Full text of Review (Federal Reserve Bank of St. Louis) : May 1978

View original document

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

FEDERAL RESERVE BANK
OF ST. LOUIS
MAY 1978

CONTENTS

il f

A Perspective on the Economy:
Three Years of Expansion
JEAN M. LOVATI

ARCH 1978 marked the third anniversary of
the current economic expansion. Over the course of
the expansion, growth in production of goods and
services has been comparable to the average output
growth in similar three-year periods of prior recov
eries. By other measures, however, this recovery has
not been average. Growth in employment, for ex
ample, has been exceptional. Moreover, the level of
unemployment and the pace of inflation both have
remained relatively high.

Production
In the most recent quarter, production of goods and
services, as measured by real gross national product,
declined at a 0.6 percent annual rate. This recent
decline reflected the pervasive effects of severe winter
weather early in 1978 combined with an extended
strike in a major portion of the coal industry. The
effects of these factors, while having a debilitating
impact on the economy, are temporary in nature. The
decline in aggregate production early this year should
not be regarded as a signal of a failing expansion, as
economic fundamentals remain strong.
Over the entire expansion, from first quarter 1975 to
first quarter 1978, real gross national product in
creased at a 5.1 percent annual rate. This is about
average compared to other postwar recoveries. Dur
ing the first three years of four previous expansions,
output grew, on average, at a 5.6 percent rate1 (see
Table I).
While growth seems to be on track with other
recoveries after three years, the path of the current
recovery has taken on its own unique characteristics.
Initially, the economy expanded at a rapid pace, ex
periencing a 9 percent rate of growth in real output
iPostwar expansions used for comparison in this article are
those whose troughs are October 1949, May 1954, February
1961, and November 1970. The expansion beginning in April
1958 was eliminated because it lasted only two years.

Page 2

1970

1971

1972

1973

1974

1975

1976

1977

1978

Source: U.S. D e partm en t of C om m erce
P e rc e n tage s are an n u a l rates of c h a n g e for p e rio d s indicated.
Q _ G N P in current d ollars.
12 G N P in 1972 d ollars.
Latest d a ta plotted: 1st q u a rt e r p r e lim in a r y

of goods and services between the first and third
quarters of 1975 (Chart I ) . During this time, inven
tory stocks continued to decline, but did so at a re
duced rate. This slowing down of inventory decumu
lation as well as the dramatic rise in real final sales
were reflected in the exceptional rate of growth early
in the recovery. Over the first year of expansion, real
output grew 7 percent.
The second year of the recovery was marked by
more sustainable growth. Production of goods and
services advanced 4 percent between first quarter 1976
and first quarter 1977. Real final sales expanded stead
ily, registering a fairly strong 4.4 percent gain in the
expansion’s second year.

MAY

FEDERAL RESERVE BANK OF ST. LOUIS

1978

Table 1

A C O M P A R IS O N O F SELECTED P O S T W A R E X P A N S IO N S 1
Rates of change after 3 years for expansions beginning:
October
1949
(IV /1949)

May
1954
(11/1954)

Real GNP

7 .8 %

4 .0 %

5.3 %

5 .1 %

5 .6 %

5 .1 %

Real final sales

6.8

3.6

4.9

4.4

5.0

4.2

Industrial production

February
1961
(1/1961)

November
1970
(IV /1970)

AVERAGE

March
1975
(1/1975)

12.2

6.5

8.1

7.9

8.7

8.1

Retail sales

8.0

5.8

5.7

11.9

7.9

11.1

Real disposable personal income

4.9

4.6

4.5

5.0

4.8

5.0

G NP deflator

3.8

2.7

1.6

5.4

3.4

5.7

Consumer price index2

4.1

1.3

1.2

5.1

2.9

6.4

Wholesale price index — all commodities

4.4

1.8

- 0.2

7.9

3.6

6.2

Wholesale price index — industrial commodities

4.1

3.0

-0 -

5.4

3.2

6.6

- 0.1

1.6

1.1

2.5

1.3

2.6

1.6

2.2

1.6

2.9

2.1

3.5

Civilian labor force/population 16 years +

5 8 .7 %

5 9 .5 %

5 8 .8 %

6 1 .2 %

6 2 .8 %

Civilian employment/population 16 years-f-

56.9

57.1

55.6

58.2

59.0

3.0

4.1

5.4

4.8

6.2

1 .1 %

Civilian labor force
Civilian employment

Levels as of 36th month of expansion

Unemployment/labor force
Change in the unemployment rate3

- 4 .9 %

1 .8 %

1 .5 %

2 .4 %

^ o e s not include expansion beginning April 1958
2Series for urban wage earners and clerical workers
3From trough to 36th month of expansion

In the latter part of 1976, the economy experienced
some inventory adjustments which induced a slow
down in the growth of total output. At the end of the
year, the ratio of inventories to monthly sales in man
ufacturing and trade industries fell precipitously. The
situation was compounded by severe weather condi
tions in the first quarter of 1977. Such complications
limited inventory rebuilding and output growth.
These temporary effects were offset, however, by
faster growth in the two subsequent quarters, so that
the recovery regained its footing and achieved a 5.7
percent increase over 1977 as a whole.
Many of the same factors were at work at the end
of 1977 and the beginning of 1978. Inventories again
declined relative to monthly sales. As noted above,
severe weather combined with the coal strike tempo
rarily hampered production. Despite these factors,
output grew 4 percent in the third year of the
expansion.

Sector Activity
The over-all expansion has been reflected in sub
stantial growth of spending by different sectors in
the economy at different stages of the recovery. In the
first year, for example, consumer spending made sub

stantial advances, while spending by businesses on
plant and equipment and by the Federal Government
were rather sluggish. As the expansion progressed in
the next two years, growth in these sectors gradually
reversed.
Personal consumption expenditures, which com
prise about 64 percent of gross national product, in
creased rapidly early in the expansion, advancing 13
percent in the first year of the recovery as substantial
gains were made in personal income. Even after ac
counting for inflation, real household spending regis
tered a 7 percent growth between first quarter 1975
and first quarter 1976. Spending on consumer durables,
which increased 25 percent in the first four quarters
of the recovery, was boosted by heavy purchases of
autos and appliances. Expenditures on nondurable
consumer goods rose 9 percent over the same period.
While remaining relatively strong, growth of consump
tion expenditures has slowed from initial rates in the
recovery. Since the first quarter of 1976, spending on
consumption goods and services grew in nominal
terms at a 10 percent annual rate.
Investment in residential structures has also made
significant gains over the course of the current expan
Page 3

FEDERAL RESERVE BANK OF ST. LOUIS

sion. Residential housing advanced at a nominal rate
of 28 percent per year over the three years of ex
pansion. Housing starts surpassed 2 million units in
1977, with single-family starts exceptionally strong.
This strong demand for housing partially reflected an
attempt to make up for new house purchases post
poned during the recession. Between second quarter
1974 and first quarter 1975, investment in residential
structures had declined at more than a 21 percent
rate. Recent strength of demand for housing is prob
ably also influenced to some extent by homebuyers’
uncertainty about future home prices. As housing
prices have risen sharply in the last several years,2
some housing purchases were based on the assump
tion that new homes would become even more expen
sive relative to income in the future.
Spending on plant and equipment, on the other
hand, demonstrated little strength during the first
year of the recovery, growing 4 percent in nominal
terms. This relatively slow growth in investment
spending was due in large part to continued busi
ness uncertainty resulting from the indefinite status
of various Government regulations, proposed changes
in Federal energy and tax programs, and the future
course of inflation. Higher replacement costs and
lower productivity of capital goods due to higher
energy costs have impeded growth of capital outlays.
Inflation, combined with the tax structure, has low
ered the yield on, and increased the cost of obtaining
the necessary funds for a given investment program,
thus eroding incentives to invest.3 Recently, however,
business fixed investment has shown signs of renewed
strength. Since the fourth quarter of 1976, plant and
equipment spending has advanced at about a 14 per
cent rate.
Government purchases of goods and services
showed little strength in the first two years of the
expansion. Between first quarter 1975 and first quar
ter 1976, real Government spending advanced 2 per
cent and over the following four quarters declined
slightly. This pattern has been reversed, however,
since the beginning of 1977. Real Government spend
ing has increased more than 4 percent in the past
year. The recent rise in the growth of Government
spending has been sharpest among state and local
governments, which coincides with increased grantsin-aid through the Federal Government’s 1977 eco2Housing prices, as measured in the consumer price index,
have risen at an 8 percent annual rate since 1970.
:lSee John A. Tatom and James E . Turley, “Inflation and Taxes:
Disincentives for Capital Formation,” this Review (January
1 9 7 8 ), pp. 2-8.

