The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
ISSN 0007-7011 Federal Reserve Bank of Philadelphia Ten Independence Mall Philadelphia, Pennsylvania 19106 MAY/JUNE 1987 The CPI Futures Market: The Inflation Hedge That Won't G ro w ................ 3 Brian R. Horrigan Several new financial instruments designed to hedge the value of investments against a variety of uncertainties have been introduced into the markets, and many have flourished. One instrument that has not had much success, however, is the CPI futures—a futures contract whose value is based solely on the Consumer Price Index, the most widely quoted measure of inflation in the U.S. Some economists have hailed it as a tool to allow households, businesses, and investors to shed their inflation risk at low cost; some even have predicted it could become the largest-volume contract in the country. Yet, the daily volume of contracts traded so far is minuscule, compared to other futures contracts. When such a good idea fails in practice, several factors are probably involved, but perhaps the overriding factor is that inflation risk itself has not been a serious concern in recent years. Explaining Long-Term Unemployment: A New Piece to An Old Puzzle . . . 15 Robert H. DeFina According to some theories about labor markets, long-term unemployment simply shouldn't happen. If the labor market is a typical auction market, people will bid for jobs, and employers will hire the lowest bidders. Therefore, whoever wants to work should be able to find a job simply by bidding the lowest wage. A new view suggests that something else is going on inside labor markets that prevents potential workers from landing employment this way. The "efficiency wage hypothesis" argues that employers may not find it most profitable to take the lowest bids on wages. Instead, they may find that by paying higher wages, they may be getting more productivity, and therefore, more profit, out of their workers. The BU SIN ESS REVIEW is published by the Department of Research every other month. It is edited by Judith Farnbach. Artwork is directed by Ronald B. Williams, with the assistance of Dianne Hallowell. The views expressed herein are not necessarily those of this Bank or of the Federal Reserve System. The Review is available without charge. Please send subscription orders, changes of address, and requests for additional copies to the Department of Research at the above address or telephone (215) 574-6428. Editorial communications also should be sent to the Department of Research or telephone (215) 574-3805. The Federal Reserve Bank of Philadelphia is part of the Federal Reserve System—a System which includes twelve regional banks located around the nation as well as the Board of Governors in Washington. The Federal Reserve System was established by Congress in 1913 primarily to manage the nation's monetary affairs. Supporting functions include clearing checks, providing coin and currency to the banking system, acting as banker for the Federal government, supervising commercial banks, and enforcing consumer credit protection laws. In keeping with the Federal Reserve Act, the System is an agency of the Congress, in dependent administratively of the Executive Branch, and insulated from partisan political pressures. The Federal Reserve is self-supporting and regularly makes payments to the United States Treasury from its operating surpluses. The CPI Futures Market: The Inflation Hedge That Won't Grow Brian R. Horrigan* Twenty years of persistent inflation in the United States have brought many proposals for coping with inflation. On June 21, 1985, the Coffee, Sugar, & Cocoa Exchange in New York offered a new way: a futures contract whose value is based solely on the Consumer Price Index (CPI), the most widely quoted measure of *Brian R. Horrigan, now with Chase Econometrics, prepared this article while he was a Senior Economist in the Research Department of the Federal Reserve Bank of Philadelphia. inflation in the U.S. This newly authorized futures contract provides investors and busi nesses a means of hedging against inflation risk, that is, the uncertainty caused by inflation. Until the CPI futures contract was invented, there was no direct means to hedge against inflation risk. While some so-called inflation hedges—such as real estate, precious metals, or stocks—may offer some long-run, imperfect protection against infla tion, they are often highly risky investments in the short run. But the final value of a CPI futures contract is determined solely by the actual value 3 BUSINESS REVIEW of the CPI, which allows investors and businesses to trade on and hedge against pure inflation risk. Academic economists have been urging the creation of a CPI futures market for over a decade, ever since economists Michael Lovell and Robert Vogel first proposed the idea in 1973.1 The benefits of having such a futures market have been noted by many eminent economists, includ ing Milton Friedman, Paul Samuelson, and Robert Barro. These benefits include giving households and businesses the means to shed their inflation risk at low cost, allowing invest ment decisions to be made on the basis of profitability without regard to possible future inflation, and permitting people to save without the concern that their savings could be hurt by inflation.2 But despite the early enthusiasm of econo mists, the CPI futures market has had very little activity. There are probably several reasons for the low activity. For example, the CPI may be an inadequate measure of inflation for many house holds, while businesses may have other ways to hedge against inflation risk. Furthermore, the CPI futures market, unlike other futures markets, has no underlying asset which is storable or traded on an active spot market, which reduces the opportunities for arbitrageurs and specula tors to participate in the market. Finally, inflation risk has become small compared to other types M. Lovell and R. Vogel, "A CPI-Futures Market," Journal of Political Economy 81 0uly/A ugust 1973) pp. 1009-1012. A futures market trading on the CPI was independently pro posed by L. Ederington, "Living With Inflation: A Proposal for New Futures and Options Markets," Financial Analyst Journal (January/February 1980) pp. 42-48, and Milton Friedman, "Financial Futures Markets and Tabular Stan dards," Journal of Political Economy 92 (February 1983) pp. 165-167. 2These eminent economists gave the CPI futures market enthusiastic endorsements in: Milton Friedman and Rose Friedman, Tyranny of the Status Quo, (San Diego, CA: Harcourt Brace, 1984); Paul Samuelson, Economic Index Futures: An Introduction to the Concept of Shifting Macroeconomic Risk, Coffee, Sugar, & Cocoa Exchange, Inc., June 1 4 ,1 9 8 3 ; and Robert Barro, "Futures Markets and the Fluctuations in Inflation, M onetary Growth, and Asset Returns," Journal of Business 59 (April 1986) pp. S21-S38. 4 MAY/JUNE 1987 of price risk for many businesses. So, though it may be a good idea in theory, in practice the CPI futures market seems unlikely to attain greater volume unless inflation risk becomes much more significant than it is currently. INFLATION RISK A typical contract in this country—be it a financial contract such as a loan, a labor contract specifying an hourly wage rate, or a standard business contract requiring payment for work completed—is written in nominal terms, prom ising the payment of cash by one party to another in the future. Inflation—the increase in the average level of prices of goods and services— reduces the purchasing power of money over time. Businesses or investors who make con tracts involving the future payment or receipt of cash are subject to inflation risk if they cannot precisely predict the future price level.3 Those making contracts try to protect themselves against inflation by negotiating their contracts in light of their anticipations of future inflation. If everybody who agrees to a contract correctly anticipates the inflation, everybody will get what he expected in real terms. But if the inflation turns out dif ferently from what two parties anticipated when they agreed to a contract, one of the parties loses wealth in real terms while the other gains. Consider, for example, a financial arrange ment in which someone borrows at a nominal interest rate of 9 percent, while the inflation rate is 6 percent. In that case, the real interest rate is 3 percent— 9 percent more dollars buys 3 percent more goods and services after the price level rises by 6 percent. The inflation rate might, however, turn out as high as 10 percent or as low as 2 percent, even though the lender's best guess for the inflation rate is 6 percent. If the inflation rate turns out to be 10 percent, the lender gets a real interest rate of —1 percent (the 9 percent 3Inflation risk here means not only that the average price level may rise unexpectedly, but also that it may fall un expectedly. FEDERAL RESERVE BANK OF PHILADELPHIA The CPI Futures Market nominal interest rate that the lender contracts for minus the 10 percent inflation rate.) Alter natively, if the inflation rate turns out to be 2 percent, the lender gets a real interest rate of 7 percent (the 9 percent nominal interest rate minus the 2 percent inflation rate). Even if the nominal interest rate is certain, an uncertain inflation rate produces an uncertain real interest rate. Unanticipated inflation creates redistributions of the real wealth of businesses and investors, and the greater the variance of unanticipated inflation, the greater the redistributions and the greater the inflation risk it creates. The newly developed CPI futures market offers one way for businesses and investors to cope with infla tion risk. THE BASICS OF THE CPI FUTURES MARKET The CPI. The CPI, a monthly measure of the average level of prices of consumer goods and services, generates the most widely quoted measure of inflation in the