The Economics of Long-Term Care
Summary and Keywords
Long-term care (LTC) is arguably the sector of the economy that is most sensitive to population aging: its recipients are typically older than 80 years whereas most care providers are of working age. Thus, a number of ongoing societal trends interact in the determination of market outcomes in the LTC sector: trends in longevity and healthy life expectancy interact with changing family structures and norms in shaping the need for services. The supply side is additionally affected by changes in employment patterns, in particular regarding the transition into retirement, as well as by cross-regional imbalances in demographic and economic conditions. The economic literature on long-term care considers many of these issues, aims at understanding this steadily growing sector, and at guiding policy. Key economic studies on long-term care address determinants of the demand for long-term care, like disability and socio-economic status; the two most important providers: informal family caregivers and nursing homes; and the financing and funding of LTC.
Population Aging Puts Economics of Long-Term Care on the Agenda
In the decades to come, long-term care for older people (LTC) is expected to be one of the most important and challenging policy issues for developed societies worldwide. LTC has aroused interest of researchers in different fields such as medicine and nursing, sociology, political science, healthcare management, as well as economics. Although it is receiving increasing attention by the research community, the role of long-term care in health economic research is still rather small compared to its role in society and political debate.
The increasing importance of the economics of LTC is apparent already in the development of LTC expenditure. Figure 1 compares the proportion of total LTC expenditure GDP over time for selected countries: Austria, France, Germany, Italy, the Netherlands, and the United States. For all countries except the United States, a steady increase is observed. This trend is expected to continue. In the EU, the share of public LTC expenditures as percentage of GDP amounted to about 1.5 % and is expected to almost double until 2060 (OECD/European Union, 2016). A similar story applies to other developed countries.
The topic of long-term care is closely related to the phenomenon of demographic change which is characterized by an increase in life expectancy and a reduction in fertility. It is mainly population aging that is responsible for the increase in LTC expenditures:
• Higher life expectancy causes an increase in the number of older people. This probably increases the demand for long-term care; in particular as life expectancy increases faster than healthy life expectancy.
• Low birth rates lead to an increase in the old-age dependency ratio (the number of older people per working age individual) inducing a higher financial burden of health and LTC costs per capita.
• A reduced number of working-age individuals leads to another challenge: Most LTC is provided informally by family members, particularly women. These women will be needed on the labor market in the decades to come, setting the supply side of LTC under pressure.
Life expectancy has been rising significantly throughout the world. Figure 2a shows trends in life expectancy at birth between 1980 and 2015 for selected countries. All countries experienced a clear positive trend in life expectancy. There has been a long debate in the health academic literature of whether these trends coincide with a compression of morbidity (Fries, 1980) or an extension of morbidity (Gruenberg, 1977). A parallel literature in health economics studies whether health expenditures increase with an individual’s age or whether it is only proximity to death that matters (Karlsson, Mayhew, & Rickayzen, 2008). While no clear consensus has emerged on this issue, it is fair to say that age does affect LTC expenditures (see “The Demand for Long-Term Care” and Karlsson, Iversen, & Øien, 2018 for more on this topic).
The challenge of increased longevity is reinforced by another phenomenon, reduction in birth rates. The natural consequence of increasing life expectancy and low birth rates is a rising old-age dependency ratio, i.e., the number of individuals aged 65 and above divided by the working-age (15–64) population (Figure 2b). The old-age dependency ratio can be interpreted as an indicator of the burden on the working-age population due to retired and care-dependent individuals.
These trends imply that not only more and more people potentially demand LTC but also that fewer people are available for provision and financing, leading to a massive burden. Further, adverse effects on health and labor market outcomes of informal caregivers such as spouses and children are possible. Not only informal care obligations but also increased labor demand in the formal LTC sector may indirectly affect other sectors of the economy.
Most economic research on LTC for older people falls within three broad areas. The first deals with social and economic determinants of demand for LTC services. This is instrumental for projections of future costs, but also for considerations of equity in national LTC funding schemes—and for policy questions related to the optimal allocation between prevention and care services. Second, the supply of different types of care services, and their effectiveness and quality in covering needs in the population, has been an important object of study. The different financing and funding regimes for LTC and their implications for economic efficiency and equity is a third important research area.
In this article, we summarize economic research in the field of LTC. We narrowly define LTC as care for older people. Hence, we do not cover issues of care for younger people with
disabilities. We address all three main research areas introduced above and point out gaps in the research. The structure of this chapter is as follows: We provide literature reviews on the demand (see “The Demand for Long-Term Care”) and supply of LTC services (see “The Supply of Long-Term Care”) as well as on financing aspects (see “Long-Term Care Financing and Funding Regimes”). Finally, we conclude with a brief summary and try to identify gaps that require further research (see “Summary and Outlook”).
It is not the aim of this article to provide an exhaustive review of all strands of the literature within the three domains. There are quite a number of excellent surveys that fulfill this goal and we mention some of them in the section “Further Reading.” Instead, we focus on selected aspects only, and aim to put most weight on the—in our opinion—most credible studies in the field. These studies are discussed and evaluated in great depth in order to point out what makes them particularly relevant. This approach comes at the expense of making some striking omissions, such as a detailed discussion of studies in industrial organization or entire subfields of the literature such as the one on formal home care. This is not to be understood as a judgment on the quality of this research; it is merely a necessary action in order to keep the review focused. Most, though not all, of the literature discussed is empirical work using micro data.
The Demand for Long-Term Care
Modeling LTC Demand
In order to structure the following exposition, we present a theoretical model of LTC demand within a family decision-making context. Our model is a slightly adapted version of the model used by Stabile, Laporte, and Coyte (2006).
Thus, consider a household with utility defined by the function(1)
where X represents market goods and services, L leisure, and A the ability of care recipients to perform activities of daily living. The variable A is in turn a function of the care recipient’s disability status and of the utilization of care:(2)
where F represents formal care and C denotes informal care provided by family members. D is the current disability status of the care recipient. The household faces the resource constraint(3)
where PX is the unit cost of X , PF the cost of formal care; V represents non-wage income and the wage rate is W. T is the total time available for leisure, care-giving, and labor market work.
The household’s decision problem thus involves choosing (F, C, X, L) to maximize household utility, given the constraints. This leads to the Lagrangian(4)
Under the additional assumption that the function U (X, L, A (F, C | D)) is concave with respect to (X, L, F, C), an interior solution will be given by the first-order conditions(5) (6) (7) (8) (9)
These FOCs provide some useful information on the determinants of LTC utilization. For example, the relationship between formal and informal care will be governed by the relationship so that a higher opportunity cost of the caregiver, W, a lower price of formal care services, PF, and greater productivity in formal care, AF will lead to less utilization of informal care. Moreover, the model can be used to evaluate the effects of various changes in the exogenous parameters. Such an exercise would determine the response in demand for informal (C* = C (H, W, V, PF, PX)) and formal (F* = F (H, W, V, PF, PX)) care. The exact shape of these demand functions will depend on the substitutability between formal and informal care in the production of performance ability, A, and on the marginal rate of substitution between different arguments of the utility function.
Evidently, the disability status of the care recipient is a key determinant of the demand for LTC services. Starting from a position where the household’s marginal valuation of consumption and leisure is independent of the functional ability of the care recipient (UAX = 0 and UAL = 0 respectively), a deterioration in physical health will result in increased demand for formal and informal care. This is evident already from the first-order conditions, since UA will take on a larger value when the care recipient is in poorer health. If a deterioration in disability is associated with a change in the marginal utility of consumption, the relationship between health and demand for LTC will either be weaker or stronger than in the baseline case.
This simple model captures some of the most salient aspects of the problem facing households with care needs, and it is helpful for the interpretation of empirical work, which often does not make a clear distinction between demand and utilization. That said, it does have some clear limitations, since, for example, dynamic issues or the family bargaining problem are left aside.
Determinants of LTC Need
Given that the disability status of the care recipient appears to be a key determinant of demand for LTC services, it deserves some further consideration. Economic research has made rapid progress in this area in recent years, by bringing in perspectives from biology. This biological literature in turn draws on reliability theory from engineering. A seminal contribution to this biological literature was made by Gavrilov and Gavrilova (1991), who model the human body as a fixed number of individual parts, which do not age. In this model, the aging process is a progressive reduction in redundancy, which ends with the failure of the organism, i.e., death. This modeling approach is able to capture a number of characteristics of the aging process; most importantly, that it is progressive, as the failure of a component forces other blocks to work harder, which in turn increases the probability of their failure.
Following these insights from reliability theory, empirical work has studied the pace at which human beings accumulate the deficits that eventually lead to physical limitations and death. Mitnitski Mogilner, MacKnight, and Rockwood (2002) show that the following equation captures an individual’s accumulation of deficits very well:(10)
where D (t) is the number of deficits at age t, µ is a physiological parameter (“the force of aging”) and E represents the impact of nonbiological factors on the aging process. Both parameters have been estimated with great precision in a number of different populations and the functional form appears to be the same everywhere.
Dalgaard and Strulik (2014) introduce this perspective on frailty in economics. In their version of the model, individuals may slow down the accumulation of deficits by undertaking health investments. Hence, Dalgaard and Strulik (2014) model deficit accumulation as(11)
where a is an autonomous parameter, reflecting influences of the environment which are beyond the individual’s control, h (t) represents health investments at time t, A represents the productivity of these investments, and γ determines the economies of scale.
