EXPLAINING THE DECLINE OF THE U.S. SAVING RATE: THE ROLE OF HEALTH EXPENDITURE

• Author: Béla Személy,Yi Chen,Maurizio Mazzocco
• DOI: http://doi.org/10.1111/iere.12405
• Published date: 01 November 2019
• Date: 01 November 2019

INTERNATIONAL ECONOMIC REVIEW

Vol. 60, No. 4, November 2019 DOI: 10.1111/iere.12405

EXPLAINING THE DECLINE OF THE U.S. SAVING RATE: THE ROLE OF HEALTH

EXPENDITURE∗

BYYICHEN,MAURIZIO MAZZOCCO,AND B´

ELA SZEM ´

ELY1

Jinan University, China; UCLA, Bunche Hall, Los Angeles, CA, U.S.A.; Ofce of the

Comptroller of the Currency, U.S.A.

The U.S. saving rate declined by 8% between 1980 and 2009. We document that the decline can be explained

by rising health expenditures. Using exogenous variation in medical expenses generated by Food and Drug

Administration drug approvals, we document that a 1 percentage point increase in health expenditure generated

a decline in saving rate of 0.9 percentage points. We then estimate a model of household decisions to evaluate

the mechanisms behind the decline. We find that the rise in health expenses and drop in saving rate are driven

by progress in health technology, reduction in copayment rates, and improvements in income processes.

1. INTRODUCTION

It is well known that the U.S. personal saving rate declined from 11% in the late seventies to

about 3% in the late 2000s. Economists and policymakers are concerned with this drop because

it may signal an increased dependence on foreign investment and a future reduction in capital

stock with negative consequences for labor productivity, wages, and national output. In the past

20 years, economists have attempted to explain this sharp drop. An examination of the related

literature indicates that the decline is still a puzzle. Parker (1999) states that “[e]ach of the major

current theories of the decline in the U.S. saving rate fails on its own to match significant aspects

of the macroeconomic or household data.” Guidolin and La Jeunesse (2007) review a number

of arguments and theories that have been proposed and conclude that “[t]he recent decline of

the U.S. private saving rate remains a puzzle.”

The main contribution of this article is to provide evidence that the increase in medical

expenditure as a share of disposable income is the major driver of the decline in the U.S.

personal saving rate. The definition of the personal saving rate used by the saving literature

and by the National Income and Personal Accounts (NIPA) includes total health expenses in

the computation of household-sector expenditure, i.e., health expenses paid by consumers plus

health expenses paid by some form of health insurance. We document that the decline is mostly

driven by health expenses paid by some form of medical insurance. We provide this evidence

in two steps.

In the first step, after having documented using a simple accounting exercise that the share of

health expenditure on its own can explain most of the drop in the U.S. saving rate, we provide

evidence of a causal relationship between those two variables. The evidence is provided using

cross-state changes in health expenditure and saving rates and, as an arguably exogenous

source of variation in medical expenses, the approval of new drugs by the Food and Drug

Administration (FDA). The idea behind the choice of this instrument is straightforward. When

∗Manuscript received February 2016; revised August 2018.

1We thank the managing editor Hanming Fan and two anonymous referees for valuable feedback and suggestions.

We are also grateful to Mary-Ann Bronson, Maria Casanova, John Kennan, Rustin Partow, and participants at various

seminars and conferences for helpful comments. Chen acknowledges financial support from the 111 Project of China

(Project Number: B18026). The views expressed are those of the authors alone and do not necessarily reflect those of

the Office of the Comptroller of the Currency. Please address correspondence to: Maurizio Mazzocco, Department of

Economics, UCLA, Bunche Hall, Los Angeles, CA 90095. E-mail: mmazzocc@econ.ucla.edu.

1823

C

(2019) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1824 CHEN,MAZZOCCO,AND SZEM´

ELY

a new drug is approved and made available to users, medical expenses will generally increase

since consumers have a new product that can help treat medical conditions. A limitation of this

variable is that, in each period, it is constant for the entire United States, whereas our analysis

requires changes across states. To generate the required cross-state variation, we interact the

FDA approval of new drugs with the demographic characteristics of a state. The constructed

variables provide the needed variation, since the approval of a new drug produces larger

increases in health expenditure in states with a larger fraction of families that make extensive

use of medical products. We find that a 1 percentage point increase in health expenses generates

a reduction in the saving rate of about 0.9 percentage points.

Using the reduced-form results alone, it is difficult to evaluate which mechanisms are behind

the rise in medical expenditure and the corresponding decline in the personal saving rate. To

deal with this limitation, in the last part of the article, we develop and estimate a model of

consumption, saving, and health decisions. In the model, health expenses are beneficial because

they improve health status and, hence, reduce the individual mortality rate. People may be

affected by two types of health conditions: nonsevere, such as high blood pressure and the

flu, and severe, such as cancer and heart attacks. Individuals can choose whether to undertake

treatment for nonsevere conditions, but must seek treatment for the severe conditions or their

mortality probability increases to one. Each person is covered by a private or public health

insurance. Thus, incurred total medical expenses can be divided into two categories: expenses

paid by the person seeking treatment and the medical expenses paid by the private or public

health insurance that covers the person. In the model, to be consistent with the definition of

the saving rate used by NIPA and in the literature, both types of expenses enter the calculation

of the household-sector saving rate. But only the health expenses and the health insurance

premium paid by a person affect individual decisions.

