Generic Drugs in Canada: Market Structure — Trends and Impacts

December 2010

1 Introduction

This report examines market-structure trends among non-patented prescription drug products in Canada. It is principally concerned with the degree of sales concentration in Canadian generic markets, as well as the possible impact of such concentration on prices.

The analysis covers a set of 298 leading generic drug products.1 The analysis is restricted to cases where there were at least two generic versions of the product sold in Canada in 2007. Markets where there is a single generic supplier will be considered separately in a subsequent study.

All data used in this report are from IMS Health's MIDAS database. The data cover only drug purchases made by the retail pharmacy sector.2

1 Note that many of the markets considered in this analysis are “patented markets”, in the sense that the market includes at least one (branded or generic) product reported to the PMPRB as patented between 2003 and 2007. This analysis draws no distinction between such “patented markets” and markets where there are no such patented drug products.

Appendix B provides a detailed description of the process by which drug products were selected for inclusion in the analysis, as well as all other aspects of the methodology. A list of the drug products included in the analysis can be found in Appendix C.

2 Throughout this report the term “price” refers to prices paid by pharmacies to wholesalers or directly to companies, not those paid by consumers or drug plans. These prices are derived from sales data based on invoiced prices and do not include off-invoice discounts, free goods and other forms of price reduction such as rebates. Appendix B provides details on how prices are derived, as well as other aspects of the methodology.

2 Generic Drug Products: Trends in Market Structure

Table 2.1 presents basic market information for the top 10 suppliers of generic drug products (by 2007 sales) in Canada. The first column shows how many of the 298 leading generic drug products each supplier sold in 2007. The second column gives the corresponding estimated value of the supplier's sales revenue. The third column gives the supplier's average share of sales in those generic markets in which it was active. Hence, the first line in Table 2.1 indicates that in 2007 Apotex sold 246 of the 298 leading generic drug products, thereby generating $978.9 million in ex-factory sales. Among the 246 leading generic drug products in which Apotex was active, it had an average market share of 48.4%.

Table 2.2 provides basic results on the sales concentration in generic markets. These results indicate that, although sales concentration may have declined slightly from 2003 to 2007, Canadian generic markets remain highly concentrated. In particular, Table 2.2 indicates that in 2007 the two leading suppliers in a typical generic market accounted for 84.5% of sales, while the four leading firms accounted for 96.7% of sales. The Herfindahl index3 reinforces the finding of high concentration: this has a value of 49.1% for 2007, which is extraordinarily high by the standards usually applied in the interpretation of this measure.

Table 2.3 provides additional information on the individual ratios summarized by the average two-firm concentration ratio reported in Table 2.2. The results in this table show that in 2007, 117 (39.3%) of the 298 leading generic drug products were sold in markets where the two leading suppliers accounted for 95% or more of sales. Another 95 drug products (31.2%) were sold in markets where the two leading suppliers had between 75% and 95% of sales. In contrast, only 5 drug products (1.7%) were sold in markets where the two leading suppliers had less than 50% of sales.

Table 2.4 provides concentration statistics by major therapeutic class. These results are entirely consistent with the aggregate results of Table 2.2. No matter the therapeutic class, all measures indicate a very high degree of sales concentration.

Table 2.5 reports concentration statistics by the size of market. As might be expected, the results in the table indicate that the larger the market, the more entrants it can support, and hence, the smaller the degree of sales concentration. However, this relationship is not nearly as strong as one might expect. For example, the table shows that in markets with total sales of less than $5 million, there were typically 4.0 suppliers, with the two leading firms accounting for 89.1% of sales. By comparison, the 20 markets where there were sales of at least $25 million typically sustained 7.6 suppliers. Despite this, the two leading suppliers still accounted for 73.9% of sales, while the top four suppliers accounted for 92.8%. Thus it appears that in even the largest generic markets, where one would expect vigorous competition for market share, sales are usually dominated by a few suppliers.

Table 2.6 gives results specifically for the 20 drug products among the set of 298 with 2007 sales of $25 million or more. Although there were five or more suppliers in most of these markets, the two leading firms accounted for more than half of the sales in every case. These results permit only one conclusion: there are factors at work in the Canadian generic sector that bring about a high degree of sales concentration even where market entry is uninhibited and the number of active suppliers is large.

