The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
april
Paying for Social Security:
Is It Time to "Retire" the Payroll Tax?
The Tax Bite:
How the U. S. Stacks Up
How Reliable Are Those
Price and Employment Measures?
FEDERAL RESERVE BANK of PHILADELPHIA
business iw ieir
IN THIS ISSUE . .
Paying for Social Security: Is It Time
To "Retire" the Payroll Tax?
. . . The Social Security payroll tax has a
number of undesirable aspects, many of
which could be eliminated or reduced
by implementing alternative methods of
financing.
The Tax Bite: How the U. S. Stacks Up
. . . Compared to their foreign counter
parts, individuals and corporations in the
U. S. tote a heavier income-tax burden, yet
as a share of national income the overall
tax take is considerably smaller.
How Reliable Are Those Price and
Employment Measures?
. . . The Consumer Price Index and the
unemployment rate— the most commonly
used economic barometers— have limita
tions and should be regarded as only rough
gauges of trends in the economy.
On our cover: The Philadelphia Museum of Art is situated at the extreme upper end of the
tree-lined Benjamin Franklin Parkway, near the east bank of the Schuylkill River and the south
ernmost end of Fairmount Park. This museum, the chief repository of art in Philadelphia, is
the architectural product of Zantzinger & Borie and Horace Trumbauer. Opened in 1928, the
neoclassical structure houses exhibits presenting a comprehensive view of the history of art
from ancient times to the present.
(Photograph courtesy of the Philadelphia Convention and Tourist Bureau.)
BUSINESS REVIEW is produced in the Department of Research. Ronald B. Williams is Art Director and Manager,
Graphic Services. The authors will be glad to receive comments on their articles.
Requests for additional copies should be addressed to Public Information, Federal Reserve Bank of Philadelphia,
Philadelphia, Pennsylvania 19101.
FEDERAL RESERVE BANK OF PHILADELPHIA
Paying for
Social Security:
Is It Time to "Retire"
The Payroll Tax?
By Donald ). Mullineaux
Many people mistakenly perceive Social
Security as insurance against future loss of
earning power. In fact, the system is simply
a method of making transfer payments from
the young to the old. There is no firm con
tractual relationship between what you pay
today and what you get tomorrow. Recog
nizing payroll deductions as "taxes" rather
than "insurance premiums" means that the
benefits and costs of Social Security can be
discussed separately. In particular, the
method chosen to finance this socially man
dated program— the payroll tax— can be
compared with other forms of taxation
which might accomplish the same ends
more equitably and efficiently.
In the last five years, the effective tax rate
for Social Security programs has increased
over 18 percent. Since most Americans ac
cord tax increases and bubonic plagues the
same order of welcome, a public outcry
might well have been expected to echo
throughout the halls of Congress. Yet, until
very recently, nary a whisper of complaint
was heard.
Why should taxpayers prove so tolerant of
increased government demands from pay
rolls to support retired and disabled work
ers and their survivors? After all, since the
days when Robin Hood and his band of
merry men popularized compulsory trans
fer-payment schemes, these systems have
typically been a ssaile d by those "called
upon" to give— regardless of the apparent
virtue of the cause. Social Security probably
has proved exempt from such a reception
because of the public's fundamental miscon
ception of the nature of the program.
THE FINANCIAL DEMANDS OF SOCIAL
SECURITY: A GROWING BURDEN FOR
THE PAYROLL TAX
The Old-Age, Survivors, Disability and
Health Insurance (OASDHI) program is a
3
APRIL 1973
BUSINESS REVIEW
pal source of income. Benefits per recipient
have increased from $157.65 a year in 1940
to $1,715.43 in 1972, or over 1000 percent.2
Since prices have increased about 200 per
cent over the same period, the "real" level
of benefits has risen substantially.
As most workers know, Social Security is
financed through compulsory payroll-tax
deductions. Currently, workers and employ
ers each pay 5.85 percent on the first $10,800
earned each year. Because of increases in
both the tax rate and the wage base to which
product of the Social Security Act of 1935.
Initially designed to guarantee a minimum
level of income support to retired industrial
and commercial workers, the program now
embraces almost all workers1 and their de
pendents and covers the additional contin
gencies of illness in old age, death, and
disability (see Box). Today, one in every
seven Americans receives a grayish-green
Social Security check each month, and for
many recipients this represents their princi1Those workers currently exempted from manda
tory participation include clergy, employees of non
profit organizations, state and lo c a l g o v e rn m e n t
workers, and Federal employees under the Civil Ser
vice Retirement System.
2The maximum pension for a couple in 1940, the
first year benefits were paid, was $820.80 annually.
In 1972, the maximum benefit was $4,668 or $389 a
month.
W HY HAVE A SOCIAL SECURITY PROGRAM?
The establishment of Social Security and growth in the program are no doubt
interpreted by many as a reflection of the humanitarian values of American society.
However, a compulsory pension system also makes good sense from an economic
(resource-saving) viewpoint. Left to their own devices, those individuals who would
fail to provide sufficiently for retirement would not be the only ones to suffer the
cost of such improvidence. Having become "public charges," society would be re
quired to support them. These "social costs" can be reduced by requiring individu
als to set aside a minimum level of current income for retirement.
Although "neighborhood effects" suggest an economic rationale for compulsory
pensions, the argument does not demand that the required portion of retirement
insurance be provided by the government rather than the private sector. The na
tionalization of the program is probably explained by the fact that Social Security
was one of several New Deal programs such as the Agricultural Adjustment Act, the
Works Progress Administration, and the National Recovery Act, all of which involved
government intervention in private-sector activities in an attempt to combat the mis
ery and economic dislocation caused by the Great Depression. Several economists
have suggested that transferring the program to profit-motivated private firms would
increase efficiency in the retirement-provision process.* However, Social Security
may well be the most popular political program since biblical times when Joseph
dispensed wheat from government storage to starving Egyptians and Israelites.