United States. The CPI is compiled and published by the Bureau of Labor Statistics (BLS), and measures the monthto-month change in the cost of buying a fixed market basket of goods and services, such as food, clothing, shelter, and transportation. In January 1978, the BLS started publishing two versions of the CPI, one new and the other a continuation of the original series which was started in 1919. The CPI futures market uses the original CPI— now called the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W) —since it is the basis for costof-living adjustments in labor contracts, many lease agreements, and Social Security benefit payments.4 4The new version of the CPI, the Consum er Price Index for All Urban Consum ers ( CPI-U), covers everyone covered by the CPI-W plus salaried workers, the self-employed, and those not in the labor force, among others. (In the text, just "C P I" is used for simplicity to represent the CPI-W.) Both price series are very highly correlated, so that there is little significance to the fact that the more narrow definition of the Brian R. Horrigan The CPI-W is released at the end of the third week of each month. The CPI is presented in index form, with the average level of prices in 1967 set at an index value of 100.0. Each CPI announcement reveals the average of prices sampled throughout the preceding month. Once released, the CPI is never revised. The inflation rate is calculated as the percentage change in the CPI. For example, the average CPI-W for 1985 was 323.4 and for 1984 was 312.2, so the inflation rate in 1985 was about 3.6 percent. The CPI Futures Contract. A CPI futures contract is a standardized agreement in which one party to the contract pays the other an amount of money determined by the level of the CPI at the time that the CPI is announced. CPI futures contracts are traded at the Coffee, Sugar, & Cocoa Exchange in New York, which sets certain rules and regulations concerning the trading of the contracts and which standardizes the characteristics of the contracts, such as the size, the method of settlement, minimum price fluctuations, and so forth.5 (See CPI FUTURES CONTRACT FEATURES, p.6, for details on the characteristics of the market.) The Exchange guarantees the performance of each contract traded at the Exchange, so that buyers and sellers need not worry about the creditworthiness of those on the opposite side of the trade. A CPI futures contract amounts to a bet between two people on the future value of the CPI. A useful way to understand how the market works is to "walk through" a typical transaction. Suppose on December 3,1986, you want to buy a CPI futures contract settling in April 1988. You contact your broker to discover the current "price" of that contract. Essentially, the "price" of the contract represents the market's consensus on price index is used in the CPI futures market. For more detailed information about the CPI, see Depart ment of Labor, BLS Handbook of Labor Statistics, (June 1985) and BLS Handbook of Methods, Vol. II (April 1984). 5A good introduction to the subject of futures markets can be found in R. Kolb, Understanding Futures Markets, (Glenview, IL: Scott, Foresman, and Company, 1985). 5 BUSINESS REVIEW MAY/JUNE 1987 CPI Futures Contract Features Exchange The Economic Index Market of the Coffee, Sugar, & Cocoa Exchange at 4 World Trade Center in New York Regulator Commodity Futures Trading Commission Trading Hours 9:30 a.m. to 2:30 p.m., New York time Trading Unit The Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W ), with the value of the index equal to 100.0 in 1967 Settlem ent Day The day that the Bureau of Labor Statistics announces the value of the CPI-W Settlement Price $100 times the value of the CPI-W End of Trading Trading in the futures contracts ends at the close of trading two business days prior to the settlement day Minimum Fluctuation Prices are quoted to two decimal places, and the minimum price fluctuation is .01— one basis point. The dollar value of one basis point is $10.00 Daily Price Limit 300 basis points ($3,000 per contract) above or below the preceding day's settlement price for contracts in the same delivery month Position Limits The maximum net short or long position is 4,000 contracts Original Margin Requirements $1500 per contract, subject to change what the value of the announced CPI for March 1988 will be. (The March CPI is announced in April; the CPI futures contract is settled after the announcement.) So your broker tells you that the market collectively predicts that the March 1988 CPI will be, rounded, 346.9.6 6The CPI futures market offers contracts going out as far as three years, so that inflation risk three years out can be hedged against. In fact, to the extent that inflation risk more than three years out is correlated with inflation risk at three years or less one can, in effect, hedge against inflation risk more than three years out. 6 The next step is to purchase the contract via the futures Exchange from somebody who is willing to sell. Even though your CPI futures contract is not settled with cash until April 1988, the Exchange requires that you post $1,500 “margin" when you agree to the contract. The margin is a “good faith" deposit made with the broker to guarantee that you comply with the terms of the futures contract. Since the margin may earn interest while it is held by the broker, you do not necessarily sacrifice earnings by posting it. In addition to posting margin, you must also FEDERAL RESERVE BANK OF PHILADELPHIA The CPI Futures Market Brian R. Horrigan Brokerage Fees Brokerage fees are negotiable, vary from broker to broker, and depend on the number of futures contracts entered into. The fee for a single "round trip"— opening and closing a position on one CPI futures contract—is likely to be in the $25 to $75 range currently. Available Contract Months CPI futures contracts continue for three years at any time. Contracts are traded for quarterly settlement (January, April, July, and October) during the 12-month period from the current trading date and for semi-annual settlement (July and January) during the period beyond 12 months but less than 36 months beyond the current trading date. For example, a user of the CPI futures market on June 2 3 ,1 9 8 6 would have the following futures contracts available: SETTLEMENT MONTH July 1986 July 1987 July 1988 O ctober 1986 January 1987 January 1988 January 1989 April 1987 The settlement of a CPI futures contract is based on the CPI-W measured in the previous month. On the first business day following the expiration of any January or July contract, trading begins in the relevant January or July contract three years hence. On the first business day following the expiration of an April or O ctober contract, trading begins in the relevant April or October contract one year hence. Extensions The Coffee, Sugar, & Cocoa Exchange has filed a request to modify the nature of the CPI futures contract. The Exchange wants a new contract that trades not on the level of the CPI-W, but on the percentage change (that is, the inflation rate) of the CPI-W. The underlying value of the contract will be $1,000,000 and prices will be quoted in thousandths of a percent with the minimum tick being 0.005 percentage points, with each 0.01 change in the projected inflation rate equaling $25 and each tick being $12.50 per contract. So if the inflation rate com es in one percentage point higher than specified in the inflation futures contract, the gain to the buyer is $2,500. The economics of the proposed contract are identical to the existing contract. pay your broker a fee for his efforts. The size of the fee is negotiable and can vary with the size of the transaction. Currently, broker fees run between $25 and $75 for entering into and settling a single CPI futures contract. Between the time you buy the contract and the day you settle the contract, the market value of your contract is likely to change. As new information about the economy comes to light, participants in the CPI futures market will revise their beliefs about the March 1988 CPI. If the participants come to expect more inflation, they will expect the March 1988 CPI to be higher than they previously thought, in which case the mar ket value of the April 1988 CPI futures contract will rise. In that case, you will receive capital gains on your contract. These gains are added to your margin account on a daily basis as the contract is "marked to market," and you are free to withdraw those gains. But suppose market participants come to expect less inflation; then the April 1988 CPI futures contract falls in value. You consequently suffer capital losses on your contract, and those losses are subtracted from your margin. If your margin shrinks below a specified "maintenance level" (which is 75 per 7 BUSINESS REVIEW cent of the initial margin in the CPI futures market), you must post additional margin to bring the margin back to its original level. If you fail to make the margin call, your broker liqui dates your position immediately at the going market price. Now comes settlement day in April 1988. You and the seller of the contract must settle with cash. If the CPI comes in higher than the CPI specified in the futures contract, you make money. The person who sold you the contract pays you the difference between the announced and contracted value of the CPI, times $1,000. For example, if the CPI comes in at 343.5 while the contracted value of the CPI is 341.5, you receive $2,000, that is, (343.5 — 341.5) X $1,000. Similarly, if the CPI comes in lower than the contracted value, you lose money; if the CPI comes in at, say 338.5, you would have to pay the seller of the contract $3,000. Thus, your gains and losses in the CPI futures market depend on the difference between the announced value of the CPI and the market's expectations of the CPI as embodied in a CPI futures contract, not just on the increase in the CPI.7 Finally, you can liquidate your position in the CPI futures market before settlement day simply by taking an offsetting futures position of the same size. For example, if you bought 20 CPI futures contracts whose settlement day is in April 1988, you need only sell 20 CPI contracts with the same settlement day in order to close out your position and realize the gain or loss on that date on your initial investment. It might seem that the CPI futures market is simply a gambling pit that belongs in Atlantic City rather than in New York, and indeed, for those who take purely speculative positions in futures markets, the analogy is apt.But the CPI 7The Coffee, Sugar, & Cocoa Exchange has filed a request to modify the nature of the CPI futures contract. The Exchange wants a new contract that trades not on the level of the CPIW, but on the percentage change (that is, the inflation rate) of the CPI-W. The economics of the proposed contract are identical to the existing contract. 