The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE BANK OF DALLAS
Second Quarter 1993
Regulatioll, Bank Compelitiz'eness, alld
EjJimdes o/Missing Money
John V. Duca
Wba/ Determines Economic Gmwtb?
David M. Gould and
Roy J. Ruffin
77Je Pricing o/Natural Gas
ill Us. Markets
Stephen P. A. Brown and
Mine K. Yucel
/711'estillg/or Grou,tb:
77Jl'il'i77g in /be World Marketplace
Fiona D. Sigalla and
Beverly J. Fox
This publication was digitized and made available by the Federal Reserve Bank of Dallas' Historical Library (FedHistory@dal.frb.org)
Economic Review
Federal Reserve Bank of Dallas
Robert O. McTeer, Jr.
'.~
f
Tony J . Salvaggio
Harvey Rosenblum
W. Michael Cox
,
I
Gerald P. O'Driscoll, Jr.
Stephen P. A. Brown
','
,~<I'
•
Economists
7solt Becs
Robert T CI,m
JollI'V Ollca
Kenneth M erllery
Robert W Gilmer
David M Gould
William C GrubeI'
Josepl] H Haslag
Evan r Koenig
O'Ann M. Pewrsen
Keith R Phillips
Fiona 0 Sigalia
LOri L Taylor
John H Welch
Mark A Wynne
Kevill J Yeats
Mille K. YOcel
Research Associates
Professor Nathan S Balke
Sou/llem Me/hodist Univelsit)'
Professor Thomas B. Fomby
Southern Methodist Umversit)'
Professor Scott Freeman
UIJIIBrslty of Texas at Austin
Professor Gre[lory W Huffman
Southern Methodist University
Professor Roy J Ruffin
University of Houston
Profess or Ping Wang
PennsylvalJla State UniverSIty
Editors
Rhonda Harris
Virglilia M Rogers
Tile [C()fIOmll Rev!I"'/1S punllslled by the feueral Reserve B,,,,, cf [Jdl as ["P ,'p",s
expressed are t!lOSl: of the a~jthors and do /lC' "'cces~ar Iy refl('~t • ,e Posltl~'nc; Of 'he Fedcra
Reserve B,,"< 01 Dallils or the fetleral Re,erve Syslerl
Su!lscrlntlOf1~ a~e d\n'lahle ~rec of rhi:jrge Please se~d reqdl!'Sts for SH19!c-coPY iH'O n l, ':p1e
copy SUh~CiutIO:IS, hi.wk ISSue., and dci(iieSS c1rangcs to tile Pl.IJIlC Affd"S Deparltl1e0t Feoeral
Reser;e Ba'\K of Dilll.;,. P 0 Bux 555906.llall~s TX 10265·5906 '2' '19725257
!~fllcles may be 'ep'·n:ed all tile LOlldJl'on that tbe souce s crl~OJ:(.J a~'(j 'ii, Rpsl!arl,;,
Dep(lnrnen· IS pro,>"deu \"ith copy 01the pl,r"catlor ro Italf1'cn "'e 'err.r- e" ate r l'
On the cover: an architectural rendering of the new Federal Reserve Bank of Dallas
headquarters.
Contents
Regulation) Bank
Competitiveness)
and Episodes of
Missing Money
John V. Duca
What Determines
Econ01nic Growth?
David M. Gould and
Roy J. Ruffin
.Johll I )UCI rL'\ iL'\\ s thrl'c cpis(JciL-s of "Illi,~sing 1ll01K') ,"
durillg \\ hich one oj" the mOJK't;II'\' ~Iggrcgatl's \\,:IS
lI111lslIally \\ e:iI,. DUCI linds that in l':ICh or these episode ... , an
ill( reasl'd rl'guLitolY hurelen Oil IXll1k~ encouraged hou . . eholds
;llld linns to h\pas . . the il:lJlkillg systl'm in Lln)r or llonh:lI1k
1'i1l:lI1ci:iI killilitie ... ;li1d a. . "ets. l'sing :1 st:lndarcl :Inah tical
rr:IIlll'\\ ()rk. Ill' .shcm·s hm\ tlll'''''' ... hifts h\ iJl\e ..,tors em il':ld
to CISl'S of mhsing mOil,'> :llllllk-clim's in h:lnks' r()k in
pnl\ iding lTedil. Duci I'urtlll'r sIH)\\'s th:lt incrcases in h:lI1k
reglilatol'\' ilurdl'1l elll ,Tl':ltL' potl'nti:iI prohk'llls for anal) st... in
usillg l'ithcr int,'I'e"t ratl'''' or re:11 time lll(lIlL'\an aggr,'gatl's as
indicators or nomill:iI C'l'ollomic :Icti\·it\. CiH'n thl'sC findings,
I )lIca ;Irgul'" tlut IX'C1U"'V r,'gulator> ch:ll1ge'i h;l\c implica
tioll'" ror condul'lillg mOlll'Un polin. the I'ederal ]{e,"c'J'\ e
..,houklu)lltilllll' to h:l\l' a roll' in lormulating hallk regUlations,
pL'J'iod.~
I )Ol'S incrca."'c'd imcstml'nt in cducation enhance longrUIl l'Conol1lic gnl\\ Ih , or docs it simply reduce currl'lll consumption' \\ ill rrCl' [r:lde stimulate gro\\th, or \\ il l it ll1ercly
incrca'ie import'>'
For:1 long time, ecollomis\.,> relied on an cconomic gnl\\th
theory th:lt ofTcrcd lillie' scope I'm undersl:inding long- rull gJ'()\\th
il10\'l'mellts. ]{ecently, IHl\\ ,'\er, the stud> or economic gnmth
11:1." hl'l'n reim'igoratccl hI' Ill'\\ dl'H'loplllents in thcor) :me!
empirical findings tll:nsuggl'st hcm long run gn)\\th e\·ol\'l's.
Bccause l'('onOlllic gr()\\'th detcrmines \\'hether our grand
chile!rcn \\ ill h:1\ e hClier li\'cs than OUl'S or \\ hl'lher poor
nations \\ ill catch up \\ ith or r:ill rurthl'r IX'hind rich nations,
1),11 id Could :Ine! I{o\ l{ulTin im'l,.,tig:lll' rlTl'nt iL'..,solls Ic:lrnnl
"hout gn)\\ th anc! apph thclll to till' ahol'l' i~sucs. Coule! ami
HufTin report on rCC,'llt research ,.,uggl'stillg that iml'stll1ellt.
p:lrticularl), hU1ll:l1l capital iJ1\ estllll'llt, increasc,> ecolloll1ic
gn)\\tiJ, They al."o illH'stigatl' e\idencl' shc)\\ ing tl1:1t political
"l:Ihilit\'. \\ ell-defined properly rights, Ic)\\ tradc h:m'il'rs, and
Ie)\\ gO\ ernlllL'nt consumption e'penditllres enhance grc)\\th
Ihrough pmiti\l' clrcc!.s Oil ill\ l'stllll'nt.
Contents
Page' II
The Pricing ofNatural
Gas in U.S. Markets
Stephen p, A, Brown and
Mine K, YOcel
Stephen 131'O\\'n and .\Iine YLkel examine hoI\' different
natural gas users and the Ill<lrket institutions serdng them
afkct the tr~lJ1smissi()n of price changes throughout \ 'arious
markets for n~ltural gas, 1 ~ lectrical utilities and inclustrial users
huy much of their natural gas in :1 competiti\'e spot market
sl'l'I 'ed by hrokers and interstate pipeline companies, In con t rast. most commercia I :lJ1d resident ia I customers a re dependent on loci! clistrihution companies. \\ 'hich earn a regulated
rate of return ancl huy their gas under long-term contracts,
l 'sing time-series ml,thocls, BrO\\'n ancl 1'('lcel fincl that
eH'n in the long run. changes in prices arc not transmitted
uniforml\' throughout the \ 'arious markets for natural gas,
Electrical and industrial customers haH' seen a greater henefit
from falling natural gas prices than commercial and residential
customers, Differences in market institutions and in the ahility
of the end users to s\\'itch fuels may account for the lack of
uniformity,
Page ')3
Investing for Growth:
J7Jriving in the
World Marketplace
Fiona 0, Sigalla and
Beverly J. Fox
rree trade is rapidly eXlxlnding the world marketplace,
prOl'iding nCII' opportunities to raise li\ing standards and
stimulate long-term economic growth, To capitalize on these
opportunities, the l ' nited States must O\'ercome problems that
reduce its competitil'cncss, such as rcgu lations that distort
market inccnti\'cs. an cducation systcm that may not be preparing a \\ 'ork force cquipped to compete in the tll'enty- first
century. and a health carc systcm \\'hose skyrocketing costs
hal'e hecomc a hurdcn to taxpayers. gOl'ernment, and business,
TIlc Federal l{csel'\'c Bank of Dallas acldressed the
opportunities pn:scntcd by frec tracle and the concomitant
need to increase L' ,S, competiti\'encss at the Bank's recent
economic conkrence . entitled "JJl\'csting for Groll,th: Thri\'ing
in thc \, 'orld .\ Iarketplace," In this article. Fiona Sigalla and
Be\'erly Fox summarize concerns and strategies re\ 'icII'cd and
debated in conf'crcnce sessions,
John V. Duca
Senior Economist and Policy Advisor
Federal Reserve Bank of Dallas
Regulation, Bank Competitiveness,
and Episodes of Missing Money
I
n setting monetary policy, most central banks
look at a number of economic indicators, including data on monetary aggregates. The motivation
for monitoring monetary aggregates comes from
the equation of exchange:
(1)
M ×V = P ×Y ,
where M = money, V = velocity [nominal gross
domestic product (GDP) / M )], P = the price level,
and Y = transactions (usually measured by inflationadjusted GDP). Typically, people reduce their
holdings of money as the spread between a riskless short-term market interest rate (such as the
three-month U.S. Treasury bill rate) and the average
yield earned on monetary assets rises. As a result,
the velocity of money rises as this spread or
“opportunity cost” of money increases. If velocity is
predictable, then money and its predicted velocity
can be used to infer nominal GDP (P × Y ). Under
these circumstances, a monetary aggregate is useful
for policymakers as an indicator of nominal GDP.
This is especially true because data on GDP are
available after a long lag, whereas information on
money and interest rates is more readily available.
However, in three of the past four recessions
(1973–74, 1979–80, and 1990–91), the monetary
aggregate most closely monitored by the Federal
Reserve has been much weaker relative to income
and opportunity cost measures than previous
experience predicted. This unusual weakness, or
“missing money,”1 poses a serious problem for
policymakers because it means that the monetary
aggregate in question is less useful as an indicator
of nominal activity at a critical point in the business
cycle. Furthermore, analysts often need at least
several quarters of data to discern whether such a
Economic Review — Second Quarter 1993
money demand shock has occurred and whether
any particular shock is permanent or temporary.
Consider a permanent downward shift in the
level of demand for a monetary aggregate; such a
shift would result in a fall of that aggregate’s growth
rate relative to GDP growth over a period of time
at each level of opportunity cost. There are two
choices that a responsible central bank would
consider. If the central bank stabilized the growth
rate of that aggregate at the previous average,
nominal GDP growth would temporarily accelerate
and then return to its previous growth rate. Eventually, the spurt in nominal GDP growth would
result in a temporary acceleration in inflation. As
a result, the price level would be permanently
raised relative to its path had the money demand
shock not occurred. While the price level would
post only a once-and-for-all rise, such an episode
would create uncertainty about whether the central
bank was committed to controlling inflation. Such
uncertainty would likely depress real economic
activity for awhile because inflation uncertainty
discourages firms and households from committing
to long-run projects.
I would like to thank, without implicating, Anne King and
Steven Prue for providing excellent research assistance,
and Michael Cox, Ken Emery, Evan Koenig, Ken Robinson,
and Harvey Rosenblum for their suggestions during the
progress of this research. Any remaining errors are my own.
1
Throughout this study, the term “missing money” describes
episodes in which the level of a monetary aggregate has
been smaller than predicted based on past relationships,
income, and the opportunity cost of money.
1
As an alternative, the central bank may accept
temporary weakness in the growth of its primary
monetary aggregate. However, it is difficult in real
time to know precisely how much of a slowdown
is appropriate. If the central bank permitted money
growth to slow too much, nominal GDP growth
would temporarily be below trend. If the monetary
authority underestimated the impact of a money
demand shock, then nominal GDP growth would
temporarily be above trend until the money
demand shock passed.2
Given that cases of missing money are problematic for central banks, it is natural to ask why
there have been money demand shifts. To help
answer this question, this study reviews research on
the three most recent episodes of missing money.
Common to each of these cases is a decline in
banks’ ability to compete with nonbanks that
stemmed from the changing impact of banking
regulations. As a result of declines in the competitiveness of banks, households and firms have
shifted toward using nonbank types of “money”
and credit, and researchers have found it helpful
to redefine money or measures of its opportunity
cost to obtain a more reliable indicator of nominal GDP.3
In establishing these findings, this study begins
with a simple macroeconomic model of how rising
bank regulatory taxes can contribute to weakness
in overall economic activity and a decline in the
share of credit provided by banks. The second
section of this article reviews the mechanics of how
2
3
2
The cases in the text analyze permanent shocks to money
demand. If the shocks were temporary, then by not altering
its long-run monetary targets, a central bank could keep the
economy growing in line with the central bank’s previous
long-run nominal GDP target. In this case, there would be
some temporary acceleration or deceleration in nominal
GDP growth that would later be reversed.
As argued later in this article, money demand shifts in the
mid- and late-1970s led the Federal Reserve to change the
primary monetary aggregate it monitored from M1 to M2.
Indeed, when M2 was officially created in 1980, it was
defined to include new financial instruments such as money
market mutual funds (MMMFs) and repurchase agreements.
a shift away from credit and deposits at banks to
substitutes at nonbanks can also result in a missingmoney phenomenon. Within this framework, the
second section then analyzes evidence on the
three most recent episodes of missing money.
Each case of missing money is found to have
coincided with declines in the ability of banks to
compete with nonbanks. The concluding section
discusses the policy implications of these findings.
A simple macroeconomic model
This section lays out a simple model of aggregate demand that can be used to analyze the
impact of regulatory burden on economic activity
and on the share of credit provided by banks.
These effects, coupled with insights provided in
the next section, are later shown to be useful in
helping policymakers detect whether a monetary
aggregate is accurately reflecting nominal GDP
growth. The model used here is presented in two
parts. First, the conditions for equilibrium in the
goods market are described in a world where firms
can borrow either from banks or directly from
open credit markets. Second, conditions for equilibrium in the credit market are derived. Using
these conditions, the equilibrium levels of output
and interest rates are derived.
A simple IS specification. A portion of firms
(Θm) obtains credit only from the financial markets,
while the remaining portion (Θ b ≡ 1–Θ m) relies
completely on bank loans. Demand by each firm
for open market (L m) or bank credit (L b) is
(2)
Ltm = α 0 + α1Yt − α 2R tm and
( 3)
L bt = α 0 + α1Yt − α 2R tb ,
where Greek letters denote positive coefficients,
Y is output, R m is the average rate on open market
credit, and R b is the average bank loan rate.
The average cost of credit (R) and total private credit demand (L p) across all firms are thus
(4 )
(5 )
Rt = Θ m R tm + Θ b R tb and
Ltp = Θ m Lmt + Θ b L bt
= α 0 + α1Yt − α 2 (Θ m R m + Θ b R b ).
Federal Reserve Bank of Dallas
On grounds that firms and households spend less
when the cost of finance rises, nominal income is
assumed to depend negatively on average credit
costs:
(6 )
Yt = φ0 − φ1R t
= φ0 − φ1Θ m R tm − φ1 (1 − Θ m )R tb .
Decisions about modeling how the costs of
bank and open market credit are determined have
been made to be consistent with several key stylistic
facts. First, firms that rely on open market credit
generally are perceived as posing little default risk.
