Financial Effects of Pharmaceutical Price Regulation on R&D Spending by EU versus US Firms

The Dorgan-Snow US Senate Bill, S.334 proposed legalized importation, often called reimportation because pharmaceuticals are often developed and manufactured in the US, exported to a price-controlled EU country or Canada, and then reimported back to the US. Reimportation is an indirect method of price control that is similar to European parallel trade. Increased parallel trade in the 1990s has decreased pharmaceutical prices in the EU.[1] Parallel trade is a more general term than reimportation because it refers to any product trade that occurs after the original manufacturer sells the product. For example, product manufactured in Germany, sold to an agent in Spain, and then resold by the agent in Spain to an agent in England is parallel trade. Reimportation fits the more common situation in the US, where a US manufacturer sells to a Canadian agent, who then resells to another agent in the US. The Medicare Prescription Drug Price Negotiation Drug Act of 2007 (S.3) proposed direct price negotiation between Medicare officials and drug firms.[2] President Clinton’s Health Security Act proposed price controls for new drugs.[3]

Maine allows reimportation. In August 2006, California legislators passed the Prescription Drug Initiative,[4] requiring manufacturers to discount pharmaceuticals by between 40% and 60% for low-income state residents.

The firms were Abbott Labs, Bristol-Meyers Squibb, Eli Lilly, Glaxo, Johnson & Johnson, Merck, Pfizer, SmithKline Beecham, Warner-Lambert, Wyeth-Ayerst (American Home Products), Ciba-Geigy, Dupont-Merck, G.D. Searle, Genentech, Hoechst-Roussel, Hoffmann-La Roche, Knoll, Marion Merrell Dow, Syntex, Upjohn and Zeneca.

US pharmaceutical and CPI (all items price 1982–4 = 100) indexes are from the Bureau of Labor Statistics.[14] EU CPI is from Eurostat Harmonized Indices of Consumer Prices (all items).[15] The EU pharmaceutical price index is from Eurostat[15] starting in 2001 and compiled from OECD Health Data 2003[16] for the years before 2001.

Owned by Standard and Poor’s Inc., Compustat[19] compiles financial and accounting data mostly from company financial reports such as 10K statements. These data are packaged into various databases, such as the segments database. Compustat data are widely used by accounting and finance researchers. Our study relies most heavily on the Compustat Global database. It is an objective source for detailed financial and market data on more than 13 000 publicly traded international companies in more than 80 countries.

These filters tend to eliminate small and young firms with few sales. Because the following results are size weighted, including these firms has little effect on the results.

The Compustat Global Industrial/Commercial data used for this table contain some errors that we have corrected. The errors were contained in the EU firms’ data, perhaps because Compustat’s US database has been available since 1950 and its foreign portion of the Global database is relatively new (since 1993). Figures for Aventis between 1999 and 2003 needed to be multiplied by 1000, as did AstraZeneca’s figures for 1997. R&D figures for GlaxoSmithKline between 1993 and 1995 and Novo-Nordisk in 1993 were incorrect due to typographical errors. All of these were changed based upon figures found in the firms’ annual reports. Finally, the very large merger between Sanofi and Aventis in 2004 posed a unique problem. Instead of Compustat figures , which reflected merger-related accounting, we simply used the firms’ combined 2003 figures for the combined firm in 2004. This likely understates the true figures somewhat, but because the firm booked a very large intangible asset to reflect the premium paid for the merger, the reported figures are large enough to cause the EU totals to be substantially overstated.

EFPIA member associations include those in Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and the UK.

Many firms voluntarily adopted price constraints in this period in response to the Clinton administration’s proposed pharmaceutical price constraints.

R&D shifts from the EU to US include Pharmacia in 1995, Aventis in 1999, GlaxoSmithKline in 2000 and Novartis in 2002. Reasons given for these moves include growing US sales compared with EU sales, and requirements that they perform clinical trials in the US, particularly FDA phase III trials. US-based trials also facilitate new drug marketing as firms establish relationships with top US physicians who set prescription guidelines for other physicians. There are also individual cases that show a direct connection between EU price controls and R&D shifts from the EU to the US. One case involves Bayer A.G. Spain and France set a price for Bayer’s second best-selling drug (Adalat®) at 40% below its price in England. When Bayer restricted supply to French and Spanish wholesalers who were reselling their supplies in England, the European Commission fined Bayer €3 million in 1996 and ordered them to stop restricting supply. Worldwide sales of Adalat®, which had been growing strongly, subsequently fell by 4% (3%) in 1998 (1999), mostly due to French and Spanish wholesalers reselling Adalat® at relatively low prices. As a consequence, Bayer cut its EU R&D spending by 1% in 1998 and by 14% in 1999. Conversely, Bayer increased its US R&D spending by 8% in 1998 and by 31% in 1999. Figures are taken from Bayer annual financial reports 1997–2001.[23]

The data on foreign firms’ US pharmaceutical R&D spending[24] and assets[25] come from the Bureau of Economic Analysis, assets by industry and country. These data are available starting in 1999. The data on US firms’ foreign pharmaceutical R&D spending come from the National Science Foundation.[26] Data for 2002 is the latest available.

Because the independent variables (\({\hat b_{i,1}}\) and \({\hat b_{i,2}}\)) are measured with error, the estimates of β₁ and β₂ could be biased. Greene[29] showed that, when one independent variable is measured with error, its estimate is biased toward zero, but when more than one independent variable is measured with error, the bias depends on a number of parameters. To the extent that the estimates are biased toward zero, the tests for statistical significance are conservative.

The following model is used to evaluate and compare R&D spending over time: where g _s is the growth in R&D spending from year t through year n; RD _{t + 1} is pharmaceutical R&D spending in year t + 1; PVRD _t is the present value of pharmaceutical R&D spending; and k is the discount rate associated with R&D spending flows.

Data on the distribution of R&D expenditures in the pharmaceutical and biotechnology industries across labour and non-labour inputs are not typically available in firms’ financial statements. We searched the financial statements filed on the US Securities and Exchange Commission’s Electronic Data Gathering, Analysis and Retrieval[37] database for those that discussed their labour inputs, and labour was mentioned as at least a majority of R&D expenses. We conducted a limited analysis of ‘pure play’ biotechnology firms with R&D expenditure and labour expense data listed on Compustat. These firms had little or no revenue from marketed products so they were solely research firms. Under reasonable assumptions about the distribution of labour expense between R&D and non-R&D employees, labour represents close to 90% of their R&D expenses.

For example, Germany sets annual pharmaceutical spending budgets for medical service providers, which in turn, tends to cap the amount that providers are willing to pay for pharmaceuticals. Annual budgets change at different rates in different years, with different implications for potential pharmaceutical price changes.