rural areas. Sylhet Sadar was considered an urban area, and Upazilla, Balagonj, and Kanaighat, were randomly selected as rural areas. Face-to-face interviews were conducted from November to December 2015 in two villages (Muhammadpur and Shankorpur) of Balagonj and three villages (Prurbogram, Jingabari, and Bhatirpoud) of Kanaighat in the Sylhet district.
To reach the desired respondents, the researchers utilized the country's voter list, which included citizens' ages. The total sample was divided proportionally according to the urban-rural ratio. In 2011, the urban population was 28.4 percent of the total population. In 2012, according to the World Bank, the rural population was 71.11 percent of the total population. In this study, 29 percent of the sample was taken from the urban area and 71 percent from the rural area. A total of 366 respondents were interviewed, 106 from the urban area and 260 from the rural area. All urban respondents were from the Sylhet metropolitan area. Of the rural respondents, 35.2 percent (129) were from Kanaighat and 35.8 percent (131) from Balagongupazilla in the Sylhet district. Respondent age, marital status, education, previous job status, current income source, living arrangement, number of children, number of siblings, and sociability were included as socioeconomic variables. Child characteristics included age, gender, marital status, education, occupation, and residence location.
Home care providers were divided into two categories: informal home care providers and formal home care providers. Informal care providers included spouse, children, relatives, friends, and neighbors. Formal home providers included government, nongovernmental organizations (NGOs), and paid home servants. Informal care included financial support, personal, and practical care. Personal care included that required for dressing, bathing, eating, getting in or out of bed, and using the toilet (Kalwaj, Pasini, & Wu 2014). Practical care included preparation of meals; delivery of medication; housework; local travel to doctor's appointments or shopping, for example; and making phone calls or writing letters. After a pilot test, the questionnaire was finalized and a full-length survey was conducted in November and December of 2015.
Data Analysis Techniques
The level of home care received was determined by the binary Probit regression model used by Cameron and Trivedi (2005) and Kalwij and colleagues (2014). Variables used in this study are defined in Table 1. The first column of Table 1 represents the variable name, the second column is the short form of each variable, and the third column is the definition of each variable.
This study used the Probit model as specified by Kadoya and Green (2013). The Probit model constrains the estimated probabilities to be between 0 and 1 and relaxes the constraint that the effect of independent variables is constant across predicted values of the dependent variable. The dependent variable of the study is home care received ([c.sub.receive]) by the elderly. The variable is dichotomous, categorized as ([.csub.receive] = 1, otherwise = 0). The model used for the analysis is described by the following equation:
probit([c.sub.receive] = 1) = a+ [beta]1(age) + [beta]2(gen) + [beta]3(ms) + [beta]4(edu) + [beta]5(pjs) + [beta]6(soi) + [beta]7(lia) + [beta]8(soci) + [beta]9(sib) + [beta]10(hs) + [beta]11(nos) + [beta]12(hd) + [beta]13(nod) + [beta]14(aca) + [beta]15(amc) + [beta]16(aec) + [beta]17(alc) (1)
If the coefficient of the independent variable has a negative sign, it indicates that the elderly are less likely to have received home care. A positive sign indicates that the elderly are more likely to have received home care. The variable home care was identified by nine indicators:
Personal care received from children and spouse
Financial support received from children or spouse
Personal care received from children or spouse (1 = PCRC or 0 = otherwise)
Practical care received from children or spouse (1 = PrCRC or 0 = otherwise)
Financial support received from relatives (1 = FCRR or 0 = otherwise)
Personal care received from relatives (1 = PCRR or 0 = otherwise)
Practical care received from relatives (1 = PrCRR or 0 = otherwise)
Financial support received from friend and/or neighbor (1 = FCRFN or 0 = otherwise)
Practical care received from friend and/or neighbor (1 = PrCRFN or 0 = otherwise)
In all models, independent variables were the same except for the model of personal care received from children and spouse. Each of the nine indicators of home care was specified as a dependent variable, resulting in nine empirical models of home care.
Results and Discussion
Descriptive Statistics: Socioeconomic Characteristics of the Respondents
Table 2 presents distributions of the demographic and socioeconomic characteristics of the respondents. The sample respondent age varied from 60 years to 105 years. The mean age of the respondents was 69.38 years (SD [+ or -] 8.74)...