The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
^ Economic iSsisi Review mammmmma t 1 FEDERAL RESERVE BANK OF ATLANTA • NOVEMBER/DECEMBER 1989 Economic Review President Robert P. Forrestal Senior Vice President and Director of Research Sheila L. Tschinkel Vice President and Associate Director of Research B. Frank King Research Officers W i l l i a m Curt H u n t e r , Basic Research Mary Susan Rosenbaum, Macropolicy G e n e D. S u l l i v a n , R e g i o n a l LarTy D. W a l l , F i n a n c i a l D a v i d D. W h i t e h e a d , R e g i o n a l Public Information Officer B o b b i e H. M c C r a c k i n Publications R o b e r t D. L a n d , E d i t o r L y n n H. F o l e y , E d i t o r i a l A s s i s t a n t J. E d w a r d R o o k s , G r a p h i c s a n d T y p e s e t t i n g M i c h a e l |. C h r i s z t , C i r c u l a t i o n The Economic Review seeks to inform t h e public about Federal Reserve policies and the economic environment and, in particular, to narrow the gap between specialists and concerned laypersons. Views expressed in the Economic Review are not necessarily those of this Bank or of the Federal Reserve System. Material may be reprinted or abstracted if the Review and author are credited. Please provide t h e Bank's Public Information Department with a copy of any publication containing reprinted material. Free subscriptions and limited additional copies are available from the Public Information Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, N.W., Atlanta, Georgia 30303-2713 (404/521-87881. Change-of-address notices and subscription cancellations should be sent directly t o the Public Information Department. Please include the current mailing label as well as any new information. ISSN 0 7 3 2 - 1 8 1 3 V O L U M E LXXIV, N O . 6, N O V E M B E R / D E C E M B E R 1989, E C O N O M I C R E V I E W 2 Interest-Rate Caps, Collars, and Floors Peter A Abken 26 Financial Asset Pricing Theory: A Review of Recent Developments Ellis W. Tallman The author surveys recent theoretical and empirical developments concerning asset pricing and its relevance to real economic phenomena. F.Y.I. U.S. and Foreign Direct Investment Patterns 42 William ). Kahley 58 Aruna Srinivasan 64 Using these interest-rate risk management instruments, investors can both hedge against uncertainties resulting from interest-rate risk and speculate on interest-rate movements. Book Review Index for 1989 FEDERAL RESERVE BANK OF ATLANTA II Bank Costs, Structure, and Performance by James Kolari and Asghar Zardkoohi Interest-Rate Caps, Collars, and Floors Peter A. Abken As some of the newest interest-rate risk management instruments, caps, collars, and floors are the subject of increasing attention among both investors and analysts. This article explains how such instruments are constructed, discusses their credit risks, and presents a new approach for valuing caps, collars, and floors subject to default risk. 2 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 S ince the late 1970s interest rates on all types of fixed-income securities have b e c o m e more volatile, spawning a variety of m e t h o d s to mitigate the costs associated with interest-rate fluctuations. Managing interest-rate risk has b e c o m e big business a n d an exceedingly complicated activity. O n e facet of this type of risk m a n a g e m e n t involves buying and selling "derivative" assets, which can b e used to offset or h e d g e changes in asset or liability values caused by interest-rate movements. As its n a m e implies, the value of a derivative asset d e p e n d s on the value of another asset or assets. Two types of derivative assets widely discussed in the financial press a n d in previous Economic Review articles are options a n d futures contracts. 1 Another derivative asset that has b e c o m e extremely popular is t h e interestrate swap. 2 This article examines a group of instruments known as interest-rate caps, collars, a n d floors, which are medium- to long-term agreements that have proven to b e highly useful for hedging against interest-rate uncertainties. In this regard, caps, collars, and floors can b e thought of as insurance poi icies against adverse movements in interest rates. Like interest-rate swaps, to which these instruments are closely related, caps, collars, a n d floors are designed to h e d g e cash flows over time rather than on a single date. The discussion below will show how caps, collars, a n d floors are related to each other, as well as how they may b e constructed from the most basic derivative asset, the option. The article also shows t h e ways in which caps, collars, and floors are created in practice, along with the different kinds of interm e d i a r i e s involved in t h e c a p market. 3 The rationale for hedging is reviewed, as are examples of how caps, collars, and floors are used by different financial institutions. The last section of the article considers the credit risk associated with buying caps, collars, or floors a n d presents a new a p p r o a c h for d e t e r m i n i n g t h e expected cost of default on these instruments. The author Atlanta is an economist Fed's Research ser of Noonan, cussions cap about Astley, in the financial Department. and Pearce, the cap market section He thanks Inc., for helpful and for providing rates. of the Igor A. Lamdata dison What Is an Interest-Rate Cap? An interest-rate cap, sometimes called a ceiling, is a financial instrument that effectively places a maximum a m o u n t on the interest paym e n t m a d e on floating-rate debt. Many businesses borrow funds through loans or b o n d s on which t h e p e r i o d i c interest p a y m e n t varies according to a prespecified short-term interest rate. The most widely used rate in both the caps a n d swaps markets is the London Interbank Offered Rate (LIBOR), which is t h e rate offered on Eurodollar d e p o s i t s of o n e international bank held at another. 4 A typical example of floating-rate borrowing might b e a firm taking out a $20 million b a n k loan on which t h e interest would b e paid every three months at 50 basis points (hundredths of a percent) over LIBOR prevailing at each payment date. Other shortterm rates that are used in conjunction with caps include commercial bank certificate of d e p o s i t (CD) rates, the prime interest rate, Treasury bill rates, commercial p a p e r rates, a n d certain taxexempt interest rates. D a t a on t h e size of t h e c a p m a r k e t are sketchy. The International Swap Dealers Association (ISDA) conducted a survey of its members in March 1989, and 44 of the association's 97 m e m b e r s responded. Almost 90 percent of the respondents reported participating in the markets for caps, collars, floors, a n d o p t i o n s on swaps. As of year-end 1988, these m e m b e r s alone held 7,521 caps, collars, a n d floors, with a total notional principal of $290 billion. The volume conducted through 1988 was reported as having notional principal of $172 billion. These figures inflate the size of the market considerably because they are not adjusted for transactions a m o n g t h e dealers themselves, such as the purchase or sale of caps or floors to h e d g e existing positions in these instruments. On the other hand, the survey did not cover the entire market. Nonetheless, t h e figures probably still greatly overstate the size of the market, net of interdealer transactions or positions. 5 The interest-rate swaps market is vastly larger at o v e r $ l trillion. Most s t u d i e s of caps concern a g r e e m e n t s offered by commercial or investment banks to borrowers seeking interest-rate protection. These instruments are often tailored to a client's II FEDERAL RESERVE BANK OF ATLANTA needs, and, particularly in the case of caps, may b e marketable or negotiable. Caps, collars, and floors can also b e manufactured out of basic derivative assets: options or futures contracts, or a combination of the two. The following discussion will define caps, collars, and floors in terms of option contracts, which are the simplest type of derivative asset. Call a n d Put Options. An option is a financial contract with a fixed expiration date that offers either a positive return (payoff) or nothing at maturity, d e p e n d i n g on the value of the asset underlying the option. At expiration, a call option gives the purchaser the right, but not the obligation, to buy a fixed number of units of the underlying asset if that asset's price exceeds a level specified in the option contract. The seller or "writer" of a call has the obligation to sell the underlying asset at the specified exercise or strike price if the call expires "in the money." The payoff on a call need not actually involve delivery of the underlying asset to the call buyer b u t rather can b e settled by a cash payment. The caps market, for example, uses cash settlement. If the asset price finishes below the exercise price, t h e call is said to expire " o u t of t h e money." Put options are analogous to calls. In this case, though, the purchaser has the right to sell, rather than buy, a fixed number of units of the underlying asset if the asset price is below the exercise price. The options discussed in this article will all b e "European" options, which can only b e exercised on the expiration date, as o p p o s e d to "American" options, which can b e exercised any time before or at expiration. As will b e seen, caps, floors, a n d collars are European-style option-based instruments, and the European interest-rate call option is the basic building block for the interest-rate cap. Options on d e b t instruments can b e confusing if it is unclear just what the option "price" represents. For d e b t instruments, t h e strike price is referred to as the strike level, reflecting an interest rate. Recall that the price of a d e b t instrument, such as a Treasury bill or CD, moves inversely with its corresponding interest rate; as the interest rate of a Treasury bill rises, its price falls. Thus, a call on a Treasury bill rate is effectively a put on its price. (To keep the exposition clear, all discussion will b e in terms of options on interest rates. The strike price will b e re4 ferred to as the strike level.) A call with a strike level of 8 percent (on an annual basis) on some notional amount of principal is effectively a cap on a floating-rate loan payment coinciding with the expiration of this option. (The notional amount of principal is a sum used as the basis for the option payoff computation. Cap, collar, and floor agreements d o not involve any exchange of principal.) Assume the call's payment date, known as the reset date, falls semiannually. If the interest rate is less than 8 percent on the reset date, the call expires worthless. If t h e interest rate exceeds 8 percent, the call pays off the difference between the actual interest rate and the strike level times the notional principal, in turn multiplied by the fraction of a year that has elapsed since purchase of t h e option. For example, if 7Cjaps, floors, and collars are Europe an-style option-based instruments, and the European interest-rate call option is the basic building block for the interest-rate cap. " the actual rate of interest six months later were 10 percent a n d if t h e notional principal were $1 million, the payment received from the call writer would b e 2 percent (the 10 percent actual rate minus the 8 percent strike level) x $ 1,000,000 x 180/360 = $ 10,000. A put option on an interest payment works in a similar way a n d is the foundation for t h e interest-rate floor. The holder of a floating-rate loan could protect against a loss in interest income from the loan by buying an interest-rate put. A fall in the interest rate below the strike level of the put would result in a payoff from the option, offsetting the interest income lost because of a lower interest payment on the loan. An option writer is basically an insurer who receives a premium payment from the option buyer when an option is created (sold). In fact, the option price is alternatively called the option premium. The same party can simultaneECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 ously write and buy options, thus creating an interest-rate collar. Before exploring this strategy further, o p t i o n pricing must b e reviewed briefly. Option Pricing. An option's price before expiration d e p e n d s on several variables, including the value of the underlying asset on which the option is written, the risk-free rate of interest (usually a Treasury bill that matures at the same time as the option), the time remaining before expiration, the strike price or level, and the volatility of the underlying asset price. 6 For later reference, readers should know how an option price changes in response to a change in an underlying variable, all other variables remaining constant. A call price rises (falls) when the underlying asset price, volatility, or time to expiration increases (decreases). It falls (rises) "A cap can... be perceived as a series of interest-rate call options for successively more distant reset dates; a floor is a similarly constructed series of put options." with an increase (decrease) in t h e exercise price. A p u t price rises (falls) with an increase (decrease) in the strike price or volatility. It falls (rises) with an increase (decrease) in the underlying asset price or interest rate. Unlike a call price, a put price is not unambiguously affected by an increase in the time to expiration, but the put price d e p e n d s at any time on how far in or out of the money the put is.7 For an interest-rate call option, the higher the strike level compared to the current interest rate, the lower the option value. Choosing a high strike level (out-of-the-money) call is less expensive than buying an at-the-money or in-themoney call. Similarly, a low strike level (outof-the-money) put is cheaper than o n e with a higher strike level. This relationship between an option's strike level and its price (the amount the option is out of the money) is analogous to a large deductible FEDERAL RESERVE BANK OF ATLANTA on an insurance policy. Such a policy is less likely to pay off and is therefore less expensive. Likewise, the cost of interest-rate "insurance" can b e reduced by taking a large deductible— that is, buying an out-of-the-money option— and thereby protecting only against large, adverse interest-rate movements. Creating an interest-rate collar is another method for reducing the cost of interest-rate insurance. The call-option p r e m i u m for an interest-rate cap may b e partially or completely offset by selling a p u t o p t i o n that sets an interest-rate floor. For a floating-rate debt holder, the effect of this dual purchase is to protect against rate movements above the cap level while simultaneously giving u p potential interest savings if the rate drops below the floor level. If the cap and floor levels of a collar are narrowed to the extent that they coincide at the current floating interest rate—that is, both put and call options are at the money—the resulting collar is so tight that it is similar to a forward contract on an interest rate, which is a derivative asset that locks in the current forward rate. When the contract expires, the change in the contract's value that has occurred since t h e inception of the contract exactly offsets t h e change in the interest payment due. A rise in the floating-rate payment is matched by an equal gain in the interest paid to the contract holder; a fall in the floating-rate payment is balanced by an equal loss on the forward contract. In effect, a forward contract converts a floating-rate payment to a fixed-rate payment. The discussion thus far has been about a single payment, yet, as m e n t i o n e d earlier, actual cap, collar, or floor agreements are designed to hedge a series of cash flows, not just one. A cap can thus b e perceived as a series of interest-rate call options for successively more distant reset dates; a floor is a similarly constructed series of p u t options. Assume that an interest payment on floating-rate d e b t falls d u e in three months, at the next reset date. If the interest rate on the reset d a t e exceeds the strike level, the cap writer would make a payment to the cap buyer on a date to coincide with the cap buyer's own payment date on the underlying floating-rate debt. A collar that consists of a series of at-themoney call and put options is equivalent to an II interest-rate swap. Buying the cap and selling the floor transforms floating-rate d e b t to fixedrate debt, whereas selling the cap and buying the floor switches fixed-rate d e b t into floatingrate debt. A swap that is constructed out of cap and floor agreements is ca11 ed a synthetic swap. Caps brokers and dealers will sometimes determine rates on floors by deriving the rate from swap and cap rates, which come from instruments that are more actively traded than floors and therefore more accurately reflect current market values. In practice, swaps are not usually p u t together from cap and floor agreements. Caps and floors are more readily tradable than swaps because credit risk is one-sided; swaps carry a credit risk that is two-sided in nature. Matching buyers and sellers for swaps is therefore more involved than for caps or floors. 8 Examples of some caps, collars, and floors should help the reader understand their operation. As the foregoing single-payment-date discussion illustrates, creating these instruments amounts to an exercise in option pricing. O n e widely used option-pricing model, known as the Black futures option model, is used in the following examples. 9 Robert Tompkins (1989) explains caps pricing in terms of Black's model, and the examples that follow are loosely patterned on Tompkins' approach. The chief virtue of the Black m o d e l is its simplicity and ease of use, even though it has a serious internal inconsistency when used t o value d e b t options: the assumption that the short-term interest rate (that is, the Treasury bill rate) is constant. Options on short-term interest rates have value, though, only if those rates are less than perfectly predictable. In the last section of this paper, a more complex model that does not suffer from this shortcoming is used to price options. 1 0 Eurodollar Futures a n d Forward LIBOR. In order to give realistic yet simple examples of caps, collars, and floors, this article assumes that the reset dates coincide with the expiration dates of Eurodollar futures contracts, which are traded at the Chicago Mercantile Exchange (CME) and the London International Financial Futures Exchange (LIFFE). Purchase of a Eurodollar futures contract locks in the interest payment on a $ 1 million three-month time deposit to b e m a d e upon expiration of the futures con6 tract. The interest rate on the deposit is threemonth LIBOR. On the other hand, the seller of a Eurodollar futures contract is obligated to pay the specified LIBOR-based interest payment at expiration. 11 Eurodollar futures expire in a quarterly cycle two London business days prior to the third W e d n e s d a y of March, June, S e p t e m b e r , a n d December. The Chicago Mercantile Exchange currently offers contract expiration months extending four years, with only March and September contracts for the fourth year.12 The interest rate implied by a Eurodollar futures price may b e regarded as a forward interest rate, that is, the three-month LIBOR expected by the market to prevail at the expiration date for each contract. 13 The Black model uses the futures price for a particular contract expiration month as an input to determine the value of a European call and put option on that contract. In the case of Eurodollar futures contracts, the add-on yield (100 minus the futures price) is plugged into Black's formula. Another crucial variable is the volatil ity, which is either estimated from the historical volatility of the Eurodollar futures yield or obtained as an i m p l i e d volatility from traded Eurodollar futures options. 1 4 Chart I shows the recent behavior of both of these volatility measures. Again, higher volatility results in highercost call a n d p u t o p t i o n s a n d h e n c e more expensive caps and floors. Table I gives two-year cap, floor, and collar prices on three-month LIBOR for two arbitrarily chosen dates, June 19, 1989, and December 14, 1987, that give reset dates which coincide with Eurodollar futures expiration dates. The first date illustrates pricing during a relatively low volatility period when t h e term structure of LIBOR rates, as given by the "strip" of prices on successively more distant contracts, was just about flat. The market was predicting virtually no change in short-term interest rates over this two-year horizon. In panel A of Table 1, the contract expiration months are given along with the forward rates or add-on yields for each futures contract. The row labeled time to expiration shows the n u m b e r of days from the creation of the cap, floor, or collar to the expiration date for each contract. Another input into Black's formula, the risk-free rate, is taken to be the Treasury bill or zero-coupon b o n d yield for which the ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 volatility in Chart 1. Implied and Historical Volatilities for Eurodollar Futures Prices percent 50 (daily data, December January 1986 Higher volatility, sive caps, Note: 1985-July 1989) - f Gaps shown January 1987 such as that exhibited in Charts in Chart 2, 3, and 1 result around January 1988 the time of the October 1987 stock-market January 1989 break, results in more expen- 4. from missing observations. Source: Chicago Mercantile Exchange. expiration falls nearest t o the futures expiration date. The first example prices a two-year 10 percent cap, which consists of t h e sum of seven call options. At 10 percent, this cap is clearly out of the money. The c o m p u t e d call option price is expressed in basis points. The calls b e c o m e progressively more expensive as t h e time to expiration increases, reflecting t h e rising t i m e value of the calls. The shorter-maturity calls have little value because they are out of the money and, given the volatility, only a slight chance exists that they might finish in the money. Although the more distant calls are also out of the money, there is more t i m e (and more uncertainty) a b o u t what LIBOR will do. Thus, their value is greater because of the higher probability that they might expire in t h e money. The FEDERAL RESERVE BANK OF ATLANTA sum of these calls is t h e cap rate, which is 147 basis points (rounded from I47.1).' 5 For a threemonth contract with a nominal face value of $ l million, a one-basis-point m o v e is worth $25 ($l million x .01% x 90/360). Translated into dollars, 147.1 basis points is $3,677.60 (147.1 x $25), which represents the dollar cost of placing a cap for two years on a $1 million loan. This example was c o m p u t e d ignoring t h e risk of default on the cap. It also a s s u m e s that p a y m e n t s at reset dates, if owed, are m a d e at t h e time of the reset date. Next, a slightly out-of-the-money 7.5 percent floor is shown. The total cost is 96 basis points, or $2,396.61. As m e n t i o n e d above, t h e cost of interest-rate protection can b e r e d u c e d by creating a collar, which is s o m e t i m e s referred to as a ceiling-floor agreement. In this example, II Table 1. Examples of Two-Year Cap, Floor, and Collar Prices on Three-Month LIBOR Panel A: June 19,1989; Volatility, 18 percent September December 1988 1988 March 1989 June 1989 September December 1989 1989 March 1990 91 9.02 8.46 182 8.84 8.47 273 8.64 8.54 364 8.71 8.56 455 8.77 8.59 546 8.87 8.59 637 8.86 8.56 Call prices (10.0 percent strike) 5.3 10.3 12.9 19.9 26.5 34.1 38.1 Put prices (7.5 percent strike) .6 4.7 11.8 15.4 18.6 20.6 24.2 Time to expiration (days) Forward rate Risk-free rate Zero-cost collar 10 percent cap implies 7.85 percent floor 7.5 percent floor Cost in basis points: 96 Cost in dollars: $2,396.61 10 percent cap Cost in basis points: 147 Cost in dollars: $3,677.60 Panel B: June 19,1989; Volatility, 18 percent September December 1988 1988 March 1989 June 1989 September December 1989 1989 March 1990 Call prices (11 percent strike) .4 2.2 3.8 7.6 11.8 16.9 20.3 Put prices (7 percent strike) .1 1.3 4.7 7.2 9.5 11.2 13.9 Zero-cost collar 11 percent cap implies 7.19 percent floor 7 percent floor Cost in basis points: 48 Cost in dollars: $1,198.08 11 percent cap Cost in basis points: 63 Cost in dollars: $1,575.84 Panel C: December 14,1987; Volatility, 25 percent March 1989 June 1989 September 1989 371 8.88 7.51 455 9.11 7.66 553 9.31 7.79 644 9.48 7.92 28.9 45.9 62.0 78.0 91.6 26.8 29.0 30.5 32.9 34.8 September December 1988 1988 March 1988 June 1988 91 8.09 6.09 182 8.34 6.79 280 8.62 7.11 Call prices (10 percent strike) 2.1 12.5 Put prices (7.5 percent strike) 16.2 23.0 Time to expiration (days) Forward rate Risk-free rate 10 percent cap Cost in basis points: 321 Cost in dollars: $8,025.53 Note: Dollar 8 amount is for $1,000,000 7.5 percent floor Cost in basis points: 193 Cost in dollars: $4,829.68 in notional Zero-cost collar 10 percent cap implies 8.05 percent floor principal. ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 selling a 7.5 percent floor would substantially reduce t h e cost of a 10 percent cap. The combination would cost a b o u t 51 basis points, or $1,281. However, by judiciously selecting the floor level—in this case, 7.85 percent—the price of the cap can b e driven to zero. 16 Marketing p e o p l e d e l i g h t in explaining that d o w n s i d e interest-rate protection (the cap) can b e obtained at no cost: just sell a floor. 17 Of course, though, this strategy carries a cost. The holder of an interest-rate collar has traded away potential savings on interest-rate declines below the floor. This caveat notwithstanding, a collar for which the floor exactly matches the cap will b e referred to as a zero-cost collar. Panel B illustrates how t h e cost of caps a n d floors falls by selecting more out-of-the-money levels. Increasing the cap by o n e percentage point to 11 percent reduces t h e cap rate substantially to 63 basis points, or $1,575.84. Decreasing t h e floor by half a p e r c e n t a g e p o i n t t o 7 percent more than halves the cost to 48 basis points, or $ 1,198.08. A zero-cost collar with an 11 percent cap effectively lowers the floor to 7.19 percent. The final example, reflected in panel C of Table 1, shows prices for caps, collars, a n d floors during the relatively high volatility period after the October 1987 stock market break. As depicted in Chart 1, Eurodollar futures' volatility surged during a n d after t h e O c t o b e r 21 crash; t h e degree of fluctuation had abated greatly by late January, although it had not returned completely to precrash levels. The implied volatility was 25 percent on D e c e m b e r 14, 1987, as compared to 18 percent on June 19, 1989, in the earlier examples. The 10 percent cap priced in panel C is substantially m o r e costly than the o n e in p a n e l A. The cost is 321 basis points, or $8,025.53. Another important factor contributing to the higher cost is t h e rising structure of LIBOR forward rates. Although t h e futures nearest to expiration indicate a forward rate of 8.09 percent as compared to 9.02 percent in the June 19, 1989, example, t h e distant futures for December 14, 1987, have forward rates that are well above those for June 19. The upward sloping term structure of interest rates for D e c e m b e r 14 reinforces the effect of higher volatility on raising cap and floor rates. The floor is more expensive as well at 193 basis points, or $4,829.68. Interestingly, the zero-cost collar with a 10 per- cent cap is only slightly more constraining with a floor of 8.05 percent as compared t o 7.85 percent in the previous example, which exhibited low volatility and flat term structure. 18 Caps, Collars, and Floors in Practice At first sight, creating caps, collars, a n d floors would a p p e a r to b e a s i m p l e matter because options are traded on the Eurodollar futures contract. Selecting the appropriate strike levels and expiration dates would a p p e a r to b e all o n e n e e d s t o manufacture a cap, collar, or floor. However, as m e n t i o n e d above, Eurodollar contracts extend into the future for at most four years (which nevertheless is an unusually large n u m b e r of m o n t h s for a futures contract). Eurodollar futures options traded at the Chicago Mercantile Exchange currently have expiration dates ranging out only two years, in a quarterly cycle that matches that of the Eurodollar futures contracts. 19 Another limitation of Eurodollar futures options is that only contracts expiring within the three months or so from t h e current d a t e are liquid, that is, they are the only ones that are actively traded so that their prices at any t i m e reliably reflect equilibrium values. The options also are limited to strike levels in increments of 25 basis points, whereas the futures have increments of o n e basis point. Unlike Eurodollar futures and options, caps, collars, a n d floors have b e e n created with maturities extending as much as 10 years. Furthermore, actual caps, collars, a n d floors can b e created on any day, not just on futures and options expiration dates. The actual use of futures and options t o fashion caps, collars, and floors is neither a straightforward nor a riskless matter. The solution to this p r o b l e m is t h e use of existing futures a n d options contracts to create the desired positions synthetically. Synthesizing an options position using options or futures contracts—or a combination of the two—requires not only taking appropriate positions in the existing liquid contracts b u t also altering that position over t i m e so that the value of t h e actual position tracks or "replicates" t h e desired position. This process is known as dynamic hedging. Theoretically, the replicating portfolio of actual II FEDERAL RESERVE BANK OF ATLANTA futures a n d options contracts can exactly match the value of, say, a cap sold to a counterparty. 2 0 In reality, managing a replicating portfolio is a risky and costly activity. 21 Tracking errors cumulate since costly trading cannot b e conducted continuously as is theoretically required and because mismatches can occur with t h e expiration dates and possibly also with t h e interest rates involved. Using E u r o d o l l a r futures t o h e d g e a cap b a s e d on the commercial p a p e r rate exemplifies t h e latter. 22 The Over-the-Counter Market In view of t h e c o m p l e x i t i e s a n d risks of dynamic-hedging strategies, most cap, collar, and floor users prefer over-the-counter instruments. Commercial a n d investment banks create these instruments themselves, possibly by manufacturing them through dynamic hedging. Nonfinancial users t e n d to rely on the expertise of these financial institutions and are willing to pay for t h e convenience of interest-rate risk m a n a g e m e n t products issued through an intermediary. The intermediaries may also b e more willing t o bear t h e risks associated with hedging because of t h e scale of their operations. In fact, Keith C. Brown and Donald J. Smith (1988) describe the increasing involvement of banks in offering interest-rate risk m a n a g e m e n t instruments as t h e reintermediation of commercial b a n k i n g . Since t h e 1970s, commercial b a n k s have played less of a role in channeling funds from lenders to borrowers. With the growth of interest-rate risk m a n a g e m e n t , though, their intermediary role is being restored, albeit in a different form. Commercial banks, particularly t h e largest money-center banks, are better a b l e to absorb a n d control t h e hedging risks associated with managing a caps, collars, and floors portfolio, and these institutions are better a b l e to evaluate t h e credit risks inherent in instruments bought from other parties. Credit risk arises b e c a u s e any counterparty selling a cap, for example, is obligated to m a k e payments if the cap moves in the money on a reset date. That counterparty could go bankrupt at s o m e point during t h e course of the cap agreement a n d would default on its obligation. (This issue is 10 examined in detail in the last section of this article.) By taking positions in caps, collars, and floors, commercial banks—and to a lesser extent, investment banks—act as dealers by buying a n d selling to any counterparties. Within their portfolio or " b o o k " of caps a n d floors, individual instruments partially net out, leaving a residual exposed position that t h e banks then h e d g e in the options a n d futures markets. Much trading of caps, collars, a n d floors consists of purchases a n d sales of these instruments t o adjust positions a n d risk exposures, so much of the caps market's v o l u m e is generated by interdealer transactions. In addition to t h e d o z e n or so commercial a n d investment banks in New York and London that d o m i n a t e t h e caps market, there are a b o u t half a d o z e n caps brokers, who d o not take positions themselves b u t instead match buyer a n d seller. 