Cohenetal_comparison.txt

A Comparison of the Burden of Universal Service in Italy
*


and the United States
Robert Cohen Rossana Scocchera
Postal Rate Commission Poste Italiane
Carla Pace Vincenzo Visco Comandini
Poste Italiane Poste Italiane
Matthew Robinson John Waller
Postal Rate Commission Postal Rate Commission
Gennaro Scarfiglieri Spyros Xenakis
Poste Italiane Postal Rate Commission

1. INTRODUCTION
Samuel Johnson said Patriotism is the last refuge of a scoundrel
and many postal observers believe that the Universal Service
Obligation (USO) is the last refuge of a postal service wishing to
protect inefficiency, monopoly profits, or economic rents. Despite
inefficient service, opaque accounts, and large rents, many postal
officials maintain that the monopoly, whatever its faults, is the
only means of ensuring universal, affordable postal service for
citizens throughout the nation. Without the monopoly, they argue,
the post office would be crushed by the cost of the USO. This paper
examines when this argument might be valid.
All posts probably have some cost associated with the USO, but
it has been difficult to quantify. We agree with John Panzar (2001)
that the cost of the USO is the cost of the services that would not
be provided in a competitive market. Under this definition, in
practice, there has been no discernable USO cost. Competition
exists in both Sweden and New Zealand and the incumbent posts have
continued to provide universal service with no subsidy.
The USO as defined by the proponents of the monopoly goes well
beyond ubiquitous service. It also includes a uniform price with
respect to location of the recipient1 and a uniform frequency of
delivery. These conditions might have been appropriate a century
ago when the post was the primary means of communication. Today
however, this definition seems unduly restrictive. It is incumbent
on defenders of the restrictive definition of USO to explain why
postal service should be treated
*
The views expressed in this paper are those of the authors and
do not necessarily represent the opinions of Poste Italiane or the
Postal Rate Commission.
In other words a firm cannot charge more for delivering similar
items to addresses in one section of the country than another.
Under this definition United Parcel Service (UPS) is not a true
universal service provider because it imposes a surcharge for
residential delivery. This is ironic because UPS delivers to every
address in the continental U.S. and delivers right to the door. The
Postal Service which is a universal service provider under the
definition used here, delivers to mail boxes that are, in many
cases, remote from the actual address.
differently from other important products such as groceries,
bank accounts and internet service.2 Notwithstanding, we use this
restrictive definition of universal service for this paper.
The obligation to provide ubiquitous service is not in itself
onerous. It has been pointed out by the American co-authors of this
paper that a post could rationalize its delivery costs by reducing
the frequency of service on unprofitable routes to the point where
they become profitable (Cohen et al. 2000). A post could also
rationalize its prices by reflecting the cost of service to
different areas. The restrictive definition of the USO does not
allow a post to rationalize either its cost or pricing structure,3
thus, creating cream-skimming opportunities under liberalization.
The burden that the restrictive USO imposes on a post is the
increase in its unit costs resulting from cream skimming under
liberalization caused by the requirements of uniform pricing and
levels of service. We use the term "burden of the USO" to
distinguish our concept from other papers that use the term "cost
of the USO."
We believe that insights into the burden of the USO can be
gained by comparing postal systems. Assuming that the keenest
insights can be gained by comparing systems with large differences,
the Italian and United States systems are good candidates. The
U.S. has the highest volume per capita in the industrial world
(739 pieces in 1999), while Italy has one of the lowest (115 pieces
in 1999).4
Previous work by Cohen et al. shows that the U.S. Postal Service
(USPS) is not likely to require the monopoly to continue to satisfy
the USO. (Cohen 1999, 2000) However, this result may not apply to
all posts and it is important to distinguish magnitudes of burden
for different posts. To this end, we develop a model to determine
the USO burden for posts with different per capita volumes. In
particular, applying the model demonstrates that the burden of USO
is very great for Poste Italiane and other posts with small per
capita volumes. On the other hand, for the U.S. and other posts
with large and medium per capita volumes the burden is small.
We first compare some statistics for Italy and the U.S. We then
examine the consequences of volume loss on postal systems in order
to provide a measure of burden in terms of increases in average
unit costs as volume is lost to competitors, or
2
Many postal services are increasingly becoming primarily
carriers of advertising. U.S. households receive far more
advertising than letters, bills or other similar matter. According
to the Household Diary Study (United States Postal Service 2000),
in 1999, 56 percent of the pieces received by U.S. households were
advertising mail.
3
An apparent exception is Consignia. It rationalizes its delivery
cost by delivering twice a day in most urban areas and once a day
in rural areas.
4
Italy has a very undeveloped direct mail market and the Italian
monopoly does not include direct mail. The
U.S. monopoly includes all addressed direct mail consisting of
fewer than 24 pages. The U.S. Postal Service handled 314 pieces per
capita of direct mail in 1999.
equivalently, the increase in average rates for the volume
remaining.5 Next we introduce the concept of delivery route profit
and quantify the relative profits from delivered mail and mail not
requiring delivery in both Italy and the U.S. We then compare the
distribution of profit margins for routes in Italy and the U.S. and
the effect this has on vulnerability to cream skimming. Finally, we
discuss two measures of the cost of universal service, the entry
pricing measure and the net avoided cost measure in the context of
delivery profits. We compare them with our burden measure for the
Poste Italiane and the U.S. Postal Service.
Our principal findings are:
The burden of the USO is much greater for small per capita
volume posts than it is for large and medium per capita volume
posts.
The likelihood of a graveyard spiral resulting from volume loss
to cream skimmers is remote for medium to large volume per capita
posts; whereas it is a real threat to small volume per capita
posts.
The measure of burden presented here combined with the
distribution of route profit margins for Poste Italiane and the
U.S. Postal Service yields results contrary to the entry price and
net avoided cost measures for the cost of the USO.
The portion of mail not requiring delivery is a very important
contributor to the finances of a post.


