The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
F E D E R A L RESERVE B A N K O F ATLANTA J A N U A R Y / E E B R l l A R Y 1991 MORTGAGE-BACKED SECURITIES Analyzing Risk and Return U.S. Population Migration New Farm Bill Provisions Economic Review President Robert P. Forrestal Senior Vice President a n d Director o f Research Sheila L. Tschinkel Vice President a n d Associate Director o f Research B. Frank King Vice Presidents W i l l i a m Curt H u n t e r , Basic R e s e a r c h Mary Susan R o s e n b a u m , Macropolicy Research Officers William Roberds, Macropolicy G e n e D. Sullivan, Regional Larry D . W a l l , F i n a n c i a l David D. W h i t e h e a d , Regional Public Affairs Officer Bobbie H. McCrackin Publications J o y c e l y n T. W o o l f o l k , P u b l i c a t i o n s C o o r d i n a t o r Lynn II. Foley, Production C a r o l e Starkey, G r a p h i c s Ellen A n h . Circulation The E c o n o m i c R e v i e w seeks t o inform t h e p u b l i c a b o u t Federal Reserve policies a n d t h e e c o n o m i c e n v i r o n m e n t a n d , in p a r t i c u l a r , t o n a r r o w t h e g a p b e t w e e n specialists a n d c o n c e r n e d laypersons Views expressed in t h e E c o n o m i c R e v i e w a r e n o t n e c e s s a r i l y t h o s e of t h i s B a n k or of t h e Federal Reserve S y s t e m M a t e r i a l m a y b e r e p r i n t e d o r a b s t r a c t e d if t h e R e v i e w a n d a u t h o r are credited Please p r o v i d e t h e Bank's P u b l i c Affairs D e p a r t m e n t w i t h a c o p y o f a n y p u b l i c a t i o n c o n t a i n i n g reprinted m a t e r i a l Free s u b s c r i p t i o n s a n d l i m i t e d a d d i t i o n a l c o p i e s a r e a v a i l a b l e f r o m t h e P u b l i c Affairs D e p a r t m e n t , Federal Reserve Bank of A t l a n t a . 104 Marietta Street, N W . A t l a n t a , Georgia 30303-2713 (404/521-80201 Change-of-address n o t i c e s a n d s u b s c r i p t i o n canc e l l a t i o n s s h o u l d be sent directly to t h e Affairs D e p a r t m e n t Public Please i n c l u d e t h e current mail- ing label a s well a s any new i n f o r m a t i o n ISSN 0732-1813 V O L U M E 76, N O , 1. J A N U A R Y / F E B R U A R Y 1991. E C O N O M I C R E V I E W 2 Analyzing Risk and Return for Mortgage-Backed Securities Stephen D. Smith 12 Population Migration in the United States: A Survey of Research William J. Kahley 22 30 FYL The author provides a nontechnical introduction to modern methods used in calculating yields on mortgage-related products. In this survey of studies on the factors that motivate migration, the author also presents tentative findings of his empirical work on some unresolved migration issues. The 1990 Farm Bill Gene D. Sullivan Book Review B. F r a n k K i n g FEDERAL RESERVE B A N K O F A T L A N T A Manias, Panics, and Crashes: A History> of Financial Crises by Charles P. Kindleberger Analyzing Risk and Return for MortgageBacked Securities Mortgage-backed securities and their derivative products have become a major component of banks' and other financial firms' investment holdings. Calculating risk and return measures for these securities is complicated by the fact that homeowners have an option to prepay their debt obligation at any time. Because this prepayment option increases the risk of lower returns, new methods have been developed for adjusting the yields on mortgage-related instruments. The author describes one of these techniques—the option adjusted spread approach—which, unlike more conventional methods, adjusts for both the timing and level of potential prepayment. Stephen D. Smith T he growth o f an active secondary market Y i e l d adjustments for mortgage-backed securities, for h o m e mortgages w a s o n e o f the or MBSs, are necessary primarily because o f the law m a n y important innovations in financial allowing h o m e o w n e r s to prepay the principal bal- markets over the past decade. Although a n c e o n their m o r t g a g e s w i t h o u t p e n a l t y . 3 Since organizations such as the Federal National Mortgage such prepayments occur primarily w h e n Association (FNMA or "Fannie Mae") have been buy- rates fall substantially b e l o w existing c o u p o n rates ing securities b a c k e d by the Veterans Administration (that is, contract rates) o n the mortgages, investors market a n d the Federal H o u s i n g Administration for decades, in the mortgages face the risk that, after h a v i n g p a i d o n l y recently h a v e mortgage-related securities be- a p r e m i u m for a high c o u p o n security, they will be c o m e an integral c o m p o n e n t of financial statements saddled with m o n e y that must b e reinvested at low- for a n u m b e r o f b a n k s a n d other intermediaries. 1 er (current market) rates. Investment banks a n d oth- The growth of these securities has led to a dazzling er f i n a n c i a l f i r m s h a v e d e v e l o p e d array of derivative a n d hybrid products p r o d u c e d by- adjusting the yields o n mortgage-related instalments methods for r e p a c k a g i n g t h e basic cash flows f r o m a p o o l o f to reflect this possibility o f p r e p a y m e n t s a n d the fixed-rate mortgages. 2 Equally bewildering to poten- corresponding lower yields. tial investors in these products is the technology invented to calculate adjusted yields or, equivalently, adjusted spreads over Treasury yields. 2 Regulators are b e c o m i n g c o g n i z a n t o f these issues as they b u i l d a f r a m e w o r k for a n a l y z i n g the risk profiles o f an increasingly large p o o l o f securi- ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 ties w i t h u n c o n v e n t i o n a l cash flow characteristics. I n d e e d , the Comptroller o f the Currency has recent- cators o f the return to b e expected from h o l d i n g mortgage-backed securities, as discussed b e l o w . ly provided s o m e specific guidelines for the holdings o f c o l l a t e r a l i z e d m o r t g a g e o b l i g a t i o n s (see W i l l i a m B. H u m m e r 1990). T h e p u r p o s e o f this article is to p r o v i d e a non- Shortcomings of Static Yield technical i n t r o d u c t i o n to the m e t h o d s used to analyze the risks a n d returns associated w i t h investing in m o r t g a g e - r e l a t e d securities. This A s s u m i n g that p a y m e n t s are guaranteed against information default by a government agency such as the Gov- s h o u l d h e l p potential investors better c o m p a r e the e r n m e n t National Mortgage Association ( G N M A or cash flow a n d yield measures for mortgage-backed " G i n n i e Mae"), a standard fixed-rate mortgage is, in securities w i t h those o n alternative investments. t h e a b s e n c e o f the p r e p a y m e n t c l a u s e , nothing m o r e than an annuity contract. G i v e n a r e m a i n i n g life, a market price, a n d the promised payments per period, it is possible to find the contract's yield to The Prepayment Problem maturity (YTM), or "static" yield. Static is used to denote the fact that an investor will earn the yield to The p r o b l e m o f prepayment o n a mortgage (an maturity per period if all of the p r o m i s e d payments asset) is in s o m e ways the reverse o f the p r o b l e m o f are m a d e w h e n d u e a n d are reinvested at the same early withdrawal o f a fixed-rate certificate o f deposit rate (that is, rates d o not change over the life o f the (CD; a liability). I m a g i n e that a banker has issued a loan). T h e latter c o n d i t i o n is a w e l l - k n o w n short- fixed-rate C D for s o m e period o f time, a n d s u p p o s e c o m i n g o f u s i n g the yield-to-maturity m e t h o d to the depositor has the right to w i t h d r a w his or her calculate the expected return o n a n y security. It is funds at any time before maturity, without penalty. the first c o n d i t i o n that m a k e s the yield-to-maturity The depositor might w i t h d r a w early for t w o general a p p r o a c h p a r t i c u l a r l y u n a t t r a c t i v e for a n a l y z i n g reasons. If market rates o n C D s rose substantially m o r t g a g e - b a c k e d securities. I n s h o r t , t h e static a b o v e the current rate o n the C D , the C D h o l d e r yield treats t h e p a y m e n t s from a mortgage-backed m i g h t c h o o s e to w i t h d r a w early a n d reinvest the s e c u r i t y as a s u r e t h i n g o v e r t i m e , w h i c h funds in a higher-yielding account. W h e t h e r f u n d s clearly are not. are actually r e m o v e d or s i m p l y rolled over into a n e w account at the current b a n k , the banker will be replacing this relatively low-cost C D with funds that will cost substantially more than the old deposit. The second reason for early withdrawal w o u l d fall into a "catch-all" category that includes noninterest factors like the depositor's m o v i n g or d e v e l o p i n g an unexpected need for funds. In either case, the b a n k suffers a cost if it imposes n o early withdrawal penalty. they The static yield approach will also distort calculations c o m m o n l y u s e d to m e a s u r e the interest-rate risk o f a security. Duration a n d convexity are t w o such measures. H o w e v e r defined, the duration o f a fixed-income security is basically a measure o f the percentage c h a n g e in a security's price if interest rates c h a n g e b y a small a m o u n t . 5 Securities with shorter durations experience smaller price decreases, i n percentage terms, for a small increase in rates Prepayment o n a mortgage is analagous to the C D than d o securities w i t h longer d u r a t i o n . Likewise, e x a m p l e a n d may occur because rates fall substan- smaller increases occur for shorter duration securities tially b e l o w the m o r t g a g e rate the h o m e o w n e r is w h e n rates decrease. paying. Prepaying the m o r t g a g e for this reason is However, duration is itself a function o f the level called "rational exercise" o f the o p t i o n . Exercising o f interest rates. In fact, duration declines as interest the prepayment o p t i o n in other cases (such as mov- rates rise (and vice versa) for standard fixed-income ing for a n e w job) is called "irrational exercise" be- securities. This relationship is simply a result of the cause such behavior is not tied directly to interest fact that price changes are not symmetric. The per- savings. 4 centage c h a n g e in b o n d prices as rates increase is Rational exercise o f p r e p a y m e n t o p t i o n s forces smaller than the price changes associated with equal m o r t g a g e holders t o reinvest their f u n d s at rates rate decreases. Thus, the risk measure (duration) is substantially b e l o w those they w o u l d have earned if p r e p a y m e n t h a d not occurred. Moreover, m o r t g a g e s typically h a v e maturities m u c h since longer t h a n those o f m o s t o t h e r assets o r liabilities, the earnings loss is felt over a longer period o f time. Uncertainty concerning repayment of principal m a k e s conventional yield measures unreliable indi- FEDERAL RESERVE BANK OF ATLANTA The author is an Associate Professor and the interim Mills Bee Lane Professor of Banking and Finance, Georgia Institute of Technology• ¡Ie is currently a visiting scholar in the financial section of the Atlanta Fed's research department. He would like to thank Richard Hall and Michelle Trahue for assistance. 3 inversely related to the interest-rate level. Convexity is Treasury term structure o f interest rates o r y i e l d the term usually applied to this "drift" in the duration. curve. Typically, the implied one-period future inter- H o w e v e r , unlike fixed-income securities, mortgage- est rates (or forward rates) from the Treasury curve related securities' p r e p a y m e n t o p t i o n m a k e s cash are used as the m e a n , or expected value, a r o u n d flows a function o f interest rates. Risk measures such w h i c h a d i s t r i b u t i o n o f f u t u r e short-term interest as duration need to be adjusted to reflect this fact. rates is constructed. 8 Future mortgage rates are ei- T o summarize, the standard yield-to-maturity ap- ther constructed as a m a r k u p over the short-term proach for calculating risk a n d return is i n a d e q u a t e rates or, in m o r e c o m p l e x models, a m a r k u p over a w h e n analyzing mortgage-related securities primari- long-term Treasury rate that does not m o v e exactly ly because prepayment risk causes cash flows to b e in concert with short-term rates. A volatility (or vari- a f u n c t i o n o f interest rates a n d o t h e r factors. By ance) estimate is also needed to construct a distribu- treating the promised cash flows as certain, an in- tion o f future interest rates. This parameter restricts vestor is likely to overstate seriously the return from the degree to w h i c h rates m a y deviate from the cur- h o l d i n g mortgage-backed securities. The o p t i o n ad- rent term structure (the m e a n ) . Estimated prepay- justed spread ( O A S ) a p p r o a c h discussed in the fol- m e n t s are critically d e p e n d e n t o n t h e v o l a t i l i t y l o w i n g section is an attempt to adjust the cash flows estimates, w h i c h m a y c o m e from historical data or to reflect p r e p a y m e n t risk. 6 more exotic forms, such as implied volatilities from options contracts. Prepayments are estimated as a f u n c t i o n o f the The Logic of the Option Adjusted Spread Approach deviation o f current c o u p o n rates in the mortgage p o o l from estimated market rates a n d other currently available information such as the average age o f the mortgage p o o l a n d other k n o w n factors (for ex- The basic premise o f the option adjusted spread a m p l e , the region o f the country in w h i c h the mort- a p p r o a c h is that prepayments, a n d therefore cash gages o r i g i n a t e d ) . Future cash f l o w s are flows, will be a function o f both the evolution of in- generated as a function o f the evolution o f interest then terest rates a n d other (for e x a m p l e , d e m o g r a p h i c ) rates a n d the d e m o g r a p h i c factors. The fact that fu- factors that c o u l d cause irrational prepayments o n ture cash flows are d e p e n d e n t o n the entire interest pools or portfolios o f mortgages. A distribution o f rate process is c o m m o n l y referred to as path future cash flows (or prices) is generated b y assign- dency. ing probabilities to plausible alternative future interest-rate scenarios. Finally, in a step a n a l o g o u s to f i n d i n g t h e d i s c o u n t rate ( t h e static y i e l d ) that e q u a t e s the p r e s e n t v a l u e o f the p r o m i s e d cash flows to the current price, a yield measure can b e f o u n d that e q u a t e s the a v e r a g e present v a l u e o f these o p t i o n a d j u s t e d cash f l o w s to t h e c u r r e n t price. T h e difference b e t w e e n this adjusted yield a n d that o n a base security—a c o m p a r a b l e duration Treasury b o n d , for e x a m p l e — i s considered the option adjusted spread. A l t h o u g h this analogy is not exactly correct unless the yield curve is flat (see the a p p e n d i x for general definitions o f o p t i o n adjusted spread), the general idea is that similar calculations result in a yield measure for mortgage-backed securities that has b e e n adjusted for the expected level ( a n d t i m i n g ) o f p r e p a y m e n t s over the life o f the mortgage pool. 7 depen- Consider the case in w h i c h a d o w n w a r d movement in mortgage rates will p r o m p t a prepayment. If rates increase next period a n d return to their original level in period two, n o p r e p a y m e n t will occur. By the same token, if rates s h o u l d fall next period a n d then increase to their original level, prepayment will occur. The level o f rates in period t w o is the same in both scenarios, but the cash flow in period t w o is not. In this case the period t w o p a y m e n t is either the promised p a y m e n t or zero a n d is clearly a function o f earlier interest rates (in this case interest rates in period one). Therefore, each path of rates can generate a different cash flow pattern for the mortgage. The next step in the process is to find a constant discount factor w h i c h , w h e n a p p l i e d to every' path of future short-term Treasury rates, equates the cash outflow's present v a l u e (the current market price o f The critical steps to be taken in the option adjust- the mortgage) to the average present value o f the ed spread process appear in Chart 1. Although each cash inflows. This constant discount factor is the op- practitioner is faced with a n u m b e r o f specific choic- tion adjusted spread. es (some of w h i c h are discussed below), the steps The final step in Chart 1 involves shocking inter- outlined in C h a n 1 must be followed for almost all of est rates u p a n d d o w n b y s o m e a m o u n t . C o m b i n e d the option adjusted spread models currently in use. Raw input is p r o v i d e d from a n u m b e r o f sources. Interest rate information is gathered from the current 4 with the current price, the n e w prices provide sufficient information to calculate option adjusted duration a n d convexity measures. ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 Chart 1 Steps in Option Adjusted Spread Calculation FEDERAL RESERVE BANK OF ATLANTA 5 Computational Choices The procedure outlined in Chart 1 has at least two different versions, depending o n the practitioners' choice o f techniques for generating interest rates and discounting the cash flows. I n t e r e s t Rates. Probably the most widely used approach for generating a distribution of interest rates is the simulation method. Using forward rates embodied in the term structure as the means, the investigator inputs a variance estimate and draws a series of short-term rate paths. Resulting cash flows are generated and the process is repeated for another drawing from the distribution of rates. The simulation approach is sometimes ad hoc in the sense that the method need not be based directly on a rigorous link to the term structure of interest rates.9 An alternative is given by the binomial, or lattice, approach, which starts with today's term structure and assigns probabilities to scenarios wherein rates increase or decrease (or possibly remain the same). Cash flows are calculated at each point in the interest rate tree. (See the next section for an example.) A volatility estimate is needed for this technique as well, because it determines the amount by which rates are allowed to vary from point to point. D i s c o u n t i n g . The most intuitively a p p e a l i n g method for discounting involves finding the expected cash flow for each period (over all possible rates) and discounting back at rates contained in today's Treasury curve. However, the most popular method in use today (see, for example, Alan Brazil 1988) involves discounting back each cash flow at the simulated rate (as o p p o s e d to today's term structure rates). To the extent that rates and cash flows are correlated—correlation being the whole premise of rate-sensitive cash flows—the two techniques will yield different results. An example in the next section illustrates the difference between these approaches. Properties o f t h e O p t i o n A d j u s t e d S p r e a d . The foremost benefit o f the option adjusted spread approach is that it provides a yield measure that more accurately reflects the timing and level of payments that an investor might expect to receive from holding a mortgage-backed security. A second advantage is that risk measures calculated from prepayment adjusted cash flows provide a better indicator of the security's true interest-rate risk properties. For example, although the price of a standard fixed-income security will vary inversely with the level of interest rates, it is possible for prepayment adjusted prices to change in the same direction, no matter which way rates move. The key to this concept is that, should rates fall, the possibility of mortgage prepayments may go up, in which case investors may bid down 6 the mortgage-backed security's price. This action is, of course, the opposite of what would happen with a truly fixed-income security like a Treasury bond. This "whipsaw" effect is particularly evident in mortgagebacked securities that are selling at a premium from par value. Finally, the option adjusted spread methodology is often put forth as one method for identifying "rich" (overpriced) and "cheap" (underpriced) mortgagebacked securities. Typically, the o p t i o n adjusted spread on securities with similar adjusted durations and coupons are compared. Matching durations is an attempt to hold constant the differences in the risks of the assets. Note, however, that such comparisons tell the investor nothing to give direction about whether he or she should purchase either of the securities. Suppose, for example, that the yield for a stream of expected cash flows is greater than that for a comparable duration Treasury security. This situation is analogous to the case of a positive option adjusted spread (OAS > 0). A risk neutral investor—one w h o demands no compensation for the variability of the cash flows (read "variability o f prepayments")— would certainly find such an investment attractive. However, a positive option adjusted spread alone would not generally provide a risk averse investor with enough information to determine whether or not the extra yield would exceed the investor's desired risk premium. Another (and equivalent) way to view this ambiguity is to recognize that if two mortgage-backed securities have the same expected cash flows but the variability is greater for, say, the second one, riskaverse investors will bid a lower price for the second mortgage-backed security. The result will be that the second mortgage-backed security has a higher option adjusted spread. The meaning is clearly not that the second security is a better one, however. In short, establishing that the expected return on a risky security is greater than the Treasury rate, or even greater than the expected return on some other security of comparable risk, does not imply that it is a good buy unless you happen to be neutral toward risk (because it could be the case that neither of the securities has a high enough premium to cover its risk). Such "risk neutral" information is exactly the sort the option adjusted spread methodology provides. Examples of Option Adjusted Spread Technology In the following examples, noninterest rate "irrational" factors that might influence prepayments are ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 ignored for simplicity. Moreover, for simplicity the examples will deal with a case where the term structure is flat. Consider a mortgage-backed security for which the underlying fixed-rate mortgages are identical. The total principal balance on the pool is $1,000,000, and the mortgages carry an 11 percent coupon and have a maturity of four years. The latter assumption, while unrealistic, allows analysis of the process without changing the qualitative results. Promised payments on this annuity contract amount to $322,326 per year. Shortly after issue, rates decline in such a fashion that the mortgage pool is now selling for a current market price of $1,044,246. It is always possible to find a rate, the static yield, that will discount back the promised cash flows to the current market price. In this case the rate turns out to be 9 percent (see the appendix). If Treasury rates are 8 percent, the "static spread" is 9% - 8% = 100 basis points. Assume that because of refinancing costs and other factors prepayment will not occur unless the spread between the c o u p o n rate and market rates on mortgages is 300 basis points (3 percent). In this case prepayment will occur if mortgage rates are less than or equal to 11% 3% = 8%. Interest Rates. A simple binomial (or two-state) process is used to model changes in interest rates.10 Rates can move up and down with equal likelihood from their current levels. The assumed change is equivalent to a rate volatility estimate. Let this be 50 basis points. In this situation, mortgage rates will be either 9.5 percent (9% + .5%) or 8.5 percent (9% .5%) next year. In year two rates will be either 9.5 percent ± .5 percent or 8.5 percent + .5 percent. So in year two there is a 25 percent [(.5)(.5)J chance that rates will be either 10 percent (go up twice) or 8 percent (go down twice) and a 50 percent chance the rates will return to 9 percent (that is, a 25 percent chance they will increase and then decline and a 25 percent chance they will decline and then increase). Chart 2 provides a graphic representation (a "tree") of all possible rates over the four-year life of the mortgage pool. C a s h Flows. In any given year the realized cash flows will be either ( a ) the p r o m i s e d p a y m e n t ($322,326), (b) the promised payment plus prepayment of the remaining principal ($322,326 + principal balance) or (c) zero. Case (b) occurs if and when mortgage rates fall below 8 percent, while case (c) is the cash flow in subsequent periods should prepayment take place. Chart 2 Interest Rate Tree Current Mortgage Rate = 9.0% Volatility = 50 Basis Points Year FEDERAL RESERVE BANK OF ATLANTA 7 Given the volatility estimate of 50 basis points there is n o chance of prepayment in period one because rates can only fall as low as 8.5 percent. The cash flow is certain to b e $322,326. However, in period two there is a 25 percent chance that rates will d r o p to 8 percent a n d p r e p a y m e n t w i l l o c c u r . Should this happen, the cash flow will b e $322,326 + $551,992 = 874,318, where $551,992 is the remaining principal balance on the mortgage. O n c e prepayment has occurred, the cash flow in years three and four will be zero. The expected (or average) cash flow from the pool can also be calculated by multiplying each possible cash flow by the probability that it will occur, in year one, for example, there is a 0 percent chance of prepayment because rates will always be above 8 percent. Alternatively, in year two there is a 25 percent chance the mortgage rate will be 8 percent and the corresponding cash flow $874,318. There remains a 75 percent chance the rates will be above 8 percent and the cash flow will be $322,326. Thus, the expected cash flow is $874,318 (.25) + $322,326 (.75) = $460,324. Table 1 provides the possible and expected cash flows by period. In this special case, the option adjusted yield can b e c a l c u l a t e d by f i n d i n g a d i s c o u n t factor that equates the discounted value of the expected cash flows (see Table 1) to the current price. The rate that solves this discounting problem is about 8.85 percent, somewhat less than the static yield of 9 percent. Finally, the o p t i o n adjusted spread in this simple example is given by subtracting the Treasury rate (8 percent) from the adjusted yield; the option adjusted spread is 8.85% - 8% = 85 basis points. This importance of the volatility measure can be seen by first considering the case in which rates in Table 1 Possible and Expected Cash Flows from Mortgage by Period Period Possible Cash F l o w Expected Cash F l o w 1 $322,326 $322,326 2 $322,326 or $874,318 $460,324 3 $322,326 or $0 $241,745 4 $ 3 2 2 , 3 2 6 or $0 $241,745 8 each period can c h a n g e by only 25 basis points rather than 50. The mortgage rates n o w have n o chance of falling to 8 percent before the mortgage m a t u r e s . T h e r e f o r e , t h e e x p e c t e d cash f l o w is $322,326 in every period a n d the option adjusted spread will be the same as the static spread (100 basis points). Alternatively, the option adjusted spread is actually negative (-12 basis points) if the volatility estimate is doubled to 100 basis points. O p t i o n adjusted spread estimates are also extremely sensitive to the discounting method used by the investigator. The calculation procedure described above can be viewed as a "discounted average cash flow" approach. Alternatively, as mentioned earlier, many investment houses prefer what could be called an "average discounted cash flow" (or average price) approach. This method differs from the discounted average cash flow approach because the cash flows are discounted back at the realized interest rates in Chart 1, as opposed to today's rate. In this case, lowcash flows are, o n average, associated with lower discount rates (the prepayment problem) and vice versa for higher rates. The net result is that the option adjusted spread will b e different (typically lower). For this particular example the option adjusted spread is actually a negative 2 percent (-200 basis points) w h e n using the average price method. The a p p e n d i x contains equations (2) a n d (3), w h i c h s h o w the mathematical distinction between the approaches. 11 This rather large difference in results comes from the fact that the high cash flows are discounted at higher rates in the average price approach. 1 2 The benefits are not symmetric, though, because the zero cash flow after prepayment is still zero n o matter h o w low the discount rate is. A final example of the o p t i o n adjusted spread method involves its use in calculating the risk an investor faces for relatively large interest-rate changes. A potentially useful exercise is to see h o w mortgage prices w o u l d c h a n g e if rates are s h o c k e d u p or d o w n by some amount in the current period. For example, if Treasury and mortgage rates increased immediately by 100 basis points, there would be a zero chance that the mortgage rate could fall to 8 percent before year four (that is, 9% + 1% = 10% and the volatility is 50 basis points a year). It is possible, using the original option adjusted spread, to calculate a new price. Consider, for simplicity, the average cash flow m e t h o d . The o p t i o n adjusted spread is 85 basis points, while current mortgage and Treasury rates are 9 percent and 8 percent, respectively. Suppose rates increase immediately to 10 percent and 9 percent, respectively, and volatility remains constant at 50 basis points. Then, using the average cash flow method, the new price is $1,025,057, which is less ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 than $1,044,246 because the present value o f the fu- average risk. T w o other caveats merit attention. It is ture cash flows has been reduced. possible, especially for mortgage securities selling at Alternatively, if rates s h o u l d immediately fall by 100 basis points, prepayment will occur as s o o n as p r e m i u m a b o v e par, to encounter interest-rate changes that " w h i p s a w " the investor; that is, prices possible (probably in period one), because the n e w m a y fall if rates either decline or increase. The ad- mortgage rate is 8 percent. In this case the n e w price justed yield measures can b e very sensitive to the in- is $1,029,208, w h i c h is also lower than the original puts u s e d — f o r e x a m p l e , the a s s u m e d volatility o f price. This w h i p s a w effect occurs because the lower interest rates. interest rate causes earlier p r e p a y m e n t , m o r e than The o p t i o n adjusted spread has b e c o m e a favored offsetting the present v a l u e a d d e d from h a v i n g a technology for dealing with this p r o b l e m because it lower rate w i t h w h i c h to d i s c o u n t b a c k the cash adjusts for b o t h the t i m i n g a n d level o f potential flows. These examples should m a k e clear that, w h i l e prepayments. In fact, the a p p r o a c h can b e u s e d in a it is conceptually an averaging technique, the actual variety o f settings because m a n y assets a n d liabili- option adjusted spread is extremely sensitive to a va- ties have o p t i o n s o f s o m e sort e m b e d d e d in their riety o f i n p u t a n d methodological decisions practi- structure. 1 3 tioners make. W h i l e potentially useful, the o p t i o n adjusted spread is not a panacea for investors h o p i n g to find u n d e r v a l u e d assets o n a risk-adjusted basis. N o form a l m o d e l currently provides a basis for decompos- Conclusion ing the o p t i o n adjusted spread into c o m p e n s a t i o n for risk a n d excess returns. V i e w e d properly as a The calculation o f yields o n mortgage-backed se- yield (or yield spread over Treasury), h o w e v e r , the curities is c o m p l e x primarily because of the difficulty option o f v a l u i n g an owner's o p t i o n to p r e p a y the mort- clearer indicators o f the likely return than conven- adjusted spread measures are probably gage. Despite their computational complexities, cur- tional static yield calculations. The trade-off here in- rent approaches to mortgage valuation are still just volves the o p t i o n a d j u s t e d spread's sensitivity to averaging techniques. As s h o w n in this article, cash inputs (such as volatility a n d the p r e p a y m e n t mod- flows, a n d therefore returns, d e p e n d o n the entire el) a n d the v a l u a t i o n f r a m e w o r k the investigator path of interest rates; that is, they are path depen- employs. Potential purchasers, as well as regulatory dent. Therefore, the interest-rate history over the life personnel, s h o u l d b e aware o f these facts. Asking of a mortgage-backed security is an important piece potential sellers for information concerning the ac- o f i n f o r m a t i o n . A n a d d i t i o n a l c o m p l i c a t i n g factor tual r i s k / r e t u r n p r o f i l e s o f p r e v i o u s l y arises from the difficulty in determining whether a mortgage-backed securities w o u l d b e useful in this mortgage that looks "cheap" is really undervalued or regard. 1 4 analyzed whether it is selling at a discount because of above- Appendix This appendix contains the general formulas used for calculations in the text. As noted below, the actual examples usually involve simplified versions of these equations (for example, the term structure is flat). M = market price of the mortgagebacked security C = promised payment per period on a fixed-rate contract Y = static yield to maturity N = remaining maturity of the contract The static yield is the rate that equates the present value of the promised cash inflows (Cs) to the current market price. That is, Ysolves FEDERAL RESERVE BANK OF ATLANTA C C (1 + Y) (1 + Y) M + ...