Dalgaard and Strulik (2014) incorporate this law of motion into an economic model, where individuals maximize their lifetime utility by means of decisions regarding consumption and health investments. They calibrate their model to the U.S. economy in 2000 and find a remarkably good fit on some variables. In counterfactual experiments, they vary a number of exogenous variables and study their impact on the accumulation of health deficit. Accordingly, variation in incomes and in the productivity of health investments has large effects on health deficit accumulation. Subsequent papers have developed the Dalgaard-Strulik model to take some further aspects into account. Dalgaard and Strulik (2017) incorporate a retirement decision and study the effects of education on aging. Education is found to have a large effect on health deficit accumulation and longevity. A paper by Schünemann, Strulik, and Trimborn (2017) studies the gender gap in mortality and finds that gender-specific preferences appear to be responsible for a large part of the gap.
The above exposition has shown that the key determinants of demand for LTC services may be summarized in a relatively parsimonious model. Accordingly, the key variables appear to be the individual’s needs, the household’s wealth and income, potential informal caregivers’ opportunity costs and the prices of formal care services. In addition, the degree of substitutability between formal and informal care, and the impact on frailty on the utility function, are important aspects of the problem as well. An empirical analysis of demand for LTC needs to take into account that most of these determinants are endogenous. This is arguably not necessarily the case for prices and wages, but needs, income and wealth, and the availability of informal caregivers will typically correlate with preferences and other unobserved personal characteristics in a way that may bias estimates of their effects.
In addition, as our treatment of the Dalgaard-Strulik model makes clear, the determinants of LTC needs exhibit a great degree of overlap with the determinants of demand: income, wealth, education, and labor market attachment are variables that likely have an impact on both variables. This adds an additional identification challenge: it is difficult, in some cases impossible, to separate direct effects on demand from those that are mediated through LTC needs. For example, to the extent that education is an important determinant of both outcomes, exogenous changes in education can only identify the combined direct and indirect effects. On the other hand, for variables that vary later in life, such as labor market earnings, it might be possible to identify two sources of exogenous variation, which in turn would allow a separation of direct and indirect effects.
Empirical Studies of Demand for LTC Services
Income as a Determinant of Care Demand
As mentioned above, income is a potential determinant of care needs and of demand for care, and it is likely to be an endogenous variable in both dimensions. Two American studies, Goda, Golberstein, and Grabowski (2011) and Tsai (2015), deal with this endogeneity issue by considering random and permanent shocks to income at older ages, given by the U.S. Social Security Benefits “notch.” This “notch” was introduced due to changes in indexation of social security benefits in the 1970s and generates an arbitrary, and strong, hump in postretirement income affecting individuals born around 1915.
Both studies base their analysis on a regression equation of the type(12)
where Ui represents utilization of some type of service in for individual i, SSi represents social security income, and Xi a set of exogenous controls. These controls include a number of exogenous characteristics of the individual such as race, education, and age, but exclude variables believed to be endogenous to social security income. To the extent that the income variable SSi can be seen as exogenous, β thus represents the total effect of income on utilization, including all behavioral responses to a change in incomes.
There are a number of reasons to expect εi to be correlated with SSi. For example, health and preferences for independent living are components of εi which are likely to be correlated with income. For this reason, an instrumental variable approach is used in order to address the endogeneity problem. The first stage is specified as(13)
where Ni is a dummy variable taking on the value 1 if the household head belongs to the pivotal birth cohorts.
Both studies report a strong first stage. According to Tsai (2015), belonging to the “notch cohorts” is associated with SSI gains of around $800; according to Goda et al. (2011), this is driven by the low-educated. Both studies also report significant income effects in the utilization of formal home care, with elasticities of around 2. In addition, both studies report substitution away from other service types: according to Tsai (2015), there is a reduced reliance on informal care provided by children (in particular sons) and according to Goda et al. (2011) there is a reduced reliance on nursing home care. Both results are consistent with the idea that independence is a valued good which leads older people to demand less restrictive LTC arrangements when they can afford so.
In terms of identification, these two studies belong to the most convincing research designs used to date in the analysis of LTC demand, and therefore their empirical approach deserves some further comments. In a traditional textbook treatment of two-stage least squares regression analysis (2SLS), two conditions would be required for identification: strength and validity. The need for strong instruments has been reinforced by the literature showing that 2SLS estimates are biased in small samples (Bound, Jaeger, & Baker, 1995) and is usually seen as satisfied when the F statistic of the excluded instruments is above 10 (Stock & Yogo, 2002). Validity would in this textbook case imply that Ni, the “notch” dummy, is independent of εi, the second-stage error term. By controlling for a polynomial in age, the authors mimic a regression discontinuity design, which makes this assumption appear plausible. Yet, given that distinct effects related to the First World War and the 1918 influenza pandemic have been discussed for the subsequent cohorts (Brown & Thomas, 2011), a remaining correlation between the instrument and the error term cannot be ruled out.
An additional methodological issue is the usage by Goda et al. (2011) of an IV probit model, where the dependent variable is a binary outcome reflecting an underlying latent variable (Ui) crossing a threshold. The IV probit model is based on residual inclusion, i.e., that the first-stage residual is included in the regression equation in the second step (Terza, Basu, & Rathouz, 2008). This approach has two clear virtues compared to the alternative of using a linear probability model (LPM) and 2SLS. First, the LPM is mis-specified in this case as it allows predicted values of utilization outside of the permissible range [0, 1]. Second, an IV probit approach can be used to estimate population average effects, whereas 2SLS only allows for estimation of local average treatment effects within the group of compliers.
However, these two virtues may also be turned around: first, the advantage of getting correct predictions for any value of SSi is limited in this case, where the instrument is binary: for a given realization of covariates, X , the income variable can only take on two distinct values. Second, the estimation of treatment effects for wider populations depends on the distributional assumptions underlying the probit estimator (cf. Angrist & Pischke, 2008; Lewbel, Dong, & Yang, 2012). Third, given that SSi responds mechanically to the instrument for anyone who has positive SSi, compliance is nearly perfect in this case and therefore a LATE estimated by 2SLS would be close to the average treatment effect in the population.
Studies Considering Multiple Determinants
A very different approach is taken by De Meijer, Bakx, Van Doorslaer, and Koopmanschap (2015). The aim of their paper is to attribute overall trends in LTC utilization to a multitude of determinants, using a decomposition technique and a small representative sample of older people in the Netherlands. The backdrop is a steady decline in the utilization of institutional care in the Netherlands in the early 2000s, and a parallel trend toward increasing reliance on formal home care.
Since LTC utilization is represented by categorical variables, an extension of the traditional Oaxaca-Blinder decomposition technique (cf. Elder, Goddeeris, & Haider, 2010) is required. This is challenging, given that linear dependent variable models are typically modeled using nonlinear functions which cannot be easily decomposed. The literature has suggested a number of solutions to this problem. Bauer and Sinning (2008) present a method that can decompose total differences into an explained and an unexplained part, in a number of different settings. Fairlie (1999, 2005) and Yun (2004, 2008) have proposed methods that allow for detailed decomposition for individual variables. De Meijer et al. (2015) use the approach developed by Yun (2004). This technique is a generalization of Oaxaca-Blinder decomposition that can be used whenever the impact of independent variables on the outcome variable may be formulated as for some differentiable function F. In the case of LTC utilization trends, the decomposition becomes
where X represents determinants of LTC use and β represents the impact of these determinants; t denotes the year 2008 and t − 1 denotes either 2000 or 2004, depending on the outcome variable under consideration. The first term in the second line represents the impact of changing values of the determinants, and the second term represents the impact of changing coefficients. In Oaxaca-Blinder decompositions, the first part is often referred to as “explained” differences. Standard errors are calculated using the delta method.
Using this approach, in a first step, the authors decompose the 2000–2008 decline in institutionalization rates. This analysis delivers a very clear result: the contribution of the explained part is negative—so that LTC utilization should have increased between the two years if coefficients had remained the same. This expected increase is primarily driven by the aging of the population and increased prevalence of disability. As regards coefficients, it is, in particular, coefficients associated with moderate disability that are responsible for the decline: apparently, moderate disability was much less likely to result in institutional care in 2008 compared to 2000.
Next, the authors consider two different service types—home care and institutional care—and base their decomposition on a multinomial logit model comparing the years 2004 and 2008. For both types of service, again most of the change is explained by changing coefficients. The authors conclude that LTC policies aimed at affecting these coefficients, and individual preferences among older people, may be responsible for this.
The contribution by De Meijer et al. (2015) represents a first attempt to adapt standard decomposition methods to the case of LTC. The finding that, in general, changes in endowments have little impact on LTC utilization is intriguing and suggest that changes in coefficients are possibly devoted too little interest in research on trends in LTC. Another reading of the results would be that there are relevant omitted variables that are essential for getting a clear understanding of how LTC utilization is determined. An additional possibility is that the method imposes too-strong restrictions on the relationship between the variables. In a recent paper, Schwiebert (2015) shows that the weights imposed by the Yun (2004) decomposition technique essentially ignore the nonlinear model structure. Therefore, results may become biased if the mean of the outcome is in the tails of the distribution, or if there are large differences in independent variables (Fortin, Lemieux, & Firpo, 2011).