2. COMPARATIVE STATISTICS
Table 1 contains some statistics for Italy and the U.S. and for
their postal administrations.6 Italy has 21 percent of the U.S.
population and 16 percent (approximately one seventh) of the U.S.
per capita volume. Most of the delivery statistics in the table's
second section are reasonably close to this 16 percent figure. The
number of delivery points in Italy, as a percentage of those in the
U.S., is a
5
To provide a basis for comparing Poste Italiane and the U.S.
Postal Service, we use a model of postal operations that follows
the structure of FY 1999 costs in the U.S. Postal Service.
Parametric analyses are conducted by varying volume under different
assumptions about the extent that institutional cost varies over
the long run. The model is validated for Poste Italiane by making
adjustments for major differences in the two posts, such as the
extent of worksharing.
6
Prices of goods and services vary between Italy and the United
States, and the prevailing market exchange rate does not
necessarily account for the relative differences in prices. We have
used the Purchasing Power Parity (PPP) of 1,677 liras per U.S.
dollar to convert the 1999 Italian statistics reported in Italian
liras. The PPP has been produced by the Organization of Economic
Co-Operation and Development (OECD) and shows the number of Italian
liras required in 1999 to buy goods and services equivalent to what
can be purchased with one U.S. dollar. In 1999, the average market
exchange rate, reported by the International Monetary Fund (IMF),
was 1,817 liras per
U.S. dollar.
disproportionately high 25 percent. The disproportionate number
of small business addresses in Italy (Scarfiglieri and Visco
Comandini 2001) may help explain this.
Poste Italiane also has disproportionately more delivery points
per route. This results from a much lower coverage (percent of
addresses receiving mail on a given day), a higher population
density and other unexplored route and topographical
considerations.7 We see in the fourth section that the U.S. Postal
Service cost per piece is much lower than Poste Italiane cost per
piece despite having a much higher labor cost per employee. This is
primarily a result of economies of scale, worksharing and the
extensive use of automation.8
Table 1. Selected National and Postal Statistics for FY
1999a
Italy as Italy U.S. Percent of U.S. Finally
Population 57,679,895 Area (Square Miles) 116,347
Population/Square Mile 496 Number of Households 21,211,334
272,691,000 21 we note that the 3,675,031 3 74 668 number of
retail 103,620,000 20 units (per

100,000
4
2
population)
3 16
operated by
17
Poste Italiane is
16 25
much larger
14
than the number
18 21
operated by the
144
U.S. Postal
14
154
Service.
24
illustrates
37 173
fact that
7
USO
206

This the the has
in
It
also illustrates that the concept of the USO has many more
dimensions than ubiquitous delivery. However, the cross-subsidy
that results from ubiquitous delivery and uniform prices and
frequency of delivery is widely regarded as the most important
aspect of the USO mitigating against liberalization. This paper
will be restricted to dealing with this aspect.
7
Motorcycles are the primary means of carrier transport in Italy
and automobiles are the primary means in the U.S.
8
It is also the result of the quantity of non-delivered mail (See
Section 4).