Clearly, the result of many more studies would be required to convince the public
of the wisdom of shifting Social Security to private hands. Thus, most critics of the
system have suggested reforms within the framework of continued government
operation.
*See, for example, Milton Friedman, Capitalism and Freedom (Chicago: University of Chicago Press,
1962), chap. 11.
4
FEDERAL RESERVE BANK OF PHILADELPHIA
level of support for increasing numbers of
retired workers in an economy where the
cost of living trends ever upward. On the
other hand, the program also has grown in
response to social concern with the problem
of destitution among the aged. (See Box
on multiple goals for a discussion of why
these objectives may conflict and how they
might be reconciled.) Reformers contend
that if the current system is restructured to
eliminate the problem of multiple goals, the
compulsory pension program can grow
more slowly and the financial stresses will
prove less severe. Such reforms probably
lie considerably in the future, however, and
even if implemented, will not solve many of
the problems associated with the payroll tax.
it applies, Social Security taxes have risen
substantially relative to national income in
recent years (see Table). Payroll taxes are
currently the second-biggest revenue gener
ator for U. S. Treasury coffers, personal in
come taxes being the largest.
Recent changes built into the Social Se
curity setup insure the continued growth of
the program. Present law provides that after
1974, the wage base subject to tax will auto
matically be adjusted upward as wage levels
rise.
Benefits, too, will be raised without prior
Congressional approval whenever the Con
sumer Price Index goes up 3 percent. If
wages rise by 5 percent annually and infla
tion averages 2.75 percent, a couple at re
tirement age (65) in 2011 can expect a
pension of nearly $33,000 a year (about
$11,500 in 1972 dollars).
Observers have in te rp re te d this rapid
growth of Social Security as reflecting con
fusion concerning the objectives of the pro
gram. On the one hand, Social Security must
expand to continue to provide a minimum
CAN THE PAYROLL TAX CONTINUE TO
"END-RUN" THE TAX REVOLT?
Since 1949, aggregate payroll deductions
for Social Security have grown like rabbits
in a lettuce patch. Overall tax collections
have increased at a 16 percent annual rate
and the maximum tax per worker has gone
A GROW ING PERCENTAGE OF NATIONAL INCOME
GOES TO SOCIAL SECURITY TAXES
BECAUSE OF BOTH RATE AND BASE INCREASES
Year
1937
1947
1957
1967
1972
1973
Source:
Employee's
Portion
Total Social
of the
Total Social
Security Taxes as
Social Security Maximum Wage Security Taxes as Percent of All
Tax Rate
Base Subject
Percent of
Federal Tax
(Percent)
to Tax
National Income
Receipts
1.0
$ 3,000
1.0
10.9
1.0
3,000
0.8
3.6
2.25
4,200
2.1
9.2
4.40
6,600
4.4
18.9
5.20
5.2
9,000
21.5
5.85
10,800
5.8 (est.)
25.0 (est.)
U. S. Department of Health, Education and Welfare;
U. S. Department of Commerce
5
APRIL 1973
BUSINESS REVIEW
RECONCILING MULTIPLE GOALS:
SOCIAL SECURITY AND THE NEGATIVE INCOME TAX
Some students of Social Security contend that the program suffers from a basic
flaw since it tries to satisfy two goals with a single policy instrument. On the one
hand, it attempts to maintain the livelihood of the aged poor, while on the other
hand, it assures all eligible recipients of an income supplement which is purport
edly related to their previous standard of living. Focusing on the former goal sug
gests it is wasteful to hand out money to those already well-off, while if the lat
ter is emphasized, the program will not lift low-income old people out of poverty.
Students of policymaking are well aware that conflicts among goals can only be
resolved through the use of additional policy instruments. Thus, both of the above
objectives might be accomplished more efficiently if the antipoverty goal were as
signed to an income-transfer scheme such as the negative income tax and the pen
sion-provision goal were attacked through a revised Social Security program.
The negative income tax is a device for making transfer payments to low-income
families. Individuals whose income falls below some "break-down" level would
"pay" taxes at a negative rate. In other words, they would receive payments from
Uncle Sam rather than sending him a check each April. The size of the payment
would be determined by applying the negative tax rate to the amount by which
their income falls short of the "break-even" level. For example, if the "break-even"
point is $4,000 and the negative tax rate is 50 percent, a family with income of
$1,000 will receive $1,500 as a negative tax payment.
Although there are some practical problems, the Social Security Administration
could manage the negative income tax for all aged persons. Those ineligible for
Social Security could qualify for payments under this scheme and the cost of their
benefits could be paid from the general fund of the Treasury. Recent estimates in
dicate the cost of a negative income tax to the aged poor would be relatively small
compared with the total cost of a comprehensive negative income tax. However,
because of the sweeping nature of the proposal and the administrative detail in
volved, it is usually suggested as a longer-run reform of the Social Security program.
from $60 in 1949 to $936 in 1972. Last year's
Social Security tax hike constituted the larg
est Federal tax increase ($7 billion) since the
Korean War as Congress settled on a $1,404
maximum figure for 1974. This event might
have proved less surprising had it not oc
curred in the midst of a much-publicized
"taxpayers' revolt."
many taxpayers believe that (1) payroll-tax
deductions represent "insurance premiums"
for vested benefits to be paid during retire
ment and (2) the tax rate for Social Security
is low in comparison with other taxes. Both
of these views are in fact misleading. As the
nature and effects of Social Security "con
tributions" become more widely recognized,
payroll-tax increases are likely to be vigor
ously debated in parlors, pubs, and parlia
mentary chambers.
