The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
C O o GO O June/July 1981 Vol. 63, No. 6 <3 T CD cd C D C O C D Od ~S. a3 3 11 W hy the M edian-Priced Hom e Costs So Much 20 Inflation— The Cost-Push M yth “ a C D Indexation of Social Security Benefits A Reform in Need of Reform The R e v i e w is published 10 times per year by the Research Department of the Federal Reserve Rank o f St. Louis. Single-copy subscriptions are available to the public free o f charge. Mail requests for subscriptions, back issues, or address changes to: Research Department, Federal Reserve Rank o f St. Louis, P.O. Rox 442, St. Louis, Missouri 63166. Articles herein may be reprinted provided the source is credited. Please provide the Rank’s Re search Department with a copy of reprinted material. 2 Indexation of Social Security BenefitsA Reform in Need of Reform NEIL A. STEVENS c I^ O C IA L security benefits for retired workers, their dependents and their survivors are indexed or linked to movements in the consumer price index (C P I).1 Price indexing, a method of making adjustments to inflation, is the linking of nominal (dollar) magnitudes, such as wages, interest rates, government expenditures or taxes, to movements in a price measure. The purpose of indexing is to ensure that the purchasing power over goods and services is not changed by movements in the general level of prices.2 Under current law, social security payments are automatically increased in June “whenever the CPI in the first quarter of the calendar year exceeds the CPI in the first quarter of the previous calendar year by at least 3 percent.”3 iThe price index specified in legislation is the consumer price index for urban wage earners and clerical workers rather than the more recently constructed and more broadly based series for all urban workers. 2For background information on indexing, see Thomas M. Humphrey, “The Concept of Indexation in the History of Eco nomic Thought,” E conom ic R eview (Federal Reserve Bank of Richmond, November/December 1974), pp. 3-16; Herbert Giersch, Milton Friedman, et al., Essays on Inflation and Indexation (Washington, D.C.: American Enterprise Institute for Public Policy Research, 1972); and Jai-Hoon Yang, “The Case For and Against Indexation: An Attempt at Perspective,” this R eview (October 1974), pp. 2-11. 3An Analysis o f the E ffects o f Indexing for Inflation on F ed eral Expenditures, Report to the Congress of the United States by the Comptroller General (GAO, August 15, 1979), p. 18. The current provision for the automatic indexing of social security benefits was contained in social security legislation enacted in 1973. Initially, however, automatic indexing of benefits was included in die social security legislation enacted in 1972. The first effective date was to have been January 1, 1975, based on increases in the CPI from the third quarter of 1972 to the second quarter of 1974. In subsequent years, possible benefit increases were to be based on second-quarterto-second-quarter changes in the CPI and made effective Jan uary 1. Legislation enacted in 1973 amended these provisions by providing the first possible automatic increase in benefits to be effective in June 1975, based on the change in CPI from the second quarter of 1974 to the first quarter of 1975. For example, effective June 1, 1981, social security payments were increased 11.2 percent, reflecting the increase in the consumer price index from the first quarter of 1980 to the first quarter of 1981.4 Social security programs are funded by a tax on wage and salary income.5 Currently, benefits for re tired workers, their dependents and survivors, are rising faster than revenues into the Old Age and Survivors Insurance (OASI) trust fund from which these benefits are paid. According to estimates by the trustees of the social security system6 and the Con gressional Budget Office (CBO), the OASI trust fund will face a financing problem beginning in late 1981 or early 1982 that will continue through the decade.7 CBO estimates show that outlays from the OASI trust fund in fiscal year 1981 will exceed income into the fund by $4.8 billion. Furthermore, estimates are that the trust fund balance will be depleted by the end of fiscal year 1983, and that the fund will be approxi mately $64 billion in deficit by the end of 1986 (table 1). 4Increases are effective June 1 and are payable July 1. 5Social security benefits are provided under three separate programs. The Old Age and Survivors Insurance program, the largest, with expenditures of $87.6 billion in 1979, pays benefits to retired workers, their dependents and survivors. The Disability Insurance program pays benefits to disabled workers and their dependents, and the Hospital Insurance program pays benefits to workers covered by the previous two programs and the railroad retirement program. 6The trustees include the Secretary of the Treasury, the Secre tary of Labor, and the Secretary of Health and Human Services. 11980 Annual R eport of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, H. Doc. 96-332, 96th Congress, 2nd session (GPO, 1980), p. 50; and Paying fo r Social Security: Funding Options fo r the N ear Term , Congressional Budget Office of the Congress of the United States (February 1981), p. 13. 3 JUNE/JULY F E D E R A L R E S E R V E BAN K O F ST. L OUI S 1981 Table 1 Outlays, Income and Balances for the Old Age and Survivors Insurance Trust Fund (billions of dollars) Projections2 A ctual1 1970 1975 1980 1981 1982 1983 1984 1985 1986 Outlays $27.3 $56.7 $103.2 $122.6 $141.4 $158.7 $178.0 $199.3 $222.6 Income 31.7 58.8 100.1 117.8 129.0 143.0 159.1 181.9 203.7 Year-end balance (fiscal ye a r) 32.6 39.9 24.6 19.7 7.4 -8 .2 -27.1 -4 4 .5 -6 3 .5 iSocial Security Bulletin (April 1981), p. 43. 2Paying for Social Security: Funding Options for the Near Term , (Congressional Budget Office of the Congress of the United States, February 1981). This short-term financing problem will begin to dissipate in the next decade but only because of sizable increases in payroll taxes scheduled as a re sult of the 1977 amendments to the Social Security Act. In the next century, however, an imbalance again will emerge between promised social security bene fits and projected revenues. The solution to this prob lem will require either larger increases in taxes than those already scheduled or a reduction in promised benefits. This article examines the role that the price index ing of benefits has played in creating these financing problems, particularly in the short term. The article shows that the use of the CPI for indexing probably has overstated the benefit increases necessary to keep the purchasing power of benefits constant, and that price indexing is inconsistent with the way benefits are funded. In addition, it discusses modifications of indexing formulas that would help eliminate the cur rent imbalance or avert the development of future financial imbalances. Table 2 Increases in Social Security Benefits and the Consumer Price Index Effective date of increase Increase in benefit level Increase in CPI from last effective date of benefit increase1 13% 0.5% 2 Before automatic indexing Septem ber 1954 January 1954 7 January 1965 7 7.8 February 1968 13 9.3 January 1970 15 10.8 January 1971 10 5.2 Septem ber 1972 20 5.9 June 1974 113 Over time, the price level varies as the stock of money changes relative to the available amount of goods and services. Price indexing first came about to rectify the fact that prices or incomes administered by government agencies or fixed by private contractual agreements, such as wage contracts or fixed payment mortgage contracts, do not adjust rapidly to changes in the price level. Market-determined prices of goods and services eventually will adjust to changes in the price level. However, social security benefits and pay ments under other government programs, which are 4 16.4 After autom atic indexing June 1975 The Rationale for Price Indexing 8.0 8.0% June 1976 6.4 June 1977 5.9 June 1978 6.5 June 1979 9.9 June 1980 14.3 June 1981 11.2 ’ Computed from not seasonally adjusted data for urban wage earners and clerical workers. 2Increase in CPI from September 1952, the previous date benefits were increased. 3Effective in two steps — a 7 percent increase in March 1974 and a 4 percent increase in June 1974. FEDE RAL . R E S E R V E BANK O F ST. L OUI S JUNE/JULY 1981 Table 3 Alternative Measures of Changes in the Price Level1 Consumer price index Period Official series2 Experimental rental equivalence series Official series minus experim ental series Personal consumption expenditures deflator Official CPI minus PCE deflator 1947-77 3.4% n.a. n.a. 3.3% 0.1% 1977 6.6 6.2% 0.4% 5.9 0.7 1978 9.0 7.8 1.2 7.5 1.5 1979 12.7 10.6 2.1 9.5 3.2 1980 12.5 10.8 1.7 10.1 2.4 1Changes are from fourth quarter to fourth quarter, except the 1947-77 period which was computed from yearly average data. 2AU urban worker series. not directly determined by market forces, must some how be adjusted to changes in the price level if Congress desires to offset the impact of these changes on the payments’ purchasing power. On a practical level, an advantage of price indexing is that it is automatic, so it relieves Congress from having to devote considerable attention to frequent increases in benefits in an inflationary environment. Also, given the tendency of Congress in the late 1960s and early 1970s to increase benefits faster than the price level (table 2), indexing may have acted to put a cap on benefit increases, and thus may have averted an even larger financing crisis. Nevertheless, there are problems with the price indexing of benefits that should be discussed. The Vexing Problem of Measuring Changes in the Price Level Price indexation requires a statistical measure of the price level, which in practice, is measured by price indexes. Typically, a price index is designed to answer the question, “How much does the cost of a basket of goods and services change over time?” Unfortu nately, there are numerous technical problems in pro viding an answer.8 8For a more thorough discussion of the problems in measuring the price level, see William H. Wallace and William E. Cullison, Measuring Price Changes: A Study o f the Price Indexes, 4th ed. (Federal Reserve Bank of Richmond, 1979); and R.G.D. Allen, “Index Numbers in Theory and Practice” (Chicago: Aldine Publishing Co., 1975). Two widely used price indexes — the consumer price index and the personal consumption expendi tures (PCE) deflator — demonstrate the differences that can arise between price measures. Though these indexes have followed similar patterns in the past, their movements diverged substantially in 1979 and 1980 (see table 3). For example, the CPI increased 3.2 percentage points more than the PCE deflator in 1979 and 1.7 percentage points more in 1980. While there are several differences in the way these two price indexes are computed, two of the more signifi cant differences are the choice of period in which to define the market basket — either current or past — and the methods used to measure housing costs. Fixed Versus Variable Weights — The CPI-W, which is used to index social security benefits, is a “fixed-weight” index. The procedure for calculat ing this index is to regularly survey prices of a large number of items consumers typically purchase (the so-called market basket) and compare the aggre gate cost of these items with the cost of the same market basket in a selected base period. The weights given to various expenditure categories in the mar ket basket are kept constant or fixed from period to period. Currently, the CPI market basket is based on a survey taken in the 1972-73 period. In the past these weights have revised approximately every 10 years. The PCE deflator, on the other hand, is essen tially a variable or current-weight index. This index, unlike the CPI, takes into account the prices of per sonal consumption items in the current period and, in effect, weighs them by the quantities currently pur 5 F E D E R A L R E S E R V E BAN K O F ST. L OUI S chased. Since expenditure patterns can change, the weights can vary from period to period. Economists have recognized that both fixed- and variable-weight indexes present problems. One prob lem is that consumption patterns change over time as people alter their preferences or respond to changes in relative prices. For example, in the 1970s the price of oil rose sharply relative to other prices. As a result, consumers curtailed their consumption of gasoline and other oil-based products and purchased substitutes for these products. Yet, in calculating the CPI, the 1972-73 consumption pattern for oil-based products continues to be used. As a result, the CPI overstates changes in the price level. On the other hand, the PCE deflator, which uses the current ex penditures for oil-based products in its calculation, understates the price level, since substitutions and curtailed energy consumption lowered the living standard of consumers. While over past periods changing expenditure pat terns have been a rather insignificant problem, as evidenced by the small difference between the CPI and PCE deflator from 1947 to 1977, the problem has been somewhat more pronounced in recent years. For example, in 1979 it is estimated that about half of the 2.9 percentage-point difference between the CPI and the PCE deflator can be attributed to changing weights or differences in weights on gasoline pur chases.9 Also important was the approximately 1.8 percentage-point difference caused by the different treatment of homeownership costs in the two indexes. Homeownership Costs — Much recent criticism of the CPI has focused on the measurement of housing costs. Durable goods, such as houses, are consumed over an extended period of time. Thus, the purchase of a durable good represents an act of saving (future consumption) with relatively minor effects on cur rent consumption. The cost of owner-occupied housing, like other items in the CPI, are calculated by using weights derived by surveys in the early 1970s. In this period, about 6 percent of households purchased homes. As currently measured, the CPI assumes that 6 percent of households will purchase and consume the total value of the house and one-half of the mortgage that usually goes along with it in the current period, while those living in previously purchased houses will spend nothing for housing services except for mainte9Alan S. Blinder, “The Consumer Price Index and The Meas urement of Recent Inflation,” Brookings Papers on Econom ic Activity (2 :1 9 8 0 ), pp. 539-65. 