23 Caps, collars, a n d floors are usually sold in multiples of $5 million, b u t because of the cust o m i z e d nature of the over-the-counter market other amounts can b e arranged. Most caps have terms that range from o n e to five years and have reset d a t e s or f r e q u e n c i e s that are usually monthly, quarterly, or semiannual. Caps b a s e d on three-month LIBOR are the most c o m m o n and t h e most liquid or tradable. From the purchaser's point of view, buying a cap that matches t h e characteristics of t h e liability being h e d g e d might seem best. Even strike levels and notional principal amounts can b e chosen to vary over the term of an agreement in a predeterm i n e d way, b u t good fit comes at a price. Transactions costs are higher for such tailored products, as reflected by the larger difference between b i d a n d offer rates on u n c o m m o n caps. This wider spread also increases t h e cost of removing caps by selling them before their term expires. Many users o p t for a liquid cap a n d are willing to a b s o r b t h e basis risk—the risk from a mismatch of interest basis or other characteristics—in order t o avoid the higher cost of a less liquid instrument. Caps a n d floors are usually available at strike levels within several percentage points of t h e current interest-rate basis and are most commonly written out of the money. Settlement dates typically occur after reset dates, u p o n maturity of t h e underlying instrument. For exa m p l e , interest on a three-month Eurodollar d e p o s i t is credited u p o n maturity of t h e deECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 posit. A cap on three-month LIBOR would have a three-month lag between a reset date and actual settlement. Most payments for caps are m a d e u p front, although they can also b e amortized. When a cap and a floating-rate loan c o m e from the same institution, the two are usually treated as a single instrument; thus, when the floating rate exceeds the strike level, payment is limited to the strike level and the cap does not pay off directly. 24 Long-Term Caps. During the mid-1980s, early in the development of the caps market, longerterm caps were created directly from floatingrate securities rather than synthetically. Two kinds of floating-rate instruments were used: floating-rate CDs and floating-rate notes. 2 5 Floating-rate notes are d e b t obligations usually indexed to LIBOR, and floating-rate CDs are medium-term deposit instruments that are also typically indexed to LIBOR. The innovation that sparked much activity in the caps market was the issuance of capped floating-rate notes and CDs that in turn had their caps stripped off and sold as separate instruments sometimes known as "free-standing" caps. As an illustration, consider the floating-rate CD. Banks use ordinary CDs as well as variablerate CDs to acquire funds for the purpose of making loans and funding other balance-sheet assets. The capped floating-rate CD was prom o t e d as a m e t h o d of raising funds below LIBOR, the rate on an uncapped CD with a variable rate of interest. The reason is that, after issuing a capped floating-rate CD to a depositor, a bank could then sell the corresponding cap into the caps market and collect premium income. Because CDs of this type typically fund floating-rate loans, the bank would b e fully h e d g e d after selling t h e cap. Funding costs would b e lowered if the premium for the cap on the floating-rate CD were less than the premium t h a t t h e b a n k c o l l e c t e d u p o n selling the cap into the market. 26 This method of creating or "sourcing" caps, floors, and collars—through capped floating-rate CDs and floating-rate n o t e s became extremely popular b u t was short-lived. Reportedly, the longer-term caps were gradually perceived to b e undervalued, such that cap writers were not being compensated for the risks of having to make payments to cap holders if interest rates rose above strike levels. 27 Also contributing to the demise of this method of FEDERAL RESERVE BANK OF ATLANTA sourcing was a flattening of the yield curve that m a d e floating-rate borrowing less attractive and reduced cap prices. Today, few caps, collars, or floors are created beyond the five-year maturity. Charts 2-4 give actual cap bid and offer rates, in basis points, q u o t e d by o n e major caps broker in New York. The bid rate is the rate at which the broker is wil 1 ing to buy a cap; the offer rate is the rate at which the broker sells a cap. The spread between the two represents the transactions costs of match ing buyer with seller. Charts 2, 3, and 4, respectively, give the rates on two-year 8 percent, three-year 10 percent, and five-year 10 percent caps. These rates are just a sample; many other strike levels are available. The strike levels quoted change over time as interest rates change. Cap strike levels that move too far in the money or out of the money are discontinued and replaced by caps with strike levels that are in greater d e m a n d . All of these series are highly correlated. They are also correlated with the volatilities shown in Chart I, which are a major determinant of cap values. 28 The Motivation for Hedging and Some Hypothetical Examples With some background on the caps, collars, and floors market, the use of interest-rate risk management instruments can now b e put into perspective by briefly considering the nature of hedging. Caps, collars, and floors are often talked about in terms of an insurance analogy. They are instruments that can b e used to hedge assets orliabilitiesand thus protect against loss resulting from interest-rate risk. In practice, though, distinguishing between hedging and speculating in interest-rate risk management is s o m e t i m e s difficult, especially with optionbased instruments. Discretion is required in selecting the timing of the hedge, the strike level, and the maturity of the instrument, all of which are usually predicated on some opinion of what interest rates and other variables are expected to do. Selling a cap or floor, for example, is a way to generate income on a fixedincome portfolio by collecting the premiums. The decision to sell often reflects a difference of opinion regarding the volatility implied by the II Chart 2. Two-Year 8 Percent Cap Bid and Offer Rates Rate in basis points 400 T (daily data, March 1987-October 1988) 300 100 I T o J March 1987 1 1 p September 1987 March 1988 The spread between the bid and offer rates represents the transactions costs Note: in Charts 2, 3, and 4 reflect rates were not available. Gaps days for which of matching September 1988 buyer and seller. Source: Noonan, Astley, and Pearce, Inc. cap or floor. If a m o n e y manager thinks a cap is overvalued because t h e market's expectation of volatility is higher than his or her own, then selling an out-of-the-money c a p might b e a g o o d move. If t h e m o n e y manager's j u d g m e n t a b o u t volatility is correct, even small upward moves in the interest rate may not w i p e out all of the prem i u m income. At t h e s a m e time, t h e sale provides a limited h e d g e against small downward moves in rates, again because of t h e p r e m i u m receipt. Even determining the effect of hedging can b e problematic, since a firm's purchase of a cap, for example, to h e d g e t h e interest-rate risk of a particular liability could increase t h e variability of the firm's net worth. The financial claim b e i n g h e d g e d may itself h e l p offset the variability of another financial claim on t h e balance sheet. 12 The net result of a specific h e d g e could b e t o increase the interest-rate risk exposure of the firm. A more fundamental issue is why firms h e d g e in t h e first place. A basic insight derived from t h e economics of uncertainty is that risk aversion leads individuals to prefer stable income a n d c o n s u m p t i o n streams t o highly variable ones. Given an assumption of risk aversion on the part of decision makers, o n e can show that their welfare or utility (that is, their economic well-being) is greater over time if they enjoy smooth income or c o n s u m p t i o n opportunities rather than erratic ones. 2 9 Hedging is a way of improving economic well-being by trading off income or c o n s u m p t i o n in g o o d times for greater income or consumption in b a d times. Thus, a hedging strategy serves a well-defined p u r p o s e ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Rate in basis points Chart 3. Three-Year 10 Percent Cap Bid and Offer Rates (daily data, March 1987-June 1989) 300 200 - 100 March 1987 The rates Chart depicted September 1987 in Chart 3 are highly March 1988 correlated with September 1988 those in Charts March 1989 2 and 4, as well as with the volatilities in 1. Source: Noonan, Astley, and Pearce, Inc. for risk-averse economic agents, such as farmers or a firm's owner-manager. The issue is less clear-cut for widely held corporations, which actually are t h e typical users of interest-rate risk-management tools. A corporation owned by a large n u m b e r of stockholders n e e d not operate like a risk-averse decision maker because each stockholder can insulate his or her wealth and consumption opportunities from risk, specific t o t h e corporation's activities, by h o l d i n g a diversified portfolio of assets. Clifford W. Smith a n d R e n é M. Stulz ( 1985) surveyed managers of widely held, value-maximizing corporations t o d e t e r m i n e t h e motivations behind hedging behavior. According to the researchers, managers engage in hedging of a firm's value for three basic reasons. The first explanation is tax-related; Smith and Stulz arFEDERAL RESERVE BANK OF ATLANTA g u e that, on average, a less variable pretax firm value implies a higher after-tax firm value than d o e s a more variable pretax value. The reasoning turns on their assumption that t h e level of corporate tax liabilities grows at an increasing rate with rising pretax firm value because of t h e progressive structure of the tax code. Hedging helps reduce the variability of pretax firm value a n d therefore raises after-tax value. Second, Smith and Stulz maintain that hedging lowers the probability that t h e firm will g o bankrupt a n d thus incur bankruptcy costs. Hedging firm value would benefit stockholders by reducing the expected future costs of bankruptcy that lower current firm value. A related p o i n t is that a firm's d e b t may often contain covenants that force the c o m p a n y to alter investment policies that the shareholders would like to see underI3 Rate in basis points 700 ^L Chart 2. Five-Year 10 Percent Cap Bid and Offer Rates (daily data, March 1987-June 1989) 600 500- 400 300 200 100 September 1987 The longer expiration date on a five-year cap results March 1989 September 1988 March 1988 in prices that are relatively higher than those on shorter-term caps. Source: Noonan, Astley, and Pearce, Inc. taken. Hedging reduces the likelihood of financial distress and the limitations on managers' discretion that b o n d covenants may impose. A third reason for hedging is that when managerial compensation is tied to the firm's value, managers may become more risk-averse in order to maintain that value. Participants in the Caps, Collars, a n d Floors M a r k e t While the precise social value of interestrate risk management products is not fully understood in the case of widely held corporations, such products are clearly becoming increasingly popular among corporate treasurers and other financial managers. End users of caps, collars, and floors typically include firms seeking to limit exposure to adverse movements in shortterm interest rates, such as a firm that sells commercial paper to fund its purchases of inventory. 14 Specific market participants are depository institutions, particularly savings and loan associations (S&Ls) ; corporations going through leveraged buyouts (LBOs) or taking on d e b t to fend off hostile takeovers; and real estate developers, who are often highly leveraged with floating-rate debt. Unfortunately, the only information about these applications is anecdotal. Also, compared to the potential market, the actual market is probably very small. Many potential users are unaware of or cautious about interest-rate risk management instruments. Any user of interest-rate swaps is potentially also a user of caps, collars, and floors. Larry D. Wall and John J. Pringle (1988) conducted a systematic search of annual reports for 4,000 firms that used interest-rate swaps in 1986. The stocks of these firms were traded on the New York ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Stock Exchange, the American Stock Exchange, or t h e over-the-counter market. Of this sample, 250 firms were identified as swaps market participants. Over 50 percent of this group were banks, savings a n d loans, a n d other financial services firms; commercial banks alone accounted for half of these. In addition, Wall a n d Pringle report that "the overwhelming majority of thrifts (59 percent), manufacturing firms (69 percent), a n d nonfinancial, n o n m a n u f a c t u r i n g firms (77 percent) are exclusively fixed-rate payers." 3 0 As a conjecture, the profile of caps, collars, a n d floor users may b e q u i t e similar to that for swaps users. The fact that credit risks for caps and floors are one-sided, however, suggests that firms with weaker credit ratings probably use caps and floors because they cannot gain access to t h e swaps market on favorable terms. Anecdotal accounts from various sources illustrate how different end users e m p l o y caps, collars, or floors in their management of interestrate risk. Many savings a n d loans, for instance, have b e e n active users of these option-based instruments. The interest-rate risk confronting S&Ls, a n d d e p o s i t o r y institutions generally, may b e considered in terms of their net interest margins, that is, the difference between t h e rates at which an institution lends a n d borrows. S&Ls are particularly vulnerable t o changes in interest rates because maturities (or alternatively, the durations) of these institutions' assets, p r e d o m i n a n t l y long-term mortgages, greatly exceed t h e maturities of their liabilities, most often short-term t i m e a n d savings deposits. Thus, a rise in rates raises t h e interest expense on an S&L's short-term liabilities with possibly little increase in interest earnings on its mortgages. The net interest margin narrows and could very well b e c o m e negative. O n e solution is to convert the floating-rate interest expense on the liabilities into fixed-rate payments via an interest-rate swap. The net interest margin would then b e c o m e much more stable. However, a weak credit standing could m a k e such a swap too expensive or unobtainable. A cap on the floating-rate liabilities could b e an effective alternative. An S&L's credit rating would b e irrelevant to a cap writer, who bears no credit exposure. 31 As another example, consider a commercial bank's portfolio manager w h o is responsible for overseeing a portfolio of floating-rate notes. Suppose this manager believes that a large drop in short-term interest rates, currently at a b o u t 8 percent, is a b o u t to occur. H e wants to protect the portfolio's earnings and therefore buys an out-of-the-money 7 percent interest-rate floor. Concerned a b o u t the cost of this protection a n d reasonably convinced that rates will not rise substantially, he also decides to sell a 9 percent interest-rate cap to create a collar on the portfolio. This e x a m p l e highlights t h e discretion involved in selecting a hedge. A floor could have b e e n in place all along, b u t maintaining a floor reduces a portfol io's return by the a m o u n t of the premium expense. Only when the manager has strong concerns a b o u t a drop in rates is the floor purchased. As a final example, the corporate treasurer of a consumer products firm is worried a b o u t t h e prospects of a rise in interest rates because her company has recently undergone a leveraged buyout. The financing strategy for the LBO inc l u d e d heavy reliance on floating-rate d e b t secured from a syndicate of commercial banks. The firm's debt-to-equity ratio has soared, and even a m o d e s t rise in rates could bankrupt the company. After the LBO the firm's credit standing was downgraded by the rating services; consequently, access t o t h e swap market is effectively foreclosed. Buying a two-year interestrate cap to cover the firm's floating-rate exposure seems to b e a p r u d e n t action. 32 The treasurer expects earnings will b e more robust after a two-year interval. Also, the protection gained for a relatively short-term horizon makes sense b e c a u s e during this period t h e firm would b e downsizing a n d reorganizing its operations. Credit Risk The earlier discussion of the over-the-counter market for caps, collars, and floors a l l u d e d t o t h e riskof default inherent in these instruments. That risk is present because the seller of a cap or floor is agreeing to fulfill a contract in the event the cap or floor moves in t h e money on a paym e n t date. Since the seller is a firm, whether a commercial bank, investment bank, or nonfinancial institution, its assets are limited, a n d thus the company is exposed to t h e possibility II FEDERAL RESERVE BANK OF ATLANTA of bankruptcy. The p r o b a b i l i t y of d e f a u l t is rather small for t h e typical caps, collar, or floor writer who also typically issues investment-grade b o n d s into the market. Moody's Investors Service, o n e of the major b o n d rating firms, recently released a study indicating that from 1970 t o 1988 t h e average annual rate of default by issuers of investment-grade b o n d s was 0.06 percent, as c o m p a r e d t o an average annual default rate of 3.3 percent for j u n k b o n d issuers. 3 3 Because t h e consequences of default can b e financially damaging, default risk receives careful analysis, particularly by counterparties entering into caps and swaps agreements. This section of t h e article takes a detailed look at how default risk is evaluated a n d how it affects the pricing of caps, collars, a n d floors. The first aspect of t h e p r o b l e m is t o consider the precise nature of t h e default risk or, alternatively, t h e credit exposure. If a cap is in t h e money on a floating-rate reset date, t h e owner of the cap expects to receive a payment from t h e cap writer, as reviewed above. If t h e writer is insolvent and thus fails to m a k e the payment, the owner is again in an u n h e d g e d position a n d must m a k e t h e full floating-rate payment, b u t this is not t h e only c o n s e q u e n c e of default. Provided t h e default d o e s not occur on t h e final reset date, t h e c a p was also hedging future reset dates, which u p o n default are also fully exposed. Thus, credit exposure d e p e n d s on the t i m e that default occurs in the life of a cap agreement. (Note that a parallel argument can b e m a d e for floors and collars.) The cost of default to t h e cap buyer is the cost of replacing the original cap with a new cap from another seller. If interest rates at the default d a t e were identical to t h e initial interest rates and t h e volatility had not changed since t h e original cap was purchased, the replacement cost of t h e cap would b e zero, ignoring transactions costs and differences in credit risks. That is, t h e cost of a new cap for the remaining reset dates would exactly equal t h e current market value of the existing cap (if default had not occurred). The next a n d rather complex aspect of t h e credit risk question to consider concerns the m e t h o d of assessing credit risk when a cap is sold. Bankruptcy of a cap writer has n o impact on cap buyers as long as the c a p stays out of t h e money a n d the cap buyer has n o intention of selling the cap before its term ends. Default 16 occurs only when a c a p is in the money a n d the cap writer is bankrupt. The likelihood of bankruptcy may also b e related to the level of interest rates and thus d e p e n d e n t on the future path of these rate movements. In addition, as just discussed, a cap's replacement cost is a function of where in the life of the cap agreem e n t default occurs. All of these factors should b e weighed in evaluating what t h e potential cost of default could b e and how that should affect t h e price of a cap. Marcelle Arak, Laurie S. G o o d m a n , and Arthur Rones (1986) propose a m e t h o d of computing credit exposure for caps, collars, a n d floors. Their approach amounts to considering different worst-case scenarios that are defined by t h e d e g r e e t o which a c a p can m o v e in t h e money. For a cap t h e c o m p u t e d exposure de- "T/ie cost of default to the cap buyer is the cost of replacing the original cap with a new cap from another seller." p e n d s on t h e size of t h e upward m o v e m e n t in t h e interest rate that could occur during each reset interval. A cap's replacement value will tend at first to increase early in the life of t h e instrument a n d then to decrease toward the e n d of the contract. The credit exposure is taken t o b e t h e maximum replacement value comp u t e d at the reset dates. For example, if the interest-rate volatility based on three-month LIBOR is 10 percent (as measured by t h e annual standard deviation), over a three-month period t h e volatility is 0.10 Xv/( 1/4) = 5 percent. 3 4 Assuming an initial 7 percent LIBOR, three m o n t h s later t h e u p w a r d m o v e w o u l d b e t o 7.35 |7.0 + (0.05 x 7.0)|. Given this rate and a further assumption that rates at all other maturities shifted in parallel, t h e cap replacement value is calculated. Another 5 percent upward move is then c o m p u t e d , giving a new LIBOR of 7.72 17.35 + (7.35 x 0.05)1 and again the replacement ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 value is c o m p u t e d , and so forth. The credit exposure is the maximum value of the replacement cost during the cap agreement. A more conservative evaluation of credit exposure might assume that rates rose by two standard deviations per year instead of one as in the previous example. At two-standarddeviation moves, the actual exposure would, on average, exceed the maximum computed amount only 2.5 percent of the time (as compared to exceeding it 16.5 percent of the time using a one-standard-deviation measure). 35 Arak, Goodman, and Rones give an example of the credit exposure on various collar agreements with a floor equal to 6 percent and a cap equal to 9 percent. For three-month reset intervals, the exposure is 0.44 percent of the notional principal (two-year collar), 0.82 percent (five-year collar), "Computations based on worst-case scenarios implicitly overstate the actual incidence of default because of the arbitrary assumption about sequential interest-rate moves only in one direction. " and 2.68 percent (10-year collar). 3 6 By these researchers' calculations, the credit exposure on collars is rather small, especially compared to similar calculations for other instruments they consider, such as interest-rate swaps and forward contracts. These calculations are intended for commercial banks, which set credit limits for particular customers in order to manage the size of potential losses in the event of default. However, the method put forth by Arak, Goodman, and Rones is not useful for pricing caps— that is, for adjusting the price or rate for the anticipated cost of default. Computations based on worst-case scenarios implicitly overstate the actual incidence of default because of the arbitrary assumption about sequential interest-rate moves only in one direction. A more desirable approach would c o m p u t e the "expected value" of default—the difference between caps not subject to default and those that are. FEDERAL RESERVE BANK OF ATLANTA Caps as Default-Risky Options. Almost all of the option pricing models used to value caps ignore default risk. An exception is the model proposed by Herb Johnson and Stulz (1987), in which they derive formulas for default-risky or "vulnerable" puts and calls. Unfortunately, their formulas cannot b e straightforwardly applied to caps, collars, or floors because of the t i m e dimension involved in these options-based instruments. As has b e e n emphasized, caps are a s e q u e n c e of options—default-risky options. Fulfilling a given option contained in a cap depends on the absence of bankruptcy at earlier reset dates. If bankruptcy occurred earlier, the current option would not b e honored by the cap writer. The sequential time dimension involved in valuing caps makes the mathematics formidably complex. 3 7 This author has tackled the complexity of cap valuation by using computer-intensive methods to handle the intricate contingencies implied in cap, collar, floor, and swap agreements (Peter A. Abken |forthcoming|). His c o m p u t e r m o d e l avoids the contradictory assumption inherent in the Black model used for short-term d e b t options—that short-term interest rates are constant—but at the cost of trading off a simple analytical formula for a complicated computer algorithm. Nevertheless, the intuition behind the new model is simple and easily explained. The value of a European option can be thought of as the average or expected value of its payoffs at expiration, discounted back to the present. Options are difficult to value because the payoff upon expiration is a "kinked," or discontinuous, function of the underlying asset price. A call option is worth zero if the underlying asset price at expiration is less than the strike price, and positive in value if the underlying asset price exceeds the strike price, increasing dollar for dollar with the amount above the strike. The Black-Scholes and Black formulas c o m p u t e the value of a call as the expected value of the future payoffs. 38 S o m e payoffs are more likely to occur than others, and the formulas account for the probabilities associated with the payoffs. M o n t e Carlo S i m u l a t i o n . O n e m e t h o d for valuing options relies on extensive computations to determine the expected payoffs. Known as Monte Carlo simulation, this process was first applied to option pricing problems by Phelim P. Boyle (1977). The standard application II involves stock option pricing. A stock price, on which an option is valued, is assumed to rise and fall randomly overtime, although its value at any point can b e described in terms of its statistical distribution, which is known or assumed. In standard problems the distribution for stock price changes is assumed to b e fully characterized by its mean and variance. Using this information, artificial future stock-price paths, also known as realizations, can b e created numerically by computer. By randomly generating a large enough number of price paths (tens of thousands, at a minimum) and evaluating the payoff on an option with a given strike price at a particular point in time—the option's expiration date—an average over these randomly generated payoffs can b e made. The option price is given by appropriately discounting the expected future payoff into current dollars. Of course, the Black-Scholes formula accomplishes t h e same thing mathematically and is conceptually equivalent. To the penny, both methods will give the same price using identical assumptions regarding the statistical characteristics of stock price movements. The Monte Carlo method, though cumbersome, pays off in cases where the asset price moves in unusual ways, such as in random jumps—for example, d u e to a stock market crash. The Black-Scholes model rules out such movements by assumption. Cap valuation is another area where Monte Carlo methods offer a simplification over approaches that may not otherwise b e mathematically tractable. Three factors taken together contribute to the complexity of default-risky cap valuation. The first is that d e b t prices on instruments like Treasury bills or Eurodollar deposits vary with interest rates. Second, each constituent option in a cap is subject to default and must b e valued as a default-risky option. Third, the payoff on a given option d e p e n d s on the nonoccurrence of default on options from earlier periods. The payoff of a vulnerable call option is the lesser of the firm's value or the default-free option payoff. The value of the firm is the market value of its equity (before including the value of its cap). If the value of the firm that sold the option is greater than the payoff, no default occurs. If the payoff exceeds the firm's value, the company defaults and the option holder receives the value of the firm—or some share of it, as determined by the bankruptcy courts—when 18 the company is liquidated. In view of the fact that a vulnerable call may pay off less, b u t never more, than a default-free call, the value of a vulnerable call must b e less than the value of an otherwise comparable default-free call. An additional consideration for cap valuation, as discussed above, is that default on a cap leaves the cap buyer unhedged. The exposure is t h e replacement value of the cap. Thus, default involves at a minimum replacement of the missed option payoff, and possibly t h e entire remaining value of the cap, if the firm wants to maintain the hedge. Thus, besides valuing default-risky call options, cap valuation must also evaluate such replacement costs. The Elements of the Caps Model. To convey the basic ideas behind construction of the caps model, this section of the article sketches out "An additional consideration for cap valuation ...is that default on a cap leaves the cap buyer unhedged. The exposure is the replacement value of the cap." the model, the technical details of which can b e found in Abken (forthcoming). Three so-called state variables are computer-generated to implement the simulation. The options making u p a cap are valued based on the underlying interest rate, as discussed earlier. The entire path of the term structure of interest rates is generated using the model developed by Stephen M. Schaefer and Eduardo S. Schwartz (1984). Two state variables are the difference or spread between the instantaneous rate and a consol rate (that is, t h e rate on a b o n d having infinite maturity), and the consol rate itself. All other intermediate-maturity d i s c o u n t b o n d s are derived by formula from these two inputs, which describe absolute and relative movements in interest rates at all maturities. The third state variable represents the value of the firm, which also fluctuates randomly over time, reflecting unpredictable changes in interest rates, earnECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 ings, and other variables that determine firm value. The example to b e considered is parallel to the one discussed earlier in Table 1, but the focus is now on credit risk. The cap model will value two-year caps on a three-month interest rate. The cap consists of seven reset dates, at each of which the firm's value is compared to the call option payoff. Default-free and default-risky caps are valued. The difference in the price or rate for these otherwise identical caps is the credit spread for default risk. The example developed below illustrates how default risk is particularly sensitive to t h e correlation over t i m e b e t w e e n firm value a n d interest-rate movements. The parameter values for t h e SchaeferSchwartz m o d e l were estimated from actual "[Djefault risk is particularly sensitive to the correlation over time between firm value and interest-rate movements. " interest-rate data on one-month Treasury bill and 30-year b o n d yields, which served as proxies for the instantaneous interest rate and consol interest rate, respectively. The rates were sampled weekly on Fridays from January 1983 to August 1989. The reader is referred to Abken (forthcoming) for details concerning parameter estimation and other technical details concerning the model. A simplification used in the simulations presented in this article is that whenever a default occurs—that is, when the firm value is less than the option payoff—the replacement value of the cap is not computed. Instead, the option payoff for that reset date is set equal to the negative of the payoff. In other words, the cap owner has to cover the full floating-rate interest payment for that date. Payoffs at future reset dates are assumed to b e zero. Valuing a new cap at current rates would increase the cost of default com- pared to the procedure used here; such valuation, however, would also require separate simulations at each occurrence of default. M o r e Examples. Table 2 gives the results of the simulations. Three panels of this table differ only in the degree that firm value is correlated with interest-rate movements. In the SchaeferSchwartz model, there are two elements to this correlation. Firm value can b e correlated with consol rate movements or spread movements, or both. (TheSchaefer-Schwartz model assumes that the spread and consol rate are uncorrelated, which is supported by empirical research.) Correlation coefficients range from — 1, perfect negative correlation, to 1, perfect positive correlation. Intuitively, a cap writer whose firm value is negatively correlated with interest-rate movements poses a greater credit risk than o n e that is positively correlated. For a given strike level, when interest rates are high, caps are more likely to b e in the money and require a payment from the writer. A negative correlation therefore means that high interest rates are associated with low firm value; hence, default is more probable than it would b e for zero or positive correlations. Also, empirically short- and long-term interest rates are positively correlated. Thus, a negative correlation of firm value and long-term interest rate would also b e associated with a negative correlation between the firm value and interest-rate spread (defined as the short rate less the long rate). Panel A gives the base case of zero correlation of firm value with the interest-rate spread and with the long-term interest rate. The annual default rate for this case is set to 0.13 percent by adjusting the initial value of the firm to give this rate as the outcome of the simulations. 