For the sake of comparison, Table 2.7 provides concentration statistics for foreign markets. These have been calculated in exactly the same way as the corresponding statistics given in Table 2.2. In particular, underlying observations are again restricted to the set of 298 leading generic drug products in Canada.

The results in Table 2.7 imply that Canadian generic markets are more concentrated than generic markets in other countries. Average Canadian two- and four-firm concentration ratios are higher — in most instances, substantially higher — than concentration ratios seen elsewhere. The value of the Herfindahl index in Canada is higher than that obtained for any country except the United Kingdom, whose value of 72.8 indicates a remarkable degree of sales concentration.4 Somewhat surprisingly, given the difference in scale of the two markets, the Canadian market structure appears to most closely resemble that prevailing in the United States.

It is of interest to know whether, on a market-by-market basis, the extent of market concentration in Canada correlates with levels of concentration in other countries. To this end, Table 2.8 gives the average number of generic suppliers by country for those markets where there are one or two suppliers, three or four suppliers and five or more suppliers in Canada.

The results in Table 2.8 provide clear evidence of a positive correlation between the number of generic suppliers active in Canada and the number of suppliers in individual foreign markets. This tendency is especially pronounced at the top end: where a drug has at least five suppliers in Canada, it will typically have at least that number of generic suppliers in every foreign market.

3 The Herfindahl index (sometimes called the Herfindahl-Hirschmann index) is calculated by squaring the market share of each supplier and then summing the resulting values. This calculation can return a value of 50% or more only if sales are dominated by one or two firms. If, in contrast, sales are evenly distributed among three firms, the Herfindahl index will have a value of 33%. This falls to 25% if sales are evenly divided among four firms, and so forth. Very high values of the Herfindahl index — values in the range of 70% to 100% — are possible, but can only occur if a single supplier accounts for at least 80% of sales.

4 The UK data includes many instances in which a single company accounts for 85% or more of sales (in which case the Herfindahl index will have a minimum value of 72%). This explains the extraordinarily high average Herfindahl index value obtained for this country.

3 Generic drug products: International Price Comparisons by Level of Concentration

Conventional economic theory acknowledges that market structure can affect price. In particular, standard models of price determination suggest that markets where sales are dominated by only a few firms may produce prices substantially exceeding the economic costs of manufacturing and distribution. Such models usually imply that the greater the concentration of sales among leading firms, the higher the price, all other things being equal. It, therefore, is of some interest to compare prices in Canada's highly concentrated generic markets with those prevailing elsewhere.

Tables 3.1a and 3.1b provide average foreign-to-Canadian price ratios for three distinct sets of generic drug products: those sold by one or two suppliers in Canada, those sold by three or four suppliers and those sold by five or more suppliers. Table 3.1a provides bilateral comparisons. It might be expected that Canadian prices would be highest relative to foreign prices where there are the fewest domestic suppliers and closer to international prices where there are more suppliers and, therefore, more competition in the Canadian market. In fact, the results Table 3.1a point in the opposite direction: where a pattern is discernible, it is that Canadian prices are relatively higher when there are a greater number of Canadian suppliers.

Table 3.1b provides multilateral comparisons for two sets of countries. The first of these encompasses the seven countries that the PMPRB normally considers in carrying out its price reviews: France, Germany, Italy, Sweden, Switzerland, the UK and the US. The second set encompasses all 11 foreign countries represented in the entire extract of MIDAS dataset available to PMPRB. Here again, we have the counterintuitive result of an inverse relationship between foreign-to-Canadian price ratios and the number of Canadian suppliers. The results for the case of five or more Canadian suppliers are especially striking: it appears Canadian prices are the highest relative to foreign prices in precisely those cases where one would expect price competition to be strongest in Canadian markets.

Tables 3.2a and 3.2b again give average foreign-to-Canadian price ratios for three sets of generic drug products, with the division in this case based on the Canadian two-firm concentration ratios. The results in these tables indicate that foreign generic prices tend to be substantially less than Canadian prices even in those cases where the Canadian sales concentration is relatively low. This is especially striking when one considers the results for the foreign mean, weighted mean and median reported in Table 3.2b.

Tables 3.3a and 3.3b give foreign-to-Canadian price ratios based on a division of drug products according to the size of the Canadian market. The results in these tables indicate that foreign generic prices tend to be substantially less than Canadian prices in even those cases where Canadian sales are the highest. Here again, the multilateral results are very striking: whether one considers the foreign mean, weighted mean or median, foreign prices appear to be no more than two-thirds of corresponding Canadian prices in those generic markets that are largest in Canada.