6 JU NE/JULY 1981 nance, taxes and insurance. Aside from numerous technical problems, this view of housing costs gives a misleading estimate of the cost to the “average” consumer.10 A sharp rise in mortgage interest rates, for example, raises the cost of housing, though few homeowners may actually bear the cost. While there is no “right” way to measure housing costs, economists generally accept a procedure called the “rental equivalence” approach. This method in volves the sampling of rents from rented houses. If homeowners charge rents that cover all the costs of maintaining a home, then the price of a house, the cost of credit, etc., are included in the rent charged by the owner to the renter. The Bureau of Labor Statistics, which computes the CPI, has experimented with several methods to calculate homeownership costs, including rental equiv alence.11 A comparison of the experimental rental equivalence series and the official CPI measure shows considerable differences in recent years (table 3). In 1980, for example, the official CPI measure grew at a 12.5 percent annual rate compared with only a 10.8 percent increase in the experimental series. Price Indexing and Real W age Declines The Financial Implications — In essence, the price indexing of social security payments is a promise by the government to keep benefits unchanged in terms of their purchasing power. As just seen, this may be hard to attain be cause of problems in measuring the price level. In addition, the promise of fixed real benefits is some times inconsistent with the methods used to finance social security benefits. If Congress wants to maintain the purchasing power of benefits received by current social security recip ients, it must levy an appropriate level of taxes to pay for these benefits. Social security benefits are currently funded on a pay-as-you-go basis; that is, benefits for currently retired workers are funded by payroll taxes on the wages and salaries of those cur rently working.12 A problem crops up whenever wages 10Ibid, p. 546. n As a proxy for rents on rented houses, this experimental series uses the CPI rent index, which includes rents on apart ments as well as rented houses. 12One-half of the payroll tax is levied directly on an employee’s earnings and the other half is paid by employers. Studies, however, indicate that the portion of the tax paid by em ployers is, for the most part, borne by employees. For a discussion of this point, see John A. Brittain, T h e Payroll Tax for Social Security (Washington, D.C.: The Brookings Institution, 1975), Chapters I and II. FEDE RAL . R E S E R V E BAN K O F ST. L OUI S and salaries rise more slowly than the price level. When this occurs, the benefits of current social security recipients, which rise with increases in the price level, increase more rapidly than revenues, which rise with increases in wages and salaries. Historically, wages have generally advanced more rapidly than prices; that is, real wages have generally risen (table 4). These increases reflect advances in labor productivity (output per manhour worked). In recent years, however, real wages have declined with the decline in labor productivity.13 The adjusted hourly earnings of workers,14 after allowances for increases in the CPI, have declined or remained unchanged in five of the past eight years and, on average, have de clined at about a 1 percent annual rate since 1972.15 JUNE/JULY 1981 Table 4 Changes in Consumer Price index and Hourly Earnings (year-over-year changes) Y ear Consumer price index1 Adjusted hourly earnings2 ______________ _______ ___ Current 1967 dollars dollars3 The CPI’s overstatement of the cost-of-living in crease in recent years magnifies the income transfer. This overstatement results in benefit increases above those necessary to maintain the same level of real benefits, and implies an even greater increase in taxes and a further reduction in the real after-tax wages of 13Several explanations have been offered by economists for the recent decline in productivity and, hence, the decline in real wages. One explanation points to the sharp rises in the relative price of energy in 1973-74, and again in 197980. According to some studies, these sharp increases in the relative price of energy made part of the capital stock eco nomically obsolete, thereby reducing workers’ productivity. For details of this and other explanations, see John A. Tatom, “The Productivity Problem,” this Review (September 1979), pp. 3-16. 14Average hourly earnings are reported before deductions for taxes, social insurance, fringe benefits, etc. The adjusted hourly earnings index takes into account such factors as variation in the amounts of overtime pay or shifts of workers into higher or lower paying industries. 15As discussed earlier in this paper, to the extent the CPI has overstated the rate of price increases in recent years, this decline in real wages is also overstated. 2.2% 4.5% 2.2% 1972 3.3 6.4 3.0 1973 6.2 6.2 0.0 1974 11.0 7.9 -2 .7 1975 9.1 8.3 -0 .7 1976 The decline in real wages has contributed substan tially to the financing problem the social security sys tem now faces. Moreover, if this financing problem is resolved by increasing payroll taxes rather than re ducing benefits, the price indexing of benefits will further redistribute the ability to consume the nation’s output from those working to those receiving social security benefits. When nominal wages rise more slowly than the price level, a tax increase imposes a further decline in real wages. In effect, this decline imposes an additional reduction in living standards for workers in order to leave the purchasing power of social security benefits unchanged. 1952-72* 5.7 7.3 -1 .3 1977 6.5 7.5 1.0 1978 7.6 8.2 0.6 1979 11.3 7.9 -3.1 1980 13.3 9.3 -3 .6 1972-804 8.8 7.8 -0 .9 1952-804 4.1 5.5 1.3 'Data are for all urban consumers. 2Total private nonagricultural earnings for production or nonsupervisory workers adjusted for overtime and for in dustry employment shifts. 3Current dollar index divided by consumer price index. 4Annualized rates of change. workers than would otherwise have been necessary.16 Alternative Indexing Rules Modifying the indexing rules to tie benefits to nomi nal wage movements rather than price index move ments appears to be a way to alleviate some of these problems. With benefits linked to nominal wage rates, real benefits would change commensurately with real wages; that is, when nominal wages rise faster than the price level, indicating rising productivity and a rising standard of living for workers, real social secu rity benefits would also increase. Conversely, when wages increase more slowly than the price level, indi16Another potential problem is the difference in expenditure patterns of social security beneficiaries, and urban wage earners and clerical workers. To the extent there are differ ences in expenditure patterns, relative price changes would affect the purchasing power of social security benefits differ ently than earnings of urban wage earners. A study by the Bureau of Labor Statistics, however, indicated that a con sumer price index based on purchasing patterns of retired workers would not be substantially different from the official CPI measure. See Janet L. Norwood, “Cost-of-Living Escala tion of Pensions,’ Monthly L ab or Review (June 1972), pp. 21-24. 7 F E D E R A L R E S E R V E BA N K O F ST. L OUI S JUNE/JULY 1981 Table 5 Outlays, Income and Balances for the Old Age and Survivors Insurance Fund, Assuming the Minimum Rule in Effect (billions of dollars)1 1979 1980 1981 1982 1983 1984 1985 1986 Outlays $ 89.8 $100.3 $114.7 $131.6 $149.0 $168.3 $189.6 $212.9 Income 86.9 100.1 117.8 129.0 143.0 159.1 181.9 203.7 Year-end balance 28.1 27.9 31.0 28.4 22.4 13.2 5.5 -3 .7 iEstimates are based on actual OASI trust fund data for income under current indexing procedures for 1981-86, puted by assuming that the minimum rule had been in rose less rapidly man the CPI. For subsequent years, it 1979-80 and Congressional Budget Office projections of outlays and as shown in table 1. Modifications of these outlay data were com effect in 1979, 1980, 1981, all years in which average hourly earnings is assumed that the hourly earnings rose more rapidly than the CPI. eating lower productivity and a declining standard of living for workers, real benefits would decline. Had social security benefits been indexed to aver age hourly earnings during the recent period of real wage declines, much of the current short-term finan cial problem would have been averted. Benefits in dexed to adjusted average hourly earnings during 1979, 1980, and 1981 would have resulted in social security benefit increases of 8.2 percent in June 1979, 8.4 per cent in June 1980 and 9.8 percent in June 1981, instead of the respective 9.9, 14.3 and 11.2 percent increases actually granted. In total, during these three years, benefits of retired workers rose 35 percent as average hourly earnings of current workers rose only 26 percent. To initiate the indexing of social security benefits to nominal wages at this point, however, would prob ably still increase future deficits in the OASI trust fund. Since, historically, nominal wages have risen faster than the price level, the continuation of this trend would result in greater increases in benefits than under a price-indexed scheme and hence greater deficits in the OASI trust fund than those currently projected. A variant of a wage indexing scheme is the “minimum-rule” scheme, which indexes benefits to the lower increase of the wage or price index. This rule implies that real social security benefits will decline when real wages decline, but remain unchanged when real wages rise.17 This rule would largely remove the fi17One objection to this rule is that real wages often decline and rise in business cycles. For instance, if real wages fell in year one but recovered that loss in year two, under the minimum rule real social security benefits would be re duced in the first year but kept at that same lower real level in the second year. However, Congress could easily monitor this kind of problem and remedy it by ad hoc in creases in benefits. 