39 The same initial firm value is then used in panels B and C, thereby yield ing new default rates d u e to different correlations with the term structure variables. The initial term structure has a spread of 2.7 percentage points, which was the average spread over the sample period. The short-term interest rate is initially 8 percent and the cap is written at 9 percent. As in Table 1, the option rates are given for each reset date. This table includes default-free and default-risky options; the sum over reset dates for each type is the cap rate. Because the default rate is so low, the discrepancies between default-free and defaultrisky option prices d o not become significant II FEDERAL RESERVE BANK OF ATLANTA Table 2. Default-Free and Default-Risky Cap Rates Estimated by Monte Carlo Simulation, 9.0 Percent Two-Year Cap Initial term structure: Short-term rate, 8.0 percent; Long-term rate, 10.7 percent Panel A: Correlation of firm value with interest-rate spread: 0 Correlation of firm value with long-term rate: 0 1 13 7.94 7.94 Reset date number: Time to expiration (weeks): Default-free option rate: Default-risky option rate: Default-free cap rate: Standard deviation: 95 percent confidence interval: 2 26 17.99 17.99 3 39 26.95 26.95 4 52 35.57 35.53 5 65 43.98 43.81 Default-risky cap rate: Standard deviation: 95 percent confidence interval: 244.16 (1.45) (241.32, 247.00) Credit spread in basis points: Standard deviation: 95 percent confidence interval: Annual default rate: 6 78 51.87 51.34 7 91 59.86 58.71 242.28 (1.43) (239.48, 245.08) 1.89 (0.14) (1.62, 2.16) 0.13 percent Panel B: Correlation of firm value with interest-rate spread: -0.5 Correlation of firm value with long-term rate: -0.5 1 7.94 7.94 Reset date number: Default-free option rate: Default-risky option rate: Default-free cap rate: Standard deviation: 95 percent confidence interval: 2 17.99 17.98 3 26.95 26.79 4 35.57 34.98 5 43.98 42.19 Default-risky cap rate: Standard deviation: 95 percent confidence interval: 244.16 (1.45) (241.32, 247.00) Credit spread in basis points: Standard deviation: 95 percent confidence interval: Annual default rate: 6 51.87 48.15 7 59.86 53.51 231.54 (1.35) (228.89, 234.19) 12.63 (0.35) (11.94,13.32) 0.71 percent Panel C: Correlation of firm value with interest-rate spread: 0.5 Correlation of firm value with long-term rate: 0.5 1 7.94 7.94 Reset date number: Default-free option rate: Default-risky option rate: Default-free cap rate: Standard deviation: 95 percent confidence interval: 2 17.99 17.99 3 26.95 26.95 Sample 20 size for each panel: 50,000 independent 5 43.98 43.98 Default-risky cap rate: Standard deviation: 95 percent confidence interval: 244.16 (1-45) (241.32, 247.00) Credit spread in basis points: Standard deviation: 95 percent confidence interval: Annual default rate: Note: 4 35.57 35.57 draws. 6 51.87 51.87 7 59.86 59.86 244.16 (1.45) (241.32, 247.00) 0.0 (0.0) (0.0,0.0) 0.0 percent Cap rates expressed in basis points. ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 until t h e later reset dates. The default-free cap rate is 244.16 basis points, whereas t h e defaultrisky cap rate is 242.28. The difference of 1.89 basis points is the credit spread. These figures are estimates and have an error associated with them. O n e can arbitrarily red u c e that error by increasing the n u m b e r of realizations u s e d t o c o m p u t e t h e o p t i o n s . Q u a d r u p l i n g t h e n u m b e r of realizations reduces the standard deviation by half. The simulations for each p a n e l were g e n e r a t e d by taking 50,000 i n d e p e n d e n t sets of realizations of t h e state variables. 40 The standard deviation and 95 percentconfidence intervals for each cap rate and the spread are reported in Table 2. The simulation used t o generate panel B was the s a m e as that for panel A in all respects except that t h e correlation between firm value and interest-rate spread is —0.5 instead of zero, a n d t h e correlation b e t w e e n firm value a n d long-term rate is —0.5 instead of zero. The results show a substantial increase in the incidence of default. The b a s e rate in panel A for zero correlations is 0.13 percent, whereas the negative correlations in panel B raise the default rate to 0.71 percent. The credit spread rises from 1.89 basis points to 12.63 basis points. As discussed previously, the reason is that firm value is likely to b e low when interest rates are high. The cap writer has a greater chance of being insolvent when a payment is required. As a final example, the correlations in panel C take the o p p o s i t e signs from those in panel B. The credit spread and annual default rate drop to zero. The greater chance of high firm value coinciding with cap payments reduces the likelihood of default by t h e c a p writer; in this case, the incidence of default drops to zero. The substantial increase in t h e credit spread exhibited in panel B may exaggerate default risk for two reasons. First, t h e cap is assumed t o b e u n h e d g e d by the firm. In other words, t h e company is taking a speculative position. Actual cap writers usually take offsetting positions in other caps or h e d g e by other methods, at least to s o m e degree. Second, the model assumes that failure to cover cap payments is t h e only factor causing bankruptcy. For actual cap writers, the contingent liability posed by a cap is probably small compared to other items on the balance sheet. O n the other hand, the c o m p u t e d credit spread may still b e a good approximation if the c a p serves as a proxy for t h e firm's overall balance sheet exposure to movements in interest rates. No data on actual credit spreads are published. In conversations with the author, cap market participants place t h e credit spreads that have occurred in the range of 5 to 10 basis points for two- to three-year caps. The estim a t e d spreads using t h e cap model are roughly in that range. Further research into actual credit s p r e a d s a n d r e f i n e m e n t s of t h e c a p m o d e l should sharpen t h e estimation results a n d m a k e the m o d e l more useful. Conclusion Interest-rate caps, collars, a n d floors are a m o n g the newest interest-rate risk managem e n t instruments. This article has given an exposition of these closely related instruments, which are options-based and designed to limit exposure t o fluctuations in short-term interest rates on floating-rate assets or debt. Their applications are not limited to hedging. Like options, they are also convenient for speculating on interest-rate movements. In practice, however, the distinction between these two applications is rarely clear-cut. Several e x a m p l e s served to illustrate how financial managers use caps, collars, a n d floors. The article also discussed the credit risks associated with caps, collars, and floors, which for the most part are over-the-counter contracts offered by o n e firm to another. Default risk is inherent in this kind of arrangement and can b e priced. A new cap valuation model produced credit spreads that are not much different from those observed in the cap market between stronger a n d weaker credit risks a m o n g c a p writers. Interest-rate risk m a n a g e m e n t has b e e n growing in importance for financial managers. This article may improve their understanding of the credit risk of caps, collars, and floors and h e l p d e t e r m i n e the cost of interest-rate protection. II FEDERAL RESERVE BANK OF ATLANTA Notes ' R e c e n t Economic Review Koch (1988), a n d Feinstein (1989). 2 Uses Incentives t o Push Rate Collars," American articles i n c l u d e A b k e n (1987), Feinstein a n d G o e t z m a n n (1988), Kawaller, Koch, a n d ' 8 T h e s e e x a m p l e s are consistent with the recent findings of Boyle a n d Turnbull (1989) in their examination of collars. S e e Wall a n d Pringle (1988) for an introduction t o interest- Using a different option-pricing m o d e l t h a n t h e Black rate swaps. 3 m o d e l , they f o u n d that a 100 percent increase in t h e For brevity, the market for caps, collars, a n d floors will b e referred t o as t h e c a p 4 volatility causes t h e floor level t o c h a n g e by less than o n e market basis p o i n t . If their f i n d i n g s are also valid for the Black S e e Kuprianov (1986): 16-20, for a discussion of Eurodollar m o d e l , m o s t of t h e difference observed in the e x a m p l e s d e p o s i t s a n d Eurodollar futures. 5 in t h e text is a t t r i b u t a b l e t o t h e d i f f e r e n c e in yield T h e i n f o r m a t i o n o n t h e 1SDA survey was r e p o r t e d in Risk 2 (April 1989): 11. 6 A d e t a i l e d discussion of o p t i o n pricing is b e y o n d t h e curves. 19 Before March 1989, contract expiration d a t e s h a d a maxim u m maturity of o n e year. S e e Chicago Mercantile Ex- s c o p e of this article. A basic overview can b e f o u n d in c h a n g e (February/March 1989): 7. A b k e n (1987). S e e Cox a n d R u b i n s t e i n (1985) or larrowand R u d d (1983) for m o r e thorough introductions t o o p t i o n 20 T h e term counterparty S e e A b k e n (1987): 6, for m o r e detail, 2 ' A n o t h e r c o m p l i c a t i o n in using futures in a replicating portfolio is that futures contracts are marked to market ^ e e H e n d e r s o n (1986) for further discussion. 9 10 daily. This situation may create cash flow p r o b l e m s since S e e Black (1976). futures positions that lose value may b e subject to fre- Tb the author's knowledge, n o p u b l i s h e d s t u d i e s have q u e n t margin calls. Even t h o u g h the replicating portfolio c o m p a r e d t h e accuracy of different option-pricing m o d e l s is u s e d to h e d g e a cap, which matches it in value, the cash for pricing caps a n d related instruments. O n e reason may flows from t h e c a p c o m e only w h e n it is sold a n d o n b e that there are n o publicly available d a t a o n t h e s e rates, a n d a n o t h e r is that t h e s e instruments are relatively new. interest p a y m e n t dates. 22 S e e A b k e n (1987) for m o r e on t h e synthetic creation of Little empirical research exists o n t h e a d e q u a c y of dif- options. Mattu (1986) gives e x a m p l e s of replicating port- ferent interest-rate o p t i o n - p r i c i n g m o d e l s . Boyle a n d Turnbull (1989) u s e t h e C o u r t a d o n option-pricing m o d e l folios for c a p s a n d floors. 23 in evaluating collar rates, b u t they d o n o t c o m p a r e their Shirreff (1986) gives an i n t e r e s t i n g t h o u g h in it. ket rates. ' B e c a u s e t h e C M E a n d m o s t LIFFE Eurodollar futures are "cash-settled," a $1 million d e p o s i t is rarely m a d e , b u t instead only t h e difference b e t w e e n t h e current, or spot, 24 L e G r a n d a n d Fertakis (1986): 134. 25 Floating-rate C D s are also called variable-rate CDs. 26 S e e Intermarket (October 1986): 14, for an account of t h e first such sale of a c a p from a c a p p e d floating-rate n o t e LIBOR a n d the contracted LIBOR t i m e s t h e notional prinl2 somewhat d a t e d overview of t h e c a p s market a n d t h e various players rates with t h o s e from other m o d e l s nor with actual mar1 is standard terminology for t h e other party in a swap, c a p , floor, or collar agreement. pricing. 7 Banker, August 2, 1989. cipal actually changes hands. (FRN). By selling a c a p off an issue of $100 million in 12- Prior to )une 1989 contract m o n t h s extended three years. year c a p p e d FRNs, B a n q u e I n d o s u e z of Paris lowered its I3 interest rate by o n e - e i g h t h of a p o i n t b e l o w LIBOR. A Eurodollar futures price is actually a n index value that e q u a l s 100 m i n u s the "add-on" yield (three-month LIBOR). U n c a p p e d , t h e n o t e s w o u l d h a v e s o l d at LIBOR. T h e Thus, t h e futures price a n d add-on yield m o v e inversely c a p p e d FRNs were issued at LIBOR p l u s three-eighths. with each other. S e e Kuprianov (1986): 16, for m o r e d e t a i l O n an a n n u a l basis, I n d o s u e z therefore collected the o n Eurodollar futures a n d short-term interest-rate futures generally. Both the add-on yield a n d t h e futures price are usually q u o t e d in t h e financial press. l4 e q u i v a l e n t of 50 basis p o i n t s o n t h e sale of its cap. 27 Shirreff (1986): 29. 28 T h e volatilities shown i n C h a r t I are p r o b a b l y not t h e s a m e S e e Feinstein (1989) for d e t a i l s o n t h e estimation, inter- as t h o s e u s e d to generate t h e c a p rates. The volatilities pretation, a n d uses of i m p l i e d volatilities. T h e Eurodollar were o b t a i n e d from a different source t h a n t h e c a p rates, futures o p t i o n s are actually American options, b u t t h e b u t t h e y s h o u l d b e highly c o r r e l a t e d with t h e actual early exercise feature has negligible value for t h e slightly volatilities u s e d to price t h e caps. out-of-the-money o p t i o n s usually u s e d in estimating the 29 N e w b e r y a n d Stiglitz (1981) give a c o m p r e h e n s i v e discus- 30 Wall a n d Pringle (1988): 22. sion of risk aversion a n d t h e rationale for hedging. i m p l i e d volatilities with a E u r o p e a n futures o p t i o n formula. l5 S u m s in Table 1 m a y n o t a d d u p d u e t o r o u n d i n g error. C a p rates a r e usually r o u n d e d t o w h o l e b a s i s p o i n t s . T h e 16 3 ' T h e e x a m p l e given was d e s c r i b e d in terms of a "flow conc e p t " of interest-rate risk, that is, t h e i m p a c t of a c h a n g e in dollar a m o u n t s are t h e exact a m o u n t s c o m p u t e d in con- interest rates o n the net interest margin. Another way to structing Table 1. view interest-rate risk is in t e r m s of a "stock c o n c e p t , " t h e A n o t h e r way to create a zero-cost collar is t o set the floor change in the net worth of t h e firm. A parallel shift in t h e first a n d then d e t e r m i n e t h e a p p r o p r i a t e cap. T h e m e t h o d term structure of interest rates w o u l d r e d u c e t h e value of discussed in t h e text is m o r e c o m m o n . an S&L's long-term mortgages m o r e than it would r e d u c e ' 7 Collars have also b e e n offered that give t h e buyer a pay- t h e value of its short-term liabilities. Net worth w o u l d b e m e n t for taking t h e collar, that is, the value of t h e floor sold r e d u c e d or possibly turn negative. Purchasing a c a p - a n exceeds t h e cost of t h e c a p purchased. S e e "NatWest asset o n the b a l a n c e s h e e t - w o u l d offset loss of net worth 22 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 t o s o m e extent b e c a u s e it would gain value as interest plexities involved in valuing securities that are c o m p o s e d rates rise. S e e Spahr, Luytjes, a n d Edwards (1988) for a g o o d exposition of this a p p l i c a t i o n of c a p s a n d how they of s e q u e n c e s of options. 38 h e d g e interest-rate risk. 32 average of all p o s s i b l e payoffs, each payoff m u l t i p l i e d by the probability of its occurring. C o m m e r c i a l b a n k s u n d e r w r i t i n g d e b t for highly levera g e d financings often require their floating-rate borrow- 39 According to Moody's study, t h e lowest investment-grade ers to buy caps for a portion of t h e d e b t . This h e d g i n g b o n d s , rated Baa (or BBB by Standard a n d Poor's), h a d r e q u i r e m e n t may b e s t i p u l a t e d in t h e loan covenant. S e e average a n n u a l d e f a u l t rates over two-year h o r i z o n s of Richardson (1989): 12. 0.25 percent. A Standard a n d Poor's BBB-rated investment b a n k was reportedly at a disadvantage in writing caps 33 S e e Moody's 34 This m e t h o d a s s u m e s that t h e interest rate follows a ran- Special (1989). Report c o m p a r e d t o stronger writers. S e e Shirreff (1986): 34. d o m walk with n o " d r i f t " (that is, d e t e r m i n i s t i c t r e n d The 0.13 default rate used in the e x a m p l e was chosen t o movements). C h a n g e s in t h e interest rate from p e r i o d to reflect t h e lower risk of default o n a c a p relative t o a period are a s s u m e d t o b e normally d i s t r i b u t e d with constant variance (or standard deviation), implying that the statistical distribution of interest-rate m o v e m e n t s may b e 35 l n a discrete t i m e m o d e l the expected value is a weighted bond. 40 T h e M o n t e Carlo s i m u l a t i o n s u s e d a variance reduction t e c h n i q u e called t h e m e t h o d of antithetic variates (see c o m p l e t e l y characterized by only its m e a n a n d variance. Boyle 11977|). The total n u m b e r of realizations was in fact These percentages are b a s e d o n t h e properties of t h e nor- 200,000 for each simulation, though only a fourth of that mal distribution, which is a s s u m e d to describe interest- n u m b e r c a m e from i n d e p e n d e n t draws from t h e r a n d o m rate m o v e m e n t s . n u m b e r generator. S e e A b k e n (forthcoming) for m o r e 36 Arak, G o o d m a n , a n d Rones (1986): 452. details. 37 C a p valuation can b e formulated as a kind of c o m p o u n d o p t i o n p r o b l e m . S e e G e s k e (1977) t o a p p r e c i a t e t h e com- References Abken, Peter A. "An Introduction t o Portfolio Insurance." Feinstein, Steven P. "Forecasting Stock-Market Volatility 72 Using O p t i o n s o n Index Futures." Federal Reserve Bank "Valuing Default-Risky Interest Rate C a p s : A Feinstein, Steven P., a n d William N. G o e t z m a n n . "The Effect M o n t e Carlo Approach." Federal Reserve Bank of Atlanta of t h e 'Triple Witching Hour' on Stock Market Volatility." Working Paper (forthcoming). Federal Reserve Bank of Atlanta Economic Federal Reserve Bank of Atlanta Economic Review of Atlanta Economic ( N o v e m b e r / D e c e m b e r 1987): 2-25. Arak, Marcelle, Laurie S. G o o d m a n , a n d Arthur Rones. "Credit Lines for New Instruments: Swaps, Over-theCounter O p t i o n s , Forwards a n d ments." In Proceedings Floor-Ceiling Agree- of a Conference on Bank Structure Federal Reserve Bank of Chicago (May and Competition. 1986): 437-56. nal of Financial Economics 3 (January/March 1976): 167- 79. 74 (May/June 1989) : 12-30. Review Geske, Robert. "The Valuation of Corporate Liabilities as C o m p o u n d O p t i o n s . " Journal of Financial and nal of Financial Economics 4 (May 1977): 323-38. H e n d e r s o n , Schuyler K. "Securitizing Swaps." Law Review International ( S e p t e m b e r 1986): 31-34. Jarrow, Robert A., a n d Andrew R u d d . Option Pricing. Home- wood, 111.: Richard D. Irwin, Inc., 1983. ing C a p p e d O p t i o n s . " Journal of Futures Markets Default Risk." Journal of Finance 42 (June 1987): 267-80. Kawaller, Ira G „ Paul D. Koch, a n d Timothy W. Koch. "The Boyle, Phelim P., a n d Stuart M. Turnbull. " P r i c i n g a n d Hedg9 (Feb- ruary 1989): 41-54. Relationship b e t w e e n t h e S&P 500 Index a n d S&P 500 Index Futures Prices." Federal Reserve Bank of Atlanta Economic Brown, Keith C., a n d D o n a l d J. Smith. " R e c e n t Innovations in Review 73 (May/June 1988): 2-10. K u p r i a n o v , A n a t o l i . "Short-Term Interest R a t e Futures." Interest Rate Risk M a n a g e m e n t a n d t h e Reintermedia- Federal Reserve Bank of R i c h m o n d Economic tion of Commercial Banking." Financial ( S e p t e m b e r / O c t o b e r 1986): 12-26. Management 17 (Winter 1988): 45-58. " C a p s a n d Floors." The Banker Review LeGrand, Jean E„ a n d John P. Fertakis. "Interest Rate C a p s : K e e p i n g the Lid o n Future Rate Hikes." Journal (February 1989): 9. Chicago Mercantile Exchange. Market Perspectives. Vari- Rate Risk." Inter- ous issues. market Quantita- 12 (1977): 541-52. tive Analysis Johnson, Herb, a n d R e n é Stulz. "The Pricing of O p t i o n s with Boyle, Phelim P. " O p t i o n s : A M o n t e Carlo Approach." Jour- Commins, 73 ( S e p t e m b e r / O c t o b e r 1988): 2-18. Financial Black, Fischer. "The Pricing of C o m m o d i t y Contracts." Jour- Review Kevin. " M a n a g i n g Interest of Ac- (May 1986) : 130-36. Mattu, Ravi. " H e d g i n g Floating Rate Liabilities: Locks, C a p s Cox, |ohn C., a n d Mark Rubinstein. Options and Floors." C h i c a g o M e r c a n t i l e Exchange Strategy Paper, 1986. (May 1987): 28-34. w o o d Cliffs, N.J.: Prentice-Hall, 1985. countancy Markets. Engle- Moody's Special Report. "Historical Default Rates of Cor- porate B o n d Issuers, 1970-1988." July 1989. II FEDERAL RESERVE BANK OF ATLANTA of Spahr, Ronald W„ )an E. Luytjes, a n d D o n a l d G. Edwards. Economics "The I m p a c t of t h e Uses of C a p s as D e p o s i t H e d g e s for Newbery, David M.G., a n d )oseph E. Stiglitz. The Theory Commodity Price Stabilization: A Study in the Financial Institutions." Issues of Risk. New York: Oxford University Press, 1981. Richardson, Portia. "Put o n Your Thinking C a p . " Intermarket Factor M o d e l of t h e Term Structure: An A p p r o x i m a t e of Financial and Quantitative Analysis Firms' H e d g i n g Policies." lournal Analysis Cashflow ()une 1988): 47-50. 2 (March 1989): 21-23, 41. of Financial and 20 ( D e c e m b e r 1985): 391-405. Swaps: A Theoretical a n d Empirical Analysis." Management (March 1986): 26-40. Smith, Clifford W „ a n d R e n é M. Stulz. "The D e t e r m i n a n t s of titative LBO Work." Corporate T o m p k i n s , R o b e r t . " T h e A-Z of C a p s . " Risk Wall, Larry D. "Alternative Explanations of Interest Rate 19 ( D e c e m b e r 1984): 413-24. Shirreff, David. " C a p s a n d O p t i o n s : T h e D a n g e r o u s New Protection Racket." Euromoney (Sum- Sutherland, L. Frederick. " S q u e e z i n g Cash: How t o Make an (March 1989): 10-13. Schaefer, S t e p h e n M„ a n d E d u a r d o S. Schwartz. "A TwoSolution." lournal in Bank Regulation m e r 1988): 17-23. Quan- Financial (forthcoming, 1989). Wall, Larry D„ a n d )ohn ). Pringle. "Interest Rate Swaps: A Review of t h e Issues." Federal Reserve Bank of Atlanta Economic Review 73 ( N o v e m b e r / D e c e m b e r 1988): 22- 37. 24 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Federal Reserve Bank of Atlanta 1989 Annual Report The Atlanta Fed's 1989 Annual Report will be available in February 1990. In addition to reviewing the Bank's activities during 1989, the report will include selections from its 75th anniversary publication, A History of the Federal Reserve Bank of Atlanta, 1914-1989. It will also contain a statement of condition, a statement of earnings and expenses, and a statistical summary of operations, along with a list of directors and officers who served during the year. For copies of the publication, please fill out the coupon below and mail to: Public Information Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, N.W., Atlanta, Georgia 30303-2713, or call 404/521-8788. Please s e n d m e c o p y (copies) of t h e Atlanta Fed's 1989 Annual Report. Name Address City State ZIP. Financial Asset Pricing Theory: A Review of Recent Developments Ellis W. T a l l m a n In this article, an Atlanta Fed economist reviews research on financial asset pricing with a special focus on the links between asset pricing and the real economy. Surveying the capital asset pricing model (CAPM), the consumption-based CAPM, and the more recent arbitrage pricing theory, he concludes that ongoing theoretical and empirical developments point toward future research that can link real economic factors to asset pricing behavior. Financial asset pricing theories have devel- firm's stock. C o m p e t i t i v e market forces t h e n b i d o p e d primarily over t h e p a s t 30 years. Scholars t h e asset price u p or d o w n to its n e w equilib- have m a d e great strides d u r i n g t h e latter half of rium price. t h e p e r i o d in t h e analysis of newer, m o r e dy- This framework provides an intuitive link be- n a m i c m o d e l s . This article surveys recent de- t w e e n t h e asset markets a n d m e a s u r e s of real v e l o p m e n t s in t h e o r e t i c a l research a n d t h e e c o n o m i c behavior. Such a m o d e l suggests that state of relevant empirical evidence. It con- t h e aggregate stock m a r k e t v a l u e reflects expec- c l u d e s that financial asset pricing research re- tations of t h e p r e s e n t d i s c o u n t e d v a l u e of cash m a i n s o p e n for a d d i t i o n a l study, yet t h e current flows from t h e future performance of t h e econ- b o d y of k n o w l e d g e presents a coherent frame- omy. D e s p i t e t h e intuitive a p p e a l of a correla- work for analyzing asset pricing in a rational tion b e t w e e n p e r f o r m a n c e of t h e stock market e c o n o m i c setting. a n d t h e real e c o n o m y , insufficient e m p i r i c a l M o d e r n c a p i t a l m a r k e t theory s t u d i e s t h e e v i d e n c e exists to s u p p o r t this relationship. d e t e r m i n a t i o n of asset prices. In a basic m o d e l For e x a m p l e , from O c t o b e r 2 to O c t o b e r 23, of stock valuation, asset prices reflect t h e pres- 1987, nearly 30 percent of perceived asset v a l u e e n t d i s c o u n t e d v a l u e of t h e projected future in t h e stock market, as m e a s u r e d by t h e Stan- d i v i d e n d p a y m e n t s to t h e stockholder. W h e n d a r d a n d Poor's 500, was lost. In contrast, t h e n e w information a b o u t a firm's prospects be- real e c o n o m y g r e w a n d c o n t i n u e d t o grow c o m e s public, expectations a b o u t future cash t h r o u g h o u t 1987 a n d 1988 a n d into 1989. In long- flows or t h e risk-adjusted d i s c o u n t rate of a er perspective, o n t h e other h a n d , t h e overall given stock change. In m o s t m o d e l s of asset growth rate of real gross national p r o d u c t (GNP) pricing, investors, lenders, a n d other e c o n o m i c averaged 4.2 p e r c e n t from 1983 through 1988, " a g e n t s " are a s s u m e d to b e rational, using n e w a n d t h e stock market b o o m has c o i n c i d e d , al- information to a d j u s t their valuation of t h e given t h o u g h not without volatility. The s o m e t i m e s a n o m a l o u s behavior of t h e stock market vis-avis real G N P growth indicates that t h e forces t h a t drive asset prices are still largely a mystery, especially with regard to t h e relationship be- The author Atlanta is an economist Fed's Research Department. and Frank King for valuable 26 in the macropolicy section He thanks Curt comments. of the Hunter t w e e n real e c o n o m i c p e r f o r m a n c e a n d asset markets. ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 To respond to a n e e d for an u p d a t e on t h e existing research on asset pricing, this article offers a general introduction to and historical survey of m o d e r n financial asset pricing theories; other surveys of asset pricing m o d e l s have focused only on a specific subset of them. This p a p e r concentrates on the recent developments in asset pricing m o d e l s under uncertainty in t h e 17 years since an early survey of the groundbreaking research by Michael C. Jensen (1972). Its primary purpose is to show that such theories have p r o v i d e d useful advances in achieving a better u n d e r s t a n d i n g of e q u i t y market pricing. 1 Presented first is a brief summary of t h e traditional capital asset pricing m o d e l (CAPM) a n d the groundbreaking work that d e v e l o p e d from it. The survey then traces t h e evolution of t h e initial work on the CAPM to asset pricing in d y n a m i c m o d e l s , s o m e of which e n c o m p a s s both the disciplines of finance and macroeconomics. While not exhaustive, t h e article conc l u d e s with suggestions for further work on asset pricing that may b e especially useful to those involved in economic policy-making or in the application of asset valuation research. Policymakers have a particular interest in better understanding linkages between real economic performance and asset pricing behavior. Financial data such as stock market prices are FEDERAL RESERVE BANK OF ATLANTA observed much more often than are most real economic data like industrial production and gross national product, which are available only monthly or quarterly. Such financial statistics may provide preliminary insights into the condition of real economic prospects, thus enhancing t h e policymaker's information set. Also, specific information on risk sources in the real economy can help policymakers promote a stable environment for financial markets and foster more efficient allocation of capital in the economy. Investment managers may also find the isolation of macroeconomic sources of asset risk a useful m e t h o d of assessing portfolio risk and a criterion for portfolio formulation. Valuation Theory, Mean Variance Efficiency, and the Traditional CAPM As long as stock markets have existed, prognosticators have tried to predict future movements in equity values. Forecasts often lacked a strong theoretical a n d analytical framework, though. Early attempts t o examine their effectiveness, notably by Alfred Cowles III (1933), suggested that they offered no perceptible advantages to investors. More formal and scientific analysis of asset price behavior began later. 