Chain Formation and Consumer Welfare
on the Retail Pharmacy Market
Jakub Červený1, Richard Kališ2 and Biliana Yontcheva3
1 Institute for Health Care Analyses, Ministry of Health of the Slovak Republic
2 University of Economics in Bratislava
3 Düsseldorf Institute for Competition Economics and CEPR
April 2022, Brno
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 1 / 26
Regulation of retail pharmacies
The regulation of the European retail pharmacy market has gone through
substantial changes in the past two decades.
Entry restrictions based on demographic and distance criteria
▶ to prevent the excess entry inherent on markets with high regulated
prices and homogeneous products (Mankiw and Whinston, 1986)
▶ often too restrictive from a social welfare perspective (Schaumans
and Verboven, 2008)
Ownership of pharmacies in two dimensions:
1 the role of pharmacist
2 chain formation
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 2 / 26
Regulation of retail pharmacies
The regulation of the European retail pharmacy market has gone through
substantial changes in the past two decades.
Entry restrictions based on demographic and distance criteria
▶ to prevent the excess entry inherent on markets with high regulated
prices and homogeneous products (Mankiw and Whinston, 1986)
▶ often too restrictive from a social welfare perspective (Schaumans
and Verboven, 2008)
Ownership of pharmacies in two dimensions:
1 the role of pharmacist
2 chain formation
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 2 / 26
Multiple ownership regulation on the pharmacy market
Table: Pharmacy regulation in a subset of OECD countries
Entry restrictions (1) Ownership (2) Pricing (3)
Country General Population Distance Pharmacist Chains Margins Regressive
Austria Y Y (5500) Y (500m) Y N Y Y
Czech Republic N N N N Y Y Y
Canada N N N N Y O O
Denmark O N N O O Y Y
Finland Y O Y Y N Y Y
France Y Y (2500) N Y Y Y Y
Germany N N N Y O Y Y
Italy Y Y (3300) Y (200m) N N Y N
Netherlands O N N N Y Y N
Norway N N N N Y Y N
Slovak Republic N N N Y Y Y Y
Spain Y Y (2800) Y (250m) Y N Y Y
Sweden N N N N Y Y Y
United Kingdom N N N N Y O N
USA N N N N Y N N
Notes: Y - yes/permitted, O - mentioned, but open to interpretation/partially allowed, N - no/not permitted. Source: compiled
by authors based on country-specific regulation.Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 3 / 26
Rationale for multiple ownership bans
Statement of Federal Union of German Associations of Pharmacists, 2021
“The ban on third-party and multiple ownership stresses the personal
responsibility and liability of self-employed pharmacists in the healthcare
sector. It separates the pharmaceutical supply from companies’
exclusive intention to maximise return.”
“There is a danger for all patients that chain pharmacies may not
provide independent advice as their owners (e.g. manufacturers) are
driven by commercial interests and only aim at selling certain products. In
particular, over-the-counter (OTC) drugs to creating additional
demand which may lead to abuse and danger to one’s health and life.”
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 4 / 26
Rationale for multiple ownership bans
Statement of Federal Union of German Associations of Pharmacists, 2021
“The ban on third-party and multiple ownership stresses the personal
responsibility and liability of self-employed pharmacists in the healthcare
sector. It separates the pharmaceutical supply from companies’
exclusive intention to maximise return.”
“There is a danger for all patients that chain pharmacies may not
provide independent advice as their owners (e.g. manufacturers) are
driven by commercial interests and only aim at selling certain products. In
particular, over-the-counter (OTC) drugs to creating additional
demand which may lead to abuse and danger to one’s health and life.”
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 4 / 26
Rationale for multiple ownership bans
Empirical evidence on misconduct
Janssen and Zhang (2020): change in dispensation policies during
the opioid epidemic in the US
▶ independent pharmacies on dispense 40.9% more opioids and
61.7% more OxyContin, with a substantial portion of this use being
categorized as recreational
▶ independent pharmacies may have a lower cost of misconduct
due to lower levels of oversight,
▶ a pharmacy owner is entitled to a higher share of the profits and
thus may prioritize profitability,
▶ integrated information systems may raise quality.
