Provisional Version
Profitability of the Universal Postal Service Provider
in a Free Market with Economies of Scale in Collect and Delivery
Prof. Dr. Gonzales d’Alcantara, University of Antwerp and Economic Expert
We propose a model for a liberalized
competitive postal market with a Universal Service Obligation Provider, the
Incumbent, and one Entrant. In this model we take into account the existence of
economies of scale in their collect and delivery activities. The study is
intended to lead to conclusions about the impact of economies of scale on the postal
market in various countries, as measured by the numbers of postal items
delivered per person and per year, and by the size of the population. We consider
delivery technologies of operators on markets of different sizes, measured both
by these two criterions. Variable and fixed costs in the collect and delivery
activities are determined by a structural cost model, which has the same form
for the Incumbent and for the Entrant. In this latter model we will also pay attention
to delivery technologies parameters which differ from country to country, such
as the number of delivery points per route stop. Given a definition of the USO,
assumptions about demand behaviour and the opening of the market to the Entrant,
the calibrated model will compute the impact of the cream skimming mechanism and
the graveyard spiral on volume, tariffs, market shares and the cumulative
balance of the Entrant needed to finance his delivery network. In a second step,
the model will also measure the effects of economies of scale on the financial requirement
of the Incumbent in charge of the USO, in case she keeps her tariffs at a
constant pre-competition level.
The existence of economies of scale has
been recognized in the literature… Key references are K. Jasinski and E.
Steggles, (1977), “Modelling letter delivery in town areas”, and, in various Michael A. Crew and Paul R.
Kleindorfer edited volumes, John
Panzar (1991), “Is Postal service a natural monopoly?”, Cathy Rogerson and William
Takis (1993), “Economies of scale and scope and competition in the Postal
services”, Michael Bradley and Jeff Colvin (1995), “An econometric model of
postal delivery”, Catherine Cazals, M. de Rycke, Jean-Pierre Florens, and S.
Rouzaud (1997), “Scale economies and natural monopoly in the postal delivery:
comparison between parametric and non parametric specifications”, Robert H. Cohen, and Edward H. Chu (1997),
“A measure of scale economies for postal systems”, Bernard Roy (1999), “Technico-economic
analysis of the costs outside work in postal delivery”.
All these references show the existence and importance of economies of
scale, which clearly exist in the delivery function and to a lesser extent in
other activities in the postal value chain. Our purpose is to examine the impact of these economies
of scale on the postal market in various countries. When the market opens and there are economies
of scale which differ country by country, what are the long term impacts of a well
defined regulatory set-up on tariffs, volumes and market shares in each case?
How do economies of scale affect the levels of the tariff differentials between
both operators? What
is the end-result in each country when scale obviously decreases for the
Incumbent with the entry of the competitor, who benefits both from increasing
scale and from relatively denser delivery zones, lost by the Incumbent? Which
parameters are crucial for the financial requirements of the Entrant, who wants
to invest in a delivery network, and for the Incumbent in charge of the USO,
when she keeps her tariffs at unchanged – affordable – levels?
The paper will treat two types of
model solutions:
a) The
regulatory authorities want to achieve a liberalized postal market without giving
any subsidies to finance the Universal Service Obligation of the Incumbent. The
Incumbent and the Entrant should behave as they would in a situation of
competition: both operators determine their tariffs according to their unit
costs including the margins to cover their fixed costs. The Incumbent therefore
modifies her tariffs so as to keep balanced or break-even accounts.
b) If the
Incumbent’s tariffs are required to remain equal to the costs per unit in the
base period before the market is opened, the starting situation under monopoly,
how large is then the financial loss of the Incumbent, provider of the Universal Postal
Service Obligation?
In this
contribution we will not take into account the possible inefficiencies of the
Incumbent, the obstacles to the access to the market by the Entrant or the
transition problems to go from an inefficient and monopolistic to a more
competitive market situation.
1.
