Changes in the Household-Level Demand for Postal Delivery
Services from 1986 to 1994
Frank A. Wolak Department of Economics, Stanford University
Motivation-Over the past decade there has been an dramatic
increase in 1) Available modes of interpersonal communication and
the range in quality of these modes of communication in terms
speed, reliability, and flexibility 2) Reductions in the price of
these modes of communication-long-distance telephone service, FAX
machines, on-line information services CompuServe and America
Online (zero price for incremental messages)
Many of these modes of communication are very attractive
alternatives to traditional postal delivery services supplied by
the United States Postal Service (USPS).
Particularly true for household sector-E-mail, Electronic bill
paying, FAX, long-distance call
Total pieces delivered and mail revenues have continued to
increase
Percentages by volume in 1996: First-Class--53%, Standard Mail A
(Formerly Third Class)--39%, Periodicals--6%, All others--2%.
Standard Mail A-Primarily advertising circulars and mail-order
catalogues-comprises an increasing share of pieces delivered
37.5% of pieces delivered 1992 versus 39% of pieces delivered in
1996
From 1993 to 1994, single-piece First-Class volume fell by 0.2
percent, but a 6.8 percent increase in pre-sorted First Class
volume resulted in a net 2.4 percent increase in First-Class
volume. From 1994 to 1995, single-piece fell by 1.0 percent and
from 1995-1996 it fell by 1.5%.
This event appears to signal a leveling-off or decline in
household-level use of USPS postal delivery services
Purpose of Talk
Reconcile growth in aggregate pieces delivered and mail revenues
with increasing relative price of postal delivery services and
growing number of substitutes at household level
Are household sector trends consistent with aggregate
trends?
Quantify decline in aggregate household sector postage
consumption.
Determine the relative extent to which the decline in household
sector
consumption over time can be attributed to
1) Increasing penetration of alternative modes of
communication
2) Increasing relative price of USPS postal delivery
services
Specific Questions Addressed
1) What has happened to the average household-level consumption
and aggregate (US Population) household consumption of postal
delivery services over the past decade?
2) What are own-price, cross-price and expenditure elasticities
of the household-level demand for postal delivery services and how
have they changed over time?
3) What is the impact of home computing technology on postal
demand?
4) What household characteristics predict differences in
household-level postal demand?
Aggregate Household Consumption of Postal Delivery
Services
Bureau of Labor Statistics (BLS) Consumer Expenditure Survey
(CES). A national probability sample of households. Sampling frame
generated from the 1980 Census 100-percent detail file.
Two surveys administered to different samples of households: (1)
Quarterly Interview Survey and (2) Diary Survey
Diary Survey-Each selected household is requested to keep two
one-week diaries of all expenditures over consecutive weeks. Source
for householdlevel postage expenditures and other non-durable goods
expenditures.
Interview Survey-Household is interviewed every three months
over a 15month period. Questions about durables holdings--cars,
housing, and personal computers.
Both surveys collect information on household
characteristics-hours of work of the head and spouse, occupation of
head and spouse, age and race of head and spouse, marital status,
number of children, dwelling type, income, and census region of
residence.
For each household in Diary Survey from 1986 to 1994, BLS
computes a sampling weight giving the representativeness of that
household in the population of US households during the year it is
sampled
1) How many US households represented by a household with
these
demographic characteristics
Using these weights and BLS-recommended procedure, can compute
estimate of US population aggregate expenditures on postal delivery
services or aggregate expenditures any other category of goods
*Preliminary figures. Nominal declines aggregate household
postage expenditures. In constant
dollars, decline would be steeper due to real postage price
increases Decline from $74.26 per household in 1989 to $54.35 in
1994
Almost double nominal increase in aggregate household telephones
services expenditures from 1986 to 1994.
Increase from $421.67 per household in 1986 to $690.65 in
1994
Shift away from consumption of postal delivery services to
telephone services for household sector
From 1987 to 1994 share of USPS mail revenues from US household
fell From a little less than 20 percent to a little more than 10
percent
Significant shift way from consumption of postal delivery
services for household sector relative US economy at large
Second research question: What are the sources of these shifts
in the consumption of postal delivery services?
Digression on Economic Model of Household Behavior
Each household is assumed to have a demand function for each
good which depends on the prices of all goods, the household's
budgeted total expenditure and observable and unobservable
characteristics of the household
Let p ,p ,...,p be the prices for goods household purchases, M
be the
12 K
household's budgeted total expenditure and A ,A ,...,A are the
attributes
12 N
(observable and unobservable characteristics) of the
household.
Define the household's demand function for the ith (i=1,...,K)
good x * = D (p ,p ,...,p ,M,A ,A ,...,A )
i i12K 12 N
the amount each good, x , the household consumes. There are K
goods and
i
N characteristics.