8 nancial problem that occurs when real wages decline. For example, if the minimum rule had been in effect in the past three years so that benefits would have increased with hourly earnings rather than the CPI, the estimated year-end balance in fiscal year 1986 would be $3.7 billion in deficit rather than the CBO estimates of $63.5 billion (see tables 1 and 5). Un fortunately, to introduce the minimum rule at this point would not solve the financing problem of the OASI trust fund, since this rule would not reduce benefits from their present level.18 Thus, the minimum rule will not cure the present budget problems, but will prevent larger deficits in a future period of de clining real wages. New measures will have to be taken to solve the current short-term financing problem. The CBO for example, has investigated a number of possible op tions, including a “partial” indexing to the CPI. It estimated that if benefits were increased by only twothirds of the actual increase in the CPI over the next several years, the financing problem over this decade would be eliminated.19 Several other proposals have also been suggested to eliminate the short-term financ ing problem. These include the reapportionment of 18The CBO has considered the effect of this rule on the im pending deficit of the OASI trust fund. According to their estimates, the minimum rule would reduce outlays of the trust fund over the next five years by $26 billion, or cut about 40 percent of the projected deficit over the next five years. Their estimate, however, is based on the assumption that the indexing rule goes into effect in 1981 when the CBO projects the nominal wages to rise about 3 percent slower than the official CPI measure. In fact, if this rule were instigated at a later date when the CBO projects wage increases to again exceed price increases, the minimum rule would do nothing to eliminate the short-run financing prob lem of the OASI trust fund. See Paying fo r Social Security: Funding Options fo r the Near Term, (Congressional Budget Office of the Congress of the United States, February 1981). 19Ibid, pp. 29-31. F E D E R A L R E S E R V E BAN K O F ST. LO UI S funds in other social security trust funds into the OASI trust fund, the taxation of social security bene fits and the reduction or elimination of certain bene fits. The administration recently has proposed a plan that would cut back the benefits of future early re tiring workers, while only marginally reducing the benefits of current recipients. While these various proposals would eliminate, to varying degrees, the estimated deficit over the next several years, they would not remove the basic inconsistency between price indexing and the pay-as-you-go financing of the social security benefits. In other words, none of these proposals would preclude additional deficits from de veloping in the OASI trust fund should future declines in real wages occur. JU NE/JULY 1981 Table 6 Social Security Taxes on Payrolls Before and After the 1977 Amendments Tax rates (p e rc e n t)1 Before Taxable base2 After Before $16,500 After $16,500 Analysts also project a long-term financial imbal ance in the social security trust funds for the next century, when outlays again are projected to exceed inflows into the social security system. One of the major factors underlying this imbalance is the in creasing ratio of retirees to workers. Since benefits are now funded on a pay-as-you-go basis, the benefits of current workers will be funded by payroll taxes paid by the next generation of work ers. This funding scheme is subject to changes in demographic patterns. Past increases in benefits were granted on the assumption that birth rates would be higher than current projections now indicate. With declining birth rates, there will be fewer workers to pay into the social security trust funds when the postWorld War II baby-boom generation begins to retire. In addition, the increased life expectancy has in creased the ratio of retirees to workers. This problem is potentially quite serious, despite the substantial increases in social security taxes that have already been scheduled as a result of the 1977 amend ments to the Social Security Act (see table 6). If the current law is left intact until 2025, taxes on payrolls would have to rise, according to estimates of the chief actuary of the social security system, by at least 8 percentage points in order to fund benefits at that point, implying nearly a 24 percent tax on taxable payrolls.20 To compensate for these scheduled in20A. Haeworth Robertson, “Financial Status of Social Security Program After the Social Security Amendments of 1977,” Social Security Bulletin (March 1978), pp. 21-30. The 24 percent rate includes the tax on both the employer and the employee. 5.85% 5.85% 6.05 6.05 17,700 17,700 1979 6.05 6.13 18,900 22,900 1980 6.05 6.13 20,400 25,900 1981 6.30 6.65 21,900 29,700 1982-84 6.30 6.70 1985 6.30 7.05 1986-89 INDEXING AND THE LONG-TERM FINANCIAL IMBALANCE 1977 1978 6.45 7.15 1990-2010 6.45 7.65 2011 and after 7.45 7.65 includes taxes for old age, survivors, disability insurance and hospital insurance. beginning in 1982, the amounts will be determined auto matically under the new law on the basis of the annual increase in average earnings on covered employment. creases in social security taxes, alternatives to reduce benefits have been suggested. One proposal is to in crease the retirement age from 65 to 68, thereby reduc ing the average period that benefits are paid out; another is to tax social security benefits, which would generate additional general revenues that could then be used to fund social security benefits.21 The modification of current indexing procedures is another alternative. Under current law, the benefits of retired workers increase with the price level, but the benefits of future retirees are tied to average wage movements.22 Wage indexing is essentially a promise 21An argument for the taxation of social security benefits can also be made on the basis of equity considerations. Taxation of social benefits, or some portion of benefits, would treat such income more equally with that of other pension in comes. Also, it would tax social security income according to the ability-to-pay criteria applied to other income. --Benefits of future retirees are indexed in two ways. First, the wage history of retiring workers is indexed to the average wages of U.S. workers so that earnings in past years are brought up to current average wage levels. For example, if the average earnings doubled from 1970 to 1980, the wage that a retiring worker actually earned in 1970, say, $10,000, would be doubled to $20,000. After a certain number of years are dropped out, the worker’s indexed earning history is averaged to obtain the average indexed monthly earnings (A IM E). A formula to compute benefits, prescribed by Conress, is then applied which in 1979 was 90 percent of the rst $180 plus 32 percent of the next $905 plus 15 percent of the excess over $1,085. This formula is indexed to average U.S. wage movements. The “breakpoints,” namely $180 and $1,085, are adjusted automatically each year to reflect changes in average U.S. wages. 9 F E D E R A L R E S E R V E BANK O F ST. L OUI S to keep the benefits of retiring workers at a certain proportion of their real income during their lifetime. Thus, as discussed earlier, when wages rise faster than prices, the initial benefits of future retiring workers will increase in real terms in step with increases in wages over their working years. This procedure itself is consistent with the current financing scheme that taxes payroll income, in the sense that future benefits and taxes will rise or fall together. However, large benefit increases were granted in the late 1960s and early 1970s based on what now appears to have been incorrect assumptions about birth and death rates. As a result, a long-term financing problem is expected to emerge if current demographic trends persist. An indexing procedure that would insure that bene fits grow at a lower rate than revenue is a “partiallyindexed” wage rule. This rule would specify that benefits be indexed to wage movements, but by a smaller percentage than the wage increase. When real wages are growing, benefits would rise over time, but not as rapidly as wages or revenues. While the purchasing power of promised benefits to future re tirees would be reduced from where they are now, they would normally be at a substantially higher level than those of current retirees. CONCLUSION Congress has indexed social security benefits to movements in the consumer price index, a reform in tended to protect the purchasing power of these bene fits. Recent U.S. experience, however, has shown that there are problems with this procedure. One problem is that the CPI has seriously overstated the rise in prices in recent years; thus, it has contributed to 10 JUNE/JULY 1981 higher benefits than were necessary to keep the pur chasing power of benefits unchanged. The Bureau of Labor Statistics is currently contemplating certain technical changes in the official CPI calculations, such as the measurement of home ownership costs by the rental equivalence method. These changes, if imple mented, may result in a better measure of changes in the price level. In addition, price indexing of benefits can be in consistent with the method currently used to finance the social security system. The system is essentially financed on a pay-as-you-go basis by taxes on wages and salaries, so when prices rise faster than wages, benefits rise faster than revenues into the OASI trust fund. To remedy such a situation, once the small trust fund balances are depleted, requires that the government either levy additional taxes on workers or reduce the growth of benefits. One reform of the indexing procedure that would greatly diminish the likelihood of such financial im balances in the OASI trust fund in a future period of declining real wages; is the so-called minimum rule. This rule would limit benefit increases to either average wage movements or the price level, whichever is smaller, and implies that benefits would decline in real terms when real wages of workers decline and remain constant when real wages rise, as the current law provides. Such a rule does not reduce benefits from those promised under current law except when real wages of workers are declining. Thus, at this point, it would neither solve the short-term financing problem faced in the 1980s nor the long-term problem that is expected to develop in the next century. It would, however, preclude them from becoming worse should prices rise faster than wages in the future. To solve these financing problems, other measures must be taken to either increase taxes or reduce benefits. Why the Median-Priced Home Costs So Much SCOTT E. HEIN and JAMES C. LAMB, JR. I n FLATION has caused many distortions that af fect the affordability of housing, especially for first time buyers. Since 1965, the price of the medianpriced house in the United States has more than tripled.1 More important, however, the annual mort gage payment for a standard financing arrangement is almost seven times as large as before for the median-priced house. As a result, the median-income family is unlikely to qualify for and presumably would be reluctant to obtain conventional financing to buy the median-priced house today. In 1965, the median before-tax family income was $7,610, the new home mortgage rate averaged 5.81 percent, and the median sale price of a new onefamily house was $20,150. With a 20 percent down payment and a 30-year mortgage to secure the bal ance, a homeowner would owe $1,136 in annual in terest and principal payment on the debt, approxi mately 14.9 percent of his income. In 1980, by comparison, median before-tax income was approximately $21,500, the new home mortgage rate for the year averaged 13.73 percent, and the median sale price of a new one-family house vaulted to $64,900. The annual interest and principal charges on a 30-year mortgage for this home, again assuming a 20 percent down payment, would be $7,249, or 33.7 percent of the median income (see table 1). As evidenced by these numbers, the change in the cost of homeownership has been drastic. This article explains why this cost has risen so sharply: why the 1965 median-income family had to pay less than 15 percent of its annual income in mortgage pay ments on a median-priced house, while the 1980 median-income family must pay more than 33 per'The median of a set of data is the number below and above which there are an equal number of observations. Table 1 Comparison of Income and Mortgage Payments for 1965 and 1980 (current dollars) 1965 1980 Median before-tax income $ 7,610 $21,500 M edian sale price of a new home $20,150 $64,900 5.81% 13.73% $ 1,136 $ 7,249 14.9% 33.7% Average mortgage rate Annual interest and principal1 Payment as a percentage of income 1Assuming 20 percent down payment. cent. Two separate issues are considered: the increase in housing prices and the increase in the cost of financing a home purchase. THE RISE IN HOUSING PRICES From 1965 to 1980, the prices of personal consump tion goods more than doubled, rising 131.8 percent. Since inflation is a sustained increase in the general level of prices, one would expect similar increases in housing prices. However, the prices of new housing for the same period rose an even higher 223.2 per cent.2 Table 2 shows the annual rate of increase in new housing prices and personal consumption goods. 