Page 4

MAY

1978

nomic stimulus program. This program includes local
public works, public service employment, and other
employment and training programs.

Labor M arket D evelopm ents
A significant characteristic of the present expansion
has been the rapid growth of employment. In the 36
months of the expansion, civilian employment has
grown at an annual rate of 3.5 percent. This growth
is exceptionally high by historical standards. Over
comparable 36-month periods in past expansions, civil
ian employment grew at an average 2.1 percent rate.
Labor force growth also has been rapid over the
recovery. A record 62.8 percent of the civilian noninstitutional population aged 16 and over were mem
bers of the labor force in March 1978. In other periods
of relatively high employment, 1956 and 1967-69 for
example, 60 and 59.8 percent, respectively, of the
working-age population were in the labor force. In
the three years of the current expansion, the civilian
labor force has grown at a 2.6 percent rate, twice the
average rate of growth achieved during the first three
years of previous expansions. Demand for labor has
been relatively strong, so that many in the labor force
have been placed in jobs. The number of employed
persons in the working-age population reached a
postwar peak of 59.0 percent in March, higher than
at any time in the prior 29 years.
Significant progress has been made in lowering the
unemployment rate over the past three years. The
unemployment rate declined from 8.6 percent in
March 1975 to 6.2 percent in March 19784 (see
Chart II). Despite such progress, the rapid growth in
the labor force left more than 6.1 million workers
currently recorded as being unemployed.
The level of the unemployment rate reflects var
ious factors, including structural changes in the com
position of the labor force, which have tended to
maintain its relatively high position. The labor force
now contains relatively more women and teenagers
than heretofore. Adult women comprised 37.1 percent
of the labor force as of March of this year, a record
high. Moreover, while the participation of women
has been increasing throughout the postwar period,
it has intensified within the past five years. Participa
tion of adult women (aged 20 and over) in the labor
force reached a peak of 49.1 percent in March, com
pared to an average participation rate of 37.8 percent
4In April, the unemployment rate fell to 6.0 percent.

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

Chart II

Labor M a r k e t Trends

1948

1950

1952

1954

1956

1958

1960

1962

1964

1966

1968

1970

1972

1974

1976

1978

Shaded areas represent periods of business recessions.
Latest data plotted: 1978, based on first four months

in prior periods of expansion.5 Much of the impetus
for the recent rise can be attributed to married
women (spouse present) whose participation in the
labor force has risen over 3.6 percentage points since
1974. Of these women, many have small children
( infant to 5 years old), a characteristic whose inhibit
ing influence on labor market participation seems to
be diminishing.
Teenagers also account for a larger proportion of
the labor force than in previous periods. Labor force
participation of workers aged 16-19 stood at 56.7 per
cent in March, nearly the highest in recent history.
Partially because of the level of skills of these
groups of workers, plus the restrictions of the min
imum wage and the tenuous nature of labor force
■’Participation rates measure the proportion of persons in a
specific population group that are in the civilian labor force.

attachment in the case of teenagers, higher rates of
unemployment are generally experienced by women
and teenagers. Due to the increased size of these
groups, their unemployment rates are weighted more
heavily which, in turn, tends to raise the average level
of the overall unemployment rate. In March 1978,
unemployment rates for adult women and teenagers
were 5.8 and 17.3 percent, respectively. In the same
month, adult men had an unemployment rate of
4.5 percent.
According to current data, the average duration of
unemployment is less than 13 weeks. This is well
within the present maximum limit for receipt of
unemployment compensation. The broad scope of
benefit programs, coupled with savings and other
sources of aid such as food stamps, tend to reduce
the economic hardship suffered by the unemployed.
However, such unemployment programs also tend to
reduce incentives for the jobless to obtain work, pre
Page 5

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

serving a high number of unemployed, even under
fairly tight labor market conditions.6

Prices
Another feature of the current expansion which
distinguishes it from others is the much higher rate
of inflation which prevails. The implicit price deflator,
a measure of the general level of prices, rose at a 5.7
percent rate over the twelve quarters of this recov
ery. In the first twelve quarters following the troughs
of past recessions, the general level of prices rose an
average of 3.4 percent per year.
C h a rt III

Prices

Ra ti o Scale
1967=100
230

Ra ti o Scale
1967=100

S e a s o n a lly Adjusted

|-------

220 ----------

1978

The difference between the inflation experience in
the current recovery and that in previous recoveries
primarily reflects differences in the growth of money.
Growth of prices is primarily determined by the trend
growth rate of money. In the five years preceding
this recovery, the money stock grew at a 6.2 percent
rate, greater than money had grown in the five years
prior to any other postwar recovery. While money
stock increased at a 5.7 percent rate between Novem
ber 1966 and November 1970, in the corresponding
periods preceding the expansions beginning May 1954
and February 1961, money had grown at only 3.1
and 1.4 percent rates, respectively. Moreover, money
growth has been accelerating since the current recov
ery began, reaching a 7.3 percent rate in the period
from the third quarter 1976 to first quarter 1978
(Chart IV ).

2 1 0 --------------

This expansion in the money supply reflects growth
in the monetary base, its prime determinant, which
has shown a marked acceleration during the recovery.
In the first two years of this expansion, the monetary
base had grown at an 8.2 percent rate, from the 7.7
percent rate recorded in the previous three quarters.
In the expansion’s third year, the base growth accel
erated to a 9.3 percent rate.
C h art IV

M o n e y Stock a n d M o n e ta ry Base

1970

1971

1972

1973

1974

1975

1976

1977

1978

P ercentages are a n nu a l rates o f cha n ge for p e rio d s indicated.
* C P I for U rb a n W a g e Earners an d C le rical W o rk e rs (revised series b e g in n in g Ja n u a ry 1978).
Note: CPI, se a so n a lly adjusted, (old series) term inated b egin n in g J a n u a ry 1978.
L a te st d a t a p lotted: C o n su m e r-M a rc h ; W h o le sa le -A p ril

Prices, as measured by the consumer and wholesale
price indexes, show a similar pattern. The consumer
price index (C P I)7 rose at a 6.4 percent rate between
March 1975 and March 1978 (Chart III). In the past,
however, the CPI averaged a 2.9 percent annual rate
of growth over similar recovery periods. On the whole
sale level, prices of industrial commodities rose at a
6.6 percent rate in this expansion, compared to an
average rate of 3.2 percent in other expansions.
6See Martin Feldstein, “The Economics of the New Unem
ployment,” T he Public Interest (F all 1 9 7 3 ), pp. 3-42.
7This CPI refers to the series for urban wage earners and
clerical workers.
Page 6

nonm em be r banks. A djustm ents are m ade for reserve requirem ent c h a n g e s a n d shifts in deposits
am on g classe s of b anks. D ata are com puted b y this Bank.
Percentages are an n u a l rates of chan ge for p erio d s indicated.

FEDERAL RESERVE BANK OF ST. LOUIS

SUMMARY
The current expansion has reached the average
length of other postwar recoveries and, in terms of
production of real goods and services, has posted
about the same rate of growth from the trough as
have other expansions. However, specific characteris
tics of the current situation are unique. Plagued by
severe winters and temporary shortages of energy,
growth of real output has been hampered at various
stages of this recovery. Inventory investment, al
though accelerating recently, has also been impeded
by various adverse factors. Business fixed invest
ment expenditures still remain low relative to other
recoveries.
While the expansion has been reflected primarily
in growth of private consumption, investment and
Government expenditures may take on more prom
inent roles in the future. Investment expenditures as
well as Federal purchases of goods and services have
rebounded from their sluggish behavior in the earlier
stages of the recovery. During 1977, gross private
domestic investment, after accounting for inflation,

MAY

1978

grew nearly 17 percent; real Federal Government
purchases advanced at a 10 percent rate over the last
three quarters of the year.
Housing growth, on the other hand, which has
shown exceptional strength during the recovery, may
be tempered due to rising interest rates and exhaus
tion of “pent-up” demand forces. Rates on short-term
market instruments, such as 3-month Treasury bills,
have risen to levels equal to the ceiling limits on sav
ing and loan time deposit accounts. If disintermedia
tion becomes a serious problem, the housing sector
may suffer more serious slowing.
Two problems, formerly an unusual combination,
are likely to continue to be characteristic of the pre
vailing economic environment. The unemployment
rate, which has been primarily influenced by struc
tural and supply factors, is generally expected to re
main relatively high. Inflation is also not likely to
show any slowing in 1978. Based on past rates of
monetary growth, inflation can be expected to run in
the neighborhood of 6 percent during this year.