The model
The postal market is modelled with two types of customers (j), retail customers and business customers. The retail customers are afforded special attention by the regulator because they have a high weight in social welfare. They are also characterized by more loyalty to the Incumbent and a less elastic demand than the business customers, who are more cost conscious. Both customers are assumed to consume one single product, for simplicity the same mix of first and second-class mail services.
There is an Incumbent operator (I) providing services for upstream and downstream. This operator has a Universal Service Obligation. This obligation includes delivering the whole population of the area to be serviced at least one time each day of the week (6 days). The tariff should be an affordable price; therefore tariff will be cost-based. No price-cap has been set-up. The tariff should be the same for all zones of delivery (tariff uniformity). The base period corresponds to a period with a reserved activity for the total volume.
In the first period, when the market is
opened, there is an alternative operator - the Entrant (E) - providing
competitive services for upstream activity and also able to provide downstream
activities. The downstream activity used to be the Incumbent’s reserved
activity in the base period, but it is opened to competition in the current
period. In other words, in the current period bypass is both technologically
and economically feasible for the Entrant.
Total market demand is related to broad
substitution possibilities for the customer among different forms of
communication. Since in the base period the Entrant is already providing
upstream services, new opportunities to create additional mail volume when the
conditions and regulations of the downstream activity are changed are not
likely. In the model, what is usually called the displacement ratio, s, will be
assumed to be equal to 1; i.e. for a given total market demand, an increase of
demand of one unit of the Entrant’s service necessarily leads to a decrease of
one unit in the demand for the Incumbent’s service. To summarize, within a
segment there is perfect substitution between the Incumbent and Entrant’s
services. No empirical evidence was found in the literature showing that the displacement ratio for the mail considered is
significantly different from 1.
Each country is divided in two delivery zones, the dense or urban delivery zone (U) and the
non-dense or rural delivery zone (R). It is assumed indeed that dense zones are
situated in urban areas and non-dense zones in rural areas. Each of these
delivery areas is characterized by a grouping
index representing the number of delivery points per route stop, in
other words the average number of apartments in a building within the urban
zone for example. It is assumed that every delivery point (apartment) is
occupied by exactly one household, counting 2.6 people in the
Combining these elements, there are four different segments served by postal service providers: following the customers, where the market shares depend on customers behaviour, and following the density of delivery, where the market shares depend on unit cost differences.
a)
Demand
model
Exactly as d’Alcantara
and Amerlynck (2004), there are two steps in the demand model: the first is
related to possible substitutions and complementarities of Mail with respect to
other communication media on the broad market. The second is related to the
Mail customer’s behaviour, given the total mail volume he has determined as his
demand. The first is a constant elasticity demand model for total mail demand of a customer in a zone,
the second a market share model for each customer, determining the
substitution between postal service providers within customers’ market demand,
assumes two coefficients: a loyalty coefficient and elasticity. One defines a customer loyalty percentage as the percentage of the Incumbent’s
tariff that a given customer will accept to be higher than the Entrant’s tariff
without any decrease in his demand addressed to the Incumbent. Given the loyalty percentage of a customer and
given total market demand of this customer, the elasticity is the percentage change
of Incumbent’s volume demanded by this customer and strictly replaced by the
Entrant’s product, as a consequence of his percentage discount. Note that there
is one elasticity for each customer type, retail
and business, and for each destination, urban
and rural. The demand model consists of two sets of equations to be solved
simultaneously: a total market demand equation for each customer type and the
market share equations of the two service providers for their services provided
to the two types of customers in the two zones. The structure of these
equations is given in the Appendix.
The total market price elasticity of demand for postal services has
been assumed to be -.3. The value cannot be rejected in various econometric
studies of time series estimation surveyed by Catherine Cazals and Jean-Pierre Florens (2002).