Balia and Brau (2014) present a comprehensive framework for LTC utilization, which includes formal and informal care as dependent variables, and allows for substitutability between them in an analysis of the impact of various determinants of LTC need. This is a significant improvement on much of the previous literature, which tends to consider formal and informal care separately, and it leads naturally to a simultaneous equations model for utilization of different types of care. In addition, LTC utilization is bounded at zero, and in most populations a significant proportion of individuals consume zero care. Therefore, the authors use a two-part model for each type of service—a modeling choice which implies that utilization on the intensive and extensive margins can be governed by different forces. The extensive margin is thus modeled as a probit model:(14)
where Uij represents utilization of service type j (formal or informal care), Xi is a set of potential determinants of utilization, and Φ (·) is the cumulative distribution function of the standard normal distribution. The second component of the model is the distribution of positive utilization, which is specified as a gamma function with two parameters: α > 0 is a shape parameter and σ = exp (Xidj + Uikgj) is the scale of the distribution. The vector Xi includes a range of potential determinants, such as age, gender, income, education, and disability. The authors also include proximity to death (PtD) and thereby speak to the large literature on whether age or proximity to death are more important predictors of health expenditure.
There are at least two empirical challenges associated with this modeling approach. First, the simultaneity of different types of care utilization, which requires further restrictions to achieve identification; second, unobserved heterogeneity associated with explanatory variables. Both problems, whenever present, would introduce a bias of unknown size in all parameter estimates. The authors’ solution to the latter problem is to use a discrete latent factor model (DLFM) which approximates the heterogeneity with a discrete multivariate distribution. The error term in each equation is specified as a composite error:(15)
The latent factors, l j, represent unobserved heterogeneity and combine elements which are common across all outcomes with those that are outcome-specific. The DLFM approximates the unknown distribution of l j using K distinct classes each occurring with probability πk. The likelihood function for each class is weighted accordingly.
Concerning the second identification problem—the simultaneity in the utilization of different types of care—the authors rely on exclusion restrictions. It is assumed that the distance between the potential care recipient and their children is a valid instrument for informal care, and it is assumed that the stated reasons for not consuming formal care is a valid instrument for formal care.
The system of equations is estimated using data on European parents aged 65 years and older, using the SHARE survey (Börsch-Supan & Jürges, 2005). The results suggest that formal and informal care services are substitutes in the sense that the utilization of one of them reduces the utilization of the other. However, these effects are very small and probably not economically significant.
Amongst the determinants of demand, age, PtD and disability all are significant, and their effects interact. The result regarding PtD is in contrast with findings by de Meijer, Koopmanschap, Bago d’Uva, and van Doorslaer (2011), who find no evidence for PtD to be a determinant of informal care demand using Dutch survey as well as registry data from 2004/2005 on individuals aged 55 and above. Their results indicate that that the effect of PtD becomes insignificant after controlling for disability. On the other hand, Weaver, Stearns, Norton, and Spector (2009) show for the United States that proximity to death affects formal care use but that this effect is reduced if informal support is taken into account. For a more extensive review of this literature, cf. Karlsson, Iversen, & Øien (2018).
The study by Balia and Brau (2014) belongs to the most sophisticated so far in terms of methods, and it addresses some of the main concerns in the previous literature. The authors make a relatively large number of assumptions regarding functional forms and distributions of variables, which makes them vulnerable to criticism in several regards. We will now briefly discuss two of the most interesting aspects of these assumptions. First, the functional form of the relationship between utilization and its determinants is an important modeling choice. The authors use a two-part model, an approach which is quite well-established in the literature on healthcare utilization (cf. Duan, Manning, Morris, & Newhouse, 1983; Mihaylova, Briggs, O’Hagan, & Thompson, 2011), where it appears plausible given that different steps in the utilization process may be exposed to different decision-makers. This assumption of different effects on extensive and intensive margins is thus closely related to physician agency. It is less clear what its rationale is in the context of LTC and in particular in the case of informal care. Given that the empirical analysis shows that most determinants of utilization have similar impact on the extensive and intensive margins, the authors could possibly have restricted themselves to a more parsimonious model.
Second, the conditions required for the identification of causal effects are not as clear as in less parameterized studies. The error structure imposed by the latent factors allow error terms to be correlated across equations in a very flexible way; however, they are in essence nonparametric random effects and thus do not address unobserved heterogeneity which is correlated with independent variables. There are estimators that do allow for fixed effects within a finite mixture specification (Deb & Trivedi, 2013) and the panel structure of the SHARE data could in principle have been used to estimate such a model.
There is also work that more directly studies the relationship between formal and informal care, for example, Orsini (2010), Mommaerts (2008), and Costa-Font, Karlsson, and Øien (2016).1 Orsini (2010) uses a reform from 1997 that induced geographic variation in the Medicare Home Health Care reimbursement rate and finds that lower rates increase the fraction of older individuals who live in shared living arrangements instead of independently. This can broadly be interpreted as a switch from informal to more formal care. Mommaerts (2008) exploits variation in Medicaid policies to identify an inverse relationship between co-residence and nursing home care in the United States. Costa-Font et al. (2016) take advantage of the financial crisis in 2008/2009 to show that the economic downturn increased informal care provision in Europe with stronger effects in the North than in the South. They argue that the crisis might have led to a substitution away from formal care services. All in all, a majority of studies find a substitutional relationship between informal and formal long-term care.
Finally, a couple of studies consider how disability affects demand for LTC. Work by de Meijer, Koopmanschap, Koolman, and van Doorslaer (2009) found that disability, both if measured (i) by number of ADLs/mobility problems and (ii) as a disability index, significantly affect the demand for LTC services. Surprisingly, general health status seems hardly to affect utilization. Luppa Luck, Matschinger, König, and Riedel-Heller (2010) apply a Cox proportional hazards model to a sample of dementia-free adults aged 75 years and above and generally confirm the importance of disability, as individuals reporting disabilities are more likely to be admitted to nursing homes in the near future.
The Supply of Long-Term Care
The provision of long-term care services differs significantly from the supply of other health services due to the important role of informal providers. Unlike formal long-term care, which can either be provided at the recipient’s home or in institutions such as nursing homes, informal care is typically received at home. Providers of informal LTC are traditionally spouses or children who do not necessarily live in the same household. We discuss issues in informal care and institutional care in turn in the following two subsections.
The decision to provide informal care is probably the result of a family bargaining process on how to provide care for a person in need. This family decision concerns both the general form of long-term care (informal, formal, both) as well as who—if anyone—takes on the burden of informal care. An important strand of the literature studies this question, often with the help of game-theoretic models and/or micro data (see e.g., Barczyk & Kredler, 2018; Byrne, Goeree, Hiedemann, & Stern, 2009; Checkovich & Stern, 2002; Engers & Stern, 2002; Fontaine, Gramain, & Wittwer, 2009; Hiedemann, Sovinsky, & Stern, 2018; Hiedemann & Stern, 1999; Rainer & Siedler, 2009). A closely related topic is that of transfers inter vivos, bequests, and their relation to informal care provision. We do not go into further detail here and instead refer to Norton (2016) for an excellent summary.
In “The Demand for Long-Term Care” we mentioned studies showing that informal and formal care are substitutes. This partly reflects the preference of most societies for informal care. It is not only preferred by most care recipients (Hajek, Lehnert, Wegener, Riedel-Heller, & König, 2017), it also imposes less direct costs on LTC insurance or other funding institutions. This makes informal care the most common source of long-term care (e.g., Karlsson, Mayhew, Plumb, & Rickayzen, 2006; Grabowski, Norton, & Van Houtven, 2012). Yet, the indirect costs might be considerable. Informal care is a challenging task for the caregiver and has potential implications for their labor market participation. Taking up the burden of care tightens the caregiver’s time constraint, but whether labor supply is reduced depends on individual preferences and, ultimately, is an empirical question. Informal care is also a physically and mentally challenging task—in particular for those who do not adjust their labor force participation and take on the double burden of care provision and work—potentially exhibiting negative health effects on the caregiver. The following two subsections summarize studies on the relationship between informal care and labor market as well as health outcomes in turn.
Informal Care and Caregivers’ Labor Market Outcomes
A large economic literature studies the relationship between informal care provision and caregivers’ labor force outcomes: the extensive margin of labor supply, the intensive margin, and/or wages. A small number of authors have studied implications for retirement entry (Meng, 2012; Jacobs, Laporte, Van Houtven, & Coyte, 2014; Van Houtven, Coe, & Skira, 2013). Almost all of the literature uses survey data, mainly the HRS, SHARE, or the English Longitudinal Study of Ageing (ELSA). These are representative samples of individuals aged 50+ and include a variety of information on health, long-term care and socioeconomic status. Other frequently used surveys that cover the full population are SOEP, HILDA, BHPS, and ECHP. We are only aware of Fevang, Kverndokk, and Roed (2012) and Løken Lundberg, and Riise (2017) as studies that use (Norwegian) register data in this field of study.
Regarding results, the literature has not yet reached a clear consensus on the labor market consequences of informal care provision. Associations between care and the probability of working for pay range from small to very large—up to 30 percentage point reductions in labor force participation (see, e.g., Bolin, Lindgren, & Lundborg, 2008; Casado-Marín, García-Gómez, & López-Nicolás, 2011; Ciani, 2012; Crespo & Mira, 2014; Carmichael & Charles, 1998; Ettner 1995, 1996; Heitmueller, 2007). Results for changes in working hours, as studied by, for example, Bolin et al. (2008), Casado-Marín et al. (2011), Ettner (1996), Johnson and Sasso (2000), Van Houtven et al. (2013), and Wolf and Soldo (1994), are quite mixed as well. Wage penalties of care provision are more consistently found (Carmichael & Charles, 2003; Heitmueller & Inglis, 2007; Van Houtven et al., 2013). There seems to be more consensus on how different care intensities are related to labor market outcomes. For example, Carmichael and Charles (2003), Casado-Marín et al. (2011), Kotsadam (2012), and Lilly, Laporte, and Coyte (2010), find that high care intensities are associated with lower rates of labor force participation or hours worked. Moreover, it seems that co-residential care is more strongly associated with (negative) labor market outcomes than extra-residential care, for example, Casado-Marín et al. (2011), Heitmueller (2007). On the other hand, however, the literature does not provide a consistent picture on gender-specific differences in these relationships. See Van Houtven et al. (2013) for a more detailed survey of results, also regarding different subgroups.