27 Harry Markowitz (1952) provided t h e source for m o d e r n portfolio theory. His research is often cited as the seminal work of m o d e r n finance from which evolved the early impetus t o describe the equilibrium relationship between assets a n d risk. The resulting asset pricing models, particularly t h e traditional capital asset pricing m o d e l , rely on the return to a "market'' portfolio, consisting of a weighted average of all assets held, as t h e benchmark from which o n e can assess asset prices relative t o the market. The CAPM provides a useful simplification a n d focus for asset pricing theory because it produces an interpretable risk measure for a riskreturn relationship and parsimoniously summarizes a great deal of information in a single variable. A weakness of the early equilibrium asset pricing m o d e l s , however, is their lack of any formal linkage b e t w e e n real e c o n o m i c performance a n d t h e behavior of asset prices. Recessions—or, m o r e generally, changes in future prospects for the economy—affect firm valuation by altering expected cash flows and t h e relevant discount rate, b u t t h e s i m p l e CAPM offers no direct m e t h o d to incorporate fluctuations in economic conditions over t i m e into the asset pricing process. The model ignores this issue. Recent theoretical advances in dynamic m o d e l s of asset pricing as well as an accumulation of e v i d e n c e in conflict with t h e s i m p l e CAPM have m a d e that basic formulation less central as an equilibrium model of asset pricing. 2 Yet, the simple CAPM provides a major part of the underlying intuition for these models. Modern financial theory has its underpinnings in the application of scientific m e t h o d s t o basic finance questions. Hundreds of different assets trade on t h e stock exchange, and each asset has characteristics, such as firm size, location, industry, and age, that distinguish it from other assets. The simultaneous analysis of so many characteristics is not feasible in a scientific realm. To focus attention on t h e most imp o r t a n t traits, financial m o d e l s simplify t h e p r o b l e m by limiting t h e n u m b e r of variables. O n e of the earliest m o d e l s of asset pricing, briefly m e n t i o n e d above, analyzes the valuation of a single stock as a function of t h e flow of future d i v i d e n d s discounted by the relevant riskrelated discount rate. The m o d e l , presented by John B. Williams (1931), is outlined below: 28 where P, 0 is t h e price of asset / at period 0, di t is the dividend per share of c o m m o n stock of firm /' from the e n d of m o n t h t - 1 to the e n d of m o n t h t, and kj is t h e risk-related rate of discount for firm /'. Eugene F. Fama and Merton H. Miller (1972) s h o w that—given a n u m b e r of s i m p l i f y i n g assumptions—firm valuations derived from disc o u n t e d firm cash flows, t h e stream of divid e n d s , or t h e firm's earnings p r o d u c e Williams's result. Williams's m o d e l , in a d e t e r m i n i s t i c world without uncertainty, thus offers a framework to g u i d e analysis. This valuation model, altered for uncertainty, suggests that asset prices vary u p o n the release of new information regarding a firm's prospects. Before t h e a n n o u n c e m e n t of t h e return on a stock, a large degree of uncertainty exists regarding t h e actual outcome, that is, t h e return ex post. 3 In modern financial theory, stock returns are viewed as random variables, and a probability distribution is associated with them. For most applications, stock return distribution has b e e n assumed t o b e approximated by the normal distribution, which is fully s u m m a r i z e d by two parameters: t h e mean, which is the measure of central tendency, a n d t h e variance, t h e measure of dispersion around t h e central tendency. 4 Markowitz p r e s e n t e d a m o d e l of investor portfolio selection under uncertainty in which investors choose asset portfolios on t h e basis of asset return a n d variance in a single period. Portfolio optimization involves t h e trade-off between reward (expected return) a n d greater risk. Investors prefer assets with higher m e a n s of expected returns b u t lower return variances, or less return variability. Thus, investors, assumed to b e risk-averse, want to balance risk and return in their portfolio choice. A further insight of portfolio theory is that t h e addition of a security a d d s to a portfolio's risk mainly by t h e contribution of its variabil ity to the variability of return from t h e entire portfolio—its "covariance" with the return stream of other portfolio assets. In the limit, as o n e increases t h e n u m b e r of individual assets in a portfolio, this covariance risk is t h e d o m i n a n t c o m p o n e n t of financial asset portfolios' variance. ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 This insight regarding covariation suggests that a collection of assets can offer a lower level of return variability than individual assets held separately. Consider two assets whose returns are perfectly negatively correlated. (The percentage increases in returns to o n e asset occur simultaneously with equal b u t o p p o s i t e movements in another asset's return.) A portfolio of these two assets would carry a fixed return a n d no risk, or variability of return. Although such neat covariation properties of assets d o not often occur, t h e portfolio helps illustrate the advantages of diversification. A primary outgrowth from the Markowitz work was the general equilibrium m o d e l s of asset pricing credited most often to William F.Sharpe (1964), John Lintner (1965), a n d Jan Mossin (1966).5 These m o d e l s assume t h e existence of an asset that is both free of default risk a n d offers a fixed one-period return. 6 In equilibrium, all assets are held, a n d the market portfolio—as d e f i n e d earlier, t h e portfolio that represents t h e return on every asset weighted by its proportion in the total value of all assets c o m b i n e d — c o m p r i s e s entirely risky assets. James Tobin (1958) shows that in Markowitz's environment, all individuals hold assets in only two types of portfolios: the riskless asset and the market portfolio. The traditional capital asset pricing m o d e l , with fixed covariance between returns on an asset and returns on the market portfol io, shows that the risk premium of an asset (that is, the difference between the return on a risky asset and the return on a risk-free asset) is determined by movements in the market portfolio's expected premium. For the purpose of this article, the following equations present t h e main implications of the m o d e l : and C o v (/?,, P / ~ o ^ ) RM) ' where E, is t h e expected return to asset i, RF is the rate of interest on the riskless asset, EM is the expected return on t h e market portfolio, (3,is the degree to which asset i's return varies with the market's return (discussed below), o 2 (RM) is the variance of t h e return to market portfolio, and Cov (/?, , RM) is t h e covariation between t h e return to asset / and the return on the market. FEDERAL RESERVE BANK OF ATLANTA The main argument of the model relies on the intuition that investors are rational a n d will undertake risk only to t h e extent that they are compensated. If, simply through diversification, risk can b e removed from a portfolio, no o n e should b e compensated for holding risk that can b e avoided—"diversifiable" risk. While diversification can b e achieved by holding assets that should have low or negative covariation, Markowitz's portfolio theory result suggests that holding a large n u m b e r of assets also results in diversification. Diversifiable risk should not b e related to a risk premium. If a firm experiences a period of poor m a n a g e m e n t or suffers a labor strike, the asset returns may b e negative. But these sources of risk are company- or firm-specific, and an investor can reduce risk by investing a proportion of wealth in other firms. In contrast, certain factors like wars or the oil price shocks of the 1970s affect t h e entire economy and, as a result, future returns to the market portfolio. Such nondiversifiable risk—that is, risk related t o covariation of an asset's return with the return to the market portfolio—will therefore b e related to a risk premium since an investor will require an incentive to hold a risky group of assets. For an individual asset /', the expected return equals the riskless rate of interest plus the product of the market risk premium and the relevant risk measure, (3,-, commonly referred to as the risk of covariation with the market, or " b e t a " risk. An implication of the model is that assets are priced relative to their sensitivity to the market portfolio returns. A portfolio in which 3 equals o n e results in t h e s a m e expected return as t h e market portfolio. Portfolios in which (3 is less than o n e are referred to as defensive since they should fluctuate relatively less than the market b u t will also have a lower expected return. In contrast, portfolios with (3 greater than o n e are d e e m e d aggressive, in that their expected returns are greater than t h e market portfol io's, b u t they incur relatively greater return volatility. Thus, according to the CAPM only nondiversifiable, or systematic, risk is relevant for asset pricing: consequently, the expected return on any asset is a linear function of the asset's (3. O n e may interpret (3 as a sensitivity measure, gauging t h e reaction of the return on asset / to a m o v e m e n t in the market return. Any returns that are significantly greater or less than predicted II by an asset's beta-risk measure are called abnormal returns. Testing t h e C a p i t a l Asset P r i c i n g M o d e l . Despite CAPM's simplicity and the measurability of its variables through market price data, empirical evidence has not supported this approach. Two important empirical tests of the CAPM—Fischer Black, Jensen, and Myron S. Scholes (1972) and Fama and James D. MacBeth (1973)—find evidence that conflicts with t h e predictions of the simple Sharpe-Lintner pricing m o d e l introduced earlier. Black, Jensen, and Scholes find evidence that low beta risk stocks or portfolios have positive abnormal returns, and high beta risk stocks or portfolios have negative abnormal returns. 7 Fama and MacBeth included other measures as additional explanatory factors in an asset pricing regression to examine the CAPM's sensitivity to variables that theory suggests should b e unimportant. The a d d e d measures—beta s q u a r e d a n d t h e average of t h e residual variance—might also indicate possible nonlinearities in t h e risk-return relationship. Although over the entire sample period their two additional variables show no significant systematic relationship to priced risk, in certain time periods these measures were associated with statistically significant risk premia-that is, the return which asset holders must b e paid in order to induce them to accept an asset with nondiversifiable risk. Fama and MacBeth concluded that these variables serve as proxies for relevant underlying risk measures. 8 However, their results suggest that a positive trade-off generally exists between risk, as measured by beta, and return. In a recent paper that reexamines and extends the Fama and MacBeth estimates, Seha M. Tinic and Richard R. West (1986) showed both significant departures from t h e linear riskreturn trade-off predicted by the traditional CAPM and significant nonlinearities not captured by the model. As a result, the researchers conclude that results of existing empirical research on the traditional CAPM is suspect. Much of the empirical literature that uses the CAPM applies beta as a risk measure in order to adjust asset returns for their degree of riskiness, prior to the examination of the impact of an event. 9 For example, beta has been used to adjust returns for risk in numerous event studies 30 dealing with finance issues like judging mutual fund performance or the price effects of stock splits and public tender offer announcements. After rigorous investigation, researchers have uncovered a number of anomalies that underm i n e the capital asset pricing model. Two of the more well-known inconsistencies, known initially as the January effect and the small firm effect, have been particularly damaging. Researchers have f o u n d that t h e returns to small firms generally outperform those of larger firms after a d j u s t m e n t by beta risk measures. In other research, stocks have shown an abnormally high excess return after risk adjustment in the month of January. A study by Donald B. Keim (1983) has shown that the small firm effect and the January effect are related; that is, smaller firms outperform larger firms in January. These results con- "Despite CAPM's simplicity and the measurability of its variables through market price data, empirical evidence has not supported this approach." tradict implications of t h e CAPM. They have fueled criticisms of its framework not only as an equilibrium model of asset pricing but also as a useful framework for risk adjustment in other applications. Recent work by Jay R. Ritter and Navin Chopra (1989), however, suggests that the capital asset pricing model's risk-return relationship is more robust when portfolios weight the individual assets by their proportion of total market value in contrast to the standard practice of weighting assets in a portfolio equally. In those circumstances, small firm effects are deemphasized and the relationship shows no January seasonal effects. If nothing else, then, the existence of anomalies has stimulated additional research and evidence on the traditional CAPM. Aside from these empirical shortcomings, the CAPM is essentially not dynamic. Though the m o d e l involves n u m e r o u s simplifying condiECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 tions, for this article t h e most relevant assumption is that all investors maximize t h e utility of terminal wealth. In other words, notwithstanding inevitable uncertainty, future investment and consumption opportunities are completely captured by t h e certain m e a n and variance of the probability distribution of the asset returns. The CAPM is hence a one-period model that cannot encompass issues like economic fluctuations and their effects on asset pricing. Fama (1970) has shown that the CAPM's assumption about maximizing terminal wealth can b e ext e n d e d to m u l t i p l e periods as long as future consumption and investment decisions are determined outside t h e m o d e l . However, the implication of Fama's work is that investors have perfect foresight with regard t o future market conditions; intertemporal factors like changes "Recent advances in the sophistication of financial asset pricing theory have moved to models that address intertemporal variations in opportunities over the business cycle." in economic performance are assumed t o have already b e e n anticipated a n d thus should not affect the investment decision. In this sense, the model is static, although empirical tests estimate the m o d e l as if its restrictions held over time. Consequently, aside from predicting t h e return to t h e market portfolio (a difficult task at best), t h e CAPM as interpreted by Fama d o e s not p r e s e n t a m e t h o d t o e s t i m a t e any dynamics. Another controversy in CAPM-based asset pricing research is whether a truly riskless rate of return actually exists. Black (1972), in fact, attacks this question by offering a CAPM without the riskless rate. Generally, though, in empirical tests, the riskless rate of return, RF, is proxied by the return on a Treasury bill with o n e m o n t h to maturity. The return to the market portfolio, RM, is approximated by s o m e equity index, usually the Standard and Poor's 500. Richard Roll (1977) criticizes empirical examinations of the traditional CAPM in what is now referred to as t h e "Roll critique." His primary concern is measurement of the market portfolio in CAPM tests. Roll argues that an u n a m b i g u o u s test of t h e model cannot b e performed with t h e typical proxies for market rate of return because t h e true market portfolio has to include all individual assets. The argument suggests that inferences a b o u t t h e model may b e sensitive to the composition of the market proxy, and any demonstrated sensitivity to various reasonable proxies for t h e market will reduce the testability of the m o d e l . Robert F. Stambaugh (1982), res p o n d i n g to this criticism, has shown that, although tests are sensitive to the selection of assets, inferences about the CAPM are insensitive to t h e use of several different proxies for the market portfolio, suggesting that the CAPM may b e less sensitive to the Roll critique than t h e argument implies. Still, Roll's analysis has contributed to deemphasizing the model in m o r e recent research. Intertemporal CAPM and the Consumption CAPM The ambiguous empirical support for the traditional CAPM as well as dissatisfaction with the restrictiveness of s o m e of its assumptions has led researchers toward m o d e l s that relax s o m e CAPM a s s u m p t i o n s . Recent advances in t h e sophistication of financial asset pricing theory have moved to m o d e l s that address intertemporal variations in opportunities over the business cycle. Empirical evidence suggests that the extension of t h e CAPM to account for intertemporal change is useful. For instance, Katherine Schipper and Rex Thompson (1981) demonstrated that equity assets may b e used to h e d g e against changes in consumption and investm e n t opportunities related to unanticipated shifts in consumption, GNP, a n d the price level, which serve as proxies for general conditions. 1 0 Robert C. Merton (1973) earlier d e v e l o p e d a dynamic asset pricing model, drawing from the initial insights of the mean-variance CAPM b u t extending the framework to incorporate intertemporal uncertainty. Merton's pricing equation describes a framework that holds in the II FEDERAL RESERVE BANK OF ATLANTA presence of a business cycle. The model allows investment and consumption opportunities to fluctuate over time so that the economy's condition is linked directly with asset price behavior. John B. Long, Jr., (1974) introduced an alternative dynamic model that specifies relevant variables, known as "state" variables because they indicate the state of the economy, useful for the pricing equation. These state variables represent external factors, such as the stock of physical capital, that determine current investment opportunities. 11 Long's model suggests that the term structure of interest rates-that is, interest rates on equally risky d e b t of successively distant maturities—is a key element in the pricing of equity assets, and recent empirical work has supported this intuition. 12 The general model in Merton (1973) also involves a vector of state variables, which represents the number (S) of relevant variables n e e d e d to describe the condition of the economy. The state vector indicates whether the economy is in a recessionary or expansionary stage of the business cycle and characterizes uncertainty in a model economy. The model solution, in the general case, implies that there will b e S + 2 number of portfolios in the equilibrium asset pricing equation. The additional two portfolios are the market portfolio and the riskless asset. 13 The resulting pricing relationship would expand the simple CAPM to include S additional betas (or sensitivity measures, o n e for each state variable), and the premia related to each state variable sensitivity. The general model requires identification of state variables that may b e unobservable, however. Thus, t h e m o d e l may not b e directly empirically testable with existing econometric methods. To give more interpretation to the model as well as to develop a potential route of inquiry, Merton assumes that the interest-rate movements of the riskless asset are the sole state variable sufficient to describe the investment opportunity set. The restricted model provides a tractable result in which the equilibrium asset pricing model involves three funds: the market portfolio, the riskless asset, and a portfolio of assets negatively correlated with movements in the riskless asset. This simplified model presents the intuition of intertemporal uncertainty more directly than 32 does the general specification. Investors are compensated for holding both market risk, just as in the static CAPM, and the risk of unfavorable movements in investment opportunities as conveyed by the riskless interest rate. In periods of poor economic opportunities, investors would 1 ike to have assets that offer large returns. In fact, an investor may h e d g e against aggregate intertemporal risk by holding a risky asset that has an expected return less than the riskless asset if the risky asset pays off a high return when the return to the riskless asset is low. Although the restricted model in Merton's work provides insights into the forms of risk presented by shifts in aggregate economic opportunities, the three-fund result remains a special case. Testing Merton's general model requires identifying a n d counting t h e state variables. "/An alternative! model allows investment and consumption opportunities to fluctuate over time so that the economy's condition is linked directly with asset price behavior." Unfortunately, theory makes no unambiguous predictions about their number or identity. 14 However, the model motivates investigation of additional variables as measures of intertemporal risk a n d offers a general structure for empirical analysis. The M a c r o e c o n o m i c Link. In an elegant and influential paper, Douglas T. Breeden (1979) provides a key link between macroeconomic growth models and financial models of asset pricing. This connection makes possible an analysis of asset price determination in a model economy that fluctuates over time. Breeden's construct, which is consistent with Merton's, shows that the growth rate of consumption is a sufficient statistic for the state of the economy; in other words, the S state variables need not b e identified for asset pricing. The resulting relationship is an equilibrium asset pricing model that uses the growth rate of (real per capita) conECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 s u m p t i o n as the benchmark return from which all other assets are priced. 1 5 Thus, covariation with consumption growth is t h e single relevant measure of risk. Breeden argues that aggregate consumption should b e a better proxy for the desired measure of consumption than the return to a market proxy is for the return t o the market portfolio. The model is commonly referred to as the consumption CAPM (CCAPM).16 The equilibrium relationship is: £/ ~Rf= Pc/ " *Fh where £(- is t h e expected return to asset /, RF is the riskless rate of return, Ec is the expected growth rate of real per capita consumption, a n d P c/ is t h e m e a s u r e of t h e covariance of an asset's return with t h e growth rate of consumption. "Its simplicity as well as its derivation from a dynamic model has made the consumption CAPM an attractive method of asset pricing." By using consumption as the benchmark, the model is implicitly concerned with fluctuations in both consumption and investment opportunities. The business-cycle behavior of consumption, therefore, directly affects the pricing of assets. The intuition b e h i n d this relationship is that the-marginal utility, or marginal contribution to valuation from an extra unit of consumption, is low in a t i m e p e r i o d that has high consumption. If consumption fluctuates, a consumer would prefer assets that will h e l p reallocate c o n s u m p t i o n across states t o t h o s e in which consumption is low. As a result, an asset return that covaries negatively with consumption growth should h e l p smooth consumption and b e associated with a negative risk premium. On the other hand, asset returns that covary positively with consumption are associated in this m o d e l with a positive risk premium. Hedging behavior on the part of the investor results from the incorporation of intertemporal uncertainty into the m o d e l . Persistent differences in average yields to a selection of assets can b e explained, therefore, by the insurance that particular assets provide against certain states. Its simplicity as well as its derivation from a d y n a m i c m o d e l has m a d e t h e c o n s u m p t i o n CAPM an attractive m e t h o d of asset pricing. As with t h e traditional CAPM, the equilibrium relationship of t h e consumption model requires estimating only o n e parameter to evaluate t h e risk characteristics of an asset or portfolio. Consumption data are also readily available on a m o n t h l y basis, so t h e m o d e l can b e t e s t e d relatively easily. It has b e e n tested often. A theoretical criticism by Bradford Cornell (1981) suggests that the consumption m o d e l is not free of the restrictions implied by Merton's intertemporal m o d e l ; direct estimation still requires t h e identification of state variables. As a result, t h e conditional distribution of cons u m p t i o n betas is random. Although Cornell notes that this situation may b e resolved with empirical evidence, t h e theory implies that distribution of the consumption betas relies u p o n the properties of the state variables. 17 Despite Cornell's criticism, s o m e empirical research has b e e n d o n e on the adequacy of t h e consumption CAPM. A recent study by Gregory N. M a n k i w a n d Matthew D. Shapiro (1986), using quarterly data from 1959 to 1982, shows that t h e traditional CAPM outperforms the consumption CAPM. Based on a large s a m p l e of equity returns, their test employs instrumental variables estimation methods. 1 8 Its results show that the expected real return has a significant linear relationship with the market beta b u t not with t h e consumption beta. Simon Wheatley (1988) criticizes the inferences m a d e from the Mankiw a n d Shapiro results because t h e instrumental variables estimations are widely different from those using t h e ordinary least squares regression technique, suggesting p r o b l e m s with t h e selected estimation strategy and weakening the resulting inferences. 1 9 Wheatley continues by estimating the cross-sectional adequacy of the consumption CAPM restrictions using 40 stock portfolios, Treasury bills, Treasury bonds, a n d corporate b o n d s as d e p e n d e n t variables. His tests found t h e CCAPM implications consistent with t h e data. 2 0 II FEDERAL RESERVE BANK OF ATLANTA To transcend Cornell's criticisms, Breeden, Michael R. Gibbons, a n d Robert Litzenberger (1989) used weaker empirical tests of the cons u m p t i o n m o d e l . Their p a p e r e x a m i n e d t h e CCAPM relative to t h e traditional CAPM using 12 stock portfolios, a Treasury bill asset, a Treasury b o n d portfolio, a corporate b o n d portfolio, a n d a junk b o n d p r e m i u m as t h e set of asset returns to b e explained. Their consumption data are adjusted for m e a s u r e m e n t p r o b l e m s associated with reported consumption statistics. The primary p r o b l e m with aggregate consumption data is that they are issued much less frequently than observations of stock returns. The CCAPM theory requires m e a s u r e m e n t of spot consumption growth rates, whereas actual consumption is measured over an interval. Given the data adjustments required, empirical evidence found in support of the m o d e l is necessary, b u t not sufficient, to accept it. Nonetheless, Breeden, Gibbons, a n d Litzenberger find that the explanatory power of t h e consumption growth rate for t h e behavior of asset returns over t i m e is significant. The results using t h e return to a portfolio of assets that has maximum correlation with consumption growth— the maximum correlation portfolio-are more c o m p a r a b l e to 1 inear regressions with t h e return t o a market proxy, since they b o t h use portfolio returns data as i n d e p e n d e n t variables. Tests of linearity between consumption beta a n d expected returns reject t h e hypothesis that consumption p a n d expected returns are linearly related for the full period covered by t h e data. However, examination of the subperiods suggests that the source of t h e rejection is t h e period 1929-39, a u n i q u e t i m e in t h e U.S. economy. In all other subperiods, the relationship c a n n o t b e rejected. Test results imply that neither t h e maximum correlation portfolio nor the market portfolio proxy has t h e lowest variance for a given mean return, that is, neither proxy is mean-variance efficient. Despite these rejections, t h e estimates for the risk premia related to consumption a n d to t h e market are quantitatively similar. Macroeconomic s t u d i e s that e m b o d y t h e consumption capital asset pricing model concentrate on t h e relationships between forecasta b l e movements in asset returns a n d in consumption. Lars P. Hansen and Kenneth J. Singleton (1982, 1983) imposed strong assumptions 34 to generate a closed-form solution that would test t h e predictability of asset returns a n d obtain estimates of t h e structural parameters of interest, namely t h e degree of risk aversion a n d t h e intertemporal discount factor.21 Unfortunately, their empirical results suggest rejection of t h e m o d e l , in part because of p r o b l e m s associated with measuring consumption data. However, further work by Sanford J. Grossman, Angelo Melino, a n d Robert J. Shiller (1987), which explicitly accounts for t h e t i m e averaging of c o n s u m p t i o n data, also failed t o s u p p o r t t h e model. 2 2 In sum, t h e empirical results for the consumption CAPM are mixed. The strong restrictions i m p o s e d by t h e macroeconomic m o d e l tests lead to rejection of the m o d e l and d o not prod u c e r e a s o n a b l e or useful e s t i m a t e s of t h e structural parameters. In contrast, recent evid e n c e on t h e c o n s u m p t i o n CAPM as a relative asset pricing construct are more hopeful, suggesting s o m e potential for its use in evaluating asset risk. However, more research is necessary t o d e t e r m i n e the robustness of the consumption beta measure as a risk gauge for assets; t h e initial tests are supportive of a linear relationship b e t w e e n consumption risk a n d asset premia b u t d o not s u p p o r t mean-variance efficiency, a prediction of the model. Thus, further research on the consumption CAPM must b e d o n e before it can b e widely a p p l i e d . Arbitrage Pricing Theory An alternative t o the traditional CAPM paradigm that has gained considerable attention is arbitrage pricing theory, d e v e l o p e d by S t e p h e n A. Ross (1976) 2 3 This model retains the distinction between diversifiable a n d nondiversifiable risk b u t imposes fewer restrictive assumptions in its derivation of asset returns than d o e s the CAPM. For e x a m p l e , t h e traditional pricing m o d e l requires that returns follow the normal distribution, implying that knowledge of the m e a n a n d variance is sufficient to describe the entire distribution. The traditional CAPM relies on t h e return t o the market portfolio as t h e benchmark variable that describes asset return behavior relative to it. In contrast, arbitrage pricing theory d o e s not require normally distribECONOM1C REVIEW, NOVEMBER/DECEMBER 1989 uted returns a n d suggests that a n u m b e r of variables, known as factors (risk sources), describe asset returns. the importance of covariance risk in asset pricing as a result of the no-arbitrage assumption, which has strong theoretical appeal. The derivation of arbitrage pricing theory requires two major assumptions. First, agents are assumed to believe that s o m e identifiable set of factors generates t h e variabil ity of al 1 asset returns a n d that their relationship is consistent across the range of variables. The second assumption is that no opportunities are available for riskless arbitrage (that is, n o unlimited profits given no net investment). The hypothesized linear factor model is: Arbitrage pricing theory is an attractive generalization of the traditional CAPM model's insight that covariance risk—risk that cannot b e diversified away—underlies t h e pricing of assets. 