3. RELATIVE IMPACT OF VOLUME LOSS ON POSTAL SYSTEMS
The burden of the USO on a postal system lies within its fixed
costs. Ceteris paribus, unit (per piece) fixed costs are higher in
a low volume per capita system than in a high volume per capita
system. We examine the effect of volume on unit total costs.
The sensitivity of cost to volume for a postal system can be
investigated directly provided the ratio of fixed to variable costs
is known by function. Some postal administrations do not have this
information and some choose not to make it available to regulators
or the public. Consequently, we have built a model to estimate the
relationship between unit average costs and per capita volume. 9
The model is described in the appendix and includes parameters for
the portion of variable costs by major postal functions. When the
average unit cost is graphed as a function of volume per capita one
obtains a curve with a parabolic shape. The exact shape of the
curve depends on the specific set of parameters used to benchmark
the model, such as the proportion of fixed costs at a specific per
capita volume. Because the cost behavior of the U.S. Postal Service
has been studied extensively and because U.S. postal costs are the
most transparent in the industrialized world, we use U.S data from
1999 as provided in the most recent omnibus rate proceeding, Docket
No. R2000-1 to benchmark the model10 Table 2 shows the U.S.
institutional/variable percentages for the major postal
functions.
Table 2 : U.S. Fixed/Variable Cost By Major Function a (FY
1999)

a
Source: Postal Rate Commission Docket No. R2000-1
b
Delivery includes in-office and out-of-office costs.
Because the U.S. Postal Service may differ greatly from other
postal systems in labor cost, automation, route topography,
worksharing, mix of mail by shape, and the percentage of
non-delivered mail, etc., we cannot expect the cost estimates
furnished by the model to be extremely accurate estimates of
particular non-U.S. systems. For our
9
We use pieces per capita rather than total volume to normalize
the volume.
10
This model has been developed on the assumption that the cost
behavior of postal administrations is essentially similar. The
functions of collection, delivery, transportation, processing and
window services are common to the postal administrations of all
industrialized countries.
purposes we do not need great accuracy, we only need reasonable
estimates in order to examine the sensitivity of total unit cost to
volume per capita.
The fixed portion of non-delivery costs would be expected to
change as we move from higher volume administrations to lower
volume ones. It would be expected that lower volume postal
administrations would try to keep them small to reduce the burden
on rate payers. Since these costs are variable in the long run and
we do not know how other postal administrations deal with them, we
parameterize them for purposes of modeling.
As a preliminary matter, using the fixed/variable ratios for
U.S. Postal Service costs from Table 2, Figure 1 shows that as per
capita volume increases, the mail processing proportion of total
cost will increase and the delivery proportion will decrease. This
is because mail-processing costs are almost all variable while
delivery costs have a very large fixed component. Since the U.S.
volume per capita is the highest in the world, the U.S. Postal
Service delivery costs proportion should be the smallest of any
postal service. On the other hand, Poste Italiane per capita volume
is among the lowest in the industrialized world, and so its
delivery costs proportion should be among the highest.

Figure 1: Functional Percentage of Total Costs
Benchmarked by U.S. Costs and Volumes with 25% of Non-Delivery
Institutional Costs Long-Run Variable
Percentage of Total Costs
70
60
50
40
30
20
10
0

Pieces per Capita
Again using the ratios for U.S. Postal Service costs in Table 2,
Figure 2 shows the variability of the major functions as per capita
volume changes. It can be seen that at low volumes, mail processing
and transportation costs have a high degree of fixity while at high
volumes they are almost all variable.
Figure 2: M odeled Variabilities by Function
B enchm arked by U .S . C o sts and V o lu m es w ith 25% of N
on-D elivery Institutional C osts L ong-R un V ariab le

0 1 00 200 3 00 400 500 600 700 800
P ieces per C apita
Figure 3 shows the impact of per capita volume on average unit
costs. The three curves represent different variabilities of
non-delivery fixed costs as shown in the legend. It appears that
unit costs are not very sensitive to changes in per capita volume
at the high end but are quite sensitive at the low end. For
example, a ten-percent reduction in pieces per capita for the U.S.
Postal Service would result in costs increasing about one cent. A
ten-percent reduction for a post at 100 pieces per capita would
increase unit costs by nine cents. The increase in unit cost
resulting from a ten-percent decrease in per capita volume for a
range of volumes is shown in Table 3.
Figure 3: Model Estimates of Unit Cos
Benchmarked by U.S. Costs and Volumes