8 MAY/JUNE 1987 futures market also allows investors and busi nesses to hedge themselves against pure inflation risk. HOW THE CPI FUTURES MARKET CAN LOCK IN A REAL INTEREST RATE Trillions of dollars worth of financial contracts in the United States are written in nominal terms and hence are subject to inflation risk. The CPI futures market offers those involved in financial contracts the opportunity to hedge against that inflation risk and thereby lock in a real interest rate on their investments. Some examples indi cate how this can be done. Hedging an Investment With the CPI Futures Market. Suppose in June 1990, a financial insti tution—let's call it the Retirees' Investment Fund (RIF)—invests $1,000,000 in 1-year Treasury bills bearing a known, fixed nominal yield of 9 percent, which means RIF receives $1,090,000 in principal and interest in June 1991. Now sup pose in June 1990, the CPI has a value of 100.0 and the CPI futures contract selling in June 1990 and settling one year later has a price of 106.0, so the expected inflation rate implicit in the 1-year futures contract is 6 percent.8 By buying the appropriate number of CPI futures contracts, RIF can lock in a 3 percent real return on its investments. To find the number of futures con tracts RIF must buy, divide the amount of money subject to inflation risk—in this example, $1,090,000—by the settlement value of one CPI futures contract—in this example, $106,000— covering the month in which the T-bills mature. So dividing, we get approximately ten contracts. Since fractions of contracts cannot be purchased, RIF buys ten contracts. Table 1 shows how the real yield on RIF invest ment is affected by inflation. Suppose that infla tion is 6 percent during the 12 months following June 1990. In that case, RIF neither gains nor 8In the example in the text, the CPI for June 1990 is set at 100.0 for convenience, a number below the current value of the CPI. FEDERAL RESERVE BANK OF PHILADELPHIA The CPI Futures Market Brian R. Horrigan Table 1 How RIF Locks in a Real Interest Rate of (Almost) 3 Percent Investment to Be Hedged: $1,000,000 1-year T-Bills, bought in June 1990 in Nominal Return on the T-Bills: 9%, yielding $90,000 in interest in addition to the princi pal repayment of $1,000,000 in June 1991 CPI in June 1990: 100.0 Price of a 1-year CPI Futures Contract in June 1990: 106.0, an implicit expected inflation rate of 6% Number of CPI Futures Contracts Purchased: Assumed Brokerage Fee Per Contract: 10 $25 Three Scenarios for Settlement After Announcement of June 1991 CPI CPI In Time 1991 June-June Inflation Rate Total Paymentb Total Real Payment' Total Real Yieldd $0 $1,089,750 $1,028,066 2.81% Payment On CPI Futures2 1. 106.0 6% 2. 110.0 10% $40,000 $1,129,750 $1,027,046 2.70% 3. 102.0 2% -$ 4 0 ,0 0 0 $1,049,750 $1,029,167 2.92% aIf the CPI rises by 6 percent, RIF receives nothing on its futures contracts and gets a real yield of approximately 3 percent on its investment in T-bills. If the CPI rises by 10 percent, RIF receives $4,000 on each CPI futures contract it bought [$1,000 X (110.0 — 106.0)], for a total of $40,000 on 10 contracts. If the CPI rises by 2 percent, RIF pays $4,000 on each CPI futures contract it bought [$1,000 X (102.0 — 106.0)], for a total payment of $40,000 on the 10 contracts. bThe total payment is the sum of the payment on the T-bills (which is always $1,090,000) and on the CPI futures contracts, less the brokerage fee for the 10 futures contracts (which is $250). The payment is rounded to the nearest dollar. Taxes are not considered here because people have different tax brackets. cThe real payment is the total payment divided by the CPI, expressing the payment in June 1990 dollar values. dThe real yield is found by dividing the real payment by $1,000,000— the initial principal— and subtracting 1. The real yield is always about 3 percent because the payment on the CPI futures contracts almost completely offsets the gain or loss in purchasing pow er on the T-bills caused by inflation being different from the 6 percent implicit in a 1-year CPI futures contract. The hedge is not perfect because fractions of contracts cannot be purchased. loses on its futures contracts because the actual CPI turns out to equal the CPI specified in the futures contract. RIF gets about a 3 percent real return on its combined investment in T-bills and CPI futures. If, instead, the inflation rate is 10 percent, RIF ends up with a real return of —1 percent on its T-bills instead of a 3 percent real return (—1 percent real interest rate = 9 percent nominal interest rate minus 10 percent inflation). But RIF gains on the CPI futures contracts that it 9 BUSINESS REVIEW bought, and the gain about offsets the real loss on the T-bills, so that the real return on the Tbills plus the CPI futures contracts is about 3 percent. On the other hand, if there is a 2 percent inflation rate, then RIF pays on its futures con tracts. At the same time, the lower inflation creates a 7 percent real interest rate on the T-bills (7 percent real interest rate = 9 percent nominal interest rate minus 2 percent inflation). The loss on the CPI futures contracts offsets the real gain RIF makes on its T-bills, so that, once again, the real return on the T-bills plus the CPI futures contracts is about 3 percent. By buying the right number of futures contracts, RIF can insure itself against inflation risk and lock in a real interest rate on its T-bills.9 Borrowers Can Hedge As Well. Who is selling the futures contracts that RIF is buying? If RIF reduces its inflation risk when it lends money by also buying CPI futures contracts, do the busi nesses that sell the contracts increase their infla tion risk? Perhaps, but not necessarily. Consider a hypothetical corporation selling 1-year dis count bonds (that is, bonds which do not bear coupons) which bear a nominal yield of 9 per cent.10 The corporation is in a position similar to RIF's—it is certain about the nominal interest it will pay, but it is uncertain about the real interest it will end up paying because it cannot forecast the inflation rate perfectly. If inflation comes in unexpectedly high, the real value of the corpo ^There is a slight problem with this example. One-year T-bills bought at the end of June 1990 mature at the end of June 1991 when the investors receive the face value of the T-bills. However, a CPI futures contract covering the CPI of June 1991 does not settle until the end of the third week in July 1991 when the Bureau of Labor Statistics announces the value of the June 1991 CPI. Thus, there is a lag between receiving the cash from the T-bills and receiving (or paying) cash for the CPI futures contract, so the CPI futures market cannot provide a perfect hedge against unanticipated infla tion. The inability to buy fractions of contracts also prevents perfect hedging. 10A discussion of coupon bonds is more complex; for details of hedging coupon bonds, see Todd Petzel and A. Fitzsimons, Bank Utilization of Inflation Futures, (NY: Economic Index Market of the Coffee, Sugar, and Cocoa Exchange, 1985). 10 MAY/JUNE 1987 ration's debts is unexpectedly reduced, and if inflation comes in unexpectedly low, the real value of the corporation's debts is unexpectedly increased—the reverse of the position of RIF. The corporation, like RIF, can shed that inflation risk in the CPI futures market—but it does so by selling CPI futures contracts, not buying them as RIF did. Borrowers, as well as lenders, can hedge them selves in the CPI futures market. In fact, busi nesses that both borrow and lend, like insurance companies and banks, can profitably use the CPI futures market to lock in a real interest rate on both their assets and liabilities.11 In addition, nonfinancial businesses also can protect them selves from the inflation risk that arises in ordinary business operations. For example, a business that submits a fixed nominal bid to construct a building is vulnerable to changes in inflation while construction is underway. Another example, examined below, concerns a business that negotiates an indexed labor contract with its employees. The CPI Futures Market Assists Indexing. Consider the case of a hypothetical firm called the American Steel Company (ASC). Because the price level grows at an uncertain rate, both the ASC and its employees, members of the Steel Union, are subject to risk about the real value of the long-term labor contract they nego tiate. Suppose ASC signs a labor contract with the union that includes a cost-of-living adjust ment (COLA) clause that automatically adjusts wages proportionately to the CPI. While this indexation protects workers' real wages, it may ^C urrently, financial institutions can use interest rate futures to reduce significantly the risk to their cash flow that comes from unexpected variations in the nominal interest rate. (See Michael Smirlock, "Hedging Bank