+ C a + Yy (i) If G is the yield on a comparable maturity (more specifically, duration) Treasury security, the static spread is just Y - G = S. For the example in the text, C = $322,126; N= 4; M= $1,044,246, so Y= 9% and 5 = 9 % - 8 % = 1%. More notation and calculations are needed to calculate the option adjusted spread. Let/, = one-period forward rate for government securities in period t, ft= (1 + R, Y/O + Rt_ j ) ' " where R, is the current spot rate for a government security that makes one payment in period t and zero otherwise. Define rt = actual interest 9 rate on one-period government securities in period /. Note that this number is unknown today if / > 1. With this notation, J\ = r, = R{. Finally, let Ct = actual cash flow realized from the mortgage in period t. Notice that there are two general cases: C t = C (promised payment) if there is no prepayment, or C) = C + remaining principal balance > C if there is prepayment. In this case Ct + j = Ct + 2 = • • • = CN = 0. For notational purposes let £ ' ( • ) denote expectation (or expected value) of the term inside the parenthesis. The expected value of x, for example, is given by the sum of the possible outcomes for x multiplied by the probability that each outcome will occur. There are two approaches to calculating the option adjusted spread. Expected Cash Flow A p p r o a c h Search for a discount factor, Oc , such that Oc solves ,v/= E[CX] (1+/,+ +. . . + Oc)+ E[C2) (l + /i + oc}(i+/2+oc) (2) 1-1C.v] (1 + / , + O c X 1 + f , + O . ) , . . (1 + f N + Oc ) The term E[CtJ, t= 1, . . . 4 is given by the right-hand column of Table 1. These terms can be calculated bymultiplying the possible cash flows by the probability that rates will be above or below the cutoff rate for prepayment. In particular, with equal probabilities (.5) of an increase or decrease in rates and volatility = 50 basis points, E[C{\ = $322,326(1). Because there is a 25 percent chance [(.5)05) = .25] that the Treasury rate will fall to 7 percent in year two (mortgage rate = 8%), the expected cash flow in period two is given by multiplying the possible cash flows in Table 1 by their respective probabilities. This calculation gives the expected value, or E\C2) = $322,326 (.75) + $874,318025) = $460,324. Likewise, because the cash flow in period three is zero if prepayment occurs, the expected cash flow in this case is E[C}\ = $322,326 (.75) + $0 025). The expected cash flow in period four will be the same, so E[C4] = $322,326 075) + $0 025). Solving (2) for the option adjusted spread yields Oc = 85 basis points. Notice that when the term structure is flat, the forward rates will be the same. Thus, solving (2) is the same, in this case only, as the simpler approach used in the text—solving for the yield first and then subtracting the (constant) Treasury rate. Expected Price A p p r o a c h Find a discount factor, O p , that solves the following equation: M=E + . . . + £ (1 + r, + Op) + ¿2 (1 + r, + Op)+(1 + r2 + Op) (3) c. (1 + t\ + Op)(1 4- r2 + Op) . . . (1 + rv + Op) In this case the rs are the realized one-period Treasury rates from the tree in Chart 2. For this example, the current Treasury rate is 8 percent and the term structure is flat; therefore, r, = 8%. However, next-period rates will be either 7.5 percent or 8.5 percent, so r? = 7.5% with probability 05) and r2 = 8.5% with probability .5. Interest rates for future periods are calculated in a similar fashion. The possible C's are found in the first column of Table 1. Using (3), Op can be calculated as Op = -2.0% or -200 basis points. Calculating New Prices The new prices are calculated by fixing 0 ( (or O p ), adding a constant amount (+ 100 basis points) to each f t (or r t ), and finding the new cash flows associated with these rates. The new M is then given by simplyusing the discount formula. For the expected cash flowapproach, a 100-basis-point increase in rates will result in a zero probability of prepayment, so E[C\) = E\C+,)=ElCl 1 = EIC\] = $322,326, where C * is the cash flow for a rate increase. The discount rates in equation (2) are (1 + /, + .01 + .0085) for period one, (1 + /J + ,01 + .0085) (1 + / , + .01 + .0085) for period two, and so forth. The new price can be calculated using equation (2). The rate decrease is the same problem except that, if rates should fall by 100 basis points, prepayment occurs immediately, and E[C\] = $1,110,000 and E[C21 = E[C$ = E[C4] = O. The discount factors are (1 + / , - .01 + .0085) and so on. An analogous approach would also be used if one were applying the expected price methodology. 12 ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 Notes 1. The book value of domestic bank holdings of guaranteed and nonguaranteed mortgage-backed securities was about $200 billion at the end of 1989. About $35 billion of these assets was held by banks in Alabama, Florida, Georgia, Louisiana, Mississippi, and Tennessee. 2. See Sullivan and Lowell (1988) for an introduction to the mechanics of the mortgage securities market and the major participants. 3. For convenience the terms mortgage and mortgagebacked security will, when there is no ambiguity, oftentimes be used interchangeably. 4. See, for example, Hendershott and Van Order (1987) for a formal options approach to modeling rational exercise. used here is based on the original probabilities. In essence, such a formulation amounts to assuming that investors are, in some sense, truly indifferent to risk. See notes 8 and 11 for additional comments on this point. 8. Such a formulation is equivalent to assuming that investors demand no risk premium for holding longerterm bonds. See Abken (1990) for a review of this "expectations" hypothesis of the term structure. 5. Interestingly, duration can also be viewed as a weighted average time to maturity, where the weights for each period / are equal to the present value of period t's cash flow divided by the total present value (the market value, or price). Shorter-term securities are those that have relatively large cash flows in earlier periods. For fixed-rate mortgages, however, the promised cash flows are a level annuity, so the weights are simply the present value factor for each period relative to the present value interest factor for an annuity. 9. Schwartz and Torous (1989) provide a good example of a method for explicitly linking interest rate processes to a theory of the term structure in the context of valuing mortgage-backed securities. 10. See Abken (1990) for a discussion of binomial processes in the context of the term structure of interest rates. 11. From a somewhat more technical perspective this choice of assumptions may be v i e w e d as one of choosing between "global" and "local" risk neutrality on the part of investors. With local risk neutrality investors expect to earn the same return from all securities over any one period of time. It can be shown byrepeated substitution that this results in the expected price approach discussed in the text. See Cox, Ingersoll, and Ross (1985) for a mathematical discussion of the "local" expectations hypothesis. 6. It should be noted that some adjustment for early prepayments is typically incorporated into the static yield framework. For example, yield quotes provided by investment banks will often incorporate a constant prepayment rate per month for the remaining mortgage balance. 12. Part of the difference may also come from the fact that in this example the average realized rate is set equal to today's rate. This approach is different from the more mathematically correct one, which would be to make the average realized one-period bond price equal to today's one-period bond price. 7. Those familiar with options pricing theory will realize that this method is only loosely based on conventional option pricing models. Although it is true, for example, that an option's value can, under certain circumstances, be viewed as the expected payoff over a transformed (or "risk neutral") probability distribution, the approach 13. See, for example, Ayaydin, Richard, and Rigsbee (1989) for a discussion. 14. For example, Toevs (1990) provides evidence on the return characteristics o f various mortgage-backed securities when compared to duration-matched Treasury securities. References Abken, Peter. "Innovations in Modeling the Term Structure of Interest Rates." Federal Reserve Bank of Atlanta Economic Review 75 (July/August 1990): 2-27. Ayaydin, Sirri, Charles Richard, and Stephen Rigsbee. "Applying an OAS Model Consistently." Financial Managers Statement 11 (November/December 1989): 65-75. Brazil, Alan. "Citicorp's Mortgage Valuation Model: Option Adjusted Spreads and Option Based Durations." Journal of Real Estate Finance and Economics 1 (June 1988): 151-62. Cox, John C., Jonathan F. Ingersoll, Jr., and Stephen A. Ross. "A Theory of the Term Structure of Interest Rates." Econometrica 53 (March 1985): 385-407. Hendershott, Patrie, and Robert Van Order. "Pricing Mortgages: An Interpretation of the Models and Results." FEDERAL RESERVE BANK OF ATLANTA Journal of Financial 1987): 77-111. Services Research 1 (September Hummer, William B. "Say Hello to the CMO." Bankers Monthly 107 (December 1990): 54. Schwartz, Eduardo S., and Walter N. Torous. "Prepayment and the Valuation of Mortgage-Backed Securities." Journal of Finance 44 (June 1989): 375-92. Sullivan, Kenneth, and Linda Lowell. "Mortgage PassThrough Securities." In The Handbook of MortgageBacked Securities, edited by Frank J. Fabozzi, 69-114. Chicago: Probus Publishing, 1988. Toevs, Alden. "Laser Brains Rejoice: Analytical Methods Can Help Shape Market Equilibrium Prices." Financial AnalystsJournal 46 (November/December 1990): 8-10. 11 Population Migration in the United States: A Survey of Research William J. Kahley Migration behavior can strongly influence a region's economic development. Understanding the motivations for migration is important to state and local policymakers because in-migrants can fuel job growth and stimulate construction activity, but they can also overburden roads, schools, and other infrastructure. This article provides insights into the determinants of migration through an overview of theories on the subject and a survey of relevant literature. The author also reports the findings of his own empirical workconcerning the influence of certain economic variables. eople m o v e from o n e c o m m u n i t y or re- P growth a n d differentials in e c o n o m i c performance. g i o n to another for m a n y reasons. Un- Because o f the contribution o f w o r k e r a n d retiree derstanding this m o v e m e n t o f p o p u l a t i o n migration to Florida's a n d Georgia's above-average is important because it alters a n area's e c o n o m i c g r o w t h c o m p a r e d w i t h the nation's, im- potential for e c o n o m i c growth by directly affecting p r o v e d u n d e r s t a n d i n g o f t h e c a u s e s a n d conse- the size a n d c o m p o s i t i o n o f regional labor forces. q u e n c e s o f m i g r a t i o n is o f s p e c i a l i n t e r e s t State or local g o v e r n m e n t officials, w h o see in- southeastern states. At the same time, recessionary in migration as fueling job growth a n d the d e m a n d for e c o n o m i c conditions have caused other southeastern h o u s i n g a n d other types o f construction, work to at- states such as Louisiana a n d Mississippi to lose pop- tract (or keep) high-income earners, skilled workers, ulation in recent years. As the a m o u n t o f natural in- or wealthy retirees with g o o d pensions. crease ( n u m b e r o f births m i n u s n u m b e r o f deaths) in- i n the U n i t e d States declines in the years a h e a d , migration can also create burdens for state a n d local Rapid population growth resulting from p o p u l a t i o n change via redistribution will b e c o m e an governments, however. Congestion, p o l l u t i o n , a n d even more important demographic influence on increased need for p u b l i c transportation, schools, state economies. a n d other infrastructure investment can drain a local The considerable research in recent decades o n economy. The goal o f planners a n d policymakers is the factors motivating migration is reviewed in this to lure migrants w h o will add less to the social bur- article along with several important issues that con- d e n than they return in taxes a n d other receipts, tinue to d e m a n d attention. Research concerning the thereby raising the area's quality o f life. overall qualitative or quantitative impact of migration A n u m b e r o f studies that deal with the effects o f is m o r e limited a n d generally b e y o n d the scope o f m i g r a t i o n s h e d light o n t h e s u b j e c t o f r e g i o n a l this article. However, this summary includes a report 12 ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 o n s o m e progress in the study o f the impact o f mi- e c o n o m i c theorizing a n d empirical findings reveal gration w h e n it is narrowly viewed as a process o f that m i g r a t i o n decisions are c o m p l e x a n d multidi- adjustment to regional labor market differences. m e n s i o n a l , m a d e b y i n d i v i d u a l s w i t h a variety o f Following an overview o f theoretical perspectives considerations. I n fact, individuals (or a family unit) o n the factors that motivate migration, the article in- can be viewed as m a k i n g not o n e but t w o decisions, cludes a brief review o f the literature dealing w i t h in sequence. The first is the basic determination o f the relationship between migration a n d e m p l o y m e n t whether or not to relocate, the second, the choice o f growth. The succeeding section contains important destination. survey results concerning specific determinants o f in- Theorists favoring a h u m a n capital a p p r o a c h to migration. The final t w o sections consist o f tentative the question o f whether to m o v e (Larry A. Sjaastad findings o n s o m e unresolved migration issues based 1962; Gary S. Becker 1964) have argued that individ- o n the author's empirical w o r k a n d a s u m m a r y o f uals will "invest" in migration only if it is profitable— the principal conclusions a n d policy implications of that is, if the present v a l u e o f perceived benefits information presented in this survey o f migration re- from m o v i n g exceeds the present value of costs as- search. sociated with m o v i n g (earnings m i n u s migration costs). 2 Individuals can b e expected to m o v e to the place that offers t h e m the highest net a d v a n t a g e , more broadly defined than in the definition b y Hicks Migration Theories cited earlier. A c c o r d i n g to Sjaastad, migration distance generates several costs b e y o n d the monetary In early migration studies researchers v i e w e d the expenses o f m o v i n g , i n c l u d i n g earnings forgone a n d issue o f labor migration primarily from either the other costs related to the job search, the psychologi- neoclassical e c o n o m i c perspective or in light of the cal cost o f leaving the current environment, a n d un- m e c h a n i c a l gravity m o d e l . As succinctly stated by certainty about prospective i n c o m e because o f the J o h n R. Hicks, neoclassical e c o n o m i c theory assumes possibility of imperfect information. that "differences in net e c o n o m i c advantages, chiefly Subsequent theorizing has a d d e d that the decision differences in wages, are the m a i n causes o f migra- to migrate is also m u l t i d i m e n s i o n a l because o f the tion" (1932, 76). I n this view, migration is seen as m a n y factors involved in perceived benefits. For ex- part o f the process whereby workers adjust to differ- ample, a family or individual's personal characteris- ences a m o n g regional labor markets. Labor is ex- tics w i l l affect t h e v a l u e p l a c e d o n w h a t a n e w pected to flow from low-wage, labor-surplus areas location has to offer. In addition, characteristics o f to high-wage, labor-short regions until w a g e levels the places themselves—the migrant's place o f origin converge. and the potential d e s t i n a t i o n — h e l p The mechanical gravity m o d e l — l i k e Newton's law determine whether or not a m o v e will occur. o f universal gravitation, w h i c h states that the force of Individual or personal qualities like age, sex, race, attraction between t w o bodies is directly proportion- education a n d skills, earnings a n d e m p l o y m e n t sta- al to their respective masses a n d inversely propor- tus, a n d health, as well as life-cycle characteristics tional such as completion o f schooling a n d entry into the to t h e s q u a r e o f t h e d i s t a n c e between t h e m — h o l d s that migration between t w o regions is work force, marriage or divorce, birth a n d aging o f related to links between regional populations a n d in- children, a n d retirement, influence the individual or tervening distances. These gravity models originally family unit m a k i n g the migration decision. Individu- hypothesized that gross migration interchange w o u l d als or families with different traits will weight the net b e directly related to the size of the p o p u l a t i o n in advantage of o n e place over another differently, de- the areas p e o p l e were m o v i n g between a n d inverse- p e n d i n g u p o n b o t h the individual's or family's traits ly related to distance. 1 Distance is seen as a barrier a n d the characteristics of the place being evaluated. a n d larger p o p u l a t i o n masses, a magnet. More recently, the original m e c h a n i c a l v i e w has b e e n exp a n d e d b e y o n d p h y s i c a l q u a l i t i e s . For e x a m p l e , p o p u l a t i o n a n d distance might really reflect the econ o m i c influences of market size a n d transportation costs, respectively. N u m e r o u s factors figure in w h e n calculating the net a d v a n t a g e o f o n e place over another. Among e c o n o m i c factors, j o b o p p o r t u n i t i e s ( o r the lack thereof) a n d potential living standards adjusted for cost differences are obvious considerations. Nonecon o m i c benefits such as topological ( m o u n t a i n , plain. Recent research has exposed both o f these views o f w h y migration occurs as oversimplifications. Comprehensive surveys by Michael J. G r e e n w o o d (1975, 1985) a n d Charles M u e l l e r (1982) o f m o r e recent FEDERAL RESERVE BANK OF ATLANTA The author is an economist in the regional section of the Atlanta Fed's research department. 