Purchase Commitments for Vaccines: Their Uses and Their Limitations

1
Purchase Commitments for VaccinesTheir Uses and
Their Limitations

• Andrew Farlow
• University of Oxford
• Department of Economics, and Oriel College
• BioDesign Institute
• Arizona State University
• 5 April 2005
• This version 16 May 2005
• This has been rewritten so as to be a stand-alone
  presentation
• without need of a presenter.
• A downloadable copy of this PowerPoint and
  supporting papers can be found at
• www.economics.ox.ac.uk/members/andrew.farlow
• The author very much welcomes feedback
  andrew.farlow_at_economics.ox.ac.uk

2
Based on

• The Global HIV Vaccine Enterprise, Malaria
  Vaccines, and Purchase Commitments What is the
  Fit?
• Andrew Farlow, Submission to Commission on
  Intellectual Property Rights, Innovation and
  Public Health, WHO, March 2005.
• Forthcoming in Innovation Strategy Today.
• An Analysis of the Problems of RD Finance for
  Vaccines And an Appraisal of Advance Purchase
  Commitments
• Andrew Farlow, April 2004.

3
The problem

• Over 40,000 people many of them children die
  every day in developing countries of infectious
  or parasitic diseases.
• Many could be saved by access to already
  developed vaccines and drugs
• A large proportion of the disease burden in
  such countries is unnecessary, since it could be
  reduced by the effective distribution of
  medicines that are currently available and
  inexpensive. International Policy Network
  Incentivising research and development for the
  diseases of poverty 2005 p17.
• Barely more than 1 of total global
  pharmaceutical expenditure goes into the research
  and development of new products for diseases
  affecting 90 of the worlds population.
• 10-15 of global pharmaceutical spending goes
  into RD, and barely 10 of this goes into
  diseases impacting 90 of the worlds population.

4
Recent strides

• Large fresh funds to purchase currently existing
  vaccines and to roll out immunisation programs
• The UK 1.8bn over 15 years
• Bill and Melinda Gates foundation 750m
• Norway 290m.
• Proposed 4bn budget over ten years for
  Immunizations, via an International Financing
  Facility for Immunizations
• Though this budget and immunization initiative
  could, and should, have been provided with or
  without the IFF possibility.
• Launch of Global HIV Vaccine Enterprise in
  2004.
• UK presidency of the G8 and EU this year
• Big opportunityor lost opportunity?
• Given the potentially controversial nature of
  the IFF proposal, policy makers also need to take
  great care that, in all their public
  pronouncements, the IFF is seen to the there to
  support the immunizations and never the other way
  around.

5
A spectrum of vaccines

• Low or non-use of many already existing, already
  cheap or even practically costless vaccines.
• Late-stage vaccines where most of the science
  is already known and a viable product is close
  to, or already at, hand.
• Early-stage highly complex and difficult
  vaccines such as those for HIV, malaria, and TB
  where there are either no viable vaccines on
  the horizon or current candidates fall well short
  of 100 effectiveness, and many scientific
  difficulties remain.
• Spectrum of vaccines often lumped together.
• But, instruments needed for each case are
  different.
• Nature of purchase commitments for each kind of
  vaccine also differs.

6
The role of purchase commitments

• Long-term purchase contracts/commitments very
  advantageous for underused and late-stage
  vaccines.
• Advance Purchase Commitments (APCs) for
  early-stage vaccines here argued much weaker and
  a great deal more problematic than is often
  suggested
• The phrase Advance Purchase Commitment, APC, is
  used here, and not Advance Market, since the
  latter needs to be proven and not prejudged to be
  the case by the choice of language used. An
  inability of APCs to perform as if a market is
  central to many of the concerns here.
• For early-stage vaccine RD, manufacture and
  access may even be harmed by the presence of
  prior precommitments (certainly as currently
  proposed) compared to alternative mechanisms.
• Where does the boundary between the cases fall?

7
Lessons from vaccine introductions 1

• Past vaccine introductions
• Hepatitis B
• Haemophilus influenzae type B (Hib)
• Smallpox.
• Recent purchase arrangements
• African trivalent meningitis vaccine
• Meningitis conjugate C.
• Future late-stage vaccines
• Pneumococcus
• Rotavirus
• Issues
• Cost of manufacture
• Safety issues
• Epidemiology.

All these case-studies are treated in detail in
Farlow, March 2005, Section 3.
8
Lessons from vaccine introductions 2

• None of these case-studies remotely matches
  anything being proposed for HIV, malaria, and TB.
• Many of the problems caused/resolved by contracts
  are very different for these case-studies
  compared to early-stage vaccines.
• Case-studies illustrate current faults in need of
  rectification.
• Current short-run contracts are inefficient a
  stable long-term market matters.
• Get rid of market risk
• The need for distribution commitments,
  vaccine/health infrastructure commitments and
  commitments to tackle market risk at many levels
• APCs (as currently designed) for early-stage
  vaccines put market risk back on to developers!

9
Lessons from vaccine introductions 3

• Good information on how to efficiently set terms
• Information extracted through competitive
  tenders, etc.
• Relatively easy to make the incentive
  additional via procurement contracts, etc.
• Commitments as coordination devices.
• In practical cases, the breakthrough was through
  lowering production costs
• Incentives/competition for this?
• Technological shifts dependent on access to
  technology, IP, know-how, especially at
  manufacture and distribution stage
• Volume and regulatory issues important in
  lowering costs.

10
Lessons from vaccine introductions 4

• Incentives to install capacity quickly and for
  use quickly.
• Product differentiation and correcting vaccine
  market distortions.
• Ability to use IP in ways to encourage
  competition, to keep the market open to many
  potential developers and producers, and to help
  create cheaper-to-produce vaccines.
• Wider finance in place for a wider set of
  players
• More open to those who cannot draw off deep
  pocket finance for long periods.