Pieces per Capita Long-Run Variability of Non-Delivery
Institutional Costs

In the per capita volume range of most Northern European
countries, Canada, New Zealand and Australia, the effect on unit
cost from a ten-percent reduction in per capita volume is not
great. At about 200 pieces per capita the effect of volume loss on
unit cost becomes much more pronounced.11
Table 3: Unit Cost with 10% Decrease in Volume Assuming 25% of
Non-Delivery Institutional Costs are Long-Run Variable
(1999 U.S. Dollars)
Pieces per Capita 700 600 500 400 300 200 100
Base unit cost 0.32 0.33 0.36 0.40 0.46 0.59 0.97 Unit cost with
10% volume loss 0.33 0.35 0.38 0.42 0.49 0.63 1.06
When volume is lost to cream skimmers because of the USO, net
revenue losses (revenue minus cost) must be made up from the
remaining volume to maintain breakeven status. The burden measure
quantifies the resulting increase in unit cost and revenue that
must be made up from each piece of mail that remains in the system.
An important question is whether the rate increases would lead to a
graveyard spiral. We think the burden measure provides an answer to
this question.
Insofar as the burden of the USO might introduce a death spiral
(Crew and Kleindorfer 2000) owing to the loss of volume because of
liberalization, the probability seems slight until per capita
volumes are in the 100 to 150-piece range. Below that point the
possibility of a death spiral increases rapidly. Thus, we conclude
that the burden of the USO is much greater for low per capita
volume posts than for medium to high volume ones.
In order to make the model more useful for comparing Poste
Italiane and the U.S. Postal Service we adjust model costs to
account for important known differences. Since worksharing
activities reduce United States costs by about 25 percent (Cohen et
al. 2001), we increase model costs to account for the lack of
worksharing in Italy. The portion of the model costs that are labor
costs is adjusted downwards to reflect hourly labor cost
differences between Poste Italiane and the U.S. Postal Service. In
addition, model costs are adjusted to reflect the Poste Italiane
proportion of low-cost letters (and cards), medium-cost flats, and
high-cost parcels. The appendix describes the adjustment factors.
Table 4 shows estimates of Poste Italiane unit costs derived using
the model.
It can be seen that the model comes fairly close to Poste
Italiane unit costs. Of course the model could be further refined
to account for many more specific differences. We leave that
project for another day. Nonetheless, we think the model is
sufficiently accurate to allow us to draw a major conclusion: the
burden of universal service is highly dependent on volume per
capita, so policies suitable for liberalizing medium and large per
capita volume posts are likely not suitable for small per capita
volume posts.
Greece, Italy, Portugal, and Spain have per capita volumes below
200.
Table 4: Model Estimates of Poste Italiane Unit Costs Based on
U.S. Costs and Variabilities Adjusted for Worksharing, Wage
Differences, and Mail Mix
Long-Run Variability Unit Costs ($ per piece) of Non-Delv. Inst.
Costs Attributable Institutional Total
0% 0.19 0.59 0.78 25 0.19 0.53 0.72 50 0.19 0.47 0.66


Actual Unit Cost 0.79



4. PROFITS FROM DELIVERED AND NON-DELIVERED MAIL
Because of the interest in the USO and cross-subsidy within the
delivery system, it is important to define the concept of route
profit (and loss). We define the profitability of a delivery route
as the revenue generated by the mail delivered on the route minus
the attributable upstream costs of the mail delivered and the total
cost of the route (Cohen et al. 1999). Some mail handled by a
postal service does not require delivery; recipients pick up mail
from post office boxes and large volume recipients pick up mail in
bulk from designated counters.12 Non-delivered mail can also be
considered to have delivery profits in the same manner as delivered
mail.13
A postal service has variable and fixed upstream costs as well
as variable and fixed delivery costs. In a breakeven postal service
total delivery profits from all routes plus profits from
non-delivered mail must equal fixed upstream costs.14 Put another
way, fixed upstream costs minus profits from non-delivered mail
must equal delivery profits.
In order to compare profits from delivered and non-delivered
mail in Italy and the
U.S. we increase Poste Italiane revenue by 24 percent so that it
equals cost and is thus at breakeven.15 Table 5 shows that
non-delivered mail accounts for a very disproportionate share of
delivery profits in both posts (and presumably in all posts). The
most obvious reason for this is that non-delivered mail incurs no
carrier delivery cost. In addition, although many delivery routes
are highly profitable, losses from unprofitable routes reduce the
net delivery profit from delivered mail.
12
In the United States, this is frequently called "firm holdout"
mail.
13
Revenue minus attributable upstream costs that includes the
small amount of costs for non-carrier box or counter delivery
activities associated with non-delivered mail.
14
An analogous identity holds for a post at any given profit
(loss).
15
Data in this paper are from 1999. In that year, the U.S. Postal
Service had profits of $400 million and total revenue of $62.8
billion.