Kuang et al. (2020): non-prescription sale of antibiotics and the
service quality of community pharmacies in China:
▶ 49% lower probability of non-prescription sale of antibiotics in
chain pharmacies
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 5 / 26
Rationale for multiple ownership bans
Empirical evidence on misconduct
Janssen and Zhang (2020): change in dispensation policies during
the opioid epidemic in the US
▶ independent pharmacies on dispense 40.9% more opioids and
61.7% more OxyContin, with a substantial portion of this use being
categorized as recreational
▶ independent pharmacies may have a lower cost of misconduct
due to lower levels of oversight,
▶ a pharmacy owner is entitled to a higher share of the profits and
thus may prioritize profitability,
▶ integrated information systems may raise quality.
Kuang et al. (2020): non-prescription sale of antibiotics and the
service quality of community pharmacies in China:
▶ 49% lower probability of non-prescription sale of antibiotics in
chain pharmacies
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 5 / 26
Rationale for multiple ownership bans
Empirical evidence on misconduct
Janssen and Zhang (2020): change in dispensation policies during
the opioid epidemic in the US
▶ independent pharmacies on dispense 40.9% more opioids and
61.7% more OxyContin, with a substantial portion of this use being
categorized as recreational
▶ independent pharmacies may have a lower cost of misconduct
due to lower levels of oversight,
▶ a pharmacy owner is entitled to a higher share of the profits and
thus may prioritize profitability,
▶ integrated information systems may raise quality.
Kuang et al. (2020): non-prescription sale of antibiotics and the
service quality of community pharmacies in China:
▶ 49% lower probability of non-prescription sale of antibiotics in
chain pharmacies
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 5 / 26
Rationale for multiple ownership bans
Empirical evidence on misconduct
Janssen and Zhang (2020): change in dispensation policies during
the opioid epidemic in the US
▶ independent pharmacies on dispense 40.9% more opioids and
61.7% more OxyContin, with a substantial portion of this use being
categorized as recreational
▶ independent pharmacies may have a lower cost of misconduct
due to lower levels of oversight,
▶ a pharmacy owner is entitled to a higher share of the profits and
thus may prioritize profitability,
▶ integrated information systems may raise quality.
Kuang et al. (2020): non-prescription sale of antibiotics and the
service quality of community pharmacies in China:
▶ 49% lower probability of non-prescription sale of antibiotics in
chain pharmacies
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 5 / 26
Rationale for multiple ownership bans
Empirical evidence on misconduct
Janssen and Zhang (2020): change in dispensation policies during
the opioid epidemic in the US
▶ independent pharmacies on dispense 40.9% more opioids and
61.7% more OxyContin, with a substantial portion of this use being
categorized as recreational
▶ independent pharmacies may have a lower cost of misconduct
due to lower levels of oversight,
▶ a pharmacy owner is entitled to a higher share of the profits and
thus may prioritize profitability,
▶ integrated information systems may raise quality.
Kuang et al. (2020): non-prescription sale of antibiotics and the
service quality of community pharmacies in China:
▶ 49% lower probability of non-prescription sale of antibiotics in
chain pharmacies
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 5 / 26
Rationale for multiple ownership bans
Rising market concentration and quality concerns
Vogler et al. (2006): after deregulation in Norway, 80% of pharmacies
are part of a chain owned by a wholesaler ⇒ market power and
foreclosure concerns
Vogler et al. (2012): excess new entry may lead to fewer
pharmacists per outlet and thus lower service quality.
Rostam-Afschar and Unsorg (2021) look at partial deregulation in
Germany and find that employment levels rose in chain
pharmacies.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 6 / 26
Rationale for multiple ownership bans
Rising market concentration and quality concerns
Vogler et al. (2006): after deregulation in Norway, 80% of pharmacies
are part of a chain owned by a wholesaler ⇒ market power and
foreclosure concerns
Vogler et al. (2012): excess new entry may lead to fewer
pharmacists per outlet and thus lower service quality.
Rostam-Afschar and Unsorg (2021) look at partial deregulation in
Germany and find that employment levels rose in chain
pharmacies.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 6 / 26
Rationale for multiple ownership bans
Rising market concentration and quality concerns
Vogler et al. (2006): after deregulation in Norway, 80% of pharmacies
are part of a chain owned by a wholesaler ⇒ market power and
foreclosure concerns
Vogler et al. (2012): excess new entry may lead to fewer
pharmacists per outlet and thus lower service quality.