11
Lessons from vaccine introductions 5

• Importance of role of developing/emerging country
  developers and manufacturers.
• More ability to share information and collaborate
  (key for HIV? Malaria? TB?).
• Relatively low levels of capital costs
  (compared to, e.g. case of HIV vaccines).
• Lower risk to biotechs.
• Many of these reasons are fungible they apply
  whatever the source of finance.
• Not committee-driven over long horizons.
• Current purchases do matter a lot.

12
Hepatitis B case-study

• Hepatitis B case-study in draft versions of the
  recent CGD report dropped from final report.
• Hepatitis B case does not support the reports
  underlying hypotheses for HIV/malaria/TB
• The original Hepatitis B vaccine developers were
  not the ones who developed and maintained the
  lower price market
• The competitive situation for hepatitis B today
  a key component in achieving long-term
  sustainable low prices reflects poorly on the
  lack of competition at a similar stage in product
  life cycle in the CGD report
• Emphasis in the success of the Hepatitis B case
  on market and competitive devices to push
  production prices lower compared to
  insufficient emphasis in CGD report.

13
Why the sudden interest in HIV, malaria, and
tuberculosis? 1

• Focus of attention increasingly on speculative,
  experimental applications to HIV, malaria, and
  tuberculosis, even though APCs never been used
  before for even the most basic of applications.
• Less than a year ago these difficult vaccines
  were not deemed likely doable by this approach
• Including by many now heavily advocating the
  approach.
• By definition APCs cannot be tested except
  through trying
• Should crash test the thinking, subjecting it
  to the harshest of possible self-critiques before
  trying
• Given the risks of abandoning the approach due to
  lack of industry response and the constant need
  to change the program, should learn by trying
  less complicated vaccines first and building up
  to more complicated vaccines.

14
Why the sudden interest in HIV, malaria, and
tuberculosis? 2

• Supporters and critics all concerned with
  incentives and rewards for private firm
  involvement.
• Disagreement is about
• The shape and timing of incentives and rewards
• The ability to set APC terms remotely
  efficiently
• Whether APCs will actually work as proposed
• The role of, and interaction with, other parts of
  the overall mechanism for developing these
  early-stage vaccines
• Order of priority, given the burden APCs place on
  the system capacity' of GAVI/VF/WHO, and use of
  political capital
• The dangers of perverse results.
• BIG CONCERN Approach feeds off (and also feeds)
  growing budgetary pressures to cut Vaccine RD
  funding especially for HIV given global
  fiscal deterioration.

15
Literature on this as an RD incentive for
early-stage vaccines

• Making Markets for Vaccines (henceforth MM in
  all references), Center for Global Development
  (henceforth CGD), Washington, D.C., April, 2005.
• Strong Medicine Creating Incentives for
  Pharmaceutical Research on Neglected Diseases,
  Kremer, M, and Glennerster, R, Princeton
  University Press, November, 2004.
• UKs No. 10 Policy Unit 1998-2001.
• Mostly the work of a small handful of authors.

16
Benchmark as an RD incentive for early-stage
vaccines 1

• (These details taken from MM)
• Legally binding contract before vaccine RD
• Sponsor(s) and all actual and potential vaccine
  developers sign-on to the contract within 36
  months of the initiation of the program
• All actual and potential developers agree to be
  monitored by the committee controlling the
  program
• Later entry of developers policed by the
  committee
• Those conducting current vaccine trials and
  failing to sign-on, and those initiating future
  vaccine trials without prior permission from the
  committee, are barred access to the eligible
  markets controlled by the committee
• Sponsors have an opt-out if contract fails to
  stimulate enough research (though current
  status of opt-out is a little unclear).

17
Benchmark as an RD incentive 2

• ISSUE Inability to identify all developers in
  advance.
• ISSUE Highly complex and evolving vaccine
  development process that is also moving
  increasingly toward emerging/developing country
  developers.
• How to avoid biasing early-stage APCs too early
  in favor of large developed-country companies,
  stymieing this evolution, and forcing later
  entrants to work through current large
  multinationals?
• Opt-outs and sunset clauses hard to incorporate
  without feeding back to harm RD incentives.

18
Benchmark as an RD incentive 3

• Sponsor(s) commit 3bn-10bn per disease
• Figure keeps falling. Now 3bn each for HIV,
  malaria, and TB (This presentation sticks to
  original figures for now).
• For the purchase of a vaccine or vaccines in a
  pre-agreed quantity (200-300 million treatments)
• Figure keeps falling. Now 200 million each for
  HIV, malaria, and TB.
• Benchmark has changed over last year. Prior to
  May 2004 a simple flat subsidy on each of the
  first X million of a vaccine.
• Now a complicated subsidy spread over first X
  million of several possible vaccines
• Rules for this unclear?and un-write-able?
• But this is largely a change in language
• Recent malaria announcements seem to target one
  vaccine.

19
Benchmark as an RD incentive 4

• To aid credibility, sponsors relinquish control
  of their funding to a committee with
  discretionary powers.
• Supposedly (since extremely difficult, if not
  impossible, to do in any practical sense for
  vaccines such as HIV, malaria, and TB) the size
  of (and distribution of) funding for the first
  200m high-cost treatments over developers is set
  precisely high enough to re-create the precise
  size of additional market needed to encourage
  the entry of the precise amount of venture
  capital and stock market finance needed for the
  remaining research and development including
  capital cost (though remaining is unclear)
  needed to produce a high quality vaccine or
  series of vaccines.
• A complex expected subsidy pattern across
  developers and over time (investors
  expectations of this are key).

