xxxxx

eytan
Print Friendly, PDF & Email

Richard Evans / Scott Hinds / Ryan Baum

203.901.1631 /.1632 / .1627

richard@ / hinds@ / baum@ssrllc.com

twitter.jpg @SecSovHealth

February 25, 2012

xxxxx

  • Mention that real pricing power erodes if CPI accels ….. also mention ex-US pricing pressures as EU GDP decels in ’12; mention that valuations imply many investors are in this for yieldxxxx

How fast have prices risen, and how much does it matter?

Strong real pricing gains have long been a driver of US pharma sales trend; however in light of recent volume deceleration (cyclical)[1] and longer-term mix erosion (secular)[2], pricing actions in the last half decade have accounted for more than 100 percent of real US sales growth (Exhibit 1). The 2.6% real price figure shown in Exhibit 1 is taken from the Bureau of Labor Statistics’ Rx PPI; this data series is very useful for long-term trend analysis, but has limitations vis-à-vis gauging the immediate real brand price trend. The index covers both brands and generics, and sampling is limited to first sales from US-based production facilities. A bottom-up approach to backing real price out of brand-only sales produces an estimate of recent real price growth closer to 6% (Exhibit 2), or roughly 2x faster than the Rx PPI

The point is often made that growing rebate obligations offset higher list prices. This is true, however the offset is only partial, i.e. rebates have not eliminated net pricing gains. Exhibit 3 compares growth in average US brand list price[3], rebates, and net prices from 2002 to 2010; average (list) brand dollars per prescription grew 7.4% (CAGR) in real terms, rebates grew 15%, and net dollars grew 6.5%. All in, the total effect of rebates was to reduce real pricing growth by only 84bp

Richard Evans / Scott Hinds / Ryan Baum

203.901.1631 /.1632 / .1627

richard@ / hinds@ / baum@ssrllc.com

twitter.jpg @SecSovHealth

February 25, 2012

xxxxx

  • Mention that real pricing power erodes if CPI accels ….. also mention ex-US pricing pressures as EU GDP decels in ’12; mention that valuations imply many investors are in this for yieldxxxx

How fast have prices risen, and how much does it matter?

Strong real pricing gains have long been a driver of US pharma sales trend; however in light of recent volume deceleration (cyclical)[1] and longer-term mix erosion (secular)[2], pricing actions in the last half decade have accounted for more than 100 percent of real US sales growth (Exhibit 1). The 2.6% real price figure shown in Exhibit 1 is taken from the Bureau of Labor Statistics’ Rx PPI; this data series is very useful for long-term trend analysis, but has limitations vis-à-vis gauging the immediate real brand price trend. The index covers both brands and generics, and sampling is limited to first sales from US-based production facilities. A bottom-up approach to backing real price out of brand-only sales produces an estimate of recent real price growth closer to 6% (Exhibit 2), or roughly 2x faster than the Rx PPI

The point is often made that growing rebate obligations offset higher list prices. This is true, however the offset is only partial, i.e. rebates have not eliminated net pricing gains. Exhibit 3 compares growth in average US brand list price[3], rebates, and net prices from 2002 to 2010; average (list) brand dollars per prescription grew 7.4% (CAGR) in real terms, rebates grew 15%, and net dollars grew 6.5%. All in, the total effect of rebates was to reduce real pricing growth by only 84bp

Over time, the cumulative effect of real pricing has been remarkable, accounting for just less than half of total US real sales growth since 1980 (Exhibit 4). However the industry relies on pricing not just for growth, but also for predictability and for longer-term returns on current R&D spending. As a largely discretionary variable pricing can be – and often has been – used to fill in periods of weak volume and/or mix growth. Where the per-capita volume trend is generally slow and predictable, the mix trend is not – on average mix is a very large part of changes in revenue, and is highly varied from one period to the next (Exhibit 1, again). Accordingly if real pricing power is lost, the industry not only sees slower sales growth (or outright decline), but also more volatility – i.e. less growth, and more risk. Elsewhere we’ve shown that real pricing gains on US small molecules account for roughly half of historic returns on R&D spending. If we assume projects currently in development enjoy historic rates of real price gain, we estimate that current returns on R&D spending are slightly (-7%) negative. More rationally, if we assume the industry loses some (or all) of its real pricing power, our estimate of returns on current R&D spending fall to as low as -57%[4]

Why can’t prices just keep rising at their historic pace?