Rostam-Afschar and Unsorg (2021) look at partial deregulation in
Germany and find that employment levels rose in chain
pharmacies.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 6 / 26
Research agenda and results overview
1 Do consumers perceive chain pharmacy services as inferior to
independent firms?
▶ there is no systematic evidence for consumers preferring
independent sellers
2 Is there an efficiency differential between chain affiliates and
independent firms?
▶ chain pharmacies require significantly lower number of employees
than independent counterparts
3 Do chains break-even more easily than independent firms?
▶ no evidence for lower fixed costs of chains
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 7 / 26
Research agenda and results overview
1 Do consumers perceive chain pharmacy services as inferior to
independent firms?
▶ there is no systematic evidence for consumers preferring
independent sellers
2 Is there an efficiency differential between chain affiliates and
independent firms?
▶ chain pharmacies require significantly lower number of employees
than independent counterparts
3 Do chains break-even more easily than independent firms?
▶ no evidence for lower fixed costs of chains
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 7 / 26
Research agenda and results overview
1 Do consumers perceive chain pharmacy services as inferior to
independent firms?
▶ there is no systematic evidence for consumers preferring
independent sellers
2 Is there an efficiency differential between chain affiliates and
independent firms?
▶ chain pharmacies require significantly lower number of employees
than independent counterparts
3 Do chains break-even more easily than independent firms?
▶ no evidence for lower fixed costs of chains
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 7 / 26
Slovak deregulation process
From 1998 to 2004 entry levels were constrained and self-regulated
by the Slovak Chamber of Pharmacists.
In 2004, Slovakia shifted to a liberalized entry regime: no geographic
restrictions and no ban on multiple ownership.
Chain formation in two formats:
▶ standard chains = pharmacies have a common owner
▶ virtual chains = pharmacies join a network and sign investment
contracts with distributors
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 8 / 26
Slovak deregulation process
From 1998 to 2004 entry levels were constrained and self-regulated
by the Slovak Chamber of Pharmacists.
In 2004, Slovakia shifted to a liberalized entry regime: no geographic
restrictions and no ban on multiple ownership.
Chain formation in two formats:
▶ standard chains = pharmacies have a common owner
▶ virtual chains = pharmacies join a network and sign investment
contracts with distributors
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 8 / 26
Slovak deregulation process
From 1998 to 2004 entry levels were constrained and self-regulated
by the Slovak Chamber of Pharmacists.
In 2004, Slovakia shifted to a liberalized entry regime: no geographic
restrictions and no ban on multiple ownership.
Chain formation in two formats:
▶ standard chains = pharmacies have a common owner
▶ virtual chains = pharmacies join a network and sign investment
contracts with distributors
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 8 / 26
Chain market shares
Freq. Market share Virtual chain Vertical int. Owner/Manager of network
Plus 444 21.01 Y Y Distributor: Unipharma
Dr.Max 276 13.06 N O Insurer: Mirakl/PENTA
Partner 235 11.12 Y Y Distributor: Phoenix
VASA lekaren 216 10.22 Y Y Distributor: Med-Art
Benu 61 2.89 N Y Distributor: Phoenix
Druzstvo lekarni 49 2.32 Y N Horizontal
Farmakol 48 2.27 N Y Distributor: Farmakol
Schneider 44 2.08 N N Horizontal
Moja lekaren 28 1.33 Y Y Distributor: Pharmos
Apotheke 5 0.24 N Y Distributor: Unipharma
Independent 707 33.46 - - Notes:
Y - yes; N - no; O - integration with other health care providers.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 9 / 26
Chain market shares
Freq. Market share Virtual chain Vertical int. Owner/Manager of network
Plus 444 21.01 Y Y Distributor: Unipharma
Dr.Max 276 13.06 N O Insurer: Mirakl/PENTA
Partner 235 11.12 Y Y Distributor: Phoenix
VASA lekaren 216 10.22 Y Y Distributor: Med-Art
Benu 61 2.89 N Y Distributor: Phoenix
Druzstvo lekarni 49 2.32 Y N Horizontal
Farmakol 48 2.27 N Y Distributor: Farmakol
Schneider 44 2.08 N N Horizontal