24 The arbitrage pricing model provides a c o h e r e n t structure, less restrictive than t h e CAPM, that allows for investigation of the sources of risk. The linear factor m o d e l framework, in addition, appears better a b l e to account for the anomalies that conflict with the traditional model's predictions. In arbitrage pricing theory, the covariance is measured relative to t h e factors that d e t e r m i n e the behavior of asset returns, whereas the CAPM gauges covariance only relative to the market return. Thus, finding a size factor or a seasonal factor that explains the CAPM anomalies would seem possible. where /?, is the uncertain return to asset /, £,- is the expected return to asset /', bf/- is t h e factor loading for asset i related to factor j, or asset /"s sensitivity to movements in factor /', is the factor j {j = 1 . . . k), and 8j is t h e error term for asset /'. In addition, the m o d e l assumes that the factors and error terms have a mean of zero. It d o e s not m a k e other assumptions a b o u t the distribution of the factors or error terms aside from requiring that the covariance between the error terms, e,- a n d e;-, is zero. In t h e derivation of the traditional capital asset pricing m o d e l , the "market" m o d e l that relates all asset returns to movements in the market return (prior to the pricing equation) follows directly from the assumption that returns are jointly normally distributed. In arbitrage pricing theory, however, the linear factor model is an assumption, although the idea that a set of forces determines the movements of all asset returns is compelling. Exact arbitrage pricing implies t h e following asset pricing relationship: where A^ is riskless or zero b return a n d represents risk premia related to factor j. A clear intuition underlies the equilibrium relation of arbitrage pricing theory. If the indicated factors truly generate t h e movements of all asset returns and if current asset prices allow n o riskless arbitrage, it follows that expected returns are approximately I ¡nearly related to covariance between the asset returns and the factors. A main contribution of this theory is that it recognizes Arbitrage pricing theory has few underlying a s s u m p t i o n s . It has b e e n criticized, though, because its initial form has few rejectable hypotheses. Refuting the theory itself, which d o e s not identify or limit the n u m b e r of factors, is difficult. 2 5 Thus, tests of t h e arbitrage pricing theory are c o m b i n e d examinations of the pricing relationship a n d the appropriateness of the set of factors chosen. 2 6 Theoretical extensions a n d refinements by a n u m b e r of researchers have provided the foundation for the substantial a m o u n t of empirical research that has b e e n produced and the many works that are still in progress. 27 Empirical m e t h o d s d e v e l o p e d to i m p l e m e n t estimation of the factors a n d factor loadings in arbitrage pricing theory have involved two distinct a p p r o a c h e s t o t h e d a t a : factor analytic techniques (or principal c o m p o n e n t analysis as in the work of Gregory Connor a n d Robert A. Korajczyk (1988|) a n d prespecification of the factors. 28 The former m e t h o d employs the estim a t e d covariance matrix of returns to determ i n e the factor structure that underlies asset return behavior. 2 9 Estimates of the factors are d e t e r m i n e d in accordance with arbitrage pricing theory; that is, factors are estimated from t h e characteristics observed in the set of returns. The second technique attempts to identify factors w i t h o u t first examining t h e structure of returns. Instead, variables are chosen as n e e d e d by economic intuition that these factors affect II FEDERAL RESERVE BANK OF ATLANTA asset pricing. The m e t h o d uses t h e prespecified factors to estimate factor loadings a n d then tests to see if t h e loadings are associated with significant risk premia. 3 0 This article will survey only the m o s t recent of the large n u m b e r of papers on arbitrage pricing theory. 31 Two recent works c o m p a r e this theory with the traditional CAPM approach as m o d e l s of asset pricing. 32 Bruce N. L e h m a n n a n d David M. M o d e s t (1988) cannot reject the arbitragebased construct when asset portfolios are formed on the basis of dividend yield or an asset's own return variance. Since CAPM research has found an anomaly with regard t o d i v i d e n d yield a n d asset pricing, the evidence can b e viewed as supportive of the arbitrage pricing theory as an alternative. O n t h e other hand, t h e researchers reject t h e arbitrage pricing m o d e l when t h e portfolios are formed on the basis of firm size. Connor a n d Korajczyk (1988) show evidence consistent with Lehmann a n d Modest that a significant relationship exists between firm size a n d asset expected return that is not captured by the arbitrage pricing theory. However, Connor a n d Korajczyk demonstrate that t h e size effect is separate from a seasonal effect (for example, the January effect), which appears to b e explained by the variation in t h e risk factors. 33 K.C. Chan, Nai-fu Chen, a n d David A. Hsieh (1985) a n d Chen, Roll, and Ross (1986) e m p l o y prespecified factors in testing t h e arbitrage pricing theory predictions. In b o t h works, t h e theory shows n o significant anomalous behavior related t o firm size. 34 Since t h e two empirical methodologies are q u i t e different, t h e conflicting evidence suggests that more research is n e e d e d to investigate t h e issue. In sum, arbitrage pricing theory represents a generalization of t h e CAPM intuition that covariance risk forms a basis for asset pricing. The arbitrage m o d e l makes few assumptions in its derivation a n d provides a m e t h o d to investigate the underlying sources of asset risk. This theory has shortcomings, however, especially with regard t o t h e n u m b e r a n d t h e identity of t h e underlying risk factors a n d empirical testing. The factor structure has b e e n estimated in different ways, although t h e techniques that estim a t e this structure from t h e e s t i m a t e d covariance matrix of asset returns seem most consistent with the theoretical model. 36 The current status of arbitrage pricing theory testing suggests its consistency with the data, although the m o d e l cannot explain the anomalous firm size effect. However, t h e capital asset pricing m o d e l cannot account for that anomaly, either. Thus, a m o n g static m o d e l s , arbitrage pricing theory seems to b e a viable alternative to t h e CAPM. Dynamic Models Although theoretical extensions of both paradigms have extended t h e basic m o d e l to an intertemporal realm in search of a productive way to link pervasive economic factors t o equity market performance, b o t h arbitrage pricing theory a n d t h e capital asset pricing m o d e l e m p l o y the fiction of a single period m o d e l . 3 5 The linkage between macroeconomic and financial markets has also b e e n explored within a dynamic general equilibrium m o d e l , notably by William A. Brock (1980, 1982), a n d |ohn C. Cox, Jonathan E. Ingersoll, Jr., a n d Ross (1985), in which asset prices are e n d o g e n o u s functions of underlying economic forces. However, t h e empirical implications of these m o d e l s are not directly testable. Brock provides a key link to understanding the relationship between t h e static and intertemporal models, a n d h e emphasizes the degree of generality in t h e m o d e l s that generate t h e testable implications. The Merton pricing m o d e l , for example, introduced intertemporal aspects to t h e traditional m o d e l , b u t Merton s is a partial equilibrium construct. The state variables that d e t e r m i n e intertemporal asset price movements are not linked to the underlying sources of uncertainty in t h e economy. Brock, on t h e other hand, derives a general equilibrium m o d e l in which the state variables—here, technological factors underlying economic uncertainty—determine behavior of asset prices. The model provides o n e interpretation of technological shocks as the arbitrage pricing theory factors presented in Ross (1976), b u t it may also apply to Merton's m o d e l , or Long's (1974). In an example, Brock shows that t h e Sharpe-Lintner capital asset pricing m o d e l conforms to the case of o n e underlying technological shock. As a further exposition, t h e Brock m o d e l has t h e ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 characteristic of consumption sufficiency—as in the Breeden model—so that the c o n s u m p t i o n CAPM m o d e l holds. Thus, Brock presents a framework in which all major asset pricing models may b e derived a n d provides a unifying syst e m to motivate research linking macroeconomic factors and asset price behavior. Empirical research on arbitrage pricing theory that employs prespecified factors to estimate factor l o a d i n g s relates closely to t h e macrofinance implications of the Brock as well as the Cox, Ingersoll, a n d Ross models. In fact, Chen, Roll, and Ross use these more general m o d e l s on which to b a s e their empirical investigation. As m e n t i o n e d above, t h e m e t h o d appears to stretch the arbitrage pricing theory's motivation of t h e determination of factor structure, yet they address the issue of the factors' identity. The motivation of linking the real economy with asset returns and the freedom to choose factors a priori has produced s o m e stimulating empirical research. Chan, Chen, and Hsieh (1985) investigated the firm size effect by prespecifying factors in a multifactor pricing equation, in which the factors are measures of economic and financial activity that may relate to asset pricing. They include the market portfolio, industrial p r o d u c t i o n , two estimates of inflation, the change in the term structure, and the risk premium. The results show that the observed firm size effect has a strong relationship with t h e risk premium measure (the difference between t h e yield on lowrated long-term b o n d s a n d t h e yield t o a portfolio of long-term government bonds). The variation in t h e risk premium reflects alterations in business conditions and, therefore, introduces an intertemporal feature into the empirical asset pricing m o d e l . These results imply that the firm size effect may b e captured by the multifactor asset pricing m o d e l . Also, the firm size effect may b e consistent with an efficient market in which small firms have higher expected returns because of higher risk that is not captured by the traditional CAPM model's risk measures. S u b s e q u e n t empirical work by Chen, Roll, and Ross investigated directly t h e role of economic forces in asset pricing, using similar econ o m i c variables as t h e prespecified factors. They found similar evidence that t h e multifactor model explains t h e pricing of a selection of asset portfolios formed on the basis of firm size. The authors s h a p e d the selection of economic state variable proxies by choosing those that influence either cash flows to firms or t h e discount rate a p p l i e d to asset cash flows in the simple stock valuation m o d e l . The formulation of prespecified factors in these studies assumes that the chosen variables constitute t h e factor structure of asset returns, which is the underlying determinant of asset return time variation. The research d o e s not present time-series evidence to suggest that these factors explain much of the timeseries variability of asset returns. In factor analytic research, time-series explanatory power is evident in t h e m e t h o d of identifying factors. In future research on economic factors and asset pricing, the time-series regressions of the prespecified factors will b e useful indicators of whether t h e chosen factors are relevant. 36 The underlying shocks (or factors) in Brock and Cox, Ingersoll, and Ross represent technological shocks that directly affect the productivity of t h e economy. Yet the main explanatory variables in these studies are financial measures, notably the term structure proxy and t h e risk p r e m i u m proxy. These variables are at least partially determined by the true underlying factors, just as asset prices are. Although financial factors have significant implications for the pricing of risky assets and provide insights into the interrelationships of macroeconomic and financial markets, the results d o not reveal the underlying sources of uncertainty. The search for these underlying sources may seem futile. However, recent studies by David Alan Aschauer (1989a, b) suggest that governm e n t s p e n d i n g behavior, primarily changes in t h e public capital stock, may b e o n e source of uncertainty which directly affects the aggregate productivity of the economy. In his 1989 paper, this researcher shows that the government stock of infrastructure capital—for example, roads, buildings, sewers, and so on—has a significant i m p a c t on t h e profitability of t h e aggregate economy. In other words, t h e p u b l i c capital stock has a positive effect on t h e aggregate value of private firms. Thus, further work on asset pricing should investigate the effects of government policy, since these shocks seem most justifiably to b e exogenous variables. 3 7 Research in this area may also provide economic policymakers with better information a b o u t the II FEDERAL RESERVE BANK OF ATLANTA long-term effects of s p e n d i n g policies at all levels of government. Conclusion Recent financial models of asset pricing derive from the initial insights into the relative riskiness of different assets and the trade-off between risk and expected return provided by the traditional capital asset pricing model. The various models surveyed here provide a coherent framework in which to analyze asset returns. Over time, the CAPM has b e e n useful for portfolio evaluation and for extending scientific analysis of financial markets. It continues to b e used as a method to evaluate the risk of assets or portfolios. The model requires only one estimable parameter per firm or portfolio. However, the CAPM's static framework, the difficulty of predicting the return to the market proxy, evid e n c e in conflict with its implications, a n d theoretical advances in financial modeling have combined to shift attention away from the simple CAPM as an equilibrium model of asset pricing. The consumption-based asset pricing model (CCAPM) presents a dynamic construct in which a single estimable parameter measures asset risk. This model has m e t criticism on matters of data measurement and inconclusive empirical evidence of its usefulness. Recent work nonetheless presents some support for further research in this area. The intertemporal CAPM and arbitrage pricing theory, though clearly different models, suffer similar empirical difficulties. Neither provides insights into the identity of the multiple sources of asset risk. For arbitrage pricing theory, empirical a p p l i c a t i o n s using factor analysis cannot interpret risk sources. However, the two models provide a motivation for investigating multiple sources of asset risk. Ongoing theoretical d e v e l o p m e n t s p o i n t toward future research that can link economic factors to asset pricing behavior. Such research should interest investors and especially policymakers, who may gain insight into the effects of alternative economic policies. For the policymaker, an appreciation of the risk sources in the economy can aid formulation of policy by linking information provided by equity market behavior to real economic performance. In addition, further research may uncover elements of economic policy as sources of macroeconomic uncertainty and provide policymakers with an improved set of policy guides. Notes 1 tribution remains t h e m o s t c o m m o n a p p r o x i m a t i o n of t h e In fact, as e v i d e n c e of their influence, financial asset pric- distribution of stock returns. ing theories have p e n e t r a t e d Wall Street a n d are currently b e i n g u s e d in t h e design of m u t u a l fund portfolios. 2 T h e traditional capital asset pricing m o d e l remains, how- 5 S e e also Treynor (1961). 6 T h e traditional CAPM involves several a d d i t i o n a l assumptions that m a y b e f o u n d in S h a r p e (1985) or other fi- ever, a useful m e t h o d t o analyze asset characteristics with n a n c e textbooks. few measures. 7 Black ( 1972) presents a m o d e l without a riskless asset that 3 This uncertainty is larger in s o m e periods than in others. 4 D e s p i t e e v i d e n c e that stock returns a p p e a r leptokurtic offers predictions that are m o r e consistent with t h e Black, ("fat-tailed," or having a higher probability t h a n a normal Jensen, a n d Scholes results. distribution of observing extreme values), the normal dis- 38 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 8 9 Levy (1978) suggests that if individuals are not well diver- t h e highest return. Roll's (1977) critique suggests that this sified their own variance of return s h o u l d b e relevant for implication has not b e e n tested a d e q u a t e l y . For arbitrage asset pricing. pricing theory, a portfolio that has only risk related t o t h e For a n intuitive discussion of event study m e t h o d o l o g y f u n d a m e n t a l factors is mean-variance efficient. This as- s e e Hunter a n d Walker (1988). p e c t of the m o d e l has b e e n tested empirically in several I(> 27 cific m o d e l . 1 studies. rhe results, however, d o not represent a test of any spe- S e e , for example, H u b e r m a n (1982), C h a m b e r l a i n a n d ' O n e may interpret these state variables to b e virtually R o t h s c h i l d (1983), a n d C o n n o r (1984). S e e H u b e r m a n anything: t h e weather, oil s u p p l y shocks, m e a s u r e s of g o v e r n m e n t policy, a n d s o o n . As will b e d i s c u s s e d ,2 1 (1986) for a m o r e exhaustive list of references. 28 Factor analysis is a statistical procedure in which "com- further, t h e identity of such variables is subject to con- m o n factors" are u n o b s e r v a b l e hypothetical variables that siderable debate. c o n t r i b u t e t o t h e v a r i a n c e of a vector of d e p e n d e n t S e e results in Chan, C h e n , a n d Hsieh (1985); Chen, Roll, variables. That is, factor analysis is a m e t h o d t o d e s c r i b e a n d Ross (1986); a n d McElroy a n d Burmeister (1988). t h e variation of a set of variables without explicit ex- ^The m o d e l has also b e e n referred to as the m u l t i b e t a planatory variables. A data series, then, will b e d e s c r i b e d CAPM, in which an e s t i m a t e d parameter is associated with as a linear f u n c t i o n of a set of c o m m o n factors a n d o n e each state variable. u n i q u e factor that contributes variance only t o that series. T h e p r o b l e m , however, is not u n i q u e to Merton's m o d e l , In the set of d e p e n d e n t variables, each variable has o n e ,4 15 as t h e discussion of arbitrage pricing theory indicates. u n i q u e factor that is uncorrelated with all other u n i q u e O n e c o u l d also use the asset portfolio that has m a x i m u m factors. The coefficients, or factor loadings, for each comm o n factor provide t h e e s t i m a t e s of b¡ in the APT. correlation with the growth rate in c o n s u m p t i o n as t h e 29 benchmark. l6 S e e also R u b e n s t e i n (1976) a n d B r e e d e n a n d A l t h o u g h factor analysis is m o r e efficient, the compu- Litzen- tational d e m a n d s of t h e m e t h o d limit the n u m b e r of berger (1978). l7 S e e Cornell (1981) for technical e l e m e n t s a n d t h e com- returns that can b e analyzed at o n e time. 30 H u b e r m a n (1986) suggests that this form of research, relat- p l e t e argument. Bergman (1985) criticizes t h e a s s u m p t i o n ing e x p e c t e d return t o covariances of asset returns with of time-separable preferences in t h e derivation of t h e other variables, is m o r e in line with the Merton intertem- CCAPM. This a s s u m p t i o n i m p l i e s that past decisions o n poral CAPM. Below, t h e Brock m o d e l is u s e d t o show t h e c o n s u m p t i o n d o n o t affect today's choices. W i t h o u t the a s s u m p t i o n , t h e Merton ICAPM still holds, b u t it can n o similarity of the two m o d e l s . 3 ' S o m e references for t h e earlier yet important works are longer b e c o l l a p s e d into t h e CCAPM. D e s p i t e this prob- (1983), a n d Dhrymes et al. (1985). S e e H u b e r m a n (1986) for time-sepa'rable preferences a n d that t h e CCAPM holds. a m o r e extensive listing. ' i n s t r u m e n t a l variables e s t i m a t i o n m e t h o d s use variables 19 20 Roll a n d Ross (1980), Brown a n d Weinstein (1983), C h e n lem, m o s t m a c r o e c o n o m i c m o d e l s a s s u m e that there are 32 L e h m a n n a n d M o d e s t (1988) use factor analytic tech- correlated with t h e regressors b u t unrelated t o t h e errors n i q u e s o n 750 asset returns to isolate a factor structure. in an a t t e m p t to r e d u c e t h e potential correlation be- Then they test t h e APT with t h e s e factors o n a selection of tween regression variables a n d t h e residual error. asset portfolios, g r o u p e d o n t h e basis of d i v i d e n d yield, Wheatley suggests that either t h e instruments are weakly an asset's own return variance, a n d firm size. C o n n o r a n d related to t h e underlying variables of interest or that t h e Korajczyk (1988) e s t i m a t e factors using asymptotic prin- u n d e r l y i n g v a r i a b l e s a r e c o l l i n e a r . In e i t h e r case, t h e cipal c o m p o n e n t s , which allows m o r e returns in t h e results in M a n k i w a n d S h a p i r o (1986) are suspect. estimation of t h e covariance matrix. The asset portfolios A l t h o u g h t h e results d o n o t suggest r e j e c t i o n of t h e u s e d as d e p e n d e n t variables are g r o u p e d o n the basis of size. CCAPM, t h e e s t i m a t e of t h e relative risk aversion parameter greatly exceeds t h e theoretical value. 2 33 B o t h p a p e r s reject t h e restriction of m e a n variance ef- 34 T h e research that uses prespecified factors to test APT will ficiency in APT as well as in t h e CAPM. ' T h e a s s u m p t i o n s are (1) joint log-normality of asset returns a n d c o n s u m p t i o n growth a n d (2) a constant relative 22 risk aversion specification of utility. b e e x a m i n e d further b e l o w in the discussion of mac- S i n c e t h e test a s s u m e s b o t h constant relative risk aver- roeconomic factors a n d asset pricing. sion utility a n d joint log-normality of returns a n d con- S e e O h l s o n a n d G a r m a n (1980) a n d Connor a n d Korajczyk (1988) for discussions of intertemporal arbitrage pricing b e t h e source of t h e m o d e l failure. Further work may b e theories. r e q u i r e d in t h i s area t o d e c i p h e r t h e i m p l i c a t i o n s of t h e results. 23 35 s u m p t i o n growth, the violation of either restriction could The intertemporal CAPM, CAPM, a n d a recent intertemporal m o d e l in Cox, Ingersoll, a n d Ross (1985) have m u c h For a m o r e d e t a i l e d survey of t h e APT, s e e H u b e r m a n in c o m m o n with m o d e r n dynamic macroeconomic m o d e l s (1986). (stochastic growth models) such as t h o s e of Lucas (1978) 24 S e e S h a r p e (1985): 199-200. a n d Brock (1982). These similarities i n c l u d e the key role of 25 S e e S h a n k e n (1982, 1985) a n d the r e s p o n s e by Dybvig a n d t h e real interest rate, concern for the changing investment Ross (1985). a n d c o n s u m p t i o n o p p o r t u n i t i e s faced by c o n s u m e r s , a n d 26 T h e m o d e l implications can b e e x t e n d e d to the c o n c e p t t h e attention to e c o n o m i c forces as the underlying sour- of mean-variance efficiency. In t h e CAPM, t h e m o d e l ces of asset risk p r e m i a . The progress in m o d e l i n g a i m p l i e s that t h e market portfolio is mean-variance effi- d y n a m i c e c o n o m y with an asset market has occurred in cient; that is, given its level of risk, t h e market portfol io has b o t h fields; empirical work on t h e issue has n u m e r o u s II FEDERAL RESERVE BANK OF ATLANTA potential applications. It is notable, however, t h a t only (1986) in a nonlinear e s t i m a t i o n m e t h o d that presents t h e ICAPM a t t e m p t s t o i s o l a t e u n d e r l y i n g e c o n o m i c b o t h time-series a n d cross-sectional pricing results of t h e variables—that is, state variables-directly; t h e c o n s u m p - multifactor m o d e l . Their results generally s u p p o r t t h e tion CAPM is similar t o t h e traditional CAPM in the u s e of a usefulness of t h e e c o n o m i c variables in explaining b o t h types of variation. reference portfolio t o price assets relative t o that port37 folio. 36 M c E l r o y a n d Burmeister (1988) e m p l o y v a r i a b l e s s i m i l a r t o Chan, C h e n , a n d Hsieh (1985) a n d C h e n , Roll, a n d Ross R e s e a r c h by Tallman (forthcoming) investigates t h e effects of g o v e r n m e n t s p e n d i n g behavior o n cross-sectional asset pricing m o r e in t h e tradition of financial research. References Aschauer, David Alan. "Is Public E x p e n d i t u r e Productive?" journal of Monetary Economics 23 (March 1989a): 177-200. . " P u b l i c E x p e n d i t u r e a n d t h e Stock Market." Federal Reserve Bank of Chicago u n p u b l i s h e d manu- Bergman, Yacov. " T i m e Preference a n d Asset Pricing Modof Financial Economics 14 (March 1985): 51 ( S e p t e m b e r 1983): 1305-23. C h a m b e r l a i n , Gary, a n d Michael Rothschild. "Arbitrage, FacM a r k e t s . " Econometrica 51 ( S e p t e m b e r 1983): 1281- 1304. Chan, K.C., Nai-fu C h e n , a n d David A. Hsieh. "An Exploratory 145-59. Black, Fischer. "Capital Market E q u i l i b r i u m with Restricted Borrowing." journal of Business Capital Asset Pricing M o d e l : S o m e Empirical Tests." In in the Theory of Capital e d i t e d by Markets, Michael C. Jensen. New York: Praeger Publishers, 1972. Bradley, Michael, a n d L. M a c d o n a l d W a k e m a n . "The Wealth Effects of Targeted Share Repurchases, "journal cial Economics of Finan- with Stochastic C o n s u m p t i o n a n d Investment Opporof Financial Economics of Financial 14 (April 1985): 451-71. Chen, Nai-fu. " S o m e Empirical Tests of t h e Theory of Arbitrage Pricing." journal of Finance 38 ( D e c e m b e r 1983): 1393-1414. C h e n , Nai-fu, Richard Roll, a n d S t e p h e n A. Ross. " E c o n o m i c Forces a n d t h e Stock Market." journal of Business 7 (July 1979): 59 duly 1986): 383-404. Connor, Gregory. "A Unified Beta Theory." journal 11 ( S e p t e m b e r 1982): 301-28. Breeden, Douglas T. "An Intertemporal Asset Pricing M o d e l tunities." journal Investigation of t h e Firm Size Effect ."journal Economics 45 duly 1972): 444-54. Black, Fischer, Michael C. lensen, a n d Myron S. Scholes. "The Studies Arbitrage Pricing M o d e l s . " Econometrica tor Structure, a n d Mean-Variance Analysis o n Large Asset script, 1989b. e l s . " journal C h a m b e r l a i n , Gary. " F u n d s , Factors, a n d Diversification in nomic Theory of Eco- 34 (October 1984): 13-31. Connor, Gregory, a n d Robert A. Korajczyk. "Risk a n d Return in an E q u i l i b r i u m APT." journal of Financial 21 Economics (March 1988): 255-89. 265-96. Breeden, Douglas T., Michael R. G i b b o n s , a n d Robert Lit- Cornell, Bradford. " T h e C o n s u m p t i o n Based Asset Pricing zenberger. "Empirical Tests of the Consumption-Oriented M o d e l : A N o t e o n Potential Tests a n d Applications."/our- CAPM." journal nal of Financial of Finance 44 Qune 1989): 231-62. Breeden, Douglas T„ a n d Robert Litzenberger. "Prices of State-Contingent C l a i m s Implicit in O p t i o n Prices." journal of Business Brock, William A. "Asset Pricing in an Economy with Production: A 'Selective' Survey of Recent Work o n Asset-Pricing Optimization and e d i t e d by Pan-Tai Liu. Vol. 19 of Economics, Concepts and Methods in Science Mathematical Mathematical and Engineering. New . "Asset Prices in a Production Economy." In The of Information and Uncertainty, e d i t e d by J.J. McCall, 1-46. Chicago: University of Chicago Press, 1982. Brown, S t e p h e n , a n d Mark I. Weinstein. "A New Approach t o Testing Asset Pricing M o d e l s : T h e Bilinear Paradigm." journal of Finance 38 (June 1983): 711-43. Econometrica I (1933): 309-24. "A Theory of t h e Term Structure of Interest Rates." metrica Econo- 53 (March 1985): 363-84. Dhrymes, P h o e b u s J., Irwin Friend, Mustafa N. Gultekin, a n d N. Bulent Gultekin. " N e w Tests of t h e APT a n d Their Implications." journal of Finance 40 (July 1985): 659-75. Dybvig, Philip H„ a n d S t e p h e n A. Ross. "Yes, t h e APT is Test- York a n d L o n d o n : P l e n u m Press, 1980. Economics 9 (March 1981): 103-08. ' Cox, John C., Jonathan E. Ingersoll, Jr., a n d S t e p h e n A. Ross. 51 (October 1978): 621 -51. M o d e l s . " In Dynamic Economics Cowles, Alfred, III. " C a n Stock Market Forecasters Forecast?" able." journal of Finance 40 (September 1985): 1173-88. Fama, E u g e n e F. " M u l t i p e r i o d C o n s u m p t i o n Decisions." The American Economic Review Investment 60 (March 1970): 163-74. Fama, E u g e n e F„ a n d James D. Macbeth. "Risk, Return a n d E q u i l i b r i u m : Empirical Tests." journal of Political Econ- omy 81 (May/June 1973): 607-36. 40 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Fama, E u g e n e F., a n d M e r t o n H. Miller. The Theory of Hinsdale, III.: Dryden Press, 1972. Finance. Merton, Robert C., a n d Stanley Fischer. "Macroeconomics a n d Finance: The Role of the Stock Market." In Essays Ferson, Wayne E., a n d John ). Merrick, )r. "Non-Stationarity Macroeconomic Implications of Financial and on Labor e d i t e d by Karl Brunner a n d Stage-of-the-Business-Cycle Effects in Consumption- Markets B a s e d Asset Pricing R e l a t i o n s . " lournal a n d Allan H. Meltzer, 57-108. Carnegie-Rochester Con- of Financial 18 (March 1987): 127-46. Economics "Estimating t h e Continuous-Time C o n s u m p t i o n Based Statistics Processes, f e r e n c e Series o n P u b l i c Policy, vol 21. A m s t e r d a m : Grossman, Sanford J., A n g e l o Melino, a n d R o b e r t J . Shiller. Asset Pricing M o d e l . " lournal and Political of Business and Economic 5 duly 1987): 315-27. North-Holland, 1984. Mossin, Jan. " E q u i l i b r i u m in a Capital Asset Market." metrica Ohlson, James A., a n d Mark B. G a r m a n . "A Dynamic Equilib- Hall, Robert E. " C o n s u m p t i o n . " National Bureau of Eco- rium for t h e Ross Arbitrage M o d e l . " lournal Hansen, Lars P., a n d Kenneth J. Singleton. " G e n e r a l i z e d Ritter, Jay R„ a n d Navin Chopra. "Portfolio Rebalancing a n d Instrumental Variables Estimation of Nonlinear Rational t h e Turn-of-the-Year Effect." lournal E x p e c t a t i o n s M o d e l s . " Econometrica 1989): 149-66. 