Page 7

Comparing Per Capita Output Internationally:
Has The United States Been Overtaken?
JAI-HOON YANG

I n 1950 the United States was generally recognized
as having the highest per capita output in the world.
Using exchange rates to convert foreign output into
dollars, the level of U.S. per capita output in 1950
was more than 50 percent higher than that of any
other industrialized country.1 During the next two
decades the conventional exchange rate-based meas
ure of comparison indicated that these industrialized
countries markedly narrowed the U.S. lead. By 1970
U.S. per capita output was still more than 15 percent
higher than the next highest industrialized country,
Sweden.
By 1974, however, that same conventional measure
indicated that Sweden and Switzerland had over
taken the United States. Reportedly, Canada, Den
mark and West Germany have joined the club.2 Citing
these developments, one critic of the U.S. economic
system speculated that “the lack of government planlThe concern here is with the relative per capita levels of
actual output of goods and services produced (such as per
capita gross national product or gross domestic product as
conventionally measured in accordance with the prevailing
United Nation’s System of National Accounts) and not with
the elusive and speculative measures of relative levels of
economic welfare. The primary reason for focusing on a
measure of production, such as per capita output, rather
than on a measure of welfare, such as consumption per
capita adjusted for length and conditions of work, is that
in most studies and popular discussions of international
comparisons, measures of per capita output have been used.
For a discussion of reasons why measures of production,
rather than of welfare, are compared, see Milton Gilbert
and Irving B. Kravis, An International Comparison of Na
tional Products and the Purchasing Power of Currencies
(Paris: Organization for European Economic Cooperation
(O E E C ), 1 9 5 4 ), pp. 72-76. Fo r a discussion of the distinc
tion between a measure of welfare and a measure of output,
see Edward F . Denison, “Welfare Measurement and the GNP,”
Survey of Current Business (January 1 9 7 1 ), pp. 13-16, 39.
“West Germany was reported to have become “just a bit
richer” than the U.S. on a per capita basis on June 24, 1973.
See J. W . Anderson, “The Relative Wealth of Nations,”
Washington Post, 2 July 1973. Denmark’s leap forward was
reported in early 1977. See Lester C. Thurow, “The Myth of
the American Economy,” Newsweek (February 14, 1977),
p. 11. These reports are based on a temporary (as short as a
day) dip in the value of the dollar. Based on an annual or
even on a quarterly average basis, Denmark and W est Ger
many have yet to pass the United States, even in terms of the
conventional measure. Canada’s entry into this exclusive club
was reported recently in “U.S. Slips to 4th in National In
come,” Washington Post, 13 May 1978.
Page 8

ning, worker participation, and social spending may
in fact be at the heart of our poor performance in
recent decades.”3
There is good reason to question the conclusion
about the comparative levels of per capita output
based on the conventional (exchange rate-based)
measure. Specifically, the method of using exchange
rates to convert output of different countries into a
common currency, such as the U.S. dollar, has sev
eral serious drawbacks. First, actual per capita out
put of goods and services in different countries does
not necessarily change every time exchange rates be
tween the countries change, although the conven
tional measure of comparison would indicate such a
change. Second, the exchange rate between curren
cies serves to equalize, at best, the prices of goods
traded between countries. However, total output in
each country also consists of goods and services which
are not traded but are consumed domestically. Price
differences in these non-traded goods are not neces
sarily captured in the exchange rate. To the extent
the exchange rate does not reflect such a difference
in the prices of non-traded goods, any comparison
based on the conventional measure would be dis
torted. In addition, the prevailing exchange rate may
not even equalize the prices of goods traded interna
tionally for a variety of reasons, including govern
ment interventions in the markets for foreign
exchange.
To overcome these shortcomings of the exchange
rate-based measure, economists have developed an
alternate measure based on the relative purchasing
power of different currencies over both traded and
non-traded goods. This alternate measure of inter
national comparison indicates that the U.S. lead in
per capita output in the earlier period (1950 through
1970) was generally much narrower than that indi
cated by the conventional measure. Also, these esti
mates indicate that the U.S. lead has yet to be
overtaken.
3See Thurow, “The Myth,” p. 11.

FEDERAL RESERVE BANK OF ST. LOUIS

T h e Conventional M easure of International
Comparisons of P er Capita Output
The conventional measure of international compar
isons using exchange rates between two countries,
say, Germany and the United States, is quite simple.
First, the per capita German output in Deutsche marks
(D M ) would be converted to dollars by the prevail
ing exchange rate. The resulting per capita German
output in dollars would be divided by the per capita
U.S. output, also measured in dollars. The resulting
quotient, expressed in percentage form, is the conven
tional measure of the relative level of per capita
output between Germany and the United States. For
example, suppose that per capita German output in a
given year is DM4,000 while U.S. per capita output is
$2,000. Assume that the exchange rate is DM4/$1.00.4
To derive the conventional measure of international
comparison, the per capita German output would be
converted to dollars (DM4000
DM4/$1 = $1000)
and then expressed as a percent of U.S. per capita
output ($1000/$2000 = .50, or 50 percent). For the
sake of illustration, if the exchange rate in this ex
ample changes to DM2/$1.00 (the dollar depreciates),
the conventional measure becomes 100 percent. In
other words, per capita output in Germany would be
estimated to be equal to that of the United States.
This conventional measure of international compar
isons is a unique number for given estimates of per
capita output of any two countries (denominated in
their respective national currencies) and a given ex
change rate. Also, the conventional measure is easy to
construct. These attractive features explain why the
conventional measure is regularly published and
widely quoted.5 However, this procedure is fraught
with conceptual difficulties. So much so that since
the early 1950s, there have been concerted attempts
to construct more appropriate measures.6 The im4D M 4/$1.00 denotes that 4 units of the German currency
(Deutsche mark) can be exchanged for one U.S. dollar.
sStatistical Yearbook, 1976 (N ew York: United Nations, 19 7 7 ),
pp. 686-88; The World Bank Atlas: Population, Per Capita
Product, and Growth Rates (W orld Bank, 1 9 7 6 ); “East vs.
W est: W ho’s Richer — and W hy,” The Morgan Guaranty
Survey (February 1 9 7 8 ), pp. 6-9; and “A Special Report —
Socialism: Trials and Errors,” Time (M arch 13, 1 9 7 8 ), espec
ially pp. 26-27.
''For a pioneering study of international comparisons based on
an extensive collaboration with the statistical agencies of the
countries involved, see Gilbert and Kravis, An International
Comparison of National Products. Fo r a follow-up study with
an expanded coverage, see Milton Gilbert and associates,
Comparative National Products and Price Levels; a Study of
Western Europe and the United States (Paris: O EEC , 1 9 5 8 ).
For a recent study, see Irving B. Kravis, Zoltan Kenessey,
Alan Heston and Robert Summers, A System of International
Comparisons of Gross Product and Purchasing Power (Balti-

MAY

1978

portant point to note for now is that the allegation
that the United States has lost its lead in per capita
output in the 1970s has been based exclusively on the
conventional measure of international comparisons.