Average out-of-pocket (OOP) spending on prescription drugs, as a percentage of total household spending, has been flat (and small) since at least 1965 (Exhibit 5). Total drug spending (OOP plus insurance) has risen rapidly; household spending has only remained flat because drug coverage has absorbed all of the real gains in total drug spend. For brand prices to continue inflating, either or a combination of two things must occur: households must spend more of their budgets on out-of-pocket drug costs; and/or plan sponsors must continue defending households from drug price inflation. We don’t think either of these essential preconditions to drug price inflation will occur, and by extension believe that drug pricing will slow

Average OOP drug expenses are quite low as a share of total household spending, but the average is not ‘typical’ – in fact the average is incredibly misleading. For example a very high percentage of total commercial insured drug spending falls on very few persons (or households) – more than 20% of spending is accounted for by just one percent of the (under 65) population, 70% of spending comes from just 10% of the population, and fully 95% of spending comes from one-quarter of the population (Exhibit 6). Translating this back into household spending terms, we see that 20% of (commercial) drug spending comes from households whose OOP drug spending is roughly the size of their total housing (rent or mortgage plus utilities) expense, and 70% of commercial drug spend comes from households spending as much OOP on drugs as on groceries (Exhibit 7). Given where OOP drug spending falls in the typical household budget, we believe that exposing households to even a constant (instead of falling) share of drug costs must lead either to falling unit demand, or falling real prices, for brand drugs

This leaves the question of whether plan sponsors will continue to absorb all or most of real price growth. We believe the answer is no, in very large part because of the emerging trend toward replacing co-pays (where patient OOP costs are a flat dollar fee for drugs on a certain formulary tier, irrespective of the drug’s underlying cost) with co-insurance (patient OOP costs calculated as a percent of retail drug price). Drug benefits with co-insurance provisions were offered by a third of employers in 2011, up from roughly 13% of employers just three years earlier (Exhibit 8). Importantly, employers offering benefit designs with co-insurance tend to be larger employers; this implies co-insurance is a strong ‘intentional’ and thus continuing trend, as opposed to a defensive reaction by less financially secure employers (Exhibit 9). To be clear, these data tell us what percent of employers offer co-insurance, but not whether they offer plans with co-pays as an alternative, so we can’t imply that a third of employees have co- insurance in their drug benefit designs. Nor do these data tell us if all of a drug benefit’s tiers are co-insurance rather than co-

pay; in some cases we know benefit designs are a mix of both flat-dollar co-pays and co-insurance. Nevertheless we think it’s far more likely that co-insurance is an emerging standard than a benefit design that either stalls at low market share or goes away entirely, and accordingly believe the logical effects of co-insurance on real drug pricing are likely to play out. There are two such effects: deceleration of average real list price increases, and falling real prices for drugs where reasonable substitutes are available at lower prices

The tendency of co-insurance (relative to fixed dollar co-pay) to decelerate real pricing is fairly obvious – if co-insurance percentages remain relatively constant, and if real drug prices inflate more rapidly than fixed dollar co-pays, then OOP expenses for patients with co-insurance grow at the rate of real drug pricing – which is considerably faster than the rate of growth in fixed-dollar co-pays. Comparing average fixed dollar co-pay amounts to the product of average retail brand pricing and average co-insurance rates since 2000, we see that OOP drug expense has grown faster under co-insurance than with traditional fixed dollar co-pays (Exhibit 10). If co-insurance becomes the standard, real changes in drug price quickly show through as proportionally higher OOP expenses for patients, with falling unit demand (or a tempering of price growth) as the probable result