50 ( S e p t e m b e r 1982): 1269-86. s u m p t i o n , Risk Aversion, a n d t h e Temporal Behavior of Asset Returns." Journal of Political Economy 91 (April Tests." lournal of Financial 4 (March 1977): Economics 129-76. Roll, Richard, a n d S t e p h e n A. Ross. ""An Empirical Investigation of the Arbitrage Pricing Theory." lournal 1983): 249-65. H u b e r m a n , Gur. "A S i m p l e Approach to Arbitrage Pricing." of Economic Theory 28 (October 1982): 183-91. Center for Research in Security Prices Working Paper Fairness of Investment Bankers' Fees." Federal Reserve Review 73 (March/April 1988): 2-7. Pricing." lournal 13 (May 1976): 341-60. of Economics and Management Science 3 (Autumn 1972): 357-98. as H e d g e s against Shifts in the C o n s u m p t i o n or Invest- of 54 (April of Business 1981): 305-28. Shanken, lay. "The Arbitrage Pricing Theory: Is It Testable?" lournal 12 (June 1983): 13-32. Economics of Schipper, Katherine, a n d Rex T h o m p s o n . " C o m m o n Stocks of Finance 37 ( D e c e m b e r 1982): 1129-40. . "Multi-Beta CAPM or Equilibrium-APT?: A Keim, D o n a l d B. "Size-Related A n o m a l i e s a n d Stock Return S e a s o n a l i t y : Further E m p i r i c a l E v i d e n c e . " lournal lournal 7 (Autumn 1976): 407-25. Economics m e n t O p p o r t u n i t y S e t . " lournal lensen, Michael C. "Capital Markets: Theory a n d Evidence." of Economics Finance R u b e n s t e i n , Mark. " T h e V a l u a t i o n of Uncertain I n c o m e S t r e a m s a n d t h e Pricing of O p t i o n s . " Bell Series No. 166 (May 1986). Hunter, William Curt, a n d Mary Beth Walker. "Assessing t h e Bank of Atlanta Economic of 35 ( D e c e m b e r 1980): 1073-1103. Ross, S t e p h e n A. "The Arbitrage Theory of Capital Asset . "A Review of t h e Arbitrage Pricing Theory." Financial 44 (March of Finance Roll, Richard. "A C r i t i q u e of t h e Asset Pricing Theory's Hansen, Lars P., a n d Kenneth J. Singleton. "Stochastic Con- The Bell lournal 35 of Finance (June 1980): 675-84. n o m i c Research Working Paper No. 2263 (1987). lournal Econo- 34 (October 1966): 768-83. Reply." lournal of Finance 40 (September 1985): 1189-96. S h a n k e n , Jay, a n d Mark I. Weinstein. "Macroeconomic Vari- L e h m a n n , Bruce N„ a n d David M. Modest. "The Empirical a b l e s a n d Asset Pricing: Estimation a n d Tests." Univer- of sity of Rochester, S i m o n School of Business, u n p u b l i s h e d l-evy, Haim. " E q u i l i b r i u m in an Imperfect Market: A Con- Sharpe, William F. "Capital Asset Prices: A Theory of Market Foundations of t h e Arbitrage Pricing Theory." lournal Financial Economics working p a p e r , 1989. 21 (January 1988): 213-54. straint o n t h e N u m b e r of Securities in the Portfolio." The American Economic Review 68 (September 1978): 643-58. E q u i l i b r i u m u n d e r C o n d i t i o n s of Risk." lournal Lintner, |ohn. "The Valuation of Risk Assets a n d t h e Selection of Risky Investments in Stock Portfolios a n d Capital Budgets." The Review of Economics and Statistics 47 (February 1965): 13-37. ture of Interest Rates." lournal Finance . Investments. Englewood Cliffs, N.J.: Prentice- Hall, 1985. S t a m b a u g h , Robert F. " O n t h e Exclusion of Assets from Tests of t h e Two-Parameter M o d e l . " lournal Long, John B„ Jr. "Stock Prices, Inflation, a n d t h e Term Struc- of 19 ( S e p t e m b e r 1964): 425-42. nomics of Financial Eco- 10 (May 1982): 237-68. 1 Tallman, Ellis W. "Real Shocks a n d Asset Premia." Federal Lucas, Robert E., Jr. "Asset Prices in an Exchange Economy." Tinic, S e h a M., a n d Richard R. West. " R i s k Return a n d of Financial Economics (February 1974): 131-70. Econometrica Reserve Bank of Atlanta Working Paper (forthcoming). 46 ( N o v e m b e r 1978): 1429-45. E q u i l i b r i u m : A Revisit." lournal Mankiw, Gregory N., a n d Matthew D. Shapiro. "Risk a n d Return: C o n s u m p t i o n Beta versus Market Beta." of Economics and Statistics Review 68 (August 1986): 452-59. Markowitz, Harry. "Portfolio Selection." lournal of Finance ing Theory as a Restricted Nonlinear Multivariate Regression M o d e l : ITNLSUR Estimates." lournal Statistics of Business 6 (January 1988): 29-42. Merton, Robert C. "An Intertemporal Capital Asset Pricing M o d e l . " Econometrica 94 Studies 25 (February 1958): 65- 85. Treynor, Jack. "Toward a Theory of Market Value of Risky (March 1952): 77-91. McElroy, Marjorie B., a n d Edwin Burmeister. "Arbitrage Pric- and Economic Economy Tobin, James. " L i q u i d i t y Preference as Behavior Toward Risk." Review of Economic 1 of Political (February 1986): 126-47. 41 ( S e p t e m b e r 1973): 867-87. Assets." U n p u b l i s h e d manuscript, 1961. Wheatley, Simon. " S o m e Tests of t h e C o n s u m p t i o n Based Asset Pricing M o d e l . " lournal of Monetary Economics of Investment Value. 22 (March 1988): 193-215. Williams, (ohn B. The Theory Cam- bridge, Mass.: Harvard University Press, 1931. II FEDERAL RESERVE BANK OF ATLANTA U.S. and Foreign Direct Investment Patterns William ). Kahley Since 1985, when this Bank's Economic Review last surveyed foreign direct investment, foreign ownership of plants, real estate, and the like in the United States has grown considerably in importance and magnitude. The 1985 article reported: "Many Americans are unaware t h a t . . . international activities wield a sizable a n d steadily growing impact u p o n t h e economic characteristics of the region and the nation." In the meantime both American awareness and direct foreign investment have m a d e significant gains. This article draws on newly available data to u p d a t e the status of foreign direct investment in the United States and the Southeast. It also probes some economic impacts of foreign direct investment, focusing particularly on employment and exports. A wide variety of observers, from policymakers to small business owners, are now interested in investments that foreigners make in the U.S. economy. As the magnitude of investment by Europeans, Asians, and others rose in the 1980s, public concern over such foreign investment accelerated, intensifying the d e m a n d for The author Atlanta is an economist Fed's Research for extensive 42 research in the regional Department. assistance. section He thanks Amy of the Bailey information on the subject. Accounts of billiondollar acquisitions of major U.S. corporations by foreigners often relayed fears that control of corporate America may b e slipping out of domestic hands. O p i n i o n polls have shown that t h e American public is troubled by the increased foreign ownership of U.S. firms a n d real estate. In contrast, government officials and other opinion leaders in southeastern states have displayed a generally positive perspective on foreign direct investment. The Southeast receives an especially large share of foreign companies' spending on new plant and equipment. Political and business leaders have come to appreciate t h e jobs, tax base, a n d diversification that foreign direct investment can bring. Many workers also welcome the foreign presence. In 1987 the number of Americans employed in foreign-owned U.S. affiliates was 3.2 mill ion. More than one out of eight of these jobs were in the Southeast, defined in this article as the states in the Sixth Federal Reserve District: Alabama, Florida, Georgia, Louisiana, Mississippi, and Tennessee. Today, southeastern employees working for foreign-owned U.S. affiliates probably number close to half a million. U.S. ownership of plants abroad generally predates foreigners' direct investments in busiECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 nesses here. The major expansion in U.S. direct investment overseas took place in the 1950s and 1960s, whereas international investments in the United States have b e e n especially fastgrowing only since the 1970s. Just as many Americans now oppose foreign direct investm e n t here, critics have faulted U.S. corporations' decisions to produce outside the United States, chiefly claiming that millions of American workers' jobs are lost in the process. For their part, U.S. multinational corporations have argued that their direct foreign presence is n e e d e d to serve foreign markets adequately and that customers in the United States benefit from lower prices on imported goods produced at lower-cost foreign subsidiaries. Based on available data and information, it is impossible to estimate accurately the net economic impact on this country of U.S. companies' foreign investment activities or of foreign direct investment here. However, the ever-expanding amount of research on the topic suggests that both types of flows probably increase U.S. production and well-being. From a theoretical perspective, most economists and policymakers accept the view that international capital flows help companies make better use of the world's resources. This result occurs because foreign investment presumably increases competition in an industry through the entry of new companies. Seeking a competitive edge, firms in an industry try to cut costs, improve efficiency, or enhance product quality to maintain or expand market share. Ultimately, consumers should benefit from a lower-priced or higher-quality product that economizes on resource usage. This same theoretical perspective also implies that the flow of capital across national borders benefits workers and company owners. If the resources are complementary, employees should b e better off because the availability of foreign capital raises labor productivity; consequently, wages should rise or the number of employed workers should increase. The availability of foreign capital also should reduce the cost of capital, making some plant investment projects cheaper and boosting t h e value of firms, thus benefiting their owners and stimulating investment. On the other hand, domestic savers and financial intermediaries may b e losers in the short run as a consequence of greater capital availability. Savers could lose interest income because of lower interest rates brought about by the a d d e d supply of capital, while entry of foreign lenders could increase competition in the financial lending business and reduce profitability or the rate of return. This article presents available information on t h e magnitude, industrial distribution, and geographic concentration of foreign investment in the United States generally and in the Southeast. 1 Along with a discussion of alternative measures of investment and conceptual and statistical p r o b l e m s associated with these measures, the article compares features of investment in the United States by "U.S. affiliates" of foreign c o m p a n i e s with U.S. corporations' investment in "foreign affiliates." The discussion next makes tentative assessments of some impacts of foreign investment in the United States, with particular attention to two questions: • How, if at all, have job growth and worker income been influenced by foreign investment? • How, if at all, has foreign investment stimulated U.S. exports? The final portion of the article focuses on the importance of foreign direct investment in the Southeast. The article concludes with a discussion of emerging trends and prospects for foreign investment in t h e nation a n d by U.S. companies during the 1990s. The Activities of U.S. and Foreign Affiliates Overall Activity. A statistical snapshot of inward and outward foreign investment for the United States at year's end 1987 reveals some important information. As measured by employment, U.S. multinationals were more active, with 6.2 million workers employed in their foreign affiliates compared to 3.2 million workers in U.S. affiliates of foreign c o m p a n i e s (see Table I). In contrast, less reliable data on book values of assets suggest that the magnitude of foreignowned operations in the United States is much closer to that of U.S.-owned operations abroad. The book value of foreign corporations' U.S. II FEDERAL RESERVE BANK OF ATLANTA The M e a n i n g a n d M e a s u r e m e n t of Foreign Investment U.S. statistics-gathering agencies define foreign direct investment in the United States and U.S. investment abroad as ownership or controldirectly or indirectly—of 10 percent or more of an enterprise's voting securities, or an equivalent interest by an individual, partnership, group, or organization. Businesses under such control are called affiliates, and the investment is said to be direct. Although another type of foreign investment in a private enterprise, known as portfolio investment, refers to the purchase of stocks or bonds by investors seeking to diversify their assets rather than exercise an effective management role, the terms foreign direct investment and foreign investment are used interchangeably throughout the rest of this article.1 In addition, the term multinational corporation is used to refer to all foreign investors even though some actually are individuals or other entities. There are several ways to measure the magnitude and importance of foreign investment. Conceptually, the best measure of importance is annual value added, or contribution to final output. However, value-added data are not available for foreign affiliates, and other measures that only approximate the importance or contribution of foreign investment activity must be used as proxies to compare inward and outward foreign investment.2 This article focuses on employment and the gross book value of property, plant, and equipment. These complementary gauges lend themselves to calculations of national and regional levels, shares, and growth rates by industry and by country (of origin or destination). Thus, these data serve as proxies for the stock, or cumulative value, of foreign investment and as measures of importance and change in such activity. Although employment and gross book value data tend to be correlated, or move together, their patterns of change can vary. Therefore, these data series are not equally suited for all purposes. For example, gross book value data do not serve as reliable measures of growth in real industrial activity because they are valued at (constant) acquisition cost. Market value would be better, with values for all years adjusted by prices for some base year. Also, an acquired firm may revalue its assets from historical book value to fair market value, thus changing the asset valuation while employment and the value of production remain static. Thus, data on the number of jobs or employment associated with foreign investment are better for measuring growth in foreign investment activity. On the other hand, the gross book value may be more accurate than employment figures in measuring industrial or regional shares if foreign investment is in capital-intensive industries and industries in which capital has been substituted for labor. Even then, gross book value comparisons are only approximations because of the shortcomings noted. Generally, when the amount of capital used per worker varies from industry to industry, the industrial and regional shares and patterns of change vary. The form that foreign investment takes can also make a difference in terms of its impact. For example, if a merger or acquisition merely involves the purchase of existing assets, the transfer of ownership may generate few or no new jobs. By contrast, capital inflows to build and equip new plants generate new jobs immediately. Foreign investment activities can be classified according to type and characteristics: • acquisitions and mergers of enterprises whereby title to stock or assets of a business are secured by a foreign investor; • a rise in percentage ownership by a foreign investor, known as equity increases; • joint ventures, in which two or more entities establish a new business according to contractual provisions; and • new plants and plant expansions, or a foreign investor's establishment of a new operating facility or addition to existing capacity. Notes 1 O t h e r m a j o r c o m p o n e n t s of foreign investment in t h e United States i n c l u d e foreign official assets in t h e 2 V a l u e - a d d e d e s t i m a t e s are available for U.S. affiliates at t h e national level for the 1977-86 period. A com- United States, such as their h o l d i n g s of U.S. Treasury parison of the value a d d e d a n d e m p l o y m e n t d a t a for securities, a n d U.S. b a n k liabilities. O t h e r major U.S. t h e s e firms for 1986 shows that manufacturing affili- investment assets a b r o a d i n c l u d e U.S. official reserve ates' e m p l o y m e n t share of all affiliate e m p l o y m e n t , assets, U.S. g o v e r n m e n t loans, a n d U.S. b a n k claims. In 47 percent, was a b o u t t h e s a m e as their share of all 1987 foreign direct investment totaled 26 percent of all affiliates' v a l u e a d d e d , 45 percent. Value a d d e d a n d U.S. assets a b r o a d a n d 17 percent of foreign-owned e m p l o y m e n t s h a r e s varied w i d e l y for o t h e r indus- assets in t h e United States. tries. 44 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Table 1. Selected Data for U.S. and Foreign Affiliates, 1987 Millions of Dollars Number of Employees Total Assets Annual Sales Annual Employee Compensation Foreign Affiliates of U.S. Corporations 6,234,600 1,098,166 1,052,260 134,715 U.S. Affiliates of Foreign Corporations 3,159,700 926,042 731,392 93,652 Sources: See U.S. Department of Commerce, Bureau of Economic Analysis (1989a, b, c). affiliates' property, plant, and e q u i p m e n t was $926 billion in 1987 compared to $1,098 billion for foreign affiliates of U.S. companies. However, U.S. investments abroad are on average much older than foreign-owned investments in the United States and were m a d e when asset prices were far lower. Comparing the magnitude of activity using book values of assets exaggerates the foreign presence in the United States compared to U.S. multinationals' activities abroad. In 1987 both sales and employee compensation of foreign affiliates of U.S. companies substantially exceeded those of U.S. affiliates of foreign companies. These differences also suggest that U.S. outward investment exceeds inward investment. These employment numbers and income and asset values seem large, but they appear less so when compared to the total national economy. In 1987, employment of foreign multinationals' U.S. affiliates accounted for 3.6 percent of the 86.6 million workers in n o n b a n k U . S . businesses, according to a survey conducted by the U.S. Commerce Department's Bureau of Economic Analysis. 2 This presence d o u b l e d the 1.8 percent share recorded in 1977. Although the overall percentage of workers employed by U.S. affiliates remains small, these shares are significantly larger for some industries. For example, U.S. manufacturing affiliates employed over 7 percent of all U.S. manufacturing workers and accounted for nearly half of the employment by U.S. affiliates in 1987. (Partly because foreign investment is concentrated in chemicals and other industries with relatively low employment-to-assets ratios, U.S. affiliates' 12 percent share of manufacturing assets was larger than the employment share.) Based on available information that shows foreign investment still growing at a fast pace, U.S. affiliates' share in manufacturing employment and assets may well b e larger today. Foreign a n d U.S. Characteristics. Foreign direct investment here and U.S. firms' direct investments abroad have distinct characteristics. Sources and applications of such investments shown in Tables 2 and 3 display these differences. The geographic pattern of U.S. direct investm e n t a b r o a d is m o r e d i s p e r s e d than is t h e pattern of country sources of foreigners' investments here. Foreign affiliates located in industrialized countries accounted for a b o u t 70 percent of employment in U.S.-owned enterprises abroad in 1987, while Canada, Japan, and the European nations accounted for almost 90 percent of employment in all foreign-owned U.S. affiliates. O n e of the more noticeable features of foreign investment in the United States over the past decade has been the growing prominence of Japan as a major foreign investor. Between 1977 and 1987 Japan's share of U.S. affiliate employment rose from 6 percent to 9 percent, and its share of assets among U.S. affiliates' rose from 4 percent to 21 percent. Based on 1987 data, Japan ranks first in terms of assets and fourth in terms of employment. However, the relatively recent vintage of Japanese investment may overstate the value of its assets vis-a-vis countries that have long been investing in factories and real estate. Japan also ranks higher by the asset measure than the employment measure because of Japanese investors' acquisiII FEDERAL RESERVE BANK OF ATLANTA Table 2. Shares of Assets and Employment for U.S. and Foreign Affiliates by Country, 1987 Employment Assets U.S. Companies' Foreign Affiliates Foreign Companies' U.S. Affiliates U.S. Companies' Foreign Affiliates Foreign Companies' U.S. Affiliates Canada 14.6 18.7 13.8 15.2 Europe France Germany Netherlands Switzerland United Kingdom 41.2 5.7 8.9 2.1 0.8 12.8 60.2 5.8 11.5 8.5 5.8 19.9 48.0 4.3 8.3 4.6 3.4 15.7 50.5 3.7 6.3 7.6 9.0 16.9 Japan 5.5 9.0 9.7 21.1 Australia, New Zealand, and South Africa 7.2 4.0 4.3 3.0 19.7 4.5 14.7 3.5 Middle East 1.6 1.0 2.2 1.9 Other Africa, Asia, and Pacific 9.7 1.4 6.1 2.1 Latin America Sources: See U.S. Department of Commerce, Bureau of Economic Analysis (1989b, c). tions of financial companies, which are assetintensive. By industry, manufacturing accounted for 65 percent of foreign affiliates' e m p l o y m e n t in 1987 but "only" 48 percent of foreign companies' U.S. affiliates' workers were in the factory sector. Retail trade accounted for a much higher share of U.S. affiliate employment, and finance was a much bigger component of U.S. affiliates' assets. Foreigners, particularly from other d e v e l o p e d countries, a p p e a r strongly attracted both to the large and affluent U.S. consumer market—with its efficient distribution network—and to its financial services industry. Some other noteworthy contrasts (not shown in the accompanying tables) appear when comparing these investment shares. As might b e expected on the basis of differences in country wealth, nations like Italy, Spain, Brazil, Mexico, and many in Asia and the Pacific are more likely to host U.S. investment than to have companies with large interests in the United States. In addition, U.S. foreign investments in developing countries, where wage rates are lower, make considerable use of labor, whereas activity in 46 advanced countries tends to b e in more capitalintensive industries. C o m p a r i n g Foreign a n d U.S. Patterns. More detailed analysis of foreign investment data can help determine whether the investment patterns of U.S. and foreign multinational corporations are similar or different. Table 4 shows concentration ratios for manufacturing industries. These ratios are c o m p u t e d by dividing each industry's share of total affiliate employm e n t by t h e overall U.S. share in t h e s a m e industry. An industryvaluegreaterthan o n e suggests a preference for the industry. Foreign investment in the United States is concentrated in resource-intensive manufacturing industries, whereas U.S. investment abroad appears to b e concentrated in technology-intensive industries. Specifically, U.S. firms' foreign affiliates' concentration ratios exceed o n e for chemicals; nonelectrical, electronic, and electrical machinery; transportation equipment; instruments; a n d rubber a n d plastic products. Of these industries, all but rubber can b e classified as technology-intensive. By contrast, in addition to their concentration in chemicals, foreign comECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Table 3. Shares of Assets and Employment for U.S. and Foreign Affiliates by Industry, 1987 Employment Assets U.S. Companies' Foreign Companies' U.S. Companies' Foreign Companies' Foreign Affiliates U.S. Affiliates Foreign Affiliates U.S. Affiliates Mining 1.6 0.8 1.4 Petroleum 4.7 3.7 17.9 8.7 65.4 6.5 9.3 4.2 19.3 26.1 48.0 4.6 12.2 5.0 10.2 16.0 38.8 3.2 8.1 2.4 10.4 14.7 23.6 2.5 8.2 2.5 3.5 7.0 Wholesale trade 7.9 9.9 9.1 10.5 Retail trade 8.7 18.0 1.4 2.9 Finance, except banking, insurance, and real estate 2.5 6.5 26.0 48.0 Agriculture 1.5 0.4 0.2 0.3 Manufacturing Food and kindred products Chemicals and allied products Primary and fabricated metals Machinery Other 1.3 Construction 0.8 1.3 0.4 0.4 Transportation, communication, and utilities 1.4 2.9 1.8 1.0 Services 5.6 8.5 2.9 3.2 Sources: See U.S. Department of Commerce, Bureau of Economic Analysis (1989b, c). panies' U.S. affiliates have e m p l o y m e n t concentrations in s t o n e , clay, a n d glass products; primary metals; and petroleum and c o a l industries that are basically resource-intensive. Impacts of Foreign Investment O n e would h o p e that foreign investment, like domestic investment, promotes economic growth, enhances productivity, a n d bolsters the competitiveness of U.S. industry. Newspaper accounts have reported anecdotally on s o m e successes of individual foreign investments, such as the revival of m o r i b u n d U.S. tire companies a n d the reinvigoration of parts of the a u t o m o b i l e manufacturing industry. 3 Besides bringing in new money a n d possibly increasing net investment and growth, foreign investors also have introduced new technology, such as process engineering, improved quality control FEDERAL RESERVE BANK OF ATLANTA m e t h o d s , a n d a c q u a i n t e d m a n a g e m e n t with innovative approaches, such as just-in-time inventory systems and quality circles. Unfortunately, these impacts of foreign investment cannot b e quantified systematically. Determining precisely how many U.S. workers' jobs are attributable to foreign investment is not even possible, although direct investment in manufacturing has u n d o u b t e d l y a d d e d jobs in s o m e industries and kept j o b losses down in others. 4 E m p l o y m e n t impacts are concentrated in manufacturing since U.S. affiliate employment is concentrated there (see Table 2). Among individual manufacturing industries, U.S. affiliates' shares are above total U.S. shares except in textiles, apparel, lumber, furniture and fixtures, rubber, and transportation e q u i p m e n t . U.S. affiliates appear to b e good employers. Compensation per worker at U.S. affiliates inc r e a s e d at an above-average rate in m o s t manufacturing industries compared tocompenII Table 4. Concentration Ratios for Manufacturing Assets and Sales of U.S. and Foreign Affiliates, 1986* Concentration Ratio for Assets Concentration Ratio for Sales Foreign Companies' U.S. Companies' Foreign Companies' U.S. Companies' U.S Affiliates Foreign Affiliates U.S Affiliates Foreign Affiliates Chemicals and allied products Stone, clay, and glass products Primary metal Petroleum and coal Printing and publishing 2.69 1.89 1.70 1.24 0.97 1.95 0.40 0.80 1.01 0.10 2.98 2.05 1.92 1.38 0.81 1.93 0.34 0.56 1.85 0.11 Electrical and electronic equipment Food and kindred products Paper and allied products Fabricated metal Instruments and related products 0.94 0.79 0.67 0.63 0.53 1.28 0.73 0.84 0.77 1.30 1.19 0.69 0.72 0.64 0.58 1.17 0.69 0.74 0.60 1.25 Machinery, except electrical Textile products Rubber and plastics products Transportation equipment Other 0.46 0.37 0.33 0.24 0.46 1.51 0.44 1.40 1.55 0.89 0.62 0.34 0.37 0.35 0.33 1.75 0.24 1.08 1.70 0.55 * Concentration businesses ratios in an are affiliate export shares of industry sales divided by comparable export shares for all U.S. industry. Sources: See Howenstine (1988): 59-75, and U.S. Department of Commerce, Bureau of Economic Analysis (1988a, b). sation increases for all firms in the s a m e industries in the 1977-86 period. Moreover, affiliate compensation per worker was in 1977 already higher in a majority of manufacturing industries, including t h e i m p o r t a n t c h e m i c a l s industry. Compensation per worker for affiliates was also higher in 1986 in every other major e m p l o y m e n t category (see Table 5) c o m p a r e d t o average compensation for all U.S. firms in those industries. However, between 1977 and 1986 compensation per worker grew more slowly in U.S. affiliates than for all U.S. firms in retail trade, agriculture, transportation, communication and utilities, and services, narrowing compensation differentials, though they still favor affiliates. It is tempting to infer from t h e foregoing information that foreign investment has improved the c o m p e n s a t i o n of workers at U.S. affiliates compared to all workers in the s a m e industry. However, this conclusion m u s t b e q u a l i f i e d . Previous research by the author has shown that foreign direct investment in the United States tends to b e attracted t o industries p o p u l a t e d 48 by large firms; other researchers have established that such firms t e n d to pay higher wages a n d offer greater fringe benefits for workers with the s a m e skills, experience, and occupations. W h e n adjusted for these factors, t h e higher a n d faster-growing worker compensation observed for U.S. affiliates p r o b a b l y reflects firm a n d industry size differences in the mix of affiliate versus domestic firms.5 This finding is consistent with an earlier analysis by the U.S. Commerce D e p a r t m e n t ' s Bureau of Economic Analysis. That u n p u b l i s h e d study, based on 1980 data covering hourly wages of production workers at U.S. affiliates a n d all businesses in an industry, concluded that there was n o evidence that industrial wage rates for U.S. affiliates a n d all businesses were different. Also, the U.S. General Accounting Office has compared wages for employees of U.S. affiliates a n d U.S. automakers a n d has concluded that wages received were c o m p a r a b l e for t h e two groups. Proponents of foreign investment often assert that it p r o m o t e s exports. As a start to ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Table 5. Employee Compensation per Worker for U.S. Affiliates and All U.S. Businesses by Industry, 1986 Annual Employee Compensation per Worker Foreign Companies' U.S. Affiliates 1986 Mining Percent Change 1977-86 Total U.S. 1986 Percent Change 1977-86 1986 Ratio of Compensation by U.S. Affiliates to Compensation by all Employers $42,849 144.0 $28,899 74.3 148.3 Manufacturing 33,513 106.5 24,441 78.7 137.1 Wholesale trade 32,656 97.0 24,147 75.5 135.2 Retail trade 13,602 37.7 10,988 57.3 123.8 Finance, insurance, and real estate 50,643 227.1 24,857 108.7 203.7 Agriculture 15,949 26.7 13,819 58.7 115.4 Construction 29,855 122.6 22,430 48.4 133.1 Transportation, communication, and utilities 32,213 64.4 26,150 71.9 123.2 Services 18,655 73.0 16,627 81.0 112.2 Sources: See U.S. Department of Commerce, Bureau of the Census (1981, 1988) and U.S. Department of Commerce, Bureau of Economic Analysis (1985, 1988a). analyzing this assertion, o n e should answer two questions: are U.S. affiliates concentrated in industries that tend to export, and d o e s foreign investment s t i m u l a t e exports within a given industry? The answer to t h e first question is "probably not." In t h e American manufacturing sector, nonelectrical machinery, instruments, transportation e q u i p m e n t , a n d chemicals are the only industries for which exports amount to 10 percent or more of the value of U.S. sales (see Table 6). Of these industries, foreign investment is concentrated only in chemicals, and, as the concentrations depicted in the last column of Table 6 show, U.S. affiliates of foreign chemical companies are less likely to export than are domestic chemical companies. Industries in which U.S. affiliates had aboveaverage ratios of exports to sales in 1986 relative to all U.S. firms in that s a m e industry included only primary metals, printing a n d publishing, and petroleum a n d coal. Overall, U.S. manufacturing affiliates were less likely to export than were U.S.-owned manufacturing firms in 1977 and 1986, and thus the boost to U.S. exports supposedly given by affiliates may not exist. Moreover, t h e concentration ratios suggest that affiliate manufacturers t e n d e d to b e less likely to export in 1986 than they were in 1977 compared to their domestic counterparts. 