Table 3.4 provides supplementary bilateral price comparisons that directly address the question of how differences in concentration influence relative prices. In this case, before calculating average price ratios, drugs were assigned to one of four groups. Each group was defined by the extent of concentration in Canada and the relevant comparator country, with concentration measured by the two-firm concentration ratio (CR2). In particular, concentration was designated as “low” where the CR2 was less than or equal to 75%, and “high” where the concentration was greater than 75%. This low/high distinction applied in turn to the Canadian and foreign market yields the four-way division mentioned above.

The third column of Table 3.4 (CR2-Canada ≤ 75%, CR2-Comparator > 75%) represents instances where the concentration was low in the Canadian market but high in the corresponding foreign market. Comparing average price ratios in this column to those in the second column (CR2-Canada ≤ 75%, CR2-Comparator ≤ 75%), it appears that a higher foreign concentration tends to be associated with relatively higher foreign prices, as might be expected. The same pattern emerges when comparing the fourth (CR2-Canada > 75%, CR2-Comparator ≤ 75%) and fifth columns (CR2-Canada > 75%, CR2-Comparator > 75%).

Note, however, that this pattern does not emerge when it is the extent of Canadian concentration that is varied. Comparing results in the second and fourth columns of Table 3.4, it appears that foreign-to-Canadian price ratios are unaffected or even tend to rise as Canadian concentration levels rise. (Comparing results in the third and fifth columns leads to the same conclusion.)

The average price ratios obtained for the United States illustrate the overall trend of the results. In this case, no matter whether Canadian concentration is low or high, the US-to-Canadian price ratios tend to be higher in markets where the US concentration is high. The converse is decidedly not true: the average price ratio in markets where concentration is high in both the US and Canada, 0.70, drops to 0.39 in markets where the US concentration is high and the Canadian concentration is low. Canadian generic prices appear to be closest to their US counterparts where market concentration is high in both countries.

4 Conclusion

This report is principally concerned with the degree of sales concentration in Canadian generic markets, as well as the possible impact of such concentration on prices.

Canadian markets for generic drug products are highly concentrated with respect to sales. The two leading suppliers in a typical generic market accounted for 84.5% of sales in 2007, while the top four suppliers accounted for 96.7%. There are high levels of sales concentration in all therapeutic classes. In even the largest generic markets, where one would expect vigorous competition for market share, sales are typically dominated by a few suppliers.

On the other hand, an analysis of various foreign-to-Canadian average price ratios does not point to any simple relationship between market structure and the relative costliness of generic drug products in Canada. These results imply that relatively low sales-concentration ratios and the existence of numerous suppliers in Canada by no means guarantee that the Canadian price of a particular generic drug will be in line with prices in other countries. It follows that while markets for generic drug products definitely are highly concentrated at the level of the supplier, this is not the fundamental source of differences between Canadian and foreign prices.5

5 A recent study of the Canadian Competition Bureau concludes that “until recently, prices paid for generic drugs across the country tended to reflect the maximum generic drug prices allowed under Ontario's drug plan”. The study also concludes that competition among Canadian companies occurs principally through rebates provided to retailers. In a market framework of this sort one would expect more vigourous competition among companies to produce larger off-invoice rebates rather than lower prices. See Canadian Competition Bureau, Canadian Drug Sector Study, 2007.

Appendix A: International Price Comparisons at Purchasing Power Parities

This appendix reproduces the tables in Section 3 using purchasing power parities to perform currency conversions rather than market exchange rates. Tables A.1a and A.1b correspond to Tables 3.1a and 3.1b; Tables A.2a and A.2b correspond to Tables 3.2a and 3.2b; and Tables A.3a and A.3b correspond to Tables 3.3a and 3.3b.

Appendix B: Notes on Methodology

The results described in this report have been calculated from a large set of data on drug sales and prices in Canada and other countries. This section describes the data, sources, concepts and methods applied in performing these calculations.