13 or seashore), climatological (sunshine a n d temperature), a n d e n v i r o n m e n t a l ( c l e a n air a n d In response to this view, an alternative theory water) emerged that held that migration itself could also in- amenities m a y also be important. Additionally, the duce job growth and development or that "jobs fol- levels of public services provided, along with the rel- low people." George H. Borts a n d Jerome L. Stein ative cost of certain items like food, housing, a n d (1964) explained regional economic growth from this taxes, m a y be key concerns influencing migration contrary, supply-side perspective. For them, migra- decisions. tion to a higher-wage region was the driving force Characteristics that lure or attract migrants to a behind regional population and employment growth. place are "pull" factors. Surf, sun, and availability of In Chart 2, the shift in labor supply (Zv — jobs, for example, invite new residents. For reasons sents migration from a low-wage region. The impor- ) repre- that will become clear in the next section, in-migration tant assumption underpinning this approach is that itself is a pull factor. Qualities that repel migrants— labor demand in area Beta will accommodate work- such as a high unemployment rate a n d stagnating re- ers f r o m the l o w - w a g e area w i t h o u t gional economy—are called "push" factors. wages. ( D e m a n d for labor is assumed to be perfectly depressing elastic, and it is as if "jobs follow people.") Also, inmigration is likely to induce more local investment, shifting the labor d e m a n d curve u p w a r d and raising Migration in Regional Growth and Development Models wages still higher. Historically, supply-driven growth may have occurred in the United States with movement of workers out of agriculture and into urban A key issue in regional growth theory is whether manufacturing. or not interregional factor mobility leads to conver- Push a n d pull influences discussed earlier corre- gence of regional per capita incomes.^ As noted, spond to these supply- and demand-driven models Hicks and the basic neoclassical economic model sug- of regional e m p l o y m e n t growth. According to the gest that convergence is likely. Assuming two regions, supply approach, low wages push workers out of a one with relatively abundant labor a n d the other with region. J o b opportunities lure workers, as stated by capital, the low-wage region with a relative abun- the d e m a n d theory. Both are equilibrium-type mod- dance of labor should attract capital. That is, because els in that the tendency is for wages between re- capital is relatively scarce in the low-wage region, it gions to converge. Another potential implication of will be comparatively productive and profitable to use the demand-driven model is that job or income op- more capital there. Equivalently, the higher-wage and portunities associated with places may better explain lower-return-to-capital region will attract labor. In in-migration than out-migration. Economic prosperity short, factor mobility arbitrages away differences in will b e a d o m i n a n t factor affecting the choice of returns to labor or capital geographically. where to move. O n the other hand, personal charac- In the 1960S, the basic approach of Cicely Blanco teristics such as age, stage of education, and beginning (1963), I.S. Lowry (1966), and Warren F. Mazek (1969) work or retiring seem more important in explaining to explain regional economic growth assumed that out-migration (within the supply-driven theoretical increases in d e m a n d for goods a region produces framework). and exports to other regions leads to an increase in In the 1970s, two theoretical developments impor- labor demand locally, inducing in-migration. In this tant for this review occurred, one by Richard F. Muth export-based view of growth, where "people follow a n d the other by Philip E. Graves and Peter Linne- jobs," in-migration or net migration gain occurs in man. The demand- a n d supply-driven models briefly the high-wage region w h e n a disequilibrium is creat- described above can be viewed as alternative per- ed via the increase in d e m a n d for goods; the key to spectives analogous to the "which came first—chick- change is the increased d e m a n d for products a n d e n or egg?" debate. In fact, M u t h (1971) m a d e a the profitability o f a region's industries that serve major contribution to the migration debate with an outside markets. In Chart 1, z j represents increased article titled "Migration: Chicken or Egg?" H e con- d e m a n d for labor in region Alpha resulting from in- cluded that attention should be focused o n the inter- creased d e m a n d for exports produced there. The dif- d e p e n d e n c e of migration a n d job opportunities in ference between Z a n d Z' represents in-migration explaining regional growth and development. While induced by job opportunities (reflected in the shift in a c k n o w l e d g i n g that either factor can generate labor d e m a n d , Ld~•Zj). growth, Muth maintained that typically both d o so si- Labor supply is assumed to be abundantly available from another region to multaneously. In Chart 3, the labor supply or labor meet the increased d e m a n d for labor in Alpha. New d e m a n d curves (or b o t h ) c o u l d shift, triggering em- jobs that become available both attract migrants and ployment growth. Additionally, the initial supply or trigger faster growth. d e m a n d shifts (L$ 14 or Ld — • z j ) cause further ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 shifts in the curves ( t o L " or L "d ). Unfortunately, comparative static-type diagrams cannot adequately capture these dynamic interactions and feedback effects. Chart 1 The D e m a n d A p p r o a c h Graves and Linneman (1979) argued that rising income levels may cause migration by prompting new demands for amenities that may be location-specific. Changes in individual or family demands for amenities related to marriage or divorce, birth, death, or retirement can also trigger migration. Moreover, according to Graves (1983), the mobility of firms as well as households is an impetus for regional development. It is also the way by which location-specific factors such as amenities are capitalized, or their values incorporated, into land and labor markets. Graves and Linneman's argument is that the only way an individual's or family's demands for nontraded amenities can be satisfied is by relocation. However, migration will also change wages and prices in resource and product markets to reflect the value of amenities. A major implication is that regional wage differences can persist in the long run because of the capitalization of regional amenities. For example, employers may be able to pay persistently lower wages for a given job and still attract workers to a region because it possesses desired amenities in abundance. A regional divergence of wages can also be created when growing incomes permit consumers to seek n e w amenities (such as a sudden popularity of snow skiing), with a wage divergence compensating for regional differences in the value of amenities. In short, labor markets can be in equilibrium in two regions, yet wages can differ because of disparities in the availability of amenities. The result is that unadjusted wage differences will not necessarily equalize as migration occurs. Thus far, the consequences of migration have been analyzed primarily within the narrow framework of regional economic development as a process of adjustment in regional labor markets. Labor tends to move from low-wage, labor-surplus regions to high-wage, labor-scarce regions unless differences in amenities compensate for wage differences. At a broader level, though, migration can be viewed as a process that significantly affects the local economies migrants leave behind and those into which they move in ways beyond the direct impact on incomes. For example, selective out-migration of young, educated people from a low-wage region can have cumulative, adverse effects on labor demand that can more than offset the positive impact of reduced aggregate labor supply in raising wages. Alternatively, crowding or other deleterious effects can reduce or eliminate the presumed benefits to an area from inmigration. The effects of migration are complex, and any in-depth discussion of them would require FEDERAL RESERVE BANK O F ATLANTA Chart 2 Chart 3 Supply and D e m a n d Interactions review o f a n e n o r m o u s segment o f e c o n o m i c litera- interregional moves" (1988, 251). Research reviewed ture that is b e y o n d the scope o f this article's focus by G r e e n w o o d (1975, 1985) a n d Mueller (1982) also o n the determinants o f migration. tends to support the idea that d e m o g r a p h i c determinants have impacts like those discussed here. 5 A l t h o u g h certain p e r s o n a l traits are associated Personal Characteristics Affecting Migration with a greater or lesser probability of migration a n d can be linked to e c o n o m i c or life-cycle forces in predictable ways (such as c o m p l e t i n g school, getting married or divorced, taking a job), other personal Demographers, economists, a n d other researchers circumstances entail c o m p l e x interactions. For exam- attracted to the study o f migration have noted sever- ple, the effect o f race or ethnicity is not self-evident. al patterns in the demographic, social, a n d e c o n o m i c Racial differences in the probability o f interstate mi- characteristics o f migrants. Comparisons o f the pro- gration based u p o n simple tabulations m a y not h o l d files o f movers a n d n o n m o v e r s have led to general- u p w h e n other influences are held constant. Aggre- izations about the relationship between m o v i n g a n d gate differences in interstate migration by race may, the life cycle. Several tendencies regularly reported for example, b e attributable to differences in educa- in Geographical Mobility, an a n n u a l report from the tion levels rather than racial propensity to migrate, Census Bureau, are illuminating, even t h o u g h they a n d the results are not always clear even in multi- are based o n simple univariate or bivariate tabula- variate m o d e l s . Similar a m b i g u i t i e s c o n f o u n d tions from U.S. p o p u l a t i o n surveys. 4 For instance, study of the effects o f gender or family status o n in- m o v i n g rates a p p e a r to b e related directly to age, terstate migration. the with persons in their twenties most likely to m o v e to another state. A secondary age spike in migration occurs a m o n g the very y o u n g ( w h o tend to be accomp a n y i n g parents in their twenties), a n d there is an uptick w h e n persons enter retirement. The census figures also s h o w that m o r e m e n tend to b e interstate migrants than w o m e n , in part because m e n are more often in the military, a g r o u p with a higher rate o f m o v i n g than p e o p l e in other occupations. Place Characteristics and Migration Labor market conditions such as j o b opportunities a n d available w a g e s in regions also contribute to w o r k e r s ' d e c i s i o n s to m i g r a t e . M o r e o v e r , local amenities, costs o f living, a n d fiscal characteristics Moving rates for civilians also vary by labor force are likely to influence migration flows, especially o f status. A m o n g the u n e m p l o y e d , migration is quite a retirees. 6 Researchers have identified several place bit higher than it is for persons with jobs. The lowest characteristics, discussed in this section, that affect m i g r a t i o n rate is for i n d i v i d u a l s n o t i n t h e l a b o r worker migration. force, such as retirees or students. A n o t h e r factor G r a v i t y E f f e c t s . Researchers w h o use m o d e l s correlated w i t h migration rates is e d u c a t i o n level. that include population a n d distance variables typi- Rates rise with educational attainment, p e a k i n g for cally find t h e m very robust in explaining migration. persons with four years o f college. Race is also a Brian J. Cushing (1986, 1987) f o u n d that population variable: non-Hispanic whites have higher rates o f a n d distance (as indicated b y beta coefficients in re- long-distance m o v i n g than Hispanics a n d blacks. gression equations) have the largest relative positive S o m e o f these differences are linked to e c o n o m i c a n d negative impacts, respectively, o n the d e p e n d e n t a n d life-cycle influences on migration patterns. For migration variable w h e n it is measured as the alloca- example, labor force members are less likely to mi- tion rate o f migrants from each state o f origin to grate as t h e y g r o w o l d e r b e c a u s e they e x p e c t a each o f forty-seven other coterminous states. Peter R. shorter remaining w o r k i n g life during w h i c h to real- Mueser (1989b), however, f o u n d that m a n y aspects ize the benefits o f migration. Similarly, the m o r e ed- o f the migration structure cannot be tied to geogra- ucation p e o p l e attain, the m o r e information a b o u t p h y e v e n t h o u g h m i g r a t i o n flows b e t w e e n states e m p l o y m e n t a n d job opportunities they are likely to d u r i n g the periods 1955-60, 1965-70, a n d 1975-80 re- have so that they are active in a job market more na- flect g e o g r a p h i c distance. His analysis also s h o w s tional in scope. that changes in migration streams, in particular, re- As Larry F„ Long has pointed out, "A rather parsi- sult from nongeographic p u s h a n d pull effects. m o n i o u s set o f categories ( j o b transfer, l o o k for C o s t o f L i v i n g . Presumably, distance influences work, take a n e w job, enter or leave a r m e d forces, migration in part because it serves as a proxy for a retirement, attend schools, be closer to relatives, or cost-of-transportation barrier to m o v e m e n t b e t w e e n to seek a change in climate) usually account for 70 places. 7 Mark C. Berger and Glen C. Blomquist (1989), or 80 percent o f the 'main' reasons for interstate or using migration data from the 1 in 1,000 Public 16 Use E C O N O M I C REVIEW, JANUARY/FEBRUARY 1991 Sample of the 1980 Census, found indirect evidence that moving costs are important in the decision to migrate. Their main focus of attention, however, was the significance of living costs. Their study suggests that people consider the cost of housing but that its importance diminishes with age, "presumably because of shorter time horizons" (2). Curiously, they found that the differences in moving costs grow in importance with age, "reflecting greater attachments to an area as one grows older" (2). State a n d Local Fiscal Structure William F. Fox, Henry W. Herzog, Jr., and Alan M. Schlottman (1989) studied the extent to which migration seems related to variations in local and state government fiscal policies. Using microdata for individuals, their empirical results indicate that high state or local gove r n m e n t taxes a n d o t h e r o w n - s o u r c e r e v e n u e enhancements have the expected effect o f discouraging in-migration, as does specific reliance on the income tax. Also as expected, good police and fire protection, as well as park and recreation facilities, are enticements; however, regional education conditions appear not to have a significant effect on in-migration. These researchers conclude that fiscal structure does seem to play an important role in the migration decision. I n c o m e . A finding c o m m o n to many migration studies is that income opportunities are an important pull factor. Statistically, this variable often has a large relative impact on the dependent migration variable as well as strong statistical significance. Thus, empirical findings support the theoretical view that alternative economic opportunities need to be considered in the analysis of migration. Simultaneity—Migration and Employment G r o w t h . Researchers have investigated the relationship between migration and employment growth using simultaneous equations techniques since Muth's pioneering work in the early 1970s. These models, estimated to allow employment growth and migration to be determined jointly with two-way causation, generally take the form Mi = f(Ejy X{) and Ej = g(Mf, Z-), where Mi and Ei are the dependent migration and employment variables and X( and Z / are vectors of location-specific factors affecting migration and employment, respectively. "In an average year two extra jobs attract about one additional net migrant, and one additional net migrant has a direct effect o n area e m p l o y m e n t o f a l m o s t 1.4 j o b s , " according to Greenwood, Gary L. Hunt, and J o h n M. McDowell (1986, 231). These findings mean that migrants d o not fully substitute for or take jobs from workers where employment is growing and suggest that migration causes job growth over and above employment of the migrants themselves. In the example above, only FEDERAL RESERVE BANK OF ATLANTA one-half of all newly created jobs are taken by inmigrants, while a