Less obvious but similarly straightforward is the likely effect of co-insurance on brand pricing within therapeutic categories. As a patient with a fixed dollar co-payment, I’m indifferent to the list price of drugs that are on the same formulary tier. However as a patient with co-insurance, I’m immediately sensitive to any difference in price; and, to the extent that a reasonable substitute is available for less out-of-pocket cost, I may prefer the lower priced option. This leads directly to the question of whether brand prices are diverse within therapeutic categories, and we find that they are. About 35% of brand drug sales come from therapeutic classes in which buying a cheaper brand can save a patient at least $10 OOP as compared to buying the most expensive brand in the class, assuming a co-insurance rate (about 24%) typical of ‘preferred tier’ status. Using the higher co-insurance rate (about 39%) typical of non-preferred status, about half of brand sales are in therapeutic categories having a brand that will cost patients at least $10 less OOP than the most expensive brand in the class. Same concept, slightly different angle: about 13% of brand sales are in classes where a cheaper brand saves consumers $10 or more OOP relative to the market share leader in that class (assuming preferred rate of co-insurance); or, assuming the non-preferred rate of co-insurance about 15% of sales are in classes where a cheaper option saves the consumer $10 or more OOP relative to the class share leader (Exhibit 11; and, please see the Appendices for a visual summary of price and market share dispersion by major drug categories). We believe this non-trivial level of in-class price dispersion will narrow if and when co-insurance becomes more predominant; and we believe that the dispersion falls to the low end of the range (people on higher priced drugs move to lower priced drugs) rather than simply collapsing to the average

Are threats to real pricing reflected in valuations, and are some companies more or less at risk than others?

Large pharma companies’ forward EPS estimates reflect slow growth (Exhibit 12); however this is almost entirely the result of known patent losses (i.e. losses of exclusivity or ‘LOE’s’). As our timeframe extends to 2014, expected EPS growth rates normalize, as do relative forward multiples v. both cap-weighted Healthcare and the SP500 (Exhibit 12, again). Because LOE’s appear to fully account for low near-term growth expectations, because companies and time periods without major LOE exposure imply future EPS growth in-line with trailing growth; and, because forward multiples suggest the market accepts forward consensus as generally credible, we believe valuations fail to reflect risks to US real pricing power

For the typical large pharma company US sales are a very large percentage of global sales (Exhibit 13) and generate an even larger share of global profits; and, over the last several years US sales have grown only because of real pricing (Exhibit 14). It follows that deceleration – or loss – of US real pricing power might reasonably be expected to eliminate earnings growth, or even to reduce earnings outright

If real pricing decelerates we believe all of the large caps are at risk of being negatively re-rated; however some companies appear more at risk than others. We believe relative risk among the large caps is a function of five variables: reliance on US pricing, upcoming LOE’s as a percentage of current sales, the ratio of common dividends to gross profits, the extent to which R&D can be trimmed as an offset to lost pricing power; and, current valuation. Exhibit 15 shows each of these variables in relative (to peers) form; Exhibit 16 compares the companies across these five dimensions graphically

In both Exhibits 15 and 16 values are relative to peers; higher values are less desirable and lower values are more desirable. Exhibit 16 shows values = to 1.00 as a black line, and each company’s specific relative values in green. Visually, the more a company’s relative values stay inside the black reference line, the less share price risk we believe that company faces if real pricing power fades. Companies with higher trailing rates of US pricing and higher percentages of global sales from the US arguably are more reliant on pricing power; we use the product of US real pricing and US as a % of WW sales as an index of reliance on US real pricing. Companies with greater LOE exposure already face gross margin pressures in the near- to mid-term; because we believe an overlap of a pricing slowdown and LOE means a compounding of gross margin pressures, we view companies with greater LOE exposure as more at risk. The ratio of common dividends to gross profits helps us gauge the relative probability of companies being able to maintain dividends if US real pricing power – a huge contributor to gross profits – is diminished. Companies with higher R&D / sales ratios have more immediately available cost savings than peers with lower R&D / sales ratios, and so are better able to offset the EPS effect of reduced pricing power. Relative forward PE’s simply help us gauge whether companies with greater or less real pricing risk are more or less subject to multiple contraction

