DOCTORAL THESIS
Modeling Global P icing and
Launching o New D ugs










Au ho : Bo ja Ga cía Lo enzo
Las Palmas de G an Cana ia, Ap il 2014
DOCTORADO EN ECONOMÍA: APLICACIONES A LAS FINANZAS Y
SEGUROS, A LA ECONOMÍA SECTORIAL, AL MEDIO AMBIENTE Y A LAS
INFRAESTRUCTURAS.
Modeling Global P icing and
Launching o New D ugs
Tesis doc o al p esen ada po D. Bo ja Ga cía Lo enzo
Di igida po D a. Bea iz González López-Valcá cel
La Di ec o a, El Doc o ando,
Las Palmas de G an Cana ia, ab il de 2014

Exis imos po que alguien piensa en noso os,
y no al e és
Acknowledgmen s
Fo emos , I would like o exp ess my since e g a i ude o my ad iso D . Bea iz
González López-Valcá cel o he con inuous suppo o my Ph.D s udy and esea ch, o
his pa ience, mo i a ion, en husiasm, cons an eedback and immense knowledge. His
guidance helped me in all he ime o esea ch and w i ing o his hesis.
My since e hanks also go o D . Izabela Jelo ac and D . Ma ga e Kyle o o e ing
me he oppo uni ies o enjoy my isi ing schola s in he G oupe d’Analyse e Théo ie
Economique (GATE) and he Toulouse School o Economics (TSE) espec i ely, o hei
encou agemen , insigh ul commen s, and ha d ques ions. I also hank Ca los J. Pé ez o
his helps in he ield o decision heo y.
I hank my ellow o icema es in he Uni e si y o Las Palmas de G an Cana ia
(ULPGC): Reinaldo, Hicham, Rubén, Te esa and Fede ico o he s imula ing discussions,
o he ha d days we we e wo king oge he , and o all he un we ha e had in he las
yea s.
I g a e ully acknowledge he unding ecei ed owa ds my PhD om he Cana ian
Agency o Resea ch, Inno a ion and In o ma ion Socie y o he Cana ian Go e nmen
(ACIISI). Also, I hank IMS o p o iding he da a o he empi ical sec ion, pa icua ly o
Miguel Ma ínez.
Las bu no he leas , I would like o hank my pa en s Roque and Pepa o
suppo ing my educa ion wi hou ega d, and Nai a, o he unde s anding, e en so she
has no a Ph.D, she has suppo ed me as i she was one. F iends a ound me ha e been a
g ea suppo o each his momen . Thank you all.


Con en s XVII
Lis o Tables
TABLE2.1PPRFOROPTIMALCOUNTRYLAUNCHSEQUENCE(PI,QI)................................................................................69
TABLE2.2COUNTRYLAUNCHSEQUENCEINFIGURE2.2..............................................................................................71
TABLE2.3.OPTIMALCOUNTRYLAUNCHSEQUENCEINFIGURE2.3................................................................................72
TABLE2.4.OPTIMALCOUNTRYLAUNCHSEQUENCEINFIGURE2.4................................................................................73
TABLE2.5.OPTIMALCOUNTRYLAUNCHSEQUENCEINFIGURE2.5................................................................................74
TABLE3.1DESCRIPTIVESTATISTICS.RETAILMARKET....................................................................................................83
TABLE3.2DESCRIPTIVESTATISTICS.HOSPITALMARKET................................................................................................84
TABLE3.3RELATIVEPRICESPEARSONCORRELATION.RETAILANDHOSPITALMARKET.........................................................85
TABLE3.4.BIVARIATETEST.ERPVS.NOERP...........................................................................................................85
TABLE3.5.BIVARIATETEST.EMAVS.NOEMA........................................................................................................85
TABLE3.6.LAUNCHDELAYEQUATIONOFTHENPLM................................................................................................103
TABLE3.7.RELATIVELAUNCHPRICEEQUATIONOFTHENPLM....................................................................................106
TABLEA.1OVERVIEWOFTHEORETICALSTUDIES.......................................................................................................120
TABLEA.2.OVERVIEWOFEMPIRICALSTUDIES..........................................................................................................121
TABLEB.1.HEALTHAGENCYSURPLUSOFCOUNTRYC................................................................................................143
TABLEB.2.HEALTHAGENCYSURPLUSOFCOUNTRYD...............................................................................................143
TABLEC.1.LAUNCHEQUATION:D&EVS.UPDATEDMODEL.......................................................................................153
TABLEC.2.LAUNCHPRICEEQUATION:D&EVS.UPDATEDMODEL...............................................................................157
TABLEC.3.LISTOFCOUNTRIES.VERNIERSETAL.VS.UM..........................................................................................165
TABLEC.4.LAUNCHWINDOWANDLAUNCHPRICEEQUATIONS.VERNIERSETAL.VS.UPDATEDMODEL.............................167
TABLEC.5.VARIABLECLASSIFICATIONOFTHENPLM................................................................................................171
TABLEC.6.PROBITSELECTIONEQUATIONOFNPML.................................................................................................171
TABLEC.7.GOODNESSOFFITOFPARAMETRICMODELS.RETAILMARKET.......................................................................174
TABLEC.8.GOODNESSOFFITOFPARAMETRICMODELS.HOSPITALMARKET...................................................................176
TABLA4.1.PPRPARALASECUENCIAÓPTIMADELANZAMIENTO(PI,QI).........................................................................202
TABLA4.2.REGIONESDESECUENCIASÓPTIMASDELANZAMIENTOA)...........................................................................204
TABLA4.3.BONDADDELAJUSTEDELOSMODELOSPARAMÉTRICOS.MERCADOAMBULATORIO..........................................212
TABLA4.4.ECUACIÓNDELPRECIORELATIVODELANZAMIENTO...................................................................................225
TABLA4.5.ECUACIÓNDERETRASOENELLANZAMIENTO............................................................................................227
Lis o abb e ia es
AIFA I alian Na ional Agency o D ug Adminis a ion and Con ol P ices
AME A e age Ma ginal E ec
ATC Ana omic The apeu ic Chemical Classi ica ion Sys em
CHEPA Cen e o Heal h Economics and Policy Analysis
CEA Cos -E ec i eness Analysis
CRES Cen e de Rece ca en Economia i Salu
DDD De ined Daily Dosage
D&E Danzon & Eps ein
EMA Eu opean Medical Agency
ERP Ex e nal Re e ence P icing
EU Eu opean Union
FDA Food and D ug Adminis a ion
GDP G oss Domes ic P oduc
GLS Gene alized Leas Squa es
GPRM Global P ice Repo ing Mechanism
HHI Hi schmand-He indähl Index
HTA Heal h Technology Assessmen
XX Modeling Global P icing and Launching o New D ugs
ICER Inc emen al Cos -E ec i eness Ra io
IMF In e na ional Mone a y Fund
IMR In e se Mills Ra io
LSE London School o Economics
MES Minimum E icacy S anda d
MEPS Medical Expendi u e Panel Su ey
MLIC Middle and Low Income Coun y
NBER The Na ional Bu eau o Economic Resea ch
NCE New Chemical En i ies
NGO No-Go e nmen al O ganiza ion
NICE Na ional Ins i u e o Heal h and Ca e Excellence
OECD O ganisa ion o Economic Coope a ion and De elopmen
OF Objec i e Func ion
OLS O dina y Leas Squa es
OTC O e - he-Coun e
PC P ice Cap
PE Public Expenses
PI Pa allel Impo e
PPI P oduce P ice Indexes
PPP Pu chasing Powe Pa i ies
PPR P elimina y Resul s
Con en s XXI
PRISMA P e e ed Repo ing I ems o Sys ema ic Re iews and
Me a-Analyses
PT Pa allel T ade
QALY Quali y-adjus ed Li e Yea s
R&D Resea ch & De elopmen
RP Re e ence P ice
SU S anda d Uni
UK Uni ed Kingdom
UM Upda ed Model
US Uni ed S a es
W P Willingness o pay
3SLS Th ee-s age leas squa es

In oduc ion
Pha maceu icals a e sold in a global ma ke . This cha ac e is ic implies a speci ic
ba gaining p ocedu e be ween pha maceu ical i ms and coun ies’ heal h agencies. On
he one hand, hese i ms make s a egic decisions when launching medicines in di e en
coun ies and o maximize hei global p o i s; and on he o he hand, coun ies’ heal h
agencies implemen p icing policies in o de o con ol hei pha maceu ical expendi u e
and o gua an ee access o medicines.
F om he pe spec i e o na ional heal h insu ances, p icing policies wi hin he
pha maceu ical ma ke a e a key ac o in con olling public expendi u e (Sche e , 1993,
Lobo, 2014)1. Pa icula ly, he o al pha maceu ical bill: ac oss he O ganisa ion o
Economic Coope a ion and De elopmen (OECD) coun ies in 2009, his bill is es ima ed
o ha e accoun ed o a ound 19% o heal h spending. In ela ion o he o e all economy,
pha maceu ical spending accoun s o an a e age 1.5% o GDP in OECD coun ies.
Howe e , he dispe sion a ound his a e age is high, pha maceu ical spending accoun s
o less han 1% o GDP in No way and Denma k, while i eaches close o 2.5% o GDP
in G eece, Hunga y and he Slo ak Republic. Expendi u e on pha maceu icals is
p edominan ly inanced h ough hi d-pa y paye s in mos OECD coun ies – ei he
h ough he public heal h insu ance, which accoun s o a ound 60% o he o al on
a e age, o h ough p i a e insu ance co e age, lea ing an a e age o mo e han a hi d
o he o al o be cha ged o households (OECD, 2011).
F om he pha maceu ical indus y iew, p icing and launching a new d ug is a
complex ask di ec ly connec ed o R&D policy, indus ial policy and heal hca e policy.
Hence, p icing and launching a e majo s a egic decisions. In many coun ies, he p ice is
ag eed wi h heal h ca e insu ance p o ide s (public o p i a e). Na ional p icing policies
1 The case o Spain as an example o he p ice egula ion LOBO, F. 2014. La In e ención de P ecios de los Medicamen os
en España, Mad id, Sp inge ..
Modeling Global P icing and Launching o New D ugs
2
and s a egies a e essen ial elemen s in se ing p ices and making medicines a ailable,
since d ug p icing should con ibu e o enhancing social wel a e and ake in o accoun he
in e es s o he indus y, consume s and public insu e s. The e o e, encou agemen mus
be p o ided o de elop new medicines, make hem a ailable o consume s and, a he
same ime, con ol pha maceu ical expendi u e.
P icing and launching in ol e ade-o s be ween public wel a e and p i a e p o i s,
be ween he in e es s o he manu ac u e and hose o he coun y. When coun ies se a
d ug p ice, hey isk he possibili y o no p o iding i a he ime hey desi e, which may
ha e consequences o he heal h and he wel a e o he popula ion (Lich enbe g, 2005).
In u n, a i m ha delays he launch o a medicine in a coun y is also delaying he p o i s
o be de i ed om his coun y. Howe e , in an inc easingly globalized wo ld, na ional
p icing/launching o d ugs has become in ac an in e na ional ma e and
in e dependencies ac oss coun ies should be aken in o accoun . Bo h companies and
coun ies mus ac locally bu hink globally. Due o mechanisms like ex e nal e e ence
p icing (ERP, hence o h) and pa allel ade (PT, hence o h) (Danzon e al., 2005,
Danzon and Eps ein, 2008, Ga cia Ma iñoso e al., 2011), se ing he p ice o a d ug in a
pa icula coun y in luences o he coun ies’ p icing and launching. The use o ERP by
coun ies may make a i m apply in e na ional p icing s a egies ha may ha m coun ies’
wel a e. On he one hand, he i m may se a single p ice2, which may bene i high-p ice3
coun ies bu ha m low-p ice ones. On he o he hand, he i m may ei he a emp o se
high4 p ices in he i s coun ies o a oid low-p ices in la e launches ia ERP, o delay
launches in low-p ice coun ies o a oid spill-o e e ec s. These s a egies can ha m low-
p ice coun ies, and may e en ha m high-p ice ones (Ga cia Ma iñoso e al., 2011).
Among exis ing d ug p icing policies, mos coun ies in he indus ialized wo ld
ha e implemen ed ei he Cos -E ec i eness Analysis (CEA, hence o h) o ERP a some
2 Two ac o s con ibu e o p ice uni o mi y be ween di e en ma ke s: a) he h ea s o pa allel impo s, and b) he use o
in e na ional e e ence p icing DANZON, P. M. & TOWSE, A. 2003. Di e en ial P icing o Pha maceu icals: Reconciling
Access, R&D and Pa en s. In e na ional Jou nal o Heal h Ca e Finance and Economics, 3, 183-205..
3 In he long un, consume s om high p ice coun ies will be wo se o i his lowe p ice esul s in lowe han expec ed
e u ns on R&D, and hence ewe new medicines han hey would ha e been willing o pay o DANZON, P. M. 1997. P ice
Disc imina ion o Pha maceu icals: Wel a e E ec s in he US and he EU. In e na ional Jou nal o he Economics o
Business, 4, 310-322..
4 This company s a egy will no wo k i he high-p ice coun y e ises i s p ices downwa ds a e launch DANZON, P. M. &
TOWSE, A. 2003. Di e en ial P icing o Pha maceu icals: Reconciling Access, R&D and Pa en s. In e na ional Jou nal o
Heal h Ca e Finance and Economics, 3, 183-205, DANZON, P. M. 1997. P ice Disc imina ion o Pha maceu icals: Wel a e
E ec s in he US and he EU. In e na ional Jou nal o he Economics o Business, 4, 310-322.
In oduc ion
3
poin in ime wi h he aim o con olling pha maceu ical expendi u e, while s ill ensu ing
access o medicines, mainly in on-pa en medicines (Espin J e al., 2011, Rawlins, 2012).
In his hesis, ERP is de ined as “ he p ac ice o se ing a p ice cap o
pha maceu icals, based on ex-manu ac u e 5 p ices o iden ical o compa able p oduc s in
o he coun ies” (Ga cia Ma iñoso e al., 2011). Mos coun ies use ERP as a
pha maceu ical p icing s a egy. The use o ERP as a mechanism o se pha maceu ical
p ices is qui e widely applied: 24 o he 30 OECD coun ies (Espin J e al., 2011) and
app oxima ely 24 o he 28 EU Membe S a es (Leopold e al., 2012) ha e used i .
Howe e , ERP is no applied homogeneously in e e y coun y. The e a e a wide a ie y o
me hods o design a o eign p ice index (Leopold e al., 2012, Espin J e al., 2011). I
mainly depends on each coun y’s baske , he ype o p ices collec ed6, he me hod used
( he lowes p ice, he a e age p ice, a pe cen age o he p e ious ones, e c.) and whe he
a weigh ed-index7 is used o no . We also no e ha some coun ies ake in o accoun ERP
as a complemen a y p icing policy oge he wi h o he p icing policies o help o make he
p ice decision, and he e o e i is no exclusi ely applied as a blind p icing policy8. ERP is
used because o i s simplici y a a echnical and analy ical le el; collec ing p ice
in o ma ion ab oad does no equi e a huge e o . Fu he mo e, ERP use s hink ha he
p ices aken as e e ence a e oughly igh , sui able o ai . Howe e , hey ecognise ha
i is di icul o assess i he esul ing p ices a e app op ia e, e icien o op imal in
acco dance wi h any objec i e c i e ion. Addi ionally, i e e encing coun ies se hei
p ices oo high o oo low, hen any coun y la e applying he ERP me hod may un he
isk o epea ing he same mis ake (Espin J e al., 2011).
CEA in heal h economics aims o es ima e he a io be ween he cos o a heal h-
ela ed in e en ion and he bene i i p oduces in e ms o he numbe o yea s li ed in ull
heal h by he bene icia ies. Cos is measu ed in mone a y uni s, while bene i needs o be
exp essed in gain o heal h measu ed by quan i a i e alues. Howe e , unlike cos –
5 P ices a e ex-manu ac u e p ices.
6 Cu en p ice s. p ice a launch
7 The mos widely me hod used o new d ugs is h ough non-weigh ed measu es; such me hods will no help o achie e he
a ge o ob aining a compa able a e age le el o p ices. The applica ion o weigh ed p ice indexes, compa able and use ul
as e e ence o he es o coun ies, has been p oposed DANZON, P. M. & CHAO, L. W. 2000. C oss-na ional p ice
di e ences o pha maceu icals: How la ge, and why? Jou nal o Heal h Economics, 19, 159-195..
8 Espín e al. s a e ha “ egula o s migh no always be able o willing o “impose” a ce ain p ice, bu ins ead use he p ice
compu ed as a benchma k o e e ence o nego ia ions, o en alongside o he c i e ia, such as cos -plus, in e nal o
he apeu ic p icing”.
10 Modeling Global P icing and Launching o New D ugs
on he coun y size and posi i ely on he coun y pe capi a income. This means ha as
long as he coun y size is la ge, he egula o will ha e g ea e ba gaining powe o
nego ia e p ices wi h he pha maceu ical indus y. By con as , a high GDP pe capi a is
ela ed o a high willingness o pay o a d ug. Also, he ask p ice depends posi i ely on
he coun y p opensi y o spillo e s due o he use o ERP policy and pa allel expo s.
This means ha whe he he coun y is a po en ial e e enced coun y (see 2.2. below) o
a po en ial pa allel expo e (see 2.3 below), he i m will inc ease i s ese a ion p ice.
The ba gaining esul s in he launch o he p oduc i he coun y’s maximum o e p ice
equals o exceeds he i m’s minimum ask p ice. I his condi ion is no me he delay
occu s, mo eo e , he g ea e his di e ence, he longe he delay in launch. Danzon and
Eps ein (Danzon and Eps ein, 2008) wo k unde he same hypo hesis as Danzon e al.
(Danzon e al., 2005). In his case, hey also con empla e he ba gaining esul s in p ice
and add mo e a iables as explana o y ac o s o p icing and launching. One o hem is
he egula o y egime, which can be an in e nal e e ence p ice (RP) o ERP; bo h a e
expec ed o posi i ely a ec p ices because subs i u e p ices, ei he a home o ab oad,
a e expec ed o inc easing p ices (see 2.2. and 2.6. below). The o he a iable added, he
i m’s loca ion, is measu ed by he ixed cos s, which a e expec ed o be lowe i he
launching i m is loca ed in he coun y analysed (see 2.5. below). Now, he ba gaining
esul s in a p ice ag eed wi hin he ange be ween he o e p ice o he coun y and ask
p ice o he i m. Then he launch is likely o occu when he o e p ice o he coun y
equals o exceeds he ask p ice o he i m. The au ho s unde line ha he ade-o s
be ween p ice and delay a e expec ed o di e ac oss ma ke s and ac oss p oduc s wi hin
ma ke s.
1.2.2 How he ERP is a ec ing he ba gaining esul s in p icing
and launching?
Bo h Danzon e al. and Danzon and Eps ein conside ha he p opensi y o being
a e e ence coun y may posi i ely a ec he p ice o a d ug (Danzon e al., 2005, Danzon
and Eps ein, 2008). This a gumen is suppo ed by he ex ended use o ERP by coun ies
as a cos -con ainmen policy (see Leopold e al. (Leopold e al., 2012)). Basically, ERP
consis s o se ing a p ice cap o pha maceu icals, based on p ices o iden ical o
compa able p oduc s in o he coun ies. Despi e he me hod o calcula ion (see Leopold e
al., Rich e , S a ga d and Sch eyögg (Leopold e al., 2012, Rich e , 2008, S a ga d and
Sch eyögg, 2006), we hink ha , whe he a coun y is aken as e e ence by o he

Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
11
coun ies when p icing medicines, i is easonable o conside ha i m’s incen i e will be
o se high p ices in he e e ence coun y. As discussed by Rich e (Rich e , 2008), he
i m is be e o launching i s d ug in high-p ice coun ies i s , o in luence p ices in o he
coun ies o i s ad an age. In his sense, Ga cía-Ma iñoso e al. (Ga cia Ma iñoso e al.,
2011) di ec ly analyse he e ec s o an ERP policy on a e e encing coun y on he
nego ia ion in his coun y and, u he mo e, he incen i e o he e e encing coun y o
apply ERP. To go in o his idea, he au ho s conside one pha maceu ical i m, one on-
pa en d ug and wo coun ies ope a ing a posi i e lis o eimbu sed pha maceu icals,
whe e pa ien s pay a ixed and exogenous co-paymen . Coun ies di e in he popula ion
size and he le el o co-paymen . A model o nego ia ion p ocess as a Nash ba gaining
game is designed h ough which he au ho s compa e independen p ice nego ia ions o
he si ua ion in which one coun y ( he e e encing coun y) engages in ERP. Two di e en
scena ios a e analysed, unde “weak h ea s”12 i he d ug is no eimbu sed, o unde
“ ough h ea s”13 i he d ug is banned. In he case o ERP wi h weak h ea s, when he
e e encing coun y engages in ERP, he p ice nego ia ed in he e e enced coun y
inc eases. The o al su plus gene a ed by he nego ia ion be ween he e e enced coun y
and he i m inc eases14. This shows ha he implici nego ia ion powe o he i m is
highe when he e e encing coun y engages in ERP as compa ed wi h independen
nego ia ions. As Danzon e al. and Danzon and Eps ein had suspec ed (Danzon and
Eps ein, 2008, Danzon e al., 2005), Ga cía-Ma iñoso e al. (Ga cia Ma iñoso e al., 2011)
show he ac ha he e e encing coun y engaged in ERP policy ha ms he e e enced
coun y in e ms o high p ice and lowe ou pu s. The same au ho s also examine he
incen i es o apply ERP policy a he han independen nego ia ions. Unde he same
hypo hesis s a ed abo e, hey s a e ha a coun y has an incen i e o engage in ERP i i s
co-paymen le els a e high when compa ed wi h he e e enced coun y. This p e e ence
dec eases as he size o he e e encing coun y inc eases, and as co-paymen s o bo h
coun ies con e ge. Fi s , he e e encing coun y size inc eases he ERP nego ia ed
12 Unde “weak h ea s”, i he nego ia ions ail, he i m can s ill sell he d ug a any p ice o i s choice, bu wi h no
subsidy.This assump ion is mo i a ed by he ac ha , in Eu ope, p ice-nego ia ing agencies ha e a mino ole in he
au ho iza ion o d ugs.
13 Some coun ies ou side Eu ope, such as B azil o Canada, a e known o h ea en he i ms wi h no au ho izing d ug
sales i nego ia ions ail o i he i m does no accep ERP.
14 In his model, ERP is based on he p ice o a single e e ence coun y. Howe e , esul s a e highly sensi i e o he
modali ies o he ERP.
12 Modeling Global P icing and Launching o New D ugs
be ween he e e enced coun y and i m in wo ways. The pie o be sha ed be ween bo h
pa ies is la ge , and he i m has a s onge disag eemen payo while he disag eemen
payo o he e e enced coun y emains he same. Second, as he nego ia ed p ice is
shown o be inc easing wi h he pa ien s’ co-paymen (see (Jelo ac, 2008)), i he co-
paymen in he e e enced coun y dec eases wi h espec o he e e encing coun y hen,
he ERP will dec ease and he e o e he di e ence be ween he p ice independen ly
nego ia ed and he ERP will dec ease. Then, hey conclude ha only small coun ies
should be obse ed o engage in ERP and/o ERP should be based on p ices in la ge
coun ies (o a la ge g oup o coun ies). The analysis yields an analogous p edic ion i
one subs i u es “la ge coun y” by ‘small co-paymen coun y’ and ice e sa.
The au ho s u he ex end hei analysis o accoun o compe i ion be ween he
i m’s pha maceu ical p oduc and a he apeu ic subs i u e ha is al eady p esen on he
ma ke in bo h coun ies. This ex ension adds ealism, pa icula ly, i makes he weak
h ea s scena io compa ible wi h he obse a ion ha , in mos Eu opean ma ke s, being
excluded om he public unding may be almos as bad as being banned, as sales ou o
he posi i e lis o eimbu sed d ugs a e negligible i subsidized he apeu ic subs i u es a e
a ailable. Now, wo d ugs, 1 and 2, wi h simila he apeu ic indica ions a e conside ed.
Each d ug is p oduced by a di e en i m ( i m 1 and i m 2). Bo h d ugs a e o pa en and
one is he gene ic subs i u e o he o he 15. The consume pe cei es hem o be di e en
bu ace he same co-paymen , al hough his co-paymen may di e among coun ies. The
d ug 2 is al eady lis ed in bo h he e e enced and he e e encing ma ke s. The wo d ugs
a e ho izon ally di e en ia ed á la Ho elling (see (Ho elling, 1929)). As he independen
p ice nego ia ions lead o a highe p ice in he e e encing coun y, he i m will ejec low
p ices in he e e enced coun ies knowing ha hey will ace a p ice cap in he
e e encing one. Main esul s con inue o hold in his ex ension: ERP bene i s he
e e encing coun y and ha ms he e e enced coun y as well as he i m.
On he o he hand, unde ough h ea s (see oo no e 13), he i m su e s a
ha she punishmen in he case ha nego ia ions ail (d ug is banned). The main esul
wi h weak h ea s emains, i.e., ERP bene i s he e e encing coun y and ha ms he i m,
15 Lobo and Feldman ha e also modelled he ole o adema ks, ad e ising and gene ic names on compe i ion FELDMAN,
R. & LOBO, F. 2013. Compe i ion in p esc ip ion d ug ma ke s: he oles o adema ks, ad e ising, and gene ic names.
Eu opean Jou nal o Heal h Economics, 14, 667-675, LOBO, F. & FELDMAN, R. 2013. Gene ic D ug Names and Social
Wel a e. Jou nal o Heal h Poli ics Policy and Law, 38, 573-597..
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
13
bu he e e enced coun y is no a ec ed by he ERP.
1.2.3 Which ole does PT play in he pha maceu ical ma ke ?
As men ioned ea lie in he sec ion abo e, he i m p e e s i s ly launching i s d ug
in high-p ice coun ies o in luence p ices in o he coun ies o i s ad an age. This
s a egic beha iou may be no so use ul i PT exis s. As Danzon e al. and Danzon and
Eps ein commen ed, o be a pa allel expo e coun y migh in luence posi i ely on p ices
(Danzon e al., 2005, Danzon and Eps ein, 2008). The eason has been clea ly explained
by Rich e (Rich e , 2008). We no e ha Rich e conside s ha PT implies a loss o
income o he i m, since i s ops selling a ce ain amoun a a highe p ice han in he
absence o PT, he au ho includes PT as a iable in o objec i e unc ion i m. As
expec ed, in o de o compensa e his loss o income, he i m will ha e o inc ease he
p ice. The au ho p oposes a ma hema ical op imiza ion p oblem o he i m, conside ing
ha he i m sells a d ug ac oss all coun ies and o e all ime pe iods and he lowes
p ice coun y will be he pa allel expo e one. This is quan i ied as he sum o he
di e ences be ween he lowes p ice among all coun ies and he p ice s a ed in each
coun y, mul iplied by he quan i y los 16 in he pa allel impo e coun y.
Gansland and Maskus (Gansland and Maskus, 2004) go u he and, no only
conside PT as po en ial loss o income o he o iginal manu ac u e , bu also in es iga e
how PT i ms beha e and how he p esence o PT a ec s equilib ium p ices in he pa allel
impo e (PI) and expo e coun ies. They de elop a simple model o pa allel impo s in
which an o iginal manu ac u e compe es in i s home ma ke (Sweden) wi h PI i ms and
all i ms se p ices simul aneously. The au ho s suppose ha he quan i y o ade is
exogenously gi en, and i is all sold. This idea is suppo ed by he ac ha in high-p ice
ma ke s he PI quan i y a ely exceeds 10%, excep in a ew majo p oduc s. A model o
wo coun ies is conside ed, wi h a high-income and a low-income coun y. The high-
income coun y is un egula ed, he low-income coun y has a p ice cap se by he
go e nmen and he d ug sold is an on-pa en d ug wi hou subs i u es. A ma ginal and a
ixed cos o engaging in PT also exis . Gi en he quan i y chosen by PI i ms, he p o i -
maximizing p ice is calcula ed. The au ho s compa e a model unde a limi ed PI quan i y
16 The quan i y los is calcula ed mul iplying he ma ke sha e los by he quan i y sold in he pa allel impo e coun y.
14 Modeling Global P icing and Launching o New D ugs
o ano he model ha allows unlimi ed PI quan i y. Unde an unlimi ed quan i y o PI, an
“a bi age- ee” p ice a ises, and consequen ly a p ice con e gence om he high-income
coun y o he low-income one. Howe e , unde a limi ed quan i y o PI, he mos eal case
as s a ed abo e, he manu ac u ing i m has an incen i e o adap o PT a he han o se
an a bi age- ee p ice in i s ma ke . In his case, he equilib ium p ice o high-income
coun y con e ges o he low-income one plus a a iable ade cos 17. As expec ed, he e
is an e ec o compe i ion, whe eby he equilib ium p ice in he high-income coun y alls
in he numbe o PI i ms. Also, he au ho s iden i y ha he equilib ium numbe o PI i ms
inc eases in he size o he ma ke bu dec eases in he low-income coun y p ice and in
he ixed and a iable ade cos .
1.2.4 How asymme ic in o ma ion on quali y o d ugs may a ec
d ug p icing and launching?
One o he ac o s in luencing p icing and launching conside ed by Danzon e al.
and Danzon and Eps ein, has been he ICER ha a ec s posi i ely d ug p ices (Danzon
e al., 2005, Danzon and Eps ein, 2008). This measu e may be in e p e ed as a
p ice/quali y indica o o he d ug. In he li e a u e, unde di e en hypo heses, we can
obse e ha in o ma ion abou quali y ma e s. Two pape s ha e conside ed he
asymme ic in o ma ion abou ei he he quali y d ugs o he demand o quali y d ugs, o
analyse he p icing and launching d ugs (A ella e al., 2012, Ga cía-Ma iñoso and Oli ella,
2012). A ella e al. (A ella e al., 2012) p opose a model o asymme ic in o ma ion on he
quali y o he d ugs, o ind ou how wo ypes o egula o y egimes, one ocused on
quali y and ano he on p ice con ol, a ec d ug p ices, and u he mo e, how he p ice
egula ion a ec s, ul ima ely, he quali y o he d ugs. On he o he hand, Ga cía-Ma iñoso
and Oli ella (Ga cía-Ma iñoso and Oli ella, 2012) p opose a sequen ial launching and
analyse how he in o ma ional spillo e s, issued om he asymme ic in o ma ion abou
he quali y d ugs, a ec s he d ug p icing. The in o ma ional spillo e s a e de ined as he
claim o lowe p ices by one coun y, gene a ed by he knowledge abou lowe p ices in
o he coun ies. The no ion ha low p ices may o e spill o o he coun ies e en in he
absence o PT o ERP is he e in oduced, di e en om he p e ious esea ch.
A ella e al. (A ella e al., 2012) compa e wo egula o y egimes. Unde he i s
17 The p ice in he low-income coun y is aken as gi en.
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
15
egime he go e nmen ixes a minimum e icacy s anda d (MES), his egime
co esponds o he egula o y s uc u e o he pha maceu ical ma ke in he Uni ed S a es.
Unde he second egime, in addi ion o a MES, he go e nmen ixes a p ice cap (PC);
his egime co esponds o he s uc u e in many o he coun ies as I aly. The model
conside s wo coun ies ha di e in hei demand o d ug e icacy, and one i m, which
p oduces wo ypes o d ug, low and high-e icacy, bu i canno dis inguish be ween high
and low- ype buye s. Rega ding he pape o A ella e al. (A ella e al., 2012), his pape
in oduces pa ien co-paymen s, he discoun ’s ac o on p o i s and on quali y om he
coun y.
Whe he he go e nmen ixes a MES ha binds he low e icacy d ug, hen he
e icacy and he p ice o he low e icacy d ug inc eases o mee he MES and o co e he
d ug ma ginal cos . Ins ead, he e icacy o he high-e icacy d ug is no a ec ed and i s
p ice may be lowe . Howe e , i he egula ion imposes oo high e icacy s anda ds, low-
ype buye s would be excluded o m he ma ke . The e exis s a minimum e icacy
h eshold ha op imally balances he highe R&D cos s wi h he highe e icacy d ugs
deli e ed o low- ype buye s. This op imal le el is jus below he le el ha excludes low-
ype buye om he ma ke .
Whe he a PC ha binds on he high- ype d ug is conside ed (bu no on he low-
ype d ug), he i ms espond p oducing high- ype d ugs lowe in quali y a a p ice
co espondingly lowe . Unde bo h egimes, he e icacy and he p ice o he low e icacy
d ug inc eases o mee he MES and o co e ma ginal cos espec i ely, howe e , he
e icacy o he high- ype d ugs may be unde mined i he go e nmen ixes a PC ha
binds he high- ype d ugs, bu also ha he consume o high- ype d ugs will sa e money.
Finally, he ne wel a e will depend upon how binding is he p ice con ol and on he
ela i e size o he wo g oups o buye s.
I has been shown abo e how di e en ypes o egula ion may a ec on d ug
quali y and p icing unde asymme ic in o ma ion abou he buye s. Ga cia-Ma iñoso and
Oli ella (Ga cía-Ma iñoso and Oli ella, 2012) assume he asymme ic in o ma ion on he
o he side. The coun ies (i.e. he buye s) do no know abou he ype o he i m, hus, he
i m may be o high o low quali y, and his in o ma ion is in he hands o he i m. As he
au ho s p opose a sequen ial launching, he coun ies will ha e hei p io belie s and

16 Modeling Global P icing and Launching o New D ugs
he e may exis in o ma ional spillo e s. Thus, whe he a low p ice is ixed in he coun y
whe e he d ug is i s launched, his e eals p i a e in o ma ion ( he quali y o he i m
conce ning he p oduc ion and dis ibu ion cos s) o subsequen playe s conce ning he
p ice, and he e o e, he ollowing coun ies will also demand o low p ices. Now, low
p ices may o e spill o o he coun ies e en in he absence o PT o ERP. The eason is
ha coun ies ha would in p inciple make gene ous p ice o e s whe he obse e he i m
accep ing a low p ice elsewhe e, hey migh change hei mind and become agg essi e.
Along a dynamic game, now i is he i m, which accep s o ejec s he o e om he
coun y hus, he game is based on a “ ake i o lea e i o e ”. Coun ies may be
agg essi e o non-agg essi e18.
Acco ding o he i m s a egic beha iou , al hough in o ma ion spillo e s can be
a oided by launching in all coun ies simul aneously, he i m will p e e o delay i ( om
mo e o less expec ed) (i) he i m is su icien ly pa ien (high discoun ’s ac o 19); (ii) he
agg essi e coun y has a su icien popula ion; (iii) co-paymen s di e enough ac oss
coun ies; and (i ) coun ies ha e ela i ely pessimis ic p io s on quali y. In e es ingly, he
au ho s p esen a coun e a gumen o he s a emen ha delay only occu s in small
coun ies, hus i could happen ha he coun y ha su e s delay is he la ges in size (as
long as he es o he ac o s men ioned go in he igh di ec ion) (Ga cía-Ma iñoso and
Oli ella, 2012).
1.2.5 A e impo an he headqua e s loca ion and he con ac s
among i ms when p icing d ugs?
Di e en om o he s udies, Cab ales and Jiménez-Ma ín (Cab ales and
Jimenez-Ma in, 2007) conside ha he i m is loca ed in he coun ies analysed. The e
a e wo coun ies, one o hem egula es p ices unde ERP ( he e e encing coun y) and
he o he does no ( he e e enced coun y). One o he main con ibu ions is ha he i m
p o i s a e now maximized oge he wi h he consume su plus by he egula o . The
au ho s compa e he maximizing p ice in wo si ua ions, when headqua e s a e loca ed in
18 The agg essi eness will posi i ely depend on he co-paymen ; he la ge is he co-paymen , he mo e agg essi e he
p ice o e s will be. We al eady men ioned ha i a coun y has obse ed a low p ice accep a ion in a p e ious coun y, i will
upda e i s belie s and become agg essi e. The hi d ac o is dynamic and o wa d-looking. Being agg essi e oday may
lead he i m o ejec he p ice o e in o de o a oid he agg essi eness o u u e agencies.
19 The discoun ac o is he ac o by which a u u e cash low mus be mul iplied in o de o ob ain he p esen alue. The
highe he discoun ac o is, he g ea e he p esen alue is assessed.
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
17
he egula ed coun y o in he un egula ed one. This heo e ical model p edic s ha he
p ice se by he egula o is sligh ly highe o he local mul ina ional han o he o eign
one, and as he size o he e e enced coun y g ows wi h espec o he e e encing
coun y, he p ice o he o eign mul ina ional con e ges o he local mul ina ional one. In
his ega d, i we assume ha he coun ies ha a e s ic e egula o s we e ela i ely
small in size; his would imply ha hey could no in luence subs an ially he p ices in
a ou o he local mul ina ionals.
No only he loca ion o i ms ma e s, bu also he con ac s be ween i ms
compe ing in he same ma ke s. The mul ima ke con ac heo y implies ha mo e
con ac s be ween i ms compe ing in he same ma ke s may induce mo e collusion. This
collusion suppo p ices abo e he equilib ium p ices. A his ega d, Co onado e al.
(Co onado e al., 2007) y o p edic he e ec o mul ima ke con ac s uc u e on he
equilib ium p ices unde wo di e en egimes, he p ice egula ion and he ee p icing,
and ul ima ely o know i p ice egula ion may a ec mul ima ke con ac e ec . Fo his,
he au ho s p opose a game in ini ely epea ed whe e p ices a e se simul aneously. The
i ms can collude and suppo p ices abo e he equilib ium p ices. In case o de ia ion,
he i ms will be penalized e e ing o he equilib ium p ices. I is also supposed ha he
maximum sus ainable p ice (in collusion) depends posi i ely on he discoun ac o , i.e.,
he u u e p o i s a e mo e aluable, and he e o e he sho un bene i s om de ia ion
a e acco dingly less p e e ed. Taking in o accoun he hypo hesis abo e desc ibed, he
model p edic s ha he e ec o mul ima ke con ac s uc u e inc eases he equilib ium
p ices bu his e ec is unde mined in egula ed coun ies.
In summa y, he heo e ical models p edic ha bo h he i m loca ion and he
mul ima ke con ac no only a ec d ug p icing bu also, hei e ec s depends on p ice
egula ion egimes.
1.2.6 Which e ec s do a ise in p icing and inno a ion when
coun ies apply in e nal RP20?
Danzon and Eps ein (Danzon and Eps ein, 2008), ex ending he pape o Danzon
20 In e nal e e ence p icing, as opposed o he ERP, compa es p oduc p ices wi hin a single coun y.
18 Modeling Global P icing and Launching o New D ugs
e al. (Danzon e al., 2005), hey conside as in luencing ac o he RP as egula o y
egime being expec ed o posi i ely a ec on p ices. A his ega d, wo pape s examine,
on he one hand, how he RP policy a ec s he equilib ium p ices and he i m p icing
s a egies (Mi aldo, 2009), on he o he hand, how he RP policy in luences he in ensi y o
esea ch and he in oduc ion o new pionee in he ma ke (Ba dey e al., 2010). Bo h
pape s compa e he ou pu s unde no egula ion and unde RP policy. Fu he mo e, o he
au ho s ha e deeply s udied he RP policy om an in e na ional pe spec i e (Lopez-
Casasno as and Puig-Junoy, 2000, Puig-Junoy, 2010a, Puig-Junoy, 2010b) and
pa icula ly he Spanish case (Mes e-Fe andiz, 2003b, Mo eno-To es e al., 2009, Puig-
Junoy, 2007)
The model de eloped by Mi aldo (Mi aldo, 2009) conside s wo pha maceu ical
i ms, a con inuum o consume s uni o mly dis ibu ed and a ma ke o d ugs ho izon ally
di e en ia ed à la Ho elling (Ho elling, 1929). Each i m p oduces wo dis inc a ie y o
d ug. Each consume is assumed o ha e a mos p e e ed d ug ha is gi en by he
loca ion on he line segmen . Indeed, he cons an ma ginal cos o dis ance is he loss in
u ili y incu ed by a consume .
Mi aldo s udies he explici RP o mula ions and conside s a di e en iming o
implemen a ion o he policy. The au ho analyses a ini e dynamic game, in which
duopolis s compe e by non-coope a i ely se ing p ices in wo subsequen pe iods. In he
i s s age, he wo pha maceu ical i ms se he p ices o one a ie y. A he beginning o
he second, he go e nmen ixes he RP le el, and hen, he i ms se p ices o he
second a ie y o d ugs. The e o e, a he las s age, he i ms’ p o i s depend ia demand
no only on he p icing s a egies bu also on he RP le el ixed p e iously by he
go e nmen .
The RP policy is in oduced as eimbu semen scheme21. Mi aldo shows ha
unde he RP policy, he equilib ium p ices a e a leas as high as he equilib ium p ices
wi hou RP. As a main con ibu ion, when bo h RP ules a e compa ed, he minimum and
he weigh ed a e age, he au ho s a es ha i m se highe p ices a i s s age when he
21 In coun ies, such as Ge many and Spain, whe e pha maceu icals a e eimbu sed h ough a RP sys em, pa ien s a e
ypically eimbu sed a lump sum amoun o any homogeneous pha maceu ical clus e , independen ly o he d ug a ie y
bough . The e a e se e al c i e ia o clus e d ugs and he eplica ed model applies o coun ies ha use chemical and
he apeu ic c i e ia. The i s c i e ion clus e s d ugs wi h he same ac i e ing edien and he e o e e e s o pa en expi ed
d ugs. The second c i e ion clus e s d ugs ha ha e he same he apeu ic unc ion and he e o e, wi hin he same clus e
one can ind pa en p o ec ed d ugs.
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
19
RP is calcula ed as a weigh ed a e age han when he minimum policy is applied. The
i m p icing s a egy in he second pe iod (when al eady he RP le el has been ixed) will
depend on he weigh s o each p ice. I hey a e high (i.e., o su icien ly high and low
alues o he weigh s), he minimum policy makes i ms o ix lowe p ices han he
weigh ed a e age ule. Bu i he weigh s a e simila , esul s will be ambiguous; hey will
depend on consume s’ p e e ences, on he deg ee o ho izon al di e en ia ion and on he
discoun ac o . Anyway, in a symme ic ma ke , in o de o a oid highe p ices, he
egula o should implemen a policy whe e he e e ence p icing consis s o he minimum
obse ed p ice.
In u n, Ba dey e al. (Ba dey e al., 2010) e alua e he long un impac o RP on
pha maceu ical inno a ion and on heal h expendi u e. The pape is based on a dynamic
model wi h h ee playe s: he i ms (inno a o s/p oduce s), he egula o and he
consume s22. Bo h ho izon al and e ical di e en ia ions a e conside ed. Ve ical
di e en ia ion has di e en le els ( he apeu ic classes). To simpli y, he e a e wo le els,
C and N, designing espec i ely cu en and new. Pa ien s ob ain u ili y om he ea men
o he d ug, pe cei e i s side e ec s, pay a p ice o he d ug and ecei e a
eimbu semen . They also assume ha d ugs a e p oduced a ze o cos and when a d ug
is in oduced in he le el N, p oduce s o le el C ha e no sales. The p ice nego ia ions a e
de eloped à la Rubins ein (see Rubins ein (Rubins ein, 1982)). The i ms choose he le el
o esea ch in es men , and hen nego ia e in oduc o y p ices o new d ugs wi h he
egula o . The inno a ion p ocess is de e minis ic and can disco e a new p oduc ei he
in he same le el as exis ing p oduc s (ho izon al inno a ion o ollowe ), o in a supe io
le el ( e ical inno a ion o pionee ). The e exis a cos o b inging an inno a ion.
Ba dey e al. au ho s compa e how he dynamic o inno a ion beha es wi hou RP
and unde RP. Thus, in e ms o delay o in oduc ion, he applica ion o RP yields he
delay o ollowe s, and he delay o pionee s i only i he p ice is abo e some h eshold.
In he long un, allowing inno a ion o occu in le el C (p io o he disco e y o he i s
le el N d ug), he ollowe may be ne e in oduced (sho sequence), o i can be
in oduced be o e he pionee o le el C (long sequence). Again, bo h sequences a e
22 The ela ion be ween he pa ien and he physician is conside ed a ela ion o pe ec agency; he e o e hey a e iewed
as a single agen , he consume .
26 Modeling Global P icing and Launching o New D ugs
coun ies, as men ioned abo e, he elas ici y is signi ican ly posi i e. This is consis en
wi h he hypo hesis ha in oducing line ex ensions is one means o achie ing a p ice
inc ease in coun ies ha do no pe mi highe p ices o es ablished p oduc s. The p ice
elas ici y wi h espec o he numbe o o ms is la ges in Japan, which p esen s he
g ea es p ice educ ion o e he p oduc li e cycle and hence whe e he e exis s ong
incen i es o in oduce new o ms and hus ob ain a highe p ice (Danzon and Chao,
2000). In addi ion, Danzon and Chao show ha he global di usion28 as an indica o o
he apeu ic alue ob ains highe p ices in un egula ed ma ke s, al hough his e ec is
insigni ican o small a bes in less egula ed ma ke s such as he UK, Canada and
Ge many, bu is signi ican ly nega i e in s ic ly egula ed coun ies such as F ance, I aly
and Japan (Danzon and Chao, 2000). Besides, Co onado e al. show ha he p oduc
ma ke sha e is expec ed o be signi ican ly posi i e only in he leas egula ed coun ies.
1.3.2.2 Compe i ion and subs i u es
In gene al, he apeu ic subs i u es do no appea o exe compe i i e p essu e on
p ice. Danzon and Chao (Danzon and Chao, 2000) show ha , compe i ion om
he apeu ic subs i u e molecules appea s o ha e small signi ican nega i e e ec s in
F ance, I aly, Ge many and he UK, bu he in e p e a ion is unclea . The mos plausible
explana ion is ha egula o s use implici e e ence p icing, se ing p ices o new
p oduc s based on p ices o es ablished p oduc s (Danzon and Chao, 2000). Timu e al.
do no ind signi ican e ec on p ice om he apeu ic subs i u es, howe e , di e en om
Danzon and Chao (Danzon and Chao, 2000). Timu e al. es ima e he pool o da a and
do no show speci ic esul s o coun ies. In his case, Timu e al. sugges ha subs i u e
molecules wi h a highe p ice migh no ecei e eimbu semen , and subs i u ion is no
always possible in egula ed coun ies because o p esc ibing o consump ion
p e e ences. Fu he mo e, he e ec o delayed en y o he apeu ic subs i u es has been
also analysed, and he coe icien s ob ained o he USA and Canada imply ha
successi e molecules en e a lowe p ices, howe e hese lowe p ices a e no
su icien ly low o ully e ode he i s mo e ’s ad an age. O he coun y in e ac ions a e
gene ally posi i e, bu signi ican only in Ge many.
On he con a y, only Ve nie s e al. (Ve nie s e al., 2011) ind ha compe i ion
28 The numbe o coun ies whe e he medicine has al eady been sold.

Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
27
d i es down launch p ices when es ima ing he pool o da a. In his case, his di e en
esul can be suppo ed by he use o a iche sample. Fu he mo e, Danzon and Eps ein
collec da a o supe io and in e io medicines. They s a e ha while he p ices o
compe i o s a e posi i ely ela ed o launch p ices, he p ices o in e io medicines do no
a ec hose o supe io ones, and ice e sa, which means ha dynamic compe i ion
be ween subclasses is based on non-p ice p oduc a ibu es (Danzon and Eps ein,
2008).
Fo middle and low income coun ies (MLICs), Lanjouw inds ha compe i ion om
o he o igina o medicines does no appea o be e ec i e a educing p ices in e ail
channels in hese coun ies (Lanjouw, 2005).
Mos pape s ound gene ic compe i ion o nega i ely a ec p ices (Cab ales and
Jimenez-Ma in, 2007, Co onado e al., 2007, Danzon and Chao, 2000, Danzon and
Eps ein, 2008, Danzon e al., 2011, Timu e al., 2011). Howe e , Danzon an Eps ein
(Danzon and Eps ein, 2008) only ind ha he e ec o gene ic p ices in in e io class,
which may e lec a selec ion e ec : la e en an s in in e io subclasses launch only i hey
expec o ecei e high p ices ela i e o compe ing gene ics. In u n, Cab ales and
Jiménez-Ma ín, and Co onado e al. (Cab ales and Jimenez-Ma in, 2007, Co onado e
al., 2007) ind a signi ican and nega i e e ec in a la ge majo i y o coun ies. Howe e ,
he p esence o gene ics on a ma ke does no mean ha b and name p oduc s will
educe hei p ices. Acco ding o Co onado e al., in some cases, he p esence o
gene ics will ha e he impac o concen a ing b and name p oduc s o e he inelas ic
po ion o he demand, which will hen inc ease he p ice o hese p oduc s. Hence, he
expec ed sign o he numbe o gene ics will be posi i e29. In e es ingly, Danzon and Chao
(Danzon and Chao, 2000) ind ha gene ic compe i ion is signi ican in un egula ed o
less egula ed ma ke s bu egula ion unde mines gene ic compe i ion in s ic egula o y
sys ems. As Co onado e al. (Co onado e al., 2007), Danzon and Chao ind posi i e
e ec s o gene ic compe i ion bu hey explain ha mul i-sou ce supplie s in hese
coun ies a e usually licensed co-ma ke e s a he han compe ing gene ic manu ac u e s
o mino ‘‘new’’ p oduc s ha en e o ob ain a highe egula ed p ice. Danzon e al.
(Danzon e al., 2011) ind ha in MLICs he numbe o gene ic compe i o s only weakly
29 The US, Ge many, The Ne he lands, he UK and F ance.
28 Modeling Global P icing and Launching o New D ugs
a ec s p ices o e ail pha macies, may be because unce ain quali y leads o compe i ion
on b and a he han p ice. Con a y, ende ed p ocu emen a ac s mul i-na ional gene ic
supplie s and signi ican ly educes p ices o o igina o s and gene ics, compa ed o p ices
o e ail pha macies.
Only Kana os and Vando os (Kana os and Vando os, 2011) ind ha gene ics is
non-signi ican . They poin ou he exis ence o he gene ics pa adox, pa icula ly in he
US, whe e p ices o o -pa en o igina o b ands do no decline pos -pa en expi y, bu ,
a he , inc ease as e han p ices o in-pa en o igina o b ands.
1.3.2.3 Regula ion Cha ac e is ics
When examining p ice egula ion egimes h ough explici egula ion a iables, he
mos o he li e a u e does no ind signi ican in luences on p ices (Kana os P and Cos a-
Fon J, 2005, Ve nie s e al., 2011, Kana os and Vando os, 2011, Danzon and Eps ein,
2008, A ella e al., 2012). Only Kana os e al. ind ha coun ies wi h “ ee-p icing”
sys ems ( he US and Ge many) p esen highe p ices signi ican ly posi i e. As explici
p ice egula ion policy, hey only ind ha he explici use o HTA (Heal h Technology
Assessmen ) has a signi ican nega i e e ec on p ices (Kana os and Vando os, 2011).
In his sense, A ella e al. p oposes ha a he han highe o lowe p ices, unde a p ice
con ol egime (as in I aly), he e is g ea e p ice a iabili y han in a ee p ice egime (as
in he US) (A ella e al., 2012).
Al hough no in oducing explici egula ion a iables in he econome ic models,
Danzon and Eps ein and A ella e al. (A ella e al., 2012, Danzon and Eps ein, 2008)
e ie e in o ma ion om p ice egula ion cha ac e is ics. In his sense, A ella e al. ind a
posi i e ela ionship be ween quali y and p ice in a ee p ice egime (as in he US), bu
also ind a nega i e ela ionship be ween d ug p ice and d ug quali y in a p ice con ol
egime (as in I aly), which sugges s ha p ice egula ions ha e c ea ed pe e se
incen i es (A ella e al., 2012). Also, Danzon and Eps ein in e es ingly obse e ha launch
p ices inc ease wi h he le el o he lowes p ice p e iously ecei ed in o he high-p ice
EU coun ies, whe eas he e ec s o a p e ious launch in low-p ice EU coun ies a e
insigni ican . This is also ue o non-EU coun ies, bu only o supe io medicines. This
esul is consis en wi h he hypo hesis ha launching i s in high-p ice EU ma ke s can
in luence p ices in low-p ice ones. Tha e idence abou a sequen ial launch p ices
alida es he heo y ha a launch delay in low-p ice ma ke s may ul ima ely yield highe
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
29
p ices in hese ma ke s h ough spillo e s om highe -p ice ones (Danzon and Eps ein,
2008). This heo y is shown by S a ga d and Sch eyögg (S a ga d and Sch eyögg,
2006). They de elop an analy ical model in hei pape , which analyses di ec and indi ec
impac due o he use o ERP om he e e enced o he e e encing coun y. The au ho s
es ima e he impac o d ug p ice changes in Ge many30 on d ug p ices in o he coun ies
using ERP in he o me EU-15. The au ho s use he o mulas applied by each
e e encing coun y and hen, hey calcula e he pa ial di e en ial o hese o mulas wi h
espec o a 1 eu o p ice educ ion in Ge many. They do no only know he o mula ( he
a e age, he minimum, a pe cen age, e c. see Leopold (Leopold e al., 2012)) bu also he
baske o coun ies used by each e e encing coun y. They di e en ia e he di ec impac
(caused by Ge many o o he coun ies) and he indi ec impac (caused by he
e e encing coun ies ha ha e aken Ge many in hei baske s, o o he e e encing
coun ies). The au ho s s a e ha he ela ionship be ween he di ec and indi ec impac
o a p ice change depends mainly on he scheme applied o se p ices. Fo ins ance, he
p ice is ei he de e mined by he lowes o o eign p ices (e.g. Po ugal), he a e age o
o eign p ices (e.g. I eland) o a weigh ed a e age o o eign p ices (e.g. I aly). I he
espec i e d ug is ma ke ed in all e e enced coun ies and p ices a e egula ly upda ed, a
p ice educ ion o €1.00 in Ge many will educe p ices in he o me EU-15 coun ies om
€0.15 in Aus ia o €0.36 in I aly. Whe he we dis inguish be ween di ec and indi ec
impac , almos mo e han (abou ) he 50% o he o al impac in Aus ia comes om he
indi ec impac (0.08 eu os), howe e , only 1% o he o al impac in I aly is due o he
indi ec impac (0.03 eu os). Bo h Aus ia and I aly include 14 and 12 coun ies in hei
ERP scheme espec i ely, bu we obse e how I aly, which uses a weigh ed a e age, is
less ha med by undue indi ec impac . Thus, o a oid he nega i e e ec s o ERP and
de e mine p ices in o de o educe he di ec and indi ec impac o indi idual coun ies, a
weigh ed o mula o p ices con aining as many coun ies as possible should be used.
Su p isingly, he li e a u e does no ind a di ec e ec on p icing when ERP is
applied. This esul is obus among di e en speci ica ions, bu i could be explained by
he da abase collec ed. In Ve nie s e al. (Ve nie s e al., 2011), he da abase was limi ed
o he d ugs launched as o Feb ua y 1994 bu he ERP had no been widely applied by
30 Ge many is chosen because is one o he la ges pha maceu ical ma ke s in he wo ld, i is cha ac e ised by ela i ely
high p ices and i is e e enced by mos e e encing coun ies scheme.
30 Modeling Global P icing and Launching o New D ugs
ha ime (Leopold e al., 2012). In Kana os and Vando os (Kana os and Vando os,
2011), he p ices o b anded o igina o s do no co espond o launch p ices; he e o e,
compe i ion can also lead o downwa d p essu e o o -pa en o igina o b ands, bu he
ERP is applied on new d ugs a launch ime. On he o he hand, when he ERP is ound o
ha e e ec on launching, he da abase comp ises p ices a launch ime be ween 1995
and 2005.
E en mo e, Kana os and Cos a-Fon ha e ound ha PT does no in luence p ices
downwa ds in impo ing coun ies (Kana os P and Cos a-Fon J, 2005). Fu he mo e,
Danzon and Eps ein show ha he p esence o PT is insigni ican o supe io medicines,
bu signi ican and nega i e o in e io ones, indica ing ha he p esence o PT educes
launch p ices mainly o la e en an s in o olde subclasses (Danzon and Eps ein, 2008).
Fu he mo e, Danzon e al. make an in e es ing con ibu ion o MLICs conce ning
p ice egula ion policies. They compa e he wo di e en ways o p o ide medicine in
MLICs: ende ed p ocu emen mechanism by NGOs and s anda d e ail channels. In
hese e ms, hey ind ha o igina o b ands pu chased h ough ende ed p ocu emen
mechanisms by NGOs end o lowe o igina o p ices, compa ed o hose ob ained
h ough s anda d e ail channels. These la ge p ocu emen e ec s may e lec no only
p ice-compe i i e ende ing bu also a g ea e willingness o o igina o s o g an discoun s
o a sepa a e dis ibu ion channel ha a ge s lowe income cus ome s and is less p one
o p ice spillo e s o o he coun ies (Danzon e al., 2011).
1.3.2.4 Coun y Cha ac e is ics
In mos s udies, coun y cha ac e is ics a e included in he econome ic models in
o de o cap u e p ice di e ences among coun ies. The a iable mos commonly included
is ha o coun y GDP pe capi a (Cab ales and Jimenez-Ma in, 2007, Danzon and
Eps ein, 2008, Bo ell, 2007, Danzon e al., 2011). This is cohe en wi h he gene ally
accep ed posi i e ela ionship be ween weal h and g ea e willingness o pay. Thus, he
coun y-le el ixed e ec s con olled by GDP pe capi a show ha al hough he lowe -
income EU coun ies such as Spain, Po ugal and G eece egula e medicine p ices, hey
ha e ela i ely high medicine p ices wi h espec o GDP, whe eas highe -income EU
coun ies ha e lowe medicine p ices ela i e o hei pe capi a GDP. Then, Danzon and
Eps ein sugges ha ERP has con ibu ed o he p ice con e gence o medicines among
EU coun ies ela i e o GDP (Danzon and Eps ein, 2008). In u n, Kana os and
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
31
Vando os (Kana os and Vando os, 2011) s a e ha as we mo e owa ds newe
molecules o e ime by launch da e, he e is upwa d p ice con e gence ac oss he s udy
coun ies o e all. This is pa ly explained by ERP and he launch sequence o new
p oduc s, whe eby new p oduc s a e i s launched in less- egula ed coun ies ollowed by
p ice- egula ed coun ies. This launch sequence in luences in pa he inal p ice in p ice-
egula ed coun ies. Acco ding o he con e gence abo e men ioned, Cab ales and
Jiménez-Ma ín (Cab ales and Jimenez-Ma in, 2007) obse e ha he US does no
p esen highe p ices han o he coun ies. Con a y o he con en ional wisdom abou
coun ies egula o y egimes, he ixed e ec o he US is signi ican ly lowe han ha o
Canada (bu no o all speci ica ions), F ance o I aly. The au ho s in e p e ha i
a e age p ices in he US a e highe han in o he coun ies i is no because o he
coun ies engage in “ ee iding egula ion”, bu because he US pe -capi a income is
highe , in ac , in many cases, he US pays less, no mo e, han coun ies o simila
income o lowe income (such as Eas e n Eu opean coun ies). The au ho s also in e p e
ha as he US ma ke size is la ge and mo e compe i i e han he o he coun ies, i
p o ides wi h some p o ec ion wi h espec o simila ly ich coun ies. This con adic o y
esul may be esul o he inno a ions in oduced wi h espec o he p e ious empi ical
li e a u e in he subjec . Mainly, hey es ima e p icing equa ions sepa a ely o each
coun y, hey do no es ic he sample in any way and iden i y he e ec o ime-in a ian
a iables by ollowing a wo-s age p ocedu e (Cab ales and Jimenez-Ma in, 2007).
Fo MLICs, Bo ell shows e idence o a pe sis en posi i e ela ionship be ween
d ug p ices and pe capi a income in MLICs (Bo ell, 2007). In addi ion, he income
dis ibu ion wi hin coun ies has also been conside ed by Danzon e al. and Bo ell
(Bo ell, 2007, Danzon e al., 2011). Bo ell sugges s ha income e ec s alone a e
unlikely o achie e a o dable p ices in low-income coun ies. Thus, al hough pe capi a
income e ec s a e posi i e, he nega i e e ec o he income dis ibu ion implies ha he
poo es coun ies ace wi h he highes ela i e p ices. Mo eo e , skewed income
dis ibu ions appea o exace ba e high d ug p ices ela i e o pe capi a incomes in
MLICs (Danzon e al., 2011), howe e , Bo ell epo s ambiguous e ec s (Bo ell, 2007).
1.3.2.5 Fi m Cha ac e is ics
Some s udies ha e examined how ce ain i m cha ac e is ics may in luence p ices

32 Modeling Global P icing and Launching o New D ugs
(Cab ales and Jimenez-Ma in, 2007, Co onado e al., 2007, Danzon and Eps ein, 2008,
Kana os and Vando os, 2011, Ve nie s e al., 2011). Co onado e al. show ha he i m
size ha e a high signi ican posi i e e ec on p ices, indica ing ha la ge companies enjoy
highe p ices ei he because i s p oduc s a e o highe quali y o pe cei ed as such
(Co onado e al., 2007). Howe e , Cab ales and Jiménez-Ma ín ind his e ec signi ican
and nega i e, bu small (Cab ales and Jimenez-Ma in, 2007). As he de ini ion a iable
and he sample is he same, excep he numbe o coun ies conside ed, we hink ha he
use o mo e a iables conce ning he i m cha ac e is ics may unde mine his issue.
Ano he ques ion widely men ioned conce ns he ype and loca ion o he i m.
Danzon and Eps ein do no ind p ice p emium o medicines launched by local i ms
(Danzon and Eps ein, 2008). In addi ion, Cab ales and Jiménez-Ma ín i s ly check
di e ences in loca ion o mul ina ional i ms be ween local and o eign mul ina ionals, and
con a y o con en ional wisdom, in mos coun ies, is ound ha coun ies do no
dis inguish be ween local and o eign mul ina ionals. Howe e , Ve nie s e al., di e en
om o me esul s, ind ha i ms ob ain highe launch p ices in hei domes ic ma ke
han hey do in o eign ones (Ve nie s e al., 2011). These di e en indings may be due o
di e ences in samples. Ve nie s e al. ha e collec ed a iche sample o coun ies whe e
we can ind mo e MLICs whe e usually headqua e s a e no held, while he mos o
coun ies collec ed by Cab ales and Jiménez-Ma ín and Danzon and Eps ein a e middle
and high-income coun ies (Cab ales and Jimenez-Ma in, 2007, Danzon and Eps ein,
2008). Addi ionally, Cab ales and Jiménez-Ma ín also compa e be ween local non-
mul ina ional i ms and mul ina ional i ms and ind ha local non-mul ina ional i ms end
o ha e lowe p ices han any mul ina ional (Cab ales and Jimenez-Ma in, 2007).
Fu he mo e, as in he heo e ical models, Co onado e al. analyse he exis ence
o a mul ima ke con ac e ec and y o ind whe he mo e con ac s be ween i ms
compe ing in he same ma ke s may induce mo e collusion. They empi ically show ha
mul ima ke con ac s ha e a posi i e in luence on p ices o he i m in less egula ed
coun ies31, and uns able e ec s in egula ed coun ies32. This sugges s ha in mo e
egula ed ma ke s he e exis dis o ions ha in e ac wi h ma ke o ces. Fo ins ance, he
p oduc ma ke sha e is signi ican and posi i e only in he leas egula ed coun ies.
31 The US, Canada, Ge many, he Ne he lands and he UK.
32 F ance, Spain, I aly and Japan.
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
33
Mo eo e , educing p ices in mo e compe i i e ma ke s compa ed o he exis ing le el
may discou age en y and may ha e a nega i e dynamic e ec in he de elopmen o he
indus y. This may help p edic he undesi able e ec s o public in e en ions (Co onado
e al., 2007).
1.3.3 Fac o s in luencing launching
1.3.3.1 D ug Cha ac e is ics
As expec ed, po en ial p ices and olumes posi i ely a ec launching (Danzon e
al., 2005), al hough acco ding o Danzon and Esp ein, he olume is no signi ican
(Danzon and Eps ein, 2008). This insigni ican e ec o olume con as s wi h signi ican
posi i e e ec s in Danzon e al. (Danzon e al., 2005). These di e en indings may e lec
di e ences in sample coun ies and d ugs, in addi ion o he use o mo e de ailed
measu es o coun y-class p ices and o he cha ac e is ics. As p e iously ound o p ices,
he speed o launch inc eases wi h he medicine’s impo ance33 bu alls wi h i s age (Kyle,
2007). Fu he mo e, peculia i ies ha e been ound depending on he ype o medicine
e alua ed. Thus, he e a e signi ican di e ences among he apeu ic classes (Danzon e
al., 2005, Ve nie s e al., 2011), o ins ance, he e is a highe p obabili y o ne e
launching o in e io medicines han o supe io ones (Danzon and Eps ein, 2008).
In e es ingly, Ve nie s e al. obse e an U-shaped e ec o launch p ice on he
launch delay; launch p ice is highes a mode a e launch delay. As expec ed, a e y long
launch delay, we will expec a ela i ely low launch p ice as a p elude o gene ic
compe i ion. This ela ionship shows he ade-o o he pha maceu ical i m be ween he
p ice and he launch delay (Ve nie s e al., 2011).
Rega ding he in e na ional con ex , om he i s global launch he launch haza d
pa e n is i s dec easing and hen inc easing34. This idea is gene ally accep ed; he e is
a h eshold a which he i m will no be wo ied abou he spillo e e ec s om he ERP
and he p esence o PT. On he o he hand, he numbe o coun ies whe e he medicine
has al eady been launched a ec s signi ican and posi i ely, wi h he excep ion o p io
33 D ug’s sha e o Medline ci a ion o he apeu ic class.
34 Quad a ic e ec .
34 Modeling Global P icing and Launching o New D ugs
launch in he h ee lowes p ice EU coun ies, Spain, Po ugal and G eece. This pa e n
con i ms ha i ms delay he launch in low-p ice EU coun ies un il i has aken place in
highe -p ice ones (al hough his is no he case o in e io medicines). Fu he mo e, a
p io launch in a high-p ice coun y has a s onge e ec on launching in a low-p ice one
han ice e sa. This e ec is e en la ge when he coun ies a e bo h EU membe s
(Danzon and Eps ein, 2008).
1.3.3.2 Compe i ion and subs i u es
Wi h espec o ma ke ac o s, he launch haza d is posi i ely a ec ed by
compe i o p ices (Danzon and Eps ein, 2008, Kyle, 2006, Kyle, 2007), al hough i does
no seem o in luence when he endogenous a iable is s udied as launch window ins ead
o launch haza d (Ve nie s e al., 2011). These di e ences may come om he de ini ion
o he a iable o measu e he compe i ion. Ve nie s e al. cons uc a He indahl–
Hi schman index35 o each d ug in each coun y, howe e , Danzon and Eps ein use he
compe i o p ices and Kyle includes he measu e by Djanko e al. (Djanko e al., 2002).
Fu he mo e, c oss-class p ice e ec s a e insigni ican , indica ing ha compe i ion occu s
wi hin subclasses a he han be ween subclasses, and ha dynamic compe i ion is d i en
by p oduc cha ac e is ics o he han p ice (Danzon and Eps ein, 2008).
Conce ning gene ic compe i ion, Danzon and Eps ein (Danzon and Eps ein, 2008)
ind ha he e ec s o numbe o gene ic compe i o s a e nega i e bu s a is ically
insigni ican , p o iding u he e idence ha a ailabili y o olde , cheape gene ic
subs i u es is no a signi ican de e en o he launch o new b and p oduc s, e en in olde
subclasses whe e gene ics a e mo e nume ous, possibly because gene ic subs i u ion is
mos ly wi hin molecules a he han be ween molecules.
1.3.3.3 Regula ion Cha ac e is ics
The ques ion o egula ion has been widely analysed. The mos common inding is
ha p ice egula ion ends o p oduce a launch delay (Danzon e al., 2005, Kyle, 2006,
Kyle, 2007, Lanjouw, 2005). When p ice egula ion educes p ices below he le el
expec ed gi en a coun y’s pe capi a income, his p oblem is exace ba ed and launch
35 This index is cons uc ed by summing he squa ed ma ke sha es (MS) (based on e enues in he IMS Heal h da a) o he
m d ugs in he same ATC4 ca ego y as d ug i a he ime o launch o d ug i in coun y j.
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
35
delays may ex end e en o high-income coun ies. Con olling o expec ed p ice, he
models show ha such delays ha e been obse ed in coun ies wi h s ic egula ion and
in hose adi ional pa allel expo e s, ei he pe o ming nega i e coun y ixed e ec
(Danzon e al., 2005), o di ec ly in oducing dummy a iables conce ning p ice egula ion
(Kyle, 2006, Kyle, 2007, Lanjouw, 2005) o ia he a e age p ice compe i o s in he
coun y (Danzon and Eps ein, 2008).
Va ious deg ees o egula ion ha e been explo ed. Lanjouw (Lanjouw, 2005)
examine sepa a ely high and low-income coun ies. Fo high-income coun ies, all p ice
egula ion – whe he mode a e o ex ensi e – ends o educe he p obabili y o a
medicine being launched wi hin wo yea s a e i s launch, while o lowe -income
coun ies, ex ensi e p ice con ols clea ly lowe he p obabili y o new medicines eaching
consume s quickly. On he o he hand, mode a e p ice con ol does no appea o ha e a
signi ican in luence on en y in his case. Howe e , as in he iche coun ies, he e ec o
mode a e p ice egula ion depends on a coun y’s income le el. Fo example, in low-
income coun ies he exis ence o a na ional o mula y posi i ely a ec s launch haza d
(which is no he case o he highe -income coun ies). One would expec i s di ec e ec
o be nega i e, bu wi hin he lowe -income coun y g oup his a iable may be ac ing as a
p oxy o bu eauc a ic compe ence.
On he o he hand, Kyle (Kyle, 2007) and Ve nie s e al. (Ve nie s e al., 2011)
show ha en y ac ually appea s mo e likely in coun ies using in e nal RP. Bo h pape s
show ha di ec p ice con ols a e no signi ican ac o s on launch delay o do a ec
nega i ely he d ug launch haza d. The e o e, he e is some e idence ha indi ec
con ols may be p e e able o di ec ones om he s andpoin o a ac ing new medicines.
Kyle (Kyle, 2007) also sugges s ha he e ec o p ice con ols is no es ic ed o
an indi idual ma ke , bu a ec s he launch o a medicine in o he ma ke s as well. F om
his idea, Heu e e al. (Heue e al., 2007) in oduce he ERP as explana o y ac o and
ind ha coun ies using ERP o de e mine hei p ices p esen a signi ican ly lowe
p obabili y o launch wi hin he i s eigh mon hs. Speci ically, he wo o ms o
in e na ional compa ison based on a o mula o o eign p ices – ei he de e mining p ices
di ec ly om such a o mula, o using i as an in o mal basis o he decision – ha e
impac s ha a e nega i e bu di e en . This di e ence may de i e om he ac ha , i a
42 Modeling Global P icing and Launching o New D ugs
(Gansland and Maskus, 2004, Rich e , 2008). Fu he mo e, in he p esence o
in o ma ional spillo e s on quali y, he launch delay should occu in he non-agg essi e
coun y. Rega ding he i m loca ion, mul ina ional i ms should se led down in la ge
coun ies in o de o be e compe e wi h local mul ina ional i ms (Cab ales and Jimenez-
Ma in, 2007).
On he o he hand, he empi ical li e a u e collec s an amoun o econome ic
models which ha e iden i ied and measu ed he in luence o he mos signi ican ac o s
a ec ing p ices and launches o medicines in di e en coun ies. The use o di e en
samples may p e en om doing compa isons.
Demog aphic and income coun y ea u es, and egula ion egimes, seem o be
he mos impo an ac o s a ec ing d ug p icing and launching in he empi ical li e a u e.
Mo eo e , he d ug cha ac e is ics as s eng h, packsize and p esen a ion o ms a e
signi ican ly ela ed o he p ice, bu i is he he apeu ic alue, which obus ly a ec
p icing and launching. The i m loca ion u ns up an impo an ac o o he launching
decision bu p ice p emiums due o headqua e s loca ion appea ambiguous. Gene ic
compe i ion gene ally d i es down p ices; howe e , he e a e some di e en e ec s,
which dese e some commen s. In u n, he b and compe i ion ac o s do no appea o
exe compe i i e p essu e on p ices.
The coun y size, he GDP pe capi a and income dis ibu ion a e h ee coun y
cha ac e is ics shown as he mos signi ican ac o s in luencing p icing and launching.
Ce e is pa ibus, highe -income coun ies pay highe d ug p ices (Cab ales and Jimenez-
Ma in, 2007, Danzon and Eps ein, 2008) and ha e a mo e apid access o medicines
han lowe -income coun ies (Danzon e al., 2011, Heue e al., 2007, Kyle, 2007,
Lanjouw, 2005); u he mo e, popula ed coun ies also enjoy a highe p obabili y o launch
(Heue e al., 2007, Kyle, 2007, Lanjouw, 2005, Ve nie s e al., 2011). Addi ionally, an
unequal income dis ibu ion posi i ely a ec s he d ug launch haza d in lowe -income
coun ies ia a weal hy “eli e”. On he o he hand, a mo e equal dis ibu ion makes he
same e ec in high-income coun ies ia a la ges “middle class”. Howe e , we should
no e ha he coun y p icing policies migh unde mine he e ec s o hese impo an
ac o s. Con a y, in e es ingly, Cab ales and Jiménez-Ma ín obse e lowe p ices in he
US han in o he coun ies as Canada, F ance o I aly, and con a y o he con en ional
wisdom abou coun ies egula o y egimes (Cab ales and Jimenez-Ma in, 2007).

Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
43
In he empi ical e iew, he mos common inding is ha p ice egula ion ends o
p oduce launch delay. Such as delays ha e been obse ed in coun ies wi h s ic
egula ion and in adi ional pa allel expo e s. Pa icula ly, Heu e e al. show ha
coun ies using he ERP ha e a lowe p obabili y o launch (Heue e al., 2007), and
S a ga d and Sch eyögg s a e ha he use o ERP ha e posi i e di ec and indi ec
impac on he e e encing coun ies. A his ega d, hey p opose o use as many
coun ies as possible in he o mula (S a ga d and Sch eyögg, 2006), which pa ially
coincides wi h he ecommenda ions om he heo e ical model o Ga cía-Ma iñoso e al.
(Ga cia Ma iñoso e al., 2011). Howe e , S a ga d and Sch eyögg u he ecommend
a oiding coun ies using ERP, in o de o p e en undue impac s; and in eg a e he
ma ke olumes o he e e enced coun ies in o he index in o de o a oid high p ices
and launch delays in coun ies wi h small ma ke s (S a ga d and Sch eyögg, 2006). In
u n, Kyle shows ha he e is e idence om he s andpoin o a ac ing new medicines
ha indi ec p ice con ols such as RP may be p e e able o he di ec ones such as
pha macoeconomic e idence o p ice eeze (Kyle, 2007). Fu he mo e, he belonging o
he EMA make launches mo e likely, pa icula ly in highe -p ice coun ies (Kyle, 2007,
Ve nie s e al., 2011). On he o he hand, as explici p ice egula ion, only he explici use
o HTA has a signi ican e ec on p ices (Kana os and Vando os, 2011).
The he apeu ic quali y, he s eng h, he pack size, he numbe o p esen a ions
numbe o he d ug and he p oduc li e cycle a e e y signi ican ac o s on p icing. The
he apeu ic quali y, he s eng h and he numbe o p esen a ions make inc ease he p ice,
he pack size and he p oduc li e cycle a ec nega i ely (Cab ales and Jimenez-Ma in,
2007, Co onado e al., 2007, Danzon and Chao, 2000, Kana os and Vando os, 2011,
Timu e al., 2011, Danzon and Eps ein, 2008). Also, he li e a u e impo an ly conside s
he sequen ial launch o a d ug and he e ec on i s p ice. Commonly, he highe a e he
p ices p e iously se , he highe will be he d ug p ice in a coun y. We no e ha when
he e is no a e age global p ice40, wha can be in e p e ed as a d ug inno a i e cha ac e ,
he p ice is highe han when p e ious p ices exis (Cab ales and Jimenez-Ma in, 2007).
The apeu ic inno a ion also in luences posi i ely he launching. I has been obse ed ha
a p io launch in a high-income coun y has a s onge e ec on launching in a low-income
40 The e is no mean global p ice because he e a e no coun ies o be calcula ed.
44 Modeling Global P icing and Launching o New D ugs
coun y han ice e sa. Also, he ac o launching i s in high-p ice EU ma ke s posi i ely
a ec s launch p ices in he low-p ice ones ia ERP (Danzon and Eps ein, 2008).
Al hough he pape s analysed appea o ag ee on d ug cha ac e is ics, di e ences in
he apeu ic ca ego ies (Danzon and Eps ein, 2008, Danzon e al., 2005, Kyle, 2007,
Lanjouw, 2005, Ve nie s e al., 2011).
F om he pe spec i e o he i ms, i has been shown ha all mul ina ional i ms
ob ain a p ice ad an age. By con as , domes ic i ms end o en e he ma ke wi h sho
delays (Danzon and Eps ein, 2008, Danzon e al., 2005, Kyle, 2006, Kyle, 2007, Ve nie s
e al., 2011), howe e , i is no clea ha hey ecei e p ice p emiums (Cab ales and
Jimenez-Ma in, 2007, Danzon and Eps ein, 2008, Ve nie s e al., 2011). These las
esul s do no suppo he heo e ical model ha p edic ha local-mul ina ional i ms
ecei e p ice p emiums compa ed o o eign mul ina ional ones (Cab ales and Jimenez-
Ma in, 2007). Only Ve nie s e al. ind ha domes ic i ms ob ain highe p ices in hei
domes ic ma ke s, bu hey do no make di e ences be ween mul ina ional and non-
mul ina ional i ms. In any case, he li e a u e shows ha he con ac among i ms in
un egula ed ma ke s induces highe p ices h ough p ice collusion (Co onado e al.,
2007).
Mos pape s ind gene ic compe i ion o nega i ely a ec p ices (Cab ales and
Jimenez-Ma in, 2007, Co onado e al., 2007, Danzon and Chao, 2000, Danzon and
Eps ein, 2008, Danzon e al., 2011, Timu e al., 2011). In e es ingly, p ice egula ion is
ound o unde mine gene ic compe i ion in s ic egula o y sys ems. Howe e , he
p esence o gene ics may inc ease he p ice o b and names p oduc s (Cab ales and
Jimenez-Ma in, 2007, Co onado e al., 2007).
In gene al, excep o he case o ende ing p ocu emen in MLICs, b and
compe i ion does no seem o exe a signi ican compe i i e p essu e on p ices, o he
con a y, se e al au ho s ind ha b and compe i o p ices a ec posi i ely launch haza d
(Danzon and Eps ein, 2008, Kyle, 2006, Kyle, 2007). Only Danzon and Eps ein indica e
ha compe i ion occu s wi hin subclasses a he han be ween subclasses, and ha
dynamic compe i ion is d i en by p oduc cha ac e is ics o he han p ice (Danzon and
Eps ein, 2008).
Con a y o he heo e ical model ha p edic s ha he p esence o PT educes
p ices in high-income coun ies (Gansland and Maskus, 2004), he empi ical li e a u e
Chap e 1: Global P icing and Launching o New D ugs: Wha Does he Theo y Say?
Wha Do he Empi ical Models Show?
45
does no ind obus e ec s on i (Kana os P and Cos a-Fon J, 2005). By con as , PT
isk is mo e likely o lead o non-launch o launch delay in he pa allel expo while i has a
lowe impac on he impo ing coun y.
2 Chap e 2: Ex e nal Re e ence P icing and
Pha maceu ical Cos -Con ainmen .
2.1 In oduc ion
Pha maceu icals a e sold on a global ma ke . This cha ac e is ic gi es ise o a
speci ic ba gaining p ocedu e be ween pha maceu ical i ms and coun ies’ heal h
agencies. On he one hand, a i m makes s a egic decisions o sequen ially launch
medicines in di e en coun ies and o maximize global p o i s; and on he o he hand,
coun ies’ heal h agencies implemen p icing policies in o de o con ol hei
pha maceu ical expendi u e and ye gua an ee access o medicines.
Among exis ing d ug p icing policies, mos coun ies in he indus ialized wo ld
ha e implemen ed ei he Cos -E ec i eness Analysis (CEA) o Ex e nal Re e ence
P icing (ERP) a some poin in ime wi h he aim o con olling pha maceu ical expendi u e
bu s ill ensu ing access o medicines, mainly o on-pa en medicines (Espin J e al.,
2011, Rawlins, 2012).
Acco ding o he OECD, ERP, also e e ed o as Ex e nal P ice Benchma king o
In e na ional Re e ence P icing, is de ined as “ he p ac ice o compa ing pha maceu ical
p ices ac oss coun ies” and i is u he indica ed ha , “ he e a e a ious me hods applied
and di e en coun y baske s used” (Pa is e al., 2008). In his hesis, we use he ERP
de ini ion om Ga cia-Ma iñoso e al. (Ga cia Ma iñoso e al., 2011): “ERP consis s o
se ing a p ice cap o pha maceu icals, based on ex-manu ac u e p ices o iden ical o
compa able p oduc s in o he coun ies”.
ERP is no applied homogeneously in e e y coun y. The e a e a wide a ie y o
me hods used o design a o eign p ice index (Leopold e al., 2012, Espin J e al., 2011). I

48 Modeling Global P icing and Launching o New D ugs
mainly depends on each coun y’s baske , da e o p ices41, he me hod used ( he lowes
p ice, he a e age p ice, a pe cen age o he p e ious ones, e c.) and whe he a
weigh ed-index42 is used o no . We also no e ha some coun ies ake in o accoun ERP
as a complemen a y p icing policy oge he wi h o he p icing policies o help o make he
p ice decision, hus i is no exclusi ely applied as a blind p icing policy43. ERP is used
because o i s simplici y a echnical o analy ical le el; i does no equi e a huge ask o
collec p ice in o ma ion ab oad. Fu he mo e, ERP use s hink ha hose p ices aken as
a e e ence a e app oxima ely igh , sui able o ai . Howe e , hey men ion ha i is
di icul o assess i he esul ing p ices a e app op ia e, e icien o op imal in acco dance
wi h any objec i e c i e ion. The e o e, i e e encing coun ies se hei p ices oo high o
oo low, hen any coun y la e applying he ERP me hod may un he isk o epea ing he
same mis ake (Espin J e al., 2011).
The basic ade-o aced by a pha maceu ical i m is he ollowing. I a i m delays
a launch in a e e encing (low-p ice) coun y, i will also delay p o i s ha could be de i ed
om his coun y. Howe e , on he posi i e side, i a oids his low p ice om o e spilling
in o o he coun ies due o ERP s a egy o pa allel ade (Danzon and Eps ein, 2008,
Danzon and Towse, 2003, Ga cia Ma iñoso e al., 2011). By con as , when coun ies se
a d ug p ice, hey isk he possibili y o no p o iding he d ug a he ime hey desi e,
which may ha e consequences o he heal h and he wel a e o he popula ion
(Lich enbe g, 2005). The use o ERP by coun ies may make i ms apply in e na ional
p icing s a egies ha can ha m coun ies’ wel a e. On he one hand, a i m may se a
single p ice44 which may bene i he high-p iced45 coun ies bu ha m he low-p iced ones,
41 Cu en p ice s. p ice a launch
42 The mos widely me hod used o new d ugs is h ough non-weigh ed measu es; such me hods will no help o achie e
he a ge o ob aining a compa able a e age le el o p ices. The applica ion o weigh ed p ice indexes, compa able and
use ul as e e ence o he es o coun ies, has been p oposed DANZON, P. M. & CHAO, L. W. 2000. C oss-na ional p ice
di e ences o pha maceu icals: How la ge, and why? Jou nal o Heal h Economics, 19, 159-195..
43 Espín e al. s a e ha “ egula o s migh no always be able o willing o “impose” a ce ain p ice, bu ins ead use he p ice
compu ed as a benchma k o e e ence o nego ia ions, o en alongside o he c i e ia, such as cos -plus, in e nal o
he apeu ic p icing”
44 Two ac o s con ibu e o p ice uni o mi y be ween di e en ma ke s: a) h ea s o pa allel impo s, and b) he use o
in e na ional e e ence p icing DANZON, P. M. & TOWSE, A. 2003. Di e en ial P icing o Pha maceu icals: Reconciling
Access, R&D and Pa en s. In e na ional Jou nal o Heal h Ca e Finance and Economics, 3, 183-205..
45 In he long un, consume s om high p ice coun ies will be wo se o i his lowe p ice esul s in lowe expec ed e u ns
on R&D, and hence ewe new medicines han hey would ha e been willing o pay o DANZON, P. M. 1997. P ice
Disc imina ion o Pha maceu icals: Wel a e E ec s in he US and he EU. In e na ional Jou nal o he Economics o
Business, 4, 310-322..
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 49
and on he o he hand, he i m may ei he a emp o se high46 p ices in i s coun ies o
a oid low p ices in la e launches ia ERP, o delay launches in low-p iced coun ies o
a oid he spill-o e e ec s. These s a egies may ha m low-p iced coun ies, and hey
may e en ha m high-p iced ones (Ga cia Ma iñoso e al., 2011). Also, ano he s a egy
exis s o i ms o a oid spill-o e e ec s om ERP. This consis s o se ing high p ices
and g an ing con iden ial eba es o discoun s o e e enced coun ies. This s a egy
allows i ms o gua an ee lowe p ices in e e enced coun ies and a oids in o ma ion
spill-o e s o low p ices o e e encing coun ies (Espin J e al., 2011).
Mos coun ies use ERP as a pha maceu ical p icing s a egy. The use o ERP as
a mechanism o se pha maceu ical p ices is qui e widely applied: 24 o he 30 OECD
coun ies (Espin J e al., 2011) and app oxima ely 24 o he 28 EU Membe S a es
(Leopold e al., 2012) ha e used i .
CEA in heal h economics aims o es ima e he a io be ween he cos o a heal h-
ela ed in e en ion and he bene i i p oduces in e ms o he numbe o yea s li ed in ull
heal h by he bene icia ies. Cos is measu ed in mone a y uni s, while bene i needs o be
exp essed as a gain in heal h measu ed by quan i a i e alues. Howe e , unlike cos –
bene i analysis, he bene i s do no ha e o be exp essed in mone a y e ms. In
pha maeconomics, i is usually exp essed in quali y-adjus ed li e yea s (QALYs)47
(Na ional Insi u e o Heal h and Ca e Excellence (NICE), 2010). The ICER is he a io
be ween he di e ence in cos s and he di e ence in bene i s o wo in e en ions. So, an
example in which he cos s and gains, espec i ely, a e 140,000 eu os and 3.5 QALYs,
would yield an ICER o 40,000 eu os pe QALY. Commonly, each coun y has a di e en
h eshold o pay o one QALY. I we suppose ha such a coun y has a h eshold o
30,000 eu os pe QALY, any d ug which has an ICER o mo e han 30,000 eu os pe
addi ional QALY gained is likely o be ejec ed and any d ug which has an ICER o less
han o equal o 30,000 eu os pe ex a QALY gained is likely o be accep ed as cos -
e ec i e (WHO, 2003). Howe e , d ugs do no always yield a single ICER. We no e ha
o he au ho s ha e s udied in dep h CEA acco ding o Bayesian models (Neg in and
46 This company s a egy will no wo k i he high-p ice coun y e ises i s p ices downwa ds a e launch
47 The QALY is a measu e o disease bu den, including bo h he quali y and he quan i y o li e li ed. The QALY model
equi es u ili y independen , isk neu al, and cons an p opo ional adeo beha iou . The QALY is based on he numbe o
yea s o li e ha would be added by he in e en ion. Each yea in pe ec heal h is assigned he alue o 1.0 down o a
alue o 0.0 o being dead. I he ex a yea s a e no li ed in ull heal h, o example i he pa ien looses a limb, o goes
blind o has o use a wheelchai , hen he ex a li e-yea s a e gi en a alue be ween 0 and 1 o accoun o his.
50 Modeling Global P icing and Launching o New D ugs
Vazquez-Polo, 2006, Neg in and Vazquez-Polo, 2008, Neg in e al., 2010, Mo eno e al.,
2010)
A pha maceu ical i m knows his h eshold48. I would also ca y ou a CEA and
ob ain he numbe o QALYs gained i i s d ug we e p o ided in one coun y. Since he
i m is awa e o bo h h eshold and numbe o QALYs, i o e s he coun y he d ug a a
ce ain p ice jus below he h eshold. Howe e , he i m may upwa dly dis o he numbe
o QALYs o ob ain g ea e p o i s. Then, i is he coun y ha may e ise he i m’s CEA
applying i s own CEA o e eal a ai p ice. Howe e , his CEA equi es esou ces and
consequen ly an in es men o money by he coun y.
P e ious li e a u e has de eloped games based on ba gaining models be ween
pha maceu ical i ms and coun ies’ heal h agencies o se d ug p ices in an in e na ional
con ex . Ga cia-Ma iñoso e al. (Ga cia Ma iñoso e al., 2011) examine he e ec s o
using ERP by e e encing coun ies on e e enced coun ies’ wel a e ia a ba gaining
model. They ind ha a coun y has an incen i e o engage in ERP i i s co-paymen le els
a e high when compa ed o o he coun ies. This p e e ence dec eases as he ela i e
size o he coun y engaging in ERP inc eases. They also ind ha hese e ec s ha m
e e enced coun ies’ wel a e. Fu he mo e, Ga cía-Ma iñoso and Oli ella (Ga cía-
Ma iñoso and Oli ella, 2012) p esen a nego ia ion model based on a “ ake-i -o -lea e-i ”
p ocedu e ha examines he condi ions unde which a i m can use launch delay and he
consequen in o ma ion spill-o e s49 o ejec low p ices. The no ion ha low p ices may
o e spill o o he coun ies e en in he absence o pa allel ade o ERP is in oduced he e
di e en ly om p e ious esea ch.
Fu he mo e, o he heo e ical pape s (Jelo ac and Houy, 2013, Rich e , 2008)
deal wi h pha maceu ical i ms’ s a egies and coun ies’ pha maceu ical p icing policies
bu a e no based on ba gaining models. In his ega d, Rich e (Rich e , 2008) p oposes
48 We no e ha his h eshold does no ha e o be a single h eshold. The e exis s cu en ly an in e es ing discussion abou
he social alue o a QALY ha de e mines such a h eshold DONALDSON, C., BAKER, R., MASON, H., JONES-LEE, M.,
LANCSAR, E., WILDMAN, J., BATEMAN, I., LOOMES, G., ROBINSON, A. & SUGDEN, R. 2011. The social alue o a
QALY: aising he ba o ba ing he aise? BMC heal h se ices esea ch, 11, 8, MASON, H., JONESLEE, M. &
DONALDSON, C. 2009. Modelling he mone a y alue o a QALY: a new app oach based on UK da a. Heal h Economics,
18, 933-950, NIHR, H. 2010. Weigh ing and aluing quali y-adjus ed li e-yea s using s a ed p e e ence me hods: p elimina y
esul s om he Social Value o a QALY P ojec . Heal h Technology Assessmen , 14, PINTO-PRADES, J. L., LOOMES, G.
& BREY, R. 2009. T ying o es ima e a mone a y alue o he QALY. Jou nal o Heal h Economics, 28, 553-562,
ROBINSON, A., GYRD-HANSEN, D., BACON, P., BAKER, R., PENNINGTON, M. & DONALDSON, C. 2013. Es ima ing a
WTP-based alue o a QALY: The ‘chained’app oach. Social Science & Medicine, 92, 92-104.
49 In o ma ion spillo e s a e essen ially he demand o lowe p ices in a coun y gene a ed by he knowledge abou lowe
p ices in o he coun ies.
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 51
a ma hema ical op imiza ion p oblem o a i m wi h examples in which in e na ional p ice
dependencies play an impo an ole. This model can help coun ies o unde s and he
implica ion o hei ERP policies on a global epea ed p icing game. On he o he hand,
Jelo ac and Houy (Jelo ac and Houy, 2013) analyse he iming decisions o
pha maceu ical i ms o launch a new d ug in coun ies using ERP. When all coun ies
e e ence he p ices in all o he coun ies and in all p e ious pe iods o ime, hen he e is
no wi hd awal o d ugs in any coun y, and in any pe iod o ime and he e is no incen i e
o delay he launch o a d ug in any coun y. Howe e , hese esul s do no hold when he
coun ies only e e ence a subse o all coun ies o when he e e ence is only on he
la es pe iod p ices.
Conce ning empi ical s udies, Danzon and Eps ein (Danzon and Eps ein, 2008)
in e es ingly obse e ha launch p ices inc ease wi h he le el o he lowes p ice
p e iously ecei ed in o he high-p ice EU coun ies, whe eas he e ec s o a p e ious
launch in low-p ice EU coun ies a e insigni ican . This esul is consis en wi h he
hypo hesis ha launching i s in high-p ice EU ma ke s can in luence p ices in low-p ice
ones. This e idence abou sequen ial launch p ices alida es he heo y ha a launch
delay in low-p ice ma ke s may ul ima ely yield highe p ices in hese ma ke s h ough
spillo e s om highe -p ice ones. This heo y is shown by S a ga d and Sch eyögg
(S a ga d and Sch eyögg, 2006). They analyse he di ec and indi ec impac o he use
o ERP om he e e enced o he e e encing coun y. They es ima e he impac o d ug
p ice changes in Ge many50 on d ug p ices in o he coun ies using ERP in he o me
EU-15. The au ho s use he o mulas applied by each e e encing coun y and hen, hey
calcula e he pa ial di e en ial o hese o mulas wi h espec o a 1 eu o p ice educ ion
in Ge many. The au ho s s a e ha he ela ionship be ween he di ec and indi ec impac
o a p ice change depends mainly on he scheme applied o se p ices. Thus, o a oid he
nega i e e ec s o ERP and de e mine p ices in o de o educe he di ec and indi ec
impac o indi idual coun ies, a weigh ed o mula o p ices con aining as many coun ies
as possible should be used.
Su p isingly, only one ou o h ee s udies inds a di ec e ec on p icing when
ERP is applied. These esul s could be explained by he di e en da abases collec ed.
50 Ge many is chosen because i is one o he la ges pha maceu ical ma ke s in he wo ld, i is cha ac e ised by ela i ely
high p ices and i is e e enced by mos c oss- e e encing coun ies schemes.
58 Modeling Global P icing and Launching o New D ugs
(
2.7
)
(
2.8
)
(
2.9
)
(
2.10
)
coun y C
Assump ion 2
The e is p ice a iabili y among he J coun ies whe e he d ug has been al eady
ma ke ed. The e o e, he a e age in e na ional p ice is highe han he minimum
in e na ional p ice.
Assump ion 3
Gi en a coun y i applying CEA, he QALYs e ealed by he CEA ca ied ou by he
coun y will no be highe han ha p oposed by he i m.
Pa icula ly, i he i m is hones and, he e o e, decla es he ue numbe o QALYs , hen
he QALYs e ealed by he CEA ca ied ou by he coun y is equal o ha p oposed by
he i m. Howe e , i he i m is no hones and he e o e does no gi e he ue numbe o
QALYs, he QALYs esul ing om he CEA is lowe han ha sugges ed by he i m.
p
F
p
F
P
I

P
I
min
YiCEA YF
i
h
e
i
mdecl
a
es
h
e uenum
b
e o
Q
ALYs
YiCEA YF o he wise
Y
F
Y
iCEA

Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 59
(
2.11
)
(
2.1
2
)
(
2.13
)
Consequen ly, he p ice esul ing om he applica ion o CEA will no be highe han ha
p oposed by he i m,
p
iCEA

p
iF
i
h
e
i
m u
s
ed
b
y
h
e
h
e
a
l
h
a
gency
p
iCEA
p
iF
o he wise
Then, owing o he scien i ic e idence showed by he CEA, assuming ha he ma ginal
cos (mc) o p oducing he pha maceu ical is ze o (mc=0) and gi en ha he i m is p o i
maximizing, he i m will always sell a CEA p ice53.
Assump ion 4
Gi en bo h coun ies i applying CEA, he numbe o QALYs e ealed by he esea ch,
YiCEA,, will be he same o bo h coun ies.
Assump ion 5
The coun ies’ belie s abou he i m’s hones y a e he same o bo h coun ies i (i = C,D).
Assump ion 6
Unde assump ions 3 and 5, he p ice expec ed by he coun y i (i = C,D) when applying
CEA is,
53 We no e ha he i m knows he coun ies i (i = C,D) W P, he e o e i i is oo low (below a gi en h eshold), he i m does
no e en ini ia e nego ia ions o launch in ha coun y.
P
(
Y
F

Y
iCEA
)


,

i
P (Y
F
Y
iCEA
)1

, i
60 Modeling Global P icing and Launching o New D ugs
(
2.14
)
(
2.15
)
(
2.1
6
)
(
2.17
)
Then, unde Assump ion 1 and 5, he p ice expec ed by coun y D is la ge han he p ice
expec ed by coun y C,
Assump ion 754
I he p ice p oposed by he i m is highe (lowe ) han he in e na ional e e ence p ice,
, hen he p ice disco e ed by coun y when applying CEA will also be highe
(lowe ). The e o e, he CEA expec ed p ice will no be lowe (highe ) han he
in e na ional e e ence p ice.
Assump ion 855
The a e age be ween he a e age in e na ional e e ence p ice and he coun y C p ice is
app oxima ely equal o he a e age in e na ional e e ence p ice. The e o e, i is assumed
ha J is la ge enough ha he a e age p ice is no a ec ed by a u he obse a ion.
54 This assump ion gua an ees he ade-o be ween choosing CEA and ERP. I coun y i applies CEA, i will pay a highe
p ice o a oid ha ing he d ug launch delayed.
55 The Assump ion 8 makes simple he esul s and does no a ec he conclusions ob ained.

p
iF
(1


)

p
iCEA

Ep
i


CEA
Ep
D

CEA
Ep
C


CEA
p
F
p
I
h
p
iCEA
Ep
i


CEA
I p
F
p
I
h
 p
iCEA
p
I
h
Ep
i


CEA
p
I
h
I p
F
p
I
h
 p
iCEA
p
I
h
Ep
i

CEA
p
I
h
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 61
(
2.18
)
(
2.19
)
Gi en p
I


p
j
j1
J

J hen,
p
j
p
C
j1
J

J1p
I

wi h p
C

p
I
min
p
F
p
CEA
C





Assump ion 9
The p ice di e ence be ween he in e na ional e e ence p ice and he p ice p oposed by
he i m is he same ega dless o he ype o coun y. Thus, he incen i e o apply ERP is
also he same ega dless o he ype o coun y (high o low W P), mo e o mally
The Fi m
The pha maceu ical indus y is cha ac e ized by high ixed cos s (F) (Mes e-
Fe andiz, 2012, Mes e-Fe andiz, 2013) and low mc (mc=0). Hence, F > 0 s ands o he
ixed cos s o R&D, sa e y app o al p ocess and ma ke ing he d ug in all coun ies. They
a e ixed and independen o he numbe o people o coun ies ha use he d ug. The
i m sells he d ug o coun y i (i = C, D) a p ice pi. This p ice pi is he maximum p ice a
which he i m and he heal h agencies ag ee56. I he coun y’s p ice comes om a CEA
policy, unde assump ion 3, he i m will always sell a CEA p ice. Howe e , i he coun y
p ice comes om an ERP policy, he i m will be able o accep o e use i . Should i
e use, he i m will delay launch in such a coun y. The e o e he i m commi s o
launching he d ug and o sa is ying he whole demand in his coun y (qi) a p ice pi.
Selling his pha maceu ical p oduc wi hou subsidiza ion is no conside ed as an op ion.
Thus, we assume ha he objec i e o a monopoly p oduce o a medicine is o maximize
56 We a e awa e ha pu chase bodies such as hospi al o pha macy bodies may achie e discoun s om his p ice pi, bu
hey a e no conside ed in his hesis.
p
I
min
p
F
p
I

p
F
62 Modeling Global P icing and Launching o New D ugs
(
2.20
)
he accumula ed p o i s unc ion du ing he leng h o he sales ( wo pe iods, s age 3 and
4, see iming), which can be w i en as,
OF
F
p
i
q
i
F
iC
D

1
2

wi h
q
i

0
i
h
e
d
u
g
i
s
n
o
m
a
k
e
e
d
i
n
c
o
u
n
y
i
i
n
i
m
e
Also, we assume ha he i m is no loca ed in any o he coun ies i (i = C,D).
Timing
The iming o his game is as ollows. The game has 4 s ages. In s age 1,
coun ies C and D choose hei p icing policies, CEA o ERP, and he i m p oposes a
p ice o he d ug. In s age 2, coun ies communica e hei p ices acco ding o hei p icing
policies. In s age 3, as launching is sequen ial (say launch i s in C and hen in D o ice
e sa), he i m chooses he coun y launch sequence, i.e., i chooses be ween delaying in
coun y D o delaying in coun y C, and sells he d ug in he i s coun y o he sequence.
In s age 4, he i m sells he d ug in he second coun y. Since he i m is p o i
maximazing, he i m sells in bo h coun ies i (i=C, D).
A p io i, no ice ha coun y i (i=C, D) can choose be ween a p icing policy ha
e en ually equi es an in es men o money, CEA, and ano he p icing policy wi h no
cha ges, ERP. Howe e , applying ERP, he coun y may ha e he d ug launch delayed i
he i m he in e na ional e e ence p ice is lowe han de p ice p oposed, whe eas,
applying CEA, he coun y isks no making a use ul in es men i he i m gi es he ue
numbe o QALYs.
In o de o show he iming mo e clea ly, see he decision ee in Appendix B.
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 63
Assump ion 10
D ug launching is sequen ial and he i m keeps selling he d ug in s age 4 o he coun y
whe e he d ug has p e iously been launched.
Assump ion 11
I only one coun y i (i= C,D) applies ERP and he in e na ional e e ence p ice is lowe
han he p ice p oposed by he i m, he i m will punish such a coun y i by delaying
launch in i . I bo h coun ies i (i= C,D) apply ERP and he in e na ional e e ence p ices
a e lowe han he p ices p oposed by he i m, he i m will delay launch only in he
coun y i (i= C,D) acco ding o he coun y ha o e s he lowes income.
2.3 P ice Se ing and Sequen ial Launch
Playe s maximize hei objec i e unc ion and we sol e he game applying backwa d
induc ion. Fo bo h cases, when he i m claims a numbe o QALYs abo e he ue
numbe o QALYs, and when i claims he ue QALYs o he d ug, hen, we can sol e o
each coun y’s p icing policy: i) no coun ies apply ERP, ii) only coun y D applies ERP, iii)
only coun y C applies ERP and, i ) bo h coun ies apply ERP, we calcula e he condi ions
o he op imal coun y launch sequence o he i m, ei he i s coun y C and second
coun y D, o ice e sa. The p oo can be checked in Appendix B. In Sec ion 4, we will
compa e he coun ies’ su plus o each p icing policy, gi en he op imal coun y launch
sequence.
2.3.1 The i m is us ed by he heal h agency
i) No coun ies apply ERP
Since bo h coun ies C and D apply CEA, coun ies C and D will pay and
espec i ely acco ding o (2.11). Since he i m is us ed by coun ies i (i= C,D),
unde assump ion 1, coun ies C and D will equi alen ly pay and co espondingly.
p
CCEA
p
DCEA
p
F
p
F