At the stage of competing for the market shares, price
sensitivity of retail customers is relatively lower then that of business
customers. Retail customers are relatively more loyal to the Incumbent’s postal
service. The customer loyalty parameters 
are considered to be
20% for the Retail Customer and -10% for the Business Customer. The market share elasticity 
is lower for retail customers (-.75) than for business
customers (-1.25). The values of such elasticities are smaller in the case
there is a positive loyalty coefficient, which represents a lower tariff
sensitivity, and larger in the case there is a negative loyalty coefficient, which
represents a higher tariff sensitivity. We have taken the elasticity with zero
loyalty / disloyalty to be minus one. Such order of magnitude of the
elasticities for all mail categories should not be rejected, when looking
at the values found for first and second
class mail separately in various microeconometric demand studies, including
their own, surveyed by the same Catherine Cazals and Jean-Pierre Florens (2002).
Figure 1 : Customer behaviour (market share model) : the Switching Function
b) Cost Model
The detailed cost model is presented in the appendix. It is an activity
based bottom up cost model. Scale is defined as the number of mail items per
person and per year. Population size is explicit in the model but was seen from
model solutions not to be significant as an additional size factor. The model
is mainly making the quantitative assessment of what is variable and what is
fixed cost, such as the park and loop delivery round. The fixed costs related
to the USO are central to assess the impact of postal market opening.
Services offered require an upstream activity, including collecting,
sorting and transporting, and a downstream
activity comprising delivering. Our downstream model is taking over the structure defined by Robert H. Cohen,
and Edward H. Chu (1997) for the USA and includes street model complements from
K. Jasinski and E. Steggles (1977), further applied by Bernard Roy (1999). The postal
activity model is divided in the collection, the sorting of the mail, the in-office
delivery, representing the sequencing of the mail into delivery routes, transportation
and the street delivery. The
street delivery decomposes itself into route travel, delivery access time and load
time. The route travel represents the walking or driving along the route
without stopping to delivery points. Route travel has 4 modes: Foot Delivery
(the cost of foot delivery depends only of the number of hour worked by the
carrier), Bicycle Delivery (the cost of bicycle delivery depends of the
hours worked by the carrier and of the maintenance of the bicycle), Park
& Loop Delivery (the cost of the Park & Loop delivery method
depends of the hours worked by the carrier and of the maintenance of the car)
and Car Delivery (the cost of car delivery depends of the hours worked
by the carrier and of the maintenance of the car). The delivery access time represents
the time to deviate from the route to access the delivery points. The load time
represents the time to place the mail in the mail receptacle of the delivery
points.
The fact
that each delivery area is characterized by different grouping indices representing the number of delivery points per
route stop has a considerable impact on the unit costs of delivery. The cost ratio between rural and urban zone
delivery precisely depends on this grouping index. This
cost differential is the driving element in the cream skimming mechanism. One assumes the average value of this parameter in the rural zone to be equal to 1, as a standard. This is
interpreted that in the rural zone one stop corresponds to a one family house. The average value of this parameter in the urban zone of the
The
When
introducing a country, we started with the
Delivery network: Delivery points per route: URBAN 488, RURAL 469. Grouping
index: RURAL 1 delivery points/stop. Delivery speed: foot 4 km/hour; bicycle 15 km/hour; park&loop car
35 km/hour. In between stop distance: foot 20 m; bicycle 100 m; park&loop car 300 m.
Vehicle cost ratio 0,02. Entrant delivery frequency 3 delivery/week.
Entrant starting geographic coverage: URBAN 10,0% RURAL 0,0%
c) The USO, Entrant’s
behavior and the dynamic tariff modeling
The Universal Service Obligation is one of the major exogenous factors
of the model we have constructed. We have defined the following USO to be
imposed by the Regulator on the Incumbent. This allows answering the major question
considered in this study, namely the impact of economies of scale on the postal
market in different countries. The USO is as follows:
i.
Daily distribution frequency (6
times)
ii.
Constant Quality (J + 1, J + n arrival
percentage, which is related to frequency)
iii.