Borrowing Costs With Financial Futures," this Business Review (M ay/ June 1986) pp. 13-23.) However, even if a financial institu tion completely immunized its cash flow against nominal interest rate risk, it would still be exposed to inflation risk because unexpectedly high inflation would reduce the real value of its cash flow— a risk that could be shed in the CPI futures market. FEDERAL RESERVE BANK OF PHILADELPHIA The CPI Futures Market expose ASC to inflation risk if the price of steel is not highly correlated with the CPI. (For some firms, the prices of their products can fall as the CPI rises, as happened in recent years with oil, some foodstuffs, and industrial metals.)12 What can ASC do? With the CPI futures market, it can offer its workers indexed contracts without increasing its inflation risk. When ASC signs an indexed contract with its workers, it can simultaneously buy the appro priate number of CPI futures contracts, and thereby protect its profits against unanticipated inflation. Suppose the inflation rate implicit in the CPI futures market is 6 percent. If foreign competition does not allow ASC to raise its steel prices by more than 6 percent, then each per centage point that inflation rises above 6 percent increases labor costs (and reduces profits) via the COLA. But A SC's extra labor costs are offset by the gains on its CPI futures contracts. And if inflation comes out unexpectedly low, ASC's lower labor costs are offset by its losses on the CPI futures contracts.13 IF CPI FUTURES ARE SUCH A GOOD IDEA, WHY IS THE VOLUME SO LOW? Shortly after the CPI futures contract was introduced, Milton Friedman stated that it could becom e “the largest-volume contract in the country."14 Yet, the market's activity has been small, especially when compared to other fu tures markets. For example, on December 2, 1986, the volume of the CPI futures market was zero, while the coffee and S&P 500 futures markets, two representative futures markets, traded 3,800 and 109,749 contracts, respectively. 12Further analysis of the advantages and disadvantages of indexation can be found in Brian Horrigan, "Indexation: A Reasonable Response to Inflation," this Business Review, (Septem ber/O ctober 1981) pp. 3-11. 13The hedge is imperfect because the payroll is paid out every week, so that complete inflation protection would involve buying a series of contracts maturing at different dates through the year. 14Quoted in Institutional Investor, (January 1986) p. 17. Brian R. Horrigan And the open interest—the total number of futures contracts not settled or closed out—on the same day was 58 on the CPI futures market, while on the coffee and S&P futures market the figures were 15,322 and 146,005 contracts, respectively. These values are typical for other days, also. Clearly, this market is not catching on. Why has the market failed in practice, so far, and under what conditions may it yet succeed? While no definitive answers are available, sev eral factors may play important roles. The CPI May Not Be a Relevant Measure of Prices For Many. The CPI measures how the price of a representative “basket" of goods and services purchased by American consumers has changed over time. But many American con sumers probably purchase a basket of goods and services that is very different from that measured by the CPI. If the prices of different goods and services rise and fall by very different amounts, the inflation rate faced by individual families may be significantly different from the inflation rate as measured by the change in the CPI. In that case, protection against unexpected changes in the CPI would have little value in stabilizing the cost of living for many families. Economist Robert Gordon gave an example of how widely spread price increases are among different commodities: “Someone who spends equal shares of his income on rent, TV sets, telephone calls, eggs, and whiskey, would have experienced a price increase since 1967 of only 51 percent, or a compounded rate of only 3.2 percent per year. Someone else who spends equal shares on steak, potatoes, coffee, fuel oil, and mortgage interest, would have experienced an increase since 1967 of 321.3 percent, or a compounded rate of 11.7 percent per year."15 Businesses May Prefer Cross-Hedges. For a new futures contract to be successful, it is not 15Robert J. Gordon, "The Consumer Price Index: Measuring Inflation and Causing It," The Public Interest (Spring 1981) p. 117. Also see Robert P. Hagemann, "The Variability of Inflation Rates across Household Types," Journal of Money, Credit, and Banking (November 1982) pp. 494-510. 11 BUSINESS REVIEW enough that the contract provide an opportunity for hedging against commodity price risk. A new futures contract must provide opportunities for reducing risk that are significantly greater than the opportunities that already exist in other futures markets. Consequently, many businesses "cross-hedge," meaning that they hedge against price risk for one commodity by buying or selling futures contracts for a similar commodity that has an active futures market. For example, one reason for the failure of the barley futures market is that grain dealers found that they could hedge against barley price risk in the corn futures mar ket. The hedge is imperfect since barley prices fluctuate relative to corn prices, but the grain dealers preferred using the corn futures market which was far more active and "liquid" than the barley futures market. An analogous situation could exist for the CPI futures market, namely, that businesses reject the perfect CPI hedges available in the CPI futures market for less per fect hedges in more liquid futures markets. Futures markets in industrial and precious metals and in financial instruments allow some hedging against CPI risk, and businesses may well prefer to work in those familiar and active futures markets to shed (imperfectly) some of their inflation risk.16 Reduced Inflation Uncertainty Retards the CPI Futures Market. Businesses face two differ ent types of price risk: inflation risk, in which all prices rise together, but at an uncertain rate, and relative price risk, in which different prices rise and fall relative to each other, even though the average of all prices may not change at all. Both types of risk affect the real income of businesses. To see this, consider the economic situation of a wheat farmer who is sure about the size of his 16This point about the role of cross-hedging in determining the success of futures contracts is argued in Deborah G. Black, "Success and Failure of Futures Contracts: Theory and Empirical Evidence," Monograph Series in Finance and Economics 1986-1, Salomon Brothers Center for the Study of Financial Institutions, Graduate School of Business Admin istration, New York University, 1986. 12 MAY/JUNE 1987 wheat crop but is uncertain about prices. The farmer's real income can be reduced either by a fall in the price of wheat he sells or by an increase in the prices of the consumer goods he buys. If the price of wheat remains very stable while the CPI rises erratically, the main uncertainty about the farmer's real income is inflation risk, and the farmer can reduce that risk in the CPI futures market. But if the CPI is stable while the price of wheat is volatile, the farmer can protect his income by shedding his price risk in the wheat futures market. Eliminating the risk from wheat prices would eliminate virtually all of his uncer tainty about his real income, so there would be little reason for the farmer to use the CPI futures market. For the CPI futures market to be considered worth the costs of using it (in time and trouble as well as in brokerage costs), there must be suffi cient uncertainty in the CPI to expose many people to significant inflation risk. A reduction in inflation risk compared to relative price risk could be an important reason for the lack of interest in the CPI futures market, since the level of inflation and expectations of inflation dropped significantly after 1980. Economists have docu mented that the level of inflation is positively correlated with inflation uncertainty both in this country and in other countries, and some eco nomic theories predict that the correlation between inflation uncertainty and the level of inflation will persist. Since people expect a lower inflation rate in the near future, people should also expect inflation to be more predictable and inflation risk to be less important.17 As inflation 17See John B. Taylor, "On the Relation Between the Vari ability of Inflation and the Average Inflation Rate," CamegieRochester Conference Series on Public Policy, (Autumn 1981) pp. 57-86. Also see Stanley Fischer, "Relative Shocks, Relative Price Volatility, and Inflation," Brookings Papers in Economic Activity, 2 (1981) pp. 381-431, and A. Steven Holland, "Does Higher Inflation Lead to More Uncertain Inflation?" Federal Reserve Bank of St. Louis Review (February 1984) pp. 15-26. Stanley Fischer, Indexing, Inflation, and Economic Policy (Cambridge, MA; The MIT Press, 1986) pp. 301-320, ob- FEDERAL RESERVE BANK OF PHILADELPHIA The CPI Futures Market and inflation risk have subsided, many, and per haps most, businesses are in the same position as the wheat farmer—they find that the risk to their real profits comes from the variability of their own prices relative to the CPI, not from the variability of the CPI, and they use futures mar kets to reduce their relative price risk rather than to reduce their inflation risk. There have been other futures markets that have become inactive because of the low level of price volatility, such as the one for fresh eggs. The same could happen, and may already be happening, to the CPI futures market. There is no underlying asset for the CPI futures contract. A futures market cannot succeed unless it attracts businesses that wish to hedge against the price risk of the commodity traded in the