The heterogeneity in results might be due to differences in institutional settings of countries studied, in the specific sample selection, or just instrumental variable estimations that lead to different local average treatment effects (Imbens & Angrist, 1994). Another reason for the large heterogeneity might be found in the research design. The early work by Ettner (1995, 1996) and Wolf and Soldo (1994) addressed the potential endogeneity of caregiving. Caregiving is a choice variable and individuals probably react differently to care obligations, partly based on unobservable characteristics. Moreover, individual care provision and labor force participation are probably determined jointly, as also laid out in the theoretical considerations in “Modeling LTC Demand.” Michaud (2010) and Carmichael, Charles, and Hulme (2010) make this explicit by estimating the (reverse) effect of labor force status on care provision. The approach by Costa-Font et al. (2016) is in a similar spirit. Thus, ordinary least squares regressions will probably not deliver the causal effect of care provision on labor market outcomes.
The most prominent suggestion for a solution is an instrumental variables approach relying on strong and valid instruments (see “The Demand for Long-Term Care”). The previous literature using instrumental variables approaches chooses among the instruments from the following list:
Parental (= care recipient’s) health status, disability status, cognitive abilities, age
Marital status/socioeconomic status of parents
One parent recently passed away
Existence/number of siblings of potential caregiver
Geographical proximity to care recipient in kilometers
Lagged work status
Some of the variables (geographical proximity, lagged work status) have been challenged as potentially invalid (e.g., Van Houtven et al., 2013), as they represent choice variables, making them unsuitable as instruments. For example, the care recipient’s distance to the nearest child might just reflect the children’s choice to either provide informal care or not. Yet, the other suggestions are not completely immune from criticism either. As one example, parental health status might have a direct effect on the caregiver’s health, thereby also affecting their labor market outcomes. Moreover, both variables are likely correlated via intergenerational transmission channels. Given that caregiver health also is a potential outcome variable (see next section), it is not clear whether accounting for it in the regression solves the problem—it is a bad control in the sense of Angrist and Pischke (2008).
Other studies (e.g., Fevang et al., 2012; Leigh, 2010) avoid instrumental variables regressions and exploit the panel dimension of the data to account for time-invariant unobserved heterogeneity. In the spirit of a literature that argues that ordinary least squares (OLS) might at least be better than instrumental variables estimations with weak and/or invalid instruments (Crown, Henk, & Vanness, 2011; Murray, 2006), it is not clear whether these studies deliver per se less credible effects than the instrumental variables estimations. Fu, Noguchi, Kawamura, Takahashi, and Tamiya (2017) use a matching approach to estimate the relationship between informal care provision and labor force participation in Japan. They find that the negative relationship was reduced after the introduction of long-term care insurance in 2000, which brought about benefits in kind for formal care. This has reduced the burden of informal care, allowing caregivers to work in the labor market in addition to providing care.
The most credible studies make use of natural experiments, that is, policy interventions such as reforms in the LTC sector. We are aware of only three published studies in this field: Geyer and Korfhage (2017), Løken et al. (2017), and Sugawara and Nakamura (2014). Geyer and Korfhage (2017) exploit the introduction of public LTC insurance in Germany in 1995 in order to learn about the relationship between informal care provision and participation in the labor market. The reform introduced cash benefits and benefits in kind for informal care, formal home care, and institutional care. While benefits in kind are directly transferred to the nursing home or the provider of formal home care, cash benefits for informal care can be forwarded to the family caregiver. The reform has ambiguous theoretical effects on the relationship between informal care provision and labor force participation. Cash benefits increase non-labor income, thereby reducing incentives to work (or increasing incentives to provide informal care). Benefits in kind open the option to finance formal care, allowing family members to reduce informal care provision and to work instead. Geyer and Korfhage (2017) expect that the positive effect on informal care provision outweighs the effect on formal care due to the general preference for informal care in Germany (Schupp & Künemund, 2004).
Geyer and Korfhage (2017) use a difference-in-difference estimation design with an indicator of labor force participation as the outcome variable. The treatment group is composed of individuals between 35 and 65 years who co-reside with a person in need of care while the control group consists of individuals in multiperson households without care-dependent individuals. Using data from the German Socio-Economic Panel (SOEP), they find that the reform reduced male labor supply through incentives for older men to leave the labor market, whereas female labor supply appears to have been unaffected. They argue that labor force participation had been low for female caregivers before the reform, leaving less room for further decreases. However, the comparatively small sample of the SOEP renders the fairly large treatment effect of a 19 percentage point reduction in male labor force participation imprecise (though statistically significant). Since actual care provision is not observed in the data (at least around the reform years), the effect should be interpreted as an intention-to-treat effect. In addition to sample size, another potential problem is the restriction of the treatment group to individuals who co-reside with a care-dependent person. While necessary due to data limitations, this certainly is a selective group and it is not clear whether the estimated effects carry over to individuals who provide care for parents outside the household (which is the majority of cases).
Sugawara and Nakamura (2014), using data from Japan, report in a similar spirit to Geyer and Korfhage (2017). Though not explicitly presented as such, their approach can be interpreted as a difference-in-differences study where the treatment group consists of women with care obligations, while the control group consists of women without care obligation. As in Geyer and Korfhage (2017), care obligations can only be observed for co-residing individuals. The estimated negative labor market effect of having a care-dependent parent is strongly diminished after the introduction of benefits in kind in 2000 in Japan (the same reform as in Fu et al., 2017, above). These benefits in kind have strengthened formal care and probably induced a shift away from informal to formal care. Thus, the results can be interpreted as a negative effect of informal care on labor force participation.2
The study of Løken et al. (2017) shares some similarities with Geyer and Korfhage (2017) and Sugawara and Nakamura (2014) in also estimating an indirect effect of care provision on labor market outcomes using a difference-in-differences-type setting. The reform exploited here expanded federal funding for formal home-based care granted to Norwegian municipalities in 1998. Increased assistance for care dependent individuals may reduce the burden on informal caregivers with potentially positive labor market effects. Yet, it might also delay nursing home entry (this was in implicit goal of the reform), thus extending the duration of informal care episodes (Løken et al., 2017).
Municipalities could apply for the federal grants, and it turned out that those with a lower pre-coverage of formal home-based care were more likely to apply. This was intended by the policymakers: reducing imbalances between municipalities in the supply of formal care was a main goal of the reform. Løken et al. (2017) use a large administrative data set for the years 1993–2005 to run the following regression:(16)
where Yit is a labor market outcome (employment, earnings, sickness absence) of individual i at time t. PreCoveragei can be seen as a measure that captures the reform intensity as municipalities with low coverage witnessed larger increases in formal care capacities. Shortt (1 in 1998–2000, otherwise 0) and Longt (1 after 2001, 0 before) are used to discriminate between short- and long-run effects. The vector Xit includes individual characteristics and municipality fixed-effects, and α4 and α5 are the parameters of interest. The sample is restricted to adult daughters with no siblings and one parent alive who is at least 80 years old. This is argued to be the group of women that is most exposed to care obligations (which are not directly observed in the data). Again, estimation results have an intention-to-treat interpretation.
“First stage” regressions show that it took some time for the policy change to reach its full potential: in the longer run but not in the short run, homecare coverage of individuals aged 80+ increased significantly. Regarding the main outcomes of interest, Løken et al. (2017) do not find long-run effects on employment probability or wages. Yet, they do find reductions in sickness absence of daughters with potential care obligations due to the reform. While obviously also a measure of health, Løken et al. (2017) interpret sickness absence primarily as a labor market outcome at the intensive margin. They argue that, in Norway, sickness absence is not necessarily used for health reasons but also as a means of flexibility for individuals with care responsibilities. In contrast to daughters, there is no evidence of effects for sons, probably because they have smaller care obligations than daughters.
The vast majority of studies evaluates short-term or contemporaneous relationships between informal care provision and labor market outcomes. It is not clear whether identified effects (even if credibly estimated and accepted as causal) call for policy action. The decision of household members to reduce labor force participation for a limited period—most informal care episodes last up to three years—to care for a dependent relative does not necessarily pose a problem. The estimates may nevertheless be of great value as they provide useful information regarding the opportunity costs of informal case.
However, caregivers are usually middle-aged and might face problems returning to the labor market after the care episode has ended. They may also miss promotions even if they stay in the labor force and end up on a different career track. Thus, a longer-term perspective might detect more severe (and more policy-relevant) problems. Skira (2015), and Schmitz and Westphal (2017) study the dynamic and long-term consequences of care provision in the United States and in Germany. Although they use fairly different methods (highly parameterized structural model vs. dynamic matching) they both find small associations which, however, persist over time.3 The findings of these studies may indeed call for policy action—for example if they either reflect labor market imperfections, or if the long-term penalties of caregiving are seen as inequitable.
Informal Care and Caregivers’ Health
As providing care to older people—especially in case of informal care where the caregiver is not only physically but also emotionally involved—can be considered a highly demanding task, recent studies have investigated the relationship between informal care supply and caregivers’ health. Coe and van Houtven (2009) use the HRS and instrumental variables estimations with similar instruments as mentioned above to find that informal care provision has negative health consequences. Using South Korean data on women aged 45 and above, Do, Norton, Stearns, and Van Houtven (2015) identify adverse effects on physical health of care-providing daughters.