newly employed in-migrant causes nearly one-half of an additional job to be created. They conclude that their results are very similar to those Muth obtained in his earlier work. They also find some evidence for the notion that migration increases during national business-cycle expansions and falls during downturns. R e g i o n a l W a g e Arbitrage. Stuart A. Gabriel, Janice Shack-Marquez, and William L. Wascher (1989) evaluated whether interregional migration arbitrages regional wage and unemployment differentials. Their results, based on simulation, suggest that interregional migration reduces regional w a g e differentials. However, their findings also indicate that the direct effects of migration are insufficient to narrow unemployment rate differences much within a period of less than several years. A s s o c i a t i o n o f In- a n d O u t - M i g r a t i o n Rates. Migration researchers have long noted the positive association between rates of in-migration and outmigration. An often-mentioned explanation for the association is that areas with extensive and sustained in-migration tend eventually to have populations that are more migration prone, thus increasing their outmigration rates. Some other explanations include migration's tendency to i n d u c e a counterstream of return migrants and the probability that residents' departure creates an environment conducive to inmigrants as jobs are vacated and homes b e c o m e available to newcomers. Mueser a n d Michael J. White (1989) explain the dynamic character o f the migration process. An area with high net migration gain experiences population gain. This gain increases the number at risk to depart while also raising the denominator of the in-migration rate. Unresolved Migration Issues Numerous factors make it difficult to derive precise estimates of the relationships between migration flows and the personal and place characteristics behind these flows. A s y m m e t r i e s . Researchers have noted that some motivation factors explain in-migration better than out-migration, while the reverse is true for other factors. For e x a m p l e , G r e e n w o o d (1975, 400) a n d Mueller (1982, 32) both note that several other researchers found that income and job opportunities generally provide a better explanation of in-migration whereas personal characteristics such as age and education seem to have more effect on migrant departures. From a theoretical economic perspective, it is not clear w h y migrants seem strongly p u l l e d to 17 destinations where employment growth is high but not pushed to the same extent from origins where employment growth is low. However, because o f these and other asymmetries, better empirical results are obtained from in-migration or out-migration models than from net migration models. (Another problem with net migration models is that the effect o f distance, for example, is neutralized because it has the same sign in gross in- and out-migration equations.) Yet, because o f data limitations, researchers often have had only net migration data to work with, at the cost of substantial loss of information and insight into the migration process. D a t a L i m i t a t i o n s . A problem with much migration research is that data refer to aggregate migration flows whereas the migration decision is an individual decision, often reflecting personal characteristics. As a consequence, the more readily identifiable place characteristics such as income or job growth have more often been cited as explanatory factors, and, by definition, the effects of personal characteristics have been underestimated. Fortunately, new microdata sets increasingly will allow for needed modeling refinements such as focusing on the decision-making unit and studying the life-cycle influences on migration decisions (including measurement of the economic return to migration). These data include the Panel Study of Income Dynamics, the National Longitudinal Survey, and Census Public Use Microdata Samples. Nevertheless, explaining aggregate migration will continue to be a goal because aggregate statistics are relevant for local governments and policymakers concerned with the consequences of migration. A n o t h e r data s h o r t c o m i n g has b e e n that researchers too often have had to rely on cross-section data rather than time-series information. This limitation has prevented study of inherent lags in migration such as the t i m e b e t w e e n an i n d i v i d u a l ' s recognizing and then acting on new job opportunities (time spent selling the old house a n d making other arrangements). Moreover, the determinants and consequences of migration both vary cyclically and in the long run, but, lacking appropriate timeseries data, researchers have not been able to explain h o w a n d why. Fortunately, new time-series state-level data from the Internal Revenue Service are helping to address this issue. M e t h o d o l o g i e s a n d M o d e l Specification. Several conceptual issues in the study of migration are unresolved. They are often linked to statistical methods or the specification of particular migration models. Mueser (1989a) has noted that cross-sectional studies may be seriously biased because the current geographic population distribution reflects previous migrants' choices. Mueser notes that population, of- 18 ten included in migration models to capture social or economic scale effects, may actually be a proxy for unmeasured characteristics such as place amenities, which might be stable and have been effective in attracting migrants earlier. Under these circumstances, population and the unmeasured variables will be correlated. If population and other explanatory variables also have a correlation not taken into account, complex statistical biases result. Because o f this interrelationship a m o n g several determinants of migration, researchers have turned increasingly to simultaneous-equations models of migration's causes. Often, though, data are inadequate for estimating migration models, for reasons beyond even those already mentioned. For example, it has been noted that variables in the gravity model—population and distance—are important factors in explaining migration. However, population and distance can be interpreted in different ways, depending on the size of the city or town in which one lives and normal travel distances. Moreover, sorting out the underlying factors these variables represent is not a straightforward exercise. For instance, population might be an indicator of economies of scale or agglomeration, meaning that larger population leads to lower production cost and thus concentration of activity. 8 In contrast, as mentioned above, population also could be simply a measure of an area's past success in attracting migrants. Similarly, although the fiscal structure of a region or amenities such as climate may cause migration, it is not clear which variables should be used in specific models. Generally, clear structural and behavioral models of the migration process are lacking in this area of economic research. At least a few other practical specification issues cloud statistical and conceptual understanding of migration relationships. Economic factors thought to be important in explaining migration behavior are often correlated with each other or with noneconomic migration influences. As a consequence, channels of influences among the factors are blurred statistically, a problem k n o w n as multicollinearity. Because the probability of migration is a function of levels of certain variables, such as movement cost, and changes in the absolute levels of explanatory variables, another practical difficulty is created. When changes in the absolute levels of variables and their levels at one place relative to all others are examined as determinants of migration, the issue of multicollinearity among variables is heightened. Statistical F i n d i n g s . Differences in methodologies used, model specification, and the functional fonn imposed on variables in migration models have resulted in some significant statistical differences of opinion concerning the direction and strength of in- ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 fluence of migration determinants. O n e of the most vexing and perplexing issues for migration scholars is the erratic finding concerning the regional unemployment rate's influence. W h e n viewed with the individual in m i n d , the notion that the u n e m p l o y e d are more likely to move than the employed has strong theoretical appeal and statistical support. However, at the aggregate level u n e m p l o y m e n t rates have frequently h a d nonsignificant coefficients a n d even unanticipated signs, suggesting either that this finding is eroded during the aggregation process or that the model specification or form of data used is faulty. As noted below, results from the author's empirical research suggest that model specification is key to the issue. Other empirical findings in dispute include the relative importance of jobs versus amenities in migration. This disagreement may be more a matter of difference in interpretation than a disagreement of substance, though. For example, G r e e n w o o d a n d Hunt (1989) conclude that location-specific amenities d o not appear to have an important direct effect o n migration w h e n employment growth is taken into account. They say that these results differ considerably from Graves's (1983) findings. G r e e n w o o d and Hunt agree with Graves that the influence of amenities is positive, but for them the primary impact of amenities is via the stimulation given to job growth. State and local government fiscal variables and climate variables are other influences for which either ambiguous statistical results have been found or for which interpretation of meaning or channel of influence is difficult to make. Author's Empirical Research In an effort to help develop insight into some of the unsettled issues in migration research, the author is analyzing n e w U.S. Internal Revenue Service estimates of historical state-to-state migration flows that are based on income tax returns. Preliminary statistical analyses of these data for the 1980-87 period are revealing. The summary results reported below are based o n regression equations estimated for each of the eight years. Altogether, these new data provide strong evidence concerning the influence of certain e c o n o m i c variables in determining migration a n d s h e d n e w light o n s o m e issues that h a v e b e e n cloudy. For example, this work has led to several conclusions: • A state's income growth, lagged o n e year to reduce simultaneity bias, positively influences inmigration. FEDERAL RESERVE BANK OF ATLANTA • Expected pay, measured as average state pay lagged one year and weighted by a state's une m p l o y m e n t rate, has a positive influence on in-migration. • The higher the cost of living index for a state, the less in-migration it will experience. • The greater the number of annual cooling degree days—the warmer the climate—the greater in-migration will be. • The greater the amount of in- and out-migration the state experienced in the 1975-80 period, the greater in-migration was in the 1980s. All of the directions of influence for these determinants are as expected. In addition, these results are very robust as measured by the stability a n d significance o f the statistical coefficients relating inmigration to these explanatory variables and the significance of the entire e q u a t i o n e x p l a i n i n g state in-migration. Altogether these variables explain from 96 percent to 99 percent of the variation in state inmigration for the eight years for w h i c h migration equations were estimated. It is encouraging that statistical results using this new data set conform so well with economic theory and with the central tendencies found in a review of the migration literature. For e x a m p l e , there is a strong t h e o r e t i c a l e x p e c t a t i o n that fast i n c o m e growth should attract migrants. Also, according to theory in-migration should tend to be toward regions that pay higher wages, other factors remaining equal. Important "other factors" that must be accounted for include the probability of getting a job (suggesting that pay should be weighted by the region's unemployment rate) and cost-of-living differentials (indicating that individuals respond to real, not nominal, wage differences). Additionally, it is obvious that amenities such as warm and sunny days should lure m i g r a n t s . Finally, m i g r a n t s are likely to i n f o r m friends and relatives about the merits and shortcomings of their n e w place of residence and, having experienced migration, are more likely to m o v e again themselves. Besides finding support for the factors thought to motivate migration, a few other noteworthy findings have surfaced in the author's work. O n e is that the determinants of in-migration d o not perform well in explaining state out-migration in the 1980-87 period. This discovery is consistent with the c o m m o n , yet somewhat perplexing, finding that economic conditions in destinations e x p l a i n in-migration better than such variables in origin localities explain outmigration. Another finding supportive of previous research is that there appears to be a tendency for migration to slow d o w n in recession and speed u p during economic expansions. This finding suggests 19 that migration can have countercyclical effects over is helpful, what is really needed for policy purposes the course o f the business cycle at the regional a n d related to r e g i o n a l e c o n o m i c d e v e l o p m e n t is to even national level. All in all, although the migration k n o w w h i c h areas individuals will choose. p u z z l e has yet to b e totally solved, it nevertheless Based o n the research reviewed in this article, it seems clear that this research supports important the- appears that e c o n o m i c d e v e l o p m e n t efforts that spur ories regarding motivations. i n c o m e growth in a region increase in-migration. I n addition, environmental policies a n d others that protect a n area's amenities w o u l d appear to have a payoff. Finally, state a n d local policies that keep d o w n Policy Implications and Conclusions the cost of living, including perhaps the p r o m o t i o n of affordable housing, can b e expected to meet with Because regional d e v e l o p m e n t is expected to b e success. influenced strongly by the mobility behavior of M u c h research into migration influences remains firms a n d i n d i v i d u a l s , u n d e r s t a n d i n g the determi- to b e d o n e . The relative importance of origin a n d nants o f their migration is important. Benefit to a re- destination characteristics in e x p l a i n i n g gion from e x p a n d i n g its export base, or its capacity still is not adequately k n o w n . Also, the full extent to to sell products a n d services outside the local econ- w h i c h migrant-attracting policies w i l l b e effective o m y , is a n underlying focus o f state a n d local pro- cannot b e quantified. Nor are the timing a n d extent grams d e s i g n e d to d r a w firms or jobs to an area. to w h i c h The assumption is that migration will follow at the k n o w n , a n d very little hard evidence exists concern- migration causes w a g e migration convergence same time that such programs will offer m e m b e r s o f ing other consequences o f migration. C o n t i n u e d re- the local w o r k force n e w job opportunities. search is needed that will increase o u r understanding Current research findings indicate that personal or o f migration's d e t e r m i n a n t s a n d its c o n s e q u e n c e s life-cycle influences strongly motivate individual mi- a n d lead to development of more effective ways to gration with or without reference to e c o n o m i c condi- address critical policy issues. tions at specific locations. A l t h o u g h this information Notes 1. See Isard (I960) for a thorough review of the gravity model. The simple formulation of the model that would be analogous to Newton's law of universal gravitation would be My = k(Pt * Pp/Cdy)2, where M{ - is migration between places i and j, Pi and Pj are the populations of i and j, dtj is the distance between them, and k. is a constant. Thus, the greater the population mass between two places, the greater will be gross migration, but the greater the distance, the less migration there will be. In modern work, functional rather than geographic measures of distance and population are used so that the distinction between gravity models and economists' econometric models incorporating influences such as labor market conditions, amenities, and so forth, has been blurred. 2. If the present discounted value of the earnings stream in place j compared to place i is X (En-Eit)K\ t=1 J + r)' and the comparable present discounted value of costs of being at these places is £ (C,-Cit)/(1 t=l J + r)', then movement to j from i will occur when earnings minus costs exceed zero. Individuals move to the place with the highest net present value(PK.), where 20 _ £ p v ij t=i (E >~ E "} _ £ (1 + r) 1 ,=i (C j'~Ci'} , (1 = r)' and E = earnings, C = costs, r = discount rate, and t = time period. 3. This issue, generally referred to as "factor-price equalization," has been central to the discussion of the effect of international trade as well. In the context of trade between nations, complete equalization of the prices of production factors internationally also requires mobility of the factors. Free trade of goods will only partially substitute for factor mobility and only tend to equalize relative factor prices internationally. 