futures market. But it is hard for a futures market to survive by attracting only hedgers. Generally, it helps a futures market to attract arbitrageurs and speculators, who add "liquidity" to the futures market. By buying and selling futures contracts continually, arbitrageurs make it cheaper for hedgers to buy or sell futures contracts whenever they want, which in turn attracts more hedgers to the futures market. Furthermore, they make the futures market more "efficient," in the sense that information is reflected in futures prices more quickly. A problem for the CPI futures market may be that there is no underlying asset for the futures contract, as there is for virtually every other futures contract traded. This feature inhibits arbitrageurs and speculators from parti cipating in the CPI futures market to the same extent they participate in other futures contracts. Consider a commodity like wheat, which is sold in both spot markets and futures markets. If serves one measure of the relative importance of inflation risk and relative price risk can be found by comparing the variance of the CPI inflation rate with the variance of the stock market rate of return; the former is smaller than the latter by a factor of the order of 100. Referring to the intro duction of CPI-indexed bonds, he comments that "inflation uncertainty is relatively trivial and insufficient to make the introduction of a new financial asset worthwhile." The same point applies to the use of the CPI futures market. Brian R. Horrigan the price of a wheat futures contract is high enough relative to the spot price of wheat, arbi trageurs can make money by buying and storing wheat and simultaneously selling a wheat futures contract. On settlement day, the arbitrageur must deliver the wheat promised in the futures contract at the price specified in the contract, but the arbitrageur has already purchased the wheat destined for delivery. Since the selling price of the wheat as specified in the futures contract is above the price at which the arbitrageur pur chased the wheat, the arbitrageur makes enough money to cover the cost of storing the wheat and still have some profit—a risk-free profit, since the selling price of wheat is set by contract. What is true for wheat futures in this regard is also true for other commodity futures and for interest rate and stock price futures. Arbitrageurs can supply futures contracts upon demand without exposing themselves to significant risk as long as they can buy the asset that underlies the futures contract. But if it were impossible to store wheat for future delivery, the arbitrageur would take on the risk of a potentially significant wealth loss when he sold the wheat futures contracts. An arbitrageur who sold wheat futures contracts would have to wait until the futures contract matured, and buy wheat in the spot market for delivery to the owner of the wheat futures con tract. If the spot price of wheat rose above the futures price of wheat by settlement day, the arbitrageur would lose wealth. This risk would inhibit arbitrageurs from selling wheat futures contracts and would depress the volume of activ ity in the wheat futures market. The ability to buy and store the asset underlying a futures market encourages the participation of arbi trageurs, which enhances the liquidity of the market. Unfortunately for the CPI futures market, it is impossible to buy and store the basket of goods and services whose price the CPI measures; there are hundreds of goods and services in the CPI basket, some of which (like ice cream) can be stored only at great cost and others (like 13 BUSINESS REVIEW haircuts) which cannot be stored at all. Since arbitrageurs cannot buy and store the CPI basket while selling the CPI futures contract, they absorb inflation risk when they sell the contract, which they may not want to do. Thus, the poten tial supply of CPI futures contracts is limited in a way that other, m ore successful futures markets are not. Speculators also prefer futures markets for which there is an underlying asset. For most of the commodities or assets traded on futures markets, there are active spot markets in which there is constant adjustment of prices as new information is discovered about the future value of the commodities and assets. The new infor mation about spot prices is useful for speculators seeking to take a risky position in futures markets. The CPI market basket is not traded in active spot markets, so there is no continuous price data available on the CPI. The CPI is announced once a month, whereas for many commodities the prices are announced (on their trading floors) every minute. The lack of continuous price infor mation discourages speculators from entering the CPI futures market, which further reduces the liquidity of the market. There is indirect evidence for the argument that the absence of an underlying asset explains why market interest in the CPI futures market is so low. The New York Futures Exchange now trades a futures contract that, like the CPI futures market, is settled on a cash basis and whose settlement value is determined by a price index. The price index is the Commodity Research Bureau's (CRB) index of futures prices for 27 key commodities. There are active futures and spot markets in each of the 27 commodities in the CRB's futures price index, so that there are underlying assets for the new futures contract introduced by the New York Futures Exchange. The volume of trading and the open interest in the CRB index futures contract has been much higher than that in the CPI futures market. On December 2,19 8 6 , for example, the volume and open interest were 225 and 1,380, respectively, Digitized 14 FRASER for MAY/JUNE 1987 in the CRB index futures market, while the vol ume and open interest were zero and 58, respec tively, in the CPI futures market. The CRB index futures market has been no more successful than the CPI futures market in attracting hedgers, so far. But arbitrageurs and speculators have been able to use the CRB index futures market because of the presence of underlying assets that are actively traded and that may be stored, and they have been more willing to use the market because of the greater price volatility, which may explain why the CRB index futures market has greater activity than the CPI futures market. CONCLUSIONS The introduction of a futures market trading contracts based on the value of the Consumer Price Index—enthusiastically endorsed by promi nent economists—has made available to busi nesses and investors a means of trading and hedging against inflation risk. The CPI futures market has not been a great success, however. Its volume and open interest have been very low. The novelty of the contract, the inadequacy of the CPI for measuring inflation risk, the existence of cross-hedges for the inflation risk of many households, and the lack of a storable underlying asset for the contract are undoubtedly some of the reasons for the market's low trading volume. But the most important reason for the failure of the market to be used in a big way is probably that current inflation risk is not very significant relative to all of the other risks that investors and businesses must manage. Given the costs of using the CPI futures market, most investors and businesses probably believe that their scarce resources are better spent coping with the more significant risks they face. How ever, if inflation becomes higher and more erratic, as it was in the 1970s, inflation risk will again become significant, and many investors and businesses may turn to the CPI futures market to cope with it. FEDERAL RESERVE BANK OF PHILADELPHIA Explaining Long-Term Unemployment: A New Piece To An Old Puzzle Robert H. DeFina* At each stage of the business cycle, there are individuals of all descriptions who want to work but who fail to find jobs for weeks or even months. Indeed, several studies find that this long-term unemployment accounts for most measured joblessness. ‘ Robert H. DeFina is a Research Officer and Economist in the Research Department of the Federal Reserve Bank of Philadelphia. Such unemployment has long been a focus of policymakers, because it raises several especially nettlesome social concerns. From an economic perspective, protracted unemployment means lost output as the skills and efforts of productive individuals go unused for months at a time. From a broader social perspective, lengthy un employment means a higher incidence of psy chological and health problems, not only among the unemployed but also among the members of their families. Several studies, for example, 15 BUSINESS REVIEW have associated increases in the U.S. unemploy ment rate with increases in suicides, increases in homicides and other crime, increases in heart, kidney, and liver disease, increases in admis sions to state mental hospitals, and increases in the incidence of child abuse.1 In order to address those concerns, policy makers need a thorough understanding of why long-term unemployment exists. Unfortunately, the underlying causes of persistent joblessness have remained something of a mystery. Recently, though, economists have begun to rethink their traditional notions of how labor markets work. In the process, they have provided novel insights into the sources of long-term joblessness. Em pirical investigation into this new line of thought, known generally as the "efficiency wage hypo thesis," is still at a relatively early stage. But the evidence that is available suggests that this new perspective has some validity. LONG-TERM UNEMPLOYMENT PRESENTS A PUZZLE Economists traditionally describe labor mar kets as they do auction markets for any other commodity, in terms of supply and demand. On the supply side of the market are individuals with a set of skills who voluntarily offer their time in return for the going wage. Economists generally believe that the number of people who wish to work falls as the wage does. On the demand side of the market are firms which need workers to produce output. In contrast to the quantity of workers supplied, the quantity of labor demanded usually rises when the wage falls, because falling wages make it increasingly profitable to hire more workers. The interaction of labor supply and demand determines the level of wages and employment that are actually observed. Lengthy unemployment, where individuals ^ h e figures on the adverse health consequences of un employment are cited in Robert J. Gordon, Macroeconomics (Boston: Little, Brown and Company, 1984) pp. 353-354. 