• 1 Terminology
• 2 Price and Sales Data
• 3 Identification and Selection of Generic Drug Products
• 4 Matching Data Across Countries
• 5 Currency Conversion
• 6 Specific Reporting Elements: Statistical Methodology

1 Terminology

In this report, the term “drug” refers to any unique combination of active ingredient, strength and form. Except where manufacturer detail is required, all statistical results are calculated at the level of individual drug products and then aggregated.

The term “product” refers to a version of a drug sold by a particular supplier.

Results are sometimes reported by “major therapeutic class”. This refers to the Level 1 classes of the World Health Organization's Anatomical, Therapeutic, Chemical (ATC) drug classification system. Level 1 codes normally refer to the anatomical system on which the drug acts.

A “branded” product is one sold under a particular trade name. A “generic” product is sold under the name of its principal ingredient(s).

The term "supplier" refers to the economic entity (usually a corporation) that controls the manufacturer of a given drug product.6

2 Price and Sales Data

IMS Health's MIDAS database is the source of all price and sales data used in this analysis. MIDAS is a summary of data obtained from IMS Health's detailed audits of pharmaceutical purchases made by retailers (in 70 countries) and hospitals (in 45 countries). MIDAS contains information on sales of individual products, measured in both currency and physical units, as well as information on the product manufacturer, active ingredient, brand, form, strength, pack size, and therapeutic class.

The data used in this report cover prescription drug sales to the retail pharmacy sector. Sales data include direct sales by suppliers to pharmacies and indirect sales via wholesalers. The retail pharmacy sector accounts for most sales in all of the countries considered in the analysis.7

2.1 Sales

MIDAS includes a field representing IMS Health's estimate of the sales revenue in local currency received by the supplier selling the product in question. This measure of sales is used throughout the analysis. IMS's estimates are based directly on the purchase information obtained in its pharmacy and hospital audits.

To estimate the value of a supplier's sales of a particular product, IMS removes its estimate of wholesalers' mark-ups from the acquisition costs reported by pharmacies and hospitals.

It is important to understand that the acquisition costs used by IMS are based on invoiced prices. Off-invoice discounts, free goods and other forms of price reduction, such as rebates, are therefore not represented in the MIDAS data.

2.2 Quantities

MIDAS provides a measure of physical quantity it calls standardized units (SU). This measure is used throughout the report. Standardized units represent IMS's estimate of the number of normal doses that a given volume of physical units constitutes. This is simply the “number of pills” in the case of oral solids.

2.3 Prices

A drug product's price for any given period is obtained by dividing the value of the manufacturer's sales of that product by the number of standardized units of the product MIDAS reports for that period. Note that the prices thus obtained are properly interpreted as estimated ex-factory prices (i.e., prices charged by suppliers) and not as final retail prices paid by consumers or drug reimbursement plans.

Prices obtained at the level of the drug are derived in a similar fashion, except that sales and quantities of the drug are first summed across suppliers and the resulting sums divided.8

2.4 Patent Status

For the purpose of this study, a drug is identified as patented if at least one version of the drug (either branded or generic) was subject to price review by the PMPRB at some point over the period 2003–2007. If no version of the drug was reviewed by the PMPRB over this period, it is considered non-patented.

Every year the PMPRB, as an addendum to its Annual Report, publishes a list of patented drug products that were subject to price review as of the end of the calendar year. These lists were used to determine the patent status of individual drug products.

3 Identification and Selection of Generic Drug Products

A master list of leading generic drug products was constructed for the analysis of 2007 generic sales and prices. The following steps describe the method by which this was done.

Step 1. Using MIDAS data, drugs sold in 2007 by leading Canadian generic suppliers were identified. Drugs with aggregate 2007 generic sales of less than $1 million were then deleted.

Step 2. Using information from Health Canada's Drug Product Database, any non-prescription drugs were eliminated from the list of drugs obtained in Step 1.

Step 3. A list of all products corresponding to the drugs remaining after Step 2 was obtained. Note that the resulting list of products encompasses all branded and generic versions of the drug.

Step 4. For each remaining product, the contents of the MIDAS database's Local Product Name and Molecule fields were compared.9 Provided that the contents of the two fields were reasonably similar, the product was retained as a generic.10

Step 5. For each remaining drug, the number of generic versions on the Canadian market in 2007, was determined. If there was only a single generic version available in 2007, the drug was deleted.11

4 Matching Data Across Countries

When performing international price comparisons, matching products across countries can be difficult. Prices for the same drug can vary appreciably by strength, form and pack size, even within a single market. This suggests that price comparisons should be limited to those drug products where an exact match can be made. On the other hand, not all pack sizes may be available in all markets, which raises the possibility that a large amount of information may be lost if this is included in the matching criteria. In the PMPRB's view, matching by drug (as defined above) strikes a reasonable compromise between rigour and coverage. This is the approach taken to produce the international price comparisons reported in the main text.

