View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

I F l!'@1 <cli
ffi

(GJ

March 4,1 983

The N ewC P I
Last week, the Bureau of Labor Statistics
published a new version of the Consumer
Price Index for Urban Consumers (CPI-U).
The new index, patterned after the experimental index CPI-U-X 1 published the last
two years, changes the statistical treatment
of owner-occupied dwellings. From measuring the investment costs of housing,
including the costs of purchase and financing, the new index shifts to measuring the
cost of current shelter. This concept is not
new as the Personal Consumption Expenditure price deflator in the national income
and product accounts already usesthe same
approach. On the basis of the revised CPI-U,
all consumer prices rose an average 0.2
percent (2.5 percent at an annual rate) in
January; the unrevised CPI-U measured virtually no change in prices.
The Consumer Price Index is the leading
means of comparing the average cost to consumers of a fixed bundle of goods in different
periods of time. Dating back to 1917, the
CPI contains within its market basket of
goods the cost of housing. In recent years,
the calculation of the cost of housing for
owner-occupied dwellings has caused a
major jump in the importance of housing
(compared to that of other goods) in the
overall CPI. The distortion has made the CPI
a less representative measure of the inflation
that consumers encounter. And the distortion has ramifications for the public and
private sectors as almost all labor contracts
and virtually all government programs that
contain a cost-of-living adjustment are
indexed to a version of the CPI.
Background
Since itoriginated as a means of determining
a "fair wage scale" in World War I shipyard
labor disputes, the CPI has undergone several major changes. Until 1978, the CPI,
designated CPI-W, included wage earners
and clerical workers and covered 40 percent
of the total population. In 1978, the CPI-U

was developed to cover all urban consumers. By adding retired persons,
professional and salaried workers, the selfemployed and the unemployed, the CPI-U
increased coverage to 80 percent of the
population. Nevertheless, the CPI-W continued to be published because almost all
cost-of-Iiving adjustment contracts were still
linked to it.
The cost of homeownership came into the
index as part of a revision in 1953. Prior to
this revision, the costs of shelter were represented solely by a rental index. The rationale
behind the decision to include the cost of
a home and the mortgage interest cost reflected for the most part the belief that they
represented current expenses. While not
immediately consumed, the purchased
home was treated in a mannersimilarto that
for new and used autos and household durable goods (also not immediately consumed) that were already part of the CPI
basket.
The effects of including the costs of home
purchase in the CPI were not immediately
evident. When only rental costs were measured, the relative importance of housing to
the total index declined from 33.7 percentof
consumer expenditures in the mid-1 930s to
33.5 percent forthe 1 947-49 consumer
expenditure survey. It increased only slightly, after the costs of home purchasing were
counted, to 34.9 percent in the 1960-61
survey. However, by the 1 972-73 expenditure survey on which the CPI-U was
based, the relative importance of housing in
the CPI for all urban consumers had risen to
nearly 43 percent. A major reason for the
increase was a rise of over 4 percentage
points in mortgage rates from the early
1960s to 1972-73.
Homeownership costs-the problems
Five separate components are involved in
determining the costs of homeownership in

IEDsl1
lliIR\
§&\l ffi

I f if

li

©

()pinions expressed in this newsletter do not
necessarily reflect the views of thernanagement
of the rpderal Reserve Bank o'f San .Francisco.
or of the Board of Covernors of th(' Federal
ReserV(: System.

Rentassheltercosts

the CPI-U. Three of them represent the routine costs of running a household: property
taxes, insurance and maintenance and
repairs. The other two components -house
prices and mortgage costs-have been the
major targets of critics of this method of
calculating homeownership costs.

The solution adopted by the BLSfor the
problems posed by direct measurements of
homeownership costs is the use of a "rental
equivalent" to homeownership. The rationale for this alternative measure is simple
enough: the cost of shelter consumed by
homeowners should represent the rental
income they forego by occupying their units
instead of renting them out. .!n this way, one
would measure the value of the flow ofshelter services rather than the asset value of a
house. House prices would not necessarily
reflect this value because houses are assets
and, consequently, are valued for their
potential for capital gains (or losses) as well
as their value as shelter.