16 MAY/JUNE 1987 go many months without a job, does not fit easily into that basic market description. If un employment arises, the reasoning goes, com petitive forces will quickly and automatically guide the wage to a level at which full employ ment prevails, that is, a situation in which every one who wants a job has one. Specifically, un employment encourages competitive job seekers to offer to work for less than other similar workers in an effort to obtain one of the relatively scarce jobs. Employers readily accept those of fers since, in the simple market setting, lower wages mean higher profits. Thus as wages get bid down, unemployment declines through two channels. First, the wage reductions spark some employers to hire more workers. Second, the wage reductions lead some job seekers to aban don their search. Job seekers will continue to bid down the wage until all individuals who want a job have one, that is, until unemployment van ishes. At that point no one has an incentive to undercut the wage offer of another, and all simi lar workers will be receiving the same market clearing wage. This scenario permits abbreviated unemploy ment due to frictions in the economy, such as incomplete information about job possibilities, that prevent instantaneous market adjustment to changing circumstances. Individuals simply may need some time to collect and assess their options before acting. But once that information is obtained, the competitive bidding process, if fully operative, should guarantee quick access to a job and rule out anything like the protracted unemployment experiences that actually oc cur. No official statistics exist on the length of individuals' completed episodes of unemploy ment, although independent estimates have been made (see UNEMPLOYMENT CAN BE QUITE LENGTHY). Those estimates show that even during periods when real gross national product (GNP) grows rapidly, such as 1962 to 1968, 1971 to 1973, and 1976 to 1978, a person counted as unemployed can, on average, remain jobless for six to eight months. Studies also reveal that FEDERAL RESERVE BANK OF PHILADELPHIA Explaining Long-Term Unemployment Robert H. DeFina Unemployment Can Be Quite Lengthy Year Real GNP Growth3* Average Total Number of Weeks That a Currently Unemployed Person Remains Jobless Akerlof and Main*5 Siderc 1959 1960 1961 5.8 2.2 28.8 25.6 na na 2.6 31.2 na 1962 29.4 na 1963 5.3 4.1 28.0 na 1964 5.3 1965 1966 1967 1968 5.8 5.8 2.9 4.1 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 26.6 23.6 na na na 20.8 17.6 16.8 na 18.2 2.4 -0.3 2.8 15.8 17.4 22.6 20.1 25.1 5.0 5.2 24.0 20.0 19.4 17.4 25.0 20.6 28.2 31.6 21.3 32.7 32.4 4.7 5.3 2.5 -0.2 28.6 26.2 28.3 24.1 na na 1.9 na na 23.1 26.4 29.2 -0.5 -1.3 4.9 -2.5 35.9 The above figures represent estimates of the total time a person counted as unemployed in the indicated year remained jobless, on average. If in 1982, for example, a person had been unemployed for the preceding 17 weeks when counted, he or she remained without a job during the next 18.9 weeks, on average, for a total duration of 35.9 weeks. NOTE: "n a" = not available for the indicated year. aYear-over-year percent change. ^George A. Akerlof and Brian G. M. Main, "An Experience-Weighted Measure of Employment and Unemploy ment Durations," American Economic Review (December 1981) Table 4, pp. 1003-1011. c Hal Sider, "Unemploym ent Duration and Incidence: 1968-82," American Economic Review (June 1985) Table 3, pp. 461-472. 17 BUSINESS REVIEW protracted unemployment is fairly pervasive, in that it is experienced by all demographic groups, and that it may account for a significant fraction of measured unemployment. For example, one estimate that focused on 1974 suggested that half of all unemployment that year was accounted for by episodes lasting more than three months.2 Economists have puzzled over the inconsist ency between the short unemployment dura tions predicted by the auction market model and the extended durations that actually occur. In trying to reconcile the two, they generally have retained their emphasis on the basic mar ket paradigm, while focusing on exogenous— that is, externally imposed—factors that might prolong or even prevent complete labor market adjustment after unemployment arises. Three commonly cited examples are minimum wage laws, unions, and unemployment insurance. Minimum wage laws and unions, it is argued, fix wages at levels that are too high to clear the labor market. Unemployment insurance potentially lengthens the duration of unemployment by defraying the cost of remaining jobless. But not only does theory suggest that such explanations of long-term unemployment are incomplete, those explanations also lack convincing em pirical support. For instance, studies find that minimum wage laws lead to higher teenage un employment, but have little or no impact on aggregate unemployment.3 At best, those ex 2Kim B. Clark and Lawrence H. Summers, "Labor Market Dynamics and Unemployment: A Reconsideration," Brookings Papers on Economic Activity (Volume 1 ,1979) pp.1372. 3For a thorough discussion of the theoretical and em pirical issues regarding those explanations, see Kim B. Clark and Lawrence H. Summers, "Labor Market Dynamics..." and Lester C. Thurow, Dangerous Currents: The State of Economics (New York: Vintage Books, 1984) chapter 7. For related discussions, see Charles Brown, Curtis Gilroy, and Andrew Kohen, "The Effect of the Minimum Wage on Employment and Unemployment," Journal of Economic Literature (June 1982) pp. 487-528, and Daniel S. Hamermesh, Jobless Pay and the Economy (Baltimore: The Johns Hopkins University Press, 1977). 18 MAY/JUNE 1987 ogenous factors appear to provide a partial accounting. Persistent unemployment is also sometimes identified with unskilled individuals and people who are displaced by structural change, such as former steel workers. But the pervasiveness of lengthy unemployment suggests that long-term joblessness cannot be accounted for totally by the problems of a few particular groups. More over, the question remains as to why such struc turally unemployed individuals cannot simply bid down wages in order to get jobs. This partial accounting has led some econo mists to question the usefulness of the basic market-clearing paradigm as a description of the labor market, and hence as a basis for under standing long-term unemployment. Rather than looking for an exogenous factor that might be interfering with an otherwise smooth-working market, those analysts have asked whether something inherent in labor markets prevents their clearing. This inquiry has produced a new view of the labor market, the "efficiency wage hypothesis," that affords an alternative expla nation of long-term unemployment. THE EFFICIENCY WAGE HYPOTHESIS: A NEW PIECE IN THE PUZZLE The efficiency wage hypothesis concentrates on the possibility that by increasing a worker's wage, an employer may increase that worker's productivity. The basic market model, in con trast, ignores this potential side benefit of a pay raise. Rather, it assumes that a worker's produc tivity is fixed by her existing skills; a worker's productivity helps determine her wages, but wages do not, in turn, influence a worker's pro ductivity. The efficiency wage view, if accurate, has a striking implication: a firm might actually in crease its profits by paying its workers more. The reasoning behind that implication is fairly straightforward. Increases in productivity mean that each worker produces more output. That is, as productivity rises, labor costs per unit of out put fall. Wage increases, by boosting produc FEDERAL RESERVE BANK OF PHILADELPHIA Explaining Long-Term Unemployment tivity, thus appear as a two-edged sword for profits. Higher wages directly raise labor costs and thereby contribute to lower profits, but they might also raise productivity, thereby cutting labor costs and contributing to greater profits. If the boost to productivity is large enough, profits will rise. The potential profitability of wage increases suggests a possible source of lengthy unem ployment. Employers will continue raising wages if doing so leads to maximum profits. By raising wages they will induce their existing workers to becom e more productive, but they will also induce additional people to enter the labor force and begin looking for jobs. Depend ing on how responsive productivity is to wage changes, firms could ultimately raise wages so high that the labor market does not clear, leaving some workers unemployed. And if they do, any subsequent unemployment will not quickly and automatically vanish as it would in the auction market arrangement, because people will not be able to bid the wage down to get a job. Unemployed individuals, whether they have quit, have been fired, or have entered the labor force for the first time, might try to get jobs by bidding down the wages of current workers. But in contrast to the simple competitive market situation, firms will not accept those offers. Firms have already weighed the benefits and costs of lower wages and decided that keeping wages high yields them their greatest profit. And be cause the unemployed cannot bid their way into jobs, they must instead wait until new openings arise from quits, firings, or increases in firms' demand for