2This indicates that individuals who owned homes over this period have experienced sizable capital gains. On this matter, see Patric H. Hendershott and Sheng Cheng Hu, “Inflation and the Benefits from Owner-Occupied Housing” (National Bureau of Economic Research, Working Paper No. 383, Au gust 1979) for a discussion of the capital gains experienced by households. The present paper does not analyze the ramifi cations of these capital gains on the demand for housing. F E D E R A L R E S E R V E BA N K O F ST. L OUI S JUNE/JULY 1981 Demographic and Lifestyle Factors Table 2 Annual Change in Housing Prices and the Personal Consumption Deflator1 Change in personal consumption deflator Year Change in housing prices 1965 2.9% 1.7% 1966 3.5 2.9 1967 3.6 2.4 1968 5.6 4.1 1969 8.0 4.5 1970 3.0 4.7 1971 5.2 4.2 3.7 1972 6.4 1973 9.5 5.6 1974 9.3 10.1 1975 9.5 8.6 5.2 1977 12.8 6.0 In addition, lifestyle changes apparently have in creased the demand for shelter, at least partially af fecting the demand for owner-occupied housing. For example, the proportion of unmarried adults has in creased with the rise in the divorce rate and the post ponement of marriage. These lifestyle changes have resulted in more and more single-person households. Today there often are two people demanding hous ing, where before there was one. 7.6 1976 Additional factors that help explain the relative increase in housing prices are demographic changes since 1965. First, the adult population — the pur chasers of homes — has grown rapidly in recent years.4 There appear to be two sources of this growth. One, individuals born in the post-World War II baby boom have moved into the homebuying age group. Two, the U.S. population now enjoys an increased longevity. 1978 13.7 6.8 1979 14.2 8.9 1980 10.1 10.2 1Data on housing prices are for new sales only. Only in 1970, 1974 and 1980 was the annual rate of increase in housing prices less than that of personal consumption goods. Thus, while general inflation ex plains most of the increase in housing prices, it leaves unanswered the question why housing prices have risen faster than the general price level. Inflation and the Favorable Tax Treatment of Homeotvnership A third factor causing the relative rise of housing prices is the favorable treatment of homeownership by the U.S. tax structure. As inflation has accelerated, this treatment has become even more favorable.5 For example, an individual can deduct mortgage interest expenses from taxable income in determining his in come tax. Thus, as nominal interest rates and mort gage rates rise with inflation, borrowers can deduct larger interest expenses, even if the real (inflation adjusted) cost of borrowing remains unchanged. In other words, the higher the anticipated future infla tion, the cheaper it is to borrow under our tax system.6 Since most people borrow to purchase a Quality Changes One possible explanation for this phenomenon is that the quality of housing has risen over the past 15 years; thus, we are comparing the price of two dis similar goods. Though this problem plagues all price index measures, it appears to be particularly impor tant in the case of housing. The average new home is larger and has more amenities, such as central air conditioning and insulation. Still, economists generally believe that these quality increases are not substantial enough to fully explain the rapid relative price rise.3 ®For example, Randall J. Pozdena, “Inflation Expectations and the Housing Market,” Federal Reserve Bank of San Francisco E conom ic Review (Fall 1980), pp, 29-47, estimates that 15 percent of the increase in the average home sales price be tween 1970 and 1979 is explained by quality considerations. Digitized for 12 FRASER 4For a more detailed discussion, see Dan M. Bechter, “How Much For a New House in the Years Ahead? Some Insights From 1975-80” (Federal Reserve Bank of Kansas City, Re search Working Paper 81-104), 5Anthony Downs, “The Low (Real) Cost of Housing,” Across the Board (February 1981), pp. 51-55; James M. Poterba, “Inflation, Income Taxes and Owner-Occupied Housing” (Na tional Bureau of Economic Research, Working Paper No. 553, September 1980). 6There is an important distinction between nominal and real interest rates. Nominal interest rates are market interest rates which state how many dollars the borrower will pay and the lender will receive on a loan. Since inflation depreciates the value of a dollar in terms of its command over resources, nominal rates are bid up by anticipated inflation. Real interest rates are rates that have been adjusted for inflation. The expected real interest rate can be measured by subtracting the expected annual rate of inflation from the nominal interest rate. The favorable treatment given to borrowers comes from the fact that individuals can deduct nominal interest expenses F E D E R A L R E S E R V E BAN K O F ST. L OUI S house, this increasingly favorable treatment has in creased the demand for the single-family dwelling. This benefit becomes even more important as infla tion pushes individuals into higher marginal income tax brackets (bracket creep). Bracket creep has in creased the marginal tax rate for the median-income family from 17 percent in 1965 to 24 percent in 1980. As individuals are pushed into higher marginal tax brackets, the value of deducting interest expenses in creases. Thus, even if the interest expense on a loan or mortgage were unaffected by inflation, individuals would pay less after-tax dollars to borrow in 1980 than they did in 1965. Since we have a progressive income tax structure, the increase in marginal tax rates has been even larger for family incomes greater than the median. In 1965, a family whose income was in the 80th percentile (who earned more dollars than 80 percent of all other families) was in the 19 percent marginal tax bracket. In comparison, by 1980 this family was in the 37 percent marginal tax bracket. Thus, high-income families have experienced even greater reductions in the after-tax cost of borrowing as inflation has moved them into higher marginal tax brackets.7 In another benefit of our tax structure, capital gains realized from the sale of a home are not taxed if they are reinvested in another home. In addition, people over the age of 55 can now realize a tax-free, one-time capital gain of $100,000 or less from the sale of their home. Consequently, some homeowners effec tively pay no tax on capital gains from home owner ship, substantially less than they would pay in taxes on capital gains from stocks or bonds. An additional favorable tax consideration concerns housing as a form of investment. Consider an investor from their incomes in determining taxable income. Compare two individuals — one in an inflationary environment with 10 percent inflation and the other in an environment with no in flation. Suppose the interest rates are 13 percent and 3 per cent, respectively, so that the real rate is 3 percent in both cases. Although the real cost of borrowing is the same for each individual, the after-tax real cost is lower for the person in the inflationary environment since that individual’s nominal interest expense is much larger. JUNE/JULY 1981 who is contemplating two alternative purchases: a purchase either of $70,000 in securities or a $70,000 house. In the first case, the investor earns taxable interest income from the investment. In the latter, he receives no direct monetary remuneration, but he does obtain certain housing services referred to as “im puted rent”— the value of these services if the investor were renting the house.8 If the expected annual interest income equals the imputed rent (and if neither in vestment is appreciating in value), tax considerations would induce the investor to purchase the house rather than the securities, because the income earned from the house is untaxed. Furthermore, inflation drives nominal interest rates up so that the interest income from securities increases (relative to interest income in noninflationary situa tions). This raises the tax burden on securities, mak ing the investor worse off. Thus, a rise in the inflation rate increases the relative attractiveness of imputed income versus income from securities. Finally, the U.S. tax structure is such that, during periods of high inflation, corporations are penalized with higher tax bills, while no such effect occurs on housing investments. Thus, individuals become wary of investing in corporate stocks. Corporations are affected because the depreciation of their assets is based on historic cost, and their inventories are valued by first-in - first-out (FIFO ) inventory accounting. With respect to depreciation, present tax accounting practices do not write off capital expenses rapidly enough. In an inflationary environment, the dollar value of depreciation for a machine should represent both the physical deterioration of the machine, and the fact that it will take more dollars in the future to re place the machine or any of its parts. Present depre ciation practices fail to recognize this latter element of depreciation and, as such, overstate corporate profits. With corporations paying more in taxes, the return to equity holders is reduced accordingly. A similar over statement of profits results when corporations use FIFO inventory accounting. A number of studies have suggested that these factors have induced investors to divert money from the stock market into the hous ing market, where as noted above, more favorable tax treatment is available.9 The point that the present tax structure favors borrowing has also been made in Lawrence H. Summers, “Inflation, the Stock Market and Owner Occupied Housing” (National Bu reau of Economic Research, Working Paper No. 606, Decem ber 1980). 8See Anthony M. Rufolo, “What’s Ahead for Housing Prices?” Federal Reserve Bank of Philadelphia Business R eview (July/ August 1980), pp. 9-15. 7See Patric H. Hendershott, “Estimates of Investment Functions and Some Implications for Productivity Growth,” in Laurence H. Meyer, eel., T he Supply-Side E ffects o f Econom ic Policy (St. Louis: Center for the Study of American Business and Federal Reserve Bank of St. Louis), pp. 149-65. 9Patric H. Hendershott, “The Decline in Aggregate Share Values: Inflation and Taxation of the Returns From Equities and Owner-Occupied Housing” ( National Bureau of Economic Research, Working Paper No. 370, July 1979) and Summers, “Inflation, the Stock Market and Owner Occupied Housing.” 13 F E D E R A L R E S E R V E BAN K O F ST. LOUIS Many factors have increased the demand for resi dential housing over the last 15 years.10 It is important to recognize that part of this stimulus to demand comes from the favorable tax treatment of housing which has worked to increase the after-tax afford ability of housing from an economic perspective; that is, the relative after-tax price of housing is being re duced by the interaction of inflation and the present tax structure.11 Looking at the ratio of mortgage pay ments to before-tax income, as many homebuyers and lending institutions do, fails to recognize this point. THE COST OF BORROWING As we have seen, housing prices have risen faster than other prices over the last 15 years. However, family incomes have also risen faster than inflation over this same period. In fact, family incomes nearly have kept up with housing prices. In 1965, the ratio of the median-priced house to the median family income was 2.6; in 1980 the ratio had risen only to 3.0. Thus, ignoring tax considerations, the 1980 house would not appear to be substantially more expensive relative to income than it was in 1965. To make this point another way, consider what would result if the 1980 median-income family could purchase the 1980 median-priced house, at the 1965 mortgage rate. If this family bought a 1980 medianpriced home, but borrowed 80 percent of the purchase price at a mortgage rate equal to the 1965 average of 5.81 percent, its principal and interest payments would have been only 17.0 percent of the median family income. Thus, the 1980 median-income family could well afford the 1980 median-priced house, if only they could obtain a 5.81 percent mortgage rate. This hypothetical case is clearly unrealistic, but it does suggest that a major culprit in the 1980 afforda bility problem is today’s high level of mortgage rates. At a mortgage rate of 13.73 percent, today’s homebuyer would be paying more than 33 percent of his current income in terms of interest and principal alone. 10In addition, other factors have retarded the supply of hous ing. Bechter, “How Much For a New House? ’ p. 13, sees government regulations “as being directly or indirectly re sponsible for holding back the rate of increase in the pace of homebuilding during the rising portion of the last housing cycle.” In this light government policies have also increased the relative price of housing by imposing stringent zoning codes and subdivision regulations. 1:lThis point has led Downs, “The Low (Real) Cost of Hous ing,” to suggest that the United States is overinvesting in housing. Also, see Hendershott, “Estimates of Investment Functions.” 