20
Benchmark as an RD incentive 5

• All RD costs repaid through the purchase of a
  successful vaccine or (since May 2004) several
  vaccines in a particular period in time (if there
  are several meeting eligibility conditions in
  any period of time), or series of vaccines over
  time (to combat resistance perhaps and to give
  incentives for follow-on innovation), and only
  the successful vaccine(s) or series of vaccines.
• This program (supposedly) funds additional
  eligible market purchases only
• Eligible and non-eligible markets are
  separatedsomehow.
• Repayment of RD costs is from taxpayers of
  richer countries, foundations such as the Bill
  and Melinda Gates Foundation, and through
  co-payments made by developing countries
  themselves (that may come from third parties)
• The program is foundation- and publicly-funded.

21
Benchmark as an RD incentive 6

• NOTE Overall cost of vaccine development should
  include all funding needed outside of the
  program, including subsidies, tax-breaks, and
  other benefits granted for research, and the
  spending of national governments and foundations,
  and any costs of vaccine enterprises.
• An APC for HIV is likely to cover only a very
  small portion of the overall costs of HIV vaccine
  development.
• This is ignored (so far) in CGD cost
  effectiveness calculations for early-stage
  vaccines
• All DALYs saved are apportioned to the purchase
  commitment even if it will represent only a small
  portion of the overall cost of development of an
  HIV vaccine.

22
What the winner gets

• (Supposedly) winner(s) repaid all of the
  privately-funded (and only the privately-funded)
  RD costs (including all capital costs) of all
  firms (both the successful and the unsuccessful)
  and only the private firms, who used such private
  funding on RD towards the vaccine since the time
  the purchase commitment had been announced (and
  only since the announcement) and only for
  eligible markets covered by the mechanism.
• Capital costs refers to the costs of the
  finance used, and includes the required return to
  cover all risk being borne, including any risk
  created by the mechanism itself (i.e it does not
  refer to physical real capital investment).
• Very different pricing strategies in eligible
  and non-eligible markets.
• Firm gets all IP to the vaccine (under current
  proposal).

23
Problems with underlying model 1

• Underlying (Kremer Appendix 3) model driving the
  logic for early-stage vaccines is highly
  simplistic. The critique here is that highly
  idealized perfectly-functioning APC models are
  contrasted with highly imperfect alternatives
  thus generating unfair comparisons, and not that
  alternatives are not themselves highly imperfect
  too.
• The science is fixed, simple, constant, static,
  linear
• Extremely simple probability structure.
• No patents on anything other than end vaccine
  products
• No financial constraints, investment hold-ups,
  strategic behaviors, constraints on flows of
  information, or concentrations of market power
  based on IP ownership
• CASE When the Malaria Venture Initiative (MVI)
  mapped the patent status of the MSP-1 antigen,
  it found 39 different families of patents with
  monopoly scope impinging on it
• No notion of near market or near scratch
  developers.

24
Problems with underlying model 2

• No benefits in sharing information across vaccine
  developers. No know-how monopoly
• Not good for describing projects involving
  science with lots of feedback loops,
  collaboration and the sharing of information
  (HIV, malaria, and tuberculosis vaccine
  research)
• Lots of incentive to hoard information
• CASE Existing developed economy patent holders,
  facing a potentially emerging-economy competitor,
  can exploit secret know-how (as well as more
  general technical know-how, and undisclosed test
  or other data), including refusing to contract to
  transfer necessary know-how, thus creating a
  barrier to entry. Given the mechanism for
  distributing payments, there is a strong
  incentive to hold out under APCs
• Lack of know-how (extremely important for
  biological products) makes many disciplining
  threats non-credible (e.g. compulsory licenses if
  vaccine developers refuse to supply).

25
Problems with underlying model 3

• No variation in the probabilities of discovery
  over the vaccine development process
• No easy or difficult stretches of science.
• No ways for technology to improve or deteriorate
  over time
• No technology shocks,scientific breakthroughs
  or deteriorations
• No need to incentivize such breakthroughs.
• No sunk costs.
• No large incumbent firms instead perfect
  competition everywhere and always.
• No strategic behaviour of any sort, and of any
  firm, based on sunk costs, patent ownership,
  finance, or any other real-world factor.

26
Problems with underlying model 4

• Extremely good understanding of the state of
  current and future (extremely simple) science.
• No coordination problems across public and
  private sectors in their research decisions at a
  single point in time and over time.
• No coordination problems across public and
  private sectors and all countries in their
  vaccine purchase decisions and in their provision
  of vaccine delivery systems.
• An idealised, non-cyclical, set of financial
  markets.
• No pipelines of products, no problems with
  vaccine resistance.
• No composite vaccines, and no therapeutic
  vaccines.
• Wide range of delivery issues ignored.
• APCs need to be designed to handle/avoid these
  issues.

27
Some very rough HIV figures 1

• 10 firms put in equal effort on an early-stage
  HIV vaccine (we maintain the fiction of
  competition for now that the program encourages
  competition needs to be proved).
• Presume this is the optimal number of firms (we
  cant).
• Expected 70 of capital costs (a guess - no
  figures releasedbut presumed high for an HIV
  APC).
• Presume one firm wins (supposedly, several
  could).
• 6.25bn (pre-April 2005 figures) goes to a firm
  having spent, in present discounted (2005) terms,
  less than 200m, on private out-of-pocket
  research costs (this is a figure for purely
  illustrative purposes).
• This is the efficient and fair outcome and not
  being critiqued hereBut it does create problems
  for firms and the committee running the program,
  as we will see later.

28
Some very rough HIV figures 2

• With no crowding out, the 6.25bn pays for
• 1.875bn of out-of-pocket HIV RD costs across
  all firms
• 4.375bn of capital (i.e. finance) costs.
• With 50 crowding out (explained below) and
  other inefficiencies, the 6.25bn would pay for
• about 900m of new out-of-pocket research costs
• about 9 months worth of what those working on
  the Global HIV Vaccine Enterprise say is actually
  needed.
• The most likely response of firms no response
  at all?
• Again, these figures are very rough, and for
  illustrative purposes only.