workers. They must then hope to be chosen over other jobless persons. On the whole, unemployed persons might remain jobless for quite some time. The potentially beneficial impact of wage in creases on productivity provides a coherent ex planation of lengthy unemployment. But the question of why such an impact on productivity might arise still remains. Efficiency wage theo rists have offered several possibly complemen tary answers.4 Robert H. DeFina Higher Wages Might Reduce Shirking. One answer stems from the difficulties employers have monitoring workers' efforts. For a variety of j obs, individuals participate in groups, such as when researchers coauthor studies or when construction crews erect a building. On other occasions, employees have some discretion over the pace of work, or work at a location physically distant from an immediate supervisor. Such is the case, for example, when employees go on business trips. Additionally, most jobs allow workers a certain amount of sick leave and time away for personal reasons. In those cases, man agers typically know only imperfectly either how hard each person works, or whether an absence from work was legitimate. Workers, as a consequence, have some chance to decrease their efforts without being detected. According to the "shirking" model, workers decide whether or not to reduce their efforts by weighing the costs and benefits of doing so.5 The model presumes that workers are fired if caught shirking, so the expected cost of shirking reflects their lost wage less any public or private assistance they might receive while unemployed, the length of time they remain unemployed, and the probability that they will be detected. The expected benefit of shirking is, of course, the value of on-the-job leisure the workers receive. Workers choose to shirk if the expected benefits exceed the expected costs. In such a scenario, wage increases boost pro ductivity because they reduce workers' incen 4Accessible surveys of particular variants of the efficiency wage hypothesis are found in Lawrence F. Katz, "Efficiency Wage Theories: A Partial Evaluation," National Bureau of Economic Research Working Paper no. 1906, Cambridge, MA (April 1986), and Janet Yellen, "Efficiency Wage Models of Unemployment," American Economic Review (May 1984) pp. 200-205. 5Formal characterizations of the shirking model are de scribed in Carl Shapiro and Joseph E. Stiglitz, "Equilibrium Unemployment as a Worker Discipline Device," American Economic Review (June 1984) pp. 433-444, and Samuel Bowles, "The Production Process in a Competitive Econ omy: Walrasian, Neo-Ffobbesian and Marxian Models," American Economic Review (March 1985) pp. 16-36. 19 BUSINESS REVIEW tives to shirk. By raising wages, employers raise the perceived cost to workers of being fired when caught shirking. Higher wages might tip the scale in favor of less shirking and, thus, greater productivity. Higher Wages Might Reduce Job Turnover. A related perspective emphasizes the job turnover that wage increases might reduce. When indi viduals join a firm they rarely commit to stay for an extended period. At a minimum, they retain a passive interest in their outside options, and their curiosity will likely grow if they perceive that their current employment situation is dete riorating. If they feel particularly short-changed, they might quit and devote all their efforts to obtaining a new job, convinced they can improve their current lot. Quits result in net productivity losses to firms. When workers quit, firms must operate with a reduced work force until suitable replacements are found. In addition, firms must devote pos sibly substantial amounts of resources to finding those replacements, such as time taken to review applications, to interview candidates, and to de cide to whom offers will be made.6 Productivity will also be lowered during an initial "start-up" period in which new employees learn the par ticulars of their jobs. A firm might succeed in reducing the quit rate of its work force, and thus raise productivity, by increasing wages. An employee's wage repre sents an important facet of his or her job, and probably figures in the decision whether to quit. By raising employees' salaries, the firm increases the relative attractiveness of their jobs, which might reduce the frequency of turnover.7 6Charles L. Schultze, "M icroeconomic Efficiency and Nominal Wage Stickiness," American Economic Review (March 1985) pp. 1-15, presents some evidence on the magnitude of turnover costs. He cites a study of Los Angeles firms which shows that costs of turnover (exit costs plus replacement costs) averaged $3,600, $2,300, and $10,400 for production, clerical, and professional and managerial workers, respec tively. 7A mathematical exposition of the turnover model can be found in Steven Salop, "A Model of the Natural Rate of 20 MAY/JUNE 1987 Higher Wages Might Boost Employee Morale. The productivity of employees can also depend on how fairly they think their employers treat them. Most firms understand the importance of good worker morale and firm loyalty for pro ductivity, and often actively strive to promote internal harmony. A firm's wage structure re presents an important concern in this regard. Workers typically have some notion of what constitutes a "fair day's work for a fair day's pay." They have some perception of where they stand relative to other workers both in their own firm and in other firms, and also have some perception of what constitutes an appropriate pay differential. Those feelings of what is an appropriate wage are partly molded by observing the treatment of other workers in positions simi lar to theirs. Although firms might find that "fair wages" are quite high, paying them is worth their while.8 Firms might be able to pay lower wages and still retain their employees, but those em ployees might be less productive. Employees who feel cheated, for instance, will not "go the extra yard" for the firm, and might spend valu able time griping to coworkers. By generally increasing wages to levels considered fair, or by raising certain workers' wages to maintain in ternal pay relationships that are deemed equita ble, firms might enjoy a more satisfied and more productive work force. But Other Factors Might Render Higher Wages Unnecessary. The preceding considerations make a link between higher wages and greater productivity appear plausible. Thus, they leave open the possibility that a wage-productivity link contributes to persistent unemployment. But even if such a link exists, firms still might not Unemployment," American Economic Review (March 1979) pp. 117-125, and Guillermo Calvo, "Quasi-Walrasian Theo ries of Unemployment," American Economic Review (May 1979) pp. 102-107. 8A sociological model is rigorously developed in George A. Akerlof, "Labor Contracts as Partial Gift Exchange," Quar terly Journal of Economics (November 1982) pp. 543-569. FEDERAL RESERVE BANK OF PHILADELPHIA Explaining Long-Term Unemployment use wage increases to raise productivity. Factors either internal or external to firms might render higher wages unnecessary, or diminish their use. If so, the cause of persistent unemployment lies elsewhere. An important internal factor is that firms can use other productivity-enhancing techniques.9 Firms can, for instance, deter shirking in ways other than by paying higher wages. One strategy involves raising the chances that workers will be caught by monitoring them closely. That might entail hiring supervisors or devising sophisti cated accountability schemes. Firms might also discourage shirking by using piece-rates, tying pay to demonstrated performance. Firms like wise might use approaches other than raising wages to reduce their turnover costs. As in ap prenticeship programs, for example, firms ini tially might pay new employees less than they are “worth" in an effort to defray, partially or totally, the costs of any needed training.10 Some firms might find that although wage increases raise profits, such alternatives increase profits more. If so, those firms will opt for the alternatives. Firms might also find that external factors al ready reduce shirking, quits, and bad morale to the point where attempts to reduce these prob lems further by any approach are uneconomical. Quit rates might remain low even without a firm's intervention because workers have strong per sonal attachments to their jobs, or because the financial costs to workers of searching for another job and relocating to another area are very high. Similarly, workers might already have sufficient incentives not to shirk because of the bad reputa tion they acquire if fired for lapses in diligence. The potential importance of those internal and external factors cannot be dismissed, at least not Robert H. DeFina at the conceptual level. Nor can the possibility that pay raises simply do not boost productivity to begin with. Thus, whether or not a wageproductivity link actually plays a significant role in explaining long-lasting unemployment must be settled empirically. EMPIRICAL EVIDENCE SUGGESTS THE NEW PIECE MIGHT FIT Empirical evidence on the question is sparse, often inferential, and reflects a variety of metho dologies. Some analysts have directly examined whether the basic premise of the theory, namely, that wage increases lead to higher productivity, is valid. In doing so, they have relied mainly on case studies of employer and employee behavior in the workplace. Several other authors have taken a more indirect approach: they develop the logical implications of the theory and then test statisti cally whether those are borne out by actual labor market experience. Support Comes from Case Studies... George Akerlof has provided perhaps the most direct evidence. In a series of articles, he reviews socio logical and psychological case studies of how wages influence worker productivity.11* He notes, for instance, an experiment in which stu dents were hired for proofreading. "One group was told that they were not qualified, but would be paid the usual rate. Another group was told that they were qualified and were also paid the usual rate. Those who were led to believe they were overpaid produced...more output per hour...than those who were told they were qualified..." (p. 82). Akerlof argues that such studies show that increased job satisfaction re sulting from higher wages results in greater worker effort, as stressed by the sociological theory. He also discusses studies which he claims reveal that employers actually use wage in- 9These alternatives have been analyzed at length by vari ous authors. Lawrence F. Katz, "Efficiency Wage Theories..." contains a good summary of those discussions. 