This paper specifically addresses and contributes to the literature on how housing wealth could play a key role in funding and cost-sharing health care for older people. How to best fund health care in the context of the aging population is a key economic and social policy question in most developed countries, including the United States (U.S.). As people get older, their utilization of health care services increases (U.S. Institute of Medicine Committee, 2008). In 2015, older Americans represented the top 5% of health care users (Zayas et al., 2016), accounting for 61% of doctor visits (Rui and Kang, 2015), 35% of hospital stays (Freeman et al., 2018) and 85% of drug prescription (Martin et al., 2019). Although two-thirds of their spending is paid by the government, through various public insurance programs (De Nardi et al., 2015), older Americans are still expected to co-pay for long hospital stays (> 60 days), doctor visits, prescription drugs, and other outpatient services, through out-of-pocket payments or private health insurance (U.S. Centre for Medicare and Medicaid services, 2021). In 2020, older Americans spent up to a fifth of their income on health care services (Carman et al., 2020). Due to its cost, older Americans are more likely to postpone or forgo health care than older adults in other high-income countries (Jacobson et al., 2021).

Housing assets, as a form of wealth, thus play an important role as an additional source of funding for many older adults living on a fixed income and serve as a “self-insurance” mechanism for health care utilization (Joint Center for Housing studies, 2018). On the other hand, liquidity constraints may prevent use of housing wealth to fund health care. Despite its important policy implication, to date, there is limited causal evidence of a wealth effect, especially housing wealth, on older adults’ health care utilization, and existing studies still exhibit four main limitations.

Firstly, most studies to date have focused on income rather than on wealth, especially housing wealth (Braveman et al., 2005). As people get older, income loses its significance, whereas wealth becomes more important as a source of funding (Boyle Torrey and Teauber, 1986, Feinstein, 1993, Alessie et al., 1997, Van Ourti, 2003). Housing equity is the most important asset for a large fraction of the older population (Doling and Elsinga, 2012), and represents a primary source of collateral borrowing (Bhutta and Keys, 2016). This is especially the case for older Americans. In 2018, 78.7 percent of Americans above 65 years old were homeowners and continued to be the owner into older age (Joint Center for Housing studies, 2018). About 80 percent of older Americans’ wealth was held in the form of housing equity (Venti and Wise, 1991). The “wealth effect” hypothesis predicts that homeowners may feel wealthier during a housing boom, thus, are more willing to spend their assets (Doling and Horsewood, 2011). Increasing housing wealth may also make borrowing easier, thus relaxing homeowners’ financial constraints (Campbell and Cocco, 2007, Iacoviello, 2011). Such wealth gains allow older households to consume health care services that could otherwise have been forgone. Therefore, changes in housing wealth may have a large impact on both welfare and consumption in older age (Case et al., 2001, Campbell and Cocco, 2007).

Secondly, most investigations of the wealth-health care nexus are limited to expenditure instead of utilization. While Acemoglu et al. (2013) find a small increase in health expenditure in response to a positive income shock (0.7%), Tsai (2018) estimate a rise between 1.4% and 3.4% in health expenditures of older Americans. Although health expenditure is a good proxy for consumption, an increase in expenditure does not always imply increased utilization. Since expenditure captures both quality and utilization, higher expenditure may reflect better quality services, but not necessarily higher utilization. Indeed, a comparative study reveals that the U.S. population does not use health care more than other countries, but users spend much more than other comparable OECD countries (Papanicolas et al., 2018). Up to now, most studies investigating the effects of wealth on health care utilization are descriptive in nature (Cooper et al., 2012, Rodrigues et al., 2018). An increase in wealth allows a person to afford more health services, but it may also improve health by providing a better lifestyle or by accessing higher quality services (Schwandt, 2018, Pool et al., 2018, Fichera and Gathergood, 2016), thereby reducing the need to seek further health care.

Thirdly, many studies, which attempted to establish such causal effects, have used exogenous variation in economic resources such as oil price shocks, changes in public policy, and housing prices as sources of identification. For example, using changes in Social Security notch, Goda et al. (2011) and Moran and Simon (2006) find that a positive permanent income shock increases utilization of paid home care and prescription drugs. Acemoglu et al. (2013) utilize regional variation in oil prices to investigate the relationship between a permanent income shock and hospitalization. These identification strategies constrain the population of interest. While the first two papers use the same strategy comparing cohorts of pensioners exposed to changes in Social Security notch to pensioners who were not, the third focuses on a modest number of economic subregions in Texas and Los Angeles. Hence, it is unclear whether the empirical findings apply more generally.

Lastly, there have been limited studies breaking down the wealth effects on different types of health services. For example, Cheng et al. (2018) use lottery wins to establish the causal relationship between wealth and health care utilization. They find that lottery winners with larger wins are more likely to choose privately provided health services than publicly provided health services. Extending from Cheng et al. (2018)’s paper, Costa-Font et al. (2019) exploit housing booms and busts to investigate the wealth effects on long-term care utilization. However, these studies are limited to services commonly covered under the public insurance program. Although hospitalization and long term care are the main services used by older people, services that are not widely covered under public insurance programs, such as dental care or prescription drugs, also make up a large proportion of older individuals’ health expenditures (De Nardi et al., 2015). Therefore, breaking down the wealth effects on different types of health services is critical in planning an effective government response to the changes in population health care use, especially during economic downturns.

To our knowledge to date, this is the first paper comprehensively investigating the relationship between housing wealth and health care utilization in a multi-payer system. Our identification strategy relies on observing changes in health care utilization in response to the unanticipated wealth gain/loss created exogenously by the Great Recession. Unlike existing studies, our identification strategy utilizes a more general population of older homeowners in the United States. We also contribute to the existing literature by breaking down our investigation into different types of health services, including services not covered by public insurance, such as dental care. We estimate a Fixed Effect model using within-county variation in house prices as an instrument for housing wealth. We find that health care utilization increases on receipt of a windfall in wealth. Our findings provide additional evidence to the existing studies regarding wealth effects, especially housing wealth. We also explore potential mechanisms, complementing papers linking wealth with health outcomes.

The structure of the paper is as follows. The next section provides a brief introduction to the U.S. health care system, while the third section describes the dataset used in the analysis. Section 4 explains the empirical framework used for identification. Section 5 reports our estimation results and conducts robustness and heterogeneity analysis. The paper ends with a discussion of the findings.