Moja lekaren 28 1.33 Y Y Distributor: Pharmos
Apotheke 5 0.24 N Y Distributor: Unipharma
Independent 707 33.46 - - Notes:
Y - yes; N - no; O - integration with other health care providers.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 9 / 26
Chain market shares
Freq. Market share Virtual chain Vertical int. Owner/Manager of network
Plus 444 21.01 Y Y Distributor: Unipharma
Dr.Max 276 13.06 N O Insurer: Mirakl/PENTA
Partner 235 11.12 Y Y Distributor: Phoenix
VASA lekaren 216 10.22 Y Y Distributor: Med-Art
Benu 61 2.89 N Y Distributor: Phoenix
Druzstvo lekarni 49 2.32 Y N Horizontal
Farmakol 48 2.27 N Y Distributor: Farmakol
Schneider 44 2.08 N N Horizontal
Moja lekaren 28 1.33 Y Y Distributor: Pharmos
Apotheke 5 0.24 N Y Distributor: Unipharma
Independent 707 33.46 - - Notes:
Y - yes; N - no; O - integration with other health care providers.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 9 / 26
Data
1 Prescription-level information:
▶ 58,005,693 fulfilled prescriptions of medication in 2017
▶ address of the prescribing physician
▶ address of the pharmacy fulfilling the prescription
▶ quantity of the product purchased (in packages)
▶ price paid by the insurer and by the patient
▶ diagnosis of the patient
⇒ total prescription revenue and output per pharmacy
⇒ location-specific catchment area and potential market size by diagnosis
2 Pharmacy-level information (N = 1956):
▶ chain affiliation
▶ location and opening hours data
▶ employment: number of pharmacists and pharmacy technicians
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 10 / 26
Data
3 Industry-level information:
▶ share of OTC drugs and medical devices in total drug sales (23%)
▶ wage costs per pharmacist (e1,744) and per technician (e1,276)
▶ regulated pharmacy retail margins for each medication
Mean SD P10 P90
Revenue from prescriptions (e1,000) 509.17 688.40 1,059.13 66,662.73
Sales of prescription medication (1,000 units) 40.87 36.20 10.33 81.99
pharmacists 2.16 1.40 1 4
technicians/assistants 1.03 1.20 0 2
work days open per year 243.26 25.97 233 256
days open per year 274.52 46.77 239 350
Nonstop (open > 50 weekends) .09 .27 0 0
Distance to closest hospital (km) 6.42 7.24 0.29 17.42
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 11 / 26
Diagnosis-level demand specification
For a given diagnosis, the utility of individual i of type c from
purchasing at pharmacy j in transaction t is:
˜uijtc = uijtc + εijtc
= γcdijt + βcxjt + ξbc + εijtc.
dijt - distance from prescribing physician to pharmacy, xjt - location
characteristics and opening hours, ξbc - chain fixed effects, εijtc - extreme
value preference shock.
Demand characteristics:
if consumer type were known ⇒ conditional logit model
no outside option ⇒ no market expansion
no variation in margins ⇒ utility is scale-free (or measured in km)
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 12 / 26
Diagnosis-level demand specification
For a given diagnosis, the utility of individual i of type c from
purchasing at pharmacy j in transaction t is:
˜uijtc = uijtc + εijtc
= γcdijt + βcxjt + ξbc + εijtc.
dijt - distance from prescribing physician to pharmacy, xjt - location
characteristics and opening hours, ξbc - chain fixed effects, εijtc - extreme
value preference shock.
Demand characteristics:
if consumer type were known ⇒ conditional logit model
no outside option ⇒ no market expansion
no variation in margins ⇒ utility is scale-free (or measured in km)
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 12 / 26
Consumer types and purchase probability
Preferences for pharmacy characteristics may vary across individuals
with different lifestyles:
parameter distribution is not be unimodal,
distribution of the parameters across latent consumer types
c = 1, · · · , C (Heckman and Singer, 1984).
Likelihood function:
L =
Nd
i
ln
C
c=1
πc
Ti
t=1
Jt
j=1
exp(xijtβc + dijtγc)
J
k=1 exp(xiktβc + diktγc)
{ιit =j}
(1)
πc - probability that consumer i fall into the latent consumer group c,
{ιit = j} - indicator variable = 1 if j is chosen.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 13 / 26
Consumer types and purchase probability
Preferences for pharmacy characteristics may vary across individuals
with different lifestyles:
parameter distribution is not be unimodal,
distribution of the parameters across latent consumer types
c = 1, · · · , C (Heckman and Singer, 1984).