10This possibility was suggested in Gary S. Becker, Invest ment in Human Capital: A Theoretical and Empirical Analysis, With Special Reference to Education (New York: National Bureau of Economic Research, 1964). 11 Those reviews are presented in George A. Akerlof, "Labor Contracts...," and George A. Akerlof, "Gift Exchange and Efficiency Wages: Four V iew s"American Economic Review (May 1984) pp. 79-83. 21 BUSINESS REVIEW creases to raise morale and achieve productivity gains. Jeremy Bulow and Lawrence Summers pre sented some evidence that gives credence to the efficiency wage hypothesis in general, and the shirking model in particular.12 They focused on historical accounts of the Ford Motor Company's pay policy, and found changes implemented in 1914 particularly noteworthy. At that time, the Ford Motor Company began paying employees $5 a day, while other manufacturers were paying their workers between $2 and $3 a day. Bulow and Summers note that observers of the change claim that it led to large increases in productivity, reductions in absenteeism, and fewer discharges for cause. They cite, for example, a contemporary engineering study which explains that, "The high Ford wage does away with all of the inertia and living force resistance... The workingmen are absolutely docile, and it is safe to say that since the last day of 1913, every single day has seen major reductions in Ford shops [sic] labor costs" (p. 378). ...And from Formal Statistical Tests. Other, less direct examinations of the efficiency wage hypothesis are also available. Those generally rely on statistical analyses which find that similar workers persistently receive different compen sation solely by virtue of their industry affilia tion or occupation.13 The auction view of labor markets cannot easily explain such differentials; competition among similar workers for the higher paying jobs should quickly eliminate dif ferences in pay. The efficiency wage hypothesis, in contrast, permits wage differentials for similar workers to exist as a result of different industry and occupational characteristics. One industry might find turnover to be more costly than 12Jeremy I. Bulow and Lawrence H. Summers," A Theory of Dual Labor Markets With Application to Industrial Policy, Discrimination, and Keynesian Unemployment," Journal of Labor Economics (August 1986) pp. 376-414. 13Lawrence F. Katz, "Efficiency Wage Theories..." pre sents a comprehensive survey of empirical studies of the efficiency wage hypothesis. 22 MAY/JUNE 1987 another industry, and thus might need to keep wages higher in order to contain the problem. Similarly, differences across occupations in shirking problems and the ability to monitor shirking could also give rise to different wage levels. As a test of the efficiency wage hypothesis, analysts have examined whether those wage differentials vary consistently with the dictates o f th e p articu lar e ffic ie n c y w ag e th e o rie s . T h e turnover view, for instance, argues that em ployers use wage differentials to reduce costly, productivity-reducing quits. In fact, several studies have found that higher wage premiums tend to be associated with lower quit rates, both at the industry and individual level, after con trolling for other factors that might influence quits.14 Further studies reveal that higher wage differentials coincide with lower absenteeism rates (again, after controlling for other factors), which may give some support to the shirking model.15 Because absenteeism can be monitored with little difficulty, but the reasons for it cannot, absenteeism may reflect shirking. Researchers have also found that when one occupational group in an industry receives a sizable wage premium relative to wages paid similar workers in other industries, it is likely that all occupational groups in an industry re ceive a wage premium.16 This result is consis tent with the idea that internal wage structures 14Examples of such studies are Richard B. Freeman, "The Exit-Voice Tradeoff in the Labor Market, Unionism, Job Tenure, Quits, and Separations," Quarterly Journal of Eco nomics (June 1980) pp. 643-673, and Alan B. Krueger and Lawrence H. Summers, "Efficiency Wages and the Wage Structure," National Bureau of Economic Research Working Paper no. 1952, Cambridge, MA (June 1986). 15Such results are presented in Steven Allen, "Trade Unions, Absenteeism, and Exit-Voice," Industrial and Labor Relations Review (April 1984) pp. 331-345, and Alan B. Krueger and Lawrence H. Summers, "Efficiency Wages and...". 16Evidence on the occupational wage structure can be found in Lawrence F. Katz, "Efficiency Wage Theories...," and William T. Dickens and Lawrence F. Katz, "Industry Wage Patterns and Theories of Wage Determination," Mimeo, University of California, Berkeley (March 1986). FEDERAL RESERVE BANK OF PHILADELPHIA Explaining Long-Term Unemployment figure importantly in firms' wage-setting deci sions, as emphasized by the sociological model. Although case studies and statistical analyses offer some support for the efficiency wage hypothesis, the question of its validity is by no means settled. Case studies, while informative about particular work settings at particular times, fail to indicate how widespread the discovered behavioral patterns are. And while statistical studies examine behavior across a much wider cross section of workers and firms, none of them directly links wage increases to productivity increases. They find a correlation between higher wages and their presumed benefits, but fail to establish either purposeful behavior on the part of firms or a causal link. Thus, the rela tions they uncover could be spurious or could arise for some other independent reason. THE PUZZLE IS NOT YET COMPLETE The efficiency wage hypothesis offers a simple insight into the operation of labor markets: by raising a worker's wage, a firm may succeed in raising a worker's productivity. That idea, while simple, also appears quite powerful because it provides a logically coherent explanation for persistent unemployment. Adding to the idea's strength is the initial empirical support it re ceives. A potentially important contribution this re search can make is the guidance it gives policy makers in dealing with unemployment. At pre sent, the policy implications of efficiency wage theories are largely undeveloped. Few conclu sions have been drawn, and those often hinge critically on the particular model studied. None theless, the theories do seem to leave some scope for policies to influence unemployment. Each model ties persistent unemployment to structural aspects of labor markets. Thus, poli cies that affect those aspects may also affect un employment. Some researchers, for example, have discussed the possible impact of increasing unemployment insurance in the context of the shirking model.17 According to their logic, the Robert H. DeFina increase in unemployment insurance might in duce workers to shirk more, since it reduces the penalty for being caught. This results in firms having to raise their wages to reduce the shirking, which in turn could lead to higher unemploy ment because it draws more people into the labor force. Studies indicate that efficiency wage theories might have implications for other struc tural labor market policies as well, such as man power training and regulations regarding job security.18 For instance, regulations that in crease job security can reduce the expected cost of being caught shirking. Thus, they might in duce firms to pay higher wages, which might increase unemployment. Efficiency wage theorists have examined the cyclical behavior of unemployment, in addition to its structural sources. They have found that, under certain circumstances, the wage "sticki ness" implied by efficiency wage theories allows variations in aggregate demand to cause swings in the unemployment rate.19 That suggests a role for monetary and fiscal stabilization policy to dampen those fluctuations in unemploy ment. Long-term unemployment obviously repre sents a complex issue. And not surprisingly, many important conceptual and empirical ques tions regarding its causes and cures remain un answered. But while much work is yet to be done, the research to date on efficiency wage theories does seem to have yielded a productive step toward understanding a major social ill. 17 See, in particular, Joseph E. Stiglitz, "Theories of Wage Rigidity," National Bureau of Economic Research Working Paper no. 1442. Jeremy I. Bulow and Lawrence H. Summers, "A Theory of Dual...," presents a somewhat different analysis, but arrives at a similar conclusion. 18See Jeremy I. Bulow and Lawrence H. Summers, "A Theory of Dual...," for a discussion of the implications of some other structural labor market policies using a variant of the shirking model. 19Both Joseph E. Stiglitz, "Theories of Wage Rigidity," and George A. Akerlof and Janet Yellen, "A Near Rational Model of the Business Cycle, With Wage and Price Inertia," Quarterly Journal of Economics (August 1985) pp. 823-838, analyze that possibility. 23 FEDERAL RESERVE BANK OF PHILADELPHIA BUSINESS REVIEW Ten Independence Mall, Philadelphia, PA 19106