29
Some very rough HIV figures 3

• We can look at HIV from another angle.
• HIV vaccines likely to take a minimum of 15 years
  to develop.
• 1.2bn per year of out-of-pocket research and
  trial costs needed (IAVA 2004), i.e. double the
  current level.
• Replacing this flow for 15 years with an APC at
  the end, would cost
• 85bn (if required nominal rate of return 20)
• 130bn (if required nominal rate of return 25)
• Uncertainty about ever getting a vaccine is
  embedded in capital costs
• Crowding out would make the figures worse
• Reputational risk would make the figures worse.
  
• These are low rates of return by venture capital
  standards, and exclude the (still likely very
  high) costs after 15 years.

30
Some very rough HIV figures 4

• So, where does the MM figure of 3bn come from?
• Where does the notion of multiple developers come
  from?
• Need for a mega-blockbuster if using APC route
  for HIV?
• Maybe this is why private firms currently spend
  so very little on HIV vaccine research in spite
  of there being a sizeable extant market for some
  clades of HIV?
• How large are politicians prepared to make APC
  funds for HIV vaccines?
• Are they prepared to massively top up later?
• Is it realistic to believe that funding levels
  for such programs for HIV will be set high enough
  with no pressure to readjust firm payoffs down
  later?
• All MM cost effectiveness figures are worked out
  on basis of the 3bn and not on the basis of
  these much larger figures for HIV.

31
Some very rough figures for firms

• For a vaccine costing 25 per course of
  treatment, the best-case scenario (no crowding
  out, but high capital costs) is
• 1-2 for production and distribution
• 6-7 for private out-of-pocket RD costs
• 16-18 for the cost of finance.
• With 50 crowding out
• About 3 for new private out-of-pocket RD costs.
  
• But it is not clear that an HIV vaccine could be
  manufactured for a dollar or so (especially in
  early days)
• Previous experiences with vaccine introductions
  suggest problems
• Too little manufacturing competition to drive
  prices that low
• IP held in too few hands
• Ex ante worries that this will be the case, will
  undermine incentives to do RD in the first place
  (More on this below).

32
Figures for currently existing and late-stage
vaccines

• The above are very, very rough figures, since
  paucity of information is such that we really do
  not have much of a handle on these issues.
• We can say, however, that the above proportions
  are completely the converse for currently
  existing vaccines, and, indeed, for many
  late-stage vaccines
• Much lower capital costs because of much lower
  risk, especially risks of the mechanism itself
• No crowding out (because of the ability to use
  competitive tenders and other separation
  devices)
• Much more easy to set efficient terms (because of
  competitive tenders and other devices to reveal
  information, and good information on technology,
  etc.).

33
Impossible to set size efficiently

• Each APC should be set commensurate with the
  difficulty of the underlying science and the cost
  of the RD of developing the vaccine at hand.
• MM suggests 3bn per disease for HIV, malaria,
  and TB. Setting this right is a crucial detail
  (MM April 2005)
• For no obvious reason, the figure is much lower
  than in draft versions of the report (of even
  just a few months ago)
• Though this was recently described, though not in
  MM itself, as for illustrative purposes only.
• Needs a methodology based on expected
• Complexity of underlying science
• RD costs (also depending on types of firms
  encouraged)
• Epidemiology
• Production costs, etc.
• NOTE Not just information on the medical
  condition itself.

34
Not a good idea to base on typical market size

• An auction and heavy monitoring suggested to
  set size
• Couldnt work so abandoned.
• Now size based on typical market size of new
  drugs and heavy monitoring to check firms are
  investing enough
• Implicitly this means that the size is based on
  the typical costs of developing such drugs,
  since, in equilibrium, investment in drug
  development should be driven to the point where
  this is the case
• This methodology is therefore essentially random
  for these early-stage vaccines
• Overestimates (per unit) innovation costs of
  developing and emerging country innovators, even
  as they struggle to take advantage of APCs, even
  as it underestimates eventual costs if dependent
  on APC and developed country developers.

35
Setting size too high is wasteful

• Racing, duplication.
• Even less incentive not to share information.
• Rent-seeking/lobbying/corruption.
• Reduced resources made available for other
  vaccines and treatments, sanitation, nutrition,
  housing, etc.
• If using an International Financing Facility,
  IFF, overly-high (and overly-low) APCs add to the
  risk the IFF bears.
• Extra deadweight losses of taxation and the
  opportunity cost of the other projects that
  foundations, governments, and the IFF are
  prevented from doing.
• If firms are not perfectly competitive,
  shareholders gain something (ceteris paribus)
  but at the expense of neglected diseases and
  other poverty alleviation projects.

36
Other reasons for overpayment

• Me-too drugs/vaccines partly discipline
  patented drug/vaccine prices.
• For purchases of underused vaccines, price
  disciplined by competitive tender, competition,
  access to IP, etc.
• These disciplining devices are lost under APCs
  for early-stage vaccines.
• Discipline in APC via committee and pre-agreed
  rules
• But pre-agreed rules are hard to set and to
  credibly follow through.
• RESULT Higher payment for given (lower) quality.
• Self-fulfilling incentive ex ante also to work on
  lower quality.
• There is also an additional ratchet effect No
  adjustment downwards if RD costs are reduced by
  technological advances or by improved publicly
  funded initiatives, etc.

37
Setting size too low is wasteful 1

• Get no, or too little, extra private funding into
  RD.
• Can raise size at the rate of interest rate. But
  no faster
• Raising the size of the APC acts like an extra
  discount factor
• Early investment becomes even more expensive
• Firms delay investment.
• But the rule about raising size is difficult to
  set
• How is the start level chosen?
• How is the speed of rise set?
• How is judgment made that not enough investment
  has taken place, without good monitoring and
  given that the result on which to base judgment
  is only provided at the end?
• Are politicians willing to sign on to such
  open-ended programs?
• Current CGD thinking is that this is too
  difficult (or politically unacceptable), and this
  is not planned (or CGD are not yet saying how
  later re-adjustment will happen).