Second, bank credit has an advantage over open
market paper in that the deposit insurance system
bears some of the default risk of bank loans. Third,
open market credit has an advantage over bank
credit in avoiding certain regulatory costs imposed
on the banking system.
Interest rates on bank and open market
credit are, respectively:
(7 )
R tb = c b + rt + (1 − d )D tb + t bf and
(8 )
R tm = c m + rt + D tm ,
where r is the riskless market interest rate, d is the
implicit default risk subsidy on bank loans from
deposit insurance, D b is the average fair market
risk premium on bank loans, D m is the average
fair market risk premium on open market paper,
t bf reflects any regulatory burdens on banks that
effectively can be treated as a constant, c b is a
constant reflecting the per-dollar costs of providing
bank loans not associated with interest costs or
default risk (primarily information and transactions
costs), and c m is a constant reflecting the per-dollar
information and transactions costs associated with
issuing open market paper.
To capture differences in default risk across
firms in a tractable way, the assumption has been
made that the fair market default risk premium
(D i ) across firm types (i ) has a uniform distribution over the interval:4
(9 )
⎡
⎛cb − cm ⎞⎤
m
D tc = ⎢t bf + ⎜
⎟⎥ = Θ ,
d
⎝
⎠
⎥⎦
⎢⎣
(10 )
which is increasing in regulatory taxes imposed
on all bank loans (t bf ), decreasing in the extent to
which bank loans have lower information and
transactions costs than open market paper (c b – c m),
and decreasing in the implicit risk-taking subsidy
(d ) provided by deposit insurance. Since D i has
a uniform distribution over [0,1], Θ m = D c and
Θ b = 1 – D c. In this model, banks lend to higher
risk firms, while lower risk firms issue open market
paper. The reason is that the cost disadvantage of
bank regulatory taxes is roughly fixed across borrowers, while the implicit benefit of deposit insurance
is increasing in default risk because indirectly taxpayers bear some of the risk.5 For example, the
implicit benefit of deposit insurance is low on a
bank loan to a firm that has low default risk, while
the regulatory burden of such a loan may be very
high. In this case, bank loans are a more costly
source of credit than open market paper. Thus, the
model is consistent with the stylized fact that only
very low default risk firms issue commercial paper.
This qualitative result can be obtained in this model
if one assumes that the information costs of issuing
open market debt are lower for firms having low
default risk because their creditworthiness is
generally more transparent to investors.6
Since default risk is distributed uniformly, the
average default risk premium on open market paper
is (D c/2) and that on bank loans is ([1 + D c ]/2).
Using these average risk premia along with
4
For ease of exposition, any rationing of credit is suppressed.
5
As stressed by Keeley (1990), the value of this implicit
subsidy depends on how well-capitalized a bank is. This
implicit subsidy declines as a bank’s capital increases
because when a bank fails, the capital invested by bank
equity and subordinated debt holders are the first funds
used to cover any losses from liquidating the bank. For ease
of exposition, all banks are treated as being equally capitalized in the model.
6
Diamond (1991) develops a theoretical model that more
rigorously and formally demonstrates this result.
D i ~ U [0, 1].
Setting equations 7 and 8 equal yields the critical
level of default risk (D c ) at which firms are indifferent between bank and open market credit:
Economic Review — Second Quarter 1993
3
equations 7, 8, and 10 yields the following expression for the average cost of credit:
(11)
Rt = Θ m R tm + Θ b R tb
⎡ ⎛ t bf ⎞ ⎤
⎡ ⎛ cb ⎞ ⎤
= rt + c b ⎢1 − ⎜ ⎟ ⎥ + t bf ⎢1 − ⎜ ⎟ ⎥
⎢⎣ ⎝ 2d ⎠ ⎥⎦
⎢⎣ ⎝ 2d ⎠ ⎥⎦
m 2
(1 − d ) (c )
t bf c b
−
+
−
2
2d
d
t bf c m c bc m
+
+
.
d
d
Differentiating equation 11 and substitution from
equation 10 implies
(12)
∂R
= 1 − Θ m > 0,
b
∂c
∂R
= 1 − Θ m > 0,
∂t bf
∂R
= Θ m > 0,
∂c m
⎡ ⎛ Θm ⎞ ⎤
∂R ⎛ 1 ⎞
= ⎜− ⎟ + Θ m ⎢1 − ⎜
⎟ ⎥ < 0, and
∂d ⎝ 2 ⎠
⎢⎣ ⎝ 2 ⎠ ⎥⎦
∂R
= 1.
∂r
Equation 12 implies that the average cost of
credit rises when either bank regulatory taxes (t bf )
or information and transactions costs (c b) increase.
This result is obtained because the rise in credit
costs to those firms that remain bank borrowers
will outweigh the effect of some firms’ switching
toward less expensive open market paper. Thus, a
rise in regulatory taxes might help induce a recession, cause a decline in the importance of banks
in credit markets, and—as will be discussed in the
next section—trigger an episode of missing money.
As equation 12 also indicates, a rise in the
information and transactions costs of commercial
paper (c b) will cause the average cost of credit to
increase. This increase occurs even though some
firms shift away from open market paper when
the information and transactions costs of open
market paper rise because the effects of this shift
on average credit costs are outweighed by the
impact of higher costs on those that remain nonbank borrowers. Thus, a rise in c m will, by raising
the cost of open market paper, cause banks to
gain credit market share and reduce the demand
for nominal output.
A rise in the deposit insurance subsidy decreases the average cost of credit by lowering the
cost of bank loans. While the sign of the effect is
theoretically ambiguous, a higher subsidy will
lower the cost of finance as long as there are some
firms that rely on bank loans (that is, Θ m < 1).
Substituting equation 11 into equation 6 yields
the following IS curve:
(13)
Figure 1
Effect of a Rise in Bank Regulatory Burden
on Goods Market Equilibrium
r
ISo
y
4
Yt = φ0 − φ1rt
⎧ ⎡ ⎛ cb ⎞ ⎤
⎡ ⎛ t bf ⎞ ⎤⎫
⎪c b ⎢1 − ⎜ ⎟ ⎥ + t bf ⎢1 − ⎜ ⎟ ⎥⎪
⎪ ⎢⎣ ⎝ 2d ⎠ ⎥⎦
⎢⎣ ⎝ 2d ⎠ ⎥⎦⎪
⎪
⎪
⎪ ⎡ (1 − d ) ⎤ ⎡ (c m )2 ⎤ t bf c b ⎪
−
−
−φ1 ⎨+ ⎢
⎥
⎥ ⎢
⎬,
d ⎪
⎪ ⎢⎣ 2 ⎥⎦ ⎢⎣ 2d ⎥⎦
⎪ t bf c m c bc m
⎪
+
⎪+
⎪
d
d
⎪⎩
⎪⎭
which implies a negative relationship between
combinations of output and the short-term interest
rate that clear the goods market. Thus, the IS
curve has the normal downward-sloping shape
(Figure 1 ). As will be shown later, equation 13
implies that a rise in regulatory taxes on banks
reduces output at each combination of the riskless
market interest rate (r ) and goods demand (Y ).
Federal Reserve Bank of Dallas
Thus, such an increase in bank regulatory burden
can be depicted as an inward shift of the IS curve
from IS0 to IS1. Of course, a rise in regulatory
taxes may indirectly affect rt when the conditions
for goods market equilibrium (IS curve) and credit
market equilibrium are solved together.
Credit market equilibrium conditions. Traditional Keynesian models depict interest rates as
determined by the supply and demand for money.
This approach may have been plausible for the
1930s and 1940s because few firms could issue
open market paper following the collapse of the
bond and commercial paper markets during the
Great Depression. Today, it is more accurate to
model short-term interest rates as determined by
the total supply and demand for short-term credit,
since commercial paper and Treasury bills have
each grown to roughly the size of commercial and
industrial (C&I) loans at banks.7
The demand for short-term credit is mainly
comprised of the demand for bank loans, commercial paper, and Treasury bills. Although it is likely
that the demand for bank loans and commercial
paper is interest-sensitive, it can be argued that
the demand by the U.S. government for Treasury
bills has generally been highly insensitive to shortterm rates. By implication, government demand
for short-term credit can be approximated by a
constant (L g), and total credit demand (Lt ) equals
private plus government demands:
(14 )
Lt = Ltp + L tg
= α 0 + α1Yt − α 2R t + L g
= α 0 + α1Yt + L g − α 2rt
⎧ ⎡ ⎛ cb ⎞ ⎤
⎡ ⎛ t bf ⎞ ⎤⎫
⎪c b ⎢1 − ⎜ ⎟ ⎥ + t bf ⎢1 − ⎜ ⎟ ⎥⎪
⎪ ⎢⎣ ⎝ 2d ⎠ ⎥⎦
⎢⎣ ⎝ 2d ⎠ ⎥⎦⎪
⎪
⎪
m
⎪ ⎡ (1 − d ) ⎤ ⎡ (c )2 ⎤ t bf c b ⎪
−α 2 ⎨+ ⎢
⎥−⎢
⎥−
⎬.
d ⎪
⎪ ⎢⎣ 2 ⎥⎦ ⎢⎣ 2d ⎥⎦
⎪ t bf c m c bc m
⎪
+
⎪+
⎪
d
⎪⎩ d
⎪⎭
In principle, the supply of short-term credit
can be depicted as the sum of credit supplies from
different sources. However, the supply of shortterm credit in the United States can be depicted in
a simple fashion because the Federal Reserve has
typically implemented monetary policy by altering
Economic Review — Second Quarter 1993
Figure 2
The Supply of and Demand for Short-Term Credit
r,re
Ls
re
Ld
y,L
a targeted level of the federal funds rate to stabilize
nominal aggregate activity. Federal funds are
reserves that banks trade with one another to meet
reserve requirements on bank deposits. By purchasing or selling reserves in exchange for Treasury
bills, the Federal Reserve tries to target a chosen
level for the federal funds rate. Under these conditions, banks would borrow from the Federal
Reserve to purchase T-bills if T-bill rates were
above the average expected level of the federal
funds rate over the remaining maturity of the
Treasury bills. Banks will continue to buy Treasury
bills until T-bill rates fall in line with expectations
of the federal funds rate, consistent with the empirical findings of Cook and Hahn (1989). This
arbitrage implies that the T-bill rate equals the
average federal funds rate target (r e) that the
market expects the Federal Reserve to use over
the life of a particular maturity T-bill. By implication, the Federal Reserve can generally target shortterm interest rates, and the supply curve of total
short-term credit (L s) is horizontal (Figure 2 ) and
7
For example, the outstanding commercial paper of nonfinancial and financial firms is roughly four-fifths the size of
C&I loans at banks ( Federal Reserve Bulletin 1993), while
the stock of U.S. Treasury bills is roughly as large as C&I
loans at banks.
5
depends on the expected path of the federal
funds rate target.8
As output rises at a given level of interest
rates, the credit demand curve shifts to the right.
However, if the Federal Reserve maintains its funds
rate target, the credit supply curve remains horizontal and the riskless short-term interest rate does
not change. As a result, the equilibrium combinations of interest rates and output at which shortterm credit demand equals short-term credit supply
can be depicted as the horizontal CC curve in
Figure 3. This curve shifts in line with expectations
about the federal funds rate target. The result that
the CC curve is horizontal under an interest rate
target parallels the flat LM curve obtained in Poole’s
(1970) model.9
The effects of altering the relative cost of
bank loans and open market paper. According
to equation 10, three types of factors affect the
relative use of bank loans and open market paper:
8
By contrast, the Federal Reserve has much less effect on
long-term interest rates, which it indirectly affects by influencing expectations about future inflation and the size of the
inflation risk premium in long-term interest rates. This premium reimburses investors for the risk that inflation may be
higher than they expect. If inflation were higher than expected, expectations of future inflation would likely rise,
then interest rates would rise so that inflation-adjusted
yields are maintained. These conditions imply that bond
prices would likely fall.
9
A flat CC curve is a reasonable approximation in the very
short run. However, any changes in the central bank’s
target rate that are made partly on the basis of money or
credit growth imply that this curve is upward-sloping in
the medium run. Thus, past changes in the federal funds
rate target that have been partly or largely based on M2
growth suggest that the CC (or LM) curve has an upward
slope in the short to medium run.
10
6
In the short run, a rise in bank regulatory taxes causes a
rise in the average cost of credit. If, however, enough innovation is induced by regulations, the average cost of credit
could fall in the long run, provided the long-run cost of
issuing open market paper declines enough. For example,
while the regulatory burden of reserve requirements likely
increased the cost of providing bank loans in the high
interest rate environment of the late 1970s and early 1980s,
it helped spur the development of the commercial paper
market. As pointed out by Post (1992), the cost of commercial paper has generally fallen over the past decade.
Figure 3
Credit and Goods Market Equilibrium
r
CC
IS1
y1
IS0
y0
y
bank regulatory taxes (t bf ), the implicit deposit
insurance subsidy (d ), and the differential between
the information costs for bank loans and for open
market paper (c b – c m ). In light of these three
factors and the forthcoming analysis of three cases
of missing money, this section presents analysis
of the effects of a rise in bank regulatory taxes, a
decline in the implicit deposit insurance subsidy,
and a decline in information costs associated with
open market paper.
The effects of raising bank regulatory taxes.
By increasing the cost of loans relative to market
interest rates, a rise in bank regulatory taxes shifts
the IS curve inward to IS1. This can be demonstrated by differentiating equation 13 and substituting a result from equation 12:
(15)
⎛ ∂Y ⎞ ⎛ ∂R ⎞
∂Y
=⎜
⎟ ⎜ bf ⎟
bf
∂t
⎝ ∂R ⎠ ⎝ ∂t ⎠
= −φ1 (1 − Θ m ) < 0.
If the central bank does not perceive the IS shift
and cuts interest rates, then smoothing short-term
market interest rates (r e ) results in a decline in the
demand for nominal output (Y ) from Y0 to Y1. As
a consequence, increases in bank regulatory taxes
can contribute to the onset of a recession.10
Changes in bank regulatory taxes thus create
disturbances to the IS curve, which create problems
in using a federal funds rate target. This result
accords with the insights in Poole (1970). Poole’s
Federal Reserve Bank of Dallas
model showed that a central bank policy of targeting a monetary aggregate will be superior to
that of targeting a short-term interest rate if IS
shocks are large relative to money demand shocks.
It is thus tempting to conclude that targeting a
monetary aggregate would yield superior results
under these circumstances.
However, as shown in the next section, the
change in banks’ regulatory burden also causes a
fall in the demand for money. In the Poole framework, as the importance of money demand disturbances rises, targeting a monetary aggregate
becomes less attractive relative to targeting a shortterm interest rate. Thus, changes in banks’ regulatory burden create problems for both types of
operating procedures.
The effects of reducing the implicit deposit
insurance subsidy. The effects of reducing the
implicit deposit insurance subsidy are qualitatively
similar to those of increasing the regulatory tax on
banks. Such a decline increases the cost to banks
of providing loans (equation 7), which makes bank
loans relatively more expensive than open market
paper for more firms. One result of this change in
relative costs is that some firms shift away from
bank loans and switch to issuing open market
paper (Θ m increases in equation 10). In addition,
much like a rise in bank regulatory taxes, a reduction in this subsidy (d ) leads to an increase in the
average cost of finance (R in equation 12). This
increase in cost, in turn, shifts the IS curve inward,
thereby reducing the nominal demand for goods
(Y ). This result can be seen by differentiation:
(16 )
∂Y ⎛ ∂Y ⎞ ⎛ ∂R ⎞
=
∂d ⎜⎝ ∂R ⎟⎠ ⎜⎝ ∂d ⎟⎠
⎛ ∂R ⎞
= −φ1 ⎜ ⎟ > 0.
⎝ ∂d ⎠
This expression is positive because ( R/ d ) < 0
and implies that reducing d will shift the IS curve
inward.