Distribution to all customers
(retail and business, urban and rural zones)
iv.
Cost based tariff uniformity for
end-to-end and for access
v.
Affordable tariffs (interpreted
as keeping its present value for the Retail customers)
vi.
Availability (interpreted to be existence)
The Entrant is given the objective to
become the first market player and therefore to deliver as much mail as possible without
endangering his competitiveness. Therefore the Entrant
will invest in a delivery network, as long as his cost to deliver mail is lower
than the access tariff of the Incumbent. Between two periods, the Entrant has the
ability to decide to extend its delivery network or not. The new geographic
coverage of the Entrant delivery network will be equal to the collected market
share of the previous period. This justifies the
financial losses made by the Entrant during the first periods of entry. This
amount tends to zero when investment tends to zero and the cumulated amount to
be financed corresponds to the value of the capital needed to establish his
business. Note
that once the network has been extended, it is not possible to reduce it
afterwards. The Entrant does not have any Universal Service Obligation.
Entrant’s
behaviour is determined in the following way, all of the parameters being
parameters and can be
changed in the calibration:
i.
Distribution 3 times per week (different in quality level then
Incumbent)
ii. Same cost structure
and parameters as the Incumbent
iii. Uses self-employed at
50% of national hourly wage cost
iv. Different full cost
tariffs for dense and non-dense zones
v.
A starting investment in a delivery network in view of distributing in
the urban (dense) zones a share of 10% of the corresponding total volume of the
urban zone
vi.
Access to the Incumbent’s delivery network for the residual volume,
namely the volume
above his bypass potential
vii.
Invest in own delivery network, according to yearly decision made in advance,
an amount
capable to deliver the amount
collected in excess above his existing bypass capacity of the precedent period, only if his delivery cost
is lower than the access tariff of the
Incumbent.
Access pricing of the Incumbent is set at an access cost level, the average variable delivery cost including fixed costs plus a standard overhead. It would be possible to make a sensitivity analysis of the model results using a wider range of well-known access pricing methods, including those of the form that Michael Crew and Paul Kleindorfer and the Toulouse School have indicated as more efficient (of the DAP form).
The demand model is solved using iterative method. The model iterates until finding the market equilibrium. The competitive process starts with the entry of the competing operator and the dynamic tariff determination works as follows: The tariffs are set and fixed prior to the iterations. The demand of the previous period is the starting point. All tariffs are cost based. The Incumbent takes the total cost incurred in the previous period without the eventual cost of Entrant re-injection and considering the total mail volume collected in the previous period. The unit cost obtained is set as the Incumbent tariff of the current period. Like its end-to-end tariff, the Incumbent’s access tariff equals its unitary distribution cost, including the fixed costs and a general expense margin. The Incumbent satisfies tariff uniformity, both for end-to-end and for access. As the mail volume re-injected by the Entrant is not predictable by the Incumbent, it has been chosen to compute her end-to-end tariffs only taking into account her end-to-end volume. The unit cost of the Entrant is computed likewise but including the cost of re-injecting his mail in the Incumbent’s network and considering all his mail volume collected.
i. The
starting point is the monopoly situation: volumes per customer and per zone
ii. Cost
based tariffs are fixed by Incumbent, according USO and following equations
defined in Appendix
iii. The Entrant determines his volume
objective, unitary cost following the cost model (see
Appendix), tariffs per zone, volume delivery
objective and corresponding investment in his network
iv. Cream
skimming competition is simulated by the demand model. Equation (1) of the Appendix gives a new market average
tariff. The global mail demand is updated by the equation (2). Then total mail volume is computed. Then, a new value is obtained for 
from the equation (3).
The value 
is the complementary
of the present demand
v. When
the market equilibrium is reached, the mechanism yields end of year volumes and
market shares
vi. On
the basis of the volume and market share results, the Entrant decides his
distribution network for next period. He wants to distribute what he was not
able to distribute and had to re-inject through access of the Incumbent’s
network a higher access tariff
vii. New cost based tariffs are set
viii.