Absent suitable instruments, Schmitz and Westphal (2015) apply a regression-adjusted matching approach and find negative effects of providing care on mental health but not on physical health. However, this effect fades out over time and, on average, there is no significant negative effect 7 years after the care episode. Schmitz and Stroka (2013) are among the few exceptions that have access to administrative data from a health insurance company. Moreover, they do not only estimate a health effect of the mere care provision but also of taking on the double burden of care provision and full-time work. They find that this double burden increases the use of antidepressant drugs as well as tranquilizers. Yet, they can only control for time-invariant unobserved heterogeneity, leaving room for potential endogeneity problems. Thus, this strand of research is another one that would benefit enormously from studies that exploit natural experiments to estimate causal effects rigorously.
Finally, Bobinac, van Exel, Rutten, and Brouwer (2010) mention another problem in this literature. It is difficult to separate health effects from care provision from the effects of observing the decline of a beloved person. In most of the studies, the estimated parameters might represent a mixture of both the “family effect” and the caregiving effect.
A central topic in academic and public debate on nursing homes is the quality of care. Public scandals regarding very bad quality of care are a recurring issue in many countries (see Hackmann, 2017, for the United States and Herr, Nguyen, & Schmitz, 2016 for Germany). An important question for policymakers is how to set incentives for nursing homes to provide adequate quality at a reasonable cost. In order to guide policy, scholars in this field aim at understanding the determinants of quality. The following potential determinants generated most of the research activities: degree of competition, ownership type, staffing, information, and reimbursement rates for nursing homes. Yet, it turns out to be difficult to single out these factors as they interact and at least some of them are also affected by quality.4
Forder and Allan (2014) find that increased competition in the English nursing home market resulted in lower prices and thus lower revenues which caused quality to decrease. In contrast, Zhao (2016) finds that competition in combination with better access to information on quality improves several quality indicators. Grabowski, Feng, Hirth, Rahman, and Mor (2013) find evidence that ownership also matters. Their results from U.S. individual-level data indicate that post-acute patients in nonprofit nursing homes have a lower hospitalization risk as well as greater improvements in health and disability outcomes.
Several studies evaluate how staffing affects nursing home quality. For example, Lin (2014) provides evidence that an increase in registered nurse staffing improves quality of care measured as deficiencies. Using U.S. data, Foster and Lee (2015) find staffing subsidies to have beneficial quality effects as they reduce the incidence of pressure ulcer worsening. Yet, hiring and retaining staff is difficult for most nursing homes as reimbursement rates are usually restricted, allowing for low salaries only.
Information might be the most effective channel amenable to policymakers to improve nursing home quality. Ever since Arrow (1963), economists have been aware of the problems that asymmetric information induces, especially in healthcare markets. Long-term care is an experience good and has characteristics of a credence good where, at least prior to buying the service, suppliers of care have more information on delivered quality. Several countries aimed at increasing transparency in the market by issuing public report cards. Werner, Norton, Konetzka, and Polsky (2012) find that public reports containing information on nursing home quality cause individuals to choose nursing homes with a higher rating. Ultimately, this should increase the incentives to offer better quality. Other important studies in this field are Grabowski and Town (2011), Lu (2012), Mukamel, Weimer, Spector, Ladd, and Zinn (2008), and Park and Werner (2011). The article “Information, Risk Aversion, and Healthcare Economics” discusses these issues in more detail. One potential problem in report cards is a “teaching to test” effect. Nursing homes improve quality that can easily be measured and is reported, at the cost of unreported dimensions of quality.
Finally, and in particular in the context of the United States, reimbursement rates seem to affect nursing home quality. The highly regulated nursing home market in the United States means that only a small fraction of care recipients pays the private rates set by the nursing homes while most reimbursement rates are set by Medicare and Medicaid (Hackmann, 2017). Low reimbursement rates may reduce the nursing home’s incentive to compete for patients in terms of quality. While several early studies based on data from the 1970s and 1980s find a negative relationship between reimbursement rates and quality, Grabowski (2001), using more detailed data, finds a positive relationship. This part of the literature often measures quality by nurse staffing. Feng, Grabowski, Intrator, Zinn, and Mor (2008) find that overall staffing increased in reimbursement rates but not necessarily of registered nurses.
The most comprehensive study seems to be the one by Hackmann (2017) who develops a structural model of demand and supply in the nursing home industry and estimates its parameters using data from 2,079 nursing homes in Pennsylvania for the years 2000–2002. Estimation is based on the methods developed by Berry, Levinsohn, and Pakes (1995) and Fan (2013), well-known in empirical industrial organization. As usual in empirical work, a crucial issue is exogenous variation in reimbursement rates for identification. Hackmann (2017) exploits the fact that in Pennsylvania regulated reimbursement rates of the single nursing homes are a function of previously reported costs of all homes of similar size in a certain region. He assumes that cost increases of nursing homes in more distant areas affect the reimbursement rate of a specific home but do not capture other impacts on nursing home quality.
As a result, he finds a considerable effect of reimbursement rates on quality, measured by skilled nurse staffing ratios. While nursing homes keep 45% of increases in reimbursement rates, 55% are passed on to increase skilled nurse staffing ratios and lower prices for residents with private insurance. An increase in Medicaid reimbursement rates by 10 %is estimated to increase skilled nurse staffing ratios by 9 %, or 10 minutes of care per day spent with each resident. Adding up consumer surplus, provider profits and higher Medicaid spending, Hackmann (2017) estimates a welfare gain of $31 million per year of higher reimbursement rates in Pennsylvania. The model also allows simulation of the effect of increased competition. Interestingly, this affects nursing home quality only to a very small degree.
Long-Term Care Financing and Funding Regimes
The final part of this article deals with economic research in the field of LTC financing and funding regimes. Two aspects have received great attention by the research community, namely (1) the small size of the market for LTC insurance and (2) fiscal implications of LTC funding regimes. We provide the most important insights from these research areas.
As current projections indicate a continuous increase in the expenditure on LTC in the decades to come and private expenditure accounts for a significant share of total expenditure, it appears puzzling that the LTC insurance market still is relatively small (Brown & Finkelstein, 2011). Several studies were able to identify possible explanations for this phenomenon.
On the demand side, three different factors were found to lead to low take-up: low preferences for existing insurance products, informal care availability and family relationships, and bequest motives. Ameriks, Briggs, Caplin, Shapiro, and Tonetti (2018) develop a structural life-cycle model to evaluate whether low long-term care insurance purchases are caused by a crowding-out due to public care provision, unmet demand and poor-quality insurance products, or simply a lack of desire to insure against the risk. The authors show that a relatively large proportion of the population generally demands LTC insurance but existing products do not fit the demanded quality. Further, heterogeneity in individual preferences explains a substantial part of the LTC insurance puzzle. The dynamic model by Mommaerts (2015) confirms both the importance of individual preferences and a lack of high-quality insurance products as informal care availability appears to reduce LTC insurance purchases especially for wealthy people, and insurance take-up might be increased if insurance products covered informal care.
Not only in theoretical but also in empirical studies, the family is considered as an important determinant of LTC insurance demand. Costa-Font (2010) studies the effect of family ties on LTC insurance take-up. Two different measures for family ties are considered; the first is simply the distance to the child that lives nearest, and the second is a composite index consisting of the distance to the closest child, the respondent’s opinion regarding caregiving duties, and familistic values. The probit model results indicate a negative association between family ties and long-term care insurance demand in Europe. Further, Costa-Font and Courbage (2015) study the effect of expecting family funding for LTC on the expectations of private insurance funding. They take the potential endogeneity of family funding into account by instrumenting it with residential distance to family members and self-reported family ties. The results indicate that family support crowds-out individual incentives to seek LTC insurance. A special role is accounted to the relationship between parents and adult children. However, the distance to the nearest child might be an invalid instrument for informal care provision and thus the interpretation of the results as causal effects becomes questionable.
But the family might not only affect an individual’s LTC insurance purchase via informal care supply. According to a study by Zhou-Richter, Browne, and Gründl (2010), children’s awareness of financial risks due to LTC needs can positively affect their parents’ demand for insurance. Further, experience with LTC histories of parents is suggested to influence insurance take-up as well. Coe, Skira, and Van Houtven (2015) estimate the probability of having LTC insurance as a function of parents’ and parents in-laws’ care receipt, finding that experiencing the previous generation’s nursing home use significantly increases the probability of a LTC insurance purchase. Providing care to an in-law is shown to significantly reduce the LTC insurance purchase probability among wealthier individuals. This is interpreted as a sign of strong family responsibility norms causing individuals to self-insure against LTC in order to leave a large bequest to relatives. Theoretical evidence for the importance of bequest motives is provided by Lockwood (2018) who develops a life-cycle model that suggests bequest motives to increase savings and decrease LTC insurance purchases.
Brown and Finkelstein (2007) show for the United States that there are gender differences in LTC insurance pricing but not in coverage. However, the small market size might also be due to supply-side market failures, as there is evidence of prices being higher than actuarially fair, and of quantity rationing. Braun, Kopecky, and Koreshkova (2018) and Hendren (2013) construct theoretical models that explain insurance rejections as consequences of private information. Hendren (2013) argues that individuals who are not eligible for LTC insurance purchase—such as those who had a stroke—may not only be aware of their own health characteristics that are observable by the insurer but also about unobservable factors and preferences which are direct consequences of their health conditions. Thus, this group would be regarded as so adversely selected that offering insurance contracts to them could not result in positive profits for the insurer, which explains the rejections. Braun et al. (2018) consider administrative costs as important supply-side factors additional to private information.