64 Modeling Global P icing and Launching o New D ugs
(
2.21
)
(
2.2
2
)
(
2.2
3
)
No ice ha , unde assump ion 11, he incomes o he coun y whe e he d ug has been
i s launched a e mul iplied by wo.
Conce ning he heal h agencies’ su pluses, we no e ha he p ice conside ed by
coun ies C and D when applying CEA is an expec ed p ice acco ding o assump ions 5
and 6, since bo h coun ies ha e unce ain y abou he numbe o QALYs s a ed by he
i m.
he i m’s p o i s a e,
and he heal h agencies’ su plus a e57,
OF
C
(W P
C
Y
F
Ep
C


CEA
)q
C
a i C,D


(W P
C
Y
F
p
F
)q
C
a q
C
o he wise





OF
D
(W P
D
Y
F
Ep
D

CEA
)q
D
a q
D
i C,D

(W P
D
Y
F
p
F
)q
D
a o he wise





57 No e ha heal h agencies do no know i he i m has gi en he ue numbe o QALYs o no . The e o e, hey only know
he expec ed p ice.
OF 2p
F
q
C
p
F
q
D
F i C,D

p
F
q
C
2p
F
q
D
F o he wise





Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 65
(
2.24
)
PPR58 1
{C,D} i p
F
p
F
q
D
q
C
{C,D} o he wise





I no coun ies apply ERP, he i m chooses o delay launch in coun y D, i and only i he
p ice a io coun y C o coun y D is g ea e han he size a io coun y D o coun y C,
o he wise he i m will delay launch in coun y C. Since he p ice a io is less han uni y,
hen a necessa y condi ion o he sequence {C,D} is ha coun y C mus be la ge han
he D’s.
ii) Only coun y D applies ERP
Coun y D decides o apply ERP in s age 1. I he a e age in e na ional p ice is
lowe han he p ice p oposed by he i m, i.e., , unde assump ion 11, he i m
delays he launch in coun y D. Since he p icing policies a e se ex-an e, i ,
coun y D will pay in any case. Howe e , as coun y C applies CEA, i will pay .
Since he i m gi es he ue numbe o QALYs, unde assump ion 1, i will equi alen ly
pay .
Conce ning he heal h agencies’ su pluses, we no e ha he p ice conside ed by
coun y C when applying he CEA is an expec ed p ice acco ding o assump ions 5 and 6,
since he coun y C has unce ain y abou he numbe o QALYs gi en by he i m.
he i m’s p o i s a e,
58 P elimina y esul .
p
I

p
F
p
I

p
F
p
I

p
CCEA
p
F
66 Modeling Global P icing and Launching o New D ugs
(
2.2
6
)
(
2.28
)
(
2.2
7
)
(
2.25
)
OF
F
2p
F
q
C
p

I
q
D
F i C,D

p
F
q
C
2p
I

q
D
F o he wise





and he heal h agencies’ su pluses a e,
PPR 2
{C,D} i p
F
p
I

q
D
q
C
{C,D} o he wise





I only coun y D applies ERP, he i m will choose o delay launch in coun y D i and only
i he p ice a io coun y C o coun y D is la ge han he size a io coun y D o coun y C,
o he wise he i m will delay launch in coun y C. Speci ically, he p ice a io is he a io
be ween he CEA p ice and he a e age in e na ional e e ence p ice. Since he p ice a io
is less han uni y, hen a necessa y condi ion o he sequence {C,D} is ha coun y C
mus be la ge han he D’s.
iii) Only coun y C applies ERP
Coun y C decides o apply ERP in s age 1. I he minimum in e na ional p ice is
lowe han he p ice p oposed by he i m, i.e., . The i m delays he launch in
OFC(W PCYFE[pC]CEA)qCa i C,D


(W PCYFE[pC]CEA)qCa qC o he wis
e





OFD(W PDYFpI

)qD qD i C,D

(W PDYFpI

)qD o he wise





p
Imin
p
F
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 67
(
2.30
)
(
2.29
)
(
2.31
)
(
2.3
2
)
coun y C. Since he p icing policies a e se ex-an e, i , he coun y C will pay
in any case. In u n, coun y D applies CEA and i will pay . Since he i m
gi es he ue numbe o QALYs, unde assump ion 1, i will pay he equi alen o .
Conce ning he heal h agencies’ su pluses, we no e ha he p ice conside ed by he
coun y D when applying CEA is an expec ed p ice acco ding o assump ions 5 and 6,
since coun y D is unce ain abou he numbe o QALYs s a ed by he i m.
he i m’s p o i s a e,
OFF2pI
minqCpFqDF i C,D

pI
minqC2pFqD- F o he wise





and he heal h agencies’ su pluses a e,
PPR 3
{C,D} i pImin
pF
qD
qC
{C,D} o he wise





pImin p
F
p
I
min
p
DCEA
p
F
OF
C
(W P
C
Y
F
p
Imin
)q
C
i C,D

(W P
C
Y
F
p
Imin
)q
C
 q
C
o he wise





OF
D
(W P
D
Y
F
E[p
D
]
CEA
)q
D
a q
D
i C,D

(W P
D
Y
F
p
F
)q
D
a o he wise





74
Modeling Global P icing and Launching o New D ugs
d) I
p
F
p
I

and p
F
p
I
min
Figu e 2.4 Op imal coun y launch sequence unde d)
Table 2.5. Op imal Coun y Launch Sequence in Figu e 2.4
Region/
Policies A B X
62
E F
i) No ERP {C,D} {C,D}{D,C} {D,C} {D,C}
ii) D ERP {C,D} {C,D} {D,C} {C,D} {D,C}
iii) C ERP {C,D} {D,C} {C,D} {D,C} {D,C}
i ) Bo h ERP {C,D} {C,D} {C,D} {D,C} {D,C}
62
Unde
p
I

p
F
p
F
p
I


Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 75
2.3.2 The i m s a es he numbe o QALYs abo e he ue alue
This pa o he ee has a simila solu ion o ha sol ed abo e. Fo easons o
b e i y, we jus highligh he di e ences.
In his case, i he coun y applies ERP, as in he p e ious case, i will no be able
o disco e he eal alue o he d ug and i will pay he in e na ional e e ence p ice ,
which may be highe o lowe han he p ice e ealed when applying CEA. Howe e , i he
coun y decides o apply CEA, i will e eal a lowe numbe o QALYs han hose
p oposed by he i m . Consequen ly, he coun y will pay a lowe p ice han he
p ice p oposed by he i m . The e o e, on he one hand, i is now less likely o he i m
o accep he in e na ional e e ence p ices han in he case o being hones , which
implies ha coun ies applying ERP will be mo e likely o expe ience launch delays. Also,
he egions unde which he i m chooses i s op imal launching sequence change (Figu es
2.2, 2.3, 2.4 and 2.5). On he o he hand, he expec ed alue o he d ug p ice o he
coun ies will be highe when he i m s a es a numbe o QALYs abo e i s ue
alue han when i does no . The implica ions o his issue will be explained in sec ion 4.
2.4 Compa ing Policies: CEA s. ERP
In his sec ion, gi en he op imal coun ies launch sequence by he i m unde PPR
1, 2, 3 and 4, we compa e he coun ies’ wel a e unde each p icing policy, CEA and ERP,
o know which o hem is mo e con enien o coun ies. Thus, since we ha e sol ed he
p oblem by backwa d induc ion, we ha e ca ied ou his compa ison o each coun y
gi en he op imal coun y launch sequence o he i m. We ha e made his compa ison in
h ee s eps. Fi s ly, we ha e compa ed he bes ou come o each coun y unde he same
p icing policy and unde he same coun y launch sequence, i.e., using CEA (ERP) unde
{C, D} and {D, C}, espec i ely. Then, in a second s ep, we ha e compa ed he bes
ou comes be ween ERP and CEA o each coun y launch sequence. In a hi d s ep, we
ha e compa ed he bes p icing policy unde each coun y launch sequence. Thus, we
ha e Condi ion 1. The p oo can be checked in Appendix B.
p
I
h
Y
iCEA
Y
F
p
iCEA
p
F
p
I
h
E[
p
i
]
CEA
76 Modeling Global P icing and Launching o New D ugs
(
2.38
)
(
2.39
)
(
2.40
)
Condi ion 1
I coun y i does no su e om delay launch unde any p icing policy o , unde bo h
p icing policies, coun y i will be be e o applying ERP when he uni a y cos o ca ying
ou CEA is highe han he di e ence be ween he in e na ional e e ence p icing and
he expec ed p ice o coun y i unde CEA (hence o h, he p ice di e ence). In addi ion,
he smalle he popula ion size is, he mo e a ac i e will be he use o ERP, since he
uni a y cos o CEA inc eases.
I coun y i su e s om delay launch when applying ERP bu no unde he use o CEA,
he delay cos ( ) will make CEA mo e a ac i e,
Analogously, i coun y i su e s om delay launch when applying CEA bu no unde ERP,
he delay cos ( ) will make ERP mo e a ac i e,
In o de o show g aphically Condi ion 1, we plo he ollowing igu e,
p
I
h
a
q
i
p
Ih
E[p
i
]
CEA
a
q
i
p
I
h
E[p
i
]
CEA

a
q
i
p
I
h
E[p
i
]
CEA

Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen
77
Figu e 2.5. ERP s. CEA
Acco ding o Figu e 2.5, we obse e unde which condi ions ega ding he uni a y
cos o applying CEA and he p ice di e ence, he coun y chooses ei he ERP o CEA.
Ei he i bo h p icing policies (CEA and ERP) a e applied wi hou expe iencing any delay
launch, o bo h su e ing om a delay launch, he coun y will be be e o applying ERP
only i he uni a y cos o applying CEA is highe han he p ice di e ence.
In ui i ely, we no e ha since he uni a y cos o applying CEA dec eases, o keep
he applica ion o ERP bene icial, he p ice di e ence should be smalle , ei he because
he in e na ional e e ence p ice is lowe o he expec ed p ice o he coun y i unde CEA
inc eases. On he o he hand, i he uni a y cos o applying CEA inc eases, o main ain
he bene i s o applying CEA, he p ice di e ence should be la ge , ei he because he
78 Modeling Global P icing and Launching o New D ugs
in e na ional e e ence p ice goes up o he expec ed p ice o he coun y i unde CEA
dec eases.
Howe e , when we compa e bo h p icing policies, on he one hand, i only he
coun y applying CEA su e s om launch delay, e en hough he p ice di e ence is
highe han he uni a y cos o applying CEA, he uni a y delay cos associa ed wi h CEA
may compensa e his highe di e ence and make ERP mo e wo hwhile. On he o he
hand, i only he coun y applying ERP expe iences launch delay, and he uni a y cos o
applying CEA is highe han he p ice di e ence, he delay cos o applying ERP may
o se a high uni a y cos o CEA and make CEA mo e a ac i e han ERP o he coun y.
Then, when he p ice di e ence is nega i e, i.e., he expec ed p ice o he coun y i
unde CEA is highe han he in e na ional e e ence p ice and he e is also a launch
delay when applying CEA, ERP will always be chosen by he coun y. Simila ly, i he e is
no delay applying CEA, in ui i ely, he uni a y cos o applying CEA mus be highe han
he uni a y delay cos induced by applying ERP o o se he nega i e p ice di e ence and
he e o e o keep ERP a ac i e, despi e he delay launch.
Impo an ly, we no e ha he expec ed p ice will be highe when he i m
decla es a numbe o QALYs abo e i s ue alue han when i does no , which implies
ha ERP will be mo e a ac i e o he coun ies when he i m is no us ed by coun ies
i (i = C, D).
2.5 Conclusions
Using a model whe e one i m sells an on-pa en d ug o wo coun ies, which
di e in hei W P (ICER), popula ion size, p icing policy (ERP s. CEA) and ERP o mula,
one o ou main esul s is ha he op imal coun y launch sequence depends on he
ela i e p ice and he ela i e coun y size. The ela i e p ice depends on he coun ies’
p icing policy (ERP o CEA) and he ERP o mula.
Gi en he op imal coun y launch sequence, ou o he o e all esul is ha a
coun y is be e o applying ERP ins ead o CEA i he uni a y cos o CEA is highe han
he p ice di e ence, i.e., he di e ence be ween he in e na ional e e ence p ice and he
expec ed p ice o he coun y unde CEA. The cos o delaying may a ec his decision i
only one o he p icing policies is applied wi h delay. Thus, i ERP is applied wi h delay,
E
[
p
i
]
CEA
Chap e 2: Ex e nal Re e ence P icing and Pha maceu ical Cos -Con ainmen 79
he delay cos will make i less a ac i e wi h espec o CEA, and analogously, he same
applies o CEA wi h espec o ERP.
Basically, he highe he cos o CEA and he lowe he in e na ional e e ence
p ice is, hen he mo e a ac i e he use o ERP is. In b ie , we ha e compa ed wo
p icing policies: one o hem, ERP, does no equi e any in es men , and he o he , CEA,
needs an in es men o money. In hese e ms, he smalle he popula ion size is, he
mo e a ac i e he use o ERP will be, since he uni a y cos o CEA inc eases.
The applica ion o ERP will be mo e a ac i e han CEA when he i m is no
hones , howe e , i will be mo e likely o expe ience launch delays when i is no hones .
The e o e, he con enience, in his case, will depend on how many mo e QALYs abo e
he ue numbe he i m s a es i s d ug has and he delay cos .
We accep ha a wide a ie y o ac o s han al eady used in he model ha may
a ec he ba gaining p ocess. Fo example, he o mulas used o apply ERP may be o he
han he a e age o he minimum, mo e han wo pe iods could be conside ed, i ms may
o e one single p ice o launch simul aneously and he e ec i eness e ealed by CEA
could be di e en among coun ies. Also, we accep ha assump ion 9 cons ains he i m
s a egy since one p ice pF is se , he o he is implici ly se as well. Besides, o he ac o s
such as he popula ion age s uc u e o he lobbying ac i i y o he pha maceu ical
indus y (Ab aham, 2002) may also need o be conside ed.
Howe e , we conside ha his chap e has p o ided insigh s in o he way a
coun y’s W P and i s p icing policies a ec he op imal launch sequence o a i m.
Addi ionally, gi en an op imal launch sequence, we p opose unde wha condi ions,
ega ding coun y size and p icing policy, i is be e o applying CEA o applying ERP o
coun y i (i = C, D).

3 Chap e 3: Global P icing and Launching o
New D ugs. An Econome ic App oach
3.1 In oduc ion
Pha maceu icals a e sold in a global ma ke ha in ol es a speci ic ba gaining
p ocedu e be ween pha maceu ical i ms and coun ies’ heal h agencies. On he one
hand, i ms sequen ially launch medicines in di e en coun ies o maximize global p o i s,
he e o e p icing and launching a e hei majo s a egic decisions. On he o he hand,
coun ies’ heal h agencies implemen p icing policies o con ol hei pha maceu ical
expendi u e and o gua an ee access o medicines. Indeed, pha maceu ical p ice
egula ion is high on policy agendas in se e al coun ies, ei he because coun ies ha e
jus e o med, in end o e o m o ques ion hei p ac ices (see Chap e 1 sec ion 1.1).
Among exis ing d ug p icing policies, mos coun ies in he indus ialized wo ld ha e
implemen ed ERP a some ime wi h he aim o con olling hei pha maceu ical
expendi u e and ensu ing access o medicines, mainly in on-pa en medicines (see
Chap e 2 sec ion 2.1).
In his chap e , we aim o analyze he ade-o be ween p icing and launching and
he impac o ERP policy on p icing and launching om an empi ical poin o iew. We
de elop a model ha ocuses on bo h issues, con olling o molecules, egula ion and
coun y cha ac e is ics. We eplica e he s udy o Danzon and Eps ein, published in 2008,
and he s udy o Ve nie s e al. published in 2011. Thus, we aim o es how he si ua ion
has changed applying he same me hodology o mo e ecen da a, and in he case o he
second s udy, o a di e en lis o coun ies.
The p e ious li e a u e conce ning he ade-o be ween p icing and launching has
been al eady discussed in de ail in Chap e 1 in Sec ion 1.2.3 a a heo e ical le el and in
Sec ion 1.2.4 om an empi ical poin o iew. Fu he mo e, li e a u e conce ning he
82 Modeling Global P icing and Launching o New D ugs
impac o ERP policy has also been heo e ically and empi ically discussed in Chap e 1 in
sec ions 1.2.2 and 1.4.2, espec i ely. Addi ionally, in chap e , 2 we de eloped a
heo e ical model ha analyses he con enience o applying ERP as an al e na i e o
CEA as a cos -con ainmen policy on pha maceu ical expendi u e. Pa icula ly, chap e 2,
sec ion 2.1, p o ides insigh s in o he implemen a ion o ERP policy.
In his chap e , we de elop a wo-equa ion empi ical model consis ing o a launch
delay equa ion and a ela i e launch p ice equa ion. P e iously, we eplica e wo s udies
(Danzon and Eps ein, 2008, Ve nie s e al., 2011) using ou da abase o compa e hei
esul s wi h hose ob ained om ou upda ed da a and di e en lis o coun ies.
We use da a om IMS Heal h da abase on 56 new molecules launched in 20
coun ies belonging o 11 he apeu ic classes, all o hem app o ed h ough he
cen alised p ocedu e by he EMA, du ing he s udy pe iod, 2004-2010. We ha e
collec ed yea ly inpa ien and ou pa ien sales in eu os a ex-manu ac u e p ice and uni
olume (IMS SU).
Ou con ibu ion o he p e ious li e a u e analysed in Chap e 1, sec ion 1.3,
consis s o he analysis o da a a p esen a ion le el63, he conside a ion o he ela i e
launch p ice64 as an endogenous a iable in he ela i e launch p ice equa ion, he s udy
o he launch delay as a du a ion ime a iable and he analysis o he inpa ien ma ke .
Addi ionally, we in oduce coun y size and coun y pu chasing powe as addi ional
explana o y a iables.
3.2 Da a desc ip ion
In Tables 3.1, 3.2, 3.3, 3.4 and 3.5, we show he desc ip i e s a is ics o ou
da abase. 75% o he coun ies belong o he EMA and 70% o he coun ies apply ERP.
No all molecules ha e been launched in all coun ies. In he e ail ma ke coun ies
belonging o he EMA expe ience sho e launch delays and pay lowe ela i e launch
p ice on a e age han coun ies ou o he EMA. In he hospi al ma ke , he pa e n o
launch delays is simila o he e ail ma ke , while coun ies belonging o he EMA and
63 We de ine wo p oduc s wi h he same p esen a ion when bo h p oduc s belong o he same molecule i and ha e he
same quan i y o ac i e ing edien pe s anda d uni (see de ini ion o s anda d uni in Chap e 3 sec ion 3.3.2).
64 De ined la e in Appendix C.3.
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 83
coun ies ou o i pay he same ela i e launch p ices on a e age. Bo h he launch delays
and he ela i e launch p ices show high a iabili y. Coun ies pay highe ela i e launch
p ices in he hospi al ma ke han in he e ail one, howe e , no co ela ion ha e ound
be ween ela i e launch p ices o he e ail and hospi al ma ke .
In bo h e ail and hospi al ma ke s, we do no ind s a is ical signi ican di e ences
in ela i e launch p ices nei he be ween he coun ies ha apply ERP and coun ies ha
do no , no be ween coun ies belonging o he EMA and coun ies ha do no . Howe e ,
s a is ical signi ican di e ences a e ound when s a is ical di e ences in launch delays
a e analysed. Then, coun ies applying ERP p esen signi ican longe launch delays on
a e age while coun ies belonging o he EMA expe ience signi ican sho e launch
delays on a e age.
Table 3.1 Desc ip i e s a is ics. Re ail ma ke
Re ail ma ke
Numbe o
Molecules
Launched
Mean Rela i e
P ice SD Rela i e
P ice Mean Delay
in Mon hs SD Delay in
Mon hs ERP
EMA 5.470597 51.26997 11.82316 11.73209
Aus ia 48 7.50811 12.50254 *
Belgium 24 0.9734428 19.43561 *
Czech Republica 32 6.151763 18.99198 *
Denma k 52 7.387367 10.90712
Finland 36 5.091636 11.36712 *
F ance 31 4.115586 17.85784 *
Ge many 58 11.03968 9.974775
Hunga ya 33 1.468164 18.55194 *
I aly 19 4.34083 20.33394 *
Ne he lands 31 1.011494 6.705952 *
No way 35 1.883906 8.131455 *
Polanda 28 1.959413 13.76601 *
Spain 23 0.873803 18.75988 *
Sweden 46 10.72532 6.21954
Uni ed Kingdom 37 1.449288 8.15
Non EMA 7.450721 40.28605 16.31287 16.42291
Aus alia 34 11.41437 22.50628 *
Canada 41 8.664639 16.12424
90 Modeling Global P icing and Launching o New D ugs
p oduc s. Also, he o de in which a molecule is launched in each coun y-subclass
in luences posi i ely he launch p ice. Howe e , as in he launch equa ion, we ha e no
been able o include hese a iables in ou upda ed model as we explain in de ail in
Appendix C.1.
In bo h models, launch p ices inc ease wi h he minimum p ice p e iously se in
o he high-p ice EU coun y. Howe e , he e ec on launch p ices o he minimum p ice
p e iously se in high-p ice non-EU coun ies is di e en unde each model, posi i e unde
he D&E model and nega i e unde ou s. Only ou upda ed model epo s a signi ican
and posi i e e ec om a minimum p ice se in low-p ice EU coun ies. Fu he mo e, i
p ices in high-p ice EU coun ies a e missing, i also a ec s posi i ely he launch p ice.
This could indica e ha coun ies se ing he launch p ice wi h no e e ence in he EU may
pay high launch p ices. We no e ha when we es ima e wi h andom e ec s, he e ec s
om low-p ice EU and high-p ice non-EU coun ies become s a is ically signi ican . Then,
we obse e ha he lowes p ice p e iously se in o he low-p ice EU coun y a ec s
posi i ely he launch p ice bu mo e sligh ly han he e ec om he high-p ice EU coun y
p ice. In e es ingly, bo h he minimum p ice p e iously se in a high-p ice non-EU coun y
and he absence o a minimum p ice om high-p ice non-EU coun ies a ec nega i ely
he launch p ice. This may indica e ha spillo e e ec s also occu ei he be ween EU and
non-EU coun ies o among non-EU coun ies. Fu he mo e, since he e ec o he o me
is g ea e , i may show ha he ne e ec on he launch p ice o a p e ious launch in a
leas one high-p ice non-EU coun y is posi i e. On he o he hand, he absence o a
minimum p ice om low-p ice EU coun ies means coun ies do no ha e a e e ence om
his ype o coun ies, and he e o e i canno be included in o hei e e ence baske ,
which esul s in paying highe p ices han i p ices om low-p ice EU coun ies we e
a ailable. The esul s om he upda ed model wi h clus e ed s anda d e o s suppo he
occu ence o spillo e e ec s om high-p ice EU coun ies o low-p ice EU coun ies, and
he e o e, ha he ERP only conce ns EU coun ies. Howe e , he upda ed model wi h
andom e ec s suppo s he sugges ion ha spillo e e ec s occu in all di ec ions.
In bo h models, pe capi a income does no seem o s a is ically a ec he launch
p ice. Conce ning he ype o i ms, unde ou upda ed model he d ugs sold by a Solo
Licensee i m ob ain lowe launch p ices han d ugs sold by o he ypes o i ms. D&E do
no ind any signi ican e ec ela ed o he ype o i m.

Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 91
Fu he mo e, ega ding he p oduc ’s cha ac e is ics, he e ec o s eng h, as
expec ed, is sligh ly s a is ically signi ican and posi i e, while he packsize a ec s
nega i ely he launch p ice. Unde he upda ed model, s eng h seems o no ha e any
signi ican e ec on he launch p ice. Howe e , we ha e no included he a iables ela ed
o packsize in he upda ed model as we explain in de ail in he Appendix C.1. When
in oducing he adminis a ion ou e, he mo e obus esul in bo h models is ha
injec able d ugs a e s a is ically mo e expensi e han o he ypes such as o al solid
o mula ions.
In he launch p ice equa ion, bo h models ound some launch p ice di e ences
among coun ies. The obse ed pa e n is ha all signi ican coe icien s a e nega i e;
Ge many seems o p esen highe p ices on a e age han he es o he coun ies. The
andom e ec s in he D&E model show ha some coun y dummies a e posi i e.
Howe e , ou upda ed model does no suppo he hypo hesis ha i ms sell d ugs a
single p ice in o de o a oid spillo e s e ec s.
As men ioned ea lie , D&E do no include he he apeu ic class ixed-e ec s due o
collinea p oblems wi h a iables. Since ou upda ed model does no include ei he he
compe i o p ices a iable o he o de o en y wi hin class a iable, he e will no be any
collinea p oblems. We ha e been able o include he he apeu ic class e ec s in ou
upda ed model. We ha e aken he ATC-A (Alimen a y ac and me abolism) as
e e ence and we ha e ound some signi ican ixed-e ec s; howe e , in none o he
models yea ixed-e ec s we e signi ican .
3.4 Replica ing Ve nie s e al. (2011)
In his sec ion, he me hodology conduc ed in a s udy published by Ve nie s e al.
in 2011 (Ve nie s e al., hence o h) (Ve nie s e al., 2011) is applied o ou da abase o
compa e whe he esul s ha e changed due o he use o mo e ecen da a (2010 s.
2008) and i he esul s a e s ill obus using a di e en choice o he lis o coun ies. They
applied hei model o a se o a la ge se o coun ies, ich and poo , and we es ic ou
applica ion o de eloped coun ies.
3.4.1 The Ve nie s e al. model
Ve nie s e al. conside , on he one hand, he launch window o d ug i in coun y j
92
Modeling Global P icing and Launching o New D ugs
(
3.1
)
(
b
)
(), de ined as he di e ence, in mon hs, be ween he i s wo ldwide launch and he
subsequen launch in he speci ic coun y j. The launch p ice is de ined as he na u al-
loga i hm- ans o med o he ex-manu ac u e p ice a launch pe g am o d ug i in coun y
j ( ). Ve nie s e al. conside ha censo ing occu s o d ug-coun y combina ions o
which we do no obse e a launch a he end o he obse a ion window. Censo ing ime
(C
ij
) is de ined as he ime be ween he d ug- and coun y-speci ic launch da e and he
end o he obse a ion pe iod. Since he ac ual alues o and a e no obse ed
because igh censo ing is p esen , obse ed alues a e deno ed by and such
ha ,
Mo eo e , we only obse e he obse a ions o which and hus .
The s uc u al equa ions a e:
LW
ij
*


1
LP
ij
*


2
(LP
ij*
)
2


'
Z
ij1
u
ij1
LP
ij
*


1
LW
ij*


2
(LW
ij*
)
2


'
Z
ij2
u
ij2





whe e Z
ij1
and
Z
ij1
a e de ined as addi ional explana o y a iables. Z
ij1
comp ises he
coun y size, he heal h expendi u e pe capi a and he use o ce ain p icing policies such
as he ex-manu ac u e p ice egula ion, he p o i con ol, he ERP, he in e nal RP and
he pha maco-economic egula ion. Also, i comp ises he s eng h o pa en p o ec ion,
he EMA and he i m’s home coun y a iables. Addi ionally, i comp ises he ou
dimensions iden i ied by Ho s ede (Ho s ede, 1984, Ho s ede, 2001): unce ain y
a oidance, masculini y, indi idualism and powe dis ance. The a iable o compe i ion and
he a iable o summe a e also comp ised in his Z
ij1
a iable. Z
ij2
includes he same
a iables o Z
ij1
excep om he summe and he EMA a iable. Howe e , i u he
includes he in la ion a e and he daily dosage (DDD).
LW
ij
*
LP
ij
*
LW
ij
*
LP
ij
*
LW
ij
LP
ij
LW
ij
*
C
ij
LP
ij
LP
ij*
(
a
)
(
3.
2
)
(
b
)
(
a
)
**
ij ij ij ij
ij ij
L
WLWi LWC
LW C o he wise


Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 93
Following Ga en (1984) (Ga en, 1984), Ve nie s e al. conside he launch window
and he launch p ice as endogenous a iables. The e o e, he i m and he egula o may
bo h decide a launch window wi h he goal o in luencing he launch p ice and selec he
le el o launch p ice also wi h he goal o in luencing he launch window. The omi ed
a iables in he e o e ms o he launch window and launch p ice equa ions include non-
obse able s a egic a iables used by he i m and he egula o o selec he op imal
alue o he launch window and launch p ice, espec i ely. These s a egic a iables
would be expec ed o co ela e wi h he launch p ice and he launch window,
co espondingly.
Ve nie s e al., o accoun o he endogenei y be ween he launch p ices and
launch window, es ima e a sys em o simul aneous equa ions using a h ee-s age leas
squa es (3SLS) p ocedu e, as in Bayus e al. (Bayus e al., 2007). Addi ionally, he
au ho s co ec o igh -censo ing and selec i i y using he p ocedu e desc ibed in Vella
(Vella, 1993) o Woold idge (Woold idge, 2002). Random coun y e ec s a e included in
he equa ions o accoun o he ac ha he e a e epea ed obse a ions ac oss
coun ies o mos d ugs.
On he one hand, o es ima e he s uc u al launch window equa ion, hey i s
es ima e he educed o m o he launch p ice equa ion by a Tobi eg ession o he
second ype ( o accoun o he ac ha we only obse e p ices i he d ug has al eady
been launched). This launch p ice equa ion con ains wo a iables ha in luence launch
p ice bu no launch window, namely he de ined DDD and he in la ion a e, which se e
as ins umen s o he launch p ice in he launch window equa ion. The gene alized
esiduals o he educed launch p ice equa ion a e added o he launch window equa ion
as a co ec ion e m. Howe e , we only use one ins umen in ou upda ed model, since
we ha e no been able o calcula e DDDs in ou da abase71. On he o he hand, o
es ima e he s uc u al launch p ice equa ion, Ve nie s e al. i s es ima e he educed
o m o he launch window equa ion by a Tobi eg ession o he i s ype ( o accoun o
igh censo ing). This launch window equa ion con ains wo a iables ha in luence
launch window bu no launch p ice, namely, summe and ema, which se e as
ins umen s o he launch window in he launch p ice equa ion. In his case, we ha e
71 Fo some molecules o ou da abase, he DDD depends on pa ien cha ac e is ics. The e o e, a unique DDD o each
molecule could no be used.
94 Modeling Global P icing and Launching o New D ugs
included bo h ins umen s in ou model. The gene alized esiduals o he educed launch
window equa ion as a co ec ion e m a e added o he launch p ice equa ion.
3.4.2 Da a
Ve nie s e al. collec da a om he IMS Heal h da abase on d ugs in 50 coun ies
(see Table C.3 in Appendix C.2) o 5 he apeu ic classes, all o which expe ienced a
launch du ing he s udy pe iod, 1994-2008. They ha e collec ed yea ly da a on ou pa ien
sales a ex-manu ac u e p ices. P ice pe g am in US dolla s o each d ug has been
calcula ed. To make d ug p ices compa able ac oss coun ies, he d ug p ices in local
cu encies we e con e ed o US dolla s using he cu ency con e sion a e a launch.
We also use da a om IMS Heal h da abase. Howe e , we only conside he new
launch d ugs in 20 de eloped coun ies o 11 he apeu ic classes du ing he s udy pe iod
2004-2010, all o hem app o ed by he cen alised p ocedu e o he EMA. We ha e also
collec ed ou pa ien sales yea ly a ex-manu ac u e p ice. Since we ha e collec ed he
p ices in eu os, eu os ha e been con e ed in o US dolla s applying he exchange a es
om he IMF. Finally, he d ug p ice has been calcula ed as done by Ve nie s e al. Thus,
we also use he p ice pe g am in US dolla s o each in o de o make esul s
compa able.
In Appendix C.2, we epo he a iable de ini ions. We dis inguish among hose
a iables ha we de ine as in Ve nie s e al and hose ha we canno use o we de ine
di e en ly. In Table C.4, we show he esul s om bo h models o be compa ed.
3.4.3 Compa ison o esul s
3.4.3.1 Launch window equa ion
The sample o Ve nie s e al. is la ge han ou sample (1711 s. 505) because
hei s udy pe iod is longe and he sample o coun ies is la ge (50 s. 20). As expec ed,
in bo h models, launch p ice a ec s nega i ely he launch window, i.e., he highe he
p ice a coun y pays o a d ug, he sho e he delay he coun y will su e . Also, bo h
models ind a posi i e quad a ic e ec ha o se s he abo e men ioned nega i e e ec
(U-shaped e ec ). Fu he mo e, bo h models ind a posi i e and signi ican coe icien o
he selec i i y a iable, sugges ing endogenei y o p ices in he launch equa ion. This
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 95
inding may indica e ha heal h egula o s ac s a egically in delaying ma ke access o
expensi e d ugs, which is agains he in e es s o he d ug company.
Conce ning he egula ion a iables, al hough esea che s ha e no examined he
di ec e ec o p o i con ol on launch window, Ve nie s e al. a gue ha i may slow
ma ke access. Howe e , in ou upda ed model, p o i con ol egula ion seems o a ec
he launch window nega i ely. Indeed, he only coun y applying p o i con ol in ou
da ase is he UK, which does no su e pa icula ly om long launch delays. ERP72 may
show coun e in ui i e e ec s (Hun e , 2005). Fi s , when a coun y applies ERP, i ms will
y o gain ma ke access as ea ly as possible o minimize he numbe o e e ence
coun ies. Second, ERP may push p ices upwa d a he han downwa d. Typically,
egula o s ha seek ea ly d ug access a e mo e willing o ag ee o highe p ices. Thus,
he likelihood o a e e ence coun y ha ing a high p ice is highe ea ly in he li e cycle
han i is la e on, as we discussed in chap e 2. Consequen ly, he e e ence se o a
coun y is likely o con ain a g ea e numbe o coun ies wi h high p ices ea ly in he li e
cycle as compa ed o la e in he li e cycle. Ou upda ed model shows a signi ican and
nega i e coe icien o his egula ion a iable; he e o e, i con i ms he hypo hesis
p oposed by Ve nie s e al. As said by Ve nie s e al., ypically, he apeu ic e e encing
delays launch because he adminis a i e p ocedu e equi es an examina ion o
he apeu ic simila i ies, delaying ma ke access. Howe e , and con a y o he esul s
ob ained by Ve nie s e al., ou upda ed model epo s a nega i e and signi ican
coe icien o his a iable. The e o e, hose coun ies using he apeu ic e e ence p icing
expe ience sho e launch delays.
Pha macoeconomic e idence, in addi ion o he clinical e idence equi ed o gain
he apeu ic app o al om ins i u es such as he FDA (Food and D ug Adminis a ion) o
EMA, also equi es e idence on he cos e ec i eness o he d ug in he local popula ion,
and i mus be submi ed acco ding o complica ed adminis a i e p ocedu es. This
equi emen o en causes a delay in ma ke access simila o he apeu ic e e ence p icing
(Wilking e al., 2005) as we discusses in chap e s 1 and 2. Resul s epo ed by ou
upda ed model seem o suppo he esul s and he hypo hesis p oposed by Ve nie s e
al.
72 This a iable is named by Ve nie s e al.as C oss-coun y e e ence p icing.

96 Modeling Global P icing and Launching o New D ugs
Conce ning he s eng h o pa en p o ec ion, i is known ha high s eng h o
pa en p o ec s he i m om bio-equi alen p ice compe i ion. Thus, a highe s eng h o
pa en p o ec ion in a coun y may yield quicke access o d ugs. In bo h models, hose
coun ies wi h s ong pa en p o ec ion show sho e launch delays.
O he coun y cha ac e is ics conside ed a e popula ion size, heal h expendi u e
pe capi a and dummies o he i m’s home coun y. The ba gaining powe gi en by he
popula ion size is shown in ou upda ed model since he e ec o his a iable is
signi ican and nega i e. The e o e, i suppo s he hypo hesis and he esul s shown by
he Ve nie s e al. model. Rega ding heal h expendi u e pe capi a, Ve nie s e al. p opose
ha i ms may be mo e eage o launch in coun ies wi h high heal h expendi u es pe
capi a, as hese coun ies may ha e a mo e a ou able a i ude owa ds new d ugs.
Howe e , highe heal h expendi u es pe capi a could lowe heal h egula o s’ aspi a ions
o p o ide quick ma ke access o new d ugs (Comano and Schwei ze , 2007). In bo h
models, he e ec o heal h expendi u e pe capi a on he launch window is signi ican and
posi i e, suppo ing he second idea p oposed by Ve nie s e al. Conce ning a i m’s
loca ion, unde he Ve nie s e al. model, i ms wi h a g ea e amilia i y wi h he home
ma ke 's he apeu ic needs o heal h egula o s' a ou i ism owa d hese i ms may lead
o a as e launch (Kyle, 2006). In his case, bo h models suppo he hypo hesis
desc ibed abo e.
A a iable exclusi ely a ec ing he launch window is he dummy equal o one i he
coun y belongs o he EMA. Acco ding o Ve nie s e al., belonging o he EMA should
a ec nega i ely he launch window. Al hough ma ke access and p ice nego ia ions ake
place a coun y le el, he d ug app o al p ocess in Eu ope is cen alized. I is expec ed
ha launch windows in EMA coun ies a e sho e han hose in non- EMA coun ies
because o di e ences in adminis a i e e iciencies. Again, he Ve nie s e al. model
suppo s his hypo hesis while ou upda ed model shows a high signi ican posi i e e ec
(coun ies no belonging o he EMA enjoy sho e launch delays han coun ies belonging
o he EMA). This could be due o he di e ences in he sample o coun ies. In ou
da abase, only coun ies no belonging o he EMA a e high-p ice coun ies. In he
Ve nie s e al. da abase he e a e a lo o low-p ice coun ies and e y low-p ice coun ies.
Ou o he ou dimensions iden i ied by Ho s ede (Ho s ede, 1984, Ho s ede,
2001)) conce ning a coun y’s na ional cul u e- unce ain y a oidance, masculini y,
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 97
indi idualism and powe dis ance (de ined in Appendix C.2) – only he masculini y and he
powe dis ance p esen he same e ec s in bo h models and suppo he hypo hesis
s a ed by Ve nie s e al., he mo e masculine he socie y is and he mo e bu eauc a ic i is
(highe powe a oidance), he longe he launch window is. Ve nie s e al. expec ed and
showed ha , on he one hand, low subjec i e heal h pe cep ions (high le el o unce ain y
a oidance) may encou age heal h egula o s o allow p omp access o new d ugs and o
be less p ice sensi i e, and on he o he hand, coun ies showing a g ea e sa is ac ion
owa d heal h ca e and spending mo e money on heal hca e (high le el o indi idualism)
enjoy sho e launch windows han collec i is coun ies. Howe e , ou upda ed model
p esen s he opposi e e ec s.
Finally, he d ug he apeu ic ixed-e ec s (no epo ed) seem o be s a is ically
signi ican unde he Ve nie s e al. model and in ou upda ed model.
3.4.3.2 Launch p ice equa ion
Rega ding he explana o y ac o s o he launch p ice equa ion, unde ou upda ed
model, he launch window does no seem o a ec he launch p ice, while he Ve nie s e
al. model inds a signi ican and posi i e e ec (nega i e o he quad a ic e m) as
expec ed unde hei hypo hesis. Indeed, Ve nie s e al. p opose an in e ed U-shaped
e ec o launch window on launch p ice in which launch p ice is highes o mode a e
launch windows. Fo hese mode a e launch windows, a i m can s ill make money unde
pa en p o ec ion i he p ice is high enough o make up o local ma ke en y
expendi u es. Fo e y sho launch windows, a i m will accep a lowe launch p ice mo e
easily because he d ug enjoys a ull li e ime unde pa en p o ec ion, so he i m can
eco e R&D expendi u e and gains esou ces o in e na ional ma ke access
immedia ely. Fo e y long launch windows, a i m and a heal h egula o will ag ee mo e
easily on a ela i ely low launch p ice as a p elude o gene ic compe i ion.
Rega ding he egula ion a iables, nei he o he models ind any signi ican e ec
on launch p ices, excep om he s eng h o pa en p o ec ion, whe e only he Ve nie s e
al. model inds a nega i e and signi ican e ec as expec ed: s onge pa en p o ec ion
may impose a downwa d p essu e on launch p ices because pha maceu ical i ms can be
mo e lenien on p ices i he e is su icien ime le unde pa en p o ec ion o eco e
R&D expendi u e.
98 Modeling Global P icing and Launching o New D ugs
Among o he coun y cha ac e is ics conside ed such as popula ion size, heal h
expendi u e pe capi a and he i m home’s coun y, only he i m home’s coun y seems
o be s a is ically signi ican and posi i e, bu only unde he Ve nie s e al. model,
suppo ing hei hypo hesis. As p e iously men ioned, a g ea e amilia i y wi h he home
ma ke 's he apeu ic needs o heal h egula o s' a ou i ism owa d hese i ms may lead
o a highe launch p ice (Wagne and McCa hy, 2004). The d ug he apeu ic ixed-e ec s
(no epo ed) a e no s a is ically signi ican .
3.5 New P icing and Launching Model (NPLM)
3.5.1 The Model
We es ima e he launch delay and he ela i e launch p ice equa ions sepa a ely,
and each o hem is es ima ed o e ail and o hospi al dis ibu ion channels. We ha e
also ied o es ima e a sys em o bo h equa ions o accoun o endogenei y. Howe e ,
he a ailable ins umen o he ela i e launch p ice equa ion is weak73.
We use a pa ame ic du a ion model o he haza d o launching in ime , gi en he
obse ed explana o y a iables, wi h igh -censo ed da a o model he launch delay o he
molecule i in coun y j, which is de ined as he ime elapsed in mon hs om he i s global
launch o he molecule i and i s launch in coun y j. We ha e speci ied he shape o he
haza d a e, i.e. i s ime-dependency, wi h a Weibull dis ibu ion ha assumes a
mono onic haza d wi h espec o ime. Since we ha e no been able o obse e all
a iables a ec ing he launch delay, we ha e con olled o he unobse ed he e ogeini y
in oducing a gamma ail y dis ibu ion o he andom e o e m, The model selec ion
has ollowed he me hod p oposed by Kie e (Kie e , 1988) (see Appendix C.4). We
es ima e a igh -censo ed model since all he d ugs in ou da a se we e launched
be ween Janua y 2004 and Decembe 2010; howe e , no all d ugs had been launched in
all 20 coun ies by he end o ou obse ed pe iod. The e o e, ou da a con ain igh -
censo ed obse a ions. Ou pa ame ic du a ion model o he haza d o launching does
no allow he use o ime- a ying co a ia es74; ins ead we ha e used he da a collec ed in
73 We selec ed he a iable in la ion as he ins umen o he launch p ice in he launch equa ion. The co ela ion be ween
in la ion and launch p ice was weak (0.02), he e o e, we should no use i as ins umen .
74 Since we ejec he null hypo hesis o he log- ank es , hen he assump ion o p opo ional haza d is no sa is ied, we
should no nei he inco po a e ime- a ying no use he s anda d Cox eg ession. Fu he mo e, he ex ended Cox model
allows inco po a ing ime- a ying co a ia es bu we should no use i since i does no allow inco po a ing censo ing.
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 99
he base yea 75. Thus, we ha e:
whe e , he subindex i=molecule, j=coun y, h is he haza d a e o launching, Xij
a e he co a ia es, he co a ia es’ pa ame e s, he ime elapsed un il he launch o
molecule i occu s in coun y j, p he shape pa ame e 76, U a andom a iable and he
a iance o he ail y77 (Keele, 2007, Jenkins, 2008). The co a ia es Xij a e he ela i e
launch p ice a molecule le el, he loga i hm o he coun y size (popula ion), he loga i hm
o he public heal h expendi u e pe capi a, he loga i hm o he pha maceu ical
expendi u e pe capi a, he dummy o he i m’s headqua e s loca ion in he launching
coun y, he dummy o belonging o he EMA and he apeu ic ixed-e ec s a ATC-1 le el.
These a iables a e de ined in de ail in Appendix C.3.
Rega ding he ela i e launch p ice equa ion, we use OLS wi h molecule-
p esen a ion-clus e ed s anda d e o s o model he log o he ela i e launch p ice o
molecule i, p oduc k in coun y j a he ime , condi ional on launching. The ela i e
launch p ice is de ined as he p ice a io be ween he launch p ice o molecule i, p oduc k
in coun y j a he ime , and he launch p ice o molecule i, p oduc k in coun y g a he
i s global launch ime 0. To accoun o unobse ed molecule cha ac e is ics, we also
epo esul s om a GLS (Gene alized Leas Squa es) andom e ec s es ima o . To
accoun o possible selec ion bias p oduced by he co ela ion be ween he p opensi y o
launch and he launch p ice, we also es ima e a Heckman selec ion model wi h a i s -
s age p obi eg ession (Heckman, 1979). Then, we ha e he p obi selec ion equa ion:
75 Fo hese co a ia es, such as coun y popula ion, he GDP pe capi a, heal h and pha maceu ical expendi u e pe capi a
a iables, we use he da a o he base yea (2004). These co a ia es a e in he model o con ol o di e ences in coun y
sizes, weal h and expendi u e, which a e well ep esen ed wi h he da a collec ed o he base yea 2004.
76 The shape pa ame e p de e mines whe he he haza d is inc easing, dec easing, o cons an o e ime.
77 By es ing he hypo hesis = 0 using a likelihood a io es , we de e mine whe he we need o wo y abou unobse ed
he e ogenei y.
h( ,X)

p(

)
p1
[U]


ij
e
X
ij




(3.3)
106 Modeling Global P icing and Launching o New D ugs
Table 3.7. Rela i e launch p ice equa ion o he NPLM
Re ail Hospi al
OLS w/ Robus Clus e ed SEs No mal Random E ec s OLS w/ Robus Clus e ed SEs No mal Random E ec s
Delay -0.0722 0.0053 -0.0726 0.0137
[0.1296] [0.1199] [0.1695] [0.1490]
Delay*delay 0.0023 0.0008 0.0000 1.41e-06
[0.0021] [0.0008] [0.0000] [6.87e-06]
Log o Coun y size (popula ion) -8.6599* -5.8246*** -10.1434** -4.2420*
[4.7556] [1.9964] [4.4531] [2.4337]
Log o GDP pe capi a -9.5648 -2.6522 7.7304 1.7151
[9.7722] [16.5960] [12.6634] [19.4184]
Log o Heal h Expendi u e pe capi a 24.6213* 14.3926*** 18.0517* 4.1666
[13.9677] [12.7490] [9.8605] [14.9830]
Log o Pha maceu ical Expendi u e pc 50.7958* 37.3478 70.6921** 29.2789
[27.4052] [13.0590] [31.7399] [18.2923]
IRP -5.8320 -2.7864 -8.9012 -4.8740
[4.1272] [3.3182] [6.0281] [4.3355]
Fi m’s home coun y -4.1632 -4.1037 -19.7583* -5.1026
[3.5854] [5.8000] [11.5931] [8.3115]
Yea
2004 RC RC RC RC
RC RC RC RC
2005 0.6371 -2.8228 3.6965 -1.4028
[2.3904] [7.3313] [3.3774] [9.1615]
2006 -0.9426 -4.1618 4.4947 1.0742
[2.8771] [8.1672] [4.3652] [10.8644]
2007 -3.6626 -6.0497 -3.1937 -0.3632
[5.1028] [8.4271] [5.5596] [11.4744]
2008 -4.4736 -8.0594 -1.3471 1.0891
[5.0712] [8.8661] [5.6821] [12.3694]
2009 -2.1013 -8.1843 14.1190 1.3525
[3.6430] [9.2742] [11.2567] [13.2468]
2010 -16.3579 -16.5441 -7.6186 -4.3132
[10.4757] [10.2073] [9.7540] [14.6237]
IMR 60.7956* 45.6041*** 80.9075** 25.3029
[33.2356] [16.1445] [36.1181] [23.8379]
Cons an -19.4495 -303.1349** -648.5746* -210.4026
[87.4215] [140.4957] [329.3245] [207.2856]
Obse a ions 1334 1334 1369 1369

Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 107
Re ail Hospi al
OLS w/ Robus Clus e ed SEs No mal Random E ec s OLS w/ Robus Clus e ed SEs No mal Random E ec s
Numbe o Molecule-p esen a ion-le el
Clus e s 69 69 70 70
R-squa ed 0.1776 0.1744 0.1952 0.1952
Signi icance(sign.)le els( wo‐sided):*:p<0.10;**:p<0.05;***:p<0.01.;[]:s anda de o ;n. .:no‐ epo ed;‐:no‐included;RC: e e enceca ego y.Seede ini iono  a iablesin
AppendixC.3.
108 Modeling Global P icing and Launching o New D ugs
Fu he mo e, o he coun y cha ac e is ics such as pha maceu ical and heal h
public expendi u e pe capi a do seem o a ec signi ican ly he ela i e launch p ice.
Indeed, as expec ed, coun ies wi h high pha maceu ical and public heal h expendi u es
pe capi a pay highe ela i e launch p ices. In addi ion, he esul s show ha coun ies
wi h a high ba gaining powe , since hey ha e a la ge popula ion, pay lowe ela i e
launch p ices on a e age (see Chap e 2 sec ion 2.1). As we accoun o unobse ed
molecule cha ac e is ics, we also epo esul s om a GLS andom e ec s es ima o . The
esul s sligh ly change when we es ima e his al e na i e speci ica ion. Pa icula ly, only
he coun y size and public heal h expendi u e pe capi a emain as signi ican ac o s
in luencing he ela i e launch p ices.
When we analyse he hospi al sales, we obse e ha , compa ed o he e ail
ma ke , signi ican esul s emain. Fu he mo e, in his analysis, he i m’s headqua e s’
loca ion has a sligh ly signi ican and nega i e e ec on he ela i e launch p ice.
The e o e, d ugs launched by i ms wi h hei headqua e s in he launching coun y se
lowe p ices han d ugs launched by i ms wi h hei headqua e s ou side he launching
coun y. This unexpec ed e ec will be discussed la e on in his chap e , and i can be
compa ed wi h he esul s epo ed in he li e a u e in Chap e 1 Sec ion 3.2.5. Simila o
he e ail ma ke , when we epo he esul s om a GLS andom e ec s es ima o , he
only signi ican e ec ha emains is he coun y size, he es o a iables a ec ing he
ela i e launch p ice become insigni ican . E en, he IMR does no a ec signi ican ly he
ela i e launch p ice, only being signi ican in he e ail ma ke o bo h speci ica ions and
in he hospi al ma ke o he OLS molecule-clus e ed es ima e.
3.5.4 Discussion
Ou con ibu ion o he p e ious li e a u e analysed in Chap e 1, sec ions 2 and 3,
i s ly consis s o he analysis o he da abase a p esen a ion le el, he analysis o he
ela i e launch p ice as endogenous a iable in he launch p ice equa ion, he s udy o he
launch delay as a du a ion ime a iable and he analysis o he inpa ien s ma ke . In his
chap e , we ha e ca ied ou an analysis o he ade-o be ween p icing and launching
and he impac o ERP policy on bo h p icing and launching.
In his ega d, we ha e obse ed ha he launch delay does no signi ican ly a ec
he ela i e launch p ice; howe e , he ela i e launch p ice does a ec launch delay, bu
he ex en o he in luence is qui e low. In addi ion, he esul s show ha he use o ERP
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 109
makes coun ies expe ience longe launch delays bu does no lead o paying lowe
ela i e launch p ices. These esul s may ha e se e al implica ions on he ba gaining
p ocess. Indeed, we may hink ha i ms do no wan o play he game in which, coun ies
ejec a i m’s o e knowing ha o e he ime hey will ob ain lowe p ices. Besides, we
obse e ha i ms delay launches in coun ies using ERP policy; howe e , hese coun ies
do no necessa ily pay lowe p ices. This las esul may indica e ha ERP policy is no
e ec i e in “p icing e ms” bu i is in “launching e ms”. I seems ha i ms do no accep
lowe p ices in exchange o delaying launches om coun ies applying ERP. These
esul s may sugges ha i ms basically delay launches because coun ies p obably
canno a o d o ha e he p oduc a ailable s aigh om he global launch, and e en no
ha ing he p oduc . In con as o p e ious li e a u e, whe e i ms some imes delay launch
o a oid spillo e e ec , in ou s udy, we show ha i ms may use a mo e agg essi e
s a egy, which does no allow coun ies o ha e he p oduc s a ailable wi h a launch
delay in exchange o paying lowe ela i e launch p ices. Unde his s a egy, i ms
would a oid he spillo e e ec s om ERP policy and PT, hough hey would lose p o i s
om sales in hose coun ies whe e he molecule is no ul ima ely launched.
Fu he mo e, among o he coun y cha ac e is ics, we obse e ha he ba gaining
powe o coun y size is e ec i e o ob ain lowe p ices; howe e , his coun y
cha ac e is ic does no seem o be an in luencing ac o on achie ing sho e launch
delays, e en mo e, unexpec edly, coun ies wi h a la ge coun y size ind lowe
p obabili ies o ha e a p oduc launched. In he same line, we ha e obse ed ha GDP
pe capi a does no a ec he ela i e launch p ice; howe e , o he coun y cha ac e is ics,
mo e speci ically ones a ec ing pha maceu ical consump ion, such as pha maceu ical
and heal h public expendi u e pe capi a, a ec posi i ely he ela i e launch p ice. Exac ly
he opposi e e ec occu s o he launch delay. The pha maceu ical and he public heal h
expendi u e do no seem o esul in coun ies expe iencing sho e launch delays.
Indeed, wha does make coun ies ha e p oduc s a ailable in he sho - e m is a high
le el o weal h pe capi a. We may say ha weal hy coun ies ha e he p oduc s a ailable
in he sho - e m, and he coun ies ha ul ima ely pay high ela i e launch p ices a e
hose ha alloca e la ge budge s o public heal h and he pha maceu ical expendi u e.
On he basis o he esul s, i ms nei he make discoun s no launch in he sho -
e m in hose coun ies whe e hey ha e hei headqua e s. Only in he hospi al ma ke
do we obse e ha coun ies ob ain lowe p ices om his ype o i ms han om o eign
110 Modeling Global P icing and Launching o New D ugs
ones. So a , he p e ious li e a u e has ei he no ound any signi ican p ice p emiums
o local i ms o has ound a posi i e signi ican and expec ed e ec . No e ha hese
s udies collec ed da a no only in high-income bu also in low-income coun ies whe e
i ms’ headqua e s a e no usually loca ed, he e o e, he posi i e e ec ound could be
due no exclusi ely o he i m’s loca ion bu also o he coun y’s weal h. Fu he mo e,
coun ies belonging o he EMA enjoy sho e ma ke access on a e age han coun ies
ou side o he EMA egime.
3.6 Conclusions
Conclusions om eplica ing he D&E model
The upda ed model p esen s some di e ences om he D&E model. We mus no e
ha ou sample s a s in 2004, immedia ely a e he D&E sample inishes. Also, he lis s
o coun ies and p oduc s a e no exac ly he same in bo h models. This may jus i y
di e ences in esul s. The D&E model is obus among di e en speci ica ions, while ou
upda ed model p esen s some al e a ions. The mos ema kable di e ences in esul s
be ween he D&E model and he upda ed one conce n he spillo e e ec s and he e ec s
o he ype o i ms.
The upda ed model inds ha he numbe o low-p ice EU coun ies a ec s
nega i ely he p opensi y o launch, which may con i m ha low-p ice coun ies a e
su e ing longe launch delays, while he D&E model does no ind any signi ican e ec .
Fu he mo e, he upda ed model inds ha he minimum p ice se in he low-p ice EU
a ec s posi i ely he launch p ice, which may show ha spillo e e ec s also exis among
low-p ice coun ies, while he D&E model does no . Mo eo e , he minimum p ice se in
high-p ice non-EU coun ies p esen s a nega i e e ec on he launch p ice (posi i e in he
D&E model) which may indica e ha spillo e e ec s occu among EU and high-p ice
non-EU coun ies. To ha e no e e ence p ices om low-p ice EU and high-p ice non-EU
coun ies only seems o ha e e ec in he upda ed model. No e e ences om low-p ice
EU coun ies yields a posi i e e ec due o wo di e en si ua ions, ei he ha spillo e
also occu s among low-p ice coun ies, o ha being i s means paying highe p ices. No
e e ences om high-p ice non-EU coun ies may indica e ei he ha hese coun ies a e
also aken as e e ence by EU coun ies o ha launch p ices in he EU could be highe
han he launch p ice ou o he EU.
Chap e 3: Global P icing and Launching o New D ugs. An Econome ic App oach 111
Fu he mo e, acco ding o he D&E model, Local Co po a ions enjoy a highe
p opensi y o launch, bu ou model does no ind any signi ican e ec om his
cha ac e is ic. The o he way a ound occu s o he launch p ice, whe e he D&E model
does no indica e any signi ican e ec on launch p ices while he upda ed model shows
ha Solo Licensee i ms ob ain lowe launch p ices. Bo h models p esen signi ican
coun y- ixed e ec s in he launch and launch p ice equa ion.
Conclusions om eplica ing he Ve nie s e al. model
The same da a ea men and me hodology conduc ed by Ve nie s e al. ha e been
implemen ed o ou da abase. Di e ences in he lis o coun ies and d ugs s udied, and
in he ime pe iod co e ed may jus i y some o he di e ences in he esul s abo e.
Ve nie s e al. ind e idence o endogenei y o bo h launch p ices and launch delays. We
only ind he e ec in a single way; launch delay is nega i ely a ec ed by launch p ices
bu no he o he way a ound.
The e ec s om some egula o y policies do no seem o coincide. Some
egula o y policies adi ionally posi i ely a ec ing he launch window, such as p o i
con ol o he he apeu ic e e ence p icing, a e no signi ican in he upda ed model.
Howe e , bo h models show ha he use o pha maco-economic e idence egula ion
leads o longe launch delays. Also, in bo h models, he s onge he s eng h o pa en is,
he sho e launch delays a e. Only ou upda ed model p esen s a signi ican and
expec ed e ec o he use o ERP egula ion. On he o he hand, he egula o y policies
do no show signi ican e ec s on he launch p ice unde any models, excep o he
s eng h o pa en p o ec ion; s onge pa en p o ec ion imposes a downwa d p essu e on
launch p ices.
Resul s om o he coun y cha ac e is ics like popula ion size and heal h
expendi u e pe capi a, when s a is ically signi ican (only in he launch window equa ion)
a e conco dan in bo h models. Among o he coun y cha ac e is ics, coun ies ha hos a
i m’s headqua e s o he i m launching he d ug expe ience sho e launch delays unde
bo h models. This e ec is no signi ican in he launch p ice equa ion unde he upda ed
model bu posi i e unde he Ve nie s e al. model, suppo ing hei hypo hesis ha i ms
se led in he launch coun y enjoy highe p ices. Fu he mo e, belonging o he EMA
shows he opposi e e ec s. Acco ding o Ve nie s e al. coun ies belonging o he EMA

112 Modeling Global P icing and Launching o New D ugs
show sho e launch delays, howe e , ou upda ed model epo s ha hese coun ies
expe ience longe launch delays han in coun ies ou side o he EMA.
Finally, bo h models p esen signi ican he apeu ic class ixed-e ec s. Howe e ,
he ou dimensions iden i ied by Ho s ede (Ho s ede, 1984, Ho s ede, 2001) conce ning a
coun y’s na ional cul u e p esen di e en e ec s in bo h models.
Conclusions o he NPLM
Unde he NPLM, he p icing and launching seem o be no longe ela ed o each
o he . Di e ences exis in p ices ac oss coun ies bu no due o he launch delay. Fi ms
do no accep lowe p ices in exchange o delaying launches, e en om coun ies
applying ERP policy, he e o e, ERP policy seems o no be e ec i e in “p icing e ms” bu
is in “launching e ms”. These esul s may lead o se e al implica ions in he ba gaining
p ocess. We sugges ha i ms basically delay launches because coun ies p obably
canno a o d o ha e he p oduc a ailable s aigh om he global launch, and ul ima ely
end up no ha ing he p oduc launched. While he i ms o en delay launch o a oid
spillo e e ec s, unde ou s udy, we show ha he i ms may conduc a mo e agg essi e
s a egy ha does no allow coun ies o pay lowe p ices in exchange o expe iencing
longe launch delays. Unde his s a egy he i ms would a oid he spillo e e ec s om
IRP policy and PT, bu hey would also lose p o i s om sales in coun ies whe e he
molecule is no ul ima ely launched.
Rega ding o he coun y cha ac e is ics, ou s udy shows ha weal hy coun ies
ha e he p oduc s a ailable in a sho e pe iod, bu he coun ies ha ul ima ely pay high
ela i e launch p ices a e hose ha alloca e la ge budge s o public heal h and
pha maceu ical expendi u e. Coun ies belonging o he EMA seem o enjoy sho e
launch delays han he coun ies ou side o i ; howe e , he e a e no signi ican p ice
di e ences be ween coun ies unde he EMA egime and coun ies ou side o i .
In gene al, he esul s in he e ail ma ke and he hospi al ma ke do no show
huge di e ences, bu we highligh he i m wi h headqua e s in he launching coun y
ba gains lowe p ices wi h he coun y conce ned jus in he hospi al ma ke .
Conclusions and u he esea ch
Ou sys ema ic e iew shows ha demog aphic and income coun y ea u es, and
egula ion egimes, seem o be he mos impo an ac o s a ec ing d ug p icing and
launching. Howe e , p ice egula ion can unde mine he e ec s o hese impo an ac o s.
D ug cha ac e is ics like s eng h, packsize and p esen a ion o ms a e signi ican ly
ela ed o he p ice. Also, suppo ed by p e ious s udies, belonging o he EMA’s
he apeu ic ca ego y obus ly a ec s he launch delay and he launch p ice. The
he apeu ic alue is shown in he p e ious li e a u e as a obus ac o -in luencing d ug
p icing and launching. Addi ionally, i m loca ion u ns ou o be an impo an ac o o he
launching decision; howe e , p ice p emiums due o headqua e s loca ion appea
ambiguous.
When eplica ing he D&E model wi h mo e ecen da a, we ind some new
pa e ns compa ed wi h he D&E esul s conce ning he spillo e e ec s and he ype o
i ms. The e o e, we de e mine ha low-p ice coun ies su e longe launch delays and
spillo e e ec s also exis among low-p ice coun ies: spillo e e ec s also occu
be ween EU and high-p ice non-EU coun ies. Fu he mo e, Local Co po a ions do no
ha e a highe p opensi y o launch any longe ; howe e Solo Licensee i ms do ob ain,
nowadays, lowe launch p ices.
Replica ing Ve nie s e al. wi h mo e ecen da a and a di e en lis o coun ies
yields new ou comes. The mos impo an ou come is ha he endogenei y o bo h launch
p ices and launch delays ound by Ve nie s e al. is no longe ound; only he launch delay
is nega i ely a ec ed by he launch p ice. The e ec s o some egula o y policies on he
launch delay do no gene ally seem o coincide. Again, simila o he D&E, in ou upda ed
model, i m loca ion loses i s e ec on d ug launching and p icing.
One o he mos impo an conclusions o his doc o al hesis is ha , in con as o
p e ious models, p icing and launching seem o be no longe ela ed o each o he .
114 Modeling Global P icing and Launching o New D ugs
Di e ences in p ices exis ac oss coun ies bu no due o he launch delay. Fi ms do no
accep lowe p ices in exchange o delaying launches, e en om coun ies applying ERP
policies, he e o e, ERP seems o no be e ec i e in “p icing e ms” bu i is in “launching
e ms”. These esul s may hold se e al implica ions o he ba gaining p ocess. We
sugges ha i ms basically delay launches because coun ies p obably canno a o d o
ha e he p oduc a ailable s aigh om he global launch, o ul ima ely no ha ing he
p oduc launched. While i ms used o delay launch o a oid spillo e e ec s, in ou s udy,
we show ha i ms conduc a mo e agg essi e s a egy ha does no allow coun ies o
pay lowe p ices in exchange o expe iencing longe launch delays. Unde his s a egy,
i ms would a oid he spillo e e ec s om ERP and PT, bu hey would also lose p o i s
om sales in coun ies whe e he molecule is no ul ima ely launched.
Rega ding o he coun y cha ac e is ics, being a la ge coun y helps o ha e mo e
apid ma ke access and ob ain lowe p ices. Fu he mo e, weal hy coun ies ha e he
p oduc s a ailable wi hin a sho e pe iod, bu he coun ies ha ul ima ely pay high
ela i e launch p ices a e hose ha alloca e la ge budge s o public heal h and
pha maceu ical expendi u e. Fi m loca ion no longe a ec s he p ice and nei he does he
launch delay. Coun ies belonging o he EMA seem o enjoy sho e launch delays han
he coun ies ou side o i ; howe e , he e a e no signi ican p ice di e ences be ween
coun ies inside he EMA’s egime and coun ies ou side o i .
In gene al, he esul s in he e ail ma ke and he hospi al ma ke do no show
huge di e ences, bu we highligh he i m wi h headqua e s in he launching coun y
ba gains lowe p ices wi h he coun y conce ned jus in he hospi al ma ke .
In iew o he o e all pe spec i e conce ning he main ac o s in luencing launch
p ices and launch o new d ugs based on heo e ical s udies, we mainly dis inguish wo
ypes o ac o s. Fi s ly, we ha e obse ed ac o s ha di ec ly a ec d ug p icing and
launching, such as he p esence o PT, he i m’s cha ac e is ics and he egula ion p icing
policies, o example ERP, in e nal RP o MES + PC. Secondly, ou e iew shows o he
de e minan s ha no only impac di ec ly bu also indi ec ly, a ec ing he measu es o he
i s ypes o ac o s ha in luence d ug p icing and launching, such as coun y size and
he le el o co-paymen s.
Acco ding o ou heo e ical model, gi en he op imal coun y launch sequence, we
conclude ha he smalle he popula ion size is, he mo e a ac i e he use o ERP will
Conclusions and u he esea ch 115
be, since he uni a y cos o CEA inc eases. No e ha ERP does no equi e any
in es men , howe e CEA does. The e o e, he use o ERP is help ul o ela i ely small
coun ies compa ed o he use o CEA. This esul con i ms some s a emen s on his
issue ha ha e no been p e iously shown. ERP is a low-cos p icing policy; howe e , we
ha e now mo e in o ma ion abou why coun ies apply his ype o p icing policy ully
awa e ha i may be no ai .
We also conclude ha he op imal coun y launch sequence depends on he
ela i e p ices and he ela i e coun y sizes. The e is a ade-o be ween p ice and
olume, which a ec s he coun y launch sequence. In addi ion, he ela i e p ice depends
on he coun ies’ p icing policies, ERP and CEA, and subsequen ly, on he ERP o mula.
Pa icula ly, a coun y is be e o applying ERP ins ead o CEA, i he di e ence be ween
he in e na ional e e ence p ice and he expec ed p ice o coun y i unde CEA is no
highe han he uni a y cos o CEA. This esul is a ec ed by he delay cos i only one o
he p icing policies is applied wi h delay.
F om he pe spec i e o he egula o , conce ning he applica ion o ERP, he
p e ious heo e ical li e a u e ecommends only small coun ies o engage in ERP and/o
apply ERP based on p ices in la ge coun ies (o la ge g oup o coun ies); he same
applies i one subs i u es “la ge coun y” by “small co-paymen coun y” and ice e sa.
Also, a minimum RP le el is ecommended o a oid majo inc eases in p ice le el wi h
espec o he a e age RP. Fu he mo e, an MES policy oge he wi h a PC should be
applied when he wel a e loss due o high- ype d ug buye s is no la ge enough o
ou weigh he wel a e gained due o he low- ype d ug.
F om he pe spec i e o he i m, he li e a u e ecommends ha he loss o income
coming om PT should be aken in o accoun and he i m should se a highe p ice han
wi hou PT. Howe e , since ERP is widely used by coun ies and PT does exis , he
ma ke s a e insepa able. The e o e, he highes d ug p ice is no always he bes op ion
o he i m and he lowes d ug p ice is no always he bes op ion in a gi en coun y.
Wha may ha e been an op imal p icing s a egy in a single coun y is no longe op imal
when conside ing ERP and PT.
Ou heo e ical model shows ha he e is a ade-o be ween p ices and olumes
a ec ing he coun y launch sequence. Also, i p o ides in o ma ion abou ERP, an issue
122 Modeling Global P icing and Launching o New D ugs
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
obse ed in he p e ious yea )
* (+) Dummy o absence o a global p ice o
e e ence
* (+) A e age global p ice o he molecule in US
eal $
* (+) A dummy aking one i he co po a ion is
local-non mul ina ional
A dummy aking one i he co po a ion is local bu
mul ina ional
Numbe o iden i ied gene ics in he ma ke
Be y index (i measu es he deg ee o
specializa ion o he co po a ion)
* (+) Fpc (F ac ion o public consump ion in
GDP)
GDP pe capi a (F ac ion o public consump ion in
GDP)
* (+) GDP pe capi a in US dola s
Reg (Le el o egula ion: 1 low, 2 medium, 3
high)
* (-) Fpc*Reg2
* (-) Fpc*Reg3
* (+) GDP*Reg2
* (+) GDP*Reg3
* (-) Molage (Time elapsed since he molecule
was launched o Decembe 31, 2003)
Log o he numbe o ma ke a molecule is p esen
Cesno mol (A dummy aking one i he molecule
was launched be o e Janua y 1, 1991)

Appendix A 123
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
Censo lag (Censo mol_lag_1)
* (-) Gene ic (A dummy aking one i he p oduc
is gene ic)
Co onado e
al. (2007)** P ice le el (ex- ac o y) Regula ion
HHI (He indähl-Hi schman Index)
*(+) Fi m size lag
*(-) New p oduc lag
* (+) Global P ice in USD o p oduc j belonging
o i m i
Bina y a iable, aking 1 i p oduc j is a
compound o molecules
* (+) Ma ke sha e o p oduc j in ma ke k
* Numbe o gene ic p oduc s in ma ke k
Time elapsed up o 2003 since molecule (ma ke )
k was launched
Co po a ion sha e in ma ke k excluding p oduc
j's sha e
Bina y a iable, aking 1 i molecule age is
censo ed in he sample
Bina y a iable, aking 1 i p oduc j was launch
da e is censo ed in he sample
Bina y a iable, aking 1 i p oduc j is a gene ic)
Weigh ed a e age mul ima ke con ac a iable
o i m i in ma ke k
Al e na i e weigh ed a e age mul ima ke con ac
a iable in ma ke k
All p oduc s a ATC-4.
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
1998-2004
IMS HEALTH
x
Danzon and
Chao
(2000a)b P ice le el (ex- ac o y) D ug and B and
Compe i ion
S eng h
Molecule Age (mon hs om he i s p oduc
launch each coun y o Sep embe 1992)
Fo m codes (The numbe o dis inc
171 molecules / 5690
p oduc s
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
10/1991-09/1992
IMS HEALTH
124 Modeling Global P icing and Launching o New D ugs
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
o mula ion o s eng hs in he molecule)
Global pene a ion (numbe o coun ies in which
he molecule is a ailable ou o he se en
coun ies in sample)
Packsize
Numbe o manu ac u e s o he molecule
The apeu ic subs i u e molecules (ATC3)
The apeu ic subs i u e molecules en y lag ( lag in
mon hs be ween each molecule launch da e and
he i s launch o he molecule, ATC3)
Danzon and
Chao
(2000b)b P ice le el (ex- ac o y) B and Compe i ion and
Regula ion
S eng h
Molecule Age (mon hs om he i s p oduc
launch each coun y o Sep embe 1992)
Fo m codes (The numbe o dis inc
o mula ion o s eng hs in he molecule)
Packsize
Numbe o manu ac u e s o he molecule
Gene ic En y Lag (lag in mon hs be ween he
p oduc ’s own launch da e and he launch da e o
he i s p oduc in he molecule)
The apeu ic subs i u e molecules (ATC3)
P oduc s pe The apeu ic Subs i u e Molecule
The apeu ic subs i u e molecules en y lag ( lag in
mon hs be ween each molecule launch da e and
he i s launch o he molecule, ATC3)
171 molecules / 5690 p oduc s
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
10/1991-09/1992
IMS HEALTH
Danzon
and
Eps ein
(2008)**
P ice a launch (ex- ac o y) and
Launch window B and Compe i ion and
Regula ion
S=Supe io ; I=In e io
* (+) S Expec ed D ug P ice Supe io B ands (lag)
Supe io B and’s P ice Missing
375 molecules
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
QI/1992-
QIV/2003
IMS HEALTH
Appendix A 125
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
* I (-) Expec ed D ug P ice In e io B ands (lag)
In e io D ug P ice Missing
Expec ed D ug Volume (lag)
Numbe o Gene ic Manu ac u e in Supe io
Subclass
Numbe o Gene ic Manu ca u e in In e io
Subclass
* I(-) Gene ics’ P ice Missing
* I(+)Fi s B and Launch in Coun y-Subclass
* S(+) I(+) Second B and Launch in Coun y-
Subclass
* S(+) I(+) Thi d o Fou h B and Launch in
Coun y-Subclass
* S(+) Min Own P ice in High-p ice EU Missing
* S(+) I(+) Min Own P ice in High-P ice EU
Min Own P ice in Low-p ice EU Missing
Min Own P ice in Low-P ice EU
* I(-) Min Own P ice High-p ice non-EU Missing
* S(+) Min Own P ice in Hi-P ice non-EU
Numbe o low-p ice EU coun ies a molecule has
al eady launched
Numbe o high-p ice EU coun ies a molecule
has al eady launched
Numbe o high-p ice non-EU coun ies a
molecule has al eady launched
Numbe o molecules in Supe io Subclass
126 Modeling Global P icing and Launching o New D ugs
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
* S(-) Numbe o molecules in In e io Subclass
PIc Sha e in Subclass
* S(+) GDP pe Capi a
* S(-) Coun y-Speci ic Qua e ly P oduce P ice
Index
* S(+) S engh
* S(-) I(-) Pack size
Fo m: O al Solid Delayed
* S(+) I(+) Fo m: Injec able
Fo m: O he
* S(-) I(-) Time since Global Launch
* S(+) I(+) Time since Global Launch Squa ed
* I(+) Fi s Global Launch be o e 1990
Fi s Global Launch in [1996-end]
* S(+) I(+) Launch by Local O igina o
Co po a ion
* S(+) I(+) Launch by Solo Licensee Co po a ion
* S(+) I(+) Launch by Local Co-ma ke e
Co po a ion
Exchange a e ( US o EUR)
* S() I() Coun y ixed e ec s
(n.a.) ATC ixed e ec s
Danzon e al.
(2005) Haza d Launch Regula ion *(+) Expec ed D ug P ice
*(+) Expec ed D ug Volume
85 molecules
Only NCE (New Chemical En i ies)
09/1994-09/1998
IMS HEALTH
x
Appendix A 127
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
*(+) Fi m’s Global Launch Expe ience (sales)
*(+) The o igina o i m’s home coun y
*(+) GDP pe capi a
* Coun y ixed e ec s
* ATC ixed e ec s
Danzon e al.
(2011)** P ice le el (ex- ac o y) Regula ion and Coun y
* (-) IMS*GENERIC indica o ( IMS: a dummy
aking one i he d ug is sold h ough s anda d
e ail channels; GENERIC: a dummy aking one
i he he gene ic is p esen in a coun y-yea )
* (-) GPRM*BRAND indica o (GPRM: a
dummy aking one i he d ug is p ocu ed by
NGOs; a dummy aking one i he d ug is he
o igina o is p esen in a coun y-yea )
* (-) GPRM*GENERIC indica o ( GPRM: a
dummy aking one i he d ug is p ocu ed by
NGOs; a dummy aking one i a gene ic is p esen
in a coun y-yea )
* (+) Pe capi a income coun y
* (-) GINI coe icien
GINI missing indica o
HIV p e . (HIV coun y p e alence a e)
* (-) The numbe o ende gene ic p oduc s in he
same he apeu ic class-coun y-yea
* (-) The numbe o e ail gene ic p oduc s in he
same he apeu ic class-coun y-yea
The numbe o o igina o p oduc s in he same
he apeu ic class-coun y-yea
* (+) O igina o molecule lag
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics) 01/2004-06/2008
IMS HEALTH / GPRMc