4. Primarily because of data limitations, generalizations about migration have tended to be drawn from crosssectional and aggregate population data rather than from information about individuals over time. 5. A recent paper by White and Mueser (1989) specifically tests for changes in the demographic determinants of U.S. residential mobility in the 1940-80 period using a multinominal logit model. Their results also are consistent with the literature. Educational attainment promotes migration, the influence of education declines with age, blacks are less migratory than whites, and women are less migratory than men. 6. As might be expected, the place characteristics that influence retiree migration differ from those that affect worker migration. Studies conducted by researchers at ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 Florida State University—Fournier, Rasmussen, and Serow (1988); Rasmussen, Fournier, and Charity (1989); and Serow (1987)—support plausible notions concerning differences in what workers and retirees are looking for in a new place. Labor market conditions such as employment opportunities, for example, are important to workers but not to retirees, whereas cost-of-living considerations are more important to retirees. Apparently, regional disparities in cost of living give elderly households on fixed incomes an incentive to move in much the same way that job opportunities motivate workers to migrate. Besides being sensitive to cost-of-living variations, the migration decisions of the elderly also seem to be especially attuned to amenity variables, such as warm, sunny weather and an absence of crime. 7. The flow of information between places may also lessen with distance, limiting migration flows. 8. Scale economies are internal economies in a firm's production at a given facility as its scale of operation increases while agglomeration economies are associated with economies that accrue to all firms in a single industry at a particular location. References Becker, Gary S. Human Capital. Princeton, N.J.: NBER, Princeton University Press, 1964. Berger, Mark C., and Glenn C. Blomquist. "GeographicMobility and Quality of Life." Paper presented at the annual meeting of the Population Association of America, Baltimore, Md., March 1989. Blanco, Cicely. "The Determinants of Interstate Population Movements." Journal of Regional Science 5 (Summer 1963): 77-84. Borts, George H., and Jerome L. Stein. Economic Growth in a Free Market. New York: Columbia University Press, 1964. Cushing, Brian J. "Accounting for Spatial Relationships in Models of Interstate Population Migration." Annals of Regional Science 20 (July 1986): 66-73. • "A Note on Specification of Climate Variables in Models of Population Migration." Journal of Regional Science 27, no. 4 (1987): 641-49. Fournier, Gary M., David W. Rasmussen, and William J. Serow. "Elderly Migration: For Sun and Money." Population Research and Policy Review 7 (1988): 189-99. Fox, William F., Henry W. Herzog, Jr., and Alan M. Schlottman. "Metropolitan Fiscal Structure and Migration." Journal of Regional Science 29 (November 1989): 523-36. Gabriel, Stuart A., Janice Shack-Marquez, and William L. Wascher. "Does Migration Arbitrage Regional Labor Market Differentials?" Unpublished manuscript, Board of Governors of the Federal Reserve System, Washington, D.C., May 1989. Graves, Philip E. "Migration with a Composite Amenity: The Role of Rents." Journal of Regional Science 23 (November 1983): 541-46. Evidence on the Spatial and Temporal Dimensions of the Linkage." Journal of Regional Science 26, no. 2 (1986): 223-35. Hicks, John R. The Theory of Wages. London: Macmillan Press, 1932. Isard, Walter. Methods of Regional Analysis: An Introduction to Regional Science. Cambridge, Mass.: MIT Press, 1960. Long, Larry E. Migration and Residential Mobility in the United States. N e w York: Russell Sage Foundation, 1988. Lowry, I.S. Migration and Metropolitan Growth: Two Analytical Models. San Francisco: Chandler, 1966. Mazek, Warren F. "Unemployment and the Efficacy of Migration: The Case of Laborers." Journal of Regional Science 9 (April 1969): 101-7. Mueller, Charles. The Economics of Labor Migration: A Behavioral Analysis. New York: Academic Press, 1982. Mueser, Peter R. "Measuring the Impact of Locational Characteristics o n Migration: I n t e r p r e t i n g CrossSectional Analysis." Demography 26 (August 1989a): 499-513. • "The Spatial Structure of Migration: An Analysis of F l o w s b e t w e e n States in the U.S.A. over Three Decades." Regional Studies 23, no. 3 (1989b): 185-200. Mueser, Peter R., and Michael J. White. "Explaining the Association between Rates of In-Migration and OutMigration." Papers of the Regional Science Association 67 (1989): 121-34. Muth, Richard F. "Migration: Chicken or Egg?" Southern Economic Journal 37 (January 1971): 295-306. Graves, Philip E., and Peter Linneman. "Household Migration: Theoretical and Empirical Results." Journal of Urban Economics (April 1979): 383-404. Rasmussen, David W., Gary M. Fournier, and Douglas A. Charity. "The Impact of Cost of Living Differentials on Migration of Elderly People to Florida." The Review of Regional Studies 19 (Spring 1989): 48-54. Greenwood, Michael J. "Research on Internal Migration in the United States: A Survey." Journal of Economic Literature 13 (June 1975): 397-433- Serow, William J. "Determinants of Interstate Migration: Differences between Elderly and Nonelderly Movers." Journal of Gerontology 24, no. 1 (1987): 95-100. • "Human Migration: Theory, Models, and Empirical Studies." Journal of Regional Science 25 (November 1985): 521-44. Greenwood, Michael J., and Gary L. Hunt. "Jobs versus Amenities in the Analysis of Metropolitan Migration." Journal of Urban Economics 25 (January 1989): 1-16. Greenwood, Michael J., Gary L. Hunt, and John M. McDowell. "Migration and Employment Change: Empirical Sjaastad, Larry A. "The Costs and Returns of Human Migration." Journal of Political Economy 70 (Supplement, October 1962): 80-93. FEDERAL RESERVE BANK OF ATLANTA White, Michael J., and Peter R. Mueser. "Changes in the Demographic Determinants of U.S. Population Mobility, 1940-1980." Paper presented at the annual meeting of the Population Association of America, Baltimore, Md., March 1989. 21 The 1990 Farm Gene D. Sullivan ongress r e s p o n d e d to the c h a l l e n g e o f farmers g r o w i n g supported crops. Each must choose cutting the cost of federal farm programs what crop to g r o w o n the newly freed land. Their C by building into the 1990 Farm Act an es- c o m b i n e d adjustments will influence not only their timated $13.6 billion cut in farm subsidies support payments received but also total production over the next five years. Most savings came from the o f various crops, w h i c h will in turn affect prices a n d addition o f a provision to the system o f target prices the actual federal expenditure savings resulting from continued from the F o o d Security Act of 1985 (also the n e w bill. These issues will b e more fully consid- k n o w n as the 1985 Farm Bill). In addition to reduc- ered later in this article. - ing estimated costs, this feature—called the triple- This article is intended to b e a s u m m a r y o f the base o p t i o n — h e l p s to improve the farm program's background a n d development o f the 1990 Farm Bill m a r k e t o r i e n t a t i o n b y f r e e i n g 15 p e r c e n t o f t h e a n d the m a i n features that distinguish it from earlier acreage in subsidized crops for unsupported produc- legislation. F o l l o w i n g a discussion o f the general tion o f any crop (except for fruits a n d vegetables). Other features o f the 1990 a n d 1985 bills are simi- framework of the bill, its major provisions for specific commodities are outlined. More general provisions lar. However, in the 1990 bill target prices are frozen are also discussed. I n the concluding sections impli- at that year's levels after h a v i n g b e e n reduced in cations o f the legislation are examined. stages by the 1985 bill. Generally, price support levels are to b e a d j u s t e d e a c h c r o p m a r k e t i n g year based o n stocks-to-use ratios. T h o u g h it enhances market orientation, the triple- Controlling Government Costs base o p t i o n presents a c o m p l e x set o f choices to Under the previous (1985) a n d current legislation, i n c o m e s u p p o r t p r o g r a m s for farmers h a v e b e e n The author is a research officer in the regional section of the Atlanta Fed's research department. 22 geared to a concept of target prices that was based o n direct payments to support incomes rather than ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 on supporting prices at levels that frequently required the government to purchase farm commodities at prices well above market clearing levels. Farmers have the option to sell their commodities at prevailing market prices or obtain a nonrecourse loan from the Commodity Credit Corporation (CCC) at a specified price per unit, usually set below the expected market price level. Direct government payments to farmers make u p the difference between the target price and the higher of the CCC loan rate or the average market price during a specified period. This deficiency payment program encourages the flow of commodities into the marketplace, thereby reducing or avoiding government storage costs. In the past the government sometimes held commodities for many years because they could not be marketed at or above their acquisition cost. Stored commodities were eventually disposed of in gifts to impoverished foreign countries or to needy groups at home, or they were sometimes bartered to countries that lacked foreign exchange but had products they could trade. The target price and deficiency payment program is intended to help avoid such losses to the government. G o v e r n m e n t payments to farmers have themselves become an expensive item in the budget of the U.S. Department of Agriculture largely because of the wide differences between target and market prices. Payments reached a peak o f $16.7 billion in 1987, compared with direct outlays to farmers totaling SI.3 billion in 1980. Increases in market prices reduced direct payments after 1987, but estimates for 1990 indicate farmers again received a relatively high level o f support—about $12 billion. Without program alterations it is estimated that payments for 1991 would have moved up sharply from 1990's level, largely because of weakening market prices for wheat and feed grains as stocks are rebuilt from the shortages 1988's major drought in the Midwest and Upper Plains engendered. P o l i c y Pressures. Negotiations for the 1990 Farm Bill began under the specter of several unusual pressures. O n e was the General Agreement on Tariffs and Trade (GATT) negotiations in which then U.S. Secretary of Agriculture Clayton Yeutter had been campaigning vigorously for the removal o f tradedistorting agricultural subsidies t h r o u g h o u t the world. The success of this endeavor would have produced revolutionary changes in American farm policy because policy is heavily based on price and income support mechanisms that presumably w o u l d have been revamped or possibly eliminated. Congress decided to proceed with the 1990 Farm Bill under the assumption that total subsidy cuts were unlikely (an assumption that proved correct when the GATT negotiations subsequently broke down). Thus, the ini- FEDERAL RESERVE BANK OF ATLANTA tial proposals for the 1990 legislation were rather closely modeled after the expiring 1985 Farm Bill. An additional pressure on bill writers was the need to develop legislation in line with the budget agreement b e i n g crafted by other congressional committees. In fact, the failure of Congress to accept the budget summit agreement on October 5, 1990, left the House-Senate conference (at that time attempting to negotiate a final version of the Farm Bill from the separate bills passed by the House and Senate) without specific guidelines or goals for cutting crop subsidies. As it turned out, the estimated $13 billion cut required by the budget summit agreement remained the requirement of subsequent budget negotiators as well. S a v i n g s E s t i m a t e s . E x p e n d i t u r e cuts w e r e achieved largely by reducing the amount of land on which farmers could collect subsidy payment by 15 percent. The cutbacks were made more palatable to farmers by giving them broad flexibility in crops they are permitted to grow on the nonsubsidized acreage. It is assumed that farmers, on balance, will not change acreages and production of crops in such a way as to increase government outlays. Although production of individual crops, such as cotton and soybeans, could expand and depress market prices so that deficiency payments or CCC outlays would increase for those commodities, the corresponding reduction in grain production presumably would lift market prices and reduce deficiency payments on grain acreage. The overall effect is expected to keep net government expenditures at levels that will maintain the estimated reductions o f $13.6 billion over the five-year period of the legislation. The savings estimate is also based on assumed continued strength in domestic and foreign demand for agricultural commodities. Because target prices have been frozen, market or support price fluctuations will directly affect costs of government agricultural programs. Provisions for Major Commodities The new bill's major provisions for specific commodities are as follows: W h e a t . The target price for wheat is to be frozen at $4 per bushel for five years. The old legislation had allowed the m i n i m u m target price to decline each year since 1987, moving d o w n from $4.38 to $4 per bushel in 1990. (See Chart 1, which compares target prices for the five major commodities in the first production season under the 1985 and 1990 bills.) The rate at which the CCC would lend money 23 Chart 1 Target Prices 1985 and 1990 Farm Bills* Dollars per Unit Wheat * For the first production Source: $11.90 Rice Sorghum Cotton U.S. Department of Agriculture, Economic Research Service, The 1990 Farm Act and the 1990 Budget Reconciliation December 1990, 6. to farmers on wheat (the price support loan or loan rate) is set at 85 percent of a five-year moving average of market prices. Previously, the loan rate had been set in the range of from 75 to 85 percent of the five-year average. The objective of this provision is to ensure that in most years the loan rate would be far enough below the market price at some period of the year to encourage wheat to flow into the market rather than remain in storage. If stocks should build up anyway because of unusual price weakness, the Secretary of Agriculture is given the authority to reduce the loan rate as much as 20 percent, as he was also allowed to do under previous legislation. Deficiency payments are to be calculated on a twelve-month average market price rather than the average of the first five months of the marketing year, as was the case for previous legislation. This change will probably have the effect of reducing the average deficiency p a y m e n t to farmers because prices are typically lower during the first five months of the marketing season when the crop is being harvested than is the average price for the whole year, which w o u l d include prices during periods when supplies are likely to be drawn down (see Table 1). Feed G r a i n s . The 1990 target price for corn of $2.75 per bushel is frozen for a five-year period (see Chart 1). Like wheat, the loan rate is set at 85 percent of a five-year moving average of market prices 24 Corn season under each bill. Act instead of in the range of from 75 to 85 percent of the average price. The secretary would also have the authority to reduce the support price by as much as 20 percent during a year when stocks build to high levels. Deficiency payments for corn will be calculated on a twelve-month average of market prices for the 1994 and 1995 crops, rather than a five-month market season average price. The latter average will continue to be used until 1994. Target prices for other feed grains are set at $2.61 per bushel for grain sorghum, $2.36 per bushel for barley, and $1.45 per bushel for oats. Loan rates for these commodities are to be held in a fair and reasonable relationship with corn. C o t t o n . The expiring legislation had reduced the target prices for cotton in annual increments from 81 cents per pound in 1986 to 72.9 cents in 1990. The new bill freezes the target price at that amount for five years. The loan rate for cotton will continue to be set at 85 percent of a five-year moving average of market prices. However, cotton producers will continue to have an option of repaying less than the loan amount to the CCC. A "marketing loan" concept is in use under which the producer can repay a commodity loan at the unit price earned at market. For example, if the prevailing world market price is below 80 percent of the loan rate, a loan repayment level may be set not in excess of 80 percent of the ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 Table 1 Commodity Price Averages* Cotton Marketing Year Average Five-Month Average 1985 2.23 2.41 0.57 1986 1.50 1.69 1987 1.94 1988 1989 Wheat Marketing Year Average Five-Month