Social Security Income and Health Care Spending: Evidence from the Social Security Notch

• DOI: http://doi.org/10.1111/sjoe.12218
• Date: 01 April 2018
• Published date: 01 April 2018
• Author: Yuping Tsai

©The editors of The Scandinavian Journal of Economics 2016.

Scand. J. of Economics 120(2), 440–464, 2018

DOI: 10.1111/sjoe.12218

Social Security Income and Health Care

Spending: Evidence from the Social Security

Notch*

Yuping Tsai

Centers for Disease Control and Prevention, Atlanta, GA 30329, USA

ytsai@cdc.gov

Abstract

In this paper, I exploit Social Security legislation changes to identify the causal effect

of Social Security income on out-of-pocket medical expenditures of the elderly. Using

the 1986–1994 Consumer Expenditure Survey and an instrumental variables strategy,

the empirical results show that health care expenditures are responsive to changes in

Social Security income for elderly individuals with less than a high-school education.

The estimated income elasticities are between 1.41 and 3.47, depending on the outcome

measures, and are statistically significant at conventional levels. The findings are in

contrast to existing studies that find a small income elasticity at the individual/household

level.

Keywords: Health care expenditures; income elasticity; Social Security benefits

JEL classification:H55; I10; J14

I. Introduction

Understanding the relationship between income and health care spending

is important for several reasons. First, the findings have implications

for how health care spending is likely to evolve and the optimality of

the growth of the health care sector. If health care spending increases

considerably with income so that rising income can explain most or all of

the rising costs of health care, it would be more likely that the increasing

share of gross domestic product allocated to health is socially optimal

(Acemoglu et al., 2013). Second, it could affect how policy evaluations

are conducted because policies that affect individual/household income

might also affect how they consume medical care. The topic has naturally

received considerable attention from researchers (Kleiman 1974; Newhouse

and Phelps, 1976; Newhouse, 1977; Maxwell, 1981; Leu, 1986; Brown,

1987; Manning et al., 1987; Parkin et al., 1987; Gerdtham et al., 1992;

*The content of this paper does not reflect the official opinion of the Centers for Disease

Control and Prevention. Responsibility for the information and views expressed in the paper

lies entirely with the author.

Y. Tsai 441

Barros, 1998; Di Matteo and Di Matteo, 1998; Fogel, 1999; Gerdtham and

Jonsson, 2000; Freeman, 2003; Di Matteo, 2004; Moscone and Tosetti,

2010; Acemoglu et al., 2013). However, their findings are inconsistent

and generally vary by the level of health care spending – at the national

level, the income elasticity of health care expenditures is greater than 1

while the income elasticity at the individual/household level is typically

near zero (Getzen, 2000).

There are relatively few studies in this body of literature that use the

individual/household-level data, in part due to the virtue that income is

usually not a policy instrument to affect the demand for health care,

and also due to the limited availability of data for both the income

and health care expenditure variables. However, the primary reason is

the potential endogeneity of individual/household income. Most previous

studies compare health care expenditures by household income or by

poverty status. Thus, omitted-variable bias is a serious concern because

income is most likely to correlate with unobservables that associate with

health care spending. One example is that healthy individuals are likely to

have higher income while their health care spending is likely to be low.

Not controlling for the health status of the individual would lead to a

downward bias in estimating the income effect on health care expenditures.

To my knowledge, only one study, the Rand Health Insurance Experiments,

has addressed this issue (Newhouse and the Insurance Experiment Group,

1993). Between 1974 and 1981, the experiment randomly assigned families

to health insurance plans with different levels of cost-sharing. The sample

included families with adults under the age of 62. Although the focus

of the study is to assess the effect of cost-sharing on medical service

use and health, a subexperiment finds that a small, unanticipated, and

temporary increase in one’s income has no significant impact on one’s

health care expenditures.1,2

The goal of this study is to estimate the responsiveness of household

health care expenditures to changes in income, focusing exclusively on

the elderly population and on the Social Security portion of their income.

To my knowledge, this is the first study to estimate the income elasticity

1The subexperiment gave an annual lump-sum payment of a maximum of US$250 to a

subsample of families and the results show that the group with the extra payment had US$4.5

lowerexpenditures than the control group. However, the result cannot truly represent the income

elasticity of health care expenditures in general because the income increase is temporary

and unanticipated.

2Previous studies have estimated the income elasticity of medical service utilization. For

example, using data from the 1963 Center for Health Administration Studies survey, Newhouse

and Phelps (1974) find that elasticities with respect to wage income are around −0.15 to

−0.35 for hospital days and 0.1 for physician visits. Newhouse and Phelps (1976) find a

small income elasticity of hospital and physician services (i.e., in the range of 0.03 to 0.14).

©The editors of The Scandinavian Journal of Economics 2016.