14 JUNE/JULY 1981 But today’s high level of mortgage rates, in and of itself, is not the problem. Though the mortgage rate was quite high in 1980, it is unlikely that this rate has substantially reduced the long-run economic in centive to own a home. Quoted mortgage rates are nominal rates. Nominal rates alone, however, have little influence on an individual’s purchasing or invest ment decisions. Both tax considerations and antici pated future inflation influence these decisions. When the anticipated inflation rate and the favor able tax treatment given to housing are taken into account, homebuying is not nearly as adversely af fected by high nominal mortgage rates as might first be thought. As we have seen, the ability to write off interest expenses reduces the true interest costs asso ciated with purchasing a home. Last year, for example, a median-income family of four was in the 24 per cent marginal tax bracket for U.S. income tax pur poses. After deducting interest expenses from the pur chase of a new home, the family’s after-tax mortgage rate was reduced (at the margin) from the market rate of 13.73 percent to the net rate of 10.43 percent [13.73 X (1-0.24)]. The 1965 median-income family, on the other hand, paid a 4.71 percent [5.81 X (1 0.19)] marginal after-tax mortgage rate. Further, when inflation expectations are considered, this after-tax rate of 10.43 percent in 1980 may not be all that high. Nominal interest rates are high today, when compared to those in 1965, because investors anticipate a higher future inflation rate than they anticipated in 1965. As such, they recognize that the dollars which will be paid back in the future will buy fewer goods, and they demand compensation accordingly. Borrowers, also anticipating inflation, rec ognize they will be paying back the loan with a depreciated currency and thus do not find high nomi nal interest rates prohibitive. For example, take the 1965 median-income family who must pay 4.71 percent after taxes to borrow at the 1965 mortgage rate. Suppose this family antici pates that future inflation will be 2 percent per year. If this family borrows $100 for a year, they would pay back, after tax deductibility is allowed, $104.71 at the end of one year. Since they expect 2 percent inflation, however, they see the foregone $104.71 as equivalent to giving up $102.66 ($104.71/1.02) in present dollars. Thus, the real after-tax interest rate is only 2.66 percent — this 1965 homebuyer expects to give up only $2.66 worth of goods and services to borrow $100. How much inflation must today’s homebuyer antici FE DE RAL . R E S E R V E BA N K O F ST. L O U IS pate to make them indifferent between the present arrangement and that of 1965? The after-tax mortgage rate for the 1980 median-income family is 10.43 per cent. Thus, after tax deductions, the family will pay $110.43 to borrow $100 for one year at the 1980 mort gage rate. If the family anticipates inflation at 8 per cent over the next year, they see the $110.43 given up at the end of one year as equivalent to $102.25 ($110.43/1.08) in 1980 dollars. The real after-tax rate is 2.25 percent. Thus, the 1980 median-income family expecting the future inflation rate to be 8 percent or more anticipates lower after-tax real borrowing costs than the 1965 median-income family that expected a future inflation rate of 2 percent. If individuals expect inflation to continue at recent levels, the 10.43 percent after-tax rate the medianincome family must pay for a 13.73 percent mortgage represents a relatively small cost in terms of the real goods and services that must be given up. It is un likely then that the recent high nominal mortgage rates alone have significantly discouraged home pur chases. Thus, when both taxes and anticipated future inflation are taken into account, the after-tax real cost of the mortgage is apparently not unduly prohibitive. THE CONVENTIONAL MORTGAGE If, as has been argued, neither 1980 housing prices nor 1980 mortgage rates are too great a burden for prospective homeowners, what has caused the signifi cant increase in the ratio of mortgage payments to income? The answer lies in restrictions resulting from conventional mortgage agreements. Conventionally, mortgage debt is amortized over the repayment period, usually 25 to 30 years, so that the periodic payment is fixed, and both the principal and interest are paid off by the end of the loan. One of the main features of the conventional mortgage is that it fixes the periodic payments in dollar terms for the duration of the loan. This feature was useful in a noninflationary environment, but is it when future inflation is expected? Consider two hypothetical cases in which a family with the median income in 1980 borrows $51,920 to purchase the median-priced house.12 In the first case, suppose the family (and the rest of the public) antici pates no inflation in the future, expecting the prices of goods and services to remain essentially unchanged. The family realistically expects its income to rise, but this expectation is based on anticipated productivity gains, not inflation. As such, the increase in expected 12We will ignore all tax considerations in the following analysis. JUNE/JULY 1981 future income implies an increased future command over goods and services. Assume the family expects their income to rise at an annual rate of 3 percent. Similarly, assume the family can borrow at a 3 per cent rate. In the second case, suppose the family (and the rest of the public) anticipates a steady 8 percent rate of inflation for 30 years. We assume that in every other way this family is similar to the first. Specifically, we assume that the family expects its income to grow in real (inflation-adjusted) terms at a 3 percent rate. This implies that the family expects their dollar income to increase at about an 11 percent rate — 8 percent due to inflation, 3 percent due to real productivity gains. Table 3 lists the two respective dollar income streams that are anticipated in these two situations. While the expected income streams are quite different, each fam ily expects its command over goods and services to be the same under each scenario. In addition, we will as sume in this second case that the family expecting 8 percent inflation can borrow at an 11 percent rate, so that in real terms the cost of borrowing is 3 percent as it was in case one. Thus, we are comparing a family in two different situations that are essentially identical when infla tion is accounted for. Each family starts with the same dollar income and buys the same dollar-priced house. With the passing of every year, each family can buy 3 percent more goods and services than it could the previous year. In addition, the real cost of borrowing is the same in each case: to borrow a dol lar today, each family promises to pay back enough money in one year to buy what $1.03 buys today. The two families should be equally happy. In real terms their situations are identical. But let us con sider what would happen if each family were to obtain a conventional mortgage. In the first case, with the family expecting zero inflation and borrowing at a 3 percent rate, the annual mortgage payment turns out to be $2,627. In the second case, the family expecting 8 percent inflation and borrowing at an 11 percent rate faces a $5,933 annual mortgage payment. Over the full term of the mortgage, the two situations are identical. The significantly higher nominal payment in the second case is due to expected inflation. If, as we assume, inflation turns out to equal the 8 percent rate expected, the interest paid on the second debt over the full 30-year period will buy exactly the same amount of goods and services as in the no inflation case. Thus, in such a case the family is simply com pensating the lender for the eroding value of money and is no worse off in a real sense. 15 JUNE/JULY F E D E R A L R E S E R V E BAN K O F ST. LO UI S 1981 Table 3 Percent of Principal and Interest to Income With and Without Inflation 8 % Inflation No Inflation Y ear Payment M edian Income 1980 $2,627 $21,500 1981 2,627 22,145 Paym ent as a percent of median income 12.2% 11.9 Payment Payment adjusted for inflation Payment as a percent of median income M edian income $5,933 $5,933 $ 21,500 5,933 5,494 23,917 24.8 27.6% 1982 2,627 22,809 11.5 5,933 5,087 26,604 22.3 1983 2,627 23,494 11.2 5,933 4,710 29,596 20.0 1984 2,627 24,198 10.9 5,933 4,361 32,921 18.0 1985 2,627 24,924 10.5 5,933 4,038 36,622 16.2 1986 2,627 25,672 10.2 5,933 3,739 40,738 14.6 1987 2,627 26,442 9.9 5,933 3,462 45,317 13.1 1988 2,627 27,235 9.6 5,933 3,205 50,410 11.8 1989 2,627 28,052 9.4 5,933 2,968 56,076 10.6 1990 2,627 28,894 9.1 5,933 2,748 62,380 9.5 1991 2,627 29,761 8.8 5,933 2,545 69,392 8.5 1992 2,627 30,654 8.6 5,933 2,356 77,192 7.7 1993 2,627 31,573 8.3 5,933 2,182 85,867 6.9 1994 2,267 32,520 8.1 5,933 2,020 95,518 6.2 1995 2,627 33,496 7.8 5,933 1,870 106,255 5.6 1996 2,627 34,501 7.6 5,933 1,732 118,198 5.0 1997 2,627 35,536 7.4 5,933 1,604 131,484 4.5 1998 2,627 36,602 7.1 5,933 1,485 142,262 4.1 1999 2,627 37,700 7.0 5,933 1,375 162,702 3.7 2000 2,627 38,831 6.8 5,933 1,273 180,990 3.3 2001 2,627 39,996 6.6 5,933 1,179 201,333 3.0 2002 2,627 41,196 6.4 5,933 1,091 223,969 2.7 2003 2,627 42,431 6.2 5,933 1,010 249,132 2.4 2004 2,627 43,704 6.0 5,933 936 277,135 2.1 2005 2,627 45,015 5.8 5,933 866 308,289 1.9 2006 2,627 46,366 5.7 5,933 802 342,939 1.7 2007 2,627 47,757 5.5 5,933 743 381,486 1.6 2008 2,627 49,189 5.3 5,933 688 424,359 1.4 2009 2,627 50,665 5.2 5,933 637 472,060 1.3 This, however, is a long-run perspective. In the short run (less than the full term of the mortgage), as table 3 indicates, the two families are treated very differently. Specifically, the proportion of income spent on the mortgage when inflation is expected to be 8 percent is much larger in the early years of the mortgage. For example, in the first year the mortgage payment is 27.6 percent of the family income with 8 16 percent expected inflation, as opposed to 12.2 percent of the family income in the zero anticipated inflation case. The explanation for this is simple enough. When everyone anticipates inflation, the borrower must not only repay the principal and interest, but must also compensate the lender for the eroding value of money. F E D E R A L R E S E R V E B AN K O F ST. L OUI S Since, under the conventional mortgage, the periodic payment is fixed in nominal terms, the borrower must compensate the lender early in the repayment period for inflation expected to occur many years down the road. However, the ratio of mortgage payment to family income falls quite rapidly as the second fam ily’s income increases because of productivity gains and inflation, so that the very high ratio early in the mortgage is counter-balanced by a lower ratio later on. The conventional mortgage thus treats the homebuyer very differently depending on anticipated infla tion. In a noninflationary environment, the family in come is expected to be relatively stable in dollar terms, and the mortgage payment plan is in complete agreement with such an expectation. However, in an inflationary environment the family expects its income to rise with inflation; the fixed dollar mortgage pay ment fails to take such an expectation into account. Thus, while nominal interest rates clearly reflect ex pectations of future inflation and require compensa tion accordingly, the payment schedule for a conven tional mortgage does not reflect such expectations. THE MORTGAGE PAYMENT-INCOME RATIO: LOAN CRITERION This example indicates quite pointedly how allow ing for a maximum ratio on the mortgage payment to family income is a dubious rule for the lender to fol low in determining whether or not to make a loan under conventional arrangements. The two families are in exactly the same situation in real terms. The current house prices are the same. The expected real income streams are the same. And the expected real interest rates are the same. Therefore, if the family in the noninflationary environment can buy the house, then the family in the inflationary environment should also be able to buy the same house. Solely considering the ratio of the nominal mort gage payment to income in the first year of the mort gage would suggest that the family expecting 8 percent inflation is less able to afford the house than the family expecting no inflation. However, this problem is a function of the interaction of anticipated inflation and the conventional mortgage; it is unrelated to whether or not the family can ultimately pay off the loan. Surely if the family could “afford” the home in the case of zero inflation, they could “afford” it in the case of 8 percent inflation. In this light, recent increases in the acceptable mortgage payment to fam ily income ratio are not seen as a major problem for the long-run solvency of mortgage lenders, and in fact JUNE/JULY 1981 is a natural response to the interaction of inflation and the conventional