Price Indexes for US Medical Care Spending, 1980–2006

• Published date: 01 March 2020
• Author: Ana M. Aizcorbe,Tina Highfill
• Date: 01 March 2020
• DOI: http://doi.org/10.1111/roiw.12379

Publishe d 2018. This article is a U.S. Gove rnment work and is in t he public domain i n the USA.

205

PRICE INDEXES FOR US MEDICAL CARE SPENDING, 1980–2006

by AnA M. Aizcorbe* and TinA HigHfill

U.S. Bu reau of Economic An alysis

We construct price indexes for medical care spending in the US economy for the period 1980–2006. Our

indexes show slower price growth than the official deflator from 1987–2001, consistent with the fact that

indexes that improve on the official statistics typically find slower price growth than the official indexes.

However, the result is reversed for the 2001–2006 time period. We develop a decomposition that parses

out the numerical differences in these indexes into three factors that are held constant in the official

price indexes but are not in our indexes: changes in the type of provider supplying care, changes in the

type of insurance plan used by the patients, and changes in the bundle of procedures used to treat

patients. Our results suggest that using the official price measures may provide misleading conclusions

about spending trends and productivity growth in this important sector over this time period.

JEL Codes: I13, I19, C43

Keywords: health care spending, health insurance, price indexes

1. inTroducTion

In this paper, we provide new price indexes for medical care in the United

States using data from nationally-representative surveys of medical expendi-

tures in the civilian non-institutional population in 1980, 1987, 1996 and 2006.

Specifically, we provide indexes called Medical Care Expenditure indexes (MCEs),

a term coined by a Committee on National Statistics (CNSTAT) panel (Schultze

and Mackie, 2002). Like the official price indexes, MCEs allow one to decompose

changes in expenditures into changes in price vs. changes in quantity. However,

unlike the official price indexes, MCEs redefine the commodity of interest as the

treatment of disease and can answer questions like: What is happening to the cost

of treating cataract? In contrast, even when reweighted by disease, the official price

indexes answer a related but different question: What is happening to prices of the

specific procedures used to treat cataract?

Various previous studies report MCEs for different patient populations and

time periods to assess potential biases in the official statistics.1 Our empirical con-

tribution is to provide these price measures for a more comprehensive population

of patients than previously done (civilian non-institutionalized population) and

1Among the recent studies that construct these measures are Song et al. (2009), Aizcorbe and

Nestoriak (2011) and Dunn et al. (2013) for commercially-insured patients; Hall and Highfill (2013) for

Medicare patients, Bradley (2013) and Aizcorbe et al. (2011) for the civilian non-institutionalized

population.

Note: Many thanks to Abe Dunn, Kyle Brown, Bonnie Murphy, and two anonymous referees for

helpful comments.

*Correspondence to: Ana M. Aizcorbe, U.S. Bureau of Economic Analysis, 4600 Silver Hill Road,

Washington DC 20233. (ana.aizcorbe@bea.gov)

Review of Inc ome and Wealth

Series 66 , Number 1, March 2020

DOI : 10.1111 /roi w.123 79

bs_bs_banner

Review of Income and Wealth, Series 66, Number 1, March 2020

206

Publishe d 2018. This article is a U.S. Gove rnment work and is in t he public domain i n the USA.

for earlier periods (1980–87 and 1987–2001). We also develop a decomposition that

parses out numerical differences in the two types of indexes. We apply the decom-

position to help interpret the differences in the growth rates in the MCEs we calcu-

late here and a weighted average of the price indexes used as deflators in the

national accounts, something we call the BEA deflator.

A comparison of our MCE indexes to the BEA deflator revealed a counterintu-

itive result. Most empirical studies that aim to improve the official statistics typically

provide new price indexes that show slower price growth than the official statistics.

However, comparing our MCEs to the price indexes used in the national accounts, we

find that, for one of the time periods that we consider, our indexes sometimes show

faster price growth than the BEA deflator. For the earliest time period, our MCE

index grows at a compound annual growth rate (CAGR) of 9.2 percent from 1980–

87, very close to the 8.9 percent price increases currently in the national accounts;

over the period 1987–2001, our indexes grow 3.7 percent, substantially slower than

the 4.8 percent growth rate in the national accounts. The counterintuitive result

occurs in the most-recent time period, where the disease-based price indexes show

faster growth than the BEA deflator (a 5.4 percent vs 3.4 percent CAGR).

We develop a simple decomposition to help us explain these results by parsing

out differences in our MCEs and the BEA deflator into three components: differ-

ences that stem from shifts in treatments across provider types, those stemming

from shifts in patients across type of insurance plans, and a residual category that

mostly captures changes in utilization. Consistent with previous studies, we find

that shifts across provider types hold down growth in the MCE indexes relative

to the BEA deflator in all three periods, with higher effects in the earlier period

than later. With regard to the other two effects, the net effect of insurance shifts

and utilization changes is positive in the first and last periods, and very small in

the middle period. We argue that the well-known shift from relatively generous

fee-for-service plans to more restrictive managed care plans between the late 1980s

and early 1990s and the subsequent backlash that began in the early 2000’s likely

played a role in generating this pattern. In particular, our results are consistent with

the notion that the sharp growth in managed care plans over the 1987–2001 period

likely held down growth in the MCEs as the arrival of less-generous managed care

plans held down utilization growth in that segment and generated spillovers that

held increases in utilization in check in other insurance segments as well. Although

our data only allow us to decompose the effect of shifts across insurance types sep-

arately in the last period (2001–2006), we find that this effect is very small, despite

the managed care backlash that prompted patients to switch back to more-gener-

ous plans. Instead, most of the growth in the MCE above that in the BEA deflator

in that period is accounted for by growth in utilization, a result consistent with

Pinkovskiy (2013) and Dunn et al. (2014). We argue that this makes it unlikely that

insurance shifts, in and of themselves, account for much of the differences between

the MCE and official price index during the earlier run-up.

The paper is organized as follows. The next section compares how the offi-

cial price indexes and the MCEs define the commodity provided by the health

sector, discusses the factors that can cause these indexes to diverge, and provides

a decomposition to parse out any observed differences. After a brief discussion

of our data sources, we apply the decomposition to our data and argue that the