Likelihood function:
L =
Nd
i
ln
C
c=1
πc
Ti
t=1
Jt
j=1
exp(xijtβc + dijtγc)
J
k=1 exp(xiktβc + diktγc)
{ιit =j}
(1)
πc - probability that consumer i fall into the latent consumer group c,
{ιit = j} - indicator variable = 1 if j is chosen.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 13 / 26
Consumer types and distance sensitivity
Algorithm converges to 2 groups for all diagnoses.
Systematic differences in distance sensitivity:
▶ Group 1: distance sensitive
▶ Group 2: low distance sensitivity, higher response to branding.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 14 / 26
Chain utility as distance reduction (−ξbc/γc)
Large chains
Group 1 Group 2 Group 1 Group 2
Dr. Max (standard chain, n = 276) Plus (virtual chain, n = 444)
Metabolic 0.028 (0.001) 1.400 (0.037) 0.004 (0.001) 0.568 (0.034)
Neurological 0.058 (0.001) 0.370 (0.035) 0.019 (0.001) 0.301 (0.032)
Cardiovascular 0.042 (0.001) 1.140 (0.016) 0.003 (0.001) 0.474 (0.015)
Respiratory 0.028 (0.001) 0.568 (0.029) 0.003 (0.001) 0.455 (0.027)
Gastrointestinal 0.043 (0.002) 0.977 (0.056) 0.008 (0.002) 0.310 (0.052)
Musculoskeletal 0.034 (0.001) 1.021 (0.034) −0.005 (0.001) 0.459 (0.032)
Genitourinary 0.026 (0.002) 0.886 (0.083) 0.008 (0.002) 0.954 (0.078)
Other 0.034 (0.001) 0.440 (0.028) 0.006 (0.001) 0.483 (0.025)
Partner (virtual chain, n = 235) Vasa lekaren (virtual chain, n = 216)
Metabolic 0.003 (0.001) 0.521 (0.040) −0.002 (0.001) 0.404 (0.042)
Neurological 0.018 (0.001) −0.040 (0.034) 0.033 (0.001) −0.169 (0.039)
Cardiovascular −0.008 (0.001) 0.387 (0.017) −0.012 (0.001) 0.164 (0.018)
Respiratory 0.008 (0.001) 0.154 (0.032) −0.004 (0.001) −0.017 (0.034)
Gastrointestinal 0.008 (0.003) 0.386 (0.060) 0.002 (0.002) 0.046 (0.064)
Musculoskeletal −0.010 (0.002) 0.390 (0.037) −0.019 (0.001) 0.262 (0.039)
Genitourinary −0.005 (0.003) 0.912 (0.088) 0.002 (0.003) 0.377 (0.093)
Other 0.000 (0.001) 0.203 (0.029) −0.014 (0.001) 0.122 (0.031)
Notes: Standard errors are in parentheses. The variable n measures the number of pharmacies in the chain.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 15 / 26
Chain utility as distance reduction (−ξbc/γc)
Large chains
Group 1 Group 2 Group 1 Group 2
Dr. Max (standard chain, n = 276) Plus (virtual chain, n = 444)
Metabolic 0.028 (0.001) 1.400 (0.037) 0.004 (0.001) 0.568 (0.034)
Neurological 0.058 (0.001) 0.370 (0.035) 0.019 (0.001) 0.301 (0.032)
Cardiovascular 0.042 (0.001) 1.140 (0.016) 0.003 (0.001) 0.474 (0.015)
Respiratory 0.028 (0.001) 0.568 (0.029) 0.003 (0.001) 0.455 (0.027)
Gastrointestinal 0.043 (0.002) 0.977 (0.056) 0.008 (0.002) 0.310 (0.052)
Musculoskeletal 0.034 (0.001) 1.021 (0.034) −0.005 (0.001) 0.459 (0.032)
Genitourinary 0.026 (0.002) 0.886 (0.083) 0.008 (0.002) 0.954 (0.078)
Other 0.034 (0.001) 0.440 (0.028) 0.006 (0.001) 0.483 (0.025)
Partner (virtual chain, n = 235) Vasa lekaren (virtual chain, n = 216)
Metabolic 0.003 (0.001) 0.521 (0.040) −0.002 (0.001) 0.404 (0.042)
Neurological 0.018 (0.001) −0.040 (0.034) 0.033 (0.001) −0.169 (0.039)
Cardiovascular −0.008 (0.001) 0.387 (0.017) −0.012 (0.001) 0.164 (0.018)
Respiratory 0.008 (0.001) 0.154 (0.032) −0.004 (0.001) −0.017 (0.034)