One household characteristic we focus on is an indicator
variable of whether or not the household owns a personal computer.
This is our measure of the potential for electronic
substitution.
This will allow us to determine extent to which decline is
household-level
demand is due to
1) Increasing relative price of postage
2) Increasing penetration of personal computing technology
Using econometric modeling techniques applied to our sample of
households, we are able recover an estimate of the demand
function
x * = D (p ,p ,...,p ,M,A ,A ,...,A )
i i12K 12 N
for each household in the sample.
Demand functions differ across household depending on observable
and unobservable characteristics of the household
Using our econometric model, we can compute an estimate of any
US household's demand for postal delivery services given its M and
A ,A ,...,A
12 N
Quantifying Demand Responses at Household-Level
Price Elasticity of Demand = One percent increase in the price
of the ith good, p , brings about what percentage change in the
household's demand for
i
good j, Dj
If i = j, then it is the own-price elasticity of demand,
otherwise magnitude is called the cross-price elasticity of
demand.
Elasticity concept can be extended to any other variable in
household's
demand function Expenditure elasticity of Demand = One percent
increase in the household's total expenditure, M, brings about what
percentage change in household-level demand for good j, Dj
For each US household can compute an estimate of its own-price
and crossprice elasticities of demand and the expenditure
elasticity of demand.
One of observable attributes that is assumed to shift the
household's demand
for postage is whether or not is owns a personal computer. Model
yields an estimate of the percentage change in a household's demand
for postage as a result of owning a computer
Aggregate Demand Function for the Household Sector
Using BLS weights, can compute an estimate of the aggregate US
population demand function for postage from the household sector as
the weighted sum of sample household level-demand functions Can
compute own-price and cross-price elasticities of aggregate
household demand for postage using this aggregate demand
function.
Estimated Aggregate U.S. Household Postage Demand Elasticity
Estimates These magnitudes give the percent change in the demand
from the household sector as result of a one percent change in any
price, household's total expenditure or fraction of US households
owning a personal computer.
Substantial increase in absolute value of own-price elasticity
of aggregate household demand from 1986 to 1994, from -0.76 to
-1.27.
Decline in total expenditure (income) elasticity of demand from
0.36 to 0.25 over same period.
Increasing degree of substitutability between postal delivery
services and telephone services over past decade, from -0.12 in
1986 (complements) to
0.11 (substitutes) in 1994
Increased penetration of personal computing technology at
household-level
initially predicts increased demand for postal delivery
services.
By 1994, predicts reduction in demand for postal delivery
services,
holding all other factors constant.
Estimated Aggregate Revenue Losses Based on 1994 Elasticity
Estimates
In general, an X% own-price increase in a product with an
own-price elasticity of , increases revenues by X(1 + ,)%.
If , is less than one in absolute value revenues increase. If it
is greater than one revenues fall. In 1994, , = -1.27.
Note that before 1989, , is less than one in absolute value, so
that aggregate household revenues increase with price
increases.
Consider January 1, 1995 price increase for a one ounce
First-Class letter from $0.29 to 0.32, a little more than a 10%
increase.
From estimated annual household postage expenditures in 1994 are
approximately $5 billion, which implies a 2.7 percent reduction in
annual aggregate expenditures or approximately a $135 million
reduction in annual revenues from household sales, holding all else
constant.
Impact of on household-level demand a 17 percent increase in
penetration of computer ownership for all households.
In 1994, fraction of households owning a personal computer is
0.25.
A 17 percent increase in this number is 1.17(0.25) = 0.29, a
plausible increase in the penetration of computer ownership by
households over the course of a single year.
From estimate in above table US aggregate elasticity of demand
with respect to computer ownership is -0.16. Therefore, 17%
increase in purchase probability brings about a 0.16(17) = 2.7
percent reduction in aggregate demand.
Assuming no postal price change, this implies a 2.7 percent
reduction in revenues from sales to household, or $135 million
reduction in revenues, same as a 10 percent price increase.
Caveats
There is uncertainty associated with all of the numbers
presented in this paper due to sampling error and estimation error
in econometric estimation procedure used to recover household-level
demand functions
Less confidence in exact magnitudes reported, more confidence in
trends
they illustrate 1) Decline in revenue share from household
sector 2) Increasing, in absolute value, own-price elasticity 3)
Increasing substitutability away from postage consumption with
computer use 4) Increasing substitutability between postage and
telephone consumption.
Economic and econometric theory underlying analysis available in
technical paper by same title that can be downloaded from
http://www-leland.stanford.edu/~wolak
Future Research
1995 BLS data recently made available. It is source of
preliminary numbers. 1996 data promised by end of summer.
Have the above trends continued for household-level demand in
1995 and 1996?
Sources of electronic substitution-USPS Postal Diary Survey
provides detailed information telecommunications capital equipment
holdings, on-line service member and electronic bill paying.
