The Big Idea
Gauging insurance prices
Stephen Stanley | July 7, 2023
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
For some items, measuring prices is quite straightforward. There is little ambiguity gauging the change in prices of gasoline or a gallon of milk. But for many services, it is not always clear exactly how to separate real output from price changes. One example is insurance. There are several different ways to think about the cost of insurance, and the measurement of health insurance and motor vehicle insurance in the CPI and PCE deflator illustrates how varying methods can lead to significant differences in measured inflation.
CPI versus. PCE deflator
Broadly, before delving into the detail of insurance prices, it may be useful to note the differences at a conceptual level between the CPI and the PCE deflator. They are designed to answer two alternative questions. The CPI is closer to what the average person thinks about when he or she considers inflation. The BLS notes that “the CPI measures inflation as experienced by consumers in their day-to-day living expenses.” We can think of it as a gauge of the cost of living.
In contrast, the PCE deflator is a means to measure real GDP. The nominal dollar value for every line item that the BEA includes in personal consumption expenditures must be allocated into real consumption and price changes. As a result, the PCE deflator is considerably broader than the CPI. For one thing, significant spending is included in PCE that does not actually come out of households’ pockets. For example, when the federal government changes reimbursement rates for Medicare or Medicaid, the PCE deflator is affected because all health care outlays, regardless of who pays, are considered consumer spending, but the CPI is not. The BEA also has to impute prices for services that consumers do not explicitly pay for, such as certain financial services.
Motor vehicle insurance
The BLS tries to measure out-of-pocket expenses in the CPI. For most types of insurance, the BLS collects detailed data from insurers to ascertain the cost of a quality-constant policy. As all drivers are aware, the cost of motor vehicle insurance has exploded in recent years, reflecting the surge in vehicle prices and repair costs. The CPI measure corresponds pretty closely with what individual carriers are reporting. Every single monthly reading since the beginning of 2022 has been up substantially, yielding double-digit year-over-year increases for some time. In the May CPI, motor vehicle insurance rose by 17.0% on a year-over-year basis. This line item has a 3.3% weight in the core CPI, so this category alone is responsible for almost six tenths of a percentage point of the year-over-year core CPI rise. That works out to just over a tenth of the total rise in the core (5.3% in May). In short, motor vehicle insurance has been a major driver of core CPI inflation.
The BEA uses a different methodology because it is seeking to measure a different concept. The consumption of motor vehicle insurance services by definition has to be some sort of a net calculation involving premiums minus payouts. The idea is that if I pay out $100 in premiums and then receive $90 in benefits, then consumer expenditures are $10, not $100. This is reflected in the BEA’s measure of motor vehicle insurance prices. For the CPI, the price is just premiums. For the BEA, it is the variation between premiums and payouts. In practical terms, the BEA does not even look at the CPI measure for motor vehicle insurance. Instead, it uses the PPI gauge, which is consistent with the BEA thought process and has been up significantly but much more restrained than the CPI version. On a year-over-year basis, the motor vehicle insurance piece of the PCE deflator was up by 8.8% in May, about half as much as the CPI gauge. More importantly, the line item weight is only 0.6% of the core PCE deflator because the spending figure is net, not gross, as described earlier. As a result, the contribution of motor vehicle insurance to the core PCE deflator is only 5 bp, less than one-tenth of the corresponding CPI contribution.
Ideally, BLS statisticians would like to measure health insurance costs the same way that they do motor vehicle insurance. However, the BLS Handbook of Methods notes that “the data needed from insurers are so extensive and so closely held that BLS has not been able to construct a constant-quality health insurance index.” In other words, BLS cannot collect enough information to measure the price over time of a theoretical constant-quality insurance policy.
As a result, BLS has been forced to use an indirect method. It tries to measure the cost of various health care services, which insurance companies are mainly paying, directly. So, BLS collects price data on hospital services, doctor visits, and so on regardless of whether the bill is paid by an insurer or the individual, and these line items get the bulk of the weight in the index. What is left is mainly health insurers’ costs and profits. BLS gathers data on premium income, benefit payments, and retained earnings. When health insurers pay out less relative to their premiums collected, which amounts to a larger profit margin, the BLS sees this as a price hike. And vice versa.
These data from health insurers are collected on an annual basis, and the CPI component basically changes once a year, in October. In 2020, benefit payments fell sharply because most people were not going to their doctors or to the hospital much—a except for Covid cases—so health insurance operations saw a massive jump in profit margins. This was reflected in the CPI from October 2021 through September 2022, when the CPI gauge of health insurance prices surged by over 2% a month or 28.2% year-over-year in September 2022. At the time, that line item, though it accounted for less than 1% of the core (because most of what insurers were paying was allocated into other categories like hospital services and physician services), still contributed almost a quarter of a percentage point to overall core CPI inflation.
Health insurers’ profit margins normalized in 2021, as benefit payments increased. Starting last October, the health insurance line item flipped from around a 2% gain per month to a 4% loss per month. By this September, the year-over-year drop will be in the neighborhood of -40%, worth about a third of a percentage point drag on overall core CPI.
In contrast to all of the excitement in the CPI, health insurance prices in the PCE deflator data have been quiet. The December year-over-year increases for the health insurance component of the PCE deflator were 1.5% in 2019, 2.6% in 2020, 1.5% in 2021, 1.9% last year. Thus, in contrast to the CPI, health insurance has been mostly an inconsequential line item for the PCE deflator.
Sometimes, the devil is in the details. Measuring prices in the services sector is often problematic. The particular methodologies used by the BLS and BEA to calculate motor vehicle and health insurance have led to big swings and widely different contributions to overall core inflation. In my view, the BLS methodology for motor vehicle insurance better reflects the true price environment, one of high inflation, than the BEA figures. In contrast, the volatility in the CPI measure of health insurance costs is, in my view, mostly noise.
In any case, the hope is that the various quirks in the data wash out and the broad aggregates provide an accurate representation of underlying inflation pressures. However, in such a volatile economic environment as we have seen through the pandemic and post-pandemic years, quirks in the way government agencies estimate prices can influence the reported inflation rates.