38
Setting size too low is wasteful 2

• Development costs highly uncertain.
• CGD after a long deliberation process did not
  narrow down beyond the range of 15-25 per
  treatment
• The upper bound being 167 of the lower bound.
• If size starts, optimistically, at the bottom of
  the range when actual costs are at the top of the
  range, and 10 interest rate it takes 8 years
  till APC has any effect (or it collapses first).
• If real RD costs also grow at 5 per year
  (starting at the optimistic end of range) it
  takes 15 years to have any effect.
• Consequence is delay, and strong pressures
  towards poorer quality (broadly defined) at any
  given APC size in order to get a result.
• Maurer S. The Right Tool(s) Designing
  Cost-Effective Strategies for Neglected Disease
  Research, Goldman School of Public Policy,
  University of California at Berkeley, March 2005.
  Figures on this page from Maurer paper.

39
Some too high, others too low

• Some vaccines set way too low. Get no or little
  response
• 3bn for HIV? No connection to reality?
• Overconfidence in ability to revise upwards
  later?
• Other vaccines set too high and wasteful.
• Other cases, strong incentive to head for more
  limited quality
• Current malaria vaccine policy?
• We never get to see the good outcomes that we
  never get because of poorly-set initial terms.
• Overall a poor deal compared to alternatives?
• It depends on how well alternatives cope with
  incentivizing effort, and dealing with failures
  in vaccine RD portfolios.
• Yet, funder of program reserves right to abandon
  the mechanism if it is not working enough?

40
Getting it wrong for HIV?

• If scientific complexity means that RD costs
  are much higher for an HIV vaccine than for other
  medicines, then 3bn may be too low to stimulate
  sufficient investment Note that, if the
  commitment is too small to stimulate industry
  investment, and therefore does not succeed, there
  is no cost to sponsors.
• CGD FAQ sheet, April 2005.

41
Would a 3bn HIV purchase commitment simply
collapse?

• Political limits to a program without any effect
  (especially if high transactions costs to GAVI,
  Vaccine Fund, WHO, and others of setting up and
  running the program).
• Collapse is self-fulfilling (investors will not
  trust that any early investment will yield a
  payback, so dont bother).
• Collapse incorporates cases where to avoid
  litigation the mechanism does nothing but
  sits ready to activate even if very destructive
  meanwhile after an alternative approach has
  been successful.
• Worse if other approaches have stalled to make
  room for this approach.
• The sole criteria of the (political) willingness
  of sponsors and recipient governments to pay (MM
  p52, word inserted) for dictating the size
  chosen, is a non-criterion.

42
More pay-as-you-go better?

• This suggests more pay-as-you-go may be more
  capable of adapting to changing environment and
  less likely to fail to get a result (if set too
  low) or to overpay (if set too high).
• It depends on how good PPPs and others can be
  made at dropping failing projects.
• Note, alternative approaches still potentially
  include commercial involvement.
• Current low levels of commercial involvement in
  PPPs is more a fault of the shoestring funding of
  PPPs than of any specific failings of PPPs.

43
Technical requirements in advance? 1

• it would be possible though complicated to
  agree to product requirements in advancea small
  number of public health experts were concerned
  that it would be difficult to establish in
  advance technical requirements that a vaccine
  would need to meet. MM March 2005 p58, and final
  report.
• Some set of technical requirements always
  possible.
• But efficiency of those technical requirements
  requires some notion of the underlying
  feasibility of HIV, malaria and tuberculosis
  science, the potential costs of manufacture and
  distribution, epidemiology, etc. (i.e. many
  factors, and not just medical issues).
• So fixed enough to avoid the danger that the
  sponsor would renege, but flexible enough to
  accommodate contingencies that were not foreseen
  at the time the rules were established. MM March
  2005 p43, and final report.

44
Technical requirements in advance? 2

• Shifts problem to a different level of having a
  good notion of potential unforeseen contingences
  when setting the original rules.
• An inability to set efficient technical
  specifications for each vaccine far in advance
• Potential for great deal of discretion and
  interference later
• Risk that term-setters are unwilling to admit
  they got it wrong and to reset the terms more
  efficiently later (though they can only correct
  themselves in the downwards direction)
• Risky for all developers, but more so for some
  than for others
• Most risky for those unable to influence the
  committee
• It is expectations of all of this that matter to
  investors.

45
Minimum vaccine requirements 1

• Minimum vaccine requirements set at the start
• But, an efficient technical specification,
  closely resembling the eventual possible vaccine,
  would be impossible to set so far out for HIV,
  malaria, and tuberculosis.
• Discretion of as few as four members of a
  committee to grant waivers and make
  modifications (MM April 2005) but only to lower
  those requirements never to raise them.
• This asymmetry is bad policy in light of future
  possible improvements in science and possible
  future epidemiology.
• But symmetric ability to raise requirements as
  well as to lower requirements creates risks for
  firms, and is irreconcilable with the current MM
  mechanism.
• A dilemma.

46
Minimum vaccine requirements 2

• Constant pressures to lower the minimum
  acceptable requirements towards the very lowest
  level of any epidemiological value.
• In successive drafts of MM, malaria vaccine
  requirements gravitated ever-lower
• Final report 50 efficacy for 24 months from up
  to four doses
• This requirement (and the 3bn size) was recently
  described as for illustrative purposes only.
• A program that militates against the development
  of useful vaccines that far exceed minimum
  requirements?

47
Problems setting the long-term price

• Contracts call for determination, at the time of
  signing, of the guaranteed long-term price, or
  of an ex ante methodology for determining the
  long-term price.
• Legal obligation to supply at this price in the
  long-term in return for having had the short-term
  advantage of initial sales at very high, heavily
  subsidized, guaranteed prices.
• The 3bn is payment, in part, for this.
• A critical component of the advance market
  commitment (MM report April 2005) and key to its
  claim as the way to end ten to fifteen year
  delays in access to new vaccines.
• No such methodology for setting long-term price
  exists (see NIH evidence).
• CGD advised price could range from 0.50 to
  15.00, and that no such guarantee could be
  inserted into contracts.
• Left blank in MM contract term sheets.