6 Explanations for these export patterns are not immediately apparent. However, if foreign companies are attracted to producing in the United States primarily to gain access to t h e U.S. market, affiliate manufacturers might b e less likely to export than U.S.-owned manufacturers. Ad hoc explanations for the particular industries' tendencies probably exist also. Unfortunately, lack of detailed information about the ownership and product composition of affiliates in the various industries precludes discussion of such explanations here. The foregoing analysis suggests that job and worker income growth may have benefited from foreign direct investment. On the other hand, U.S. affiliates' export-generating benefits d o not appear positive compared to those for domestic producers. II FEDERAL RESERVE BANK OF ATLANTA Table 6. Exports of U.S. Affiliates and All U.S. Manufacturing Industries (shares of sales, in percent) Foreign Companies' U.S. Affiliates Manufacturing Concentration Ratios Total U.S. 1977 1986 1977 1986 1977 1986 5.3 5.7 6.5 7.2 0.83 0.80 1.20 0.72 1.68 1.51 0.58 0.87 1.38 0.95 4.5 6.1 5.6 8.7 2.0 8.8 5.5 4.3 3.5 8.4 3.4 5.7 3.5 10.2 4.0 4.5 15.5 7.4 11.9 2.4 9.9 4.2 4.2 1.5 16.0 8.0 4.6 1.4 16.1 9.4 5.4 1.2 0.97 0.93 2.59 1.69 0.62 0.98 0.77 1.29 Rubber and plastics products Stone, clay, and glass products Transportation equipment Instruments and related products 1.9 1.7 9.0 9.1 3.3 0.8 10.0 7.5 4.3 3.1 11.0 14.0 4.9 2.9 11.2 13.5 0.43 0.56 0.81 0.66 0.67 0.26 0.89 0.56 Petroleum and coal* Other (textile, tobacco, leather, apparel, lumber and furniture products, and miscellaneous) 4.0 3.7 0.7 1.4 5.82 2.70 12.6 6.1 5.2 5.5 2.41 1.11 Food and kindred products Chemicals and allied products Primary metals Fabricated metals Machinery except electrical Electrical and electronic equipment Paper and allied products Printing and publishing * U.S. affiliates' category exports in total U.S. in the petroleum category are assumed to correspond to the petroleum and coal products exports. Sources: See U.S. Department of Commerce, Bureau of the Census (1981, 1988) and U.S. Department of Commerce, Bureau of Economic Analysis (1985,1988a). Foreign Investment in the Southeast Foreign investment's impact in the Southeast also interests analysts and others: Where have the investments b e e n m a d e , a n d in what industry? What are the impacts of such activity? As with t h e nation, t h e task of describing where investment has occurred in t h e region is fairly s i m p l e compared to gauging its economic impacts. Nevertheless, previous analysis by t h e author of factors motivating foreigners to invest in particular industries a n d geographic areas has generated s o m e information to help understand t h e regional impacts of this activity. From these studies of the region, o n e can conclude that the Southeast has attracted an especially hefty share of such investment and that the region's lures have b e e n favorable business a n d meteorological climates, above-average economic growth, and plentiful profit oppor- 50 tunities, owing in part to the availability of lowcost resources such as labor a n d energy. 7 Comparing aspects of foreign investment in the region with foreign investment nationally using e m p l o y m e n t data for t h e 1977-87 period confirms the special favor t h e Southeast enjoys with foreign investors. Employment growth for all U.S. affiliates d u r i n g this 10-year s p a n was 159 percent, yet regional affiliate e m p l o y m e n t growth was 239 percent (see Table 7). This disparity was a b o u t as large as the e m p l o y m e n t growth difference between the region a n d the nation over the s a m e period. The fast-paced growth of U.S. affiliate e m p l o y m e n t e n a b l e d the region's share of total U.S. affiliate e m p l o y m e n t t o rise from 10 percent in 1977 t o m o r e than 13 percent 10 years later. If regional a n d national growth rates of U.S. affiliates are compared using b o o k values of assets, a somewhat different a n d surprising picture is revealed. The nation displays si ightly fastECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 er growth over the 1977-87 period: 418 percent compared to 378 percent for the region. The disparity between e m p l o y m e n t and asset-value comparisons may b e related to differences in the industry mix of foreign investment at the national and regional levels. The region's foreign investment differs from the nation's in other ways. To a greater extent than in t h e United States as a whole, foreign investment in t h e S o u t h e a s t has t e n d e d t o entail building new manufacturing plants and boosting employment-intensive service sector industries such as wholesale a n d retail trade. Fifteen percent of t h e value of foreign investment in the Southeast in 1986 was for new plants, compared to less than 6 percent in the nation. 8 Merger and acquisition investments, often involving large capital-intensive U.S. companies, accounted for 61 percent of foreign investment in the region a n d 63 percent in the nation. The region continues to b e an active host to foreign investment in a wide spectrum of industries and from investors all around the globe. A relatively high share of direct investment in the Southeast has taken the form of new construction in nontraditional industries. Between 1977 and 1987 affiliate e m p l o y m e n t growth in t h e region e x c e e d e d c o m p a r a b l e growth rates for t h e nation in all industries except chemicals a n d real estate. In the chemical industry, regional a n d national growth rates were e q u a l . In real estate, national affiliate employment growth was more than one-third faster than affiliate growth in t h e region (see Table 8). Employment shares within the region changed somewhat during this span. The biggest shifts b e t w e e n 1977 a n d 1986 were a decline in manufacturing's share of affiliate employment to 45 percent from 60 percent in 1977, a rise in retail trade's share from 9 percent to 18 percent, a n d a rise in the share of other services to 12 percent (from just 0.1 percent). For asset values, industrial affiliate shares within t h e region shifted similarly. Reliable data are not available to summarize the impact of e m p l o y m e n t and asset shifts on regional worker compensation levels or on diversifying the southeastern economy. However, growth of foreign investor interest in t h e region clearly has shifted toward employmentintensive industries such as services and retail trade. If regional shifts in affiliate versus total Table 7. U.S. Affiliate Employment and Growth, 1977-87 Total Affiliate Employment Amount Alabama Florida Georgia Louisiana Mississippi Tennessee Southeast United States - Percent Change 1977 1987 14,313 28,250 30,693 18,367 5,734 26,215 35,100 116,800 117,700 50,800 17,600 80,700 145.2 313.5 283.5 176.6 206.9 207.8 123,572 418,700 238.8 1,218,711 3,159,700 159.3 Sources: See U.S. Department of Commerce, Bureau of Economic Analysis (1985, 1989c). compensation levels by industry followed the national pattern in the 1977-87 period, s o m e relative shift p r o b a b l y occurred o u t of t h e generally higher-paying jobs in capital-intensive industries and other manufacturing jobs into lower-paying service sector jobs. However, this shift may also have hastened growth in affiliate e m p l o y m e n t because the service sector is more labor-intensive. Differences in national and regional growth rates of employment and assets by industry caused a fewsignificant shifts in the Southeast's industry concentration ratios in t h e 1977-86 period. Most importantly, t h e region's concentration in chemicals disappeared, while specialties developed in metals, machinery, and "other" manufacturing industries. In services, an above-average concentration emerged in retail trade, reflecting t h e especially fast growth of t h e southeastern market. Shifts in concentration of b o o k values of industry assets showed similar patterns of change. In 1987 t h e regional affiliate employment distribution by country of origin of foreign direct investment in the Southeast was similar t o the nation's (see Table 9). Latin American and Midd l e East investors favored the Southeast relative to the rest of the country. Though Japanese investment is often regarded as more widespread in the Southeast than elsewhere, in fact t h e concentration of Japanese investment is less in the region than in the nation. However, II FEDERAL RESERVE BANK OF ATLANTA Table 8. Foreign Companies' U.S. Affiliate Employment and Growth by Industry, 1977-87 Southeast United States Percentage Change in Employment, 1977-87 Employment Share, 1987 (in percent) Employment Share, 1987 (in percent) Percentage Change in Employment, 1977-87 0.9 88.7 Mining 0.6 Petroleum 4.1 169.5 3.9 35.7 45.2 3.9 10.3 5.1 11.2 12.9 153.9 175.5 92.2 208.5 215.4 215.6 44.3 5.1 12.0 4.9 10.3 12.0 104.1 124.3 91.8 81.2 103.3 122.2 7.9 145.5 9.6 99.1 20.5 665.5 18.3 307.7 Finance 0.6 770.8 1.7 446.0 Insurance 2.6 371.2 2.3 122.3 Real estate 1.6 243.0 1.1 330.8 11.7 351.3 100.0 159.3 Manufacturing Food Chemicals Metals Machinery Other Wholesale trade Retail trade Other 12.4 Total 100.0 Components individual do not add to totals because — — 238.8 some firms. The effect of data suppression Also, meaningful calculation of percentage detailed data are not published may be to lower the calculated changes in employment to prevent shares in mining disclosure for the Southeast and other industries of information on in a few instances. was not possible. Sources: See U.S. Department of Commerce, Bureau of Economic Analysis (1985,1989c). Japan's southeastern presence has grown rapidly from a substantial base. Japanese affiliates' e m p l o y m e n t has e x p a n d e d at a b o u t two times t h e national pace. Assets and sales of Japanese companies' U.S. affiliates have also grown at a substantially faster rate than comparable indicators for other U.S. trading partners. By year's e n d 1987 m o r e t h a n o n e j o b o u t of ten at Japanese U.S. subsidiaries was in this region. 9 The major shifts in t h e Southeast's affiliate shares by country of origin in the 1977-87 period included sharp increases in e m p l o y m e n t shares for C a n a d a , Japan, Australia, New Z e a l a n d , South Africa, and t h e M i d d l e East, and a drop in share and concentration for Latin America. The latter is u n d o u b t e d l y related to d e b t p r o b l e m s in Latin American countries. Industry and country-of-origin specializations are, of course, linked; for example, European 52 chemical producers own chemical plants across the region. Similarly, t h e Southeast has b e e n a favorite place for foreign ownership of agricultural land (see Table 10). Much of that ownership is in the region's vast forests, and several of the biggest foreign owners are headquartered in Canada. Currently, Japanese investment interest in Florida citrus c r o p l a n d a n d A l a b a m a ranchland is growing, and Japanese activity in t h e a u t o m o b i l e assembly and parts supplying industries is large, particularly in Tennessee. Emerging Trends and Future Prospects As shown in this article, foreign investment patterns are exerting a discernible influence on ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Table 9. Foreign Companies' U.S. Affiliate Employment by Country, 1987 Southeast Number of Employees Canada United States Employment Share (in percent) Number of Employees Employment Share (in percent) Southeast Share of U.S. Southeast Concentration Ratio 85,500 20.4 590,500 18.7 14.5 1.09 234,100 25,800 30,600 39,300 17,300 82,800 55.9 6.2 7.3 9.4 4.1 19.8 1,903,700 183,600 363,300 269,500 183,400 630,100 60.2 5.8 11.5 8.5 5.8 19.9 12.3 14.1 8.4 14.6 9.4 13.1 0.93 1.06 0.64 1.10 0.71 0.99 Japan 29,300 7.0 284,600 9.0 10.3 0.78 Other Asia and Pacific 19,500 4.7 149,500 4.7 13.0 0.98 Latin America 34,400 8.2 143,600 4.5 24.0 1.81 16.0 1.21 Total Europe France Germany Netherlands Switzerland United Kingdom Middle East 5,200 1.2 32,500 1.0 Africa 1,900 0.5 19,900 0.6 9.5 0.72 United States 3,800 0.9 35,500 1.1 10.7 0.81 418,700 100.0 3,159,700 100.0 13.3 All Countries Components The effect do not add to totals because of data suppression some detailed may be to lower data are not published the calculated shares to prevent for the Southeast — disclosure on individual in a few instances. firms. Source: See U.S. Department of Commerce, Bureau of Economic Analysis (1989c). the e c o n o m i c l a n d s c a p e of t h e nation a n d t h e S o u t h e a s t , e v e n if their i m p a c t is n o t easily specified. T h e current t r e n d toward globalization of markets suggests that multinational corporations will b e as p r o m i n e n t d u r i n g t h e 1990s Table 10. Land Owned and Mineral Rights Leased or Owned by Foreigners, 1987 as in t h e d e c a d e n o w concluding. T h e con- (thousands of acres) ditions that have drawn foreign investment t o the S o u t h e a s t d o n o t s e e m to b e d i s a p p e a r i n g . Foreign investors c o m i n g to t h e United States Acres of Land Owned will p r o b a b l y c o n t i n u e t o favor t h i s region. However, several t r e n d s are e m e r g i n g that might dramatically affect future investment flows from a b r o a d t o t h e U n i t e d S t a t e s a n d the Southeast, a n d vice versa. The S o u t h e a s t has b e e n a large a n d fastgrowing s e g m e n t of t h e U.S. market. This has h e l p e d draw foreign c o m p a n i e s as they have l o o k e d to l o c a t e s u b s i d i a r y p l a n t s . Foreign investors w h o have b u i l t n e w p l a n t s in t h e re- Alabama Florida Georgia Louisiana Mississippi Tennessee Southeast United States Acres of Mineral Rights Leased or Owned 625 893 709 720 369 108 405 737 70 889 593 98 3,424 2,792 13,829 42,531 gion as well as governors of southeastern states (who participate in t h e g r o u n d b r e a k i n g cerem o n i e s for m a n y of t h e s e plants) have pro- Source: See U.S. Department of Commerce, Bureau of Economic Analysis (1989c). II FEDERAL RESERVE BANK OF ATLANTA claimed the importance of regional growth in attracting investment to the region. Future Foreign Investment. Several international economic and political developments suggest that inward and outward foreign investment will continue to grow in the 1990s. • The six-year-old Caribbean Basin Initiative was d e s i g n e d to h e l p Caribbean nations d e v e l o p their e c o n o m i e s by giving t h e m duty-free access to the U.S. market and to encourage American businesses to invest in that region. Now, most U.S. firms can benefit from low labor costs there by operating subsidiaries in the region and exporting to the United States duty-free. Although the apparel industry is excluded from favorable treatment, the very low cost of labor and the requirement to pay duty only on the value a d d e d abroad has encouraged growing numbers of U.S. apparel manufacturers to establish operations in the Basin. • The Mexican government a n n o u n c e d in May a liberalization of its direct investment regulations that will permit total foreign ownership of companies with assets of up to $100 million. Mexican officials also promised to remove most restrictions on foreign investm e n t in t h e tourist industry a n d to give foreign investors access to previously restricted sectors such as glass, cement, iron, steel, and cellulose. foreigners, the standardization brought about by the initiative will enable U.S. companies to operate more freely and efficiently, thereby reducing their production and distribution costs. Removal of geographic barriers will lower transportation costs and encourage d e v e l o p m e n t of pan-European marketing efforts, while eliminating technical barriers via uniform regulations and standards should enable companies to reap economies of scale in production. • S o m e other countries also have a strong potential for absorbing U.S. foreign investment. South Korea, like Mexico, is liberalizing its investment regulations, although majority Korean ownership will still b e required in technologically advanced industries a n d others d e e m e d critical (such as those which involve large imports of raw materials for processing with a high value a d d e d ) and defense-related industries. Other developing countries in Asia, Africa, and Latin America offer investment opportunities as well, as d o some of the socialist countries, including the Soviet Union under its current policy of openness and reform. • The U.S.-Canada Free Trade Agreement, which took effect at the beginning of this year, will phase out all tariffs between the United States and Canada over the next 10 years; ensure "national" treatment so that U.S. and Canadian businesses are free of discriminatory laws at the state, provincial, or municipal level; and loosen Canadian restrictions on U.S. investment. The agreement is likely to boost growth in Canada and spur U.S. investment there. Foreign multinationals' interest in acquiring or establishing American operations also is likely to continue to b e strong. Numerous large foreign companies d o not yet have a strong direct presence here. Some may want to buy or develop U.S. enterprises that can improve their global market positions. Besides wanting to augment their manufacturing capability, foreign multinationals also are seeking access to new technology and operations that complement existing product lines or furnish a well-known brand n a m e . U.S. a n d foreign firms also are likely to enter into more partnerships and temporary deals that will increase foreign investment in the United States. All of these factors suggest that the amount of foreign investment is likely to remain high through the next decade. • The phenomenon that has come to b e known as Europe 1992 will result in a single European Community (EC) market which will replace a dozen separate national markets. The EC now boasts 320 million p e o p l e with production capacity about equal to that of the United States. If the EC market stays open to Foreign investment in the Southeast, in addition to having grown rapidly, has expanded and matured in interesting directions. Geographically, nonmetropolitan areas have increasingly b e e n affected. More and more second-tier companies, especially among Japanese firms, have recently been established to supply larger firms 54 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 with facilities here or to carve out i n d e p e n d e n t market niches. At the same time growing trade, transportation, and investment linkages between the Southeast and countries around the world are creating new investment opportunities in a broad array of industries. Even though southeastern states will probably keep benefiting from foreign investment, surveys of affiliate managers have revealed several issues of concern, some of which relate to regional shortcomings. For example, the accounting firm Peat Marwick annually compiles data on foreign-based companies with U.S. headquarters in Georgia. 10 Responses before 1988 identified the lack of quality education in Georgia as o n e of the top two concerns. (In some instances this problem was perceived by the Peat Marwick authors to b e one of image rather than a situation that actually required attention.) Moreover, inadequate labor quality and availability has moved u p in importance as a major concern in recent years. Beyond p o s s i b l e regional drawbacks lie a host of country-specific and some broader inter- national influences that could restrain growth. 11 Protectionist legislation, the tax environment, and the availability of investor incentive programs, for example, are factors of major concern to foreign investors. Summary The world economy is in the midst of a direct foreign investment surge. Outlays by foreign investors to acquire or establish U.S. businesses have risen sharply since 1977 and are now producing at record rates. Large foreign multinational corporations are seeking to expand and diversify in world markets, including especially the large, fast-growing, and stable U.S. economy. The Southeast has captured an aboveaverage share of foreign investment, particularly for new plants and activities. Newly developing investment opportunities in the United States and abroad suggest that the globalization process will continue in the 1990s. Foreign Investment from the Southeast Just how active internationally are companies headquartered in the Southeast? According to data compiled by the Conference Board, large manufacturing companies based in Alabama, Florida, Georgia, Louisiana, Mississippi, and Tennessee with activities abroad had about $33.4 billion in total sales and $5.4 billion in foreign sales in 1987. Regional companies with foreign investments and with sales greater than $100 million numbered only 32, or about 3.2 percent of all large U.S. multinational corporations. Among the southeastern states, Florida was home to the greatest number of these multinational manufacturersjust 12 firms compared to 150 headquartered in California. Based on these data, southeastern companies do not appear especially active internationally visa-vis the rest of the nation. However, the size distribution of manufacturing firms in the Southeast compared to the rest of the nation is unknown; the region may simply not be the headquarters for many large companies. Moreover, the region could be home to a significant number of companies that are active internationally, but which are small or medium-sized firms. Although 32 large southeastern-based manufacturing companies hold investments abroad, they do not necessarily have foreign plants. The data showthat the region's multinationals own 967 principal U.S. plants yet only 82 foreign plants. Many of the firms may hold licensing agreements with host country companies or simply have sales and service departments abroad. In general, the southeastern multinationals, which are distributed all over the world, tend to be concentrated in technology-intensive industries. This pattern stands in contrast to foreign investors in the region. These seem to be predominantly European, Canadian, or Japanese and tend to specialize in the resource-intensive industries. A state-by-state summation of activity abroad is presented below. Alabama. Four companies headquartered in Alabama are active internationally. These four firms have made investments abroad in the clothing and apparel, machinery, electronics, industrial II FEDERAL RESERVE BANK OF ATLANTA chemicals, concrete, and plastic products industries. Only one owns foreign plants—one in Singapore and one in the United Kingdom. Florida. All but one of Florida's 12 multinational firms have plants abroad. One manufacturer of general industrial machinery and equipment, as well as optical instruments and lenses, maintains the highest number of foreign facilities, with plants in Belgium, Canada, France, Ireland, and theUnited Kingdom. Florida's other multinational firms chiefly produce various types of machinery and equipment, electronics, plastics products, and fabricated rubber. The two most common sites for investment seem to be Canada and the United Kingdom. Georgia. Georgia has the second highest count of multinationals (10), but the largest foreign sales among the region's states. In 1987, multinational corporations headquartered in Georgia earned about $20.3 billion in sales, or three-fifths of total sales by regional multinationals. The Coca-Cola Company, with over half its $7.7 billion sales from foreign markets, has an encompassing global presence. The corporation owns at least 19 foreign plants. Its closest state rival in terms of foreign facilities is a manufacturer of fabrics and carpets with 13 foreign plants in five countries. Popular countries for foreign investment by Georgia's multinationals are Canada, the United Kingdom, and the Dominican Republic. Louisiana. With only two companies maintaining foreign affiliates, Louisiana hosts the second least number of internationally active firms among the states in the region. Two companies produce construction machinery and equipment and build and repair ships and boats. These two companies have a total of nine foreign plants in Singapore, the United Kingdom, Canada, Egypt, Indonesia, Nigeria, and the United Arab Emirates. Mississippi. Mississippi appears to be the least active internationally of the southeastern states. In 1987 the state was not credited as home to any large multinational. However, one company, described as the world's largest sound-systems manufacturer, is an example of a smaller-sized firm with international activities. The company operates 17 facilities in the state, as well as a video production studio in Los Angeles, and owns subsidiaries in Canada, England, and the Netherlands. Tennessee. Tennessee, like Alabama, has a total of four large firms with investments abroad and just two foreign plants among them. The plants are located in Canada and Mexico. The companies are involved in a variety of industries, including machinery and equipment, household furniture, drugs, detergents and cosmetics, and miscellaneous apparel and wood products. Worldwide sales are about $1.1 billion, or just 3.4 percent of total regional multinational sales. Notes 'Discussion of t h e motivations for foreign direct invest- TWO excellent articles in this vein a p p e a r e d recently in t h e York Times (see Hicks, "The Takeover of American New m e n t motivations are related to e x p e c t e d return, risk, a n d Industry" a n d "Foreign Owners Are Shaking u p t h e Com- information considerations. Moreover, certain e c o n o m i c p e t i t i o n , " May 28, 1989). T h e s e articles d e s c r i b e h o w a n d political forces may h e l p explain longer-run global foreign c o m p a n i e s have altered t h e c o m p e t i t i v e dynam- t r e n d s in t h e types a n d a m o u n t s of foreign investment ics in industries such as chemicals, b u i l d i n g materials, activity while s o m e other factors influence t h e precise tim- tires, a u t o m o b i l e s , a n d steel. ing, geographic location a m o n g a n d within countries, a n d 2 3 m e n t is b e y o n d the s c o p e of this article. Generally, invest- 4 W h e t h e r or not there is a net e m p l o y m e n t gain d e p e n d s industrial patterns of investment. For d e t a i l e d discussion u p o n t h e e x t e n t t o which U.S. affiliates r e p l a c e U.S. of t h e s e issues s e e U.S. D e p a r t m e n t of C o m m e r c e (1984) imports (or h o m e country exports) c o m p a r e d t o U.S. jobs a n d Kahley (1987). lost or d i s p l a c e d b e c a u s e of increased c o m p e t i t i o n or T h e Federal Reserve System regularly collects asset a n d b e c a u s e foreign firms' affiliates use m o r e capital at the b a l a n c e s h e e t data o n t h e b a n k i n g industry and, t o avoid e x p e n s e of labor relative t o practices of d o m e s t i c U.S. redundancy, t h e U.S. D e p a r t m e n t of C o m m e r c e a n n u a l surveys d o not collect data o n banks. However, t h e Fed d o e s not collect e m p l o y m e n t data for banks. 56 firms. ^ h e r e also are s o m e technical p r o b l e m s in c o m p a r i n g affiliate e m p l o y m e n t a n d total U.S. e m p l o y m e n t . At t h e ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 d e t a i l e d industry level, c o m p a r i s o n s of e m p l o y m e n t may 8 not b e a p p r o p r i a t e b e c a u s e of differences in industry by t h e U.S. C o m m e r c e D e p a r t m e n t ' s International Trade classification b e t w e e n U.S. affiliate a n d all U.S. business Administration. e m p l o y m e n t d a t a . T h e affiliate d a t a are classified by 9 industry at t h e enterprise or c o m p a n y level, whereas total were o p e r a t i n g in the Southeast at t h e b e g i n n i n g of 1989, m e n t level; consequently, affiliate a n d all industry com- e m p l o y i n g over 41,000 workers. The a m o u n t of Japanese pensation levels could also b e affected by an "industry investment in t h e region (excluding Louisiana) at that t i m e mix" effect. In a d d i t i o n , U.S. affiliate c o m p e n s a t i o n inc l u d e s any p a y m e n t s t o workers d u r i n g the year, while l0 1 total U.S. e m p l o y m e n t a n d compensation are as of March. 7 was e s t i m a t e d at nearly $5 billion. e m p l o y m e n t is as of the e n d of t h e year, a n d t h e data for T h e statistical finding that U.S. affiliates' likelihood of D a t a c o m p i l e d by t h e C o n s u l a t e G e n e r a l of J a p a n in Atlanta s h o w that 496 lapan-affiliated e s t a b l i s h m e n t s U.S. e m p l o y m e n t is classified by industry at t h e establish- 6 This information is b a s e d o n transactions data reported S e e KPMG Peat Marwick (1989). 'Generally, t h e current surge in foreign investment m a y reflect an a t t e m p t by foreign c o m p a n i e s to establish a presence in U.S. industries. For example, eight Japanese exporting has d r o p p e d while foreign investment has risen manufacturers are increasing capacity t o b e a b l e to pro- sharply in the 1977-86 p e r i o d m a y b e related to t h e strong d u c e a r o u n d 2 million vehicles per year here. To the extent value of the dollar in 1986 c o m p a r e d to 1977. Rather than that foreign investment represents an a t t e m p t to erase a export from the United States, foreign m u l t i n a t i o n a l cor- g a p b e t w e e n actual a n d d e s i r e d stocks, future investment p o r a t i o n s m a y h a v e s o u r c e d " e x p o r t s " in s o m e o t h e r activity can b e expected to slow as t h e stock a d j u s t m e n t country, including plants in their h o m e countries. process matures. S e e Kahley (1985, 1986, 1987). References Burnside, Yvonne. Key Company New York: The Directory. Conference Board, August 1988. Consulate General of Japan, lapanese Subsidiaries in the U.S. Southeast. Firms, Offices, of Japanese Firms General of lapan the Consulate Within and the Jurisdiction in New New Orleans. of U.S. Affiliates Orleans, La., April I, 1989. U.S. Direct Abroad: U.S. Business. Operations and Their Foreign Affiliates, of Pre- Washington, D.C.: U.S. Govern- 1986 Estimates. Survey of Current Washington, D.C: Business. U.S. G o v e r n m e n t Printing Office, July 1988c. " S t a t e Personal I n c o m e , of Current Pre- m e n t Printing Office, June 1988b. May 28, 1989, section 3, 1, 8. O p e r a t i o n s in 1986." Survey Investment U.S. Parent Companies May 28, 1989, section 3, 9. Howenstine, N e d G. "U.S. Affiliates of Foreign C o m p a n i e s : States: Companies, m e n t Printing Office, June 1988a. liminary "The Takeover of American Industry." NewYork in the United of Foreign Washington, D.C.: U.S. Govern- 1986 Estimates. of Hicks, Jonathan P. " Foreign Owners Are Shaking U p t h e Competition." New York Times, Washington, D.C.: U.S. 1977-80. Foreign Direct Investment Operations liminary Consulate General of Japan in New Orleans. Offices Times, of U.S. Affiliates, and Atlanta, G a „ S e p t e m - ber I, 1989. Subsidiaries tions G o v e r n m e n t Printing Office, 1985. E s t i m a t e s . " Survey of Current 1985-87: Revised Washington, Business. D.C.: U.S. G o v e r n m e n t Printing Office, August 1988d. D e p a r t m e n t of C o m m e r c e . Bureau of E c o n o m i c Analysis. "U.S. Direct I n v e s t m e n t A b r o a d : D e t a i l for Washington, D.C.: U.S. G o v e r n m e n t Printing Office, May Position a n d Balance of Payments Flows, 1 9 8 7 . " S u r v e y o f 1988. Current Kahley, William J. "Foreign Direct Investment - A Bonus for t h e Southeast." Federal Reserve Bank of Atlanta nomic Review Business. Eco- 70 (June/July 1985): 4-17. " U n i t e d S t a t e s D e p a r t m e n t of C o m m e r c e News." News Release BEA 89-28. Washington, D.C.: U.S. "What's B e h i n d Patterns of State Job G r o w t h ? " Federal Reserve Bank of Atlanta Economic Review G o v e r n m e n t Printing Office, June 27, 1989a. 71 (May 1986): 4-18. "Selected Data of Foreign Affiliates of U.S. C o m p a n i e s . " C o m p u t e r printouts. July 1989b. "Direct Investment Activity of Foreign Firms." Federal Reserve Bank of Atlanta Economic Review 72 (Summer 1987): 36-51. KPMG Peat Marwick. Georgia panies. Com- S e p t e m b e r 1989. Business Patterns. "Selected Data of N o n b a n k U.S. Affiliates." Photocopies. July 1989c. U.S. D e p a r t m e n t of C o m m e r c e . International Trade Admini- 's 1989 Survey of Foreign stration. International and U.S. D e p a r t m e n t of C o m m e r c e . B u r e a u of t h e C e n s u s . County Washington, D.C.: U.S. G o v e r n m e n t Printing Office, August I988e. 1978, United States. the 1981. Investment: Global Trends Printing Office, August 1984. CBP-78-1. Washington, D.C.: U.S. Government Printing Office, April Direct U.S. Role. Washington, D.C.: U.S. G o v e r n m e n t Foreign Direct Investment 1986 Transactions. in the United States: Washington, D.C.: U.S. G o v e r n m e n t Printing Office, S e p t e m b e r 1987. County Business Patterns, 1986, United States. U.S. General Accounting Office. Foreign Investment: Grow- CPB-86-1. Washington, D.C: U.S. G o v e r n m e n t Printing ing lapanese Office, O c t o b e r 1988. ington, D C.: U.S. G o v e r n m e n t Printing Office, March U.S. D e p a r t m e n t of Commerce. Bureau of Economic Analysis. Foreign Direct Investment in the United FEDERAL RESERVE BANK OF ATLANTA States: Presence in the U.S. Auto Industry. Wash- 1988. Opera- II Book Review Bank Costs, Structure, and Performance by james Kolari and Asghar Zardkoohi. Lexington, Mass.: D.C. Heath, 1987. 