Social commentators and economists of
ten explain the apathetic public response to
Social Security tax increases by noting that
6
FEDERAL RESERVE BANK OF PHILADELPHIA
Insurance? No—Taxes? Yes. Some years
ago the Social Security Administration told
Americans that:
currently retired. Private insurance firms,
however, must accumulate “ reserve funds"
to meet future obligations because their ex
pected cash flows are uncertain.4 Thus, So
*
cial Security taxes are determined by the
level of benefits currently paid, while insur
ance premiums depend on the desired level
of future benefits.
Recognition of Social Security as a taxfinanced system for making transfer pay
ments rather than a process of purchasing
claims of future income permits a com
parison of the program's current financing
p ro ced u res w ith other methods of ac
complishing the same thing. Such a study
suggests the payroll tax possesses some un
desirable aspects relative to other types of
taxes which make its true burden consider
ably greater than appears on the surface.
As these liabilities become more widely rec
ognized, payroll-tax increases will no doubt
become a tougher row for Congress to hoe.
Your account number on your Social
Security card identifies your old-age
and survivors insurance account. Your
card is the symbol of your insurance
policy under Federal Social Security
law.
The Administration no longer uses this lan
guage, and the public is none the worse for
it. While private insurance and Social Se
curity have some elements in common, the
concepts differ substantially. For example,
the benefits from private insurance are pre
determined in a firm contractual relation
ship and depend on the “ premium" paid.
Social Security benefits, however, are subject
to the whim of Congress and historically
have proved only tenuously related to ex
penses. In addition, the U. S. Supreme Court
has decided that Social Security recipients
have no “ property rights" in the system, but
rather possess only a moral claim which
Congress could theoretically deny.3 Finally,
participation in Social Security is compul
sory, while private insurance purchases are,
of course, voluntary.
Clearly, Social Security payments more
readily satisfy the definition of a tax— a
forced contribution of income to meet pub
lic needs— than that of an insurance pre
mium. In fact, since payroll deductions are
taxes rather than voluntary contributions,
the program can be financed on a “ pay-asyou-go" basis. Current payroll taxes are
used to cover the costs of benefits for the
The Burden of the Payroll Tax. When
compared to the minimum personal income
tax rate of 14 percent, the 5.85 percent So
cial Security tax on wages hardly seems bur
densome. Such a c o n clu sio n ignores a
number of factors, however. The most im
4 As originally conceived, OASDHI payments were
to be made from an accumulated reserve fund, but
amendments to the Social Security Act in 1939 aban
doned the reserve fund approach for "cash” or "payas-you-go” financing. This allowed pensions to be
paid to the first crop of retirees in 1940, despite the
fact they had paid practically nothing into the sys
tem. Under the present arrangements, payroll taxes
are transferred from the Internal Revenue Service to
OASDHI trust funds from which disbursements are
consequently made. If taxes and interest on trustfund investments (Treasury securities) exceed bene
fits paid, the size of the trust fund increases. The
purpose of maintaining this reserve is apparently to
provide for continuity of benefit payments if current
tax receipts decline sharply because of a recession.
Reserves in the trust fund are presently to be kept
high enough to cover a total of nine months in
benefits.
3 In Fleming v. Nestor, 363 U. S. 603 (1960), the
Supreme Court upheld the constitutionality of a pro
vision which prohibits Social Security payments to
persons deported for subversive activities. The ma
jority opinion declared:
The noncontractual interest of an employee cov
ered by the Act cannot be soundly analogized to
that of the holder of an annuity, whose right to
benefits are based on his contractual premium pay
ment.
7
APRIL 1973
BUSINESS REVIEW
Social Security taxes also fall forcefully on
multiple-worker families. A working hus
band and wife earning $10,800 each will pay
over $2,500 in OASDHI taxes, while a wage
earner with a nonworking spouse making
$21,600 will pay only half as much. Yet de
pending on future earnings, the two couples
in this example might well draw the same
amount of Social Security benefits. Since the
public files no payroll-tax returns, none of
these inequities are highly visible.
While Social Security taxes are clearly re
gressive, low-income retirees receive larger
benefits relative to income than their wellto-do counterparts. The entire program is
therefore usually considered progressive.
Nevertheless, a part of this income redis
tribution is a transfer from the youthful poor
to the elderly poor. By operating in this
way, the system makes it more likely that
today's young poor family will be tomor
row's old poor family. It does so by leav
ing the youthful poor with fewer dollars to
invest in their own earnings capability ("hu
man capital") or that of their children.
All of these arguments suggest that the
present method of financing Social Security
conflicts with both (1) the generally ac
cepted principle that taxation should be
based on ability to pay and (2) society's de
cision to attempt to eradicate poverty as
expressed in the waning War on Poverty.
Thus, consideration of alternative financing
schemes is merited.7
portant is that many economists believe that
workers also bear the employer's portion of
the tax in the form of foregone wages. For
example, Milton Friedman has written:
[The employer's portion] . . . isn't paid
by the employer; it is, in effect, paid
by the wage earner. It is part of his
wages that is sent to Washington in
stead of going to him.5
If this view is valid, the effective payroll-tax
rate is of course 11.7 percent in 1973, rather
than 5.85 percent.6
Another factor boosting the burden of
payroll taxes is that no deductions or exemp
tions are permitted. Given the ceiling on
taxable wages, this makes the payroll tax
highly regressive. Such taxes take a larger
piece of the poor man's muffin than the rich
man's pie. For example, a worker earning
less than $10,800 pays 11.7 percent of his
wages (if labor bears the full burden of the
tax), while an executive earning $100,000
pays only about 1.25 percent for Social Se
curity. A family of four earning $10,800
will pay a higher combined rate (Social Se
curity and income tax) than a family earning
$20,000. Similarly, a family of six earning a
poverty-range income of $5,000 typically
pays no income tax, yet surrenders $585 to
OASDHI trust funds.