48
What if firms wont/cant supply? 1

• Firm can always refuse to supply at short-term
  price (e.g. in order to supply a more lucrative
  market for HIV first).
• If cost not low enough, or firm prefers not to
  sell to eligible countries at the long-term
  price, contract allows sponsor to acquire right
  to produce
• Supplier(s) turn over IP to the sponsor but
  supplier may not have the right to sublicense all
  the IP
• Conflicts because supplier retains IP rights to
  non-eligible markets
• Sponsor has difficulties acquiring know how and
  production
• Threat undermines incentives to invest in vaccine
  RD
• Threat undermines incentives to invest in vaccine
  delivery systems
• Severe supply shortages and damaging access
  delays
• Reputational damage issues for supplier and
  sponsor(s)
• Not credible way to discipline firms.

49
What if firms wont/cant supply? 2

• Contract term sheets say other penalties such
  as liquidated damages provisions imposed on
  supplier
• Contract leaves details blank
• Threat feeds back to weaken ex ante incentives to
  invest in RD in the first place (especially if
  provisions are so vague)
• Threat can create perverse incentive not to
  supply eligible market in the first place or
  delay supplying eligible market
• HIV vaccine in particular, given richer
  non-eligible markets and the option value of the
  APC given the different HIV clades.
• Threat weakens incentive to invest in vaccine
  delivery systems
• Not credible way to discipline firms.
• Contract term sheets leave blank those sections
  specifying remedies in the event of a breach.

50
What if firms wont/cant supply? 3

• IP and know-how barriers have been principal
  causes of delays in achieving flexible,
  cost-effective manufacturing and quick access to
  vaccines for the poor in the past.
• Now control of IP and know-how part of a threat
  mechanism to drive the contracts!
• If the threats do not work, how do production
  costs get low enough to supply the ex post
  market?
• Lack of competition
• Weakened price signals due to the presence of the
  APC
• Lack of access to technology
• Insufficient pressures to lower production costs
• This is all contrary to, e.g., the Hepatitis B
  case, and to application of APC-type arrangements
  to current late-stage vaccines.

51
What if firms wont/cant supply? 4

• Legal advisors inform this author that threats in
  contracts at 20 year horizons are not used in
  practice.
• Long-term price and sustainable production
  capacity for eligible countries thereby
  unresolved in MM report.
• Excessive emphasis on getting the 3bn to the
  supplier(s) of the first 200m or so eligible
  treatments, and insufficient attention to
  follow-on suppliers and the post-200m treatments.
• A mechanism that relies on this presumption in
  order for it to work should be treated with a
  great deal of caution, indeed skepticism
  especially when, after 8 years since the idea
  first surfaced, the contract writers havent a
  clue how to do it.
• Vaccine developers facing many risks that a
  viable mechanism should seek to remove from them.

52
Follow-on innovation 1

• Advance purchase commitments may also stifle
  incremental innovation. Because they create a
  winner takes all solution, it would be
  difficult for incremental, follow-on competitors
  to emerge, thus dulling the benefits of
  competition on cost and improvements. The
  innovation that wins will crowd out competing
  inventions because it is being given away free by
  the public sector. This crowding out effect
  means that no improvements will be made to the
  winning formulation, and this may have negative
  consequences for resistance and effectiveness in
  subpopulations. International Policy Network
  Incentivising research and development for the
  diseases of poverty, 2005 p15.

53
Follow-on innovation 2

• "It is difficult to get the right quality, in
  particular to reward follow-on products that
  offer higher quality. Our view is that it should
  be possible to set an effective quality
  threshold, and that the terms of the APC must
  allow for superior quality follow-on products to
  be used(However) there may not be enough money
  left in the initial APC to reward the RD
  involved in developing some of the superior
  follow-on products. This is quite possible, as
  the commitment is only designed to generate at
  least one product that meets the quality
  threshold. Clearly a view would have to be taken
  by the donors as to whether they wished to
  finance follow-on products with additional money.
  This would be a separate investment decision from
  the original APC." Towse, A. and Kettler, H, A
  Review of IP and Non-IP Incentives for RD for
  Diseases of Poverty.What Type of Innovation is
  Required and How Can We Incentivise the Private
  Sector to Deliver It? April 2005, p87.

54
Follow-on innovation 3

• However, so as not to harm investor incentives,
  this additional money for follow-on products
  should be credibly promised in advance if not
  part of the original APC but that makes this
  additional money, by default, part of an
  original APC-type arrangement!
• Efficient incentives require each generation of
  products to cover its expected RD costs. How
  achieved?
• MM removes Quantity guarantee to (supposedly)
  remove the risk that the sponsor will end up
  funding a non-used product, and to give
  incentives for follow-on products.
• Multidimensional quality problem a quality
  surface over vaccines, over time. Set terms
  wrong and disincentivize (dynamic) investment.
  Expectations are everything to investors.

55
Follow-on innovation 4

• MM suggests potentially very complicated rules
  about qualities of acceptable vaccines, and
  variation in allocations and prices of vaccine
  purchases across multiple developers and
  purchasers, and over time
• But no practical details of how this is done
• Hard to visualize that policy makers could derive
  the optimal way to hold back on early products
  so as to leave funds for later products.
• Hugely aggravated by the fact that the likely
  quality improvements, science, and costs are all
  highly uncertain.
• Also no ex ante grasp of the potential timing of
  anything, or characteristics of firms such as
  access to finance, risk tolerance, etc.
• On average, achieving quality is more
  expensive.
• Will investors expect the distribution of the
  subsidy to be efficient?