For two reasons, one should avoid inferring
from this example that reductions in the implicit
deposit insurance subsidy are not necessarily beneficial. First, any effect on aggregate demand can be
offset by a monetary easing action, which would
push down short-term market interest rates. Second,
because deposit insurance implicitly shifts much
Economic Review — Second Quarter 1993
of the default risk on high-risk loans from stockholders in banks and thrifts to taxpayers, deposit
insurance results in excessively risky lending and
thus creates inefficiencies. The high cost of the
savings and loan association bailout provides a
justification for implementing risk-based capital
standards. The point of this example is to illustrate that monetary policy should take into
account any macroeconomic impact of curtailing
the risk-taking incentives of deposit insurance.
The effects of a decline in the information
costs of open market paper. A decline in the
information costs of open market paper can stem
from reductions in costs associated with investors’
learning about firms, improved computer technology that reduces the transactions costs of buying
and selling open market paper (such as bonds and
commercial paper), and the deepening or increased
liquidity of open paper markets. Like a decline
in the deposit insurance subsidy to banks and a
rise in bank regulatory taxes, a fall in the costs of
providing open market paper (c m) will reduce the
cost of open market paper (R m) relative to that of
bank loans (R b) and thereby induce a rise in the
share of open market paper (Θ m in equation 10).
The effects on aggregate demand, however,
differ greatly because a decline in the cost of open
market paper lowers the average cost of finance
(equation 11). As a result, the gap between the
average cost of finance to firms and the riskless
open market rate paid by the U.S. Treasury narrows
at each level of income. Thus, a decline in c m
causes the IS schedule to shift to the right, thereby
driving up aggregate nominal demand. This effect
can be demonstrated by differentiating equation
13 and substituting a result from equation 12:
(17 )
⎛ ∂Y ⎞ ⎛ ∂R ⎞
∂Y
=⎜
⎟⎜ m⎟
m
∂c
⎝ ∂R ⎠ ⎝ ∂c ⎠
= −φ1Θ m < 0.
The stabilization of aggregate demand thus calls
for increasing the federal funds rate, in contrast to
the prescription of cutting the federal funds rate if
bank regulatory taxes rise or if the deposit insurance subsidy is lowered.
Thus, the monetary policy implications of
these examples are not the same. The analysis
above implies that to discern among these types of
7
Episodes of money demand instability
Figure 4
The Supply of and Demand for Money
Opportunity
cost
Ms
Opportunity
cost implied
by the federal
funds rate
target
Md
Money
shocks, policymakers should check for not only a
decline in bank credit market share, but also for
whether such a decline is accompanied by a rising
regulatory burden on banks, a reduction in the
implicit deposit insurance subsidy to banks, and a
weakening in income growth or, alternatively, a
reduction in the information and transactions costs
of using open market paper and a strengthening
of nominal income growth. The next section
discusses how an unpredictable decline in the
measured money supply may also accompany
such shocks.
11
8
This assumption is consistent with the Federal Reserve’s
use of a target range for M2. This policy allows M2 balances
to move within a range for three reasons. First, it implicitly
recognizes that moderate M2 growth will, barring money
demand shocks, need to accompany moderate growth in
nominal income. Second, this policy recognizes that changes
in interest rates will affect the velocity of M2 and thereby
alter the pace of M2 growth needed for moderate growth in
nominal income. Third, the policy also recognizes that shifts
in money demand may occur and that the money supply
curve may shift owing to bank behavior. In the latter case,
the opportunity cost of money associated with a particular
federal funds rate could vary, depending on how actively
banks bid for M2 deposits. This practice, in turn, alters the
velocity of M2 and the growth rate of M2 that is needed to
stabilize aggregate demand.
The case of a nonrange money target is discussed
later in this article.
This section begins by describing how, in a
simple framework, reduced bank competitiveness
can lead to a missing-money phenomenon. Then,
in terms of this framework, three episodes of missing money are discussed: the missing M1A in the
mid-1970s, the missing M1A and surge in money
market mutual funds (MMMFs) during the late 1970s
and early 1980s, and the current case of the missing M2. Finally, the implications of these shocks
for M2 targeting are discussed.
How reduced bank competitiveness can create
a missing-money phenomenon. This section outlines a simple supply and demand model of money
that can be used to analyze cases of missing money.
In Figure 4, the money demand (M d ) curve is drawn
with a downward slope that reflects that households and firms reduce their holdings of money as
the opportunity cost of money (OC ) rises (if all else
remains the same). The opportunity cost of money
is the extra yield that investors forgo by holding
money, which provides convenience and transactions services over other assets that have higher
pecuniary yields. In practice, opportunity costs are
typically measured as the difference between a
riskless short-term market interest rate and the
average yield earned on monetary assets. The money
demand curve is drawn for a given level of income.
This curve shifts to the right as income rises because the transactions demand for money will rise
at each level of the opportunity cost of money.
The money supply curve (M s ) has an upward
slope to reflect that banks would be willing to
supply more deposits as the spread between
market interest rates and deposit rates increases
because banks can earn more profit by supplying
deposits when the yields on securities (or loans)
rise relative to deposit rates. Here it is assumed
that the Federal Reserve does not rigidly target
money balances (otherwise, the money supply
schedule would be vertical).11 The money supply
schedule is partly a derived demand for funding
loans because banks will bid up deposit rates if
loan demand rises and loan interest rates rise as a
result. Thus, a monetary easing action that boosts
loan demand through interest rate or wealth effects
will shift this curve to the right, while an exogenous decline in demand for bank loans will shift
the curve to the left.
Federal Reserve Bank of Dallas
Money demand models, such as the Federal
Reserve Board’s M2 model (henceforth, the FRB
model ),12 tend to estimate M2 growth well as long
as changes in the level of income are the only
source of shifts in the money demand schedule.
The following examples illustrate this point. First,
consider what happens if the Federal Reserve eases
monetary policy. The easing shifts the money
supply schedule to the right in the short run, causing a decline in M2’s equilibrium opportunity cost.
This movement along the money demand curve is
picked up by the opportunity cost measures in
the FRB model. Then, as the decline in short-term
interest rates causes aggregate demand to pick
up, nominal income will rise, causing the money
demand schedule to shift to the right, which induces a measured rise in M2’s opportunity costs.
This shift of the money demand curve is picked up
by the income and consumption spending variables
in the FRB model. In the past, M2’s growth rate
primarily reflected movements of the money supply
curve along the money demand curve and shifts
of the money demand curve owing to income
changes. As a result, the coefficient estimates of
the FRB model reflect a positive effect of income
on M2 growth and a negative correlation of M2’s
opportunity costs with M2 growth. By implication,
the coefficient estimates of the FRB model will
yield good predictions of M2 growth as long as
income and money supply shocks are the only
sources of change affecting M2.
An example of a money demand shock. Now
consider an example in which some changes in
the economic environment not reflected in M2’s
measured opportunity cost simultaneously cause
firms to shift from C&I loans to bonds and induce
households to shift out of M2 deposits into bond
mutual funds. Income is held constant in this
example to reduce the number of curve shifts for
ease of exposition.
This case is shown in Figure 5. One result is
that the demand for bank credit falls, and with it,
bank demand for issuing M2 deposits. By implication, the M2 supply curve shifts inward (shift 1),
and the FRB model should pick up this decline
through its opportunity cost measures. However,
because the demand for M2 deposits also falls at
every combination of OC and Y, the money demand
curve shifts inward (shift 2). Since coefficient estimates for most M2 models are based on a past
Economic Review — Second Quarter 1993
Figure 5
A Bypassing of the Banking System
Opportunity
cost
MS1
MS0
1
2
OC″
OC0
A
1
2
Md1
M2″ M2′ M20
Md0
Money
negative correlation between M2 and its opportunity cost, the money demand model implicitly
assumes that a shift in the money supply rather
than demand curve has occurred. In terms of
Figure 5, money demand models assume that the
demand curve is unchanged and that the supply
curve intersects the demand curve at point A. As
a result, money demand models would estimate,
once interest rate and income data are available,
that M2 moved to a level like M2′ rather than
having declined all the way to M2″.13 In this instance,
a case of missing money would occur.
Alternatively, if the demand curve shift is
large enough relative to the supply curve shift, then
OC could be lower in equilibrium and the money
demand model would predict a rise in M2 even
though M2 would actually fall (since both the M d
and M s curves shift to the left). In either case, the
money demand model overpredicts the equilibrium level of M2 because it fails to pick up the
money demand curve shift by assuming that only
12
For a discussion of this model, see Moore, Porter, and Small
(1990) and Small and Porter (1989).
13
Note that income is being held constant in this example.
While changes in income will shift the Md curve, the inclusion
of income variables in money demand curves controls for
Md shifts stemming from changes in income.
9
Figure 6
Figure 7
Transfers of Funds
Balance Sheet Changes
Bond Mutual Fund
Firm
(a)
Firms
A
Bond mutual funds
L
A
L
–100 C&I loans
+100 bonds
+100 mutual
fund shares
+100 bonds
(d)
(b)
Bank
A
–100 C&I
loans
Banks
(c)
Households
changes in nominal income shift the money
demand curve.14
Fundamentally, the money demand shift
occurs because both borrowers and depositors
substitute away from banks for credit and deposit
services. To illustrate this point, Figure 6 depicts
the transfers of funds in the movement to a lower
equilibrium level of M2. Suppose a firm issues a
$100 bond purchased with $100 by a bond mutual
fund, which in turn gives $100 in mutual fund
shares to a household in exchange for a $100 check
drawn on a bank account. Suppose also that the
household moved the $100 from a nonreservable
small time deposit to a checking account to make
the transaction.15 The firm takes the $100 raised by
issuing bonds (exchange a) to pay off $100 in C&I
10
14
From a Marshallian point of view, money demand changes
predicted by typical econometric models can be interpreted as movements along short-run money demand
curves that may shift with income levels, whereas money
demand shocks can instead be interpreted as movements
along a long-term money demand curve in so far as these
“shocks” represent the endogenous response of firms and
households to changes in the opportunity cost of using
money.
15
This assumption enables us to avoid changing bank reserves and is sensible, given that bond funds seem to be
more substitutable for small time accounts than for other M2
balances that are less useful as savings vehicles.
Household
L
–100 small
time
deposits
A
L
–100 small
time
deposits
+100 bond
mutual
fund shares
loans to the bank (exchange d). The bond mutual
fund pays the firm with the $100 it raised from
selling mutual fund shares to households (exchange
b). The household, in turn, obtains the $100 used
to purchase bond fund shares by withdrawing
$100 from its bank checking account (exchange c).
In essence, the $100 that the household shifts into
bond funds eventually goes back to the bank
when the firm issuing bonds pays off its C&I loan.
Another way of showing this equilibrium is
to review each party’s balance sheet, as depicted
in Figure 7. On the firm’s balance sheet, total
liabilities are unchanged as the $100 increase in
bonds issued matches the $100 decline in C&I loans.
For the household, the $100 increase in bond fund
holdings matches the $100 decline in M2 balances.
Notice that total assets and total liabilities are unchanged for the firm and household. The bond
fund, however, experiences a $100 increase in both
assets (the $100 rise in bonds) and liabilities (the
$100 increase in mutual fund shares). By contrast,
the banking industry is hit with a $100 decline in
C&I loans on the asset side that is matched by a
$100 decline in M2 deposits on the liability side. If
the only source of inflows into bond funds came
from M2 balances, then one way to solve this case
of missing money would be to add bond funds to
M2, much like adding MMMFs to M2, provided
that one or more variables could be found to consistently measure the desire to hold bond funds.
This example illustrates how a case of missing
money can arise when M2’s demand curve shifts,
and the shift is not the result of a change in income.
Federal Reserve Bank of Dallas
Thus, missing money likely can be accounted for by
unusual events that cause either possible changes
in the elasticity of money demand with respect to
income or opportunity costs, or permanent declines
in money demand for given levels of income and
opportunity cost measures.
The mid-1970s case of missing money. Until
the early 1980s, the monetary aggregate most
closely monitored by the Federal Reserve was
M1A, which was defined as currency plus demand
deposits.16 Based on its prior relationship to income
and interest rates, M1A was unusually weak in the
mid-1970s, leading one monetary economist to
call this episode “The Case of the Missing Money”
(Goldfeld 1976).
The mid-1970s were characterized by shocks
to both sides of bank balance sheets and by a
severe recession. On the asset side, there was
unusual weakness in C&I loans. Many large firms
shifted from C&I loans to commercial paper and
finance company loans for three reasons. First,
because market interest rates rose above deposit
rate ceilings set under Regulation Q, depositors
shifted funds away from banks and thrifts toward
investments bearing market interest rates (see the
box entitled “Regulation Q and the Competitiveness
of Banks and Thrifts”). Owing to a shortage of
loanable funds, banks and thrifts rationed credit
with nonprice terms, which drove larger firms to
credit sources that were unaffected by Regulation Q
(Figure 8).17 (Mortgage borrowers had fewer alternatives and as a result, there was a sharp decline
in housing construction [ Jaffee and Rosen 1979
and Hendershott 1980]). Second, partly to free up
funds for other borrowers, some banks provided
lines of credit to back up commercial paper issuance
to encourage their largest borrowers to make this
shift.18 Third, the rise in short-term rates increased
the reserve requirement tax on banks, and banks
passed this extra cost onto borrowers by raising
the prime rate relative to market interest rates.
This increase in the cost of C&I loans relative to
commercial paper encouraged many large firms to
shift to paper as a source of finance.19 This shift
was permanent for many firms because once they
incurred the fixed costs of becoming a paper issuer,
it was cheaper to bypass bank loans whose cost
was inflated by reserve requirements.
On the liability side of bank balance sheets,
there was unusual weakness in business holdings
Economic Review — Second Quarter 1993
Figure 8
The Nonbank Share of Short-Term Credit
(Seasonally Adjusted)
Percent
55
50
45
40
35
30
25
20
’60 ’62 ’64 ’66 ’68 ’70 ’72 ’74 ’76 ’78 ’80 ’82 ’84 ’86 ’88 ’90
of demand deposits at a time when the interest
prohibition on demand deposits was at a thenrecord binding level. These conditions were accompanied by firms’ entering repurchase agreements
(RPs) and purchasing overnight Eurodollars (Tinsley,
Garrett, and Friar 1981), likely reductions in compensating balances owing to firms’ borrowing less
from banks and shifting to commercial paper
(Duca 1992b), and firms’ incurring fixed costs to
initiate cash management techniques that reduced
their need for demand deposits (Porter, Simpson,
and Mauskopf 1979).
Thus, this missing-money episode occurred
at a time when both bank assets and bank liabilities
were unusually weak. Moreover, the missing money
16
Demand deposits are noninterest-bearing deposits that
are checkable.
17
Finance companies raise funds mainly by issuing commercial paper.
18
By providing liquidity to firms in a pinch, such backup lines
reduce the risk to investors holding commercial paper.
19
Most commercial paper is not subject to reserve requirements because it mostly is held directly by firms and
households or is held by money funds.
11
of the mid-1970s can be interpreted as having
stemmed from a decline in the competitiveness of
the banking system that resulted from an interaction
between high interest rates and bank regulations
(such as Regulation Q and reserve requirements).
In terms of the model in the first section, replace
M2 with M1A, whose opportunity cost is some
short-term T-bill rate. In reference to Figure 5, the
C&I loan shock to bank balance sheets reduced the
need of banks to issue deposits, thereby shifting
the supply of deposits curve leftward (shift 1),
and the shift away from compensating balances
and demand deposits toward RPs, Eurodollars,
and cash management can be represented by an
inward shift in the demand curve for M1A (shift 2).
As a result, the level of M1A is lower than suggested
by its opportunity cost and by nominal income.
The missing M1A and growth of money funds
in the late 1970s. In the late 1970s and early
1980s, another case of missing M1A coincided
with large inflows into nonbank types of deposits,
namely MMMFs, overnight repurchase agreements,
and overnight Eurodollars. During the late 1970s,
these new instruments grew rapidly, while demand
deposits were unusually weak (Wenninger, Radecki,
and Hammond 1981 and Dotsey, Englander, and
Partlan 1981). Owing to high nominal rates, a high
reserve requirement penalty was in effect, and
Regulation Q ceilings were binding on many smaller
banks and thrifts that were not well established
enough to issue large time deposits that were not
subject to interest rate ceilings.