"'Milton Friedman, "Transfer Payments and the So
cial Security System," Conference Board Record, Sep
tember 1965, p. 8.
7The payroll tax has a number of other poten
tially undesirable aspects. For instance, it has little
stabilizing effect on the economy, and, in fact, illtimed increases have sometimes proved destabiliz
ing. In addition, to the extent that employers bear
some portion of the tax, it may induce firms to sub
stitute capital for labor or high-wage employees for
low-wage workers. These artificially induced substi
tutions will result in smaller output or higher prices,
as well as additional unemployment. However, econo
mists believe this is not a serious problem because
labor bears the full burden of the tax. Thus, the
discussion has focused on the equity aspects of the
payroll tax.
fl Economists use the term "incidence" when ana
lyzing the way the burden of a tax ultimately gets
borne. For statistical evidence supporting the view
that labor bears the full burden of the employer's
portion of the payroll tax, see John A. Brittain, The
Payroll Tax for Social Security (Washington: Brook
ings Institution, 1972). Martin Feldstein criticizes
Brittain's analysis in "The Incidence of the Social Se
curity Payroll Tax: A Comment," American Economic
Review 72 (1972): 735-38. In the following discus
sion, workers are presumed to bear the burden of
both portions of the tax.
8
FEDERAL RESERVE BANK OF PHILADELPHIA
a single ceiling were applied to pooled
family income. The most comprehensive
reform, therefore, would appear to be re
placement of the payroll tax by the income
tax. In 1969, a complete substitution of in
come for payroll taxes would have required
a boost in income-tax yield of about 45
percent. One economist has shown that this
could be accomplished by restructuring ex
isting rate schedules to fall less heavily on
low-income earners (relative to the com
bined Social Security-income tax rate), and
generally more heavily on incomes above
the ceiling on taxable wages.8
Proposals to shift the financing of Social
Security to general revenue funding have
drawn fire from several quarters. Business
groups have opposed the idea on the
grounds that financing benefits out of ear
marked taxes has tended to temper Con
gress's generosity. The more likely case,
though, is that misconceptions about the
nature of the payroll tax have made it easier
for Congress to expand the program. Some
economists have also resisted such a change,
arguing that foreign countries which use
general revenue financing usually have the
lowest levels of benefits. While this is true,
it should be remembered that the purpose
of the program is not to maximize benefits,
but to fix them at levels which satisfy the
basic objectives of the program. No evi
dence has been presented that suggests that
general revenue financing will prove unable
to satisfy this objective.
MOVING TO GENERAL
REVENUE FINANCING?
Many critics of the payroll-tax method of
financing Social Security have suggested that
OASDHI trust funds draw upon the general
funds of the U. S. Treasury to meet their ob
ligations. In effect, this would mean inte
grating income and p ayro ll ta xa tio n , or
substituting the former for the latter. Several
different means of easing the distributive
burden of the payroll tax in this manner have
been advocated.
Integrating the Payroll and Income Tax.
According to this plan, individuals receive
credit against their income taxes for each
dollar of Social Security taxes paid. Payroll
deductions would be counted as withhold
ings for in co m e-tax purposes, and the
amount exceeding the total income-tax lia
bility would be refunded. Partial integration
could be achieved by allowing taxpayers to
count only a portion of payroll taxes as in
come-tax withholdings. A melding of pay
roll and income taxes would, of course,
result in a loss of government revenues, so
that unless a tax cut were in order, incometax rates would have to be increased to
maintain the flow of government receipts.
Allowing Exemptions from the Payroll Tax.
An alternative proposal is to allow wage
earners exemptions for themselves and their
dependents in computing their payroll taxes.
A minimum standard deduction might also
be permitted. The cost of those exemptions
in foregone OASDHI revenues is consider
able, but could be significantly reduced by
denying exemptions to those with incomes
above a certain level.
WHAT THE CRYSTAL BALL SEES
FOR THE PAYROLL TAX
As the public becomes more aware of the
true nature of the co m p u lso ry pension
scheme and of the inequitable and regres
sive aspects of the payroll tax, the present
method of financing Social Security will no
doubt prove considerably less flexible than
Replacing the Payroll Tax with the Income
Tax. Neither of the previous proposals
eliminates all of the inequities associated
with the payroll tax. The tax would remain
regressive above the ceiling on taxable
wages. D is c rim in a tio n against multipleearner families would also continue, unless
See Brittain, op. cit., pp. 147-49.
9
BUSINESS REVIEW
APRIL 1973
out and replaced with the income tax. The
types of reforms are difficult to predict, but
their inevitability becomes more apparent
with each increase in the government's bite
on workers' paychecks to feed the Social
Security trust funds. While these reforms
won't make the bite any smaller in the ag
gregate, the hope is the bite will be felt
more evenly.
■
in the past. Yet some benefit increases have
recently become "automatic" and Congress
will probably find it difficult to eschew the
temptation to boost benefits near election
times. These considerations will eventually
force some modification in the payroll tax,
such as allowing personal exemptions, a
minimum standard deduction, or both. In
the long run, the payroll tax may be phased
SELECTED READINGS ON SOCIAL SECURITY
Brittain, John A. The Payroll Tax for Social Security. Washington: The Brookings In
stitution, 1972.
Economic Aspects of the Social Security Tax. New York: The Tax Foundation, Inc.,
1966.
Issues in Future Financing of Social Security. New York: The Tax Foundation, Inc.,
1967.
Pechman, Joseph A., Aaron, Henry J., and Taussig, Michael K. Social Security: Per
spective for Reform. Washington: The Brookings Institution, 1968.