128 Modeling Global P icing and Launching o New D ugs
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
* (+) Gene ic molecule lag
Heue e al.
(2007)** Haza d Launch Regula ion
Coun y GDP pe capi a
Size o he coun y popula ion
A dummy aking one i he coun y use ERP
A dummy aking one i he coun y use o he
di ec p ice con ols such as cos -e ec i eness,
e c.)
A dummy aking one i he coun y use RP
* (-) A dummy aking one i he coun y use ERP
explici ily
* (-) A dummy aking one i he coun y use ERP
as a basis o hei decision making c i e ia
35 molecules
Only NCE
01/1995-12/2005
IMS HEALTH
Kana os and
Cos a-Fon
(2005) P ice le el (wholesale) Regula ion
To al ma ke size, de ined as sales o all p oduc s
Ma ke sha e o each PI p oduc wi hin each
p oduc ma ke and each impo ing coun y
A e age Euclidean Dis ance o la i ude and
longi ude be ween each impo ing and expo ing
coun y capi als
* (-) Exchange a e $
* (-) Pu chasing Powe Pa i ies in impo ing
coun y
* (-) Ma ke sha es o gene ics consump ion in a
coun y (i )
* (-) Dummy a iable o in oduc ion o he
clawback;
P ice egula ion (Dummy a iable o p ice
egula ion de ined as he in e en ion o hi d
pa y paye (na ional insu ance company) o he
go e nmen in e ms o se ing p ice o each
p oduc ( j )
19 molecules
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
QI/1997-
QIV/2002
IMS HEALTH x
Kana os and P ice le el (ex- ac o y and * (+) Numbe o yea s since molecule’s launch in 68 molecules / 100 p oduc s 2004,2007 IMS HEALTH
Appendix A 129
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
Vando os
(2011) e ail) he local ma ke
Age squa ed
gene ics (dummy a iable 1 i he e is a gene ic
compe i o p esen in he ma ke )
* (+) Dummy a iable o Uni ed S a es
* (+) Dummy a iable o Uni ed Kingdom
* (+) Dummy a iable o Mexico
Dummy a iable indica ing he impac o Heal h
Technology Assessmen being explici ly used as a
policy measu e
Dummy a iable. Indica es he p esence o
e e ence p icing
* (+) Dummy a iable. Indica es he p esence o
ee p icing
Dummy a iable. Indica es he explici use o
ERP)
Exchange a e
* () The apeu ic ixed e ec s
All ype o p oduc s (NCE, Pa en ,
o -pa en ,gene ics)
Kyle (2006) Haza d Launch Fi ms
* (+) D ug impo ance (d ug’s sha e o s ock o
Medline ci a ions o class)
* (+) Numbe o coun ies launched (Numbe o
coun ies whe e he molecule has been launched)
* (-) Numbe o coun ies launched squa ed
* (+) Mul ina ional (Fi m has launched d ugs in
10+ coun ies)
* (+) Domes ic i m ( aking 1 i headqua e s a e
loca ed in he coun y)
* (-) Po olio ( o al numbe o i m’s d ug)
1482 molecules
Only NCE and pa en p oduc s
1980-2000
PJB Pc
130 Modeling Global P icing and Launching o New D ugs
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
* (+) Common language
Common bo de
Common egula ions
* (+) Coun y expe ience (Coun o i m’s o he
d ugs launched in coun y)
Coun y-class expe ience (coun o i m’s d ug in
coun y-class ma ke )
* (+) Expe ience yea s (numbe o yea s i m has
ma ke ed in coun y)
* (-) P ice con ols (dummy a iable: coun y
using p ice con ols)
Popula ion (coun y popula ion)
Popula ion squa ed
GDP pe capi a
* (-) Numbe o new d ug in he ma ke (Coun o
d ugs in ma ke launched less han 5 yea s ago)
* (+) Numbe o new d ug in he ma ke squa ed
(Coun o d ugs in ma ke launched less han 5
yea s ago)
* (-) Numbe o old d ugs in ma ke (Coun o
d ugs in ma ke launched mo e han 5 yea s ago)
* (+) Numbe o old d ugs in ma ke squa ed
(Coun o d ugs in ma ke launched mo e han 5
yea s ago)
Numbe o po en ial compe i o s (Coun o d ugs
launched in class elsewhe e in he wo ld)
Numbe o domes ic incumben s
Numbe o o eign incumben s
The apeu ic ixed e ec s
Appendix A 131
Re e ence Dependen Va iable Focus E ec aIndependen Va iables Sample Pe iod Da a Sou ce TFc
(n.a.) Coun y ixed e ec s
(n.a.) Yea ixed e ec s
Kyle (2007) Haza d Launch Regula ion
* (+) Numbe o d ugs in he ma ke (Coun o
d ugs in he apeu ic classcoun ma ke )
Numbe o po en ial en an s
* (+) D ug impo ance (D ug's sha e o s ock o
Medline ci a ions o class)
* (+) Numbe o coun ies launched in ( Numbe
o coun ies whe e he molecule has been
launched)
* (+) P io launch in a high-p ice coun y
* (+) P io launch in a low-p ice coun y
* (-) Fi m is headqua e ed in a p ice-con olled
coun y
* (-) Po olio ( o al numbe o i m’s d ug)
* (+) Domes ic i m ( aking 1 i headqua e s a e
loca ed in he coun y )
* (+) In e na ional expe ience (Coun o coun ies
in which i m has launched any d ugs)
* (+) Coun y expe ience ( Coun o i m’s o he
d ugs launched in coun y)
* (-) P ice eeze
* (-) P ice con ols
* (-) Supply-side con ols
* (-) P ice Rank
* (+) P ice Rank*pos -1995 pe iod
P esc ibing budge s
1444 molecules
Only NCE and pa en p oduc s
1980-1999
PJB P

Re e ences
ABRAHAM, J. 2002. The pha maceu ical indus y as a poli ical playe . Lance , 360, 1498-
1502.
ATELLA, V., BHATTACHARYA, J. & CARBONARI, L. 2012. Pha maceu ical P ice
Con ols and Minimum E icacy Regula ion: E idence om he Uni ed S a es and
I aly. Heal h Se ices Resea ch, 47, 293-308.
BARDEY, D., BOMMIER, A. & JULLIEN, B. 2010. Re ail p ice egula ion and inno a ion:
e e ence p icing in he pha maceu ical indus y. J Heal h Econ, 29, 303-16.
BAYUS, B. L., KANG, W. & AGARWAL, R. 2007. C ea ing G ow h in New Ma ke s: A
Simul aneous Model o Fi m En y and P ice*. Jou nal o P oduc Inno a ion
Managemen , 24, 139-155.
BERNDT, E. R., COCKBURN, I. M. & GRILICHES, Z. 1996. Pha maceu ical Inno a ions
and Ma ke Dynamics: T acking E ec s on P ice Indexes o An idep essan
D ugs. B ookings Pape s on Economic Ac i i y: Mic oeconomics 1996,.
BERNDT, E. R., DANZON, P. M. & KRUSE, G. B. 2007. Dynamic compe i ion in
pha maceu icals: c oss-na ional e idence om new d ug di usion. Manage ial and
Decision Economics, 28, 231-250.
BORRELL, J. R. 2007. P icing and pa en s o HIV/AIDS d ugs in de eloping coun ies.
Applied Economics, 39, 505-518.
CABRALES, A. & JIMENEZ-MARTIN, S. 2007. The De e minan s o P icing in
Pha maceu icals: A e U.S. P ices Really Highe han Those o Canada? SSRN
eLib a y.
CLAXTON, K., MARTIN, S., SOARES, M., RICE, N., SPACKMAN, E., HINDE, S.,
DEVLIN, N., SMITH, P. C. & SCULPHER, M. 2013. Me hods o he Es ima ion o
he NICE Cos E ec i eness Th eshold. Yo k: Cen e o heal h Economics,
Uni e si y o Yo k.
COCKBURN, I. M. & ANIS, A. H. 1998. Hedonic Analysis o A h i is D ugs. Na ional
Bu eau o Economic Resea ch Wo king Pape Se ies, No. 6574.
COMANOR, W. S. & SCHWEITZER, S. O. 2007. De e minan s o d ug p ices and
expendi u es. Manage ial and Decision Economics, 28, 357-370.
CORONADO, F. J., JIMÉNEZ-MARTÍN, S. & MARÍN, P. L. 2007. Mul ima ke Con ac in
Pha maceu ical Ma ke s. SSRN.
DANZON, P. M. 1997. P ice Disc imina ion o Pha maceu icals: Wel a e E ec s in he
US and he EU. In e na ional Jou nal o he Economics o Business, 4, 310-322.
DANZON, P. M. & CHAO, L. W. 2000. C oss-na ional p ice di e ences o
pha maceu icals: How la ge, and why? Jou nal o Heal h Economics, 19, 159-195.
DANZON, P. M. & EPSTEIN, A. J. 2008. E ec s o Regula ion on D ug Launch and
P icing in In e dependen Ma ke s. Na ional Bu eau o Economic Resea ch, Inc,
NBER Wo king Pape s: 14041.
236 Modelling Global P icing and Launching o New D ugs
DANZON, P. M., MULCAHY, A. W. & TOWSE, A. K. 2011. Pha maceu ical P icing in
Eme ging Ma ke s: E ec s o Income, Compe i ion and P ocu emen . Na ional
Bu eau o Economic Resea ch Wo king Pape Se ies, No. 17174.
DANZON, P. M. & TOWSE, A. 2003. Di e en ial P icing o Pha maceu icals: Reconciling
Access, R&D and Pa en s. In e na ional Jou nal o Heal h Ca e Finance and
Economics, 3, 183-205.
DANZON, P. M., WANG, Y. R. & WANG, L. 2005. The impac o p ice egula ion on he
launch delay o new d ugs - e idence om wen y- i e majo ma ke s in he 1990s.
Heal h Economics, 14, 269-292.
DJANKOV, S., LA PORTA, R., LOPEZ-DE-SILANES, F. & SHLEIFER, A. 2002. The
Regula ion o En y. Qua e ly Jou nal o Economics, 117.
DONALDSON, C., BAKER, R., MASON, H., JONES-LEE, M., LANCSAR, E., WILDMAN,
J., BATEMAN, I., LOOMES, G., ROBINSON, A. & SUGDEN, R. 2011. The social
alue o a QALY: aising he ba o ba ing he aise? BMC heal h se ices
esea ch, 11, 8.
EKELUND, M. & PERSSON, B. 2003. Pha maceu ical p icing in a egula ed ma ke .
Re iew o Economics and S a is ics, 85, 298-306.
ESPIN J, ROVIRA J & OLRY DE LABRY A 2011. WHO/HAI P ojec on Medicine P ices
and A ailabili y. Re iew Se ies on Pha maceu ical P icing Policies and
In e en ions. Wo king Pape 1: Ex e nal Re e ence P icing.
FELDMAN, R. & LOBO, F. 2013. Compe i ion in p esc ip ion d ug ma ke s: he oles o
adema ks, ad e ising, and gene ic names. Eu opean Jou nal o Heal h
Economics, 14, 667-675.
GANSLANDT, M. & MASKUS, K. E. 2004. Pa allel impo s and he p icing o
pha maceu ical p oduc s: E idence om he Eu opean Union. Jou nal o Heal h
Economics, 23, 1035-1057.
GARCIA MARIÑOSO, B., JELOVAC, I. & OLIVELLA, P. 2011. Ex e nal e e encing and
pha maceu ical p ice nego ia ion. Heal h Economics, 20, 737-756.
GARCÍA-MARIÑOSO, B. & OLIVELLA, P. 2012. In o ma ional spillo e s and s a egic
launch delay o pha maceu ical d ugs. Mimeo.
GAREN, J. 1984. The e u ns o schooling: A selec i i y bias app oach wi h a con inuous
choice a iable. Econome ica: Jou nal o he econome ic socie y, 1199-1218.
GINARTE, J. C. & PARK, W. G. 1997. De e minan s o pa en igh s: A c oss-na ional
s udy. Resea ch Policy, 26, 283-301.
GRABOWSKI, H., VERNON, J. & DIMASI, J. A. 2002. Re u ns on esea ch and
de elopmen o 1990s new d ug in oduc ions. Pha macoeconomics, 20, 11-29.
HECKMAN, J. J. 1979. Sample selec ion bias as a speci ica ion e o . Econome ica:
Jou nal o he econome ic socie y, 153-161.
HEUER, A., MEJER, M. & NEUHAUS, J. 2007. The Na ional Regula ion o
Pha maceu ical Ma ke s and he Timing o New D ug Launches in Eu ope. Kiel
ad anced s udies wo king pape s, No. 437.
HOFSTEDE, G. 1984. Cul u e's consequences: In e na ional di e ences in wo k- ela ed
alues, sage.
HOFSTEDE, G. 2012. Ho s ede Cen e [Online]. A ailable: h p://gee -ho s ede.com/ he-
ho s ede-cen e.h ml.
HOFSTEDE, G. H. 2001. Cul u e's consequences: Compa ing alues, beha io s,
ins i u ions and o ganiza ions ac oss na ions, Sage.
HOTELLING, H. 1929. S abili y o compe i ion. Economic Jou nal, 39, 41-57.
Re e ences 237
HUNTER, D. 2005. Gua an eed u u e pain and su e ing: The ecen esea ch on d ug
p ice con ols. He i age Founda ion WebMemo.
JELOVAC, I. 2008. On he ela ionship be ween he nego ia ed p ice o pha maceu icals
and he pa ien s’ co-paymen . Cen e o Economic De elopmen , Hong Kong
Uni e si y o Science and Technology WP 0208.
JELOVAC, I. & HOUY, N. 2013. D ug launch iming and in e na ional e e ence p icing.
GATE Lyon S E ienne WP 1301.
JENKINS, S. P. 2008. Su i al Analysis
. Unpublished manusc ip : Ins i u e o Social and Economic Resea ch, Uni e si y o
Essex, Colches e . Ohio S a e Uni e si y.
KANAVOS P & COSTA-FONT J 2005. Pha maceu ical pa allel ade in Eu ope:
s akeholde and compe i ion e ec s. Economic Policy, 20, 751-798.
KANAVOS, P. G. & VANDOROS, S. 2011. De e minan s o b anded p esc ip ion
medicine p ices in OECD coun ies. Heal h Economics Policy and Law, 6, 337-
367.
KEELE, L. 2007. C oss alida ion es s o ail y models. Unpublished manusc ip , Ohio
S a e Uni e si y.
KIEFER, N. M. 1988. Economic du a ion da a and haza d unc ions. Jou nal o economic
li e a u e, 26, 646-679.
KLEIN, J. P. & MOESCHBERGER, M. L. 1997. S a is ics o biology and heal h. Su i al
analysis: echniques o censo ed and unca ed da a.
KYLE, M. K. 2006. The ole o i m cha ac e is ics in pha maceu ical p oduc launches.
RAND Jou nal o Economics, 37, 602-618.
KYLE, M. K. 2007. Pha maceu ical p ice con ols and en y s a egies. Re iew o
Economics and S a is ics, 89, 88-99.
LANCASTER, T. 1992. The econome ic analysis o ansi ion da a, Camb idge uni e si y
p ess.
LANJOUW, J. O. 2005. Pa en s, P ice Con ols, and Access o New D ugs: How Policy
A ec s Global Ma ke En y. Na ional Bu eau o Economic Resea ch Wo king
Pape Se ies, No. 11321.
LEE, L.-F. 1983. Gene alized econome ic models wi h selec i i y. Econome ica: Jou nal
o he econome ic socie y, 507-512.
LEOPOLD, C., VOGLER, S., MANTEL-TEEUWISSE, A. K., DE JONCHEERE, K.,
LEUFKENS, H. G. & LAING, R. 2012. Di e ences in ex e nal p ice e e encing in
Eu ope: a desc ip i e o e iew. Heal h Policy, 104, 50-60.
LICHTENBERG, F. R. 2005. The Impac o New D ug Launches on Longe i y: E idence
om Longi udinal, Disease-Le el Da a om 52 Coun ies, 1982–2001.
In e na ional Jou nal o Heal h Ca e Finance and Economics, 5, 47-73.
LOBO, F. 2014. La In e ención de P ecios de los Medicamen os en España, Mad id,
Sp inge .
LOBO, F. & FELDMAN, R. 2013. Gene ic D ug Names and Social Wel a e. Jou nal o
Heal h Poli ics Policy and Law, 38, 573-597.
LOPEZ-CASASNOVAS, G. & PUIG-JUNOY, J. 2000. Re iew o he li e a u e on
e e ence p icing. Heal h Policy, 54, 87-123.
LU, Z. J. & COMANOR, W. S. 1998. S a egic p icing o new pha maceu icals. Re iew o
Economics and S a is ics, 80, 108-118.
MASON, H., JONESLEE, M. & DONALDSON, C. 2009. Modelling he mone a y alue o
a QALY: a new app oach based on UK da a. Heal h Economics, 18, 933-950.
MESTRE-FERRANDIZ, J. 2003b. Re e ence p ices: The Spanish way. In es igaciones
Económicas, 27, 125-149.
238 Modelling Global P icing and Launching o New D ugs
MESTRE-FERRANDIZ, J. 2012. The R&D Cos o a New Medicine London: O ice o
Heal h Economics.
MESTRE-FERRANDIZ, J. 2013. Inc easing R&D cos s o new medicines. Chemis y &
Indus y, 77, 44-44.
MIRALDO, M. 2009. Re e ence p icing and i ms' p icing s a egies. J Heal h Econ, 28,
176-97.
MOHER, D., LIBERATI, A., TETZLAFF, J., ALTMAN, D. G. & GROUP, A. T. P. 2009.
Rep in —P e e ed Repo ing I ems o Sys ema ic Re iews and Me a-Analyses:
The PRISMA S a emen . Physical The apy, 89, 873-880.
MORENO, E., GIRON, F. J., VAZQUEZ-POLO, F. J. & NEGRIN, M. A. 2010. Op imal
heal hca e decisions: Compa ing medical ea men s on a cos -e ec i eness
basis. Eu opean Jou nal o Ope a ional Resea ch, 204, 180-187.
MORENO-TORRES, I., PUIG-JUNOY, J. & BORRELL, J. R. 2009. Gene ic En y in o he
Regula ed Spanish Pha maceu ical Ma ke . Re iew o Indus ial O ganiza ion, 34,
373-388.
MUTHOO, A. 1999. Ba gaining Theo y wi h Applica ions, Camb idge Uni e si y P ess.
NATIONAL INSITUTE FOR HEALTH AND CARE EXCELLENCE (NICE). 2010.
Measu ing e ec i eness and cos e ec i eness: he QALY [Online]. NICE.
A ailable:
h p://www.nice.o g.uk/news oom/ ea u es/measu inge ec i enessandcos e ec i
eness heqaly.jsp.
NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE (NICE). 2010.
Measu ing e ec i eness and cos e ec i eness: he QALY [Online]. NICE.
A ailable:
h p://www.nice.o g.uk/news oom/ ea u es/measu inge ec i enessandcos e ec i
eness heqaly.jsp.
NEGRIN, M. A. & VAZQUEZ-POLO, F. J. 2006. Bayesian cos -e ec i eness analysis wi h
wo measu es o e ec i eness: he cos -e ec i eness accep abili y plane. Heal h
Economics, 15, 363-372.
NEGRIN, M. A. & VAZQUEZ-POLO, F. J. 2008. Inco po a ing model unce ain y in cos -
e ec i eness analysis: A Bayesian model a e aging app oach. Jou nal o Heal h
Economics, 27, 1250-1259.
NEGRIN, M. A., VAZQUEZ-POLO, F. J., MARTEL, M., MORENO, E. & GIRON, F. J.
2010. Bayesian Va iable Selec ion in Cos -E ec i eness Analysis. In e na ional
Jou nal o En i onmen al Resea ch and Public Heal h, 7, 1577-1596.
NIHR, H. 2010. Weigh ing and aluing quali y-adjus ed li e-yea s using s a ed p e e ence
me hods: p elimina y esul s om he Social Value o a QALY P ojec . Heal h
Technology Assessmen , 14.
OECD 2011. Heal h a a Glance 2011: OECD Indica o s. Pa is.
PARIS, V., DOCTEUR, E. & MOÏSE, P. 2008. Pha maceu ical p icing polices in a global
ma ke . OECD Heal h Policy S udies. Pa is.
PARK, W. G. & WAGH, S. 2002. Index o pa en igh s. Economic eedom o he wo ld:
2002 annual epo , 33-43.
PINTO-PRADES, J. L., LOOMES, G. & BREY, R. 2009. T ying o es ima e a mone a y
alue o he QALY. Jou nal o Heal h Economics, 28, 553-562.
PUIG-JUNOY, J. 2007. The impac o gene ic e e ence p icing in e en ions in he s a in
ma ke . Heal h Policy, 84, 14-29.
PUIG-JUNOY, J. 2010a. Policies encou aging p ice compe i ion in he gene ic d ug
ma ke : Lessons om he Eu opean expe ience. Gace a Sani a ia, 24, 193-199.
Re e ences 239
PUIG-JUNOY, J. 2010b. Impac o Eu opean Pha maceu ical P ice Regula ion on Gene ic
P ice Compe i ion A Re iew. Pha macoeconomics, 28, 649-663.
PUIG-JUNOY, J. 2012. Lec u as sob e Economía del Sec o Bio a macéu ico, Mad id,
Sp inge Heal hca e.
PUIG-JUNOY, J. & GONZÁLEZ LÓPEZ-VALCÁRCEL, B. 2014. Launch p ices o new
pha maceu icals in he hea ily egula ed and subsidized Spanish ma ke . Heal h
Policy, Fo hcoming.
RAWLINS, M. D. 2012. C ossing he ou h hu dle. B J Clin Pha macol, 73, 855-60.
RICHTER, A. 2008. Assessing he impac o global p ice in e dependencies.
Pha macoeconomics, 26, 649-659.
ROBINSON, A., GYRD-HANSEN, D., BACON, P., BAKER, R., PENNINGTON, M. &
DONALDSON, C. 2013. Es ima ing a WTP-based alue o a QALY: The
‘chained’app oach. Social Science & Medicine, 92, 92-104.
RUBINSTEIN, A. 1982. Pe ec Equilib ium in a Ba gaining Model. Econome ica, 50, 97-
109.
SCHERER, F. M. 1993. PRICING, PROFITS, AND TECHNOLOGICAL-PROGRESS IN
THE PHARMACEUTICAL-INDUSTRY. Jou nal o Economic Pe spec i es, 7, 97-
115.
SCHERER, F. M. 2004. The pha maceu ical indus y-P ices and p og ess. New England
Jou nal o Medicine, 351, 927-932.
STARGARDT, T. & SCHREYÖGG, J. 2006. Impac o C oss-Re e ence P icing on
Pha maceu ical P ices: Manu ac u e s’ P icing S a egies and P ice Regula ion.
Applied Heal h Economics and Heal h Policy, 5, 235-247.
STATACORP 2009. S a a S a is ical So wa e. College S a ion, TX: S a aCo p LP.
SUSLOW, V. Y. 1996. Measu ing Quali y Change in he Ma ke o An i-Ulce D ug. In:
HELMS, R. M. & (ED) (eds.) In Compe i i e S a egies in he Pha maceu ical
Indus y. Washing on, DC: Ame ican En e p ise Ins i u e: AEI P ess.
TIMUR, A., PICONE, G. & DESIMONE, J. 2011. Has he Eu opean union achie ed a
single pha maceu ical ma ke ? In e na ional Jou nal o Heal h Ca e Finance and
Economics, 11, 223-244.
VELLA, F. 1993. A simple es ima o o simul aneous models wi h censo ed endogenous
eg esso s. In e na ional Economic Re iew, 441-457.
VERNIERS, I., STREMERSCH, S. & CROUX, C. 2011. The global en y o new
pha maceu icals: A join in es iga ion o launch window and p ice. In e na ional
Jou nal o Resea ch in Ma ke ing, 28, 295-308.
WAGNER, J. L. & MCCARTHY, E. 2004. In e na ional di e ences in d ug p ices. Annu.
Re . Public Heal h, 25, 475-495.
WHO 2003. Making Choices in Heal h: WHO Guide o Cos -E ec i eness Analysis
Gene a, Wo ld Heal h O ganiza ion.
WHO. 2012. ATC/DDD Index 2012 [Online]. A ailable:
h p://www.whocc.no/a c_ddd_index/.
WILKING, N., JÖNSSON, B. & I STOCKHOLM, H. 2005. A pan-Eu opean compa ison
ega ding pa ien access o cance d ugs. Ka olinska Ins i u e .
WOOLDRIDGE, J. M. 2002. Econome ic analysis o c oss sec ion and panel da a, The
MIT p ess.