On the asset side of depository balance sheets,
many firms shifted toward commercial paper. In
addition, the advent of market-rate-based money
market and small-saver certificates reduced the
funding cost advantage of banks over nonbanks.
As a result, bank auto loan rates rose toward
finance company rates, and banks lost market share
to finance companies. Once again, unusual weakness in demand deposits coincided with declines
in depository assets and liabilities that can be traced
to regulatory effects. The decline in the competitiveness of depositories was also accompanied by
a surge in MMMFs so large that removing MMMFs
from M2 before 1980 would have reduced M2
growth in the late 1970s by 1 to 3 percentage
points (Figure 9 ).
In terms of Figure 5, the reduced demand
for C&I loans (shift 1) and for demand deposits by
12
Figure 9
The Growth Rates of M2 and M2 Minus MMMFs
Percent
20
M2
M2 Less MMMFs
16
12
8
4
0
’60
’62
’64
’66
’68
’70
’72
’74
’76
’78
’80
firms (shift 2) during this episode is similar to the
mid-1970s case of the missing money, as is the
combination of reduced demand for loans (shift 1)
and M2 deposits (shift 2) by households.
In response to this episode of missing money,
the Federal Reserve redefined the monetary aggregates and expanded the definition of M2 to include
the new innovations. Before 1980, there was no
published monetary aggregate that resembled the
current definition of M2. Instead, the Federal
Reserve published several aggregates that reflected separations of banks from thrifts and some
aggregations of small and large time deposits. In no
case were MMMFs, RPs, and Eurodollars included
in aggregates published by the Federal Reserve
until the official redefinition of M2 in early 1980
(Simpson 1980).
The current case of missing money. The current
episode of missing M2 has been accompanied by
1. rapid inflows into bond and equity
funds that are not consistently
linked to spreads between shortand long-term interest rates,
2. heavy corporate bond and equity
issuance,
3. edit-check reports to the Federal
Reserve Board staff of large paydowns of C&I loans by corporations
issuing bonds,
Federal Reserve Bank of Dallas
4. Resolution Trust Corporation (RTC)
activity, which has reduced the
demand for M2 in unusual ways and
which has reduced both the asset
and liability sides of depository
balance sheets, and
5. the institution of new risk-based
capital standards that may have
widened the spread between the
prime and short-term interest rates.
These phenomena have arguably reduced or been
a reflection of a decline in the competitiveness of
depositories as financial intermediaries.
On the asset side of depositories, the wide
spread of prime over short-term market rates has
encouraged many firms to shift away from C&I
loans. Bond issuance has surged the past two
years, and although commercial paper has grown
less robustly, it has grown faster than bank loans.
The discrepancy between bond and paper issuance
partly reflects that corporations are refinancing
long-term debt. In addition, a reduction in the
competitiveness of prime rate financing could
affect more firms that could issue bonds than
firms that could issue commercial paper. The
reason is that only a subset of firms that can issue
bonds are well-known enough to issue commercial paper, especially since the Securities and
Exchange Commission (SEC) restricted the extent
to which money market mutual funds could purchase commercial paper with ratings below A1/P1
(Crabbe and Post 1992).
Nevertheless, there has been some shift away
from bank loans to commercial paper that may
reflect the wider spread between the prime rate
and high grade commercial paper that has persisted
since year-end 1990 (Figure 10 ). That widening
coincided with the implementation of new and
tougher risk-based capital standards on banks that
increased the cost of C&I loans (see the box titled
“The Impact of Risk-Based Capital Standards and the
FDIC Insurance Premium Hike”). Aside from commercial borrowing, a wide spread between consumer loan rates and M2 deposit rates is encouraging
households to withdraw M2 funds to pay off consumer loans (Feinman and Porter 1992). Nevertheless, for firms without access to the bond markets
and for households that cannot self-finance purchases, the increased regulatory burden on banks
Economic Review — Second Quarter 1993
Figure 10
The Spread Between the Prime and
One- and Two-Month Commercial Paper Rates
Percentage points
6
P
T
PT
P
T
P T
New
Capital
Standards
4
2
0
–2
’71
’73
’75
’77
’79
’81
’83
’85
’87
’89
’91
has likely lowered investment and consumption
spending by increasing the cost of bank financing.
On the liability side of depository balance
sheets, several unusual factors are affecting the
demand for M2. First, RTC activity has created a prepayment risk on M2 deposits that is not measured
by spreads between market and deposit interest
rates. As a result, these measures of M2’s opportunity cost are understating M2’s true opportunity
cost, thereby leading money demand models to
overpredict M2 growth (Duca, forthcoming). Second,
RTC resolution activity also has accelerated the
adjustment of deposits to a lower interest rate
environment by prematurely ending small time
deposit contracts. Third, RTC effects and the recent
large spread between short-term and long-term
interest rates may have induced the public to gather
information about long-term, non-M2 assets such
as bond and equity funds (Feinman and Porter
1992). Fourth, the same factors apparently led
mutual funds to increase their advertising of their
products and induced several large banks to begin
marketing bond and equity funds to their depositors
(Cope 1992). These actions may have led to a discontinuous portfolio reallocation by households
from M2 toward bond and equity mutual funds,
thereby causing unusual weakness in M2 as it is
currently defined.
13
Figure 11a
Figure 11b
Money Market Equilibrium
OC (r)
r
Ms
LM
OC1
r1
r0
OC0
Md(y1)
Md(y0)
m
y0
their demand for consumer loans, which causes
an inward shift of the bank supply of deposits
curve (shift 1).
Similar shifts can also plausibly arise from RTC
resolution activity. RTC resolutions effectively swap
Treasury debt and thrift assets for thrift deposits,
thereby shifting inward the deposit supply curve
(shift 1).
What missing money implies for monetary
policy. As stressed by Poole (1970), unusual
changes in the demand for money reduce the
ability of a monetary aggregate target to stabilize
aggregate demand. This can be shown by deriving
In terms of Figure 5, the shift by corporations
from C&I loans to bonds and equity finance is
similar to the shift toward commercial paper in the
mid-1970s, and substitution by households from M2
into bond and equity mutual funds is similar to the
shift toward MMMFs in the mid-1970s. Incentives
for households to reduce their assets and liabilities
also can induce similar supply and demand curve
shifts. As discussed in Feinman and Porter (1992),
wide spreads between loan and deposit rates offered
to households encourage them to reduce their
demand for M2 at given levels of income and
opportunity cost measures (shift 2), and to reduce
Figure 12a
y
y1
Figure 12b
A Money Target Reduces
the Impact of IS Shocks
OC
Ms
(money target)
r
Ms
LM1
r2
OC2
LM0
r1
OC1
C
B
OC0
rA
IS1
A
d
Md(y0)
Ms0
14
M (y1)
IS0
m
yA
yC yB
y1
y
Federal Reserve Bank of Dallas
Figure 13a
Figure 13b
A Money Target Increases the Impact
of Money Demand Shocks
OC
r
Ms (money target)
LMMA
Ms
LMA
LMM
C
OCA
rA
OC1
r1
LMB
A
B
C
Md0(yA)
OC2
r2
IS0
Md1(yA)
Ms0
m
the conditions under which the supply and demand
for money are in equilibrium (the LM curve) and
then solving for nominal output by combining the
LM curve with the IS curve from the first section
of this article.
In terms of the money demand and money
supply curves in Figure 5, a rise in income will
shift the money demand curve (M d ) out and to the
right (Figure 11a). This change implies that the
combination of opportunity cost terms and income
levels at which the demand for money equals its
supply can be depicted by the upward sloping
line in Figure 11b. Figure 11a assumes that the
opportunity cost of money increases with the shortterm market interest rate. An assumption implicit
in the upward sloping money supply curve in
Figure 11a is that the central bank will allow a
rise in income to boost money balances.
If, however, the central bank prevents the
money balances from changing, then the money
supply schedule is essentially made vertical. This
effect can be shown to make the LM curve steeper.
Figure 12a indicates that an increase in income
from Y0 to Y1 will shift the money demand curve
to the right. If the money supply curve is vertical,
the opportunity cost of money would rise to OC 2 ,
whereas if the money supply curve had a nonvertical upward slope, the opportunity cost of
money would rise to only OC1. Since the opportunity cost of money increases when the market
interest rate is higher, when income is Y1, the
equilibrium market interest rate that clears the
Economic Review — Second Quarter 1993
yA
yB yC
y
money market under a fixed-money policy is
higher (r2 ) than when the money supply curve
slopes upward (r1 ). Thus, fixing money balances
makes the LM curve steeper.
A steeper LM curve has important policy
implications. If we combine this steeper LM curve
with the IS curve from the first section of this
article, we can see that the impact of a given IS
shock on nominal output is smaller. In Figure 12b,
the economy is initially at point A with nominal
output at YA. The initial equilibrium is at point A
because point A is the only combination of nominal
output (Y ) and the market interest rate (r ) at which
both goods market (IS ) equilibrium and money
market (LM ) equilibrium occur. If the IS curve
shifts rightward from IS0 to IS1, then the new equilibrium under a fixed money rule is at point C,
whereas the new equilibrium is at point B when
under the flexible money supply policy. Notice how
output is less affected by an IS shock when the
LM curve is steeper (YC is closer than YB to YA ).
But if some change in the economic environment also affects the money demand curve, then
stabilizing the level of the money supply may further
destabilize nominal output. Consider the increased
popularity of bond funds, which causes the public
to demand less money at each combination of
income and opportunity cost of money. Then, as
shown in Figure 13a, the money demand curve
shifts inward. As a result, the level of nominal
income must be higher for the same initial level of
money to be held in equilibrium for a given level
15
Figure 14
How a Bypassing of the Banking System Is
Problematic for Interest Rate and Money Targeting
r
LM0
LM1
A
rA
D
rB
B
C
IS0
IS1
yD
yB yA
yC
y
of opportunity cost. This implies that both LM
curves in Figure 13b shift to the right by the same
horizontal distance and that the level of income
rises. However, if money balances are stabilized,
the money demand shock pushes the economy to
point C rather than point B. As a result, nominal
output is more affected by a money demand shock
when the LM curve is steeper (YC is further away
from YA than is YB ). Because a steeper LM curve
implies that money demand shocks have a greater
destabilizing effect on output, money targeting
becomes less useful when the demand for money
is unstable.
Now consider the impact of a rise in bank
regulatory burden on both the IS and LM curves,
assuming that the Federal Reserve does stabilize
money held (in other words, that the LM curve is
steep). As demonstrated in the first section of this
article, the resulting increased cost of credit to
borrowers causes the IS curve to shift inward to
the left. In addition, as discussed earlier in this
section, the reduced competitiveness of the banking
20
16
This qualitative result is also obtained if the central bank
does not rigidly target money balances, thereby making the
LM curve less steep. Quantitatively, however, the problem
of inference using a monetary targeting perspective is
smaller when the LM curve is less steep.
system will likely be accompanied by an unusual
decline in the demand for money (that is, a leftward shift of the money demand curve) that causes
a rightward shift in the LM curve. In Figure 14, the
economy is initially at point A. If the IS shift is
large enough to outweigh the LM shift, then increased regulatory burden on banks will result in
weakness in nominal income and a decline in
short-term interest rates (point B) that accompanies
both a case of missing money and a decline in the
share of credit supplied by banks. These results,
which are broadly consistent with recent events,
are also obtained if the LM curve is less steep.
Unusual shifts in either the IS or LM curves
complicate policy-making aimed at stabilizing aggregate demand because policymakers readily observe
interest rates, whereas estimates of nominal GDP
are available after a lag and are subject to substantial revision. In the context of Figure 14, notice how
the new level of the short-term market interest
rate under states economic weakness if it is assumed
that the IS curve did not shift. In this case, consider what would happen if an analyst mistakenly
assumed that a money demand shock (an unusual
decline in money demand) caused the LM curve to
shift rightward so that the new LM curve intersected
the IS curve at point C. Based on this assumption,
one would mistakenly infer from the new interest
rate level (rC ) that the economy was at point C
rather than point B and that nominal output was
YC rather than YB .
At the same time, the unusual weakness in
money balances held over states economic weakness if one assumes that a money demand shock
did not occur but that the IS curve shifted. In this
case, one would infer from the original LM curve
and a market interest rate of rC that the IS curve
shifted left to put nominal output at point D and
that income has fallen to YD , which is less than
YC .20 Thus, because changes in bank regulations
can, in principle, shift both the IS and LM curves,
they can create problems for the use of either an
interest rate or money target.
Furthermore, the dichotomy of overestimating GDP growth from an interest rate targeting
perspective and underestimating GDP growth from
a money targeting perspective from this example
may have relevance for recent events. In particular,
this dichotomy parallels the tendency of most major
economic forecasters to have overpredicted GDP
Federal Reserve Bank of Dallas
growth during 1991–92 (in other words, they
assumed that the economy was at a point like C in
Figure 14), while M2 growth suggested that GDP
growth should have been weaker than it actually
was (point D).
Conclusion
Evidence from three cases of missing money
indicates that factors reducing the competitiveness
of banks accompanied each episode. In the two
earlier cases (1974–75, 1979–80), the interaction
between controls on deposit rates and high market
interest rates spawned innovations and reactions
that reduced M1 growth. The subsequent policy
responses of deregulating deposit rates and of preventing inflation from accelerating have prevented
these types of factors from spawning further innovations that destabilize money demand.
The most recent case of missing money also
reflects how regulatory and nonregulatory factors
have encouraged firms and households to bypass
banks and thrifts. On the asset side of bank balance
sheets, risk-based capital standards have raised
the cost of bank loans, which in turn has encouraged firms to shift away from bank finance and
some households to pay down consumer debt by
drawing down their bank deposits. On the liability
side of bank balance sheets, the steepening of the
yield curve, the depressing effects of higher FDIC
insurance premiums on deposit rates, and RTC
resolution activity have encouraged households to
shift away from bank deposits to bond (and perhaps
equity) mutual funds (Duca 1993). As a result of
these factors, both sides of bank and thrift balance
sheets have declined in unusual ways. This combination of influences is suggestive of leftward
shifts in the supply and demand for money, and
thus may account for why money demand models
are overpredicting M2 growth.
The appropriate regulatory response to the
recent case of missing money is less clear-cut than
Economic Review — Second Quarter 1993
in earlier episodes. While capital standards and
increases in risk-based deposit insurance premiums
have ostensibly induced banks to widen spreads
between loan and deposit rates, they also have
the desirable effect of shifting the downside risk
of lending away from taxpayers to bank equity
holders. Determining whether risk-based capital
standards and deposit insurance premiums are
appropriate is beyond the scope of this article.
Nevertheless, this study has several implications for monetary policy. First, changes in the
competitiveness of the banking system can alter
the information content of monetary aggregates.
Second, the demand for money can be altered by
factors affecting long-term market interest rates.
As argued by Feinman and Porter (1992), both
considerations suggest that the Federal Reserve
does not have as much direct control over M2 as
previously thought, implying that the monetary
aggregates need to be interpreted in more complicated ways than previously thought.
Third, by causing IS (goods market) and LM
(money market) disturbances, changes in the regulatory burden on banks have created problems for
both interest rate and monetary aggregate targeting in three recent recessions. By implication, conducting a sound monetary policy is not as easy as
either hindsight or ex post monetary indicators
suggest. As a result, achieving broad economic
goals requires that a central bank exercise a good
deal of judgment and discretion in conducting its
operating procedures.
Finally, the previous and current episodes of
missing money imply that the Federal Reserve
should take an active role in policy actions that
affect the competitiveness of the banking system
and ensure that the consequences of such actions
for the implementation of monetary policy are
taken into account when formulating these policies.