Gastrointestinal 0.008 (0.003) 0.386 (0.060) 0.002 (0.002) 0.046 (0.064)
Musculoskeletal −0.010 (0.002) 0.390 (0.037) −0.019 (0.001) 0.262 (0.039)
Genitourinary −0.005 (0.003) 0.912 (0.088) 0.002 (0.003) 0.377 (0.093)
Other 0.000 (0.001) 0.203 (0.029) −0.014 (0.001) 0.122 (0.031)
Notes: Standard errors are in parentheses. The variable n measures the number of pharmacies in the chain.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 15 / 26
Chain utility as distance reduction (−ξbc/γc)
Small chains
Group 1 Group 2 Group 1 Group 2
Benu (standard chain, n = 61) Small chain (n < 50)
Metabolic −0.052 (0.003) −1.198 (0.077) −0.004 (0.001) −0.366 (0.047)
Neurological 0.011 (0.003) −1.784 (0.073) 0.018 (0.001) −0.572 (0.043)
Cardiovascular −0.026 (0.002) −1.203 (0.035) −0.004 (0.001) −0.565 (0.022)
Respiratory −0.028 (0.003) −0.857 (0.049) −0.004 (0.001) −0.628 (0.037)
Gastrointestinal −0.012 (0.007) −1.126 (0.108) −0.009 (0.003) −0.760 (0.074)
Musculoskeletal −0.007 (0.004) −1.115 (0.069) −0.007 (0.002) −0.434 (0.043)
Genitourinary −0.010 (0.007) −0.683 (0.139) −0.013 (0.003) −0.459 (0.104)
Other 0.017 (0.003) −1.171 (0.051) −0.017 (0.001) −0.692 (0.034)
Notes: Standard errors are in parentheses. The variable n measures the number of pharmacies in the chain.
Smaller chains offer lower perceived quality:
possible specialization in OTC products?
endogenous outcome regarding chain size? (chains with low
perceived quality may have a higher rate of bankruptcy and thus fail
to grow)
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 16 / 26
Chain utility as distance reduction (−ξbc/γc)
Small chains
Group 1 Group 2 Group 1 Group 2
Benu (standard chain, n = 61) Small chain (n < 50)
Metabolic −0.052 (0.003) −1.198 (0.077) −0.004 (0.001) −0.366 (0.047)
Neurological 0.011 (0.003) −1.784 (0.073) 0.018 (0.001) −0.572 (0.043)
Cardiovascular −0.026 (0.002) −1.203 (0.035) −0.004 (0.001) −0.565 (0.022)
Respiratory −0.028 (0.003) −0.857 (0.049) −0.004 (0.001) −0.628 (0.037)
Gastrointestinal −0.012 (0.007) −1.126 (0.108) −0.009 (0.003) −0.760 (0.074)
Musculoskeletal −0.007 (0.004) −1.115 (0.069) −0.007 (0.002) −0.434 (0.043)
Genitourinary −0.010 (0.007) −0.683 (0.139) −0.013 (0.003) −0.459 (0.104)
Other 0.017 (0.003) −1.171 (0.051) −0.017 (0.001) −0.692 (0.034)
Notes: Standard errors are in parentheses. The variable n measures the number of pharmacies in the chain.
Smaller chains offer lower perceived quality:
possible specialization in OTC products?
endogenous outcome regarding chain size? (chains with low
perceived quality may have a higher rate of bankruptcy and thus fail
to grow)
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 16 / 26
Consumer surplus gains from chain formation
Consumer surplus estimate:
CS =
8
d=1
C
c=1
πc
Nd
i=1
Tid
t=1
−
cdist
γc
ln


j∈Jt
exp(uidtj )

 qit
Value of travel time in Slovakia is approximately e7.7 per hour
(Wardman et al., 2016)
▶ If walking speed is 15 minutes per km ⇒ cost per kilometer to and
back from a pharmacy is e3.85.
e29.9 million consumer surplus gains from chain formation per
year (≈ approximately 3% of industry revenue).