The model is estimated using data from the Medical Expenditure Panel Survey (MEPS), the

NIPA, the Current Population Survey (CPS), the National Health Interview Survey (NHIS),

and the National Death Index (NDI) Mortality Files. The corresponding simulations indicate

that the rise in medical expenditure explains most of the decline in the saving rate experienced

by the U.S. economy. In our model, there are three main mechanisms that can generate the

decline: progress in health technology, which induces higher medical expenses; changes in

copayment rates and the corresponding variation in the probability of seeking treatment; and

changes in income with the associated changes in treatment rates. Using counterfactuals, we

find that technological progress accounts for about 50% of the decline in the saving rate and

is mainly responsible for the drop experienced by the U.S. economy between 1995 and 2009.

Changes in copayment rates explain about 25% of the reduction in the saving rate and account

for most of its decline during the period 1986–1994, when the copayment rates experienced a

steep drop. Finally, improvements in the income process explain the remaining 25% and are

responsible for most of the reduction in the saving rate during the years 1980–1985 and part of

the drop for the period 1986–1994.

Our results indicate that the decline in saving rate was mostly driven by the increase in

health expenses paid by some form of health insurance. But who bore the cost of the increase?

In the article, we provide suggestive evidence that the rise in private health expenditure—

medical expenses covered by some form of private health insurance—was mainly financed by

an increase in employer contributions to health plans. We then provide suggestive evidence

that the employers transferred part of the increase in contributions to the household sector in

the form of lower wages and salaries. The corporate and household sectors therefore shared the

burden of the rise in private health expenses. Finally, we document that the rise in public health

expenditure—expenses paid by some form of public health insurance—was mainly funded by a

reduction in expenditure on other government items and by issuing debt.

There is one paper that is particularly relevant to understanding our results. De Nardi et al.

(2010) study the effect of out-of-pocket medical expenses on savings of older households and

find, using the Health and Retirement Study, that higher out-of-pocket medical expenses in-

crease their savings. At first sight, this finding appears to contradict our results. But this is not

EXPLAINING THE DECLINE OF THE U.S.SAVING RATE:THE ROLE OF HEALTH EXPENDITURE 1825

the case, since the two papers focus on the effect of two different variables on saving decisions.

De Nardi et al. (2010) analyze the effect of out-of-pocket expenses, whereas we consider the

effect of total health expenditure, since the aggregate saving rate is computed using all health

expenses. We document that during the period 1980–2009 out-of-pocket expenses—health ex-

penses plus health insurance premiums paid by consumers—remained approximately constant

in the aggregate. It was the rise in medical expenditure paid by some form of health insurance

that produced the decline in the saving rate. Our results therefore do not contradict the findings

in De Nardi et al. (2010) but complement them in understanding the impact of medical costs on

saving decisions.

Our results have policy implications. They indicate that if policymakers intend to raise the

saving rate, it is essential to reduce the rate at which medical expenses grow. The recent debate

on health costs has focused on two sets of policies that have the potential of reducing the effect

of increasing medical expenses on the U.S. economy. The first set includes policies aimed at

eliminating inefficiencies in the provision of health care. This is clearly a good starting point, to

the extent that it is able to generate significant reductions in medical costs. The second set of

policies requires health institutions to be more selective in the adoption of newer technologies.

These policies can have a large impact on health expenses and, hence, on the saving rate. But

they come at a cost. First, as pointed out by Hall and Jones (2007), if health care is a luxury

good, in a rich country such as the United States, it may be welfare improving to adopt the latest

health technology. Second, the constant adoption of new health technologies has the positive

effect of fostering a large number of innovations in the health sector. A more frequent use

of older technologies might slow this progress and could have negative welfare effects in the

long run.

The rest of the article proceeds as follows: The next section provides a discussion of related

papers. In Section 3, we describe the data sets and define the variables used in the article.

Section 4 uses an accounting exercise to document that most of the decline in the U.S. household

saving rate can be explained by the rise in health expenditure. In Section 5, we use cross-

state variation to provide evidence on a causal relationship between health expenditure and

the saving rate. Section 6 develops and estimates a model to evaluate which mechanisms are

behind the increase in medical expenditure and the corresponding decline in the saving rate.

Section 7 concludes.

2. RELATED PAPERS

A large number of studies have analyzed the decline in the U.S. saving rate. In this section,

we will discuss the papers with findings that are related to ours. For a thorough review of the

literature, see Browning and Lusardi (1996), Parker (1999), and Guidolin and La Jeunesse

(2007).

One of the first papers to address the decline in the U.S. saving rate is the work by Summers

and Carroll (1987), where they study the changes in the national and household saving rate from

the 1950s to 1986. Their main conclusion is that the decline in the private sector U.S. saving rate

is real and not a result of measurement issues and that the most likely cause of the decline is the

increase in expected income after retirement that has induced the younger cohorts to reduce

the rate at which they save. In our article, we only focus on the household saving rate that

experienced most of its decline after 1986, the last year considered by Summers and Carroll.

The paper by Gokhale et al. (1996) is the first one to discuss the steep increase in health

expenditure in the past 50 years and to suggest a possible relationship with the drop in the

saving rate. They report that medical consumption as a percentage of disposable income was

3.9 in the 1950s, 5.2 in the 1960s, 7.3 in the 1970s, 10.1 in the 1980s, and 12.8 in the early 1990s.

This pattern suggests that it is difficult for medical expenditure to explain the decline in the

saving rate. Health expenses were already growing in the 1960s and 1970s, whereas the saving

rate started its decline at the end of the 1970s. Probably, for this reason, Gokhale et al. (1996)

do not directly explore the effect of the increase in medical consumption on household savings.