For this reason, the Federal Reserve must have
significant input into the regulation of banks if it
is to fulfill its mission as a central bank.
17
Regulation Q and the Competitiveness of Banks and Thrifts
The degree to which Regulation Q put
banks and thrifts at a competitive disadvantage in raising loanable funds can be gauged
by measuring the extent to which market
interest rates rose above deposit rate ceilings. The measurement of Regulation Q
effects raises three issues:
1. which retail deposit rate to use,
2. whether rate ceilings for thrifts or
banks should be used, and
3. how to handle the introduction of
market-rate-based deposit instruments prior to the lifting of all rate
ceilings on nontransactions deposits
in 1983.
With respect to issue 1, for two reasons
the Regulation Q variable presented here
reflects regulations affecting small time deposits. First, because small time deposits
serve more as savings rather than transactions instruments, the small time deposits are
more sensitive to their opportunity cost than
are other types of household M2 deposits.
Second, most market-based deposit instruments that were introduced in the late 1970s
were, by design, substitutes for small time
deposits.
In handling issue 2, rate ceilings on
thrifts were used because regulations tended
to favor thrifts since rate ceilings on thrift
accounts were as high as, if not higher than,
those on bank deposits. In addressing issue
3, there were two basic types of partially
regulated, deposit-type instruments that were
introduced before 1983 by law: small-saver
certificates and money market certificates.
Small-saver certificate regulations were used
in constructing a Regulation Q variable be-
18
Figure A
The Bindingness of Regulation Q Ceilings
on Deposit Rates
Percentage points
3.5
3
2.5
2
1.5
1
.5
0
–.5
’59
’62
’65
’68
’71
’74
’77
’80
’83
’86
’89
’92
cause minimum balance requirements on
small-saver certificates ($500 to $1,000) were
much more similar to those on retail deposits
than were the requirements on money market
certificates ($10,000) over most of the late
1970s and early 1980s.
Given these considerations in dealing
with issues 1, 2, and 3, the Regulation Q
measure here is defined using spreads between market interest rates and rate ceilings
on small time deposits and/or small-saver
certificates. Between 1960 and the second
quarter of 1978, this measure (REGQ) equals
the quarterly average spread between the
three-year Treasury rate and the rate ceiling
on three-year small time deposits when the
ceiling was binding, and zero otherwise. Starting in the third quarter of 1979 when small(Continued on the next page)
Federal Reserve Bank of Dallas
Regulation Q and the Competitiveness of Banks and Thrifts—Continued
saver certificates were created, REGQ equals
one of the following based on quarterly averages of monthly data:
a. any ceiling spread set by legislation
between market interest rates and
rates on small-saver certificates,1
b. the maximum of zero and the difference between the 21/2 -year Treasury
yield (constant maturity) and any legislated cap on small-saver rates,2 or
c. zero since August 1981 when rate
ceilings on small-saver certificates
were removed.
For details on deposit regulations, see Mahoney, White, O’Brien, and McLaughlin (1987).
As can be seen in Figure A, Regulation
Q was very binding in the 1974–75 and 1979–
Economic Review — Second Quarter 1993
80 periods when missing-money problems
were arising for M1. These disintermediation
effects were largely ended when rate ceilings
were dropped on small-saver certificates in
August 1981 and were completely eliminated
with the lifting of all ceilings on nonbusiness
deposit rates in 1983. The earlier episodes of
binding Regulation Q ceilings (the early 1960s
and again in 1967) were not accompanied by
missing-money episodes, mainly because
they were not accompanied by innovations
(such as the creation of money substitutes)
that affected the demand for money.
1
These set spreads ranged from zero to 50 basis points.
2
Between January 1980 and August 1981, ceilings on smallsaver yields were based on the 21/2-year constant maturity
Treasury yield.
19
The Impact of Risk-Based Capital Standards
and the FDIC Insurance Premium Hike on Banks’ Costs
While U.S. commercial banks were not
subject to a minimum capital rule before
year-end 1990, they attempted to meet an
unofficial goal of maintaining a minimum
ratio of 6 percent total capital (equity plus
subordinated debt) to assets. Using that ratio
as a base, the capital standards that were
fully implemented at year-end 1992 raised
the effective minimum ratio of total capital to
loans from 6 percent to 8 percent. In light of
emerging loan quality problems in 1990, many
large banks acted as though capital standards were fully implemented at year-end
1990 to reassure market investors that the
banks could meet the final phase-in of capital
standards at year-end 1992.
The effect of new capital standards on
the marginal cost of lending roughly equals
the additional capital banks need (0.08–0.06
percent) multiplied by the extent to which the
yield on capital (ROE ) exceeds that on insured deposits (r d ). Because most banks
cannot issue subordinated debt, assume that
capital costs roughly equal a targeted yield on
the return on equity (ROE ) capital. Based on
anecdotal evidence, let’s use a target ROE
goal of 15 percent. For the yield on deposits,
let’s use 4 percent, which roughly approximates the average rate on six-month time
deposits over most of 1992. Based on these
figures, the new capital standards raised the
cost of funding C&I loans by 0.22 percent
(0.02 x 0.11). In addition, in the second half of
20
1990 the Federal Deposit Insurance Corporation (FDIC) announced that it would increase
the insurance premium levied on insured deposits by 0.075 percent, from 12 to 191/2 cents
per $100 of deposits. To remain profitable,
banks eventually would need to pass on the
extra costs of tougher capital standards and
higher insurance premiums (0.295 percent)
to their customers in the form of a wider
spread between loan and deposit rates.
How does 0.295 percent compare with
the pricing of the prime lending rate? Banks
typically set the prime rate equal to the cost of
borrowing overnight funds in the federal funds
market plus some spread to compensate themselves for administrative costs, default risk,
and some target return to equity holders.
Because default risk varies with the business
cycle, the spread between interest rates on
bank loans and a competing source of credit
can be used as an indicator of how competitive banks are in providing C&I loans, provided that the spread moves with default risk.
One increasingly popular interest rate
spread is that between the prime rate and
commercial paper (Friedman and Kuttner
1992). Compared with the calculated impact
of both capital standards and deposit insurance changes (0.295 percent), the spread
between the prime rate and the one- to twomonth prime commercial paper rate rose by a
(Continued on the next page)
Federal Reserve Bank of Dallas
The Impact of Risk-Based Capital Standards
and the FDIC Insurance Premium Hike on Banks’ Costs—Continued
somewhat higher 0.50 percent near year-end
1990, as indicated earlier in Figure 10. The
somewhat greater rise in this spread partly
reflects the requirement that many banks
have risk-based capital ratios greater than 8
percent, based on regulator assessments of
the bank’s soundness. For this reason, the
calculated effect of the new bank capital standards presented here likely understates their
average effect on banks’ cost of funding loans.
The remainder of the increase in the
spread may also reflect a slight increase in
the default risk on bank loans compared with
Economic Review — Second Quarter 1993
commercial paper. While both rates rise with
a pervasive increase in default risk, this
spread tends to widen temporarily during
recessions, perhaps because the issuers of
commercial paper generally are more established firms and, during tough times, are less
prone to default on loans. Nevertheless, this
spread has not narrowed to prerecessionary
levels. This evidence and the fact that the
spread widened during the phasing-in of the
new bank capital standards strongly suggest
that the new bank capital standards have
raised the cost of prime-based bank loans.
21
References
Cook, Timothy, and Thomas Hahn (1989), “The
Effect of Changes in the Federal Funds Rate
on Market Interest Rates in the 1970s,” Journal of Monetary Economics 24 (November):
331–51.
Cope, Debra (1992), “Banks Making Inroads on
Mutual Funds,” American Banker (December
9): 1, 12.
Crabbe, Leland, and Mitchell A. Post (1992), “The
Effect of SEC Amendments to Rule 2A–7 on
the Commercial Paper Market,” FEDS Working
Paper no. 199, Board of Governors of the
Federal Reserve System (May).
Diamond, Douglas W. (1991), “Monitoring and
Reputation: The Choice between Bank Loans
and Directly Placed Debt,” Journal of Political
Economy 99 (August): 689–721.
Dotsey, Michael, Steve Englander, and John C.
Partlan (1981), “Money Market Mutual Funds
and Monetary Control,” Federal Reserve Bank
of New York Quarterly Economic Review
(Winter): 9–13.
Duca, John V. (forthcoming), “RTC Activity and
the ‘Missing M2’,” Economics Letters.
——— (1993), “Should Bond Funds Be Added to
M2?” unpublished manuscript, Federal Reserve
Bank of Dallas.
——— (1992a), “The Case of the Missing M2,”
Federal Reserve Bank of Dallas Economic
Review (Second Quarter): 1–24.
——— (1992b), “U.S. Business Credit Sources,
Demand Deposits, and the ‘Missing Money’,”
Journal of Banking and Finance 16 ( July):
567–83.
Federal Reserve Bulletin (1993), (Washington,
D.C.: Federal Reserve System), January.
Feinman, Joshua, and Richard D. Porter (1992),
“The Continued Weakness in M2,” FEDS
22
Working Paper no. 209, Board of Governors
of the Federal Reserve System (September).
Friedman, Benjamin, and Kenneth Kuttner (1992),
“Money, Income, Prices, and Interest Rates,”
American Economic Review 82 ( June): 472–92.
Goldfeld, Stephen M. (1976), “The Case of the
Missing Money,” Brookings Papers on Economic Activity 7(3): 683–730.
Hallman, Jeffrey J., Richard D. Porter, and David
H. Small (1991), “Is the Price Level Tied to the
M2 Monetary Aggregate in the Long Run?”
American Economic Review 81 (September):
841–58.
Hendershott, Patrik (1980), “Real User Costs and
the Demand for Single-Family Housing,”
Brookings Papers on Economic Activity vol. 2:
401–44.
Jaffee, Dwight M., and Kenneth Rosen (1979),
“Mortgage Credit Availability and Residential
Construction,” Brookings Papers on Economic
Activity vol. 2: 333–76.
Keeley, Michael C. (1990), “Deposit Insurance,
Risk, and Market Power in Banking,” American
Economic Review 80 (December): 1183–1200.
Mahoney, Patrick I., Alice P. White, Paul F. O’Brien,
and Mary M. McLaughlin (1987), “Responses
to Deregulation: Retail Deposit Pricing From
1983 through 1985,” Staff Study no. 151, Board
of Governors of the Federal Reserve System
( January).
Miller, Merton H., and Daniel Orr (1966), “A Model
of the Demand for Money by Firms,” Quarterly
Journal of Economics 80 (August): 413–35.
Moore, George R., Richard D. Porter, and David
H. Small (1990), “Modeling the Disaggregated
Demands for M1 and M2 in the 1980’s: the
U.S. Experience,” in Financial Sectors in Open
Economies: Empirical Analysis and Policy
Issues, eds. P. Hooper, K. H. Johnson, D. L.
Federal Reserve Bank of Dallas
Kohn, D. E. Lindsey, R. D. Porter, and R. Tryon
(Washington, D.C.: Board of Governors of the
Federal Reserve System): 21–105.
Poole, William (1970), “Optimal Choice of Monetary
Policy Instruments in a Simple Stochastic Macro
Model,” Quarterly Journal of Economics 84
(May): 197–216.
Porter, Richard D., Thomas D. Simpson, and Eileen
Mauskopf (1979), “Financial Innovation and
the Monetary Aggregates,” Brookings Papers
on Economic Activity 10 (1): 213–29.
Post, Mitchell A. (1992), “The Evolution of the U.S.
Commercial Paper Market since 1980,” Federal
Reserve Bulletin (December): 879–91.
Economic Review — Second Quarter 1993
Simpson, Thomas D. (1980), “The Redefined Monetary Aggregates,” Federal Reserve Bulletin
(February): 97–114.
Small, David H., and Richard D. Porter (1989),
“Understanding the Behavior of M2 and V2,”
Federal Reserve Bulletin (April): 244–54.
Tinsley, P. A., B. Garrett, and M.E. Friar (1981),
“An Exposé of Disguised Deposits,” Journal of
Econometrics 15 ( January): 117–37.
Wenninger, John, Lawrence Radecki, and Elizabeth Hammond (1981), “Recent Instability in
the Demand for Money,” Federal Reserve
Bank of New York Quarterly Economic Review
(Summer): 1–9.
23
24
Federal Reserve Bank of Dallas
David M. Gould
Roy J. Ruffin
Economist
Federal Reserve Bank of Dallas
Professor of Economics
University of Houston
What Determines Economic Growth?
S
ince 1973, per capita income growth in the
United States and other advanced countries has
slowed to 2.2 percent a year, or almost half the
3.9-percent annual rate of the preceding quarter
century. If the United States had maintained the
level of growth experienced in the 1950s and 1960s,
real per capita income today would be about 11
percent ($2,200 in 1987 dollars) greater than it
actually is. In contrast, it has been estimated that
eliminating the variability in U.S. consumption
since World War II would be equivalent to boosting current real consumption by only about 4.8 percent ($420 in 1987 dollars).1 If the choice is between
long-term growth policies and further short-term
stabilization policies, long-term growth policies
clearly have the potential for vastly higher benefits.
Perhaps the reason why economists have
neglected long-run economic growth is that, for a
long time, the profession relied on a theory that
offered little scope for policy to influence important sources of growth. According to traditional
growth theory, the main determinants of long-run
economic growth are not influenced by economic
incentives. Recently, however, the study of economic growth has been reinvigorated by new
developments in theory and empirical findings
that suggest growth is in the sphere of policy.
This new literature, referred to as endogenous
growth theory, helps to explain movements in
long-term growth and why some countries grow
faster than others.
Because long-term economic growth is the
fundamental determinant of whether our grandchildren will have better lives than ours or whether
the poor nations will catch up with or fall further
behind the rich nations, this article attempts to
summarize what economists have learned about
economic growth and applies recent empirical
findings to the above issues.
Economic Review — Second Quarter 1993
The first section examines the long-term
growth record, focusing on the extent of growth
variations across countries and across decades.
The second section presents the traditional growth
model and recently developed endogenous growth
models. The next section discusses whether poor
countries are catching up with richer nations or
whether the rich are getting relatively richer. The
fourth section examines factors that have been
found to influence long-run economic growth, and
the last section presents lessons for the future.
A historical perspective on economic growth
Despite the recent slowdown in economic
growth, long-run growth not only has persisted
since the early nineteenth century but has accelerated. Maddison (1991) has documented the persistence and acceleration of economic growth for
14 advanced capitalist countries (Table 1). The
annual growth rate of the 14 countries averaged
only 0.9 percent from 1820 to 1870 but rose to 1.6
percent between 1870 and 1989. For the forty
years from 1950 through 1989, growth in these
John V. Duca offered many helpful comments as the reviewer for this article. We also benefited from the discussions and comments of W. Michael Cox, Dani Ben-David,
Ping Wang, and Mark A. Wynne. All remaining errors are
solely our responsibility.
1
Lucas (1987, 27). Lucas estimates that eliminating the
variability in U.S. consumption would be equivalent to
increasing average real consumption about 0.1 percent per
year. However, if current income volatility has effects on
future productivity, as allowed for by Ramey and Ramey
(1991), the long-run costs of volatility may be higher.
25
Table 1
Growth Rates of Per Capita Real GDP, by Country
(Annual Averages)
1820–1870
1870–1989
1950–1973
1.9
1.2
2.4
1.7
Austria
.6
1.8
4.9
2.4
Belgium
1.4
1.5
3.5
2.0
Denmark
.9
1.8
3.1
1.6
Finland
.8
2.3
4.3
2.7
France
.8
1.8
4.0
1.8
Germany
.7
2.0
4.9
2.1
Italy
.4
2.0
5.0
2.6
Japan
.1
2.7
8.0
3.1
Netherlands
.9
1.5
3.4
1.4
Norway
.7
2.2
3.2
3.6
Sweden
.7
2.1
3.3
1.8
United Kingdom
1.2
1.4
2.5
1.8
United States
1.5
1.9
2.2
1.6
.9
1.6
3.9
2.2
Australia
Average
1973–1989
SOURCE: Maddison (1991).
countries was at an even higher rate (3.2 percent),
despite the slowdown since 1973.