240 pages. $29.00. C h a n g e s in b a n k i n g regulations in recent years have created new opportunities for commercial banks and other financial institutions to expand their operations. Restrictions on interstate b a n k i n g a n d intrastate branching have b e e n liberalized in many states. In addition, legislators and regulators have relaxed many limitations formerly constraining t h e types of services financial institutions could offer. Along with these developments, though, have c o m e new questions about t h e future structure of the financial services industry. The industry's composition will d e p e n d to a great extent on the types of financial institutions that can remain profitable over time, a n d profitability will b e determined largely by the extent to which banks achieve production economies and resultant cost reductions while expanding their operations. Two types of production economies are generally available to banks—economies of scale and economies of scope. Economies of scale exist if average production costs decline as outp u t increases. Scope economies are present if two or more products can b e jointly produced at a lower cost than is incurred in their independ e n t production. In s o m e industries, such as utilities, it is efficient for a single firm to supply 58 the entire industry output. These industries are termed natural monopolies and are characterized by economies of scale at every o u t p u t level consumers are likely to d e m a n d . Recent empirical evidence suggests that t h e cond itions sufficient for natural monopoly in banking are not satisfied over the relevant range of output. Rather, overall economies of scale appear t o exist only at low levels of output, and diseconomies, at large o u t p u t levels, suggesting a U-shaped cost curve for t h e industry. 1 Against this b a c k d r o p , (ames Kolari a n d Asghar Zardkoohi—professors of e c o n o m i c s and public policy, respectively, at Texas A&M University—undertake an important and timely task. Bank Costs, Structure, and Performance provides a good introduction to the topic of cost economics in banking and an in-depth review of the pre-1985 literature on production economies. The authors reexamine t h e issue of scope e c o n o m i e s in b a n k i n g a n d i n t r o d u c e a new measure designed to estimate the extent of these economies. In contrast to past research on bank costs, Kolari a n d Zardkoohi perform separate analyses (using 1979-83 data) for banks with differing market (product) characteristics. Unfortunately, t h e researchers fail to e m p h a s i z e the qualified nature of the evidence they use to ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 drawsome of their major conclusions and policy implications. In particular, their cost complementarity and scope economy results should b e interpreted with caution. Allen N. Berger addresses these and other shortcomings in another, earlier analysis of Kolari and Zardkoohi's work. 2 In the first chapter, Kolari and Zardkoohi set the stage for their discussion of costs by describing the external and internal pressures on banking that continue to affect bank profits: increased competition from n o n b a n k institutions, the spread of interstate banking, product line deregulation, and t h e rapid pace of technological advancement in the industry. These developments have brought about a more competitive environment in which banks must operate as efficiently as p o s s i b l e . The authors' historical overview of twentieth-century banking leads them to suggest that legal and regulatory c h a n g e s have greatly influenced t h e structure of U.S. banking. Deregulation has already p u s h e d banks to b e c o m e more efficient, but its lasting effects cannot b e foreseen without s o m e idea of how bank costs behave. Kolari a n d Zardkoohi begin their study of bank costs with a review of microeconomic production theory a n d how it may b e appl ied to the special p r o b l e m s of banks. The authors also delineate sources of scale and scope economies and describe how these are measured. In their discussion of conceptual and methodological issues that arise in estimating such economies, Kolari and Zardkoohi find that t h e a p p r o p r i a t e d e f i n i t i o n of b a n k o u t p u t a n d choice of functional form are especially important. Accurate measurement of b a n k o u t p u t is necessary because economies of scale are defined in terms of t h e v o l u m e of t h e bank's output. The researchers analyze the choice of o u t p u t measure a n d c o n c l u d e that dollar values of loans and deposits, rather than n u m b e r of accounts, should b e used. Their argument is that "banks c o m p e t e to increase market share of dollar amounts as o p p o s e d to the n u m b e r of accounts . . . |and| in a c o m p e t i t i v e b a n k i n g environment the cost of an additional dollar of both small and large accounts should b e t h e same." Specification of the functional form of the cost function (and therefore the underlying production function) is also closely related to FEDERAL RESERVE BANK OF ATLANTA the measurement of scale and scope effects. The production function identifies the relationship between the quantities of o u t p u t that result from the use of various quantities of inputs. Comparing three economic production functions— the Cobb-Douglas, constant elasticity of substitution, a n d translog f u n c t i o n s - t h e authors assert that the last is the preferred form. It is flexible enough to yield U-shaped cost curves (diseconomies as well as economies of scale and scope) and it allows for banks' characteristic multiple inputs and outputs. 3 Chapter 3 presents a comprehensive survey of previous literature on bank costs, divided into three parts: (1) early studies that relied on s i m p l e financial ratios to calculate bank costs, (2) analyses from the mid-1960s and the 1970s that used the Cobb-Douglas function and specified only o n e output, and (3) more recent works that focus on the translog function. Notwithstanding the various studies' differences in methodologies, output definitions, and data sources, Kolari and Zardkoohi conclude from a review of the earlier studies that "small banks were at a cost disadvantage compared to large b a n k s b u t that t h e difference was not so large as to prevent them from competing effectively " Recent research using the translog cost model yielded results somewhat contrary to those obtained earlier: very small banks were found to b e cost-efficient for the most part. All of the studies indicate that most scale economies are exhausted when bank size reaches about $25 million in deposits a n d that diseconomies of scale exist at large o u t p u t levels, leading to the familiar U-shaped cost function. However, the evidence on scope economies was ambiguous. Even studies that found positive evidence in favor of joint p r o d u c t i o n c o n c l u d e d that scope benefits were not substantial enough to alter the scale results. In chapter 4 Kolari and Zardkoohi present the econometric results of their own research. Using Federal Reserve Functional Cost Analysis (FCA) data for 1979-83, they e s t i m a t e three m o d e l s : (1) d e m a n d deposits and time deposits, (2) loans and securities, and (3) loans and deposits. 4 The d e p e n d e n t variables are the allocated costs for the specific outputs appearing in each regression model. The two researchers find average cost curves to b e relatively flat in most cases, so scale is apparently not an important cost deterII minant. The major implication of a flat cost curve is that many different sizes of banks should b e a b l e to coexist. The authors perform jointness tests a n d find that significant cost complementarities exist only in the joint production of loans a n d deposits. Kolari a n d Zardkoohi also use their new measure of scope economies, which gauges t h e decrease in costs from producing o u t p u t jointly, as compared to expanding total o u t p u t by increasing each of t h e bank's products o n e at a t i m e (from the m i n i m u m level for b a n k s of a b o u t t h e s a m e size). O n average, Kolari and Zardkoohi find that banks can reduce expansion costs a b o u t 30 percent to 50 percent by increasing o u t p u t s at t h e s a m e time, as o p p o s e d to increasing each output separately. zero. A sufficient condition for cost complementarity requires that their cross-product term b e not only negative b u t also greater in absolute value than the product of their o u t p u t elasticities. 5 However, t h e cross-product terms reported by Kolari and Zardkoohi are positive in most cases, suggesting that cost complementarity d o e s not hold or, if it does, that negative e s t i m a t e d marginal costs are generating it. Since the authors d o not provide the level of complementarities or t h e estimated marginal costs, it is impossible t o determine which condition exists. 6 The fact that Kolari a n d Zardkoohi d o not investigate t h e scope economy results for statistical significance further detracts from their results. Several issues a n d problems, both conceptual and methodological, may have influenced the results reported in chapter 4. As a consequence, t h e usefulness of Kolari and Zardkoohi's conclusions in drawing policy implications, although not eliminated, is limited. First, the FCA data used in the analysis are heavily skewed toward small banks. As of 1986, only 490 banks participated in t h e program; o f t h i s n u m b e r , 4 1 6 held under $200 million in total deposits. To draw conclusions a b o u t t h e cost structure of large banks (over $1 billion in total deposits) b a s e d on FCA data is not meaningful and can b e misleading. Also, the FCA procedures for allocating costs are sometimes imprecise a n d may induce bias in t h e results. In chapter 5 t h e authors test t h e hypothesis that differences in product mix influence b a n k cost structures. Based on balance sheet ratios, banks are clustered into four types: farm, retail, city, and wholesale. Only farm banks were found to have u n i q u e cost characteristics. (They exhibited flat cost curves where other groups had U-shaped cost curves; they also h a d s c o p e e c o n o m i e s related to d e p o s i t size.) All four b a n k g r o u p s h a d higher scope economies in the joint production of loans and deposits than in t h e other two models. Kolari and Zardkoohi's results are puzzling, nonetheless. Several researchers who h a n d l e d differing product mixes by specifying more outputs in the cost function have rejected t h e h y p o t h e s i s that different asset a n d liability categories can b e aggregated. 7 Ideally, each bank product should b e included as a distinct output, b u t the availabil ity of data and use of the translog functional form usually limit t h e level of disaggregation. 8 A second p r o b l e m that may distort t h e results is that Kolari and Zardkoohi exclude interest payments from their cost measure. Berger, Gerald A. Hanweck, and David B. Humphrey (1987) have shown that studies using dollar measures of o u t p u t s a n d total o p e r a t i n g costs as t h e d e p e n d e n t variable are biased toward finding scale e c o n o m i e s b e c a u s e b a n k s can fund a larger asset portfolio by increasing purchased funds. Thus, Kolari and Zardkoohi's analysis is biased by a bank's choice to gather deposits through a branching network or to purchase funds from other retail banks. A third problem arises with interpreting the authors' cost complementarity and scope economy results. When t h e results from the translog cost function are used, a necessary condition for the existence of cost complementarity between two products is that their cross-product term (8, 2 ) b e negative and statistically different from 60 Kolari a n d Zardkoohi turn from static costs to the impact of technological improvements in banking a n d how they have affected production costs. To test whether larger scale allows more cost-efficient use of technology, t h e authors regress d e m a n d d e p o s i t costs on the ratio of computer-related costs to labor costs and s o m e o u t p u t variables. The closer this ratio is to zero, t h e greater the cost savings by substituting computer technology for labor. Kolari a n d Zardkoohi in fact find that t h e coefficient lies between zero and one, suggesting that cost savings result. They find n o evidence of a trend toward greater cost gains by large banks. However, whether this model is sufficiently c o m p l e t e to draw such a ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 conclusion is unclear. In particular, the authors have a very narrow view of technological change and t h e a p p r o p r i a t e costs that are affected. They ignore the possibility that technological innovation can take the form of new production processes rather than equipment. Another problem with Kolari and Zardkoohi's model is that only d e m a n d d e p o s i t costs are included in t h e d e p e n d e n t variable. 9 Finally, t h e authors examine the relation between cost efficiency a n d b a n k failure using Call Report data for 1984. They develop an earlywarning-system model based on commonly used financial ratios and individual b a n k cost measures (scale economies and residual costs) generated in the research reported earlier in t h e book. The cost measures were found t o improve the predictive power of the failure model substantially when a d d e d to financial ratios. Problems exist with the analysis, though, because Kolari a n d Zardkoohi fall into t h e trap that earlier writers did. By regressing identical operating expenses on loans a n d deposits separately, t h e authors' analysis suffers from t h e same drawbacks as t h e study by Thomas W. Gilligan a n d Michael Smirlock (1984): i n p u t prices are assumed to b e constant a n d other bank services are excluded even though they affect total operating expenses. This practice gives a bias toward finding both scale and scope FEDERAL RESERVE BANK OF ATLANTA economies a n d leads t h e m to conclude that failing banks were smaller than average. From a policy perspective, the evidence presented in the b o o k appears to minimize any concern that the banking industry will b e dominated by a few large institutions. The lifting of restrictions on interstate banking and intrastate branching might help consolidate resources in states that have limited branch banking and thereby permit small banks to achieve a more efficient scale of production. O n t h e whole, Bank Costs, Structure, and Performance is a useful g u i d e to future work in this area and is of interest to academicians, policymakers, a n d practitioners. It provides an ind e p t h look at the literature, introduces a new measure of scope economies, and o p e n s s o m e new lines of research. The book's biggest failing is the absence of necessary qualifications in regard to the econometric evidence it presents and the consequent potential to mislead readers. Aruna Srinivasan The reviewer Atlanta is an economist Fed's Research in the financial section of the Department. II Notes offered by t h e bank, a n d costs i n c l u d e b o t h interest a n d 'See, for e x a m p l e , Hunter a n d T i m m e (1989) a n d Lawrence a n d Shay (1986). o p e r a t i n g expenses. The i n t e r m e d i a t i o n a p p r o a c h uses a 2 S e e Berger (1988). broader definition of costs a n d is c o n s i d e r e d t o b e m o r e 3 T h e m a i n d i s a d v a n t a g e of t h e Cobb-Douglas production relevant for a d d r e s s i n g i s s u e s relating t o t h e long-run function is that it only allows for uniform scale characteris- viability of b a n k s (Hunter a n d T i m m e 119891). 4 tics, while t h e constant elasticity of substitution function is 5 S e e Clark (1988). highly restrictive in cases where firms p r o d u c e m o r e than 6 O t h e r s t u d i e s (Benston et al. 119831, Mester |I987|) ex- o n e o u t p u t or u s e m o r e than o n e factor input. plicitly report negative marginal costs for s o m e products, A l t h o u g h FCA data p r o v i d e information o n t h e n u m b e r of attributing t h e m t o e s t i m a t i o n p r o b l e m s such as the pres- accounts, Kolari a n d Zardkoohi prefer t o use dollar a m o u n t s e n c e of multicollinearity a n d loss of degrees of freedom. for t h e reasons m e n t i o n e d earlier. In general, researchers 7 S e e Kim (1986) a n d Lawrence a n d Shay (1986). take o n e of two a p p r o a c h e s in defining b a n k costs a n d out- 8 S e e Clark (1988). p u t : the p r o d u c t i o n a p p r o a c h or the i n t e r m e d i a t i o n ap- 9 I t is also i m p o r t a n t to n o t e that Kolari a n d Zardkoohi's con- proach (Berger, Hanweck, a n d H u m p h r e y |I987|). U n d e r clusions are l i m i t e d to t h e smaller b a n k s in t h e economy. t h e p r o d u c t i o n approach, o u t p u t is m e a s u r e d in terms of H u n t e r a n d T i m m e (1986, 1988) e x a m i n e the relation be- t h e n u m b e r of loan a n d d e p o s i t accounts, a n d costs are tween technological change, p r o d u c t i o n e c o n o m i e s , a n d d e f i n e d a s total operating e x p e n s e s exclusive of interest firm size for a s a m p l e of large b a n k s a n d find that larger costs. The i n t e r m e d i a t i o n approach, o n t h e other h a n d , b a n k s e n j o y p r o p o r t i o n a t e l y h i g h e r cost savings from m e a s u r e s o u t p u t as t h e d o l l a r v a l u e of t h e p r o d u c t s technological change. References Benston, G e o r g e )., Allen N. Berger, Gerald A. Hanweck, a n d David B. H u m p h r e y . " E c o n o m i c s of Scale a n d S c o p e . " In of Money, Credit, and Banking Com- Hunter, William C., a n d S t e p h e n G. T i m m e . "Technological C h a n g e in Large U.S. Commercial Banks." Federal Re- of a Conference on Bank Structure and serve Bank of Atlanta Working Paper 88-6 ( D e c e m b e r 432-55. Berger, Allen N. In "Book Reviews." tournai and Banking of Money, Credit, 20 (May 1988): 283-87. 1988). Hunter, William C , a n d S t e p h e n G. T i m m e . " D o e s Multi- Berger, Allen N., Gerald A. Hanweck, a n d David B. Hum- p r o d u c t P r o d u c t i o n in Large Banks R e d u c e C o s t s ? " phrey. " C o m p e t i t i v e Viability in Banking: Scale, Scope, Federal Reserve Bank of Atlanta Economic a n d Product Mix Economies." Journal (May/June 1989): 2-11. nomics 18 (May 1986): 152-66. Federal Reserve Bank of Chicago (May 1983): Proceedings petition. d u c t i o n . " /ournal of Monetary Eco- 20 ( D e c e m b e r 1987): 501-20. pository Financial Institutions." Federal Reserve Bank of Review 74 Kim, Moshe. "Banking Technology a n d t h e Existence of a Clark, leffrey A. " E c o n o m i e s of Scale a n d S c o p e at DeK a n s a s City Economic Review (September/October 1988): 16-33. C o n s i s t e n t O u t p u t A g g r e g a t e . " / o u r n a l of Economics Monetary 18 ( S e p t e m b e r 1986): 181-95. Lawrence, Colin, a n d Robert Shay. "Technology a n d Financial I n t e r m e d i a t i o n in M u l t i p r o d u c t Banking Firms: An Gilligan, T h o m a s W., a n d Michael Smirlock. "An Empirical Econometric Study of U.S. Banks, 1979-1982." In Tech- Study of Joint Production a n d Scale E c o n o m i e s in Com- nological mercial B a n k i n g . " tournai 8 my, e d i t e d by Lawrence a n d Shay. C a m b r i d g e , Mass.: Hunter, W i l l i a m C , a n d S t e p h e n G. T i m m e . "Technical Mester, Loretta). "A M u l t i p r o d u c t Cost Study of Savings a n d of Banking and Finance (March 1984): 67-77. Change, Organizational Form, a n d Structure of Bank Pro- 62 Innovation, Regulation and the Monetary Econo- Ballenger, 1986. Loans." / o u r n a l of Finance 42 (June 1987): 423-45. ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Working Paper Series Available T h e Research Department of the Federal Reserve Bank of Atlanta publishes a working paper series to stimulate professional discussion and exploration of economic subjects. We welcome readers of the Economic Review to complete and return the form below in order to receive our recently released working papers. If you would like a copy of any of the current papers, simply check the box next to the paper's number and return the form to the Public Information Department, Federal Reserve Bank of Atlanta. 89-1 PeterAAbken A Survey and Analysis 89-2 of Index-Linked Certificates of Deposit Aruna Srinivasan Costs of Financial Development 89-3 Intermediation under Banks and Commercial Regulation: Banks W i l l i a m C. Hunter, S t e p h e n G. T i m m e , a n d Won Keun Yang An Examination 89-4 of Cost Subadditivity and MuItiproduct Production in Large U.S. Banks Preston J. Miller a n d William R o b e r d s How Little 89-5 We Know about Budget Policy Effects Peter A. Zadrozny Analytic Derivatives for Estimation 89-6 of the Matrix of Continuous-Time Exponential ARMA Models William C. H u n t e r a n d S t e p h e n G. T i m m e The Demand for Labor at the World's Largest 89-7 Banking Organizations Ellis W. Tall m a n Macroeconomic 89-8 Factors and Asset Excess Returns Joseph A. Whitt, Jr. Nominal 89-9 Exchange Rates and Unit Roots: A Reconsideration Larry D. Wall a n d David R. Peterson The Effect of Continental Illinois' Failure on the Financial Performance of Other Banks Previous working papers are also available. For more information, contact the Public Information Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, NW, Atlanta, Georgia 30303-2713 (404/521-8788). • Please start my subscription to the Working Paper Series. | | Please send me a copy of the following working papers: • 89-1 Q89-2 • 8 9 - 3 • 8 9 - 4 089-5 • 8 9 - 6 • 8 9 - 7 Q89-8 Q89-9 Name Address City State ZIP Return to the Public Information Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, NW, Atlanta, GA 30303-2713 J Economic Review BOOK REVIEWS Bank Costs, Structure, and Performance by lames Kolari and Asghar Zardkoohi (Lexington, Mass.: D.C. Heath, 1987) Aruna Srinivasan, November/December, 58 Breaking the Bank: The Decline of BankAmerica by Gary Hector (Boston: Little, Brown, 1988) B. Frank King and Sheila L. Tschinkel, September/October, 48 Breaking Up the Bank: Rethinking an Industry under Seige by Lowell L. Bryan (Homewood, III.: Dow Jones-Irwin, 1988) B. Frank King and Sheila L. Tschinkel, September/October, 48 The Cold Standard and the International Monetary System 1900-1939 by Ian M. Drummond (London: MacMillan Education Ltd., 1987) Thomas J. Cunningham, March/April, 50 Migration and Residential Mobility in the United States by Larry Long (New York: Russell Sage Foundation, 1988) William J. Kahley, May/June, 52 Memoirs of an Unregulated Economist by George J. Stigler (New York: Basic Books, 1988) Mary Susan Rosenbaum, July/August, 48 The Netherlands and the Gold Standard, 19311936: A Study in Policy Formation and Policy edited by Richard T. Griffiths (Amsterdam: Nederlandsch Economisch-Historisch Archief, 1987) Thomas J. Cunningham, March/April, 50 CORPORATE FINANCE "Bank Merger Motivations: A Review of the Evidence and an Examination of Key Target Bank Characteristics" William C. Hunter and Larry D. Wall, September/October, 2 "Capital Requirements for Banks: A Look at the 1981 and 1988 Standards" Larry D. Wall, March/April, 14 64 "Financial Asset Pricing Theory: A Review of Recent Developments" Ellis W. Tallman, November/December, 26 "Forecasting Stock-Market Volatility Using Options on Index Futures" Steven P. Feinstein, May/June, 12 "Interest-Rate Caps, Collars, and Floors" Peter A. Abken, November/December, 2 "A Plan for Reducing Future Deposit Insurance Losses: Puttable Subordinated Debt" Larry D. Wall, |uly/August, 2 FEDERAL DEFICIT "The Federal Budget Deficit and the Social Security Surplus" Thomas J. Cunningham, March/April, 2 FINANCIAL INSTITUTIONS "Bank Merger Motivations: A Review of the Evidence and an Examination of Key Target Bank Characteristics" William C. Hunter and Larry D. Wall, September/October, 2 Book Review: Bank Costs, Structure, and Performance by lames Kolari and Asghar Zardkoohi Aruna Srinivasan, November/December, 58 Book Review: Breaking the Bank: The Decline BankAmerica by Gary Hector B. Frank King and Sheila L. Tschinkel, September/October, 48 of Book Review: Breaking Up the Bank: Rethinking Industry under Seige by Lowell L. Bryan B. Frank King and Sheila L. Tschinkel, September/October, 48 an "Capital Requirements for Banks: A Look at the 1981 and 1988 Standards" Larry D. Wall, March/April, 14 "Commercial Bank Profitability: Improved in 1988" Robert E. Goudreau and David D. Whitehead, )uly/August, 34 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 Index for 1989 "Does Multiproduct Production in Large Banks Reduce Costs?" William Curt Hunter and Stephen G. Timme, May/June, 2 "Interstate Banking Developments in the 1980s" B. Frank King, Sheila L. Tschinkel, and David D. Whitehead, May/June, 32 INTERSTATE BANKING "Interstate Banking Developments in the 1980s" B. Frank King, Sheila L. Tschinkel, and David D. Whitehead, May/June, 32 MACROECONOMIC POLICY "A Plan for Reducing Future Deposit Insurance Losses: Puttable Subordinated Debt" Larry D. Wall, July/August, 2 Book Review: The Gold Standard and the International Monetary System 1900-1939 by Ian M. Drummond Thomas J. Cunningham, March/April, 50 FOREIGN INVESTMENT IN THE UNITED STATES Book Review: The Netherlands and the Gold Standard, 1931-1936: A Study in Policy Formation and Policy edited by Richard T. Griffiths Thomas J. Cunningham, March/April, 50 "U.S. and Foreign Direct Investment Patterns" William J. Kahley, November/December, 42 GOLD STANDARD Book Review: The Gold Standard and the national Monetary System 1900-1939 by Ian M. Drummond Inter- "The Federal Budget Deficit and the Social Security Surplus" Thomas J. Cunningham, March/April, 2 "Money and the Economy: Puzzles from the 1980s' Experience" William Roberds, September/October, 20 Thomas J. Cunningham, March/April, 50 Book Review: The Netherlands and the Gold Standard, 1931-1936: A Study in Policy Formation and Policy edited by Richard T. Griffiths Thomas j. Cunningham, March/April, 50 INTERNATIONAL ECONOMICS Book Review: The Cold Standard and the International Monetary System 1900-1939 by Ian M. Drummond Thomas J. Cunningham, March/April, 50 "Purchasing-Power Parity and Exchange Rates in the Long Run" Joseph A. Whitt, Jr., July/August, 18 "U.S. and Foreign Direct Investment Patterns" William J. Kahley, November/December, 42 MANUFACTURING "Southeastern Manufacturing: Recent Changes and Prospects" Gene D. Sullivan and David Avery, january/ February, 2 MIGRATION Book Review: Migration and Residential in the United States by Larry Long William J. Kahley, May/June, 52 Mobility "Interregional Migration: Boon or Bane for the South" William J. Kahley, January/February, 18 MONEY SUPPLY "Money and the Economy: Puzzles from the 1980s' Experience" William Roberds, September/October, 20 II FEDERAL RESERVE BANK OF ATLANTA POVERTY "Poverty in the South" jon R. Moen, January/February, 36 PUBLIC FINANCE "Measuring State and Local Fiscal Capacities in the Southeast" Aruna Srinivasan, September/October, 36 "Public Finance and Economic Growth in the Southeast" Aruna Srinivasan, January/February, 48 REGIONAL ECONOMICS "Interregional Migration: Boon or Bane for the South" William J. Kahley, lanuary/February, 18 "Measuring State and Local Fiscal Capacities in the Southeast" Aruna Srinivasan, September/October, 36 "Poverty in the South" Jon R. Moen, January/February, 36 "Public Finance and Economic Growth in the Southeast" Aruna Srinivasan, January/February, 48 "Southeastern Manufacturing: Recent Changes and Prospects" Gene D. Sullivan and David Avery, January/ February, 2 "U.S. and Foreign Direct Investment Patterns" William J. Kahley, November/December, 42 SECURITIES "Financial Asset Pricing Theory: A Review of Recent Developments" Ellis W. Tall man, November/December, 26 66 "Forecasting Stock-Market Volatility Using Options on Index Futures" Steven P. Feinstein, May/June, 12 "Interest-Rate Caps, Collars, and Floors" Peter A. Abken, November/December, 2 SOCIAL SECURITY "The Federal Budget Deficit and the Social Security Surplus" Thomas J. Cunningham, March/April, 2 STOCK MARKET "Financial Asset Pricing Theory: A Review of Recent Developments" Ellis W. Tallman, November/December, 26 "Forecasting Stock-Market Volatility Using Options on Index Futures" Steven P. Feinstein, May/June, 12 U.S. ECONOMY Book Review: Migration and Residential in the United States by Larry Long William J. Kahley, May/June, 52 Mobility "The Federal Budget Deficit and the Social Security Surplus" Thomas J. Cunningham, March/April, 2 "Money and the Economy: Puzzles from the 1980s' Experience" William Roberds, September/October, 20 "Tracking the Economy: Fundamentals for Understanding Data" R. Mark Rogers: March/April, 30 "U.S. and Foreign Direct Investment Patterns" William J. Kahley, November/December, 42 ECONOMIC REVIEW, NOVEMBER/DECEMBER 1989 H I » mmsmm • > , * , iMämmMimiimmmmm Iii II S S « ' » I i i mwm m MM m mgrnxasmm M ¡ismm m m . • o-- âêêÊé ' f\ Atlanta Fed History Now Available In conjunction with the 75th anniversary of the Federal Reserve System, the Atlanta Fed has published a retrospective on its 75 years of service. A History of the Federal Reserve Bank of Atlanta, 1914- 1989, by Richard H. Gamble, recounts the Bank's evolution in response to developments in the region, the nation, and the financial services industry. The story is told through personal recollections, as well as excerpts from official records, and is illustrated with photographs and other items from the Bank's archives. Beginning with Atlanta's race with New Orleans to become the headquarters for a Reserve Bank, Gamble follows the Atlanta Fed through the Depression, World War II, postwar technological developments, and the Southeast's rapid economic expansion starting in the 1960s. It concludes with a brief recap of the effect of the Monetary Control Act and the challenges of the 1980s. The illustrated 146-page volume is now available for $4.50 and may be obtained by writing the Public Information Department, Federal Reserve Bank of Atlanta, 104 Marietta Street, N.W., Atlanta, Georgia 30303-2713, or calling 404/521-8268. Economic Review Federal Reserve Bank of Atlanta 104 Marietta St, N.W. Atlanta, Georgia 30303-2713 Address Correction Requested Bulk Rate U.S. Postage PAID Atlanta, Ga. Permit 292