10
FEDERAL RESERVE BANK OF PHILADELPHIA
BUSINESS REVIEW
APRIL 1973
CHART 1
COMPARED TO THE AVERAGE TAXPAYER OF OTHER INDUSTRIALIZED COUNTRIES OF
THE NONCOMMUNIST WORLD, AMERICANS PAY A GREATER SHARE OF THE NATION’S
TOTAL TAX TAKE IN THE FORM OF PERSONAL INCOME TAXES
Percent
—
PERSONAL INCOME TAXES/TOTAL TAXES
40
■
30
20
10
0 --Source:
Organization for Economic Cooperation and Development, Revenue Statistics of OECD
Member Countries, 1968-1970
12
FEDERAL RESERVE BANK OF PHILADELPHIA
CHART 2
SO DOES THE CORPORATE SECTOR WHOSE SHARE IS WELL ABOVE THE AVERAGE
OF NEARLY ALL ITS FOREIGN COUNTERPARTS
Percent
---------------------------------------------------------------------------------------------------------------------------------------
Bill
25
13
BUSINESS REVIEW
APRIL 1973
CHART 3
BUT THE PORTION OF TOTAL COLLECTIONS ACCOUNTED FOR BY SOCIAL SECURITY
IS CONSIDERABLY LOWER THAN THE FOREIGN AVERAGE . . .
Percent
SOCIAL SECURITY TAXES/TOTAL TAXES
40
30
(10 NATION AVERAGE-23.59)
20
10
Qo
lli c j
i- a
z3—
=
O
<
Z
_ U _
14
FEDERAL RESERVE BANK OF PHILADELPHIA
15
APRIL 1973
BUSINESS REVIEW
CHART 5
YET, WHEN OVERALL CONTRIBUTIONS ARE CONSIDERED IN TERMS OF THE TOTAL
TAX PICTURE,* TAXES IN THE UNITED STATES TAKE A SMALLER BITE OF THE GNP
THAN IN MOST OF ITS INDUSTRIALIZED COUNTERPARTS
Percent
TOTAL TAXES/GROSS NATIONAL PRODUCT
40
(10 NATION AVERAGE-31.87)
30
20
5
10
o
a
e
>
z
*
a
LU
H
Z
=
)
LU
O
z
<
a:
>
z<
5
cc
LU
u
<
a
<
z<
at
<
L
—
CO
Q
LU
a
z
<
_i
cc
LU
N
H
5
CO
z
<
Q
.
These figures are for 1968, 1969, and 1970. Since 1970 the United States has had a reduction
in the Federal income tax and corporate profits tax, but this has been offset by increases in
Social Security taxes and many state and local taxes. Thus, the U. S. tax burden may be a little
higher, but the ranking is probably the same.
16
FEDERAL RESERVE BANK OF PHILADELPHIA
How Reliable Are
Those Price and
Employment Measures?
By David B. Thomas
“ Price Index Is Up Sharply on Record
Grocery Rise/'
New York Times, February 23, 1973
ton complain that conventional policy
instruments can't seem to handle both
problems simultaneously. Recent experi
ence bears this out. Unemployment has
responded only sluggishly to the stimula
tive economic policies of the last two
years, while at the same time inflationary
pressures have continued to plague the
economy.
Curing this malady would be difficult even
with perfect statistics. The problem is com
pounded by the fact that our most com
monly used measures of inflation and
unemployment— the Consumer Price Index
(CPI) and the unemployment rate— may
not be infallible. Unemployment occasion
ally charts a substantially different course
than the official rate suggests. And the
Consumer Price Index, built into many wage
settlements, may significantly overstate the
level of inflation.
“ Unemployment Rate Dips to 5% ,
Best in 2Vi Years,"
Philadelphia Inquirer, February 3, 1973
“ Consumer Price Increase Narrowed
0.3 % ,"
Wall Street Journal, November 22, 1972
“ Prices Rise 0.4% ; Increase Biggest in
Last 5 Months,"
New York Times, August 23, 1972
Striking an acceptable balance between
inflation and unemployment has become,
in recent years, a herculean task for policy
makers who must ride herd on the Amer
ican economy. With the public clamoring
for slower inflation and lower unemploy
ment, furrow-browed officials in Washing
17
BUSINESS REVIEW
APRIL 1973
UNEMPLOYMENT RATE: A PROBLEM OF
SAMPLES AND SEASONS
attributed to sampling errors. And even
though a .2 percent decline is significant
“ statistically," it doesn't necessarily indicate
a large decrease in the percentage of the
labor force out of work.2
The unemployment rate, which shows
what portion of the labor force is out of
work at any given time, is one of the most
widely used and closely watched barome
ters of economic conditions. Policymakers,
concerned with providing jobs for all those
willing, able, and wanting to work, use it in
determining and evaluating programs geared
toward this goal. The jobless rate also is
often used to gauge the utilization of pro
ductive capacity. As such it acts as a ther
mostat by signaling when the economy is
either straining or operating beneath its
productive capacity. While the unemploy
ment rate (see Box) can be useful in assess
ing the state of the economy, it has many
shortcomings.
Predictable Patterns? Because factors such
as school opening and closing dates, crop
seasons, production schedules, and holidays
cause employment to fluctuate regularly
from month to month, published unemploy
ment figures are most often seasonally ad
justed. The fluctuation which usually occurs
during a particular month is removed in the
adjusted rate, making it easier to discern
how the jobless picture has changed relative
to previous months.
Additional error may creep into monthly
unemployment figures, almost gremlinlike,
through the seasonal adjustment process.
This is because the method used relies main
ly upon seasonal patterns observed in past
years and cannot pick up any new pattern
before it actually occurs. Variations in wea
ther conditions, holidays, production sched
ules, and school openings and closings
insure that seasonal patterns will never be
constant. To the extent that such events
occur, the adjustment process may com
pound the error already present.
Predicting the size of this “ seasonal ad
justment" error for any particular month is
impossible. After one or two years, how
ever, when any new seasonal patterns have
been identified, the official rate is revised.