56
Follow-on innovation 5

• Constant danger of poor vaccines driving out
  better vaccines (aggravated if the overall fund
  is set too low to start with).
• Need for follow-on instruments (Towse and
  Kettler, ibid.) but these need to be fully
  articulated and credibly promised in advance if
  investors are not to be harmed.
• Dangers that fund becomes unbounded at top, yet
  still highly uncertain killing dynamic
  incentives.
• Original (Kremer) model is static and ignores
  these issues.
• In particular it presumes one vaccine target.
• Recently, these dynamics are simply presumed
  perfectly performed by the committee
• In truth the committee would fall massively short
  and investor and researcher expectations would
  respond accordingly.

57
Follow-on innovation 6

• Recent malaria case so far no attention at all
  to dynamic incentives in CGD and UK policy
  announcements.
• End up with a committee with discretion, risk to
  investors, higher risk premia, more pressure to
  go for lower quality, and rent seeking behaviour
  of big players?
• Follow-on products also covers cases of
  complete vaccine replacement.
• Incentives to actually carry out this
  replacement?
• Incentives to expediently create capacity when
  vaccines are replaced?
• Replacement is only statistical. How is this
  handled in size and distribution rules of APC?
• All the time there is the need for developing
  country trust of vaccine products and of ethical
  trials.

58
Follow-on problems especially severe for HIV,
malaria and TB

• An Arms war between virus and drugs/vaccines.
• First vaccines not necessarily the best
• More so if only therapeutic and not preventative.
• Expectations about payments to the first must not
  stifle incentives for later needed vaccines.
• Expectations about payments to non-composite
  vaccines must not stifle incentives for composite
  vaccines
• HIV vaccines coordination, information sharing,
  and timing worries.
• Expectations about payments for therapeutic
  vaccines
• Needing monitoring and even more long-term
  follow-on vaccines.
• Worries that better vaccines arrive after
  subsidies are gone.
• Second-generation vaccine costs even harder to
  work out in advance than first-generation costs.
• It is all about INVESTOR EXPECTATIONS.

59
The paradox of market risk 1

• For most late-stage and underused vaccines the
  objective of purchase commitments is to remove
  market risk.
• Early-stage vaccine APCs (as currently designed)
  put market risk back on to vaccine developers
  (supposedly) to, ex post, drive the quality
  rules and ex ante effort incentives. BUT
• These are resource-poor markets
• Most buyers are relatively uninformed (about
  current vaccines never mind about expected future
  vaccines)
• There are no marketing budgets
• Vaccine usage needs a good distribution system,
  with such systems generally not under the control
  of vaccine companies
• There are heavy knock-on costs to purchase
  decisions
• There are multiple organizational problems

60
The paradox of market risk 2

• There is a severe lack of qualified personnel on
  the ground
• There are multiple political interests
• There are cultural barriers
• There are plenty of ways to encourage
  decision-makers to take one firms product over
  another firms product (even more so if the
  other firms product does not exist yet)
• There are strong self-fulfilling pressures in
  the co-payment mechanism driving towards
  lower-quality outcomes
• This, ex ante, feeds investor expectations and
  RD incentives towards the low-quality
  outcomes.
• Why put all these risks back on to vaccine
  developers in order to try get this mechanism to
  work for early-stage vaccines and to try to
  create follow-on incentives?
• Less of an issue if system collapses down to
  just the one vaccine but that, supposedly, was
  not the point.

61
The paradox of market risk 3

• Maybe intention is that mechanism collapses down
  to a limited number of developers?
• In case-studies, production scale and competition
  were key to low-cost vaccine production.
• No sense using (expected) restrictions and
  holding back of sales of a single supplier to
  discipline quality.
• Why inflict uncertainty on investors seeking to
  scale up manufacturing capacity?
• It is dynamically inconsistent anyway hard to
  hold back on a bird-in-the-hand vaccine to
  supposedly discipline quality and to leave
  funds for later vaccines.
• Leads to further pressure to lower quality with
  risk to quality investors.
• More likely just lots of uncertainty, higher
  capital costs, and fewer developers in the end
  game.

62
This paradox undermines RD incentives

• Hard to see how any quality control over the
  whole development process could be done in the
  end-game in this way without conflicting with the
  need to get the manufacturing costs low.
• RESULT A mechanism that disciplines quality en
  route is better able to achieve larger capacity,
  multiple suppliers, and low prices, than is the
  mechanism that disciplines quality via holding
  back in the end market. A FUNDAMENTAL CONFLICT.
• Paradoxically, undermines investors who will not
  believe that manufacturing costs will be pushed
  low enough ex post to make the whole investment
  exercise worthwhile ex ante (see forthcoming
  slide for more on this).

63
Incentives to block second generation products

• Unlike patents, the first to market is likely to
  get all the subsidy if it can hold off
  follow-on vaccines just long enough. Even
  better if it can create the reputation for this
  it becomes self-fulfilling and the firm gets the
  subsidy quicker (if there is discretion over how
  to split the subsidy).
• With standard patent-based systems and marketing,
  firms can more easily agree to split the market,
  so there is incentive to share/license/split the
  higher price for the better product.
• This has all gone.
• This gives strong incentives for the
  first-generation developers to block (advertently
  or inadvertently) second generation vaccines (via
  not sharing IP, know-how, etc.).

64
The sums dont add up either

• Pricing structure can be designed to provide
  substantial insurance against demand risk for
  prospective vaccine developers so as to yield a
  net present value of revenue comparable to
  commercial products even under pessimistic uptake
  scenarios, MM March 2005 p50 and p114, emphasis
  added
• Indicates the complicated tradeoff that needs to
  take place, not that it would take place or ever
  could take place (it could not, because of the
  intractable information problems needed to make
  it work)
• Commercial return means ex ante commercial
  return. What happens when that is gone?
• No way to know how to set any of this many years
  in advance.
• In reality, according to recent announcements,
  seems little intention to use any of these
  dynamic rules anyway.