Average Marketing Year Average Five-Month Average 0.56 3.08 3.00 6.53 7.74 0.52 0.50 2.42 2.31 3.75 3.87 1.74 0.64 0.65 2.57 2.45 7.27 5.92 2.54 2.34 0.56 0.55 3.72 3.61 6.83 6.96 2.33 2.27 0.63 0.63 3.72 3.77 7.30 7.25 * The "marketing year average" as follows: corn, September The "tive-month average" Marketing Year Five-Month Average Average Rice column shows the weighted average price for the crop sold during the marketing year defined 1 to August 31; cotton, August 1 to July 31; wheat, June 1 to May 31; rice, August column shows the average price during the first five months of the commodity's Source: U.S. Department of Agriculture, Agricultural 1 to July 31 marketing year. Prices Monthly and Annual Reports. loan rate. This provision may sometimes bestow substantial financial benefits to growers in addition to the direct deficiency payments collected. Rice. The rice program's general provisions are the same as those for cotton. The target price for rice will be $10.71 per hundredweight (cwt) for five years. As with corn, beginning in 1994 and continuing through 1995 deficiency payments for rice are to be calculated on a twelve-month average market price rather than a five-month average, which will continue in use until that time. O i l s e e d s . Target prices are not p r o v i d e d for oilseeds because they are not included in the socalled basic program crops. The support price for soybeans has been set at $5.02 per bushel for the five-year period of 1991 through 1995, less a 2 percent loan origination fee to be deducted from the proceeds at the time the loan is granted. Producers will also be provided a marketing loan that allows CCC loans to be repaid at the smaller of the loan rate or the prevailing world market price for soybeans. Other oil seeds, including sunflower seed, canola, flax, and safflower, will have loan rates set at levels comparable to soybeans. The Secretary o f Agriculture may also establish loan programs for additional oilseed crops as he deems necessary. D a i r y P r o d u c t s . The price support level for milk will be frozen at $10.10 per cwt of product with 3.67 percent butterfat content. This move interrupts the trend of declining support prices that had been in process since 1986, w h e n the support price was $11.60 per cwt. Previous legislation had specified that price supports would decline periodically to reduce the incentive for surplus milk production. The CCC was required to buy manufactured milk products in sufficient volume to keep the price of milk from dropping below the legislatively specified minim u m level. If surplus production levels should decline sufficiently during the life of the new legislation, price support levels might increase. To make producers share a portion of the program's cost, the new legislation requires an assessment of 5 cents per cwt o f milk produced in 1991 and 11 cents per cwt for 1992 through August 31, 1995. However, a producer may apply for a refund of this assessment by certifying that his o w n milk production has not increased from the previous year. To further defray program costs, the Agriculture Department may level assessments on farmers to recover the full costs o f purchasing surplus dairy products in excess of 7 billion pounds a year, provided legislation has not been enacted by January 1, 1992, to limit dairy surpluses. For 1989 and 1990, net supply removal by CCC purchases is indicated to have totaled 9.0 billion and 8.5 billion pounds, respectively, suggesting that surpluses are likely to FEDERAL RESERVE BANK OF ATLANTA 25 continue to exceed the trigger point for assessments unless new dairy legislation is enacted. Sugar. The sugar program has operated basically to restrict imports to quota levels in order to maintain domestic prices at or above the support price level. Although previous legislation allowed the support price to rise to reflect increases in the cost of production, the new bill has frozen the price-support loan for sugar at 21 cents per pound, and the legislation specifies that the programs continue to operate at no cost to the government. Duties on imported sugar have covered the government's cost of operating the p r o g r a m . H o w e v e r , imports have b e e n shrinking over time, and domestic production has expanded to supply a larger share of total consumption. In addition, sugar substitutes and fructose (sugar from corn) have been claiming growing shares of the market. To preserve a minimum level of imported sugar, the new legislation requires domestic marketing controls on cane and beet sugar if imports fall below 1.25 million short tons (about 18 percent of total domestic production). The growth of corn sugar marketings is also controlled by the imposition o f a 200,000 ton sugar-equivalent limit on annual marketings. To ensure that sugar producers share the burden of the mandatory reductions in costs of government programs along with other major crop producers, a 1 percent fee will be assessed on sugar loans from the CCC. General Provisions The new legislation contains a number of other provisions that are not c o m m o d i t y or producergroup specific. P a y m e n t l i m i t a t i o n s . A limitation on the amount of payments that an individual producer can receive is carried forward from previous legislation with some changes. The maximum amount a fanner may be paid from all government programs has been cut in half, from §500,000 to $250,000. The expiring legislation's limit of $50,000 per producer on direct deficiency payments is maintained. However, payments in larger amounts may be secured from such programs as disaster relief when crops are destroyed by drought or other calamities and from marketing loan gains. Nevertheless, the total of these payments may not exceed $250,000 per farmer for the major commodities listed above. For producers of honey, wool, and mohair, the n e w legislation specifies a p a y m e n t limit o f $200,000 in 1991, decreasing to $125,000 by 1994. The stricter limits for these producers impose a way they will share in the reductions in program costs. 26 F o o d f o r Peace. The Food for Peace (P.L. 480) program, the chief instrument for delivering food aid abroad and helping develop markets for U.S. farm products, has been extended in the new legislation, with a few changes. The authority for grant programs, including a new government-to-government grant program, is specifically assigned to the Agency for International Development, with the Department of Agriculture retaining responsibility for concessional sales of products abroad. The 1985 act contained a provision authorizing commodity donations to support agricultural policy reform in developing countries. The new legislation restricts use of the program for foreign policy purposes and attempts to direct commodities to countries with the greatest need and market potential. To assist the U.S. maritime industry, previous legislation specified that 80 percent of the gross tonnage of agricultural commodities exported under the P.L. 480 program must be transported on U.S.flag commercial vessels. The new legislation allows foreign vessels to reflag as U.S. ships to qualify as carriers under the existing "cargo preference" requirements for certain ports. E x p o r t E n h a n c e m e n t . Previous legislation required the Secretary of Agriculture to provide $325 million annually, or the equivalent in CCC commodities, for export assistance to counter adverse effects of a foreign country's unfair trade practices. The program involved selling export certificates to commercial exporters o n a c o m p e t i t i v e b i d basis. The certificates could be redeemed for CCC-owned commodities that could be exported at the prices necessary to p e n e t r a t e f o r e i g n m a r k e t s . The n e w legislation enlarges the Export Enhancement Program subsidies to not less than $500 million per year. In addition, a Marketing Promotion Program is funded at not less than $200 million per year to target markets where unfair trade practices are cited. Legislation viewed these measures as additional weapons for use against countries that continue to refuse to reduce agricultural trade impediments. C o n s e r v a t i o n . Growing environmental concerns are largely responsible for a number o f new and revised stipulations of the conservation section of the new legislation. Previous legislation provided penalties intended to discourage the expansion of agricultural production on land defined as highly erodible or as wetlands, which are important wildlife habitats. The new legislation enlarges previous programs by creating an Agricultural Resources Conservation Program to protect these fragile lands as well as land that in cultivation may contribute to water pollution. The new program includes three separate components: the existing Conservation Reserve Program, which assists owners and operators in conserving E C O N O M I C REVIEW, JANUARY/FEBRUARY 1991 and improving soil and water resources on highly erodible land; a n e w Wetlands Reserve Program; and a new Water Quality Incentives Program. The Conservation Reserve Program specifies that by 1995 not more than 45 million or less than 40 million acres of environmentally sensitive lands are to be enrolled in the program. The new Wetlands Reserve Program authorizes the subsidized enrollment of u p to one million acres of wetlands into easements of thirty years or longer, with priority given to wetlands that enhance wildlife habitat. Landowners may participate in this program on a voluntary basis. The new Water Quality Incentives Program would pay farmers u p to $3,500 a year to carry out governmentapproved plans to reduce water pollution, and more if wildlife habitat is also improved. The program specifies an enrollment goal of 10 million acres. C r e d i t . During the latter half of the 1980s the Farmers H o m e Administration experienced severe problems with loan delinquencies and defaults. For a period of time the agency was prevented from disposing of this property, and the foreclosed property inventory ballooned. That development motivated the authors of the 1990 agricultural legislation to include stipulations to address the Farmers H o m e Administration's property inventory problems. The new legislation calls for disposing of the land within one year. Leasers o f property acquired by the Farmers Home Administration before January 6, 1988, would have the right o f first refusal on the sale of that property. Purchase preference would also be given to beginning farmers and ranchers. Write-downs o f both loan principal and interest have been used to bring a borrower current on his indebtedness to the Farmers H o m e Administration. The n e w legislation limits borrowers to a single write-down for loans made after January 6, 1988. It also places on individual borrowers a lifetime cap of $300,000 in principal and interest, a stipulation that grew out of losses to some unusually large fanning operations during the late 1970s and early 1980s. R u r a l D e v e l o p m e n t . In conjunction with the liquidations and consolidations of farming operations during the period of unusual financial stress on the farm sector during the 1980s, the economies of rural communities shrank and, in some cases, virtually disappeared. This development has motivated afflicted regions to secure government assistance to preserve and develop rural economies. The new law contains provisions to create a Rural Development Administration within the Agriculture Department. O n e of its specified functions is to establish in as many as five pilot states panels of state officials w h o will rank state and local applications for Farmers H o m e Administration funding through that agency's loans for community development. FEDERAL RESERVE BANK OF ATLANTA Implications of the Farm Bill The major change in the 1990 Farm Bill is the triple-base option, which calls for the division of a farmer's acreage base into three categories. As already noted, that provision reduces the acreages of program crops on which farmers can receive deficiency payments by 15 percent, but it allows farmers the freedom to plant the crop of their choice (except fruits and vegetables) on those acres. It also permits farmers to remove voluntarily an additional 10 percent of the base acreage program crop production under the same stipulations. Farmers may still be required to remove (idle) some acreage from all crop production to qualify for program benefits according to the annually announced Acreage Reduction Program based on stocks-to-use ratios of the specific crop in question. The portion of the crop base left is the permitted acreage on which the program crop is grown. Fach permitted acre o f the crop is eligible for income deficiency payments as previously described. The major u n k n o w n impact of the new legislation concerns what farmers will actually d o with the 15 percent of their acreage base the new legislation has freed. Farmers can be expected to make individual decisions based on the productivity of their land (expected yields) in various crops and the expected price once a particular commodity has been produced. Farmers w h o select options different from the majority o f their counterparts will p r o b a b l y make the most rewarding choices. For example, if most corn farmers choose to switch their triple-base acreage to soybeans, a substantially larger harvest might well push d o w n the price of soybeans while reduced production o f corn might drive its price u p so that farmers w h o continue to produce corn on their freed acreage could end u p receiving a higher income than those w h o switch. In that case, deficiency payments to corn producers w o u l d shrink because of the rise in market prices. In the South, reduced winter wheat and rice production in favor o f increased soybean and cotton output could have the same result. That is, those w h o stay with the base program crop might receive higher returns after all. O b v i o u s l y , s u c h p o t e n t i a l rewards for second-guessing competitors make it unusually difficult to forecast what most farmers will d o in 1991 and subsequent years. In addition, the USDA has not released detailed rules and regulations governing all of the specific commodity programs. These stipulations could also influence farmers' decisions in the coming crop year. Indications thus far are that acreage shifts may be relatively small for the nation as a whole. The 27 American Farm Bureau has projected that total cotton and wheat acreages are unlikely to change m u c h as a result of the 1990 Farm Bill, and relatively minimal changes are projected for corn and soybeans. 1 A n n u a l corn acreage may decline by a n average 800,000 acres, a n d a n n u a l s o y b e a n acreage will probably increase by an average 1.2 million acres over the five-year life of the program. As a result, the level of soybean production could rise by 30 million bushels (1.5 percent) and the amount of corn output diminish by 90 million bushels (1 percent) during that time. The reduced output should raise the fiveyear average corn price by 4 cents per bushel (1.8 percent) from what it w o u l d have been under previous legislation. Increased soybean output is expected to lower the five-year average soybean price by 31 cents per bushel (5.1 percent) from the baseline forecast. The bill should have little impact on cotton and wheat prices. The 15 percent reduction in acreage eligible for government payments will cut producers' program crop returns. Farmers w h o produce other more marketable crops o n acres n o longer eligible for deficiency payments could recoup at least some of that loss. However, if market prices of crops decline, the government's deficiency payments would have to increase to maintain income levels o n program base acres. For example, if corn production does not decline as anticipated, prices could fall from the approximately $2.25 per bushel level at year's end 1990 to the projected loan rate of $1.90 per bushel, necessitating increased payments o n program acreage. In addition, CCC outlays could grow as farmers repay certain commodity loans at market prices that have dropped rather than at the higher loan rate. If, as expected, a significant amount of triple-base acreage is shifted to soybean production so that prices drop, soybean producers would be the ones most likely to exercise the marketing loan option, but the price w o u l d have to decline by more than the 31 cents per bushel projected by the Farm Bureau to drop the year-end 1990 price of about $5.60 per bushel below the $5.02 per bushel loan rate. The likelihood that deficiency payments will increase over the next five-year period is somewhat reduced by the program's movement to the twelvemonth-marketing-year average as the base for calculating payments. Table 1 shows that for corn in the 1985-89 period, the marketing-year average price was higher than the five-month average in three out of the five years. For wheat, the marketing-year average was higher in four out of five years (the exception occurred in 1989, w h e n the five-month average was slightly higher). Corn a n d wheat account for the major share of the government's deficiency payment expenditures. 