Over the past few years these revisions have
averaged about .1 percent— a reasonable
approximation of the error caused by the
adjustment process.
The Sampling Scheme. If Bureau of Labor
Statistics measurements indicate the unem
ployment rate dropped from 5.5 to 5.3
percent last month, this would be heralded
by the news media and public as proof that
labor market conditions are improving. In
such a case, however, it is possible -that the
percentage of unemployed didn't change a
bit— it may even have risen!
Such potential discrepancies between the
measured and actual unemployment rates
arise because jobless figures are derived
from only a sample of the population.
These sample results may differ from the
figures that a full census would produce.
For example, to be certain that any mea
sured change in the unemployment rate
isn't only the result of variations inherent
in the sampling process, it must change by
.2 percent or more between the consecu
tive months.1 Movements in the jobless rate
of less than this amount can reasonably be
2 The chances are nine out of ten that the true
change in the unemployment rate will be within .2
percent of the measured change. Thus, when the
measured rate declines by .2 percent between two
consecutive months, the true decline could be any
value between zero (since .2 - .2 = zero) and .4
(since .2 + .2 = .4).
‘ J o h n E. Bregger, "Unemployment Statistics and
What They Mean," Monthly Labor Review, November
1971, p. 24.
18
FEDERAL RESERVE BANK OF PHILADELPHIA
EMPLOYED OR UNEMPLOYED?
CONFLICTING CONCEPTIONS
Employment figures are culled from the largest monthly sampling in the world,
the Current Population Survey. Conducted for the Bureau of Labor Statistics (BLS)
by the Bureau of Census, this survey gathers data concerning the employ
ment status of some 105,000 people, 16 years or older, across the country. During
1971 some 145 million Americans fitted this description. Thus, each person in the
survey represents about 1,350 in the total population.
The Bureau of Labor Statistics is responsible for defining what is meant by "em
ployed" and “ unemployed"*:
Job Holders. Employed are persons 16 years or older who, during the week
previous to the monthly Current Population Survey, either
• did any work at all for pay or profit, or
• worked a minimum of 15 hours without pay for a family business, or
• have a job but are temporarily out of work because of such factors as strikes,
vacations, bad weather, or illness.
Job Seekers. To be classified as unemployed, a person must
• be 16 years or older, and
• be currently available for work, and
• have engaged in some specific job-seeking activity during the previous four
weeks, and
• not have worked at all for pay during the week previous to the survey.
Many critics claim that BLS requirements for being classified as unemployed are
too stringent. Because of this, they state, there's a small army of the "disguised" or
"hidden" unemployed— those who would like to work but have had such bad
luck finding jobs in the past that they have given up pounding the pavement.
According to the BLS classification scheme, discouraged workers such as these
would not be counted as members of the labor force, since they have not en
gaged in a specific job-seeking activity within the last month. Some believe that
because of this, the official unemployment rate tends to understate the actual job
less rate.
* Source:
U. S. Department of Labor, Bureau of Labor Statistics; BLS Handbook of Methods, Bulletin
1711.
example, fall from 5.8 to 5.5 percent). Yet,
during the past year it varied by this amount
during only two months.
Does this make movements like those
reported for the remaining ten months
completely meaningless? Perhaps. If the
measured unemployment rate fluctuates by
small amounts around a particular level for
several months in a row, chances are that
Unemployment Figures in Perspective.
The total error for BLS monthly estimates of
changes in the unemployment rate might be
as high as .3 percent— about .2 percent from
sampling error and .1 percent from seasonal
adjustment error. To be absolutely certain
that the job scene differs between two con
secutive months, then, the unemployment
rate must change by at least this much (for
19
BUSINESS REVIEW
APRIL 1973
these changes are only the result of mea
surement error. If it shows small movements
in the same direction for several consecu
tive months, however, then the jobless pic
ture is actually changing. Trends in the rate
give a better reflection of what is happen
ing in the job market than the change for
a single month (even if none of the monthto-month movements forming the trend are
"statistically" significant in themselves).
Finally, inasmuch as the amount of sam
pling error decreases when the size of a
sample increases, quarterly figures for un
employment, which are derived from sam
ples three times the size of the monthly
samples, are much more reliable than
monthly ones. If the official rate falls from
5.9 to 5.7 percent between two quarters,
as it did during 1972, for example, the
chances are greater than 19 out of 20 that
the jobless rate actually decreased.
adjustment process account for only a small
portion of the potential inaccuracies in the
CPI.4 Two major problems with the CPI
are caused by improved goods and services
and by changes in the buying habits of
consumers.
Quantifying Quality. Technological prog
ress improves the quality of many goods
and services. A color TV made in 1973 is
quite different from one built ten years
earlier: It has a wider screen, clearer recep
tion, sharper color, and hopefully will last
longer. Some consumer advocates, notwith
standing, there are few people who would
trade a '73 color set for an unused '63 model.
If the CPI is to measure changes in pur
chasing power accurately, the prices of
goods and services included in the market
basket must be adjusted to represent such
quality changes. Suppose the price of a
particular color model rose by 20 percent
since 1963. To some consumers this might
indicate a 20 percent inflation in the price
of color TVs during this period. But the '73
model offers the consumer more quality
and satisfaction than the '63 one. The price
of color TVs may not have risen at all, if
GAUGING INFLATION WITH THE CPI
Coming to grips with slippery unemploy
ment figures is just one aspect of the
problem economists and policymakers face
in analyzing the tradeoff between rising
prices and joblessness. The other side of
the problem is that of measuring inflation.
The Consumer Price Index is the yardstick
commonly used.3 The "cost of living" index,
the CPI's popular name, suggests the ration
ale undergirding this choice. Of all price
indices, this one best gauges the effect of
rising prices on the workers' purchasing
power. Thus, it reflects many of the head
aches and hassles caused by inflation.