▶ we assume that all pharmacies would continue to be active on the
market as independent firms
▶ if chain affiliation results in lower costs, removing the networks may
lead to changes in consumer surplus via decreased availability
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 17 / 26
Provision of pharmaceutical services
Employee levels and output
Firms hire quality-adjusted pharmacists to satisfy demand:
lj = 40(npj +
wt
wp
ntj )
npj - number of fully qualified pharmacists, ntj - number of technicians
Notes: the median number
of transactions fulfilled
by pharmacies adjusts with
employment levels: similar
output goals with varying
ratios of each employee
type
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 18 / 26
Production function specification
We assume the production function with constant returns to scale:
(1 + sOTC
j )qM
j = exp(α0 + αb + ωj )lj
qM
j - total sales of medication, sOTC
j - OTC sales as a fraction of
medication sales qOTC
j /qM, α0 - productivity of independent pharmacies,
αb - chain fixed effect, ωj - location-specific shock, lj number of
quality-adjusted pharmacists.
Labor demand:
ln lj = −α0 − αb + ln(1 + sOTC
j ) + ln qM
⇒ higher OTC sales result in lower estimated productivity estimates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 19 / 26
Production function specification
We assume the production function with constant returns to scale:
(1 + sOTC
j )qM
j = exp(α0 + αb + ωj )lj
qM
j - total sales of medication, sOTC
j - OTC sales as a fraction of
medication sales qOTC
j /qM, α0 - productivity of independent pharmacies,
αb - chain fixed effect, ωj - location-specific shock, lj number of
quality-adjusted pharmacists.
Labor demand:
ln lj = −α0 − αb + ln(1 + sOTC
j ) + ln qM
⇒ higher OTC sales result in lower estimated productivity estimates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 19 / 26
Marginal productivity per pharmacist
MPb MP0 = MPb
(units) (p-value)
Independent 13,124 (443) Dr
Max 16,068 (871) 0.003
Small chain 15,769 (873) 0.007
Plus 15,616 (505) 0.000
Partner 14,703 (629) 0.041
Vasa lekaren 14,029 (685) 0.274
Benu 8,092 (1,571) 0.002
N 1,635
Note: Standard errors in parentheses.
⇒ marginal productivity per pharmacist is on average higher for chains.
⇒ effect is underestimated if OTC sales are higher for chains.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 20 / 26
Fixed costs
In the final step of our analysis we aim to quantify the potential fixed
cost savings available to pharmacies who join a chain
As in Eizenberg (2014), we infer these gains following a revealed
preference approach to estimate the bounds of fixed costs across
different pharmacy types.
Based on the demand and costs model we can calculate expected variable
profits of each pharmacy outlet:
πV
j (A) = ˆrnet
j − wp
ˆlj
where A is a set of potential entry locations wich can be divided into two
subsets:
1 A1 contains all locations in which entry occured
2 A0 is the set of locations without entry
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 21 / 26
Fixed costs
In the final step of our analysis we aim to quantify the potential fixed
cost savings available to pharmacies who join a chain
As in Eizenberg (2014), we infer these gains following a revealed
preference approach to estimate the bounds of fixed costs across
different pharmacy types.
Based on the demand and costs model we can calculate expected variable
profits of each pharmacy outlet:
πV
j (A) = ˆrnet
j − wp
ˆlj
where A is a set of potential entry locations wich can be divided into two
subsets:
1 A1 contains all locations in which entry occured
2 A0 is the set of locations without entry
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 21 / 26
Fixed costs
In the final step of our analysis we aim to quantify the potential fixed
cost savings available to pharmacies who join a chain
As in Eizenberg (2014), we infer these gains following a revealed
preference approach to estimate the bounds of fixed costs across
different pharmacy types.
Based on the demand and costs model we can calculate expected variable
profits of each pharmacy outlet:
πV
j (A) = ˆrnet
j − wp
ˆlj
where A is a set of potential entry locations wich can be divided into two
subsets:
1 A1 contains all locations in which entry occured
2 A0 is the set of locations without entry
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 21 / 26
Fixed costs
For each brand b operationg pharmacies in location set A1
b ∈ A1 the
expected variable profits are given by sum across branch locations
Πb
V (A) ≡
j∈A1
b
πV
j (A).
Firms play a two stage game:
1 In the first stage, each firm forms expectations regarding variable
profits from entering a location j
2 In the second stage, demand shocks are realized and firms earn profits
which depend on the equilibrium distribution of entrants.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 22 / 26
Fixed costs
For each brand b operationg pharmacies in location set A1
b ∈ A1 the
expected variable profits are given by sum across branch locations
Πb
V (A) ≡
j∈A1
b
πV
j (A).