T h e Nature of the Difficulty
W ith the Conventional M easure
International comparisons of per capita output
must be based, in principle, on a comparison of the
quantities of both internationally traded goods (such
as radios) and non-traded goods (such as haircuts)
produced in different countries.7 The basic difficulty
with the conventional measure, which uses prevailing
exchange rates, is that this measure is known to be
valid only when (a ) the relative prices of traded and
non-traded goods are identical between the countries
(a haircut costs the same amount in terms of radios
in both countries), and (b) the prevailing exchange
rate is such that the prices of traded goods are
equalized (an American-made radio costs just as
much as one of similar quality made in Germany).
These conditions, especially the one calling for iden
tical price structures (relative prices) in each coun
try, are unlikely to be met. Therefore, there can be
no presumption that the procedure underlying the
conventional measure would yield a valid measure
of comparison.
To clarify this point, consider the example given
in Table I. There are two hypothetical countries,
Alpha and Beta. Prices in Alpha are denominated in
pounds, denoted by £ . Beta’s prices are denom
inated in dollars, denoted by $. Country Beta is
assumed to produce greater amounts per capita
of both traded goods, such as radios, and non-traded
goods, such as haircuts. Country Alpha produces %
as many radios per capita as country Beta and % as
many haircuts per capita. If we compared only the
traded goods, Alpha’s output would be % or 66.7
percent of Beta’s; on the basis of non-traded goods,
Alpha’s output would be % or 33.3 percent of Beta’s
output. However, the task of comparing per capita
output internationally is to express country Alpha’s
more and London: the Johns Hopkins University Press for the
World Bank, 1 9 7 5 ). The first two studies are based on the
comparison of GNP and its components. The study by Kravis,
et al. deals with the GDP and its components.
7Non-traded goods, of course, include government provision of
goods and services which are not priced in the market place.
Valuation and comparison of such government output present
difficult conceptual and measurement problems which are
ignored in this paper. For a recent attempt to deal with the
problem of valuing government output, see Keith Leffler,
“Government Output and National Income Estimates: The
Effect on International Comparison” (forthcoming in Carnegie-Rochester Conference Series, 9, 1 9 7 8 ).
Page 9

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

Table 1

H ypothetical E x am ple W ith

Identical Relative Prices
Traded Goods: Radios (per capita)

Non-Traded Goods: Haircuts (per capita)

(1)

(2)

(3) = ( 1 ) X (2)

(4)

(5)

(6) = ( 4 ) X ( 5 )

Country

Price

Quantity
(per capita)

Expenditure
(per capita)

Price

Quantity
(per capita)

Expenditure
(per capita)

(7) = ( 3 ) + (6)
Total
Expenditure
(per capita)

Alpha

£10

£100

£50

£200

£300

$30

$60

Beta

total output per capita as a percentage of country
Beta’s total output per capita. Therefore, a valid
measure of comparison would fall somewhere within
the upper and lower limits of the ratios (expressed in
percentages) of the quantities of each good and
service produced in one country to those of another.
For this example, a valid measure of comparison must
place the total per capita output of Alpha somewhere
between 33.3 percent and 66.7 percent of that of the
total per capita output of Beta. Conversely, any meas
ure that does not fall within these limits is not a
valid measure.
In Table I, the relative price of radios and haircuts
is assumed to be identical in both countries, that is,
radios cost 5 times as much as haircuts in both coun
tries (£ 5 0 / £ 1 0 and $5/$l). The average level of
prices in Alpha (in £ ) is assumed to be 10 times
higher than that in Beta (in $). If the exchange rate
is £10/$1.00, as is likely under free trade, the under
lying condition that justifies the use of the conven
tional measure is met. The relative level of per capita
output of Alpha would be computed by first dividing
the aggregate value of its per capita output (£.300)
by the exchange rate and then expressing the result
ing figure of $30 (£ 3 0 0
£10/$1.00) as a percent
age of the aggregate value of Beta’s per capita output
($60). In this instance, the resulting conventional
measure is 50 percent which falls within the limits
required for a valid measure.
However, there is no presumption that the proce
dure underlying the conventional measure would
yield a valid measure of comparison when relative

prices differ between countries. To analyze a more
likely case where relative prices of traded and non
traded goods are different between countries, con
sider the example given in Table II (which is identi
cal to Table I except that the price of radios in Alpha
is now £ 1 5 rather than £ 5 0 ). Relative prices in the
two countries are no longer identical. Radios cost 5
times as much as haircuts in Beta whereas radios cost
only 1% times as much as haircuts in Alpha.
Since the prices of traded goods are assumed to be
equalized internationally through adjustment in the
exchange rate, the equilibrium exchange rate would
be £3/$1.00 for this example. The conventional
measure for comparing per capita output in this in
stance would be 88.9 percent [( £ 1 6 0 -r- £3/$1.00)
-f- ($60) = .889]. Thus, the conventional measure is
immediately seen to be an invalid measure for com
paring per capita output since it is even higher than
the highest of the relative quantities in the example
(66.7 percent for radios, the traded good).
As these examples demonstrate, only under the twin
assumptions of (a) identical domestic price structures
(relative prices) across countries and (b) a market
determined exchange rate which equalizes the prices
of traded goods would the procedure underlying the
conventional measure yield a valid measure for com
paring per capita output between countries. These
special assumptions are not generally met for a
variety of reasons. Inter-country productivity differ
entials across commodity groups (such as traded vs.
non-traded goods) would result in different domestic
price structures. Also, government interference,

Table II

H ypoth etical E x am ple W ith

Different Relative Prices

Non-Traded Goods: Haircuts (per capita)

Traded Goods: Radios (per capita)

(1)

(2)

(3) — (1) X (2)

(4)

(5)

(6) = (4) X (5)

Country

Price

Quantity
(per capita)

Expenditure
(per capita)

Price

Quantity
(per capita)

Expenditure
(per capita)

Alpha

£10

£100

£15

£60

£160

$30

$60

Beta

10
Digitized for Page
FRASER

(7) = (3) + (6)
Total
Expenditure
(per capita)

MAY

FEDERAL RESERVE BANK OF ST. LOUIS

through such devices as exchange controls and import
quotas, distorts the exchange rate such that it may not
equalize the prices of traded goods between the two
countries.
Even where the exchange rate is allowed to be
determined freely in the foreign exchange market,
one is not likely to observe equal prices for traded
goods. Some prominent reasons given for this are
the differences in (a) the cost of transportation, proc
essing and distribution, both between and within
countries, (b) tax structures (indirect vs. direct taxes),
and (c) selective subsidies on certain classes of
commodities.8
Hence, the conventional measure based on the ex
change rate is not necessarily a valid measure for
comparing per capita output between countries. The
point to note is that the conclusions regarding the
comparative levels of per capita output since 1950
and the allegations about the United States falling
behind in the 1970s are founded on no more substan
tive basis than the conventional measure of interna
tional comparison discussed and illustrated in this
section.

Table III

C o m p u tatio n o f PPPs a n d M e a su re s o f
In tern ation al C o m p a riso n 1
C o m p u tatio n o f PPPs
For Table I
£ 1 0 (1 0 ) + £ 5 0 (4 )
PPPOt =
$1 (10) + $ 5 (4 )
=
[PPPOC denotes an estimate using
(1 0 haircuts and 4 radios)

A generally valid measure for comparing per capita
output between countries can be constructed by using
what is known as the purchasing power parity (PPP)
of currencies. PPP is defined as the ratio of the num
ber of units of one country’s currency (say Deutsche
mark) to the number of units of another country’s
currency (say the U.S. dollar) which are required to
purchase the same bundle of both traded and non
traded goods. To estimate PPP, the total per capita
output of a given country is priced first by the prices
prevailing in the given country and then by the other
country’s prices. An estimate of PPP is obtained by
the ratio of the resulting market values of the total
per capita output which has been priced. An alterna
tive estimate of PPP may be obtained by pricing the
total per capita output of the other country. Ex
amples of PPP calculations are shown in Table III.
For the example in Table I where the price struc
tures are identical, PPP is unique (at £10/$1.00)
8Fo r a discussion of the limitations of the conventional measure,
see Gilbert and Kravis, An International Comparison of Na
tional Products, pp. 14-17. Also see Paul A. David, “Just How
Misleading Are Official Exchange Rate Conversions?” E co
nomic Journal (September 1 9 7 2 ), pp. 979-90; Robin Rarlow,
“A Test of Alternative Methods of Making GNP Compari
sons,” Economic Journal (September 1 9 7 7 ), pp. 450-59.