Maddison (1991) and Romer (1986) have
shown that the leading technological country,
defined in terms of productivity per worker hour,
has experienced increasing rates of growth since
1700. Intuitively, this pattern may be a consequence
of the creation of new technology in the leading
country. According to Maddison, there have been
only three technological leaders in the past four
centuries: the Netherlands, the United Kingdom,
and the United States. Maddison’s research shows
that the growth rate of the leading country in
three successive periods increased relative to that
of the leader in the preceding period (Table 2 ).
Using annual data spanning up to 130 years
26
across 16 countries, Ben-David and Papell (1993)
find evidence of increasing per capita growth
rates of gross domestic product (GDP). They find
that in the more recent periods, trend GDP per
capita growth rates were, on average, 2.5 times
higher than the growth rates in the earlier periods.
Just by examining the GDP growth record in
the United States, we can see that growth over the
long run has tended to accelerate. Table 3 shows
the per capita growth of the United States in five
successive periods. The annual growth rate has
risen from 0.58 percent for 1800–1840 to 1.82
percent for 1960–91. Romer (1986) also found a
similar upward drift in the growth rates of per
capita GDP for all countries in the Maddison (1991)
sample for which data were available.
Federal Reserve Bank of Dallas
Table 2
GDP Growth Rates of Leading Technological Countries
Period
Leading country
at beginning
of period
Average annual
compound growth
rate (Percent)
1700–1820
Netherlands
–.05
1820–1890
United Kingdom
1.2
1890–1989
United States
2.2
SOURCE: Maddison (1991).
How reliable are the Maddison–Romer conclusions? Two factors—changes in household
production and upward biases in measures of the
rate of inflation—would seem to strengthen their
conclusion that growth has been persistent and
accelerating.
First, GDP does not cover household production and leisure. Because hours of work have
steadily decreased, it would seem that nonmarket
output should have increased relative to measured
Table 3
Per Capita Real GDP Growth
in United States
Period
Average annual compound
growth rate (Percent)
1800–1840
.58
1840–1880
1.44
1880–1920
1.78
1920–1960
1.68
1960–1991
1.82
SOURCES:
Economic Report of the President (1992).
Romer (1986).
Economic Review — Second Quarter 1993
GDP. Hours of work seemed to be roughly constant
at about 3,000 worker hours per year until about
1870, when they began to drop (Maddison 1991,
270–71, 276). Today, annual worker hours are about
1,600 in most of the advanced industrial countries,
which suggests a substantial increase in leisure
time. Leisure time, however, has probably not increased as much as worker hours have fallen because
labor force participation rates have increased.
Nonetheless, had the measures of GDP included
estimates of the value of household production
and leisure time, it is likely that the growth rates
would have been higher in the period since 1870.
Another factor understating the pickup in
living standards in this century is the overestimation of inflation. The rate of growth in real GDP is
the rate of growth in nominal GDP less the rate of
inflation as measured by some price index. But
price indexes tend to overstate changes in the cost
of living (Gordon 1992). They are biased upward
partly because they do not incorporate new products
in a timely fashion; for example, the consumer
price index (CPI) did not include automobiles
until 1940, several decades after production of the
Model T. Price indexes also do not completely
incorporate quality changes in existing products,
account for the substitution of cheaper goods for
more expensive goods over time, or take into consideration the availability of discount outlets. If the
CPI is off, say, 0.5 percent per year, then official
measures can understate the real rate of GDP
growth by the same magnitude. The substitution
27
bias has been estimated to be about 0.18 percent
per year in the United States, and quality changes
in U.S. consumer durables have biased the rate of
price increase in these products by about 1.5 percent per year (Gordon 1992). Thus, a bias of 0.5
percent or even 1 percent per year would not be
too farfetched.
What can explain accelerating growth over
nearly two centuries? Is the recent slowdown in
per capita growth a new trend or a temporary
setback? The following section addresses these
questions in discussing the two main theories of
economic growth.
role in determining the rate of technological innovation and long-run growth. The following subsections discuss the structure of the two models.
The Solow growth model. The traditional
growth model advanced by Robert Solow (1956),
a Nobel Prize winner, is perhaps the most famous
one.2 The key idea of this framework is that growth
is caused by capital accumulation and autonomous technological change. Solow views the
world as one in which output, Y, is generated by
the production function
Theories of economic growth
where K is the capital stock and L is the labor
force. Solow postulated that the production function displays constant returns to scale, so that
doubling all inputs would double output. However,
holding one input constant—say, labor—and
doubling capital will yield less than double the
amount of output. Referred to as the law of diminishing marginal returns, this is one of the distinguishing elements of the Solow model.
The Solow model is driven by savings and
variations in the ratio of capital to labor. Suppose
that k = K/L is the capital–labor ratio. It is convenient to begin with the observation that the percentage change in k equals the percentage change
in K less the percentage change in L; that is,
Growth is a complicated process, but the
main theories of economic growth are conceptually simple. There are basically two categories of
economic growth theories—those based on the
traditional Solow (1956) growth model and those
based on the concept of endogenous growth. The
Solow model emphasizes capital accumulation
and exogenous rates of change in population and
technological progress. This model predicts that
all market-based economies will eventually reach
the same constant growth rate if they have the
same rate of technological progress and population growth. Moreover, the model assumes that
the long-run rate of growth is out of the reach of
policymakers.
The recent proliferation of endogenous
growth models began with the work of Paul Romer.
Romer (1986) observed that traditional theory
failed to reconcile its predictions with the empirical
observations that, over the long run, countries
appear to have accelerating growth rates and,
among countries, growth rates differ substantially.
Endogenous growth theories are based on
the idea that long-run growth is determined by
economic incentives. The most popular models of
this type maintain that inventions are intentional
and generate technological spillovers that lower
the cost of future innovations. Naturally, in these
models an educated work force plays a special
2
28
For a recent exposition, see Wynne (1992).
(1)
(2)
Y = F (K, L),
k/k = K/K – L/L .
The change in the capital stock equals investment,
and investment equals the output that is saved
rather than consumed. Thus,
(3)
K = sY,
where s is the savings rate. Solow assumed that
both the savings rate, s, and the growth rate of
population, L/L , are constant. Substituting (3)
into (2) and multiplying by k yields
(4)
k = sY/L – k ( L/L).
Equation 4 has a simple interpretation. The
term sY/L is the amount of investment per unit of
the labor force. The term k( L/L) is the amount of
investment per worker that is necessary to maintain the capital–labor ratio, k = K/L . For example,
Federal Reserve Bank of Dallas
suppose K equals $5 million and L equals 100, so
k equals $50,000. Then, a growth rate of 0.02
percent for the population requires $1,000 of new
investment per worker ($100,000) to keep k equal
to $50,000.
The Solow model is depicted in Figure 1,
which measures investment per worker on the
vertical axis and the population capital–labor ratio
on the horizontal axis. The amount of investment
per worker, sY/L , increases at a decreasing rate
because the law of diminishing returns implies
that per capita output, Y/L , declines as k rises. On
the other hand, investment per worker required to
keep capital intact, k(∆L/L), rises steadily because
it is just proportional to k . Therefore, the two
curves are likely to intersect at some equilibrium
capital–labor ratio, k僓. When the capital–labor ratio
is less than k僓, actual investment per worker exceeds
that required to keep k constant, so k rises. When
the capital–labor ratio is more than k僓, investment
per worker falls short of that required to keep k
constant, so k falls. Thus, the economy gravitates
toward k僓. This is called a steady state because the
economy can persist forever at this point. The
capital stock and the level of output are rising at
the same rate as the growth in the population, so
per capita income, Y/L, does not change.
In the Solow growth model, where technological progress is exogenous, income will rise with
the level of physical or human capital (accumulated human knowledge), but the rise will not
generate ever-increasing growth rates. Skilled
workers increase the level of income, just like any
other productive factor, but they do not increase
growth in the long run because technological
progress does not depend on the presence of a
skilled work force.
The basic conclusion of the model is that the
rate of growth of the economy in the long run
simply equals the rate of growth in the labor force
plus the rate of exogenously determined technological progress. It is important to note that the
rate of savings affects only the level of GDP, not
the long-run rate of growth. A larger rate of savings
will cause the rate of growth to increase temporarily because greater capital accumulation increases
the productivity of labor and the level of GDP.
But in the long run, the rate of growth will settle
down to the rate of change in the labor force plus
the rate of technological progress.
Economic Review — Second Quarter 1993
Figure 1
The Solow Growth Model
sY
L
k ∆L
L
sY
L
k
k
The Solow model implies that if rates of
growth differ among countries, it is only because
the countries are at different stages of movement
toward the steady state. Rich countries should
grow at a slower pace than poor ones; accordingly,
over time, the per capita incomes of the rich and
poor countries should converge.
Endogenous growth models with innovation.
The Solow model suffers from its assumption that
technological progress is not explained by economic forces. However, while the Solow model is
silent on the mechanism of technological progress,
some recently developed endogenous growth
models have attempted to articulate the economic
process behind technological development. Joseph
Schumpeter (1950) and Jacob Schmookler (1966)
have argued forcefully that technological progress
takes place because innovators find it profitable to
discover new ways of doing things. Technological
progress does not just happen as a result of disinterested scientists operating outside the profit
sector. Schmookler reviewed the record of important inventions in petroleum refining, papermaking,
railroading, and farming and found “not a single,
unambiguous instance in which either discoveries
or inventions” were solely the result of pure intellectual inquiry (p. 199). Rather, the incentive
was to make a profit.
The implication is that productivity growth
might be related to the structure and policies
followed by the economy, rather than to the
exogenous forces of nature and luck. If growth is
29
endogenous, we would expect to find a wide
variation in the rates of growth of different nations,
with no apparent correlation with their levels of
per capita income.
Research and development are carried out to
make a profit on a new product. But every new
product adds to the stock of human knowledge, so
the cost of innovation falls as knowledge accumulates. To use an old metaphor, we stand on the
shoulders of those who precede us. Obviously,
the car required the prior invention of the wheel
and the gasoline engine. Panati (1987) gives some
interesting examples. The potato chip followed
french fried potatoes; detergents followed soap
(by 3,000 years); the hair dryer was suggested by
the vacuum cleaner; and athletic shoes required
vulcanized rubber. These examples suggest that
the rate of growth of the economy will vary directly
with the rate of introduction of new products:
think of the automobile, the airplane, the personal
computer, or the television set.
Some recently developed endogenous growth
models have tried to capture the process behind
the introduction of new products.3 In these models,
technological progress is faster, the larger is the
level of accumulated human knowledge. The
explanation is that the cost of innovation falls as
the level of human knowledge increases.4 As
opposed to the Solow model, there are no diminishing returns to capital when other factors are
held constant; so, raising the level of capital can
lead to ever-increasing growth rates. Therefore,
income growth will tend always to be faster among
countries that have a relatively large stock of
capital, a large educated population, or an economic environment that is favorable to the accumulation of human knowledge.
The convergence hypothesis
A stark prediction of the Solow model is that
countries with similar preferences and access to the
30
3
See, for example, Lucas (1988), Romer (1990), and
Grossman and Helpman (1991).
4
See the Appendix for a formal presentation of this model.
Figure 2
Real GDP Growth Per Capita
and 1960 Real GDP Per Capita
Average annual per capita GDP growth rate, 1960–85
(Percent)
8
7
●
6
●
●●
5
●
● ●
●●
●
●
●
●
●●●
●
●●
●●
●●
● ●
●
●
●
● ●
●
● ●●
●
●
●
●
●
●
● ●●
●
●
●●
●●
● ●
● ●
●
●
●
● ●
●
●●●
●
●
●
●
●●●
● ●
●
●
● ●●
● ●
4
3
2
1
0
–1
●
●
●● ●
●●
●
●
●
●
●
●
●
●
●
●
–2
●
–3
0
2
4
6
8
10
12
14
1960 GDP per capita (Thousands of dollars)
SOURCE OF PRIMARY DATA: Summers and Heston (1991).
same pool of technology should eventually reach
the same per capita income level. Consequently,
poor nations will tend to grow faster than richer
nations until their income levels catch up with, or
converge to, the income levels of rich countries.
In contrast to the Solow model, the endogenous growth model makes no such predictions.
The model allows for the possibility that countries
that start off richer and have more resources, such
as human or physical capital, may always be
ahead of less developed countries.
What is happening? Are the poor countries
catching up with the richer nations, or are the rich
getting relatively richer? Comparing their incomes
is difficult because nations use different currencies
and may have large variations in costs of living. If
one uses market exchange rates to convert official
GDP statistics into a common currency, the poorest
60 percent of the world’s nations received only
about 5 percent of the world’s income in 1988,
down from about 10 percent in 1960. It appears
the poor countries are losing out. But the cost of
living in poor countries is lower than in rich countries. To correct for this difference, it is necessary
to use a measure of purchasing power parity—
Federal Reserve Bank of Dallas
that is, the exchange rate that would make the
costs of living of countries comparable. Robert
Summers and Alan Heston (1991), therefore, recalculated the incomes of nations, using estimates
of purchasing power parities. On this basis, the
Solow model apparently has the correct predictions because income convergence appears to be
taking place. From 1960 to 1988, the share of the
poorest 60 percent of the world’s population rose
from about 17 percent of world income to almost
21 percent, while the share of the richest tier of
countries fell from 68 percent of world income to
about 60 percent.
This analysis may be misleading, however,
because changes in income shares are highly sensitive to how one defines income classes. If one
compares the richest 10 percent of countries with
the poorest 10 percent of countries, convergence
does not appear to be taking place. Generally, the
middle-income countries, which are sometimes
grouped with the very poor countries, are experiencing convergence with the rich nations.
Another way of determining the degree to
which incomes are converging across countries is
to observe the relationship between growth rates
and levels of income. If income levels of countries
tend to converge, poor countries should grow
faster than richer countries as they catch up to
reach the higher level of income. Figure 2 shows
the relationship between income in 1960 and
growth rates between 1960 and 1985 for 98 countries of the Summers–Heston data set. There does
not appear to be any strong negative relationship
between growth rates and the level of income,
which may indicate that convergence is not taking
place. If it was, the diagram should show a negative, or downward-sloping, relationship between
the level of income and growth rates, rather than
the relationship pictured.
Some have argued that a problem with Figure
2 is that it does not hold constant other factors
that determine growth. If we examine the relationship of income levels in 1960 and economic growth
between 1960 and 1985, holding constant human
capital, we find that the poor countries appear to
be catching up with the rich countries (Figure 3).
Although income convergence conditional on
human capital and other variables has been used
as evidence against endogenous growth theory
(Mankiw, Romer, and Weil 1992), it is not necesEconomic Review — Second Quarter 1993
Figure 3
Real GDP Growth Per Capita and 1960 Real GDP
Per Capita: Human Capital Held Constant
Average annual per capita GDP growth rate, 1960–85
(Percent)
5
4
●
●
3
●●
●
●●
●
●
●
● ● ●● ●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●
●
● ●●● ● ●
● ● ● ●
● ●●
●●
● ●●
●
●
●
●
● ● ●● ●
●
●
●
● ●●●
●
●
●
●
●●
●
●
●
●●
●
2
1
0
–1
–2
–3
–4
●
●
●● ●
●●
●
●
●
●
●
●
●
●
●
–5
●
–6
0
2
4
6
8
10
12
14
1960 GDP per capita (Thousands of dollars)
SOURCES OF PRIMARY DATA: Summers and Heston (1991).
United Nations (1971).
sarily inconsistent.5 We pointed out earlier that
endogenous growth theory suggests that countries
with higher levels of education (human capital)
might provide greater incentives for invention and,
therefore, much higher rates of growth. But holding
human capital constant, endogenous growth theory
may also predict convergence. Endogenous growth
theory merely says that countries may diverge if
they have different levels of human capital, all
other factors constant. There is evidence suggesting
divergence because countries do have different
levels of human capital, and human capital tends
to be positively correlated with economic growth.