All foreign-to-Canadian price comparisons are based on foreign prices encompassing foreign generic products only.12 Foreign generic products were identified by comparing the Local Brand Name and Molecule fields.

5 Currency Conversion

As noted above, the MIDAS sales data used in this report are expressed in local currencies. Sales were restated in Canadian dollars to allow for meaningful international price comparisons. Currency conversions were done using (1) annual average spot-market exchange rates (as reported by the Bank of Canada)13 and (2) purchasing power parities (as reported by the OECD).

6 Specific Reporting Elements: Statistical Methodology

The following subsections briefly describe the construction of the statistics provided in the report.

6.1 Concentration Statistics

The concentration statistics presented in Section 2 measure the extent to which markets are dominated by a small number of suppliers. In keeping with the PMPRB's general approach, concentration is first measured at the level of individual drugs. Overall statistics are then obtained by aggregating the results across all drugs. Aggregation is done by taking a simple average of drug-level results.

N-firm sales concentration ratios are commonly used in economic analysis. An N-firm ratio is calculated by first identifying the N suppliers with the largest sales in a given market, adding up their sales, and then calculating the proportion of overall market sales accounted for by the N-firm total.

Standard criticisms of N-firm ratios are that they say nothing about the extent of concentration among leading sellers, and that their construction entirely excludes the remainder of the market. The Herfindahl index represents an attempt to correct for these supposed deficiencies. To calculate the Herfindahl index one first calculates market shares for every manufacturer and then takes a sales-weighted average of these market shares. The resulting statistic is the sum of squared market shares taken across all suppliers.

6.2 International Price Comparisons

This report presents a variety of average foreign-to-Canadian price ratios, similar to those published in the PMPRB's Annual Report. These average ratios are constructed as Canadian-sales-weighted arithmetic averages of foreign-to-Canadian price ratios at the level of individual drug products. Algebraically, let

i = 1 … N, each number identifying a drug included in the calculation

p(i) = the Canadian price of drug i

pf(i) = the foreign price of drug i (restated in Canadian dollars)

w(i) = the proportion of Canadians' spending on the drugs 1 to N accounted for by drug i

Then the sales-weighted arithmetic mean of foreign-to-Canadian price ratios is given by:

RA = S w(i)[pf(i)/p(i)]

where S signifies summation over drug products 1 to N.

With currency conversion at market exchange rates, average ratios constructed this way have a simple intuitive interpretation: they indicate how much more or less Canadians would have paid for the medicines they purchased in a given period had they paid foreign prices.

Multilateral foreign-to-Canadian price comparisons are also presented using the following indicators as comparators: minimum foreign price, maximum foreign price, the simple mean of foreign prices, the weighted mean of foreign prices (with weights based on sales by country) and the median of foreign prices. Each of these indicators is an aggregate measure encompassing all countries where comparable drug prices could be calculated.

6 These controlling entities are identified in the MIDAS database's Corporation field.

7 Retail pharmacy purchases account for more than 85% of drug sales in Canada, based on 2004 results from IMS's Hospital and Pharmacy Audit./p>

8 Note that, under this approach, a drug's price can be interpreted as a sales-weighted average of the product prices it encompasses.

9 For a large majority of generic products, the ingredient name actually appears in MIDAS's brand-name field (along with a company identifier). Sometimes the brand-name field contains the ingredient name in a modified form, in which case the product was retained as a generic. Only products sold under names clearly different from that of their principal ingredient were eliminated.

10 Note that whether the product is sold by a “generic” or a “brand-name” manufacturer does not matter here. This means our list of generic products will include “pseudogenerics” sold by companies principally engaged in selling branded products.

11 Markets with a single generic supplier will be considered separately in a subsequent report.

12 Because generic versions of some drug products are not available in all markets, the number of matches will vary from comparator to comparator.

13 This is the Bank of Canada's annual average of the daily noon rates.

Appendix C: Drug Products Included in the Analysis