mortgage. The Real Mortgage Expenses Table 3 further shows that the family in the infla tionary situation pays a substantially larger inflationadjusted mortgage payment in the early years of the mortgage than the family in the non-inflationary case. In our example, the real payments ( in terms of actual command over goods and services) are higher for the family in the inflationary environment for the first 11 years and lower thereafter. Thus, the early pay ments are high, not only relative to the family in come, but in real terms also. In real terms, the family in the inflationary environ ment is saving more in the early part of the mortgage than the family expecting no inflation. Relative to the family expecting no inflation, the family anticipating 8 percent inflation is postponing consumption in the early years of the mortgage so it can make the high nominal payments. This postponed consumption early in the life of the mortgage is, of course, offset by lower real payments later on. It is important to recognize that this savings de cision was dictated by the interaction of inflation and the conventional mortgage. It may not be a choice that the family would prefer. For example, our family ex pecting 8 percent inflation may not like the idea of spending 27.6 percent of their 1980 income on mort gage payments. But, to the extent that only fixed nominal payment plans are being offered, their choice becomes either to accept this savings schedule or to forego buying the home. In this light, it is entirely likely that we could see a reduced demand for hous ing in periods of high anticipated inflation, as a result of the pattern of real costs imposed by the conven tional mortgage. Families and individuals may forego buying homes in inflationary periods, not because housing is no longer a worthwhile long-term invest ment, but because of the disproportionate real mort gage payments forced on them in early years by the conventional fixed-payment mortgage. THE CONVENTIONAL MORTGAGE AND EXPECTED INFLATION Our example has shown that in the face of expected inflation the conventional mortgage acts to frontend load the mortgage payments both in real terms and relative to a family’s current income. Moreover, as expected inflation accelerates, the front-end load ing problem becomes more severe. The larger the 17 JUNE/JULY F E D E R A L R E S E R V E BAN K O F ST. L OUI S 1981 Table 4 Percent of Principal and Interest to Income Under Different Rates of Inflation 13% Inflation 8% Inflation Paym ent as a percent of median income 27.6% Payment Payment adjusted for inflation Median income Paym ent as a percent of median income 32.1% Y ear Payment Payment adjusted for inflation 1980 $5,933 $5,933 $ 21,500 $6,892 $6,892 $ 21,500 1981 5,933 5,494 23,917 24.8 6,892 6,099 25,024 27.5 1982 5,933 5,087 26,604 22.3 6,892 5,397 29,125 23.7 Median income 1983 5,933 4,710 29,596 20.0 6,892 4,777 33,899 20.3 1984 5,933 4,361 32,921 18.0 6,892 4,227 39,454 17.5 1985 5,933 4,038 36,622 16.2 6,892 3,741 45,921 15.0 1986 5,933 3,739 40,738 14.6 6,892 3,310 53,448 12.9 1987 5,933 3,462 45,317 13.1 6,892 2,930 62,208 11.1 1988 5,933 3,205 50,410 11.8 6,892 2,592 72,403 9.5 1989 5,933 2,968 56,076 10.6 6,892 2,294 84,269 8.2 1990 5,933 2,748 62,380 9.5 6,892 2,030 98,082 7.0 1991 5,933 2,545 69,392 8.5 6,892 1,797 114,159 6.0 1992 5,933 2,356 77,192 7.7 6,892 1,590 131,570 5.2 1993 5,933 2,182 85,867 6.9 6,892 1,407 154,645 4.5 1994 5,933 2,020 95,518 6.2 6,892 1,245 179,990 3.8 1995 5,933 1,870 106,255 5.6 6,892 1,102 209,493 3.3 1996 5,933 1,732 118,198 5.0 6,892 975 243,830 2.8 1997 5,933 1,604 131,484 4.5 6,892 863 283,793 2.4 1998 5,933 1,485 142,262 4.1 6,892 764 330,306 2.1 1999 5,933 1,375 162,702 3.7 6,892 676 384,443 1.8 2000 5,933 1,273 180,990 3.3 6,892 598 447,453 1.5 2001 5,933 1,179 201,333 3.0 6,892 529 520,791 1.3 2002 5,933 1,091 223,969 2.7 6,892 468 606,151 1.1 249,132 2.4 6,892 415 705,484 1.0 0.8 2003 5,933 1,010 2004 5,933 936 277,135 2.1 6,892 367 821,115 2005 5,933 866 308,289 1.9 6,892 325 955,693 0.7 2006 5,933 802 342,939 1.7 6,892 287 1,112,344 0.6 2007 5,933 743 381,486 1.6 6,892 254 1,294,657 0.5 2008 5,933 688 424,359 1.4 6,892 225 1,506,831 0.5 2009 5,933 637 472,060 1.3 6,892 199 1,753,812 0.4 expected rate of inflation, the higher the nominal mortgage rate and the higher the first-year conven tional mortgage payment — both in real terms and relative to family income. Table 4 shows this for a case in which inflation accelerates from 8 percent to 13 percent.13 13At 13 percent inflation, the family’s mortgage payment in 1980 is 32.1 percent of their 1980 income. If inflation was expected to be 40 percent or more, the family’s mortgage payment would have exceeded their income. This feature explains in large part why we had no “affordability” problem from 1976 through 1978.14 14There is cursory evidence that the front-end loading problem was evident in 1974. That year saw a sharp acceleration in both inflation and nominal interest rates. The FHA mortgage rate averaged 9.55 percent that year. As such, conditions were conducive for the imposition of significant real mort gage payments early in the loan. Along these lines it is appropriate to note that 1974 was one of only two years from 1965 to 1980 in which housing prices did not rise as rapidly as consumption goods. F E D E R A L R E S E R V E BAN K O F ST. L OUI S Over this period, both inflation and nominal interest rates were fairly low — suggesting relatively low ex pected inflation. For example, the FHA mortgage rate was below 10 percent from 1976 through 1978. Be ginning in 1979, however, inflation and nominal inter est rates rose sharply. Since the early part of 1979, the front-end loading problem has likely become an important one for homebuyers, especially first-time buyers.15 For previous homeowners, this problem is not as severe since they have realized significant capi tal gains from homeownership which can offset the front-end loading problem. SUMMARY AND CONCLUSION A family with the 1980 median income must pay over 33 percent of their income to buy the 1980 median-priced house. In comparison, the 1965 medianincome family paid less than 15 percent of their in come to buy the 1965 median-priced house. To some 15There is an alternative mortgage arrangement that can be implemented to avoid this problem. For a discussion of such a mortgage, see Donald Lessard and Franco Modigliani, “In flation and the Housing Market: Problems and Potential Solutions,” in Donald Lessard and Franco Modigliani, eds., New M ortgage Designs for Stable Housing in an Inflationary Environment ( Proceedings of a Conference Sponsored by the Federal Reserve Bank of Boston), pp. 13-45; and Henry J. Cassidy, “Price-Level Adjusted Mortgages ( PLAMs): A Comparison with other Home Mortgage Instruments” (Fed eral Home Loan Bank Board, Working Paper No. 90, Janu ary 1981). Note, however, that the variable or renegotiable mortgage rate arrangements will not resolve the current cash-flow prob lems for first-time buyers. JUNE/JULY 1981 extent, this drastic change is due to the fact that housing prices have risen faster than inflation, spurred on by demographic factors and the preferential tax treatment of housing which has accelerated with infla tion. Family incomes, however, have to a large extent kept up with housing prices, so this phenomenon is not as crucial as may first appear. The main culprit in causing the significant increase in the proportion of income a new buyer must pay to purchase a house is the combination of the expec tation of higher future inflation and the conventional mortgage. Expected inflation requires that lenders be compensated for the expected deterioration in the purchasing power of money. Moreover, the conven tional mortgage requires that payments, including those due to future inflationary effects on the value of money, be spread evenly over the duration of the mortgage so that dollar payments are constant. As such, today’s conventional mortgage imposes a signifi cant cash-flow problem for the homebuyer, especially the first-time buyer. It is thus likely that many prospective new home buyers recently have postponed home purchases or have bought a lower-priced home than they originally desired, either because of the significant real costs of the mortgage in the early years, or because of the limitation on mortgage debt to income imposed by credit institutions. In this regard, any actions taken to reduce inflation will benefit the long-term future of the housing industry. 19 Inflation: The Cost-Push Myth DALLAS S. BATTEN I NFLATION continues to be our greatest economic problem. This is not a particularly new revelation — policymakers have called it “public enemy No. 1” at least four times in the past decade. What is puzzling is that inflation has persisted (and worsened) even though its reduction has been a primary goal of both Federal Reserve and administration policy for over 10 years. The explanations for persistent inflation are many: uncontrollably rising wages; OPEC oil-price increases; droughts or poor harvests; large government budget deficits. The list of “causes” of inflation changes with the circumstances. If we were to take them seriously, we would conclude that inflation may be caused by nearly everything. None of these causes, however, can explain inflation consistently over time or across countries.1 This article analyzes a frequently given cause of inflation — cost-push — within a monetary framework. The cost-push view of inflation is based on the notion that prices are set by the costs of production and that prices rise only when costs rise, regardless of demand. Inflation, in this framework, is the result of the sellers of productive inputs (including labor) persistently and unilaterally raising their selling prices, causing producers’ costs, and subsequently prices, to rise. definition must be distinguished from an increase in relative prices (e.g., a rise in the price of wheat or oil) which, as argued below, is not inflation. Some advocates of the cost-push view confuse relative price changes with changes in the overall price level. Conse quently, they view the increase in a particular price as a contributor to inflation when in reality it is not.2 For example, in a study of CBS Evening News broadcasts, 61.5 percent of the reports that dealt with the topic of inflation either explicitly or implicitly identified the rising prices of individual goods as the cause of infla tion. A typical report: “Inflation continued to steam along at a double-digit annual rate. . . . The major factor in the surge continues to be food.”3 In other words, food price increases cause the overall price level to rise. Changes in the prices of individual goods do not cause inflation, although they do affect its measurement.4 Individual price increases accompany increases in the measure of inflation, but tell us little about the cause of inflation. There are an infinite number of individual prices consistent with any given overall price level. At any time, some prices are increasing, some are decreasing, while others remain unchanged. Inflation — a persist -For a more thorough discussion of this point, see Hans H. Helbling and James E. Turley, “A Primer on Inflation: Its Conception, Its Costs, Its Consequences,” this R eview (Janu ary 1975), pp. 2-8. WHAT IS INFLATION? 3Tom Bethell, “TV, Inflation and Government Handouts,” The W all Street Journal, July 8, 1980. Inflation is a persistent rise in the overall ( or aver age) level of prices of all goods and services. This 4Since there are many prices in an economy and since these prices do not necessarily move together, some type of price index must be constructed in order to capture changes in the general level of prices (the overall price level). Two of the most popular price indices are the consumer price index and the implicit GNP deflator. For a discussion of the problems associated with measuring the overall price level, see Denis S. Karnosky, “A Primer on the Consumer Price Index,” this R eview (July 1974), pp. 2-7. iSee, for example, Scott E. Hein, “Deficits and Inflation,” this R eview (March 1981), pp. 3-10; and Michael Parkin, “Oil Push Inflation?” Banca Nazionale del Lavoro Quarterly R e view (June 1980), pp. 163-86. 