£300
„
,
W
=
£ , 0 / $ ’ -00
Alpha's quantities
as weights]

£ 1 0 (3 0 ) + £ 5 0 (6 )
£300
.
^
PPPli =
$ 1 (3 0 ) + $ 5 ( 6 )
=
130" =
[PPPP denotes an estimate using Beta's quantities
(3 0 haircuts and 6 radios) as weights]
For Table II
£ 1 0 (1 0 ) + £ 1 5 (4 )
PPPOt =
$ 1 ( 1 0 f + $ 5 ( 4 ) ... =

pppP

£ 1 0 (3 0 ) + £ 1 5 (6 )
$1 (3 0) + $ 5 ( 6 )

—

£160
-$ 3 T =
£390
$60

^-33 /$l .00
£ 6 .5 / $ 1 .0 0

Com pu tation o f P P P -b ase d M e asu re s of C o m p a riso n 2
For Table I
£300

PPP-based measure =
£300

Alternative M easure of International
Comparisons Based on Purchasing
Power Parities

1978

= "W

$30

£ 10/$1.00 = $ ^ = -5 or 50 percent

For Table II
£160

Estimate Based on PPPCX — _

£ 16 0

pppcc

1
£ 5 . 3 3 / $ 1.00 =

Estimate Based on PPPp —
=

£ 1 60

~$£0~

PPP|3

£160

~$6cT

£ 6 .5 / $ 1 .0 0

° r 50 percen'

' 41 ° r 41 percen‘

C o m pu tatio n o f the C o n ve n tio n a l M e a su re 3
For Table I
Conventional measure —
£300
=

“ $^o~

£300
$$0

1
$30

£ 1 0 /$ 1.00 =

$15 -

-5 or 50 percent

For Table II
Conventional measure =
£160
$60

1
£ 3 / $ l .00

£160

1
^

$53.33
$60

.889 or 88.9 percent

*A detailed explanation of the procedures used in this table is pro
vided in the unpublished technical Appendix which is available
from the author upon request.
2Obtained by deflating the ratio of per capita total expenditure by
the estimated P PPs.
3Obtained by deflating the per capita expenditure ratio by the
exchange rate. The exchange rate is assumed to be such that the
prices of traded goods are equalized.

Page 11

FEDERAL RESERVE BANK OF ST. LOUIS

and would be equal to the exchange rate if the ex
change rate is such that the prices of traded goods
are equalized.9 When the price structures are different
as in Table II, however, the PPP is not, in general,
equal to the exchange rate.10
Hence, the alternative measure for comparing per
capita output based on PPP is not in general equal to
the conventional measure based on the exchange rate.
To illustrate the difference in the exchange rate-based
conventional and the PPP-based alternative measures
of international comparisons of per capita output,
refer to Table III. Computations reported in Table III
illustrate that the PPP-based measures are valid for
comparison in the sense that they (being 41 percent
and 50 percent) lie between the upper and lower
limits (66.7 percent and 33.3 percent) of relative quan
tities of traded and non-traded goods. As noted
above, the conventional measure derived by the use
of the exchange rate, on the other hand, is inappro
priate in that it, being 88.9 percent, lies outside of
these limits. In general, the use of PPP first to con
vert the per capita output of a given country into
common currencies and then to express it relative
to the per capita output of the base country yields
without exception a valid measure for comparing the
per capita output of the two countries.11
9For a discussion of the relationship between the exchange rate
and the purchasing power parities (P P P ) of currencies, see
Bela Baiassa, “The Purchasing-Power Parity Doctrine: A
Reappraisal,” Journal of Political Economy (Decem ber 19 6 4 ),
pp. 584-96. When PPP is equal to the exchange rate, the
absolute version of the purchasing power parity doctrine is
said to hold. The doctrine posits a relationship between the
rate of exchange between two currencies and the purchasing
power of currencies over both traded and non-traded goods
in two countries. The absolute version links the level of
exchange rates one-for-one to the purchasing power parities,
whereas the relative version of the doctrine relates the re
quired adjustment in the exchange rate (from the posited
base-period equilibrium exchange rate) to the relative changes
in the general price levels. There are no theoretical reasons
for either one of these versions of the doctrine to hold, unless
the relative price structures either remain unchanged or
become identical. See Baiassa, “Doctrine,” pp. 584-87.
10Computations given in Table III show that PPP is not equal
to the exchange rate (assumed to be £ 3 / $ 1.00) but ranges
between £ 5 .3 3 /$ 1 .0 0 and £ 6 .5 /$ 1 .0 0 . This reflects the fact
that, given the lower relative price of traded goods in Alpha,
the comparative purchasing power in Alpha’s currency over
both traded and non-traded goods is lower than that indicated
by the exchange rate which takes into account only the
prices of traded goods. Thus, whereas the exchange rate
indicates £ 3 is equivalent in purchasing power ( over the
traded good) to $1.00, the PPP indicates that £ 5 .3 3 (or
£ 6 . 5 ) is equivalent to $1.00 in buying power over both
the traded and non-traded goods. Further, the use of own
quantity weights, rather than the quantity weights of Beta,
results in a larger estimate of the purchasing power of £
( that is, a smaller estimate of P P P ) because relatively
cheaoer traded goods get a greater weight in the estimation
of PPP. When PPPs are not unique, geometric averages of
the different estimates of PPP are often used.
U A more general substantiation of these points, made so far

Digitized forPage
FRASER
12

MAY

1978

International Comparisons of P er Capita
Output Based on PPPs
The alternative PPP-based measures for comparing
per capita output have not been widely used in the
past, however, primarily because of the relatively
high cost of constructing them. Conventional meas
ures can easily be constructed with data routinely
available in regularly published statistical releases.
The data collection and processing requirements for
the construction of the PPP-based measures are stag
gering. Such measures require, in principle, price and
quantity information on each individual good and
service produced in different countries.12 In addition
to the cost of data collection, the existence of com
modities which are unique or not identical in quality
(such as a Rolls-Royce vs. a Volkswagen) poses both
conceptual and measurement problems.13
The high cost of using the PPP method explains
why the estimates based on this method are available
for only a selected number of countries and only for
selected periods.14 Essentially, PPP-based estimates
are available for 1950, 1960 and 1970. Table IV lists
PPP-based estimates relative to the United States for
a sample of countries for which data are available for
some years through 1970 (with the exception of Switzwith the aid of simple examples, is provided in an unpub
lished technical Appendix which is available from the author
upon request. It is shown in that Appendix that the basic diffi
culty with the conventional measure is that the procedure un
derlying its computation does not yield, in general, a measure
called a quantity index (th at is, a weighted average of the
quantity ratios or relatives of each distinct group of commod
ities). It is a quantity index which must be constructed in
order to make an international comparison of per capita out
put of goods and services. The procedure underlying the
conventional measure is shown to be equivalent to deflating
the ratio of per capita expenditures (expressed in different
currency units) by the prevailing exchange rate. An alter
native approach by which the per capita expenditure ratio is
deflated by an estimate of the purchasing power parity ( P P P )
of currencies is shown to yield — without exception — a
quantity index. Hence, the alternative, PPP-based estimates
of comparative levels of per capita output must be used to
make a valid comparison of per capita output across countries.
12In practice, however, price and quantity data on aggregate
categories — such as men’s clothing, passenger cars, furni
ture, and physicians’ services — are constructed by decom
posing the more readily available expenditure data, using
primarily the price data on selected components of each
category, for example, dress shirts and business suits for the
men’s clothing category. The most recent U.N. study uses
for example, 153 such expenditure categories. See Kravis,
et al., A System of International Comparisons, p. 171.
13For a discussion of the statistical procedures used to deal
with these problems, see Gilbert and Kravis, An International
Comparison of National Products, pp. 79-91; Kravis, et al.,
A System of International Comparisons, pp. 31-34.
14Fo r an exhaustive list of the countries and the time periods
for which the PPP-based measures are available, see Irving
B. Kravis, “A Survey of International Comparisons of Pro
ductivity,” Economic Journal (M arch 1 9 7 6 ), p. 19.

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

Table IV

Intern ational C o m p a riso n s o f Per C a p ita O utput
1950,

1960,

(U .S. =

1970

100)

Conventional Measure
Via Exchange Rates
1950
United Stales
Canada
Switzerland
Sweden
Belgium
Denmark
United Kingdom
France
Norway
Netherlands
West Germany
Italy
India
Japan

100
66
52
47
43
39
38
37
34
26
25
16
N.A.
N.A.

1960
100
79
52
67
44
46
49
48
46
35
46
25
N.A.
16

Alternative Measure Via
Purchasing Power Parities
1970

1950

1960

1970

100
80
72
86
55
66
46
58
60
51
64
36
2
39

100
75
N.A.
58
52
55
55
46
52
45
37
25
N.A.
N.A.

100
77
N.A.

100
83*
N.A.
78'
72
72 *
60
75
63 *
72
75
46
7
62

69
57
63
64
58
55
52
64
33
N.A.
N.A.