Firms play a two stage game:
1 In the first stage, each firm forms expectations regarding variable
profits from entering a location j
2 In the second stage, demand shocks are realized and firms earn profits
which depend on the equilibrium distribution of entrants.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 22 / 26
Fixed costs
The equilibrium concept implies that for each location in the entry set A1
b,
the expected additional variable profits from entry for chain b must be
sufficient to offset the fixed costs of operating:
∆Πb
V (Ab, Ab − 1j ) ≡ Πb
V (Ab) − Πb
V (Ab − 1j ) ≥ f b
j for ∀j ∈ A1
b
while for the entry locations A0 this is not true:
∆Πb
V (Ab + 1j , Ab) ≡ Πb
V (Ab + 1j ) − Πb
V (Ab) < f b
j for ∀j ∈ A0
b
The above conditions imply bounds for the fixed costs of each brand.
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 23 / 26
Estimated bounds by chain (in e1,000)
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 24 / 26
Conclusion and discussion
Estimated fixed costs margins are not indicative of extra-normal
profits from OTC drugs
Productivity levels indicate that chains result in cost savings for the
industry.
Pharmacy chains do not appear to perform worse than
independents in terms of perceived quality.
Discussion:
foreclosure at the medication level
dynamic effects due to exit of smaller chains
selection issue for fixed cost estimation
financial statement data on overall revenues
regional variation in catchment areas
modelling price rabates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 25 / 26
Conclusion and discussion
Estimated fixed costs margins are not indicative of extra-normal
profits from OTC drugs
Productivity levels indicate that chains result in cost savings for the
industry.
Pharmacy chains do not appear to perform worse than
independents in terms of perceived quality.
Discussion:
foreclosure at the medication level
dynamic effects due to exit of smaller chains
selection issue for fixed cost estimation
financial statement data on overall revenues
regional variation in catchment areas
modelling price rabates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 25 / 26
Conclusion and discussion
Estimated fixed costs margins are not indicative of extra-normal
profits from OTC drugs
Productivity levels indicate that chains result in cost savings for the
industry.
Pharmacy chains do not appear to perform worse than
independents in terms of perceived quality.
Discussion:
foreclosure at the medication level
dynamic effects due to exit of smaller chains
selection issue for fixed cost estimation
financial statement data on overall revenues
regional variation in catchment areas
modelling price rabates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 25 / 26
Conclusion and discussion
Estimated fixed costs margins are not indicative of extra-normal
profits from OTC drugs
Productivity levels indicate that chains result in cost savings for the
industry.
Pharmacy chains do not appear to perform worse than
independents in terms of perceived quality.
Discussion:
foreclosure at the medication level
dynamic effects due to exit of smaller chains
selection issue for fixed cost estimation
financial statement data on overall revenues
regional variation in catchment areas
modelling price rabates
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 25 / 26
Production function estimates
Variable OLS GMM OLS GMM MPb MP0 = MPb
(1) (2) (3) (4) (units) (p-value)
Dr Max -0.196 -0.202 16,068 0.003
(0.065) (0.065) (871)
Plus -0.244 -0.174 15,616 0.000
(0.049) (0.047) (505)
Benu 0.434 0.484 8,092 0.002
(0.164) (0.197) (1,571)
Vasa lekaren -0.092 -0.067 14,029 0.274
(0.063) (0.060) (685)
Partner -0.183 -0.114 14,703 0.041
(0.060) (0.055) (629)
Small chain -0.214 -0.184 15,769 0.007
(0.069) (0.066) (873)
Constant -5.822 -5.891 -5.697 -5.793 13,124 (Independent)
(0.018) (0.017) (0.032) (0.034) (443)
N 1,635 1,635 1,635 1,635 1,635 1635
Notes: Standard errors in parentheses. MPb measures the chain-specific marginal productivity per
“quality-adjusted” pharmacist. MP0 denotes productivity estimates at unbranded pharmacies. Productivity
estimates are based on the GMM model reported in Column (4).
⇒ GMM based on demand instruments from the spatial demand model
back
Červený, Kališ & Yontcheva Chain formation and welfare 13.04.2022 26 / 26