20 F E D E R A L R E S E R V E BAN K O F ST. LOUIS ent rise in the overall price level — can be detected only by observing changes in an aggregate measure of prices, not by changes in individual prices. Since inflation is a continuous rise in the average price level, a one-time increase caused by some ran dom shock (e.g., a drought or a reduction in the quantity of oil supplied by OPEC) is not considered inflation. Of course, this one-time increase will result in a higher overall price level, but the rate of increase of the overall price level (i.e., the rate of inflation) will be unaffected if the economy adjusts to this shock immediately. Consider the example in figure 1. Over time, the overall price level is rising at a rate equal to the slope of line AB. (This rate of price increase is usually called the trend or underlying rate of inflation.) At point t0, the trend is interrupted by the occurrence of a random shock (e.g., OPEC na tions reduce their rate of supplying oil). It the econ omy adjusted to this shock instantaneously, the over all price level would increase (from B to C ), but the trend rate of inflation would be unaffected. (The slopes of AB and CE are identical.) However, such adjustments are not instantaneous. During the adjust ment period (t0 to tj), the overall price level will rise at a rate (the slope of BD) that is greater than the trend rate, giving the appearance that the shock has actually increased the rate of inflation. This higher rate of price change during the adjustment period is not a continuing phenomenon, however, but simply a transitory deviation of the rate of inflation from its trend. Since these deviations do not persist, they are not considered inflation.5 JUNE/JULY 1981 F ig u r e 1 E ffe c t of Tra nsito ry N o n m o n e ta ry Shock on the T r e n d R a te o f Inflation PA N E L B WHAT CAUSES INFLATION? As noted above, considerable confusion exists about the relationships among changes in individual prices, random shocks and the cause of inflation. Political leaders attempt to persuade us that inflation is caused primarily by either random shocks or greedy busi nesses and labor unions raising their prices and wages unilaterally. As we have seen, the random shock argu ment is fallacious. Placing the blame for inflation on business and labor is the central tenet of the cost-push 5Of course, these random shocks do cause the prices of some commodities and consequently the overall price level to rise. Other things equal, individuals will experience a decline in their purchasing power. However, these shocks are typically a temporary phenomenon and, by definition, uncontrollable. To place the blame for persistent price level increases (i.e., infla tion) on continually occurring random shocks is, in essence, contending that inflation is uncontrollable. This is an undesir able approach, for if inflation is ever to be eliminated, it must be considered a result of controllable events. argument. This argument typically holds that busi nesses continually raise their prices in an effort to earn higher profits. Presumably, their ability to do this successfully stems from monopoly power. A similar argument can be made for labor unions. Specifically, unions are alleged to exercise some mo nopoly power in labor markets to procure wage in creases for their members greater than those dictated by market conditions. Then, the firms that employ these workers must raise their prices in order to cover these labor costs. Once this occurs, union members realize that their increased wages do not buy as many goods and services as they did before. As a result, they ask for another raise. This continuing scenario is the familiar “wage-price spiral.” 21 FEDER AL. R E S E R V E BAN K O F ST. L OUI S These explanations of inflation conveniently ab solve government from having any role in creating inflation. The “culprits” are identified by observing which components of the overall price level rise the most at any particular time. Needless to say, the list of those contributing to inflation quickly becomes quite large: farmers (rising food prices), participants in financial markets (rising interest rates), foreigners (rising oil prices), etc. The public then believes that almost everyone is responsible for inflation, and a myriad of government agencies are formed to regu late prices in various markets, protecting some people from the presumed excesses of others. The Council on Wage and Price Stability is one such example. Money and Inflation To understand the fallacy of the cost-push argu ment, the actual cause of inflation must be identified. The ultimate source of inflation is persistent excessive growth in aggregate demand resulting from persistent excessive growth in the supply of money. This isn’t a particularly novel idea — eighteenth century econ omists aptly described inflation as the result of “too much money chasing too few goods;” that is, the overall price level in any economy is determined by the relationship between the demand for and the supply of money. In particular, it depends on the supply of money relative to the amount that individ uals desire to hold. The quantity of money supplied is essentially a policy variable controlled by the monetary authority, the Federal Reserve System in the United States. The Fed can affect the stock of money either by changing the fraction of commercial bank and thrift institution deposits that must be held in reserve ac counts with the Fed or by directly changing the level of reserves in these accounts. The Fed most frequently employs the latter method, participating in the gov ernment securities markets. Specifically, when it wants to inject reserves, it buys government securities; when it wants to drain reserves, it sells government securities. The demand for money is the individual’s desire to hold a portion of his wealth in the form of money. In the aggregate, it is determined by permanent in come (the expected flow of income over one’s life time), interest rates, prices and price expectations. An increase in permanent income motivates individ uals to demand a larger stock of money. An increase Digitized for22 FRASER JUNE/JULY 1981 in permanent income results in an increase in wealth, other things equal. Since individuals want to hold a certain percentage of their wealth in the form of money, they will add to their money balances (i.e., demand more money) as their permanent income rises in order to maintain the desired relationship between money and wealth. The interest rate is the opportunity cost of holding money, the income fore gone by holding money instead of an interest-earning asset. As interest rates rise, holding money becomes relatively more costly; consequently individuals hold smaller money balances. The demand for money is positively and proportionately related to the overall price level. For example, if prices double, individuals will hold twice as much money since it will take twice as many dollars to conduct any real transaction. Finally, rising prices erode the purchasing power of the money held by individuals. If expectations of future inflation rise, individuals will attempt to hold less of their wealth in the form of money and more in some asset that will maintain its value in terms of other goods as prices rise (e.g., land or gold). The equilibrium overall price level is the one ( given the level of permanent income, interest rates and price expectations) that induces individuals to hold the exact quantity of money that the monetary authority supplies. Any other price level will motivate individ uals to demand more or less money than is being supplied. If individuals are satisfied with the amount of money that they are holding, they will have no desire to increase or decrease their spending on goods and services; in other words, they are in equilibrium and the existing price level is the equilibrium one. If the money supply changes, other things equal, in dividuals will alter their spending in order to reach equilibrium again and, consequently, the price level will change. For example, if the amount of money supplied is greater than the amount that individuals desire to hold, an excess supply of money exists. In dividuals will attempt to rid themselves of the excess money by increasing their purchases of goods and services. Thus, the existence of an excess supply of money necessarily implies a corresponding excess de mand for goods and services. As individuals increase their spending, they bid up the prices of goods and services. This rise in the price level continues until individuals are motivated to hold the existing stock of money supplied by the monetary authority, that is, until equilibrium is regained. If the monetary authority continues to supply more money than is demanded, excess aggregate demand will persist and prices will continue to rise. Thus, inflation is the result of a persistent excess supply of money. F E D E R A L R E S E R V E B AN K O F ST. LO UI S The link between money and inflation is not con fined to the United States. In fact, it is the “tie that binds” the inflationary experience of the industrialized world during the past decade.6 Table 1 provides a cross-country comparison of the rate of money growth and inflation over the 20-quarter period from IV/1975 to IV/1980 for the major industrial nations.7 The countries are ranked in descending order accord ing to the rate of money growth experienced during the period. If the demand for money is relatively stable across countries, the analysis above predicts a positive relationship between money growth and inflation. This relationship can be clearly identified in the table. In particular, Italy had the highest rate of money growth and the highest rate of inflation; the United Kingdom experienced the second highest growth rates of money and prices, and so forth.8 In fact, if this comparison is continued, only West Ger many violates the ordering of inflation with the rate of money growth. These results are extremely robust when one considers the heterogeneity of this group of countries. The Cost-Push Myth Though the cost-push argument is appealing on the surface, neither economic theory nor empirical evidence indicates that businesses and labor can cause continually rising prices. All firms, regardless of the degree of competition in their industry, produce a quantity and charge a price that they expect will yield the highest profit. This price is higher in a more monopolistic market than in a more competitive one. If a firm with some monopoly power chooses to raise its price arbitrarily, the quantity that it can sell will decrease — since a monopolist faces a downward-sloping demand curve — and its profits will fall. Conse quently, since profits would actually fall as prices are arbitrarily increased, a monopolist has no incentive to raise its price continually.9 A monopolist may charge • ’For additional support, see “Inflation and money — the tie that binds,” Citibank Monthly Econom ic L etter (December 1980), pp. 8-11. 7The choice of a 20-quarter period is supported by evidence presented by Denis S. Kamosky, “The Link Between Money and Prices — 1971-76,” this Review (June 1976), pp. 17-23; and Albert E. Burger, “Is Inflation All Due to Money?” this R eview (December 1978), pp. 8-12. 8The Spearman rank correlation coefficient with Germany in cluded is .829; the calculated value when Germany is excluded is .997. The critical values are .700 and .738, respectively; that is, the hypothesis that money growth and inflation are unrelated is rejected for both cases. 9In fact, one study has demonstrated that prices in highly con centrated industries increased less rapidly during the period 1954-73 than did other prices. See Steven Lustgarten, lndus JUNE/JULY 1981 Table 1 Money Growth and Inflation in the Major Industrial Nations (IV/1975-IV/1980) Annual rates of money growth1 Annual rates of inflation2 Italy 20.5% 17.1% United Kingdom 12.3 13.7 France 10.0 10.7 7.8 4.1 United States 7.5 9.1 Canada 7.5 9.0 Japan 7.2 6.3 Netherlands 6.8 5.8 Switzerland 5.3 2.5 Country West Germany X M1 for all countries except the United States for which M1B is used. 2Consumer price index used as a measure of inflation. higher prices than a competitive firm, but this does not imply constantly rising prices. Unfortunately, realizing that monopolies (which wish to maximize their profits) cannot unilaterally contribute to inflation is insufficient to lay this argu ment to rest. A similar argument has been developed based on changes in the degree of competition within markets. Since monopolies do charge higher prices than competitive firms, prices will continue to rise if the economy becomes less and less competitive. In other words, it is often argued that inflation is the result of the acquisition of additional market power by the firms within the economy. If the economy is becoming less and less competitive, then the con tinually declining rate of growth of real output that results will cause prices to rise; that is, inflation caused by the acquisition of more and more monopoly power must be accompanied by less and less output being produced and sold. Chart 1 contains a com parison of a trend rate of inflation (as measured by the consumer price index) with a trend rate of growth of real output (real gross national product). Since the trend rate of growth of real output does not show a continuously decreasing pattern, the hypothesis that increased monopolization has caused the rising in flation during the past decade can be rejected. trial Concentration and Inflation ( American Enterprise Insti tute for Public Policy Research, 1975), pp. 25-29. 