Source: International Monetary Fund, In tern a tio n a l F in a n c ia l S tatistics (May 1977), O EEC, Balassa and the United Nations.
* Based on extrapolation using the growth rates of real per capita output.
N .A .: not available.

erland). Compared to the conventional measure
given in the same table, the PPP-based measure
indicates that the U.S. lead in per capita output was
not as great in 1950. For example, the U.S. per capita
output was a trifle more than twice that of the Nether
lands according to the PPP-based measure, rather
than the nearly four fold lead indicated by the ex
change rate-based measure. Both measures indicate
that the gap has narrowed over the 20-year period
and that the U.S. lead has yet to be overtaken
through 1970.15 The alternative (PPP-based) measure
indicates, however, that the narrowing of the gap in
most cases has not been as dramatic as indicated by
the conventional measure.16
15For those countries for which the PPP-based measure was
available for 1960 and earlier but not for 1970, the 1970
estimate was obtained by extrapolating the 1960 measure by
the differential growth rates of real per capita output be
tween the given country and the United States.
laTable IV also indicates that the conventional measure based
on the exchange rate tends to be lower than the PPP-based
estimate over this period. However, given the varying de
grees of intervention in the foreign exchange markets and
changes in the role a national currency (such as the U.S.
dollar) plays as international means of payment and store of
value, there is no necessity for the conventional measure to
be systematically lower than the PPP-based measure, espec
ially for countries with relatively similar price structures and
productivity. The fundamental point to note is that the
PPP-based measure is a valid measure without exception and
that the conventional measure can be on either side of the
PPP-based measure, depending upon the particular histori
cal circumstances that happen to prevail. Even the direction
of bias in the conventional measure cannot be assessed a

In order to assess the allegation, based on the con
ventional measure, that the United States has lost its
lead in per capita output in the 1970s, we proceed as
follows. For each selected country the latest available
PPP-based estimate of per capita output is projected
forward in time by using the various countries’ actual
growth rate in per capita output. The PPP-based
estimate for the United States is similarly projected.
Comparison of the projections for each country, ex
pressed as percentages of the projection for the United
States, is then made.
As an illustration of the method, the PPP-based
estimate for West Germany in 1970 can be used. The
PPP-based estimate of per capita output for Germany
in that year is 75 percent of that of the United States.
To derive the PPP-based estimate for 1976, for ex
ample, the German per capita output for 1970 was
calculated by first multiplying the 1970 U.S. per capita
output by the PPP-based measure of comparison for
priori but must be determined on a case-by-case and countryby-country basis, that is, only after the conventional measure
is compared to the PPP-based measure. The observed ten
dency of the conventional measure to be lower than the
PPP-based estimate over this period has been explained in
terms of greater inter-country productivity differential in the
traded-goods sector (such as manufacturing and agriculture)
than in the non-traded goods sector (such as personal serv
ices and government). The presumption has been that there
is a greater opportunity for technological innovations in the
traded goods sector. Fo r details, see Balassa, “Doctrine.”
Page 13

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

Table V

Intern ational C o m p a riso n s o f Per C a p ita O utput
1970,

1 974-76

(U .S. =

100)

Conventional Measure
Via Exchange Rates

Alternative Measure Via
_________ Purchasing Power Parities

1970

1974

1975

1976

1970

1974

1975

1976

100

Canada

100

105

Switzerland

115

122

117

N.A.

Sweden

103

118

114

Belgium

N.A.

Denmark

United Kingdom

59
N.A.

United States

France

N.A.

Norway

Netherlands

75
76

West Germany

Italy

N.A.

India
Japan

2
39

Source: The same as Table IV . Computations of PPP-based measures are based on extrapolation using relative rates of growth of real
capita output.

Germany for 1970. The German per capita output in
1970 (in 1970 U.S. prices) is $3596 (75 percent of
$4795 for the U.S. in 1970). This 1970 German per
capita output was extrapolated to 1976 by using the
average growth rate (2.3 percent per year) of the
German real per capita output over the 1970-1976
period. The resulting figure of $4125 was 76 percent
of the 1976 per capita U.S. output (in 1970 U.S.
prices). Consequently, the estimated PPP-based meas
ure in 1976 was 76 percent for West Germany.17
Table V provides comparisons based on these extra
polated estimates and on the conventional measure
17This method admittedly involves the assumption that the
price structure in one country has not changed relative to
that of the other country over the extrapolation period. It
may appear that the objection raised to the use of the
exchange rate, that it assumes identical price structures
across countries, is applicable to this method of extrapolat
ing the PPP-based estimate. However, this extrapolation
method is superior for the purpose of comparing the levels
of per capita output since the use of the PPP provides a
correct gauge of the comparative levels of per capita output
for the initial period. The use of the exchange rate would
not lead to a correct measure of relative levels of per capita
output at any time over the entire period of comparison, as
long as the price structures are not identical. The basic
reasons of course, are that the PPP-based method, unlike the
exchange rate method, is not distorted by the different de
grees of intervention in the exchange market and that the
method allows for the differences in the prices of both
traded and non-traded goods. The available evidence indi
cates that this extrapolation method is superior to the
exchange rate-based method. See the discussion below and
Kravis, et al., A System of International Comparisons, pp. 8-9,
especially footnote 13.
14
Digitized for Page
FRASER

for 1970, and for 1974 through 1976.18 The numbers
are in index form, with the United States scaled at
100. Therefore, if a number exceeds 100, it sustains
the claim that the United States has been overtaken
by that country.
The conventional measure indicates that some coun
tries have overtaken the United States in the 1970s.
Per capita output in Switzerland, for example, was
72 percent of the United States in 1970. By 1974,
Swiss per capita output exceeded that of the United
States, standing at 115 percent of that of the United
States. Such a reversal in the comparative position is
hard to accept, however, because of what it implies
about the relative growth rates of per capita output
over the 1970-1974 period. The conventional measure
implies that per capita output in real terms in Switzer
land has grown at the annual rate of 14.9 percent over
this period while U.S. per capita output has grown
at a 2.3 percent annual rate. Per capita output in
18Contrary to the pattern observed in Table IV, the conven
tional measure using the exchange rate is higher after 1970
than the PPP-based measure in most cases. However, as
noted in footnote 16, there is no necessary reason for the
conventional measure to fall on either side of the PPP-based
measure. In this paper, there is no discussion of the factors
producing this difference in pattern. It can be noted, how
ever, that key factors explaining the difference are ( a ) a
steady increase over the period in the number of convertible
currencies; ( b ) varying degrees of intervention in the foreign
exchange markets; ( c ) varying degrees of constraints on
goods and capital movements; and (d ) the changing role of
the U.S. dollar as a reserve currency.

FEDERAL RESERVE BANK OF ST. LOUIS

Switzerland, in fact, has grown even slower than that
of the United States over this period, at 2.1 percent
per year compared to 2.3 percent. Such an anamolous
result holds in general. Per capita output in Sweden,
for example, has grown at the same rate as that for
the United States over the same (1970-1974) period.
The conventional measure, however, shows that per
capita output in Sweden has increased substantially
(close to 20 percent) relative to that for the United
States.
Comparison of estimates derived from the more
appropriate PPP-based measure does not support the
claim that the U.S. lead in per capita output has been
overtaken in the 1970s. For example, whereas the
conventional measure for Sweden reached its high of
118 in 1975, the PPP-based measure was only 80 per
cent of U.S. per capita output. The conventional meas
ure also placed Canada ahead of the United States in
1976; however, its PPP-based measure was below
that of the United States, at 91 percent. As noted in
Table V, PPP-based estimates of per capita output
for Switzerland are not available. However, in view
of the low growth rate of per capita output in Switz
erland relative to the United States since as far back
as 1960 (1.98 percent vs. 2.36 percent), when the
U.S. lead was judged to have been substantial (see
Table IV ), it is not likely that Switzerland had over
taken the lead by 1976.

CONCLUSION
The allegation that the United States has been over
taken in its per capita output by a number of indus

MAY

1978

trialized countries in the 1970s is based on an ex
change rate-based measure which has been shown to
be unreliable. When estimates of the more appropriate
PPP-based measure are constructed, the allegation is
not supported. That does not mean that one should be
complacent about the performance of the U.S. econ
omy. But it denies the factual basis for the claim that
“[rjelative to achievements in the rest of the world,
the U.S. economy no longer ‘delivers the goods.’ ”19
Although the United States has yet to lose its lead
in per capita output, the estimates of PPP-based
measures through 1976 indicate that the U.S. lead
has indeed narrowed since 1950. One could cite many
contributing factors to this development — such as the
imperatives for reconstruction provided for some
countries by the ravages of World War II, and the
opportunities for adapting available production tech
nologies.20 However, the fact that the U.S. lead has
been narrowed does not necessarily mean that it is
bound to be overtaken in the future. The future is
not necessarily an extrapolation of past trends. The
comparative levels of U.S. per capita output in the
future will be determined solely by this country’s
relative success in harnessing opportunities for growth
in per capita output.
19Thurow, “The Myth.”
20See Edward F . Denison, “The Contribution of Capital to the
Post-War Growth of Industrial Countries,” in U.S. Economic
Growth From 1976 to 1986: Prospects, Problems, and Pat
terns — Volume 3 Capital, Studies prepared for the use of
the Joint Economic Committee, 94th Cong., 2nd Sess., No
vember 15, 1976.