This article concludes with an overview of studies analyzing fiscal implications of LTC funding regimes. Karlsson, Mayhew, and Rickayzen (2007) as well as Wouterse and Smid (2017) apply scenario analyses to determine the fiscal effects of introducing new LTC funding schemes in the United Kingdom and the Netherlands, respectively. The paper by Karlsson et al. (2007) evaluates hypothetical adoptions of the Swedish, German, and Japanese LTC regimes in the United Kingdom to find that each system is associated with an increase in taxes. Different population groups in terms of gender and age would benefit differently from the alternative systems and the results are highly sensitive to the measure of benefits. Wouterse and Smid (2017) compare alternative policies to finance the growth in LTC utilization. Their results indicate that adverse economic affects via lower labor supply caused by higher premium rates are unavoidable. However, there are differences in the timing of the effects. Direct substantial increases in the premium rates lead to large short-term effects but smaller effects in later years, whereas gradual increases have small short-run and large long-term effects.
Several papers evaluate more modest changes in the national LTC funding schemes. Bergquist, Costa-i-Font, and Swartz (2015) estimate the effect of the LTC partnership program—an initiative to encourage individuals to purchase private LTC insurance in the United States—on Medicaid long-term care expenditures. Their OLS and GLS fixed effects results do not show significant effects on expenditures, which is possibly due to a problematic design in the program and a target group limited to middle-class individuals. Kopecky and Koreshkova (2014) build a life-cycle model calibrated to the United States to determine the welfare effects of extending nursing home cost coverage by social insurance programs. The authors find individuals to benefit from social insurance extensions financed via increases in income taxes. In a study on Germany, Geyer, Haan, and Korfhage (2017) look at the fiscal effects of informal care applying a structural model. According to their results, an increase in formal care financed by a reduction of informal care subsidies leads to positive effects on the government budget.
Summary and Outlook
This article has provided a broad overview of economic research in the field of long-term care. Starting with a brief introduction explaining the relationship between demographic change and long-term care, we identified the increase in life expectancy, the decrease in birth rates, and the increase in healthy life expectancy below the increase of total life expectancy as the main factors making long-term care one of the most important topics in developed countries in the 21st century. A detailed literature review summarized economic research on the three main areas of LTC demand, LTC supply, and LTC financing and funding. Further, as economic research on LTC is rapidly growing but still rather small compared to its potential importance, there are several remaining gaps in the literature which we would like to spell out.
In terms of substance: (1) Analyses of some crucial aspects of LTC are inadequate or missing. For instance, not much has been published about nursing home residents in general, probably due to lack of data on this particular group. There is also little work on assisted living and continuing care facilities. Furthermore, there is much work to be done in industrial organization regarding the long-term care market (e.g., entry/exit of nursing homes and home healthcare providers). Another example is that the literature largely neglects regional economic issues for demographic aging and LTC. Moreover, there is little (theoretical) understanding of the behavioral mechanisms behind the emergence of LTC needs and means over the individual’s life-cycle or, indeed, how they are linked to different modes of LTC provision and funding as well as the resulting outcomes. (2) Many of the studied research questions are far from being answered; the puzzle of low demand for LTCI being one example. Also, little is known about inequality in the needs and means of LTC and how it could be addressed. (3) Even now, most of the empirical and calibrated work on LTC based on studies in and of the United States.
In terms of methodology and data: (4) The number of studies exploiting natural experiments to achieve results of highest credibility could be increased. (5) There is also more work to be done using administrative data covering individuals and/or nursing homes. (6) While there is a growing body of research that incorporates dynamics (see Sovinsky & Stern, 2016, for a review), the inclusion of dynamics in models and empirical analyses of long-term care is still in its infancy.
Other gaps in the literature might be attributable to a lack of concerted research efforts, leading to a highly disparate field of study. This is expressed in (7) strict separation between empirical reduced-form and structural models with few attempts to exploit the best of both worlds, (8) few overarching models and frameworks for the analysis of LTC.
The field of economics of long-term care is still developing. Future studies will address these gaps.
Angrist, J. D., & Pischke, J.-S. (2010). The credibility revolution in empirical economics: How better research design is taking the con out of econometrics. The Journal of Economic Perspectives, 24(2), 3–30.Find this resource:
Bom, J. A. M., Bakx, P. L. H., Schut, F. T., & van Doorslaer, E. K. A. (2018). The impact of informal caregiving for older adults on the health of various types of caregivers: A systematic review. The Gerontologist, November 3.Find this resource:
Bauer, J. M., & Sousa-Poza, A. (2015). Impacts of informal caregiving on caregiver employment, health, and family. Journal of Population Ageing, 8(3), 113–145Find this resource:
Grabowski, D. C., & Norton, E. C. (2012). Nursing home quality of care. In A. Jones (Ed.), The Elgar Companion to Health Economics, 2nd ed. (Chapter 29). Cheltenham, U.K.: Edward Elgar.Find this resource:
Grabowski, D. C., Norton, E. C., & Van Houtven, C. H. (2012). Informal care. In A. Jones (Ed.), The Elgar Companion to Health Economics, 2nd ed. (Chapter 30). Cheltenham, U.K.: Edward Elgar.Find this resource:
Norton, E. C. (2000). Long-term care. In J. Culyer, & J. P. Newhouse (Eds.), Handbook of Health Economics. Vol. 1B. Amsterdam, The Netherlands: North Holland.Find this resource:
Norton, E. C. (2016). Health and long-term care. In J. Piggott, & A. Woodland (Eds.), Handbook of the Economics of Population Aging (pp. 951–989). Amsterdam, The Netherlands: Elsevier.Find this resource:
Sovinsky, M., & Stern, S. (2016). Dynamic modelling of long-term care decisions. Review of Economics of the Household, 14(2), 463–488.Find this resource:
Ameriks, J., Briggs, J., Caplin, A., Shapiro, M. D., & Tonetti, C. (2018). The Long-Term Care Insurance Puzzle: Modeling and Measurement. NBER Working Paper Series, No. 22726. Cambridge, MA: National Bureau of Economic Research.Find this resource:
Angrist, J. D., & Pischke, J.-S. (2008). Mostly harmless econometrics: An empiricist’s companion. Princeton. NJ: Princeton University Press.Find this resource:
Arrow, K. J. (1963). Uncertainty and the welfare economics of medical care. American Economic Review, 53(5), 941–973.Find this resource:
Balia, S., & Brau, R. (2014). A country for old men? Long-term home care utilization in Europe. Health Economics, 23(10), 1185–1212.Find this resource:
Barczyk, D., & Kredler, M. (2018). Evaluating long-term-care policy options, taking the family seriously. Review of Economic Studies, 85(2), 766–809.Find this resource:
Bauer, T. K., & Sinning, M. (2008). An extension of the Blinder–Oaxaca decomposition to nonlinear models. AStA: Advances in Statistical Analysis, 92(2), 197–206.Find this resource:
Bergquist, S., Costa-i-Font, J., & Swartz, K. (2015). Long term care partnerships: Are they “Fit for Purpose”? CESifo Working Paper 5155. Munich, Germany: CESifo Group.Find this resource:
Berry, S., Levinsohn, J., & Pakes, A. (1995). Automobile prices in market equilibrium. Econometrica, 63(4), 841–890.Find this resource:
Bobinac, A., van Exel, N. J. A., Rutten, F. F., & Brouwer, W. B. (2010). Caring for and caring about: Disentangling the caregiver effect and the family effect. Journal of Health Economics, 29(4), 549–556.Find this resource:
Bolin, K., Lindgren, B., & Lundborg, P. (2008). Your next of kin or your own career? Caring and working among the 50+ of Europe. Journal of Health Economics, 27(3), 718–738.Find this resource:
Börsch-Supan, A., Hank, K., & Jürges, H. (2005). A new comprehensive and international view on ageing: Introducing the “Survey of health, aging, and retirement in Europe.” European Journal of Ageing, 2(4), 245–253.Find this resource:
Bound, J., Jaeger, D. A., & Baker, R. M. (1995). Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variable is weak. Journal of the American Statistical Association, 90(430), 443–450.Find this resource:
Braun, R. A., Kopecky, K. A., & Koreshkova, T. (2018). Old, frail, and uninsured: accounting for features of the U.S. long-term care insurance market. Federal Reserve Bank of Atlanta Working Paper Series. Atlanta, GA: Federal Reserve Bank of Atlanta.Find this resource:
Brown, J. R., & Finkelstein, A. (2007). Why is the market for long-term care insurance so small? Journal of Public Economics, 91, 1967–1991.Find this resource:
Brown, J. R., & Finkelstein, A. (2011). Insuring long-term care in the United States. Journal of Economic Perspectives, 25(4), 119–142.Find this resource:
Brown, R., & Thomas, D. (2011). On the long term effects of the 1918 US influenza pandemic.Find this resource:
Byrne, D., Goeree, M. S., Hiedemann, B., & Stern, S. (2009). Formal home health care, informal care, and family decision making. International Economic Review, 50(4), 1205–1242.Find this resource:
Carmichael, F., & Charles, S. (1998). The labour market costs of community care. Journal of Health Economics, 17(6), 747–765.Find this resource:
Carmichael, F., & Charles, S. (2003). The opportunity costs of informal care: Does gender matter? Journal of Health Economics, 22(5), 781–803.Find this resource:
Carmichael, F., Charles, S., & Hulme, C. (2010). Who will care? Employment participation and willingness to supply informal care. Journal of Health Economics, 29, 182–190.Find this resource:
Casado-Marín, D., García-Gómez, P., & López-Nicolás, A. (2011). Informal care and labour force participation among middle-aged women in Spain. SERIEs, 2(1), 1–29.Find this resource:
Checkovich, T. J., & Stern, S. (2002). Shared caregiving responsibilities of adult siblings with elderly parents. Journal of Human Resources, 37(3), 441–478.Find this resource:
Ciani, E. (2012). Informal adult care and caregivers’ employment in Europe. Labour Economics, 19(2), 155–164.Find this resource:
Coe, N. B., Skira, M. M., & Van Houtven, C. H. (2015). Long-term care insurance: Does experience matter? Journal of Health Economics, 40, 122–131.Find this resource:
Coe, N. B., & van Houtven, C. H. (2009). Caring for mom and neglecting yourself? The health effects of caring for an elderly parent. Health Economics, 18(9), 991–1010.Find this resource:
Costa-Font, J. (2010). Family ties and the crowding out of long-term care insurance. Oxford Review of Economic Policy, 4(1), 691–712.Find this resource:
Costa-Font, J., & Courbage, C. (2015). Crowding out of long-term care insurance: Evidence from European expectations data. Health Economics, 24(Suppl. 1), 74–88.Find this resource:
Costa-Font, J., Karlsson, M., & Øien, H. (2016). Careful in the crisis? Determinants of older people’s informal care receipt in crisis-struck European countries. Health Economics, 25(S2), 25–42.Find this resource:
Crespo, L., & Mira, P. (2014). Caregiving to elderly parents and employment status of European mature women. The Review of Economics and Statistics, 96(4), 693–709.Find this resource:
Crown, W. H., Henk, H. J., & Vanness, D. J. (2011). Some cautions on the use of instrumental variables estimators in outcomes research: How bias in instrumental variables estimators is affected by instrument strength, instrument contamination, and sample size. Value in Health, 14(8), 1078–1084.Find this resource:
Dalgaard, C.-J., & Strulik, H. (2014). Optimal aging and death: understanding the Preston curve. Journal of the European Economic Association, 12(3), 672–701.Find this resource:
Dalgaard, C.-J., & Strulik, H. (2017). The genesis of the golden age: Accounting for the rise in health and leisure. Review of Economic Dynamics, 24, 132–151.Find this resource:
De Meijer, C., Bakx, P., Van Doorslaer, E., & Koopmanschap, M. (2015). Explaining declining rates of institutional LTC use in the Netherlands: A decomposition approach. Health Economics, 24, 18–31.Find this resource:
de Meijer, C., Koopmanschap, M., Bago d’Uva, T., & van Doorslaer, E. (2011). Determinants of long-term care spending: Age, time to death or disability? Journal of Health Economics, 30, 425–438.Find this resource:
de Meijer, C. A. M., Koopmanschap, M. A., Koolman, X. H. E., & van Doorslaer, E. K. A. (2009). The role of disability in explaining long-term care utilization. Medical Care, 47, 1156–1163.Find this resource:
Deb, P., & Trivedi, P. K. (2013). Finite mixture for panels with fixed effects. Journal of Econometric Methods, 2(1), 35–51.Find this resource:
Do, Y. K., Norton, E. C., Stearns, S. C., & Van Houtven, C. H. (2015). Informal care and caregiver’s health. Health Economics, 24, 224–237.Find this resource:
Duan, N., Manning, W. G., Morris, C. N., & Newhouse, J. P. (1983). A comparison of alternative models for the demand for medical care. Journal of Business & Economic Statistics, 1(2), 115–126.Find this resource:
Elder, T. E., Goddeeris, J. H., & Haider, S. J. (2010). Unexplained gaps and Oaxaca–Blinder decompositions. Labour Economics, 17(1), 284–290.Find this resource:
Engers, M., & Stern, S. (2002). Long-term care and family bargaining. International Economic Review, 43(1), 73–114.Find this resource:
Ettner, S. L. (1995). The impact of “parent care” on female labor supply decisions. Demography, 32(1), 63–80.Find this resource:
Ettner, S. L. (1996). The opportunity costs of elder care. The Journal of Human Resources, 31(1), 189–205.Find this resource:
Fairlie, R. W. (1999). The absence of the African-American owned business: An analysis of the dynamics of self-employment. Journal of Labor Economics, 17(1), 80–108.Find this resource:
Fairlie, R. W. (2005). An extension of the Blinder-Oaxaca decomposition technique to logit and probit models. Journal of Economic and Social Measurement, 30(4), 305–316.Find this resource:
Fan, Y. (2013). Ownership consolidation and product characteristics: A study of the US daily newspaper market. The American Economic Review, 103(5), 1598–1628.Find this resource:
Feng, Z., Grabowski, D. C., Intrator, O., Zinn, J., & Mor, V. (2008). Medicaid payment rates, case-mix reimbursement, and nursing home staffing: 1996–2004. Medical Care, 46(1), 33–40.Find this resource:
Fevang, E., Kverndokk, S., & Roed, K. (2012). Labor supply in the terminal stages of lone parents’ lives. Journal of Population Economics, 25(4), 1399–1422.Find this resource:
Fontaine, R., Gramain, A., & Wittwer, J. (2009). Providing care for an elderly parent: Interactions among siblings? Health Economics, 18(9), 1011–1029.Find this resource:
Forder, J., & Allan, S. (2014). The impact of competition on quality and prices in the English care homes market. Journal of Health Economics, 34, 73–83.Find this resource:
Fortin, N., Lemieux, T., & Firpo, S. (2011). Decomposition methods in economics. In Orley Ashenfelter, & David Card (Eds.), Handbook of labor economics, Vol. 4A (pp. 1–102). Amsterdam, Netherlands: Elsevier.Find this resource:
Foster, A. D., & Lee, Y. S. (2015). Staffing subsidies and the quality of care in nursing homes. Journal of Health Economics, 41, 133–147.Find this resource:
Fries, J. F. (1980). Aging, natural death, and the compression of morbidity. New England Journal of Medicine, 303(3), 130–135.Find this resource:
Fu, R., Noguchi, H., Kawamura, A., Takahashi, H., & Tamiya, N. (2017). Spillover effect of Japanese long-term care insurance as an employment promotion policy for family caregivers. Journal of Health Economics, 56, 103–112.Find this resource:
Gavrilov, L. A., & Gavrilova, N. S. (1991). The biology of life span: A quantitative approach. Chur, Switzerland: Harwood Academic Publishers.Find this resource:
Geyer, J., Haan, P., & Korfhage, T. (2017). Indirect fiscal effects of long-term care insurance. Fiscal Studies, 38(3), 393–415.Find this resource:
Geyer, J., & Korfhage, T. (2017). Labor supply effects of long-term care reform in Germany. Health Economics, 27, 1328–1339.Find this resource:
Goda, G. S., Golberstein, E., & Grabowski, D. C. (2011). Income and the utilization of long-term care services: Evidence from the Social Security benefit notch. Journal of Health Economics, 30(4), 719–729.Find this resource:
Grabowski, D. C. (2001). Medicaid reimbursement and the quality of nursing home care. Journal of Health Economics, 20(4), 549–569.Find this resource:
Grabowski, D. C., Feng, Z., Hirth, R., Rahman, M., & Mor, V. (2013). Effect of nursing home ownership on the quality of post-acute care: An instrumental variable approach. Journal of Health Economics, 32, 12–21.Find this resource:
Grabowski, D. C., Norton, E. C., & van Houtven, C. H. (2012). Informal care. In A. M. Jones (Ed.), The Elgar companion to health economics, 2nd ed. (pp. 318–328). Cheltenham, U.K.: Edward Elgar Publishing.Find this resource:
Grabowski, D. C., & Town, R. J. (2011). Does information matter? Competition, quality, and the impact of nursing home report cards. Health Services Research, 46(6pt1), 1698–1719.Find this resource:
Gruenberg, E. M. (2005 ). The failures of success. Milbank Quarterly, 83(4), 779–800.Find this resource:
Hackmann, M. (2017). Incentivizing better quality of care: The role of Medicaid and competition in the nursing home industry. NBER Working Paper 24133. Cambridge, MA: National Bureau of Economic Research.Find this resource:
Hajek, A., Lehnert, T., Wegener, A., Riedel-Heller, S., & König, H.- H. (2017). Factors associated with preferences for long-term care settings in old age: Evidence from a population-based survey in Germany. BMC Health Services Research, 17, 156.Find this resource:
Heitmueller, A. (2007). The chicken or the egg? Endogeneity in labour market participation of informal carers in England. Journal of Health Economics, 26(3), 536–559.Find this resource:
Heitmueller, A., & Inglis, K. (2007). The earnings of informal carers: Wage differentials and opportunity costs. Journal of Health Economics, 26(4), 821–841.Find this resource:
Hendren, N. (2013). Private information and insurance rejections. Econometrica, 81(5), 1713–1762.Find this resource:
Herr, A., Nguyen, T.-V., & Schmitz, H. (2016). Public reporting and the quality of care of German nursing homes. Health Policy, 120(10), 1162–1170.Find this resource:
Hiedemann, B., Sovinsky, M S., & Stern, S. (2018). Will you still want me tomorrow? The dynamics of families’ long-term care arrangements. Journal of Human Resources, 53(3), 663–716.Find this resource:
Hiedemann, B., & Stern, S. (1999). Strategic play among family members when making long-term care decisions. Journal of Economic Behavior & Organization, 40(1), 29–57.Find this resource:
Imbens, G. W., & Angrist, J. D. (1994). Identification and estimation of local average treatment effects. Econometrica, 62(2), 467–75.Find this resource:
Jacobs, J. C., Laporte, A., Van Houtven, C. H., & Coyte, P. C. (2014). Caregiving intensity and retirement status in Canada. Social Science & Medicine, 102, 74–82.Find this resource:
Johnson, R. W., & Sasso, A. T. L. (2000). The trade-off between hours of paid employment and time assistance to elderly parents at midlife. Urban Institute Working paper. Washington, DC: The Urban Institute.Find this resource:
Karlsson, M., Iversen, T., & Øien, H. (2018). Aging and healthcare costs. Oxford Research Encyclopedia of Economics and Finance.Find this resource:
Karlsson, M., Mayhew, L., Plumb, R., & Rickayzen, B. (2006). Future costs for long-term care: Cost projections for long-term care for older people in the United Kingdom. Health Policy, 75(2), 187–213.Find this resource:
Karlsson, M., Mayhew, L., & Rickayzen, B. (2007). Long term care financing in four OECD countries: Fiscal burden and distributive effects. Health Policy, 80, 107–134.Find this resource:
Karlsson, M., Mayhew, L., & Rickayzen, B. (2008). Individualised life tables. Journal of Population Ageing, 1(2–4), 153–191.Find this resource:
Kopecky, K. A., & Koreshkova, T. (2014). The impact of medical and nursing home expenses on savings. American Economic Journal: Macroeconomics, 6(3), 29–72.Find this resource:
Kotsadam, A. (2012). The employment costs of caregiving in Norway. International Journal of Health Economics and Management, 12(4), 269–283.Find this resource:
Leigh, A. (2010). Informal care and labor market participation. Labour Economics, 17(1), 140–149.Find this resource:
Lewbel, A., Dong, Y., & Yang, T. T. (2012). Comparing features of convenient estimators for binary choice models with endogenous regressors. Canadian Journal of Economics/ Revue canadienne d’économique, 45(3), 809–829.Find this resource:
Lilly, M. B., Laporte, A., & Coyte, P. C. (2010). Do they care too much to work? The influence of caregiving intensity on the labor force participation of unpaid caregivers in Canada. Journal of Health Economics, 29, 895–903.Find this resource:
Lin, H. (2014). Revisiting the relationship between nurse staffing and quality of care in nursing homes: An instrumental variables approach. Journal of Health Economics, 37, 13–24.Find this resource:
Lockwood, L. M. (2018). Incidental bequests and the choice to self-insure late-life risks. American Economic Review, 108(9), 2513–2550.Find this resource:
Løken, K. V., Lundberg, S., & Riise, J. (2017). Lifting the burden formal care of the elderly and labor supply of adult children. Journal of Human Resources, 52(1), 247–271.Find this resource:
Lu, S. F. (2012). Multitasking, information disclosure, and product quality: Evidence from nursing homes. Journal of Economics & Management Strategy, 21(3), 673–705.Find this resource:
Luppa, M., Luck, T., Matschinger, H., König, H.-H., & Riedel-Heller, S. (2010). Predictors of nursing home admission of individuals without a dementia diagnosis before admission: Results from the Leipzig Longitudinal Study of the Aged (LEILA 75+). BMC Health Services Research, 10(186).Find this resource:
Meng, A. (2012). Informal caregiving and the retirement decision. German Economic Review, 13(3), 307–330.Find this resource:
Michaud, P.-C. (2010). A dynamic analysis of informal care and employment in England. Labour Economics, 17, 455–465.Find this resource:
Mihaylova, B., Briggs, A., O’Hagan, A., & Thompson, S. G. (2011). Review of statistical methods for analysing healthcare resources and costs. Health Economics, 20(8), 897–916.Find this resource:
Mitnitski, A. B., Mogilner, A. J., MacKnight, C., & Rockwood, K. (2002). The accumulation of deficits with age and possible invariants of aging. The Scientific World Journal, 2, 816–1822.Find this resource:
Mommaerts, C. (2008). Are coresidence and nursing homes substitutes? Evidence from Medicaid spend-down provisions. Journal of Health Economics, 59, 125–138.Find this resource:
Mommaerts, C. (2015). Long-term care insurance and the family. 2017 ASSA Annual Meeting, Chicago, IL, January 6–8, preliminary paper.Find this resource:
Mukamel, D., Weimer, D., Spector, W., Ladd, H., & Zinn, J. (2008). Publication of quality report cards and trends in reported quality measures in nursing homes. Health Services Research, 14(4), 1244–1262.Find this resource:
Murray, M. P. (2006). Avoiding invalid instruments and coping with weak instruments. Journal of Economic Perspectives, 20(4), 111–132.Find this resource:
Norton, E. (2016). Health and Long-Term Care. In John Piggott, & Alan Woodland (Eds.), Handbook of the economics of population aging (pp. 951–989). Amsterdam, Netherlands: Elsevier.Find this resource:
OECD/European Union (2016). Projections of public expenditure on health and long-term care. In Health at a Glance: Europe 2016: State of Health in the EU Cycle (pp. 190–191). Paris, France: OECD Publishing.Find this resource:
Orsini, C. (2010). Changing the way the elderly live: Evidence from the home health care market in the United States. Journal of Public Economics, 94, 142–152.Find this resource:
Park, J., & Werner, R. M. (2011). Changes in the relationship between nursing home financial performance and quality of care under public reporting. Health Economics, 20(7), 783–801.Find this resource:
Rainer, H., & Siedler, T. (2009). O Brother, Where Art Thou? The effects of having a sibling on geographic mobility and labour market outcomes. Economica, 76(303), 528–556.Find this resource:
Schmitz, H., & Stroka, M. A. (2013). Health and the double burden of full-time work and informal care provision evidence from administrative data. Labour Economics, 24, 305–322.Find this resource:
Schmitz, H., & Westphal, M. (2015). Short and medium-term effects of informal care provision on female caregivers’ health. Journal of Health Economics, 42, 174–185.Find this resource:
Schmitz, H., & Westphal, M. (2017). Informal care and long-term labor market outcomes. Journal of Health Economics, 56(C), 1–18.Find this resource:
Schünemann, J., Strulik, H., & Trimborn, T. (2017). The gender gap in mortality: How much is explained by behavior? Journal of Health Economics, 54, 79–90.Find this resource:
Schupp, J., & Künemund, H. (2004). Private Versorgung und Betreuung von Pflegebedurftigen in Deutschland. DIW Wochenbericht, 71(20), 289–294.Find this resource:
Schwiebert, J. (2015). A detailed decomposition for nonlinear econometric models. Journal of Economic Inequality, 13(1), 53–67.Find this resource:
Skira, M. M. (2015). Dynamic wage and employment effects of elder parent care. International Economic Review, 56(1), 63–93.Find this resource:
Sovinsky, M., & Stern, S. (2016). Dynamic modelling of long-term care decisions. Review of Economics of the Household, 14(2), 463–488.Find this resource:
Stabile, M., Laporte, A., & Coyte, P. C. (2006). Household responses to public home care programs. Journal of Health Economics, 25(4), 674–701.Find this resource:
Stock, J. H., & Yogo, M. (2002). Testing for weak instruments in linear IV regression. In D. W. K. Andrews, & J. H. Stock (Eds.), Identification and Inference for econometric models: Essays in honor of Thomas Rothenberg (pp. 80–108). Cambridge, U.K.: Cambridge University Press.Find this resource:
Sugawara, S., & Nakamura, J. (2014). Can formal elderly care stimulate female labor supply? The Japanese experience. Journal of the Japanese and International Economies, 34, 98–115.Find this resource:
Terza, J. V., Basu, A., & Rathouz, P. J. (2008). Two-stage residual inclusion estimation: Addressing endogeneity in health econometric modelling. Journal of Health Economics, 27(3), 531–543.Find this resource:
Tsai, Y. (2015). Social security income and the utilization of home care: Evidence from the social security notch. Journal of Health Economics, 43, 45–55.Find this resource:
Van Houtven, C. H., Coe, N. B., & Skira, M. M. (2013). The effect of informal care on work and wages. Journal of Health Economics, 32, 240–252.Find this resource:
Weaver, F., Stearns, S. C., Norton, E. C., & Spector, W. (2009). Proximity to death and participation in the long-term care market. Health Economics, 18, 867–883.Find this resource:
Werner, R. M., Norton, E. C., Konetzka, R. T., & Polsky, D. (2012). Do consumers respond to publicly reported quality information? Evidence from nursing homes. Journal of Health Economics, 31, 50–61.Find this resource:
Wolf, D. A., & Soldo, B. J. (1994). Married women’s allocation of time to employment and care of elderly parents. Journal of Human Resources, 29(4), 1259–1276.Find this resource:
Wouterse, B., & Smid, B. (2017). How to finance the rising costs of long-term care: Four alternatives for the Netherlands. Fiscal Studies, 38(3), 369–391.Find this resource:
Yun, M.-S. (2004). Decomposing differences in the first moment. Economics Letters, 82(2), 275–280.Find this resource:
Yun, M.-S. (2008). Identification problem and detailed Oaxaca decomposition: A general solution and inference. Journal of Economic and Social Measurement, 33(1), 27–38.Find this resource:
Zhao, X. (2016). Competition, information, and quality: Evidence from nursing homes. Journal of Health Economics, 49, 136–152.Find this resource:
Zhou-Richter, T., Browne, M. J., & Gründl, H. (2010). Don’t they care? Or, are they just unaware? Risk perception and the demand for long-term care insurance. Journal of Risk and Insurance, 77(4), 715–747.Find this resource:
(1.) Many more papers analyze this matter. However, we would like to focus on those that use the most convincing exogenous variation in this part of the text.
(4.) How to adequately measure quality of care is another field of research that is outside the scope of this article.