PFE, MRK, and NVS are average on almost every measure, though PFE’s pricing actions are fairly rapid (1.16). Roche, SNY, JNJ, GSK and ABT are relatively less at risk; Roche and GSK have relatively high dividend to gross profit ratios, but are less reliant on price or exposed to LOE’s than peers, and so seem to be relatively safe. AZN, BMY, and LLY face relatively greater risk; all are particularly reliant on rapid real price gains, and all face relatively high near- to mid-term LOE exposure, particularly BMY

Exh 16: Comparison of metrics that may indicate exposure to real pricing risks

In all cases relative refers to large cap peers; Rate of US pricing is measured across the period ’07-’11; LOE exposure is measured as the cumulative % of current sales expected to lose US patent protection by 2014; Ratio of div / gross profit is simply common dividends divided by gross profit (2011); ‘Availability of R&D offsets’ is simply 1 – R&D as a % of sales; ‘fPE’ is the ratio of current share price to consensus EPS estimate for 2014

Sources: BLS; Company filings; FactSet; FDA Orange Book; IMS Health; S&P CapIQ; Wolters Kluwer Price Rx; SSR analysis and assumptions

  1. ”US Healthcare Demand Slow for Cyclical (i.e. Temporary) Reasons…”, January 12, 2012, Sector & Sovereign Research LLC
  2. “Below Zero and Falling Fast: R&D Productivity as an Enterprise-Wide Crisis”, October 11, 2011, Sector & Sovereign

    Research LLC

  3. Growth in average US brand list price is driven both by list pricing ‘actions’ and by changes to product mix, i.e. because mix effects are included, average list price growth values in this exhibit are greater than the rate at which manufacturers have increased real list prices
  4. Ibid 2.

Over time, the cumulative effect of real pricing has been remarkable, accounting for just less than half of total US real sales growth since 1980 (Exhibit 4). However the industry relies on pricing not just for growth, but also for predictability and for longer-term returns on current R&D spending. As a largely discretionary variable pricing can be – and often has been – used to fill in periods of weak volume and/or mix growth. Where the per-capita volume trend is generally slow and predictable, the mix trend is not – on average mix is a very large part of changes in revenue, and is highly varied from one period to the next (Exhibit 1, again). Accordingly if real pricing power is lost, the industry not only sees slower sales growth (or outright decline), but also more volatility – i.e. less growth, and more risk. Elsewhere we’ve shown that real pricing gains on US small molecules account for roughly half of historic returns on R&D spending. If we assume projects currently in development enjoy historic rates of real price gain, we estimate that current returns on R&D spending are slightly (-7%) negative. More rationally, if we assume the industry loses some (or all) of its real pricing power, our estimate of returns on current R&D spending fall to as low as -57%[4]

Why can’t prices just keep rising at their historic pace?

Average out-of-pocket (OOP) spending on prescription drugs, as a percentage of total household spending, has been flat (and small) since at least 1965 (Exhibit 5). Total drug spending (OOP plus insurance) has risen rapidly; household spending has only remained flat because drug coverage has absorbed all of the real gains in total drug spend. For brand prices to continue inflating, either or a combination of two things must occur: households must spend more of their budgets on out-of-pocket drug costs; and/or plan sponsors must continue defending households from drug price inflation. We don’t think either of these essential preconditions to drug price inflation will occur, and by extension believe that drug pricing will slow

Average OOP drug expenses are quite low as a share of total household spending, but the average is not ‘typical’ – in fact the average is incredibly misleading. For example a very high percentage of total commercial insured drug spending falls on very few persons (or households) – more than 20% of spending is accounted for by just one percent of the (under 65) population, 70% of spending comes from just 10% of the population, and fully 95% of spending comes from one-quarter of the population (Exhibit 6). Translating this back into household spending terms, we see that 20% of (commercial) drug spending comes from households whose OOP drug spending is roughly the size of their total housing (rent or mortgage plus utilities) expense, and 70% of commercial drug spend comes from households spending as much OOP on drugs as on groceries (Exhibit 7). Given where OOP drug spending falls in the typical household budget, we believe that exposing households to even a constant (instead of falling) share of drug costs must lead either to falling unit demand, or falling real prices, for brand drugs