Page 15

Rising Farmland Prices and Falling Farm
Earnings: Is Agriculture in Trouble?
N EIL A. STEVENS

U.S. farm real estate values have risen rapidly in
recent years.1 In the five-year period from 1972 to
1977, the average price of an acre of U.S. farmland
increased 114 percent, or at an annual rate of 16.5
percent. This rate of gain for farmland is the most
rapid for any five-year period in this century and
compares with a 6 percent annual rate of increase
in the thirty years ending in 1972. At a time when
some alternative investments, notably many common
stocks, have performed less spectacularly, and when
assets which offer a hedge against inflation are
highly desired, farm real estate has been very at
tractive. In fact, ownership of farmland has provided
more than simply a hedge against inflation as in
creases in farmland values have substantially out
paced the rate of general price inflation.
These increases in farmland values have brought
about substantial gains in the wealth of landowners
who made land purchases prior to, or in the early
stages of, the present farmland price boom. Yet,
while most landowners have become wealthier, lower
farm commodity prices and sharply higher costs of
production in recent years have substantially de
pressed farm earnings from the level of three to four
years ago. These conditions, in turn, have led to sub
stantial cash flow problems for those farmers who
financed purchases of land at the elevated prices of
recent years and who must continue to meet periodic
large interest payments as well as other fixed costs.
Such a financial “squeeze” apparently underlies the
recent farmers’ strike movement.

to changes in its price, especially in the near term,
changes in the price of land reflect primarily changes
in the demand for the services it provides. In addi
tion to agricultural uses, land is demanded for many
other uses including residential, industrial, commer
cial, and recreational.
The price of most agricultural land, however, re
flects primarily the value of the agricultural prod
ucts that it can produce. That is, the demand for
farmland is derived primarily from its role as an
input into the production process for food and fiber.
Thus, the earnings accruing to farmland are affected
by numerous supply and demand factors affecting
agricultural products. Among these factors are ag
gregate incomes in the economy, population, export
demand, prices of nonagricultural goods, prices of
non-land inputs into agricultural production, agricul
tural technology, and government farm programs.2
Farmland is a durable asset, yielding a stream of
services, or earnings, over time. Consequently, the
price of farmland, although influenced by current
earnings, reflects the stream of earnings which are
expected over its life.:i Investors must, therefore, ana
2For a summary discussion of the outlook for agriculture in
1978, see Clifton B. Luttrell and Neil A. Stevens, “Outlook
for Food and Agriculture,” this Review (January 1 9 7 8 ), pp.
15-22. Fo r a discussion of the problems and effects of Gov
ernment price supports on agriculture, see Clifton B. Luttrell,
“Farm Price Supports at Cost of Production,” this Review
(Decem ber 1 9 7 7 ), pp. 2-7.
3In investment theory terminology, the stream of earnings is
discounted or capitalized in order to determine its present
value. The present value ( P ) of a constant stream of earnings
i

Determ ination of Land Values
Land is an asset which is relatively fixed in sup
ply. Since the quantity of land is not very responsive
'F arm real estate values and farmland values are used inter
changeably. Farm real estate values is the more appropriate
term for the U.S. Department of Agriculture data used in this
paper since it includes the value of buildings and other per
manent structures on the land.
16
Digitized forPage
FRASER

-r.

can be written as P =

1^°

jq j.

+ ••• +

(l+i)n

where E 0 is net earnings, i is the opportunity cost of credit,
and n is the number of periods over which the earnings are
expected. This formula can be written in shorthand form as
t
£
2
------"— . For example, earnings of $100 per year for the
n=l (l+i)n
next ten years discounted at a 5 percent interest rate is worth
approximately $772 today. When n becomes very large, the
T7

formula reduces to the simple formula P =

~ y . When a

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

lyze the demand and supply factors which can in
fluence the future income stream of the asset and
form some judgment as to the probable pattern of
that income stream. If, for example, population
growth is expected to increase from 2 percent a year
to 3 percent a year, investors would probably raise
their expectations of future earnings from farmland,
other things equal, and the price of farmland would
be bid up immediately to incorporate this change.4

Farm land Values and Earnings
T h e Historical R ecord

—

Farmland has been a “good” investment over the
past thirty-five years in the sense that the average
value of an acre of U.S. agricultural land has increased
more rapidly than the rate of inflation. Farmland
values, deflated by the GNP implicit price deflator,
rose at an average rate of 2.6 percent per year from
1942 to 1972. Beginning in 1972 farmland values ac
celerated sharply, rising almost 9 percent per year
faster than the general inflation rate from 1972 to 1977
(Chart I).
Farmland investments have also performed well
when compared with most common stock invest
ments. In the past ten years, farm real estate values
in nominal terms have advanced at an 11 percent an
nual rate while common stock prices, as measured by
the Standard and Poor’s 500 Index, have remained
about unchanged.5 In the previous twenty years com
mon stock prices rose at a 9.4 percent rate, compared
with a 5.2 percent rate for farmland prices. Such a
divergent pattern for these two types of investments
is symbolic of a substantial shift in investors’ expecstream of returns is expected to increase g percent per year,
the formula can be rewritten as

Z
E "( ^
When n
n= l
(l+i)»
becomes very large and i is greater than g, the formula bep
comes P0
—

4For studies which have attempted to determine which factors
underlie real estate movements, see Marvin Duncan, “Farm
Real Estate Values — Some Important Determinants,” Federal
Reserve Bank of Kansas City Monthly Review (M arch 1 9 7 7 ),
pp. 3-12; Luther G. Tweeten and Ted R. Nelson, “Sources
and Repercussions of Changing U.S. Farm Real Estate
Values,” Technical Bulletin T -120, Oklahoma State University
(April 1 9 6 6 ); Robert W . Herdt and Willard W . Cochrane,
“Farmland Prices and Farm Technological Advance,” Journal
of Farm Economics (M ay 1 9 6 6 ), pp. 243-63.
“While different growth rates were observed for the two assets,
the rates of return on investments are not necessarily differ
ent. In fact, over the longer term, rates of return for assets
tend to equalize when differences in risks are taken into ac
count. If, for example, the rate of return of a particular asset
is substantially above that of other investments, investors will
tend to switch into the higher-yielding assets, thereby bidding
up the price of the asset relative to that of other assets.

L l_ N o m in a l farm real e state v a lu e s a re d e fla t e d b y the G N P im plicit p ric e d eflato r.

tations about the future earnings of these two
investments.
Historical data on earnings from farmland as a sep
arate factor of production are not readily available.
However, a measure of earnings on total farm assets
is computed by the U.S. Department of Agriculture
and can be used here as a proxy measure for farm
land earnings.6 As shown in Chart II, these earnings
on farm production assets (in nominal terms) have
trended upward since 1950. Especially noticeable is
the sharp rise in earnings in the early 1970s when
earnings from farm assets rose 200 percent from 1971
to 1973, then subsequently fell. Earnings adjusted
for changes in the general price level also increased
sharply in the 1971-73 period, but by 1977 real earn
ings were only 15 percent above the 1971 level.

An Analysis of Price and Earnings
The ratio of the value of farm production assets
to earnings on these assets is a useful tool for analyz
ing the behavior of farm earnings and farm real
estate values. This ratio, similar to the price-earnings
ratio used in stock market analysis, is a measure of
8Farm real estate comprised approximately 80 percent of all
farm production assets in 1977. Earnings on farm assets are
total net income of farm operators from farming plus net rent
to nonfarm landlords and interest on farm debt, less adjust
ments for farm operators’ labor and management.
Page 17

FEDERAL RESERVE BANK OF ST. LOUIS

MAY

1978

however, the ratio has continued to rise and by 1977
the value of farm assets was estimated to be 31
times current earnings, well above the average ratio
of the 1950s and 1960s.
One interpretation of this most recent rise in the
P/E ratio is that investors revised upward their ex
pectations concerning long-run prospects for future
eamings on farm assets following the surge in eam
ings in 1972 and 1973. An acceleration in real estate
prices would accompany such an upward revision in
eamings as these expectations are capitalized into cur
rent land prices. In this case, the recent shortfall in
farm earnings would appear to reflect an investor
view that such low eamings were a temporary
phenomena.