Although calculation of the CPI is a rela
tively straightforward task (see Box), the
index may not accurately reflect inflation
ary pressures in the economy. Errors infil
trate the CPI just as they do jobless figures.
However, sampling errors and the seasonal
4 Seasonal adjustment isn't as significant a problem
in the all-items CPI, so the index isn't presented in
the adjusted form. The prices of goods included in
the market basket have different seasonal variation pat
terns which largely cancel each other. To the extent
that this is true, the adjusted numbers would provide
no new information.
It is inevitable that a certain amount of sampling
error be present in the CPI, because the index is
constructed from monthly samples of retail outlets.
Fortunately, however, the amount of error originating
here is insignificantly small. The chances are 19 out
of 20 that sampling error will not exceed .08 percent
for any particular month. When the "rounding-off" of
the price figures is taken into consideration, this makes
a .2 percent change in the CPI between two consecu
tive months statistically significant. For example, if
BLS measurements indicate the CPI rose from 100.0 to
3
The CPI is officially called "The Consumer Price 100.2 during some months, we can be 95 percent
confident that prices actually did rise.
Index for Urban Wage Earners and Clerical Workers."
20
FEDERAL RESERVE BANK OF PHILADELPHIA
THE "BASE PERIOD MARKET BASKET":
H O W THE BLS KEEPS TABS ON PRICES
The method used by the Bureau of Labor Statistics to measure changes in the
price level of consumer goods and services involves two basic steps. First, about
once every ten years a large sample of consumers falling into the category of “ ur
ban wage earners and clerical workers" are interviewed concerning their spending
habits.* (The most recent survey was undertaken in 1960-61.) From the results of
this Consumer Expenditure Survey, the BLS is able to construct the index's "base
period market basket." For the sake of simplicity, consider the market basket as
representing a single budget that shows how Norman Normal, the typical con
sumer, spent his income of $10,000 during the base year.
About 400 different goods and services are included in the market basket— every
thing from apples to washing machines. These items are divided into 52 expendi
ture classes of similar products or services. A fixed weight is assigned to each class
on the basis of how Norman Normal divided his income among the classes. For
example, if he budgeted 5 percent of his income for fresh vegetables, this expen
diture class would receive a weight of .05. Weights are assigned to products within
each class on the same basis, with these weights showing how Mr. Normal bud
geted his money among the products included in a particular class.
After the "base period market basket" has been constructed, the second step,
that of monitoring prices, is relatively easy. Each month the BLS sends surveyors to
selected retail outlets in major cities across the country to collect data concern
ing current prices of goods and services included in the market basket. When this
information has been gathered, the BLS calculates exactly how much it would now
cost to purchase the same set of goods and services (the market basket) that Norman
Normal bought in the base period with his $10,000 income.
The level of the CPI is calculated by dividing the market basket's current cost
by its base period cost and multiplying this figure by 100. Suppose we find that it
now costs $11,000 to buy Mr. Normal's market basket. The new level of the index,
then, is t($11,000/$10,000) x 100} or 110. This indicates that the prices of goods
and services purchased by consumers have risen, on the average, by 10 percent
since the base period.
* Some people critically point out that this group represents only about 40 percent of the total popula
tion. In recent years, however, this occupational classification has lost most of its significance, as the
spending habits of this group have become similar to those of the rest of the population. Thus, although
the index does not claim to represent all consumers, it probably represents a large majority of them.
But assigning a dollar and cents value to
things such as TV set durability or better
color reception is a tricky business.
Many economists claim that BLS methods
of adjusting the prices of goods for quality
changes have failed to capture the full ef
fects of quality improvements. This short
price increases reflect the changes in qual
ity.
The BLS is confronted with the need, in
such a case, to decide what part of a price
increase is the result of quality improve
ment and what part is purely inflationary.
Methods have been developed to do this.
21
BUSINESS REVIEW
APRIL 1973
annually.6
coming, they maintain, causes the CPI to
overstate the rate of inflation by .5 to 1.5
percent annually.5
A Relatively Accurate Index. In the final
analysis, the CPI, like most other economic
indicators, has its strong and weak points.
The index's strength lies in its ability to mea
sure relative changes in the rate of inflation
on a month-to-month basis. For example,
when the CPI rises by .8 percent in July and
.4 during August, the rate of inflation has
dropped by about half during August. How
ever, inadequate adjustment for quality
changes and new consumption patterns may
cause the index to o ve rstate a b so lu te
changes in the level of consumer prices by
as much as 1 to 2 percent a year. Thus, al
though the rate of inflation fell by about half
in the example, it is much less certain that
prices increased by .8 percent in July and
.4 percent in August. And this is the CPI's
basic weakness.
Changing Consumption Patterns. Addi
tional upward bias in the CPI is caused by
the relative inflexibility of the "market bas
ket" used in its calculation (see Box). The
current weight structure of the expenditure
classes is based upon the results of the
1960-61 Consumer Expenditure Survey. Ac
cording to the index, then, consumers today
have the same basic spending patterns as
they did more than a decade ago.
The "fixed" construction of the CPI's mar
ket basket ignores the tendency of consum
ers to substitute relatively low-priced goods
for relatively high-priced ones. Since the
early 1960s, for example, families might have
budgeted a higher percentage of their in
come for chicken and a lower percentage
for pork, because the price of chicken has
risen less than the price of pork.
Overlooking such "substitution effects"
gives too much weight to goods and services
with rapidly rising prices. This, in turn,
causes the index to overstate the rate of in
flation. Judging by the experience gained
from past revisions of the market basket, the
amount of upward bias coming from this
source is in the neighborhood of .5 percent
BETTER CAUTIOUS THAN CONFIDENT
Policymakers continue to base their deci
sions on changes in the Consumer Price In
dex and the unemployment rate, even with
the awareness that these indicators may not
be totally accurate. There is little choice,
for even if the two measures are less than
perfect, they are among the best we have
at present. Given these circumstances, the
indices' limitations, with respect to accuracy,
must be recognized and considered when
basing policy decisions on their movements.