Go to point iv, if % variation of one tariff or demand is higher then
convergence criterion e, this means as long as market equilibrium is not reached .
The calculation is stopped after 10 periods (i.e. 10 years), since it
marginal tariff changes become smaller then a small convergence criterion after
5 or 6 periods.
Clearly the
model describes the Entrant taking away mail through bypass: this erodes Incumbent’s
economies of scale in those zones in which bypass occurs. Such erosion
changes the volumes delivered by each operator in each zone and this determines
average costs of both operators, according to the zones in which bypass
occurs. The zone-specific impact of the Entrant is eroding delivered Incumbent
mail volumes, first in the dense low-cost zones, and therefore diminishes her
economies of scale and therefore leads to her tariff’s increase in view of
restoring her eroded profitability. In the case of the Entrant, the increase of
his volume decreases his unit cost because of increased economies of scale but
the increase of his volume in less dense zones increases his unit costs.
d) Results
Let’s first
summarize the assumptions made to calibrate our model about the standard
country, the customers, the competitor and the access regulation. One
introduces the number of items delivered per person and per year and the population
for the
Cost
elements: Urban area = 80 %. Labour cost per hour = 20,32 €. Retail ratio =
0,5. Delivery frequency Incumbent = 6. Household size = 2,6. Foot delivery = 6%.
Bicycle delivery = 0%. Park & Loop delivery = 39%. Car delivery = 55%. Exchange
rate € per dollars (2003) = 0,88... CPI (2003) = 1,40... Grouping index rural =
1. Grouping index urban = 2.
Retail and Business Customers: Market elasticity = -0,3. Retail loyalty = 20% Business loyalty =
-10%. Retail substitution = -0,75. Business
substitution = -1,25. Sigma = 1.
Competitor: Freelance cost ratio= 0,5. Delivery frequency Entrant = 3. E
starting coverage urban = 10% E starting
coverage rural= 0%.
Price regulation: Access pricing = accessed cost. Including other costs = Yes. Tariff
uniformity on access = Yes
With these
values the process converges. In a second step the values of the Mail delivery
Density are introduced corresponding to fixed intervals of 50. In this way one
obtains results corresponding to the sizes of most country: in each case the
model follows a process where scale obviously decreases for the Incumbent with
the entry of the competitor, and when the Incumbent looses dense urban zones
because of her high uniform tariffs. The Entrant benefits both from increasing
scale and from cream skimming the relatively denser delivery urban zones, lost
by the Incumbent. In Table 1. “Model Results per
Country Scale” countries are placed in view of the corresponding scales. After
5 or 6 periods the model always converges, but the period reported is period
10. One should not interpret these results as final values for the countries
since the calibration of the data used should still be considerably improved.
However the large variations obtained in the results are of interests to
understand the impact of economies of scale.
1° Solutions
with break-even tariffs for the Incumbent
When
the market opens and with the definition of the USO, what are, according to economies of scale which differ country by country,
the long term impacts
on tariffs, volumes and market shares? These
variables of the converged model result are shown following the number
of mail items per person and per year.
§ Unit cost variation under monopoly (unit cost
= 1 for
The cost curve is non-linear. While the medium
size countries
|
gU = 2 |
|
|
Austria-France |
Finland-UK |
|
|
|
|
|
|
|
Scale |
1,00 |
0,85 |
0,70 |
0,55 |
0,45 |
0,30 |
0,25 |
0,20 |
0,15 |
0,08 |
|
Unit costs |
1,00 |
1,04 |
1,10 |
1,19 |
1,29 |
1,56 |
1,71 |
1,95 |
2,35 |
3,76 |
Graphically
presented, one has
Figure 2 : Standardized Unit Cost in function of Scale
§
Incumbent tariff variation after competition (Incumbent
tariff after competition convergence divided by Incumbent tariff under
monopoly)
The tariff increase to be introduced by the
Incumbent as a consequence of the arrival of the Entrant, her loss of size and
of less costly urban zones, results to be relatively important, but much more
so in small size countries: the USA tariff increases by 53% while the French
one increases by 72% and the Italian increases by 126%. This corresponds to the
price the retail customer has to pay for the USO after the opening of the
market.