23 F E D E R A L R E S E R V E B AN K O F ST. LO UI S JU NE/JULY 1981 C h a rt 1 Trend Rates of Inflation a n d O u tp u t G ro w th Percent --- 110 1 1965 1966 1967 19 68 19 69 1970 1971 19 72 1973 1974 1975 1976 19 77 19 78 1979 1980 1981 Sources: U.S. D e p a rtm e n t o f L a b o r a n d U.S. D e p a r tm e n t o f C o m m e rc e Q T r e n d s a r e 2 0 - q u a r t e r m o v in g a v e r a g e s . In fla tio n is m e a s u r e d b y th e C o n s u m e r P ric e In d e x . The cost-push argument is even less credible when analyzed in a macroeconomic framework. In particu lar, other non-monopolized sectors of the economy adapt to the exercising of monopoly power in one sector. As a result, they tend to neutralize the monop oly’s impact on the entire economy. To understand this more clearly, assume that the union in industry A succeeds in obtaining a wage increase for its mem bers that is higher than that dictated by market con ditions (i.e., the demand for A’s product and the productivity of the workers in A). As a result, the firms in A raise their prices in an attempt to cover the increased labor costs.10 Other things equal, these higher prices cause the overall price level to rise. Be cause of this price increase, individuals in the aggre gate demand larger money balances. If the money supply remains unchanged, however, there is no ad 10It should be noted that an increase in wages need not be the motivation for higher prices; higher prices could have re sulted from the firms in A exercising their monopoly power. The crucial point is that prices in A have risen unilaterally, independent of market conditions. 24 ditional money for them to hold. Consequently, in order to increase their balances to the new desired level, they must decrease their spending on goods and services. This decreased aggregate demand will ultimately cause prices in other industries to fall until the over all price level returns to what it was prior to the wage increase. The price level must return to its original value because, other things equal (especially the money supply), it is the only price level at which the quantity of money supplied equals the quantity demanded. The wage increase in A has induced higher prices in A, but lower prices in other industries. The union s action has caused relative prices to change, but has not affected the overall price level.11 u The inability of a labor union (that doesn’t represent the entire labor force) to affect the overall price level can be seen through the quantity equation: MV = PQ, where M is the money stock; V is the velocity of money (i.e., the average number of times that the money stock F E D E R A L R E S E R V E B AN K O F ST. L O U IS Critics of the above scenario state that “nowadays, . . . compensatory price declines tend not to occur.”12 As a result, they conclude that “the rules of economics don’t seem to be working any more.”13 The rules of economics, however, always work despite attempts to frustrate them. The point missed by these critics is that the monetary accommodation of a price shock prevents the occurrence of a compensatory price de crease. In the scenario above, prices in other industries fell because the money stock was held constant. This price adjustment does not occur immediately. During the adjustment period, the cost of adjusting is reflected by reduced output. If the monetary au thority confuses this loss of output (and the cor responding decline in employment) with a permanent decline in aggregate demand, he may increase the money supply. This then precludes the compensatory price declines that one expects to observe in other industries. The price level does not return to its original level and the success of the labor union in industry A in obtaining a higher than warranted wage increase for its members is termed a cause of inflacirculates within the economy during a year); P is the over all price level; and Q is real gross national product (GNP). Using the quantity equation, the overall price level can be determined as follows: p = -MX_ Q ' Suppose that there are only two industries (A and B) in this economy. Labor in A is unionized; labor in B is not. In this simple world, the overall price level and real GNP can be rewritten as: P = W aP a + wB Pb Q = Qa + Qb, where PA and PB are the prices in industries A and B, re spectively; W a and w b are the percentages of the average consumer’s consumption bundle composed of A’s output and B’s output, respectively; and Q and Q are the output of A and B, respectively. If the action of the union in A causes wages and prices in A to rise, then the overall price level must also rise, other things equal. Since nothing has occurred that causes M or V to change, the new, higher P is consis tent with the quantity equation only if Q declines. Total out put (Q ) must decline because Qa decreases as consumers react to the higher Pa by moving up their demand curves for A. This new situation, however, cannot be one of equilibrium because there are unemployed workers that are willing to work at the current market wage. The union’s action has precluded their employment in A; consequently, these workers must search for work in B. As they search for employment in B, wages in B decline, causing PB to fall and Qb to rise until there are no unemployed workers at the current wage. Since none of these occurrences change the equilibrium num ber of employed workers ( economy-wide) or the relationship between the number of workers and the quantity of output produced economy-wide, this equilibrating process must con tinue until Pb has decreased (and Qb has increased) suffi ciently for the overall price level ( P ) and real output ( Q ) to return to their original levels. a b 12“Needed: A New Perspective on Inflation,” The Morgan Guaranty Survey (November 1980), p. 2. 13Ibid. JU NE/JULY 1981 tion. In fact, the actual cause of inflation has been the accommodation on the part of the monetary authority, not the monopolist, labor union, or an in herent price rigidity built into the economy. It is difficult to support the cost-push hypothesis. Gordon, in a study of inflation in the United States, Canada, France, West Germany, Italy, Japan, Sweden and the United Kingdom for the period 1958-76, could find no support for the wage-push hypothesis; “The wage-push hypothesis appears to be alive and well as an explanation of wage rates, but not as a theory of inflation or of monetary growth.”14 In an analysis of post-World War II inflation in the United States, Barth and Bennett concluded that “there is evidence of unidirectional causality that runs from consumer prices to wages.”15 In other words, higher wages do not lead to higher prices as the cost-push hypothesis predicts; instead, higher prices lead to higher wages. The Cost-Push Illusion If cost-push inflation is really a myth, why do consumers hear businessmen rationalize their price increases with: “I have to raise my price because my costs have risen.” Are businessmen simply trying to pass the buck? No, most businessmen (especially those operating relatively small businesses) believe that higher costs of production are the motivation for their raising prices. They seldom identify the real cause — increased aggregate demand resulting from increased money growth. The translation of increased aggregate demand into higher prices is frequently concealed in the marketplace by the existence of inventories. As a result, a “cost-push illusion” is created.16 No merchant sells his product at a constant rate; sales in some time periods are larger than normal, while sales in other time periods are smaller. In order to hedge against running out of their product dur ing periods of larger than normal sales, merchants 14Robert J. Gordon, “World Inflation and Monetary Accommo dation in Eight Countries,” Brookings Papers on Econom ic Activity (2 : 1977), p. 433. Since changes in wages are the predominant causes of changes in costs of production, test ing the wage-push hypothesis is tantamount to testing the cost-push hypothesis. 15James R. Barth and James T. Bennett, “Cost-push versus Demand-pull Inflation: Some Empirical Evidence,” Journal o f Money, C redit and Banking (August 1975), p. 397. 16This phrase was coined by Armen A. Alchian and William R. Allen in University Econom ics, 3rd. ed. ( Wadsworth Publish ing Company, Inc., 1972), p. 95. This discussion follows theirs. 25 F E D E R A L R E S E R V E B AN K O F ST. L OUI S typically hold inventories (or buffer stocks). If ag gregate demand increases, merchants cannot immedi ately distinguish this phenomenon from a period in which sales are temporarily above normal; that is, they do not realize immediately that they could raise their price and still make the normal amount of sales. Consequently, they will not raise their price im mediately, but instead, will draw down their in ventories held for such an occasion as this. If these higher than normal sales persist, merchants will increase their purchase rate from suppliers in order to maintain their inventories at the desired level. The firms that supply these merchants thus will ex perience higher than normal rates of sales, and their inventories will be depleted more rapidly than desired, motivating them to increase the rates at which they purchase from their suppliers. This process continues filtering down the network of markets until it finally reaches the market of raw materials (the primary inputs used to produce this commodity). In the raw materials markets, the amount available is insufficient to meet the increased amount demanded at the old price.17 Since aggregate demand has increased (not just the demand of one or a few manufacturers), all manufacturers want additional raw materials. As a result, all offer higher prices to suppliers until the price of raw materials is bid up enough to clear the market. Because the higher price for raw materials increases their cost of produc tion, manufacturers will charge wholesalers a higher price for their product, citing increased raw material costs as the reason. Wholesalers will say that the in creased manufacturers’ price makes it necessary to charge retailers a higher price. And finally, the re tailer (merchant), being completely truthful, will tell the consumer that he must charge a higher price because his costs have risen. 17That is, existing inventories of raw materials are insufficient to meet the increased demand. 26 JUNE/JULY 1981 Though it appears that increased raw material costs have caused a higher final product price, the actual cause of the higher prices at every level of the manufacturing and distribution network is the initial increase in aggregate demand for the final product. The price increase is delayed until the impact of the increased demand reaches the raw materials market by the existence of inventories at each level that are sufficient to buffer transitory, but not permanent, changes in demand at each level. SUMMARY AND CONCLUSION The focus of this paper has been to separate the cost-push myth from the reality of inflation. The costpush argument views inflation as the result of con tinually rising costs of production — costs that rise unilaterally, independent of market forces. Such an hypothesis (1) confuses changes in relative prices with inflation, a continuously rising overall level of prices, and (2) neglects the role that the money supply plays in the determination of the overall price level. The idea that greedy businesses and/or labor unions can cause a continual rise in prices cannot be supported by either the conceptual development or the empirical evidence provided. Alternatively, the hy pothesis that inflation is caused by excessive money growth is well supported. In the major industrial countries, those with the highest rates of inflation have the highest rates of money growth, and vice versa. Consequently, inflation cannot be eliminated by attacking those sectors of the economy that have experienced the most rapid increase in prices, by imposing wage and price controls, or even by em ploying some type of tax-based incomes policy. Inflation will be eliminated only when the long-term rate of money growth is approximately the same as the long-term rate of real output growth. F E D E R A L R E S E R V E BAN K O F ST. L O U IS JU NE/JULY 1981 “The Supply-Side Effects of Econom ic Policy 3 9 Single copies of this publication, the proceedings of a conference co-sponsored by the Federal Reserve Bank of St. Louis and the Center for the Study of American Business, Washington University, are available in limited supply to our readers. If you are interested in obtaining a copy, please address your request to Editor, Review, Federal Reserve Bank of St. Louis, P.O. Box 442, St. Louis, Missouri 63166. 27