The determinants of economic growth
The Solow and endogenous growth models
have different implications for what is, or is not,
5
Furthermore, the methodology of regressing average growth
rates against initial income levels does not necessarily
provide statistical evidence of convergence. For a description of this problem, see Danny Quah (1990).
31
Figure 4
Figure 5
Solow Model:
Increase in Income Due to Educational Subsidy
Endogenous Growth Model:
Increase in Growth Due to Educational Subsidy
Income per capita
Income per capita
Income with
educational
subsidy
Income with
educational
subsidy
Income
without
educational
subsidy
Income without
educational
subsidy
Transition
period
Subsidy
begins
Years
important in determining the rate of growth. Using
these models as guides, economists have tried to
estimate the role of various factors suspected of
determining the rate of economic growth.
Does the real world behave like the endogenous growth model, in which technological
progress and long-term growth are influenced by
economic factors, or like the Solow growth model,
in which the determinants of technological progress
and growth are exogenous? The question is important because the answer can tell us how countries
may influence their growth rates. It is a difficult
question to answer because technological progress
is a long-run phenomenon and may take centuries
to observe. Furthermore, over the relatively short
period for which data are available, factors that
apparently influence growth rates may, in reality, be
changing only income levels. Hence, an observed
increase in growth may be just a short-run transition to a higher income level and not a permanent
increase.
For example, suppose Indonesia decides to
subsidize college education by providing free tuition.
6
32
One-third is typically found to be capital’s share of output
across countries. This assumes a Cobb–Douglas production technology of the form Y = KαL1–α, where α is capital’s
share of output.
Subsidy
begins
Years
If the Solow model accurately reflects reality, Indonesia will experience faster growth in the transition to a higher level of income (because of more
investment in the accumulation of human knowledge), but income growth in Indonesia will not
permanently increase (Figure 4 ). If the endogenous
growth model is a better reflection of reality, the
greater accumulation of human knowledge will
result in not only higher income but also a permanently higher growth rate (Figure 5 ). The problem
is distinguishing between models in the short run.
Both models make the same short-run prediction
that free college tuition increases Indonesia’s growth.
It is only in the long run, when growth either
speeds up or does not change, that distinguishing
between these two models becomes possible.
Plosser (1992) points out that the Solow
growth model, even in the transition to a higher
income level, cannot satisfactorily describe the
changes in growth rates across countries. Imagine,
for example, that Indonesia increases its rate of
investment by 50 percent. As discussed above, the
model predicts that the growth rate would immediately increase but would gradually decline over
time until the new higher income and level of
capital were reached. Assuming that the share of
total capital in output is one-third, the Solow model
predicts that income per capita would only rise
about 22 percent.6 If the country completed the
transition to the new higher level of income in
thirty years, then the increase in the average annual
Federal Reserve Bank of Dallas
growth rate would be about 0.7 percent per year.7
Consequently, large increases in investment rates
have little ability in the standard Solow growth
model to explain the observed large differences in
growth rates across countries.
Studies have stressed different reasons why
economic growth varies across countries. Because
of the current popularity of endogenous growth
models, most recent studies have focused on the
role of human capital accumulation. However,
human capital as an input to production is also
important in the Solow model. In addition to human
capital, a country’s economic environment can
play an important role in influencing economic
growth. For example, internal competitive structure, a country’s openness to trade, its political
stability, and the efficiency of its government can
influence innovative activity and economic
growth, as discussed below.
Human knowledge. Does educating a work force
increase a country’s growth? Barro (1991), Mankiw,
Romer, and Weil (1992), Levine and Renelt (1992),
and Gould and Ruffin (1993), among others, have
found evidence suggesting an educated populace
Figure 6
Partial Association Between Real GDP Growth
Per Capita and School Enrollment Rate
is a key to economic growth. A larger educated
work force may increase growth either because of
faster technological progress, as individuals build
on the ideas of others, or by simply adding to the
productive capacity of a country. For example, in
1960, only 7 percent of Guatemala’s children of
secondary school age actually attended secondary
school. Barro (1991) estimates that had the Guatemalans invested in education to increase attendance to a relatively modest 50 percent in 1960,
the country’s growth rate per capita from 1960 to
1985 might have increased an amazing 1.3 percent
per year.
Figure 6 depicts the empirical relationship
between GDP growth rates and secondary school
enrollment as a proportion of the working-age
population for 98 countries between 1960 and
1985.8 On the vertical axis are average annual per
capita growth rates for 1960–85, and on the horizontal axis is the log of secondary school enrollment rates as a proportion of the working-age
population in 1960. Held constant are income levels
in 1960, as well as capital savings rates. The slope
of the fitted line in Figure 6 implies that increasing
the secondary school enrollment rate a modest 2
percent, from 8 percent to 10 percent, raises the
average growth rate an estimated 0.5 percent per
year, holding other factors constant. The figure
suggests that one important way for poor coun-
Average annual per capita GDP growth rate due to schooling, 1960–85
(Percent)
6
●
7
Plosser (1992) notes that there is considerable controversy
over how fast an economy moves to its new level of income.
Barro and Sala-i-Martin (1992) and Mankiw, Romer, and
Weil (1992) estimate that it takes between 25 years and 110
years for one-half of the transition to be completed, depending on the sample and other characteristics considered.
King, Plosser, and Rebelo (1988) compute the half-life of the
transition as ranging from 5 years to 10 years under their
parameter assumptions.
8
School enrollment rates are often used as a proxy for
accumulated knowledge because of the lack of data on the
size of the educated population. A problem with using
school enrollment rates, however, is that they measure the
increase in the size of the educated population rather than
the actual stock of educated people. Using literacy rates
across countries may be more attractive because they are
a measure of the stock of educated people. They, too,
present problems, however, as literacy rates are sometimes measured differently across countries.
●
5
4
●
●
●
2
●
1
●
0
●
●
●
●
●
●
●
●
●
●
●
●
●
●
–1
●
●
–2
●
●
●
●
●
●●
●●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
–3
–1
●
●
●
● ●●
●
●
●● ●
● ●
●
● ● ●●
●●
● ● ●
●
●
●
●
●
●
●● ● ●
● ●● ●● ●
●
●● ●
● ● ●
●
3
–.5
0
.5
1
1.5
2
Log of secondary school enrollment
as percent of working-age population, 1960
SOURCE OF PRIMARY DATA: Mankiw, Romer, and Weil (1992).
Economic Review — Second Quarter 1993
2.5
33
Figure 7
Partial Association Between Real GDP Growth
Per Capita and Government Consumption in GDP
Average annual per capita GDP growth rate, 1960–85
(Percent)
5.0
2.5
●●
●
●
●
●●
● ●●
0
●
●
●● ●
●
●
●●
●
●
●
●
●
●●
●
●
● ●●
●
●
● ●
●
●
●
●
● ●●
●●
● ● ●
● ●● ●
●
●
●●
●●
●
●
● ●●●
● ●
● ●
●
●
●
●
● ●
●
● ●● ●● ●
●
●
●
–2.5
●
●
●
●
●
●
●
●
●
●
–5.0
0
5
10
15
20
25
Percent of government consumption in GDP, 1960
SOURCE: Barro (1991, 431).
tries to advance is through greater investment in
education.
Political and governmental factors. How much
does the political environment help or hinder economic growth? One can imagine that an extremely
unstable government (one that is susceptible to
rapid policy reversals) would create insecurity
about the future and decrease the incentives to
invest in future development. For example, oil
exploration would disappear if private investors
believe that the government will expropriate oil
wells. Likewise, people will not invest in developing new products if they cannot reap the rewards
of their ideas. This is why copyright and patent
protection can be an important factor in economic
growth. Although it is difficult to measure property rights in a country, a safe assumption is that
9
34
As measured by assassinations and revolutions, political
instability can also be associated with the direct destruction
of a country’s capital stock. This, by itself, can certainly
reduce a country’s growth rate.
they suffer when a country experiences a large
number of revolutions and political assassinations.
Holding constant the levels of education,
income, and government consumption, Barro (1991)
estimates that political instability, as measured by
the number of revolutions and political assassinations in a country, decreases GDP growth per
capita. For example, with greater political stability,
South Korea’s growth rate would have been 6.25
percent per year, rather than 5.25 percent, from
1960 to 1985. More dramatically, El Salvador may
have lost almost 7 percent per year in per capita
growth because of its extreme political instability.9
If political instability has a negative influence
on growth, how does the size of government affect
economic growth? Barro (1989, 1990, 1991) finds
that the larger the share of government spending
(excluding defense and education) in total GDP,
the lower are growth and investment. Barro also
finds that government investment has no statistically significant effect on economic growth. A
government may attempt to increase private productivity through government spending, but the
evidence suggests it has no such effect and may
even decrease growth. Growth appears to fall with
higher government spending because of lower
private savings and because of the distortionary
effects from taxation and government expenditure
programs. Figure 7 shows the negative relationship between per capita growth and the share of
government consumption in GDP. Held constant
are income levels in 1960, as well as the level of
education and indicators of the political stability.
As Figure 7 shows, increasing the share of government consumption in GDP from 10 percent to 15
percent would decrease economic growth about
0.6 percent per year.
International trade. Do countries that are open
to international trade grow faster than closed
economies? Evidence suggests that the answer is
yes. From 1960 to 1985, economies that have pursued outward-oriented pro-trade policies—such as
the four so-called Asian Tigers (Singapore, Hong
Kong, South Korea, and Taiwan)—experienced
growth rates between 8 percent and 10 percent a
year. In contrast, the relatively closed economies
of Africa and Latin America experienced growth
rates rarely exceeding 5 percent a year. Ben-David
(1991) finds that when countries in Europe joined
the European Community and dropped their trade
Federal Reserve Bank of Dallas
barriers, incomes increased and approached those
of the wealthier nations.
A country open to international trade may
experience faster technological progress and
increased economic growth because the cost of
developing new technology falls as more hightech goods are available. In other words, trade
increases growth because it makes a greater variety
of products and technologies available. De Long
and Summers (1991) find that relatively closed
countries with high effective rates of protection
have productivity growth rates that, on average,
are 1.1 percentage points below those of other
countries. Roubini and Sala-i-Martin (1991) find
that a country that moves from being a strongly
outward-oriented trade regime to a strongly inwardoriented trade regime would experience a 2.5percentage-point decrease in its annual growth
rate. Gould and Ruffin (1993) attempt to distinguish
between human capital as an input to production
and human capital as the source of long-term
growth in open and closed trading regimes. They
find that when human capital, as measured by
literacy rates, is relatively high, open economies
experience growth rates 1 to 2 percentage points
higher than the growth rates of closed economies.
Equipment investment. De Long and Summers
(1991) have argued that equipment investment
has potentially large effects on economic growth.
They explain that new technologies have tended
to be embodied in new types of machines. For
example, at the end of the eighteenth century, steam
engines were necessary for steam power, and
automatic textile manufacture required power looms
and spinning machines. In the early twentieth
century, assembly-line production was unthinkable
without heavy investments in the new generations
of high-precision metal-shaping machines that
made parts interchangeable and assembly lines
possible.
In examining a cross-sectional distribution of
growth rates in the post–World War II period, De
Long and Summers find evidence suggesting that
investments in machinery and equipment are a
strategic factor in growth and possibly carry large
positive benefits in generating further technological
progress. Holding constant such factors as relative
labor productivity, labor force growth, school
enrollment rates, and investment other than in
machinery and equipment, De Long and Summers
Economic Review — Second Quarter 1993
find that each extra percentage point of total output
devoted to investment in machinery and equipment is associated with an increase of 0.26 percentage point per year in economic growth. Other
investment also has a positive impact on growth,
but the effect is only one-fourth as large as that
for machinery investment.10
Overview of factors behind growth. Table 4
summarizes factors that have been shown to
influence growth rates. As the table indicates,
factors that are associated with increasing human
or physical capital investment tend to enhance
technological progress and economic growth. On
the other hand, factors that reduce incentives to
invest, or interfere with well-functioning markets,
tend to reduce growth.
Conclusion
For centuries, economists have been trying
to answer questions about what determines economic growth and to make predictions about the
future. Malthus, an economist who wrote in the
late eighteenth century, predicted that expanding
population growth combined with limited resources
and declining productivity would result in only a
subsistence income. Certainly, in the slowly growing agrarian era in which Malthus lived, it would
have seemed impossible for the land to provide
for everyone with unbounded plenitude. However,
with the technological advances in the latest century, it is difficult to be pessimistic. New products
appear to beget other products, so technology
seems to be advancing at ever-increasing rates.
Endogenous growth literature arose out of
the desire to explain why, over long periods, economic growth appears to be accelerating and why
some countries grow faster than others. The traditional Solow model left unanswered too many
questions about growth differentials across countries and the mechanism of technological progress.
10
Whether equipment investment generates positive externalities that influence technological progress is subject to
some debate. Auerbach, Hassett, and Oliner (1992) argue
that economic growth due to equipment investment is
completely consistent with the basic Solow model.
35
Table 4
Determinants of Economic Growth
Growth enhancing
Growth reducing
Schooling, education investment 1, 2
Government consumption spending 1
Capital savings, investment 2
Political, social instability 1
Equipment investment 3
Trade barriers 3,4,5,6
Level of human capital 1,6
Socialism1
1
Barro (1991).
Mankiw, Romer, and Weil (1992).
De Long and Summers (1991).
4
Ben-David (1991).
5
Roubini and Sala-i-Martin (1991).
6
Gould and Ruffin (1993).
2
3
Technological progress, however, is what ultimately determines growth, and growth determines
whether our grandchildren will have better lives
than ours.
We are just beginning to understand theoretically and empirically the mechanisms of economic
growth, and much work has yet to be done. But
so far, there appears to be a strong relationship
between investment, particularly human capital
investment, and growth. Other factors also are
positively related to investment and growth, such
36
as political stability, well-defined property rights,
equipment investment, low trade barriers, and low
government consumption expenditures. These
findings are consistent with the long-run growth
predictions of endogenous growth models but
are also consistent, in the short run, with Solow
models. It may be several decades before we have
enough detailed long-run data to distinguish clearly
between these theories. In the meantime, maintaining policies consistent with long-run growth
can have significant benefits.
Federal Reserve Bank of Dallas
Appendix
Endogenous Growth Model with Innovation
This Appendix explains the basic endogenous growth model found in Grossman and
Helpman (1991). Suppose there are n products. To simplify, each product sells for the
same price and has the same cost of production. Each product is the property of a single
firm. We assume that each unit requires only
one unit of labor so that the marginal cost of
production is simply the wage rate, w. Every
firm sets a price, p, that is the same markup
over costs,
(A.1)
p = w (1/α),
where 1/α is the markup. The parameter α is
between 0 and 1. In effect, α is the cost of
production per dollar’s worth of the product.
Accordingly, the profit on $1 worth of sales will
be (1 – α). If the economy sells $E worth of
products, then the total profit of all n firms is
(A.2)
Π = E (1 – α).
Research and development take place
in the form of new products. Firms invent new
products in an effort to capture some fraction
of the profits given in (A.2). It takes a units of
labor to invent a new product. Accordingly,
firms will enter the market as long as the
present value of future profits, called v, exceeds innovation costs, wa (that is, v wa).
Economic Review — Second Quarter 1993
Firms will enter the market until w rises or v
falls up to the point that
(A.3)
wa = v.
The rate of growth of new products is
g = ∆n /n. Each firm’s profit is E (1 – α)/n;
accordingly, the profit of any existing firm falls
as new firms develop new products. The
stock market valuation of any set of n firms
must be reduced by the growth rate, g. If g is
zero, the aggregate value of all n firms would
be vn = E (1 – α)/r, where r is the rate of
interest. But if g is greater than zero, then the
aggregate value of the stock market is
(A.4)
vn = E (1 – α)/(r + g).