Table 5: Distribution of Volume and Profits between
Delivered and Non-delivered Mail
(Percent)
Italy U.S. Volume Profits Volume Profits
Delivered 8658 79 33 Non-delivered 1442 21 67
Because total delivery profits from delivered and non-delivered
mail must equal fixed upstream costs, non-delivered mail plays an
important role in reducing the revenue needs of a breakeven post.
Ceteris paribus, the higher the proportion of non-delivered mail,
the lower the rates that a breakeven post must charge. Thus, the
higher the proportion of non-delivered mail, the less opportunity
there is for a cream skimmer to undercut the incumbent.16,17
Moreover highly profitable non-delivered mail is much less
vulnerable to diversion.18
5. DISTRIBUTION OF ROUTE PROFIT MARGINS AND VULNERABILITY TO
CREAM SKIMMING
Volumes, revenues and route profits differ widely on routes in
both Italy and the
U.S. The uniform pricing constraint of the USO creates cream
skimming possibilities because an entrant can target only highly
profitable routes and charge a price below the uniform price.
Consequently, it is not cross-subsidy per se that allows cream
skimming. All routes could be above incremental cost and the
problem would still exist. Cream skimming stems from the uniform
pricing and service constraints given that routes have disparate
profits, and not necessarily from the need to cross-subsidize
routes.
16
We would expect that the USPS's percentage of non-delivered mail
to be among the highest of industrialized country posts because the
U.S. payment system involves a large fraction of financial
transactions paid by check and mailed to large volume recipients.
This mail is generally picked up by them (or their agents) at a
post office (i.e., it is non-delivered mail).
17
In a small per capita volume post, the fixed cost of delivery
represents a larger share of total cost than in a large per capita
volume post. Ceteris paribus, per piece profit from non-delivered
mail is greater in a small per capita volume post than in a large
one.
18
Many of the 18 million boxholders in the United States do not
want mail delivered to their premises. We cannot quantify the
number, but presumably, this is an important reason why mail
recipients rent a post office box. These box holders would not be
good candidates for cream skimmers. Fifty-six percent of
non-delivered mail is single piece. It is not likely that cream
skimmers would be able to successfully put a collection system in
place to capture much of this mail. Of the 44 percent of
non-delivered mail that remains, 36 percent is First Class, much of
which is time sensitive. According to most models of cream
skimming, delivery frequency would be reduced. Time sensitive First
Class would be less likely to make use of a service with less
frequent delivery. Finally, non-single piece mail addressed to box
holders is being transported and delivered over large areas of the
nation. A delivery cream skimmer would have to have a national
presence to compete for this mail. This means much of the mail
could not be handled by local niche players or even regional cream
skimmers.
The more revenue that leaves the system because of cream
skimming, the more rates must be increased on the remaining mail to
maintain a breakeven status. Consequently, the greater the share of
system-wide revenue that is from high margin routes, the greater
impact that cream skimming could have on a postal administration.
In order to compare the possible effect of cream skimming on the
Italian and U.S. postal administrations, we have arrayed their
routes by profit margin using FY 1999 data. Again, we have adjusted
the Poste Italiane revenue to breakeven in order to facilitate
comparison.
Figure 4 displays the profit margin19 of routes arrayed by
semi-deciles. The most profitable Italian semi-decile has a margin
of just over 30 percent while the most profitable U.S. semi-decile
has a margin of over 50 percent. However, a greater percentage of
Italian routes are profitable than U.S. routes.
We do not know the profit margin that would actually attract
cream skimmers. If it is 30 percent, then 10 percent of U.S. routes
would be vulnerable and less than 7 percent of Italian routes would
be vulnerable.
Virtually all profitable routes in Italy are urban. In the U.S.,
only about half of the routes served by city carriers are
profitable while three quarters of the routes served by rural
carriers are profitable.20 It may well be that competition is much
more likely to develop in urban areas. If so, Poste Italiane would
be at greater risk from cream skimmers than the U.S. Postal
Service.
19
For purposes of this discussion, profit margin is defined as the
ratio of delivery profits to route costs (i.e. upstream variable
plus total delivery cost).
20
In the U.S. about one fifth of the routes served by rural
carriers are in high-density suburban areas.
60
50
40
30
20
Average Profit margin Percentage
10
0


(10)


(20)


(30)


(40)


(50)


(60)