§
Entrant Urban and Rural tariffs after competition (Entrant tariff in
Urban and Rural areas after competition convergence divided by Incumbent
uniform tariff under monopoly)
The discount given by the Entrant is considerably
higher in small size countries then in large size countries: extremes vary with
66% against 39% in the Urban zones and with 36% against 29% in the Rural zones;
this also reflects a relatively more aggressive behavior of the Entrant in the
Urban zones of the small countries. Scale is less significant in Rural zones
because higher unit costs prevents the Entrant to capture market share.
§
Total volume variation after competition (volume
after competition convergence divided by volume under monopoly). The impact of
competition on reduced average tariffs and on general communication market
substitution is positive and stronger in small scale countries, the extremes
varying from a 10% increase to a 25% increase.
§ Entrant Collect
market shares on Retail customers after competition
Retail markets are more difficult to be
captured by the Entrant then Business markets because of the loyalty of the
customers and their lower tariff sensitivity, but small size again is
relatively more dangerous for the Incumbent: in the smallest country the
Entrant captures 48% of the Retail market share for a 66% discount in the Urban
zones and a 36% discount in the Rural
zone against 29% in the USA for a 39%
discount in the Urban zones and a 29%
discount in the Rural zone.
§ Entrant Collect
market shares on Business customers after competition
Business markets are much easier to be captured
by the Entrant then Retail markets, because of the readiness of business
clients to switch to the Entrant and their higher tariff sensitivity. In small
size countries total 100% market share goes to the Entrant. In the large ones
the Incumbent only keeps 14% of the business market. Clearly the business market
structure has turned upside down with the assumptions made. It is
understandable, because, from a methodological point of view, when in reality a
50% market share is reached, there is a duopoly: it is not possible to keep our
same model assumptions about the market structure and parameters. For example,
a game theoretic model should be used. The Entrant’s labor cost discount
assumed would not be realistic. No further study was done about this.
§
Financial resources needed by Entrant to invest into his own
delivery network after competition (Entrant’s cumulative profit or account
balance after competition divided by Entrant’s revenue after competition). It
is interesting to see the impact of the market opening on the financial
resources needed by the Entrant to invest into a delivery network: it is
obviously related to the importance of the delivery network he has created to
take over a high delivery market share from the Incumbent. In the case of
. Assuming a self-employed labor cost ratio of 0,5, one have 
{retail, business} represents the two customer segments
{urban, rural} represents the two destination zones
= customer j’s
demand for postal service in zone z
from the Incumbent 
= customer j’s
demand for postal service in zone z
from the Entrant
= customer j’s total
demand for delivery zone z 
= 
= 
= 
= 
= 
:
{urban, rural}. The demand model consists of two sets
of equations to be solved simultaneously: a total market demand equation for
each customer type and the market share equations of the two service providers
for their services provided to the two types of customers in the two zones. 
. Let R be the
radius of the pedestrian circular round. The number of possible stops on a
pedestrian circular round in zone z
is equal to 
. 
, one will consider the limit case d = 2R. The average time traveled by foot (movement speed of 4 km
per hour) on a park & loop route in zone z equals
. While the average time traveled by car (movement speed of
35 km per hour) on a park & loop route in zone z equals
. The vehicle cost is computed taking the same relative cost
where
b = 0.6587 and PPSoz is the number of pieces per stop in the
delivery zone z for the operator o. Variable PPSoz is linked to the volume delivered by
the operator o in the zone z (
. 
. 