This leaves the question of whether plan sponsors will continue to absorb all or most of real price growth. We believe the answer is no, in very large part because of the emerging trend toward replacing co-pays (where patient OOP costs are a flat dollar fee for drugs on a certain formulary tier, irrespective of the drug’s underlying cost) with co-insurance (patient OOP costs calculated as a percent of retail drug price). Drug benefits with co-insurance provisions were offered by a third of employers in 2011, up from roughly 13% of employers just three years earlier (Exhibit 8). Importantly, employers offering benefit designs with co-insurance tend to be larger employers; this implies co-insurance is a strong ‘intentional’ and thus continuing trend, as opposed to a defensive reaction by less financially secure employers (Exhibit 9). To be clear, these data tell us what percent of employers offer co-insurance, but not whether they offer plans with co-pays as an alternative, so we can’t imply that a third of employees have co- insurance in their drug benefit designs. Nor do these data tell us if all of a drug benefit’s tiers are co-insurance rather than co- pay; in some cases we know benefit designs are a mix of both flat-dollar co-pays and co-insurance. Nevertheless we think it’s far more likely that co-insurance is an emerging standard than a benefit design that either stalls at low market share or goes away entirely, and accordingly believe the logical effects of co-insurance on real drug pricing are likely to play out. There are two such effects: deceleration of average real list price increases, and falling real prices for drugs where reasonable substitutes are available at lower prices

The tendency of co-insurance (relative to fixed dollar co-pay) to decelerate real pricing is fairly obvious – if co-insurance percentages remain relatively constant, and if real drug prices inflate more rapidly than fixed dollar co-pays, then OOP expenses for patients with co-insurance grow at the rate of real drug pricing – which is considerably faster than the rate of growth in fixed-dollar co-pays. Comparing average fixed dollar co-pay amounts to the product of average retail brand pricing and average co-insurance rates since 2000, we see that OOP drug expense has grown faster under co-insurance than with traditional fixed dollar co-pays (Exhibit 10). If co-insurance becomes the standard, real changes in drug price quickly show through as proportionally higher OOP expenses for patients, with falling unit demand (or a tempering of price growth) as the probable result

Less obvious but similarly straightforward is the likely effect of co-insurance on brand pricing within therapeutic categories. As a patient with a fixed dollar co-payment, I’m indifferent to the list price of drugs that are on the same formulary tier. However as a patient with co-insurance, I’m immediately sensitive to any difference in price; and, to the extent that a reasonable substitute is available for less out-of-pocket cost, I may prefer the lower priced option. This leads directly to the question of whether brand prices are diverse within therapeutic categories, and we find that they are. About 35% of brand drug sales come from therapeutic classes in which buying a cheaper brand can save a patient at least $10 OOP as compared to buying the most expensive brand in the class, assuming a co-insurance rate (about 24%) typical of ‘preferred tier’ status. Using the higher co-insurance rate (about 39%) typical of non-preferred status, about half of brand sales are in therapeutic categories having a brand that will cost patients at least $10 less OOP than the most expensive brand in the class. Same concept, slightly different angle: about 13% of brand sales are in classes where a cheaper brand saves consumers $10 or more OOP relative to the market share leader in that class (assuming preferred rate of co-insurance); or, assuming the non-preferred rate of co-insurance about 15% of sales are in classes where a cheaper option saves the consumer $10 or more OOP relative to the class share leader (Exhibit 11; and, please see the Appendices for a visual summary of price and market share dispersion by major drug categories). We believe this non-trivial level of in-class price dispersion will narrow if and when co-insurance becomes more predominant; and we believe that the dispersion falls to the low end of the range (people on higher priced drugs move to lower priced drugs) rather than simply collapsing to the average

Are threats to real pricing reflected in valuations, and are some companies more or less at risk than others?