S o u rc e : U.S. D e p a rt m e n t o f A g ric u ltu re
LL A n n u a l a v e r a g e o f 1 9 5 0 -5 4 p e rio d is 6 6.4
12 A n n u a l a v e r a g e o f 1 9 5 0 -5 4 p e rio d is 2 0.6

the confidence that investors have in the future earn
ings of one investment relative to another. For ex
ample, the higher the price-earnings ratio (P/E),
the greater is the expected growth rate of earnings
from the current level and the more certain investors
are that a given stream of earnings will be realized.7
As shown in Chart II, earnings on farm production
assets and the value of these assets generally trended
upward together in the 1950s and 1960s. Fluctuations
in the P/E ratio occurred, but the price of farm assets
averaged 26 times earnings from 1950 to 1971. The
P/E ratio fell sharply in 1972 and 1973, but has risen
significantly since then.
These recent movements in the P/E ratio reflect
the unusual pattern of earnings in this period. The
sharp rise in earnings in 1972 and 1973 was not im
mediately and completely incorporated into future
expectations. Thus the P/E ratio fell to an abnormally
low level when compared with historical values. Earn
ings subsequently began falling and the P/E ratio
began rising. Instead of reestablishing a value at
around the average ratio of the 1950s and 1960s,
7The greater the degree of certainty the less is the return re
quired to compensate for risk. The investors’ perception of risk
may be based, for example, on the variability of eamings.
To compare P /E ratios among alternative investments, the
useful life of the investment must be similar since assets with
a longer life span will have higher P /E ratios than shorterlived investments.

Page
18

The fact that farmland prices relative to other
prices have grown at an accelerated rate beginning
in 1972 strongly suggests that investors’ expectations
of eamings, indeed, have increased in real terms.
Because of higher eamings expectations, investors
have bid up the price of farmland. This has meant
substantial wealth gains to landowners, but the rate
of return to new farmland owners depends upon the
correctness of these expectations of higher eamings
which have been incorporated into land values. Should
these expectations be revised downward in the market,
thus leading to a general decline in farmland values,
new owners will either have to sell their land at a
loss or continue to farm the land at a lower rate of
return than was anticipated when they made their
original investment.
Cash flow problems can develop when asset values
are bid up considerably above that level which is
consistent with current earnings. When purchases are
heavily financed, as is usually the case for most farm
land purchases, large interest payments, as well as
other fixed costs, must be covered by current earn
ings, unless other sources of income are available.
Farmers who borrowed heavily to purchase land or
who have borrowed on the increased market value
of their land and who do not have other sources of
income will experience financial trouble when realized
earnings are considerably below the level anticipated
at the time of purchase.
The higher expectations for farm earnings reflected
in the recent upsurge in land prices may be traced
back to the 1972-73 period when farm commodity
prices and farm asset earnings rose dramatically. The
sharp rise in eamings in this period reflected changed
supply and demand conditions for agricultural prod
ucts. Unexpected sales of wheat and feed grains to
the Soviet Union in mid-1972 served to reduce domes

FEDERAL RESERVE BANK OF ST. LOUIS

tic stocks and increase prices. A sharp decline in the
production of Peruvian fish meal led to a shortfall
in world protein supplies and an unanticipated in
crease in export demand for soybean meal. The un
expected decline in world crop production in 1972
and a realignment of world currencies led to large
increases in export demand for U.S. crops. U.S. farm
exports rose from about 15 percent of farm commod
ity sales in 1971 to 25 percent in 1975. In addition
to sharply increasing export demand, domestic de
mand for food in the early 1970s was boosted by
increases in government food assistance programs
such as the food stamp program. Also, a number of
factors adversely affected U.S. agricultural production
in the early 1970s, such as wage-price controls, envi
ronmental regulations, sharp increases in energy prices,
and a drought-induced shortfall in U.S. crop produc
tion in 1974.
C h a rt III

Price Trends
ANNUAL
-D A T A -

Q U ARTERLY
---- D A T A ------

Farm Product Prices;

Consumer Price I
- T ( A I I Items)

1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978
So u rc e : U.S. Departm ent of L a b o r a n d U.S. D e p a rtm e n t o f Agricu ltu re

All of these factors contributed to a sudden, large
increase in agricultural earnings. Most of the factors
behind this bulge in earnings were temporary as is
indicated by the decline in earnings and farm com
modity prices in recent years. As shown in Chart III,
farm commodity prices rose much faster than con
sumer prices from 1971 to 1973, but have since come
back in line with the general price level in the
economy. World grain production, aided by more
favorable weather, has increased 14 percent since
1972-73, and oilseed and meal production across the
world has risen 38 percent. U.S. crop production also
snapped back from the 1974 disaster as weather con
ditions normalized. In addition, funding for domestic

MAY

1978

food assistance programs has levelled off and many of
the factors, such as mandatory wage-price controls,
which had disruptive effects on agricultural produc
tion in the early 1970s have disappeared in recent
years.
If the bulge in earnings in the early 1970s was
mostly temporary, then current earnings may not be
too far out of line with longer-run supply and de
mand forces. To the extent this bulge was the basis
for the upward revision in investors’ expectations
about farmland earnings, then farm assets, and farm
land in particular, have become overvalued on the
basis of fundamental supply and demand conditions
in the market for agricultural products.8

Implications
The farmers’ strike movement has brought consider
able attention to the current “low” earnings in agri
culture. One might infer from this movement that
U.S. agriculture is in trouble and is near widespread
bankruptcy. However, the fundamental factors affect
ing U.S. agriculture appear relatively sound. The
health of U.S. agriculture is heavily dependent on
its ability to compete effectively in world markets.
U.S. agriculture is very efficient and enjoys a com
parative advantage in trade with most countries of
the world.
While the long-run prospects appear sound for
U.S. agriculture, “hard times” may be experienced by
some farmers who possibly made incorrect decisions
based, perhaps, on misinterpretation of the bulge in
earnings in the early 1970s. Cash flow problems have
already developed for some farmers. Should earnings
not rise in accordance with the expected earnings now
built into land values, then agricultural land values
will decline. Continued “low” earnings, if maintained,
would eventually prompt a change in expectations by
investors since farmland must compete with other in
vestments. If farmers and other investors in farmland
begin to doubt the future prospects for earnings
growth, they will lower their bid prices for farmland
coming into the market or attempt to sell land in order
to take advantage of higher-yielding investment oppor
tunities elsewhere. Farmland values would then de
cline until the return on farmland has risen to a com
parable level with returns on alternative investments.9
8Based on 1977 data, either a 20 percent rise in earnings or
a 17 percent fall in the value of farm assets, or some com
bination, would be necessary to reestablish the average ratio
between farm assets and earnings which prevailed in the
period from 1950 to 1971.
9The rate of increase for nominal farm real estate values has
shown a tendency to slow in the past year. From February
Page 19

If land values do decline, owners of farmland will
experience losses in wealth. For most, this would
simply reduce some of the gains experienced as prices
rose. But those farmers who bought at the higher
prices of the past few years will realize a lower rate
of return on their initial investment than they ex
pected, and some who are highly leveraged may be
forced to leave agriculture, and the rate of bank
ruptcy might increase for a time.
Yet, equity in agriculture is large. In fact, the
ratio of farm real estate debt to the value of farm
real estate has actually fallen in recent years, from
1977 to February 1978, the average price of farmland rose
about 9 percent compared with a 16.5 percent increase per
year in the previous five years.

about 14 percent in 1971 to about 11 percent in 1977.
It would appear that most farmers would be able to
weather some decline in land values without incur
ring bankruptcy. In recent years over one-half of all
farmland transfers have been to existing owner-operators of farms, where equity is often substantial in
existing acreages, so cash flow problems are not as
severe for these farmers.
In the final analysis, the health of the agricultural
industry reflects its efficiency in producing food and
fiber products and the level of demand for these
products. While investors’ expectations determine the
value of farm real estate as well as other investments,
these expectations cannot stay out of line with the
fundamental supply and demand conditions for these
investments for very long.