Overreacting to small m onth-to -m onth
changes in unemployment figures, as the
news media and public often do, could lead
to perverse policy decisions. The same
should be said of relying too heavily on
changes in the price level as measured by
the CPI. Finally, it should be remembered
that monthly price and unemployment figuses are only rough gauges of economic
trends and, as such, do not lend themselves
to "fine tuning" the economy.
‘’ The argument that unaccounted-for improvements
in quality cause an upward bias in the CPI is by no
means universal. Some economists, although probably
less than the majority, believe the reverse is true—
that unaccounted-for deterioration of quality causes
the CPI to understate the actual rate of inflation.
For a more thorough analysis of the upward bias in
the CPI resulting from quality changes, see William H.
Wallace, "Measuring Price Changes," Monthly Review
of the Federal Reserve Bank of Richmond, November
1970; and Richard Ruggles, "Measuring the Cost of
Quality," Challenge, November 1961. A fuller dis
cussion of possible downward bias in the CPI appears
in Jack E. Triplett, "Quality Bias in Price Indexes and
New Methods of Quality Measurement," Zvi Griliches,
ed., Prices Indexes and Quality Change: Studies in
New Methods of Measurement (Cambridge, Mass.:
Harvard University Press, 1971), pp. 180-214.
6See "Needed: A New Dimension for the CPI,"
Monthly Economic Letter, First National City Bank of
New York, November 1970.
22
FOR THE R E C O R D ...
Billion, ol Doll.f.
2 Y EA R S AGO
Y EA R AGO
APRIL 1973
Third Federal
Reserve District
United States
Percent change
Feb. 1973
SU M M A R Y
from
mo.
ago
year
ago
Percent change
2
mos.
1973
from
year
ago
Feb. 1973
from
mo.
ago
year
ago
2
mos.
1973
from
year
ago
MANUFACTURING
LO C A L
CH AN GES
Standard
Metropolitan
Statistical Areas*
Wilmington......................
+ 4 + 10 +10
Electric power consumed . . . - 4
0
Man-hours, total*....................
Employment, total........................
0
0
Wage income*................................
CONSTRUCTION**.......................... -52
COAL PRODUCTION....................... + 7
BANKING
(All member banks)
Deposits............................................
Loans.................................................
Investments....................................
U.S. Govt, securities...............
Other.............................................
Check payments***.....................
Manufacturing
+ 5
+ 3
+ 2
+ 11
-3 2
0
+ 9
+ 4
+ 2
+12
+ 9
- 4
+
+
+
+
+
2 N/A N/A
1 N/A N/A
2 N/A N/A
1 +22 +18
7 + 1 -•5
Employment
Percent
change
Feb. 1973
from
Banking
II
Percent
change
Feb. 1973
from
Percent
change
Feb. 1973
from
Percent
change
Feb. 1973
from
month year month year month year month year
ago ago
ago ago ago ago ago ago
0
+11
Atlantic City................... + 5 + 8
0
+22
+ 4 +12
Bridgeton......................... + 1 + 7
N/A
Trenton.............................
- 2 + 9
0 + 2
Altoona............................. + 1 + 3
Harrisburg.......................
Check
Total
Payments** Deposits***
0 + 5
N/A
0 + 6
+ 7 + 25
0
-
1 +17
9 -F
N/A
7 -
N/A
+ 1
+27 + 55 -
2
-8 8
N/A
+15
- 1 4 + 11 + 1 +16
+11 .
+18
+ 2
- 2
+ 4
+ 2 3f
+ 8
+18
+ 3
- 3
+ 6
+ 27f
+
+
-
1
3
3
8
0
+ 3
+13
+22
+ 4
- 3
+ 7
+23
+12
+21
+ 5
0
+ 8
+23
+ 15 + 3 +23
- 2 + 9
- 11
+
Lancaster......................... + 1 + 9
0
1
2
3
1
6f
-1 0
Johnstown....................... - 1 + 1
+
-
+ 1 +17
+ 1 +17
-1 6
+127 + 1 +17
Lehigh Valley.................
0 + 2
+ 2
-
Philadelphia...................
0
Reading............................
0 + 2
+13
+ 1 + 1 + 8
0
+11
8 + 3 +17
5 + 19 + 1 +16
-1 0
+ 22
-12
-
Scranton........................... + 1 - 2
+ 3 + 8
-
0
+13
6 + 3
+22
6 + io + 1 +14
Wilkes-Barre.................. + 1 + 1
Consumer......................................... + I t
♦Production workers only
♦♦Value of contracts
♦♦♦Adjusted for seasonal variation
+ 4f + 4t
+ 2 + 8 + 8
+ 1 + 4 + 4
fl5 S M S A s
^Philadelphia
+ 1 + 6
-1 0
+ 25 + 2
Williamsport................... - 1 + 5
PRICES
N/A
N/A
-2 0
+ 23 + 1
N/A
York...................................
+ 1 +13
-1 1
- 46 + 2
+17
0 + 4
+34
♦Not restricted to corporate limits of cities but covers areas of one or more
counties.
♦♦All commercial banks. Adjusted for seasonal variation.
♦♦♦Member banks only. Last Wednesday of the month.
rC D X R A L R X 0 E R V E B ANK
FEDERAL REHEHVI BANK
FEDERAL RESERVE BANK of PHILADELPHIA
PHILADELPHIA, PENNSYLVANIA 19101
b u sin e ss review
FEDERAL R E S E R V E BA N K
OF PH ILAD ELPH IA
PHILADELPHIA, PA. 19101
@FedFRASER