1 2 3 4 5 6 7 8 9 1011121314151617181920
Semi-decile (Routes by 5% Increments)
a
Revenue for Poste Italiane is increased by 24% to create
breakeven conditions.
The percent of total revenue by semi-decile is shown in Table 6.
At every semidecile except the first, Poste Italiane receives a
greater percentage of total revenue than the U.S. Postal Service.
This is due to the fact that a much larger percentage of U.S.
revenue comes from undelivered mail.
The distribution of profitable routes shown on Figure 4
indicates that the U.S. Postal Service's revenue (volume) per route
is much more skewed than Poste Italiane's, as measured by the
relative size of the bars at the extremes. This would tend to make
the
U.S. Postal Service more attractive to cream skimmers than Poste
Italiane (at most profit thresholds). This skewness results from
mail volume being highly correlated with income and U.S. income
distribution being much more skewed than Italian income
distribution. (Berthelemy and Toledano 2000; Kolin and Smith 1999)
In fact the U.S. has the most skewed income distribution of any
industrialized country as measured by the World Bank's Gini Index.
The U.S. Gini Index is 40.8 while Italy's is 27.3.21 Thus, Italy
should be fairly typical of the distribution of route profit
margins of industrial countries and the U.S. should be an
outlier.
"World Development Indicators 2001, Table 2.8" The World Bank,
Washington, DC. The unweighted average of Gini Indices for
industrialized countries is 30.6. Austria has the lowest index at
23.1.
Table 6: Delivered Mail Revenue As Percent of Total Revenue
Routes by Semi-decile (Ranked by Profit) 1 2 3 4 5 6 7 8 9
10
Cumulative Percent of Total Revenue
U.S. Italy
10 10 16 18 21 24 26 30 31 36 35 41 39 46 43 50 46 54 49 58
Routes by Semi-decile (Ranked by Profit) 11 12 13 14 15 16 17 18
19 20 Cumulative Percent of Total Revenue
U.S. Italy
53 62 56 66 59 69 61 72 64 74 66 77 69 79 71 81 72 83 73 84
Because volume and revenue are highly correlated, if volume is
lost to cream skimming, then revenue is also lost. Figure 4 and
Table 6 can be used to estimate likely profit and revenue losses to
cream skimmers and Figure 3 to estimate the cost impact of volume
losses. Depending on the threshold selected, the potential impact
on the respective posts can be compared.
We doubt that cream skimmers could capture even a majority of
the volume on any route owing to the limited amount of contestable
volume in the U.S. and perhaps other countries.22 Nevertheless, to
the extent that volume would be lost to cream skimmers, the impact
on Poste Italiane would be greater than on the U.S. Postal
Service.
As mentioned above, the Italian data have been adjusted for
breakeven. Had we not done this, very few Italian routes would be
profitable. A postal service that is below breakeven will be less
attractive to cream skimmers than otherwise. On the other hand, if
a postal service is above breakeven, its routes will be even more
profitable and hence, more attractive to cream skimmers.



6. ENTRY PRICING
The entry pricing measure defines the cost of the USO as the sum
of lost profits from entry by a competitor. The form of entry that
we consider is delivery cream skimming. If an entrant were more
efficient than the incumbent, then the USO would not cause the lost
profit. Under the restrictive definition of the USO, however, it is
possible for the incumbent post to be more efficient than the
entrant and still lose profits because the incumbent must charge a
uniform price. The entrant would most likely attack routes with
much higher than average profit margins. The incumbent would not be
able to
Cohen et al. (2000) found that less than 16 percent of total
U.S. Postal Service volume is contestable.
respond because of the restrictive USO. Ceteris paribus, the
more skewed the incumbent's route profit margin distribution, the
more vulnerable the post is to entry.23
Figure 4 shows that the Poste Italiane profit margin
distribution is much less skewed than the U.S. Postal Service's
distribution. Thus, according to the entry price measure the cost
of the USO is higher in the U.S. than in Italy.24 This is contrary
to the conclusion that we have drawn in Section 3 (i.e., that the
burden of the USO is much higher for Poste Italiane than for the
U.S. Postal Service).
The possibility of an equal or less efficient entrant capturing
significant volume depends on the fraction of mail on profitable
routes that is actually contestable. Given the huge scale economies
of a universal service provider and the limited amount of
contestable mail, it is very difficult for an entrant, at least in
the U.S., to charge prices lower than an incumbent (Cohen et al.
1999).