Large pharma companies’ forward EPS estimates reflect slow growth (Exhibit 12); however this is almost entirely the result of known patent losses (i.e. losses of exclusivity or ‘LOE’s’). As our timeframe extends to 2014, expected EPS growth rates normalize, as do relative forward multiples v. both cap-weighted Healthcare and the SP500 (Exhibit 12, again). Because LOE’s appear to fully account for low near-term growth expectations, because companies and time periods without major LOE exposure imply future EPS growth in-line with trailing growth; and, because forward multiples suggest the market accepts forward consensus as generally credible, we believe valuations fail to reflect risks to US real pricing power

For the typical large pharma company US sales are a very large percentage of global sales (Exhibit 13) and generate an even larger share of global profits; and, over the last several years US sales have grown only because of real pricing (Exhibit 14). It follows that deceleration – or loss – of US real pricing power might reasonably be expected to eliminate earnings growth, or even to reduce earnings outright

If real pricing decelerates we believe all of the large caps are at risk of being negatively re-rated; however some companies appear more at risk than others. We believe relative risk among the large caps is a function of five variables: reliance on US pricing, upcoming LOE’s as a percentage of current sales, the ratio of common dividends to gross profits, the extent to which R&D can be trimmed as an offset to lost pricing power; and, current valuation. Exhibit 15 shows each of these variables in relative (to peers) form; Exhibit 16 compares the companies across these five dimensions graphically

In both Exhibits 15 and 16 values are relative to peers; higher values are less desirable and lower values are more desirable. Exhibit 16 shows values = to 1.00 as a black line, and each company’s specific relative values in green. Visually, the more a company’s relative values stay inside the black reference line, the less share price risk we believe that company faces if real pricing power fades. Companies with higher trailing rates of US pricing and higher percentages of global sales from the US arguably are more reliant on pricing power; we use the product of US real pricing and US as a % of WW sales as an index of reliance on US real pricing. Companies with greater LOE exposure already face gross margin pressures in the near- to mid-term; because we believe an overlap of a pricing slowdown and LOE means a compounding of gross margin pressures, we view companies with greater LOE exposure as more at risk. The ratio of common dividends to gross profits helps us gauge the relative probability of companies being able to maintain dividends if US real pricing power – a huge contributor to gross profits – is diminished. Companies with higher R&D / sales ratios have more immediately available cost savings than peers with lower R&D / sales ratios, and so are better able to offset the EPS effect of reduced pricing power. Relative forward PE’s simply help us gauge whether companies with greater or less real pricing risk are more or less subject to multiple contraction

PFE, MRK, and NVS are average on almost every measure, though PFE’s pricing actions are fairly rapid (1.16). Roche, SNY, JNJ, GSK and ABT are relatively less at risk; Roche and GSK have relatively high dividend to gross profit ratios, but are less reliant on price or exposed to LOE’s than peers, and so seem to be relatively safe. AZN, BMY, and LLY face relatively greater risk; all are particularly reliant on rapid real price gains, and all face relatively high near- to mid-term LOE exposure, particularly BMY

In all cases relative refers to large cap peers; Rate of US pricing is measured across the period ’07-’11; LOE exposure is measured as the cumulative % of current sales expected to lose US patent protection by 2014; Ratio of div / gross profit is simply common dividends divided by gross profit (2011); ‘Availability of R&D offsets’ is simply 1 – R&D as a % of sales; ‘fPE’ is the ratio of current share price to consensus EPS estimate for 2014

Sources: BLS; Company filings; FactSet; FDA Orange Book; IMS Health; S&P CapIQ; Wolters Kluwer Price Rx; SSR analysis and assumptions

  1. ”US Healthcare Demand Slow for Cyclical (i.e. Temporary) Reasons…”, January 12, 2012, Sector & Sovereign Research LLC
  2. “Below Zero and Falling Fast: R&D Productivity as an Enterprise-Wide Crisis”, October 11, 2011, Sector & Sovereign

    Research LLC

  3. Growth in average US brand list price is driven both by list pricing ‘actions’ and by changes to product mix, i.e. because mix effects are included, average list price growth values in this exhibit are greater than the rate at which manufacturers have increased real list prices
  4. Ibid 2.
Print Friendly, PDF & Email