Home prices. The major criticism directed
against the series of home prices that the BLS
uses in the CPI-U has been that the data
sample is biased downward. The sample is
taken from the list of persons obtaining F H A
home mortgages, but because there is a ceiling on the size of such mortgages, higher
priced houses are excluded from the
sample. As Dr. Alice Rivlin, director of the
Congressional Budget Office, has pointed
out, the upper limit on F H A eligible mortgages was not raised as rapidly as the surge
in home prices in the late 1 970s. As a result,
the home prices from this source became
less and less typical of home prices at large.

The Bureau of Labor Statistics has used an
index of rental costs since the inception of
the CPI. For rental units, the BL5 merely
compiles data on rent payments. But for
owner-occupied units, the BLS has had to
solve the problem of finding an adequate
sample of neighborhoods that contain both
houses that are rentals and houses that are
owner-occupied. Furthermore, the houses
from each group must be roughly equivalent
in quality. Then, the BLS must estimate the
rental equivalent of owner-occupied houses
on the basis of actual rent payments for the
rental units.

Mortgage interest-the major problem. The
costs of financing a new home present more
statistical problems to the construction of
the CPI than do house prices. The mortgage
cost component of the CPI does not simply
measure the current mortgage rate. It is
based upon the mortgage cost of half of
(what used to be) the standard 30-year fixed
rate mortgage, or the mortgage cost for 15
years, given the current market interest rate.

The BLS has been working for several years
with alternative approaches to the cost of
shelter, and the one that appeared most
likely to solve the homeownership problem
is the rental equivalence alternative, called
C PI-U-X1 .ln Chart 1, the annual rates of
change for the two major parts of the homeowner component of the CPI-house prices
and contract mortgage costs -are compared with changes in the residential rent
index. The annual rate of change in mortgage interest costs ranged from -7 percent in
1971 to +44 percent in 1 980. The annual
rate of change in home purchase price
dropped as low as one percent in 1 973, later
rose to over 15 percent at the end of 1 979,
and fell to an annual rate of change of 6.5

In the past two to three years, housing finance has changed to such an extent that
there simply is no "typical" mortgage package as far as interest rates and maturity are
concerned. The variable rate mortgage and
shorter mortgage maturities have been
adopted by lenders to protect themselves
against capital losses in the event of future
inflation. The mortgage market has been
further complicated by the advent of
"creative"financing,
wherein home sellers
assume a part of the financing package that
was generally at rates below those offered by
the usual sources of mortgage funds.
2

Chllrt1

Chart2

COMPONENTSOF THE CONSUMERPRICE INOEX,
ANNUAl. flATES Of CHANGE

40

DIFFERENTIAL IN
ANNUAL RATES OF CHANGE
Percenl

Mo<tgege
Conlr8tl

CPI·W

minus CPW·X1

2.0

0!-'l-fJ"

o H H --"---"----'-I

."

1970

1972

1974

1976

1976

1900

J--..3.r.f-----\

-

1!:lS2

CPI-U were constructed on the same basis,
the annual rate of change has rarely varied
between the two, and when it did, it has
never varied by more than 0.2 percent. If
one were to examine the differential
between the rate of change in the CPI-W and
the rate of change in the experimental
CPI-U-X 1 for the period 1970-1 982, the
CPI-W would seem to have overstated the
rate of inflation by 2 percent in 1980 arid·to
have underestimated the rate of inflation by
0. 8 percent at the end of 1982 (seeChart 2).

percent in the second half of 1982. The'
range of annual rate changes in the last
12 years for residential rent was much narrower, running from a little over 3 percent to
just over 9 percent. It was averaging about
7V, percent in the second half of 1982.
Clearly, house prices and mortgage interest
costs were major factors in the increase of
theCPI in 1974-75 and again in 1 979-80.
The switch to an index using the rental
equivalence measure of the costs of current
shelter for homeowners would also change
the relative importance of other components
of the CPt. A comparison of their respective
weights in the CPI-U and CPI-U-X1 at the
end of 1981 (when house prices and mortgage interest costs had already begun to fall)
indicates that the costs of homeownership,
which had a relative weight of 26.1 percent
when treated as the cost of acquiring and
financing an asset, would have had a relative weight of 13.8 percent had they been
treated on an imputed rental basis. The
importance of other items in the consumer's
market basket would have risen. The estimated importance of food (17.5 percent to
2004
percent), apparel (4.6 percent to 504
percent), transportation (19.3 percent to
22.5 percent) and energy (11.1 percent to
13.0 percent) would have played a greater
part in determining the cost of living.