7. NET AVOIDED COST
The net avoided cost measure (NAC) of the USO is the sum of the
losses from unprofitable routes (Dobbs and Golay 1995; Elsenblast
and Stumpf 1995). As stated earlier, the profits from delivered and
non-delivered mail must equal the upstream fixed cost in a
breakeven post. If the profits from the money making routes and
non-delivered mail exceed the upstream fixed cost, then there must
be an offset of money losing routes. This is simply an accounting
identity. The more skewed the distribution of delivery route profit
margins, the greater will be the losses from unprofitable routes.
Figure 4 shows that the NAC would be greater for the U.S. Postal
Service than for Poste Italiane (assuming adjustments for scale).
Again, this is contrary to our finding in Section 3 about the
relative burden of the USO.
Furthermore, we believe that the logic of the NAC measure is
flawed. Suppose that a government decides to provide a subsidy to
its breakeven postal service to offset losses on unprofitable
routes because it wishes taxpayers and not rate payers to fund the
USO. The postal service would then have a surplus. In order for the
postal service to return to breakeven, prices would be reduced. 25
Because of the price reduction unprofitable routes would become
even more unprofitable and some formerly profitable routes would
become unprofitable. Because the total losses from unprofitable
routes are greater than before, the government would have to
increase its subsidy.
23
For purposes of the discussion of entry pricing and net avoided
cost, we ignore the impact of the percentage of non-delivered
mail.
24
Bradley and Colvin (2000) estimated the entry pricing cost of
the USO for the United States. That estimate assumed that the
entrant would capture all volume on all profitable routes. It is a
worst case scenerio.
25
The same logic would hold for a postal service earning any given
profit (loss).
This process would continue until there is but one route not
subsidized by the government and that route would be the route that
initially was the most profitable one in the system. Thus, the net
avoided cost measure implies the absurdity that a government must
fund the entire delivery network if it wants to fund the cost of
universal service.


8. CONCLUSION
We introduce and quantify the concept of the burden of the USO,
which is the upward impact on unit costs that would result from
competitive entry. We find that the burden on low per capita postal
systems is much greater than on medium to high volume postal
systems. This is contrary to results obtained when using the net
avoided cost or entry pricing measures of the cost of the USO. We
also find that the burden decreases as the percentage of
non-delivered mail increases. The distribution of route profits is
much more skewed for the U.S. Postal Service than for Poste
Italiane. This would tend to make the U.S. more attractive to
potential cream skimmers. On the other hand, a greater percentage
of Poste Italiane revenue is generated by delivered mail, and this
would cause cream skimming to have a greater impact on its unit
costs.
APPENDIX
Model Definition.
The model is designed to define average unit cost as a function
of volume (Q), based on the USPS cost structure for FY 1999.
AC = TC / (Q * P)
where:
TC = DC + NDC
DC = (Vd / Q0) Q + Fd
NDC = (Vnd / Q0) Q + [ (Ev Fnd) / Q0 ] Q + (1 - Ev) Fnd
The functions can be combined and expressed as:
TC = { [ V + (Ev Fnd) ] / Q0 } Q + Fd + (1 - Ev) Fnd
The terms are defined as:
AC = Average Unit Cost d = Subscript that indicates delivery
component of a cost or volume DC = Total Delivery Cost Ev =
Long-run variability of non-delivery institutional (fixed)
costs.
Parametric values of 0, 0.25 and 0.5 used in analysis. F = USPS
Fixed Cost for FY 1999 nd = Subscript that indicates non-delivery
component of a cost or volume NDC = Total Non-Delivery Cost P =
1999 U.S. Population Q = Quantity (Volume) per capita Q0 = USPS
Volume per capita for FY 1999 TC = Total Cost V = USPS Variable
Cost for FY 1999
Adjustments for differences between the USPS and Poste
Italiane
Significant differences in the composition of mail and labor
costs exist between USPS and Poste Italiane. To make the comparison
of the unit cost from our model (which is based on USPS cost data)
and the actual unit cost of Poste Italiane more meaningful, we
apply three factors to the results of our model. The unit cost
assuming 25% of institutional (fixed) costs vary with volume in the
long run is calculated in the example.



1.
First, the attributable (variable) cost from the model
(V/Q0*P = $0.194) is adjusted to reflect the different shape mix of
mail handled by Poste Italiane. The proportions of letters (88%)
and nonletters (12%) for Poste Italiane are applied to the USPS
variable unit cost of each (13.46 and 26.60 cents, respectively).
The weighted average unit cost that results (15.04 cents) is then
divided by the actual USPS unit variable unit cost of 16.88 cents.
The resulting ratio (0.891) is applied to the variable cost from
the model.