28 If price relationships during the next five years remain approximately the same as during the last fiveyear period, using the marketing-year average base could cut deficiency payments an additional 1.5 to 2.5 percent. Because the target price is frozen for the term of the n e w legislation, however, a plunge in commodity prices w o u l d require that deficiency payments increase substantially, regardless of the market price series used. The new program has been designed to keep American grains competitive in world markets, though, so total d e m a n d for grains is expected to remain strong e n o u g h to prevent major price declines. The American Farm Bureau projects modest increases in corn and wheat prices during the five years of the program. Whether government costs rise or not, farmers in s o m e regions fear d a m a g e from potential m a j o r acreage shifts. For example, southeastern soybean farmers worry that midwestern corn growers might plant soybeans instead of corn on their nonsubsid i z e d acreage because the s o y b e a n c r o p may promise greater marketplace returns in their region. If prices fall because of extra soybean production, southeasterners w o u l d fail to cover their relatively high production costs. For these farmers, few profitable alternatives are available for a good portion of the poor land n o w devoted to soybeans. Southeastern farm spokesmen point out that the region's farm failures might well increase in the wake of the newlegislation. Conclusion The 1990 Farm Bill has the overriding objective of cutting the government's costs for agricultural programs, costs that had grown large and threatened to grow m u c h larger if allowed to continue on their course. The chief cost-cutting measure is the triplebase feature, which in effect reduces the acreage on which eligible crop producers can collect income deficiency payments. Such farmers n o w have a third category of land in addition to the acreage on which they receive subsidy payments and the acreage the government can require them to idle as a condition of receiving these payments. O n land in this third category, farmers can plant whatever crop offers the greatest return in the marketplace (except fruits and vegetables) and still preserve the acreage history for the basic crop. If market prices for the crops g r o w n o n triplebase acreage remain high or increase, farmers' incomes m a y not d e c l i n e m u c h as a result o f the program change. However, market prices could sag under the weight of additional production and possi- ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 bly softening d e m a n d abroad as world production o f crops increases. If farm incomes d r o p sharply u n d e r the n e w program, Congress will c o m e u n d e r intensifying pres- A reduction in market prices a n d support prices sure to raise subsidies again. O n the other h a n d , the w o u l d lead to increased government expenditures to l i k e l i h o o d o f c o n t i n u i n g excessive b u d g e t deficits maintain i n c o m e s o f farmers, w h i c h w o u l d reduce the estimated savings projected for the n e w legisla- will almost certainly stiffen resistance to enlarging s p e n d i n g programs. tion. O n the other h a n d , if d e m a n d s h o u l d prove stronger than expected, g o v e r n m e n t s u p p o r t costs c o u l d decline a n d a greater budget savings than the $13-6 billion projected w o u l d be realized. FEDERAI. RESERVE BANK OF ATLANTA Note 1. The American Farm Bureau Federation, Economic Review 7 (October 1990): 1-6. 29 Book Review Manias, Panics, and Crashes: A History of Financial Crises, revised edition. by Charles P. Kindleberger. New York: Basic Books, Inc., 1989. 320 pages. $21.95 (cloth). $12.95 (paper). A s the extensive set o f sources cited in C h a r l e s P. K i n d l e b e r g e r ' s tests, f i n a n c i a l crises h a v e The author, w h o s e grasp o f the subtleties of inter- at- national e c o n o m i c relations has long been appreciat- claimed ed, has m a d e the case for an international lender of book e c o n o m i s t s ' a n d historians' intermit- last resort in m a n y forums. I n Manias, Panics, and of Crashes, Kindleberger's basic thesis is again that the well-publicized gyrations in financial markets b o t h w o r l d needs a central bank, but the large n u m b e r o f here a n d abroad in the past few years, the expan- stories he tells to illuminate this idea often serve to sion phase o f a n e w cycle o f interest seems to b e mask the message. His arguments are less succinct occurring. Kindleberger's b o o k , in the revised edi- a n d m o r e colorful t h a n the straightforward recount- tent a t t e n t i o n for c e n t u r i e s . I n the a f t e r m a t h tion reviewed here, is o n e o f several recent systemat- ing o f t h e m in this review might suggest. D r a w i n g a ic considerations. broad descriptive outline o f the process of m a n i a , The d e v e l o p m e n t s o f the 1980s have, it seems, panic, a n d crash from the w o r k o f H y m a n Minsky a d d e d evidence to support Kindleberger's view o n (1977, 1982), the A m e r i c a n economist best k n o w n the progress a n d impacts o f financial crises. His ar- for explorations o f financial fragility, Kindleberger g u m e n t has several basic points. Financial crises, he p u t s m e a t o n the b o n e s o f M i n s k y ' s stylized de- contends, occur at o d d times. Such events may cause s c r i p t i o n w i t h e x a m p l e s f r o m a p l e t h o r a o f sec- real harm in the economies in w h i c h they take place. o n d a r y sources covering financial panics from the I n a d d i t i o n , financial crises m o r e a n d m o r e often S o u t h Sea B u b b l e o f 1720 t o t h e s t o c k have global dimensions. Kindleberger emphasizes crashes o f 1987. market that central banks can alleviate s o m e of these events' I n Kindleberger's version o f Minsky's m o d e l , fi- costs. Furthermore, these crises' global implications nancial fragility results m a i n l y from high levels of a n d the absence o f a strong international financial d e b t b o r r o w e r s h a v e t a k e n o n to b u y assets for p o w e r call for the establishment o f a n international short-term profit. Following Minsky, Kindleberger di- lender o f last resort to maintain the stability o f the vides the events o f a crisis into several stages. The world's economies. first is a "displacement," an event that changes profit 30 ECONOMIC REVIEW, JANUARY/FEBRUARY 1991 opportunities in at least o n e important sector of the economy. W h e n businesses and investors flock to these opportunities, the second phase—the b o o m — gets under way. The b o o m is typically fueled, at least in part, by expanding bank credit. A b o o m often progresses into euphoria, a phase characterized primarily by investors' buying assets for short-term resale, often o n the basis of overestimations of future profits. This speculation results in further credit expansion a n d leveraging. Euphoria draws in people w h o d o not usually enter speculative markets; such investors also have o v e r b l o w n profit expectations. Kindleberger calls the stage that ensues from euphoria a mania or a bubble, implying that the phase's speculative activities are irrational and precarious. U n d e r Kindleberger's scenario, a crisis begins w h e n some investors, emerging from euphoria, begin to cash in o n their speculations and prices resist further increases. After a period of "financial distress," Kindleberger explains, speculators realize that prices may not always rise. Then panic selling sets in, asset prices fall, bankruptcies increase, lenders fail, credit becomes unavailable, and even projects with high projected returns find n o lenders. Panicfeeds itself until it burns out or is extinguished by some policy response. Ultimately, real output and welfare decline as a result of the crisis. After delineating the model, Kindleberger embarks on a series of chapters that flesh out his outline. He draws stories from a 275-year series of financial crises. This anecdotal method, the author contends, is an appropriate technique for conveying a sense of events that are unpredictable, chaotic, and unique. Kindleberger's stories, selected and written with wit and a sense of the bizarre, are a trove of information o n the varieties o f financial panics' p r o p a g a t i o n , spread, and resolution through the years. They show clearly that events similar enough to fit the term "financial crises" have occurred with some frequency but have also been quite varied in all their aspects. As with most models that concentrate o n stages, Kindleberger's stylized description is somewhat unsatisfactory. H e tends to fit his stories into the model's stages rather than to probe systematic relat i o n s h i p s that m i g h t e x p l a i n the d y n a m i c s that drive the e c o n o m y from stage to stage. The m o d e l is presented as a one-way progression with little explicit consideration of the role of the real (or nonfinancial) e c o n o m y or expectations about the real economy. For example, Kindleberger clearly considers the initial displacement stage exogenous. Booms sometimes m o v e to euphoria; sometimes they d o not. In addition, n o variables in the model account for transitional phases from one stage o f a financial crisis to another. Some of the explanations the mod- FEDERAL RESERVE BANK OF ATLANTA el does provide, such as a discussion of the timing of financial distress, are inadequate. Furthermore, in considering the impacts of financial crises o n the rest of the economy, Kindleberger fails to apply the insights of recent macroeconomic thinking, particularly the perspectives of real business cycle and rational expectations theorists. H e assumes without discussion that real costs occurring after a crisis are the result of irrational behavior during the crisis. He might well have considered the insights o f economists such as Gary Gorton (1984), Charles Jacklin and Sudipto Bhattacharya (1988), and Ellis Ta 11 m a n (1988), w h o relate behavior d u r i n g b a n k i n g panics to the public's expectations about real p h e n o m e n a in an ensuing recession or depression. Their approach makes the crisis at least partially a function of expected future real losses rather than a cause of these losses. Such a view undermines Kindleberger's arguments for intervention in the latter stages (financial distress and panic), calling, rather, for policies designed to maintain a stable economy. The author's failure to provide adequate explanations o f his model and to consider alternative app r o a c h e s ultimately detracts from the b o o k ' s credibility. Despite the marvelous stories, o n e begins to w o n d e r why, if this broad model of financial crises' progression is truly systematic, documentation for each phase o f the process is selected from only a sample of historical crises. Is the selection criterion only that the tale fits the model? Could o n e support a different m o d e l with a different set o f anecdotes from the same universe of crises or, as Gorton's work implies, from the same set of samples? Kindleberger a p p e n d s two observations from his o w n earlier w o r k to the general descriptions o f Minsky's model. These points are vital to his argument for an international lender of last resort. In the most important o f these he "opens" Minsky's o n e - e c o n o m y m o d e l by noting that history gives many examples of situations in which manias, panics, crashes, and their effects have been transmitted a m o n g national economies, becoming international in scope. Interestingly, he points out, the impacts of internationally transmitted panics have not always been the same in different countries. For example, a building b o o m in Germany and Austria, which the author interprets as mania, removed credit from the U.S. economy in 1871 and 1872, restraining credit to American railroad speculators and contributing to a national panic. Kindleberger also argues that international transmission of crises has expanded in recent years with the growing integration of the world's economy, but he admits that, even as long ago as the 1720 South Sea Bubble, crises had international implications. 31 T o s u p p o r t his a p p e a l for a lender o f last resort sively? W o u l d it not b e m o r e effective for n a t i o n a l the a u t h o r c o m p a r e s panics that have b e e n a l l o w e d central b a n k s to coordinate u n d e r pressure o f crisis? to stop w i t h o u t intervention to those that have b e e n W h a t c o u l d a n international central b a n k d o to han- arrested w i t h i n t e r v e n t i o n . K i n d l e b e r g e r is clearly- dle a potentially d a n g e r o u s one-country m a n i a that c o n v i n c e d that crises in w h i c h n o lender o f last re- threatened to b e c o m e a n international panic? W h e n sort intervened have run longer a n d b e e n costlier in faced w i t h nations in different phases o f the business real terms. A n a l y z i n g the historical record o f the in- cycle, h o w w o u l d the international central b a n k de- ternational p r o p a g a t i o n o f financial crises, the a u t h o r cide o n its course o f action? c o n t e n d s that the depression f o l l o w i n g a crisis has Kindleberger's w i d e array o f stories demonstrates b e e n shorter w h e n a lender o f last resort has inter- that m a r k e t e c o n o m i e s are n o t , a n d for centuries vened. To accept this p o i n t fully, o n e must b e con- have not been, i m m u n e to manias, panics, a n d crash- vinced b y the author's a r g u m e n t that a g o v e r n m e n t , es. His cautionary tales shake a n y c o m p l a c e n c y that b a n k , or private organization has stepped in to per- m a y still exist a b o u t the stability o f today's w o r l d f o r m c e n t r a l b a n k f u n c t i o n s in a m a j o r i t y o f t h e e c o n o m y ; in fact, K i n d l e b e r g e r clearly s h o w s that crises he discusses. In international crises s u c h a n or- i n t e r n a t i o n a l transmission o f e c o n o m i c crises is at ganization has, often w i t h cooperation from similar least frequent a n d p r o b a b l y b e c o m i n g m o r e so. For institutions from other countries, b e c o m e a n interna- this reason his b o o k is w o r t h the attention o f those tional lender. O n l y in infrequent crises, Kindleberger interested in e c o n o m i c policy for financial avers, has there b e e n n o strong organization to inter- a n d the real e c o n o m y . T h e reader c o u l d w i s h , how- vene, a n d such events h a v e resulted in greater wel- ever, that this b o o k ' s m a s s o f e v i d e n c e fare losses. m o r e e n l i g h t e n m e n t o n the e n g i n e operating in the These discussions lead to the key p o r t i o n of markets provided c h o s e n m o d e l a n d at least sortie c o n s i d e r a t i o n of Kindleberger's b o o k — t h e a r g u m e n t for establishing the role o f real shocks a n d expectations in alterna- a n i n t e r n a t i o n a l l e n d e r o f last resort. T h e current tive m o d e l s . global e c o n o m y , h e maintains, lacks a strong, active T h e increasing integration o f the w o r l d ' s econo- lender o f last resort able to act in the international mies m a y s o m e d a y b r i n g m o r e c o o r d i n a t i o n o f na- sphere: the U.S. federal deficit w e a k e n s tional e c o n o m i c conditions a n d demonstrate that a n American ability; n e i t h e r J a p a n n o r G e r m a n y are yet strong international institution designed to respond to finan- e n o u g h , a n d each is h a m p e r e d b y internal d e m a n d s ; cial crises is vital to e c o n o m i c well-being. the International Monetary F u n d does not have suffi- berger breaks s o m e g r o u n d t o w a r d p e r s u a d i n g his cient resources or s c o p e o f powers. Consequently, in readers that s u c h a n o u t c o m e is desirable a n d in- t h e face o f increased threat o f i n t e r n a t i o n a l crisis evitable, b u t h e falls short o f his ultimate goal. Kindle- transmission, K i n d l e b e r g e r p r o m o t e s the establishm e n t o f a n institution c a p a b l e o f s u p p l y i n g financial resources n e e d e d to stop international panics or mit- B. F r a n k K i n g igate their results. This c o n c l u s i o n p r o m p t s several q u e s t i o n s that the a u t h o r answers o n l y briefly, if at all. W o u l d a n international central b a n k increase m o r a l hazard in a w o r l d in w h i c h risk taking is often subsidized exces- The reviewer is vice president and associate director of research at the Federal Reserve Bank of Atlanta. References Gorton, Gary. "Banking Panics and Business Cycles." Federal Reserve Bank of Philadelphia Working Paper, 1984. Jacklin, Charles J., and Sudipto Bhattacharya. "Distinguishing Panics and Information-Based Bank Runs: Welfare and Policy Implications." Journal of Political Economy 96 (June 1988): 568-92. Minsky, Hyman P. "A Theory of Systematic Fragility." In Financial Crises: Institutions and Markets in a Fragile Environment, edited by E.I. Altman and A.W. Sametz. New York: Wiley International, 1977. 32 . "The Financial Instability Hypothesis: Capitalistic Processes and the Behavior of the Economy." In Financial Crises: Theory, History, and Policy, edited by C.P. Kindleberger and J.P. Laffargue, 13-29- Cambridge: Cambridge University Press, 1982. Tallman, Ellis. "Some Unanswered Questions about Bank Panics." Federal Reserve Bank of Atlanta Economic Review 73 (November/December 1988): 2-21. E C O N O M I C REVIEW, JANUARY/FEBRUARY 1991 tmïmïym? • •. •• S T . íí;> ' - ! # - • - - s Syi; . .... : riîv ïi/x:•> S: «ÍS^ÍÍ-K . >v