In the past dozen years, the CPI-W has consistently overestimated the rate of inflation
by an average annual rate of 0.6 percent
when compared to the CPI-U-X1. A 1979
study by the General Accounting Office estimated that federal payments to individuals
(about 30 percent offederal budget expenditures, 90 percent of which are indexed)
increased between $1.5 billion and $2.5
billion for each additional percentage point
of inflation. Thus, the use of the CPI-W asthe
basis for cost-of-living adjustments (COLAs)
probably resulted in a significantoveradjustment in the cost of living. Moreover, there
are questions of equity involved when one
population group is over-compensated for
inflation while other groups not so protected
suffer a loss of real income due to inflation.
The new Consumer Price Index being introduced by the Bureau of Labor Statistics
appears to eliminate much of the overstating
(and occasionally understating) of the
observed rate of inflation. However, since
virtually all cost-of-living adjustments in
labor contracts and in federal budgetexpenditures are presently still tied to the CPI-W,
the individuals affected by the new index
may face lower COLAs than those not similarly affected and be undercompensated for
the inflation rate; workers, for example, may
face pay cuts. Still, since a better statistical
method has been devised for measuring the
rate of inflation, there is every reason to
change existing institutional and pol itical
arrangements to take advantage of the new
index.
•

Impact on the economy

The ultimate impact of the revised CPI-U
must await the test of experience. However,
we can expect the influence of interest rate
changes to be much smaller on the revised
index. The only use scheduled to be made of
the revised index is in the indexation of
income brackets for the federal income tax
in 1985, but the BLS also intends to discontinue the CPI-W in that year.
As mentioned earlier, almost all labor contracts and virtually all government programs,
such as social security and government
pensions, that contain a cost-of-living
adjustment, are indexed to the CPI-W.
Because both the CPI-W and the unrevised

Herller! Runyon

3

S S\t1 :> .LSl:lI:l
• yl'ln • uo8,:uQ• epeAaN• oyepl
!!PMl?H •. '

•

euozp\!

•

e>jsl?IV

J!,ill
JO:>SPUl?J:I

Iill'@q;r[
TI'@J!@jp)@,ill
<\l)

ues

ZSt 'ON
a1
Vd
'S'n
llVW SSVD

{)\\JJ<OJ\ll!l!{)JlW@@«JI

BANKINGDATA-TWELFTH
FEDERAL
RESERVE
DISTRICT
(Dollar amountsin millions)

Selected ssets Liabilities
A
and
Large
Commercial
Banks
Loans
(gross,
adjusted) investments*
and
loans(gross,
adjusted) total#
Commercial industrial
and
Real
estate
loansto individuals
Securities'
loans
U.s.Treasury
securities*
Other
securities"
Demand
deposits total#
Demand
deposits
-adjusted
Savings
deposits total
Timedeposits total#
'Individuals, & corp.
part.
(largenegotiableCD's)
Weekly
Averages
of DailyFigures
Member
Bank
Reserveosition
P
Excess
Reserves )jOefidency
(+
(-)
Borrowings
Netfreereserves )jNetborrowed{
(+
-)

Amount
Outstanding
2/16/83
163,341
142,076
45,014
57,319
23,671
2,365
7,570
13,694
40,042
27,420
62,170
72,481
64,037
24,263
Weekended
2/16/83
95'
73
22

Change

Change
from
from
year ago
Dollar
Percent
2/9/83
283
4,739
3.0
231
4,971
3.6
32
2,185
5.1
72
677
1.2
279
1.2
- 17
424
21.9
- 192
29
1,306
20.9
- 10.1
23
1,538
1,894
1,490
- 3.6
- 255
853
3.2
881
31,615
103.5
-1,601
- 17,822
- 19.7
-1,384
- 20.9
- 16,945
- 10,391
- 30.0
- 926
Weekended
Comparable
2/9/83
period
40

o

40

151
63
88

'" Excludes
trading
account
securities.
# Includes
items shownseparately.
not
Editorial
comments beaddressed theeditor
may
to
(Gregory
Tong) theauthor••.. Free
orto
copies thisand
of
otherFederal
Reserve
publications beobtained callingor writingthePublicInformation
can
by
Section,
Federal
Reserve
Bank SanFrancisco, Box7702,San
of
P.O.
Francisco
94120.
Phone
(415)