The U.S. Pharmaceutical Market: Expenditures, Insurance, Innovation and Generic Prescribing from 1960 to 2016

Introduction. Between 1960 and 2016, the U.S. pharmaceutical market can be characterized by a series of trends. First, the real, per capita expenditures on retail pharmaceuticals has risen by almost tenfold, from $85.12 to $922.50 in 2009 prices. Second, the share of drug spending covered by private and public health insurance programs increased from barely 1.5% to almost 86%. Over the same period, the annual FDA approval of new molecular entities (NMEs) has seen a relatively weak upward trend. Finally, since 1984, the share of retail prescriptions dispensed as generics has gone from 19% to almost 90% in 2016, and yet the pharmaceutical price index has continued to outpace general inflation. This paper attempts to empirically assess the interdependence among these time series variables.

Method and Data. We employ an econometric method commonly used in macroeconomics, but has never been applied to examine the pharmaceutical market. We construct a three-equation Vector Auto-Regressive (VAR) model, and apply the Granger Causality Test to determine whether the data are consistent with hypothesized cause-and-effect relationship between spending and insurance coverage, NME introduction and insurance coverage, and spending and NME introduction. The analyses are conducted on two time periods: from 1960 to 2016, without the inclusion of generic prescribing variable; and from 1984 to 2016 including generic prescribing as an exogenous shock variable. All variables are transformed to ensure stationarity in the VAR model. Publicly available data on annual drug expenditures, insurance coverage, and population are obtained from the National Health Expenditure Accounts. Data on generic prescribing are obtained from IMS Health (now IQVIA). Price indices are taken from the Bureau of Labor Statistics.

Results. We find a two-way, positive feedback relationship between increase in insurance coverage and increase in real, per capita spending during the overall period and from 1984 to 2016. We also find evidence that past NME introductions (positively) Granger-cause current insurance coverage, but not the other way around, in both periods. Evidence on the causality direction between spending and NME introduction is less consistent. Approximately one-fourth to one-third of the annualized growth rate of real spending may be attributed to insurance expansion, with the insurance effect being less pronounced during the period from 1984 to 2016. Generic prescribing has a highly significant and negative effect on real spending from 1984 to 2016, but its effect is largely offset by rapid insurance expansion in the same period.

Conclusion. Increase in private and public insurance programs has contributed significantly to spending growth on pharmaceuticals, which in turn has led to further insurance expansion. Likely due to enhanced negotiating power of large payers, the insurance effect on spending has lessened since 1984. Introduction of novel drugs has contributed to higher insurance coverage. The unanticipated influence of replacing self-paying consumers with payers as the responsible parties for drug purchases may have largely countered the anticipated effect of increased generic prescribing following passage of the Hatch-Waxman Act. Public policies should strike a balance between improving access and constraining moral hazard attributable to insurance expansion.