65
Two-stage game problems

• Always better to compete for contract before
  sinking costs.
• Award after costs are sunk runs the risk of
  forcing firms to compete twice
• RD stage
• Market stage.
• Bygones are bygonesAfter costs are sunk,
  always worth spending up to the value of the
  contract in rent-seeking behavior to win the
  contract.
• In such situations, firms expect to face a
  negative rate of return from the overall project.
• CONSEQUENCE The incumbent firm signals/behaves
  in ways to increase the expected risk of later
  firms. No second firm bothers to enter whatever
  the size of the commitment. Competition is
  stifled.

66
Avoiding two-stage game problems

• Must fix terms at the start to make it clear that
  no amount of lobbying/spending at the second
  stage will change the payouts.
• BUT
• Contradicts the need for discretion to change
  terms
• Becomes mechanistic based on expectations at the
  very start
• E.g. HIV science that is 10-20 years out of date!
• Real problems if a market test is used since
  firms can try to influence purchase decisions
  (illegal kickbacks, bundling to hide discounts.
  There are sanctions against the former if
  detected. Hard to detect the latter and fewer
  sanctions)
• Biased against small biotechs (cant bundle and
  cant hide other subsidies), not-for-profits (who
  may not be allowed to behave in these ways), and
  emerging developers. Again, expectations of this
  will disadvantage the latter groups when trying
  to acquire finance for RD.

67
Ex ante versus ex post information problems an
unhelpful caricature

• Sponsors of APCs said to need little knowledge
  about the likely success of particular
  approaches.
• But, they do need a huge amount of qualitative
  and quantitative information about overall set of
  potential scientific, epidemiological, expected
  research and manufacturing costs, existing
  market possibilities, etc. well in advance of
  product development in order to get the terms
  vaguely efficient.
• To be credible and to minimise the risks to
  firms, firms need to trust that policy-makers
  have this ex ante information. Otherwise, lots of
  discretion later and risk.
• Mechanism cannot claim it solves information
  difficulties it has ruled out at the start.

68
Crowding out weakens pull 1

• The proposal is that private investment would
  underpin RD by private firms Barder, O., CIPIH
  Forum, 19 November 2004.
• Has to be additional incentive
• Additional to current RD
• Additional to current market.
• Has to be believed that it will be additional.
• This is a big challenge.
• Easy to achieve additionality for underused
  vaccines
• It is a total non-issue.
• More difficult to achieve for late-stage vaccines
  though still possible
• Run a competitive tender (with more access to IP)
  and use known scientific information, etc.

69
Crowding out weakens pull 2

• Extremely difficult to achieve for early-stage
  vaccines.
• Referring to HIV/AIDS, perinatal conditions,
  maternal conditions, childhood diseases, malaria,
  and tuberculosis
• All of these are health problems in the first
  world too, and, ignoring the special case of
  mutant strains, there are substantial incentives
  for the development of effective therapies.
  Malaria is perhaps the principal exception, but
  there is a lot of work on it currently. To be
  sure, some diseases occur primarily in the third
  world, but the magnitude of the problem that is
  uniquely without solution ought to be brought
  into sharper perspective. F.M. Scherer, CIPIH
  May 2005. Emphasis added.
• Pull payment to any firm would need to be reduced
  by many times the push payments it had ever
  received to achieve a level playing field between
  firms (unless unequal access to push efforts can
  be made to be part of the efficient solution).

70
Crowding out weakens pull 3

• For scientific areas with a complicated interplay
  of push and pull and a large push element, there
  is great risk that the pull-motivated will lose
  out to the push-motivated and to those not
  relying on an APC for their funding
• Allowing heavily push-motivated and
  PPP/philanthropic-financed players to have access
  to the fund, effectively weakens the expected
  value of the fund to the APC players.
• SIMPLE EXAMPLE
• 10 firms working equally hard on HIV vaccines,
  70 capital costs, no splitting of 6.25bn APC
  fund
• Winning firm has 50 subsidies, grant support,
  non-private funding, etc.
• Winning firm should be denied just over 3bn of
  the fund
• 3bn left in the pot for competing and
  follow-on vaccines
• But every 1m of hiding, worth nearly 17m to
  the firm.

71
Crowding out weakens pull 4

• Needs a great deal of monitoring. Paradoxical
  given that those advocating the program often
  highlight the opposite.
• Need for high-quality historical evidence (20-30
  years).
• Need for committees, discretion, treaties?
• Impossible to correctly price streams of other
  payments (e.g. appropriate capital costs?).
• Repayment side-contracts that may not unfold
  for ten or twenty or more years.
• The mechanism claims that to work out an optimal
  strategy, every firm needs to know how much
  privately-funded activity is taking place. How is
  this achievedif what is going on is so opaque?
• Private firms need to trust that push is being
  efficiently handled.

72
Crowding out weakens pull 5

• Favors large pharma players for early-stage
  vaccines
• Smaller firms, biotechs, not-for-profit firms,
  etc. have many fewer ways to hide research
  supports (if they can get them)
• Many biotechs work on one area only their
  funding flows are less opaque than large pharma
  players
• Easier for large pharma to avoid monitoring
  generally.
• Standard procurement contracts capable of paying
  only for additional private funds to finish a
  project
• Competitive tender separates out the push from
  the pull funding.
• Here, Framework Agreement is the tender.
• Here, highly complicated side device has to be
  appended to the tender to achieve a standard
  property.

73
Crowding out weakens pull 6

• Near-scratch versus near-market
• APC (if not appropriately adjusted)
  disproportionately benefits those nearer to
  market, even if not the ultimately best
  vaccine
• Disincentivizes near-scratch developers
• This reduces the average expected quality of
  vaccines.
• PPP funding
• How exactly do PPP activities complement and
  not conflict with APC-based private activities?
• Should an MVI- or IAVI-funded vaccine be denied
  APC funds?