Example: 0.194 * 0.891 = $0.173



2.
Next, the resulting variable cost is adjusted to account
for the savings from worksharing reflected in the USPS costs, which
do not exist for Poste Italiane. The USPS variable costs in FY 1999
were 40% lower than they would have been without worksharing,
(Cohen et al. 2001) so the variable cost from step one ($0.173) is
increased by that amount.

Example: $0.173 * 1.4 = $0.242


3.
Finally, both variable and institutional costs are
adjusted for labor cost differences. Since 78.6% of USPS costs are
labor costs, that portion of the variable cost from step two
($0.242) and institutional cost from the model ($0.662) is adjusted
by the ratio of Poste Italiane labor costs to those of USPS
(74.7%).


Example: [ (0.786) ($0.242 + $0.662) (0.747) ] + [ (1 - 0.786)
($0.242 + $0.662) ] = $0.724
REFERENCES
Berthélémy, Francoise L., and Joëlle Toledano. 2000. "In France,
Mail Goes Where the Money and Businesses Are." In Current
Directions in Postal Reform, edited by Michael A. Crew and Paul R.
Kleindorfer. Boston, MA: Kluwer Academic Publishers.
Bradley, Michael D., and Jeff Colvin. 2000. "Measuring the Cost
of Universal Service for Posts." In Current Directions in Postal
Reform, edited by Michael A. Crew and Paul R. Kleindorfer. Boston,
MA: Kluwer Academic Publishers.
Cohen, Robert H., William W. Ferguson, John D. Waller, and
Spyros S. Xenakis. 1999. "An Analysis of the Potential for Cream
Skimming in the United States Residential Delivery Market." In
Emerging Competition in Postal and Delivery Services, edited by
Michael A. Crew and Paul R. Kleindorfer. Boston, MA: Kluwer
Academic Publishers.
Cohen, Robert H., William W. Ferguson, John D. Waller, and
Spyros S. Xenakis. 2000. "Universal Service without a Monopoly." In
Current Directions in Postal Reform, edited by Michael A. Crew and
Paul R. Kleindorfer. Boston, MA: Kluwer Academic Publishers.
Cohen, Robert H., William W. Ferguson, John D. Waller and Spyros
S. Xenakis. 2001 "The Impact of Using Worksharing to Liberalize a
Postal Market." Papers presented at Wissenschaftliches Institut für
Kommunikationsdienste GmbH, 6th Köenigswinter Seminar on Postal
Economics, Liberalization of Postal Markets.
Crew, Michael A., and Paul R. Kleindorfer. 2000. "Liberalization
and the Universal Service Obligation in Postal Service." In Current
Directions in Postal Reform, edited by Michael A. Crew and Paul R.
Kleindorfer. Boston, MA: Kluwer Academic Publishers.
Dobbs, Ian, and Jeanne Golay. 1995. "Universal Service
Obligation and Reserved Sector." In Cost of Universal Service,
edited by Ulrich Stumpf and Wolfgang Elsenbast. Papers presented at
the 3rd Königswinter Seminar, WIK Proceedings.
Elsenblast, Wolfgang, and Ulrich Stumpf. 1995. "The Cost of
Universal Service Obligations in a Competitive Environment." Papers
presented at the 3rd Königswinter Seminar. Wissenschaftliches
Institut für Kommunikationsdienste.
Kolin, Marshall and Edward J. Smith. 1999. "Mail Goes Where the
Money Is: A Study of Rural Mail Delivery in the United States." In
Emerging Competition in Postal and Delivery Services, edited by
Michael A. Crew and Paul R. Kleindorfer. Boston, MA: Kluwer
Academic Publishers.
Panzar, John C. 2001. "Funding Universal Service Obligations:
The Costs of Liberalization." In Future Directions in Postal
Reform, edited by Michael A. Crew and Paul R. Kleindorfer. Boston,
MA: Kluwer Academic Publishers.
Rodriguez, Frank, Stephen Smith and David Storer. 1999.
"Estimating the Cost of the Universal Service Obligation in Postal
Service." In Emerging Competition in Postal and Delivery Services,
edited by Michael A. Crew and Paul R. Kleindorfer. Boston, MA:
Kluwer Academic Publishers.
Scarfiglieri, Gennaro, and Vincenzo Visco Comandini. 2000.
"Postal Profits Arise where People Are." In Future Directions in
Postal Reform, edited by Michael A. Crew and Paul R. Kleindorfer.
Boston, MA: Kluwer Academic Publishers.
United States Postal Service. 2000. "The Household Diary Study,
Fiscal Year 1999." Volume 1.