Combining (A.2), (A.3), and (A.4), we
have
(A.5) p = v /aα = E (1 – α)/aαn (r + g).
To determine the rate of growth of new
products, we need to know how much labor is
devoted to their production and how much is
devoted to research and development. The
amount of labor devoted to production is the
total output of products (because one unit of
(Continued on the next page)
37
Appendix
Endogenous Growth Model with Innovation—Continued
output requires one unit of labor). In turn, the
total output of products is the physical sales
volume, E /p. The amount of labor devoted to
research and development is the number of
new products, ng, multiplied by the labor
required per new product, a, or ang. Thus, the
growth rate, g, is determined by the equation
ang + E /p = L,
(A.6)
where L is the total amount of labor available.
Substituting (A.5) into (A.6) yields
(A.7)
ang + aαn (r + g)/(1 – α) = L.
Multiplying by (1 – α), we get
(A.8)
g (1 – α) + α (r + g) = L(1 – α)/an,
The key assumption in the theory of
endogenous growth is that there are technological spillovers. A simple way of capturing
this idea is to let the cost of invention fall as
human knowledge accumulates. In this model,
human knowledge can be regarded as the
number of products, n. Accordingly, it is assumed that a = c /n, where c is a positive
constant. Substituting this into (A.9) yields the
final growth equation:
(A.10)
g = L[(1 – α)/c ] – αr.
An important implication is that the larger the
stock of people capable of carrying out research and development, L, the larger the
rate of growth. Because human capital is
growing, the endogenous growth model implies accelerating growth rates.
which equals
(A.9)
38
g = L(1 – α)/an – αr.
Federal Reserve Bank of Dallas
References
Auerbach, Alan J., Kevin A. Hassett, and Stephen
D. Oliner (1992), “Reassessing the Social
Returns to Equipment Investment,” Working
Paper Series, no. 129 (Board of Governors of
the Federal Reserve System, Division of Research and Statistics, Economic Activity Section,
December).
Barro, Robert J. (1991), “Economic Growth in a
Cross Section of Countries,” Quarterly Journal
of Economics 106 (May): 407–43.
——— (1990), “Government Spending in a Simple
Model of Endogenous Growth,” Journal of
Political Economy 98 (October, pt. 2): S103–25.
——— (1989), “A Cross-Country Study of Growth,
Saving, and Government,” NBER Working Paper
Series, no. 2855 (Cambridge, Mass.: National
Bureau of Economic Research, February).
———, and Xavier Sala-i-Martin (1992), “Convergence,” Journal of Political Economy 100
(April): 223–51.
Ben-David, Dan (1991), “Equalizing Exchange: A
Study on the Link Between Trade Liberalization and Income Convergence” (University of
Houston, October, Photocopy).
———, and David H. Papell (1993), “The Great Wars,
the Great Crash, and the Unit Root Hypothesis”
(University of Houston, January, Photocopy).
De Long, J. Bradford, and Lawrence H. Summers
(1991), “Equipment Investment and Economic
Growth,” Quarterly Journal of Economics 106
(May): 445–502.
Economic Report of the President (1992) (Washington, D.C.: U.S. Government Printing Office,
February).
Gordon, Robert J. (1992), “Measuring the Aggregate
Price Level: Implications for Economic Performance and Policy,” NBER Working Paper
Series, no. 3969 (Cambridge, Mass.: National
Bureau of Economic Research, January).
Economic Review — Second Quarter 1993
Gould, David, and Roy J. Ruffin (1993), “Human
Capital Externalities, Trade, and Economic
Growth,” Federal Reserve Bank of Dallas
Research Paper no. 9301 (Dallas, January).
Grossman, Gene M., and Elhanan Helpman (1991),
Innovation and Growth in the Global Economy
(Cambridge: MIT Press).
King, Robert G., Charles I. Plosser, and Sergio T.
Rebelo (1988), “Production, Growth and
Business Cycles: I. The Basic Neoclassical
Model,” Journal of Monetary Economics 21
(March/May): 195–232.
Levine, Ross, and David Renelt (1992), “A Sensitivity
Analysis of Cross-Country Growth Regressions,”
American Economic Review 82 (September):
942–63.
Lucas, Robert E., Jr. (1988), “On the Mechanics of
Economic Development,” Journal of Monetary
Economics 22 ( July): 3–42.
——— (1987), Models of Business Cycles (Oxford
and New York: Basil Blackwell).
Maddison, Angus (1991), Dynamic Forces in Capitalist Development: A Long-Run Comparative
View (Oxford and New York: Oxford University Press).
Mankiw, N. Gregory, David Romer, and David N.
Weil (1992), “A Contribution to the Empirics
of Economic Growth,” Quarterly Journal of
Economics 107 (May): 407–37.
Panati, Charles (1987), Panati’s Extraordinary
Origins of Everyday Things (New York: Harper
and Row, Perennial Library).
Plosser, Charles I. (1992), “The Search for Growth”
(University of Rochester, August, Photocopy).
Quah, Danny (1990), “Galton’s Fallacy and Tests
of the Convergence Hypothesis” (Massachusetts Institute of Technology, May 10, Photocopy).
39
Ramey, Garey, and Valerie A. Ramey (1991),
“Technology Commitment and the Cost of
Economic Fluctuations,” NBER Working Paper
Series, no. 3755 (Cambridge, Mass.: National
Bureau of Economic Research, June).
Romer, Paul M. (1990), “Endogenous Technological Change,” Journal of Political Economy 98
(October, pt. 2): S71–S102.
——— (1986), “Increasing Returns and Long-Run
Growth,” Journal of Political Economy 94
(October): 1002–37.
Roubini, Nouriel, and Xavier Sala-i-Martin (1991),
“Financial Development, the Trade Regime,
and Economic Growth,” NBER Working Paper
Series, no. 3876 (Cambridge, Mass.: National
Bureau of Economic Research, October).
Schmookler, Jacob (1966), Invention and Economic
Growth (Cambridge: Harvard University Press).
40
Schumpeter, Joseph A. (1950), Capitalism, Socialism, and Democracy, 3d ed. (New York:
Harper and Brothers).
Solow, Robert M. (1956), “A Contribution to the
Theory of Economic Growth,” Quarterly
Journal of Economics 70 (February): 65–94.
Summers, Robert, and Alan Heston (1991), “The
Penn World Table (Mark 5): An Expanded Set
of International Comparisons, 1950–1988,”
Quarterly Journal of Economics 106 (May):
327–68.
United Nations, Department of Economic and
Social Affairs (1971), World Economic Survey,
1969 –1970, E/4942 ST/ECA/141.
Wynne, Mark A. (1992), “The Comparative Growth
Performance of the U.S. Economy in the Postwar Period,” Federal Reserve Bank of Dallas
Economic Review, First Quarter, 1–16.
Federal Reserve Bank of Dallas
Stephen P. A. Brown
Mine K. Yücel
Assistant Vice President and Senior Economist
Federal Reserve Bank of Dallas
Senior Economist and Policy Advisor
Federal Reserve Bank of Dallas
The Pricing of Natural Gas in U.S. Markets
R
ecent patterns in natural gas prices have raised
public concern about the pricing of natural
gas in U.S. markets ( Johnson 1992). In recent
years, prices paid by industrial and electrical end
users have fallen more than wellhead prices,
while residential and commercial prices have
fallen less than wellhead prices (Yücel 1991).1
The lack of uniform changes in natural gas
prices may arise from differences in end users and
the market institutions that serve them. Industrial
and electrical users of natural gas can switch
easily between oil products and natural gas, while
residential and commercial users cannot. Industrial
and electrical users generally bypass local distribution companies (LDCs), relying heavily on spot
supplies in a competitive market served by brokers
and pipeline companies. In contrast, residential
and commercial users typically purchase their gas
from LDCs, which earn a regulated rate of return
and obtain their supplies under long-term contracts.
These observations about U.S. natural gas
markets lead us to ask two questions. Do differing
characteristics in end users and the market institutions serving them lead to differences in pricing
behavior? Are changes in natural gas prices uneven
in the long run, or is popular concern about natural
gas prices unwarranted? To answer these questions,
we examined econometrically how price shocks are
transmitted across various markets for natural gas.
characteristics of the final customer—that is, as
residential, commercial, industrial, and electrical
prices for natural gas.
Natural gas is first sold at the wellhead,
where it is produced. Both pipeline companies
and brokers use the collection and pipeline systems
to transport gas from the field to their customers,
with brokers using the pipelines as contract carriers.
Pipeline companies and brokers sell their natural
gas directly to some end users and to LDCs, which
pay the city gate price. In turn, the LDCs distribute
gas throughout localities and sell it to additional
end users.
When comparing end-use markets for natural
gas, several differences stand out. Industrial and
electrical users of natural gas generally can switch
easily between fuels to seek the lowest cost energy
source. As a consequence, most of these end users
rely heavily on spot supplies purchased directly
from pipeline companies and brokers. For the
most part, these suppliers seem to behave competitively in serving this market (Brown and Yücel
1993). In contrast, most residential and commercial consumers are tied to a single fuel. These end
users purchase their natural gas from LDCs, which
Natural Gas Markets and Prices
As natural gas journeys downstream from the
wellhead, it travels through collection systems,
pipelines, and local distribution systems before it
reaches its consumers. Natural gas prices are
observed in six separate markets. Prices in these
markets include wellhead, city gate, and four enduse prices. End-use prices are identified by the
Economic Review — Second Quarter 1993
The authors thank Nathan Balke, Phil Drake, Tom Fomby,
Bill Gruben, Shengyi Guo, Joe Haslag, Tim Smith, Lori
Taylor, and Mark Wynne for helpful comments, but do not
implicate them in the conclusions. Shengyi Guo provided
able econometric and programming assistance. The views
expressed are those of the authors and should not be
attributed to the Federal Reserve Bank of Dallas or to the
Federal Reserve System.
1
Electrical utilities use natural gas to generate electricity.
41
earn a regulated rate of return and obtain much
of their gas under long-term contract. (For a discussion of why electrical and industrial end users
rely more heavily on spot markets for natural gas
than commercial and residential customers, see
the box titled “Development of a Spot Market for
Natural Gas.”)
To some extent, differing reliance on spot
and contract supplies may account for the longterm difference in the way price shocks are transmitted through the market for natural gas. When
average wellhead prices change, spot prices
generally change more than those specified in
long-term contracts. Thus, when average wellhead
prices fall, as has been the trend since 1985, spot
prices generally fall more than those specified in
long-term contracts.
The extent to which the supply in a market
comprises spot gas determines how responsive its
prices are to changes in the average wellhead
price. With a greater than average reliance on
spot supplies, electrical and industrial customers
stand to see a change in their gas prices that is
greater than the market average. With a less than
average reliance on spot supplies, commercial and
residential customers stand to see a change in
their gas prices that is less than the market average.
42
2
During the estimation period, there were extensive changes
in federal regulation of the natural gas pipeline industry.
These changes raise concern about estimating stable relationships between the wellhead price and the electrical,
industrial, and city gate prices. Because we do find
cointegrating relationships in all cases, we treat the regulatory changes as primarily endogenous or irrelevant to the
transmission of price shocks.
Monthly data for average wellhead, city gate, electrical,
industrial, commercial, and residential prices were obtained
from the U.S. Department of Energy. The price series were
deflated with a monthly GNP deflator series and then
seasonally adjusted with the X–11 procedure in SAS.
The monthly GNP deflator was obtained by using the
Chow–Linn procedure on quarterly data. The consumer
price and producer price indexes were used as monthly
reference series in the Chow–Linn procedure.
3
See Balke (1991), Sims, Stock, and Watson (1990), and
Stock and Watson (1988).
Empirical Analysis of Natural Gas Pricing
The institutional arrangements in natural gas
markets suggest that seven pairs of natural gas prices
have an upstream–downstream relationship. These
are the wellhead price with electrical, industrial,
city gate, commercial, and residential prices, and
the city gate price with commercial and residential
prices. For each pair of upstream and downstream
prices, we conceptualize the long-run relationship
as a simple markup model of natural gas prices in
which price shocks are transmitted:
(1)
PDt = a + βPU t ,
where PD is a downstream price for natural gas,
and PU is an upstream price for natural gas.
To examine how changes in price are transmitted across the markets for natural gas, we
utilize time-series methods. In the absence of a
specific theory to be tested, we use the statistical
tests, together with identifying assumptions, to
assess in which markets shocks to natural gas
prices originate and how they are transmitted
across natural gas markets.
Our econometric work involves a number of
steps. We check whether the price series are
stationary and find that all of them have stochastic
trends (or are integrated). For each of the seven
pairs of prices, we then test for cointegration and
use a series of reduced-form vector-error-correction
models to test for causality and adjustment to
equilibrium error. We then identify the sources of
long-run price shocks and calculate their persistence. Estimation and testing uses monthly data
from January 1984 through March 1992.2
Integration. As an initial step in our econometric
work, we check whether our price series are integrated or stationary. A time series that is integrated is said to have a stochastic trend (or unit
root). Identifying a series as an integrated, nonstationary series means that any shock to the
series will have permanent effects on it. Unlike a
stationary series, which reverts to its mean after a
shock, an integrated time series does not revert to
its preshock level.
Applying conventional econometric techniques
to an integrated time series can give rise to misleading results.3 Therefore, we use both augmented
Dickey–Fuller and Phillips–Perron tests to test for
Federal Reserve Bank of Dallas
Development of a Spot Market for Natural Gas
As consumers of large quantities of energy, industry and electrical utilities have found
it attractive to invest in the ability to switch
between residual fuel oil and natural gas. This
ability may have contributed to the development of spot markets for natural gas. Without
the ability to switch fuels, an electrical or
industrial user would find it undesirable to rely
on spot supplies because the pipeline company or LDC providing connections to the
natural gas transportation and distribution
system could appropriate the end user’s capital investment.
Energy consumption involves relatively
high capital costs. After an end user makes a
capital investment that is specific to natural
gas consumption, the pipeline company or
LDC providing the connection could exploit
monopoly power over the end user’s capital
investment. Government regulation and longterm supply contracts negotiated before the
investment is made are two ways to protect a
capital investment that has a specific use.
Competition among suppliers also protects
the energy user’s capital investment and allows the user to rely on spot supplies (Ellig
and High 1992).
A spot market for natural gas may not
provide enough competition to protect the
capital investment of its end users.1 End users
stochastic trends. We find that all price series are
integrated of order one—that is, the first differences of all series are stationary.4
Cointegration. After determining that each price
series is integrated of order one, we test each of
the seven pairs of natural gas prices described
above for cointegration. Two integrated time series
are cointegrated if they move together in the long
run. Cointegration implies a stationary long-run
relationship between the two series. As such, the
cointegrating term provides information about the
long-run relationship.
Economic Review — Second Quarter 1993
must still rely on a specific LDC or pipeline
company for a hookup to the natural gas
transportation and distribution system. The
ability to switch fuels provides end users with
competitive energy supplies, allowing them to
rely on spot supplies of natural gas.
In contrast, most commercial and residential customers find it too expensive to
invest in the ability to switch fuels, given their
relatively small consumption of energy. Their
low levels of consumption combined with their
inability to switch fuels reduces the attractiveness of relying heavily on spot supplies of
natural gas. In that sense, LDCs protect their
customers from potential upstream monopolies by obtaining a greater share of their
natural gas supplies under long-term contract. In turn, state governments regulate
LDCs, giving them a regulated rate of return
and preventing them from exercising monopoly power over their customers. Combined with the regulators’ concern for security
of supply, however, this regulated rate of
return may induce LDCs to overcommit to
long-term contracts (Lyon 1990).
1
The spot market for gas creates a competitive demand for the
transportation and distribution of natural gas, protecting the
capital investments of the pipeline and LDCs involved in contract carriage.
If cointegration is not accounted for, any
model involving the two cointegrated variables
could be misspecified, and/or the parameter estimates could be inefficiently estimated.5 Therefore,
we employ the Johansen procedure to estimate
4
In other words, shocks to first differences of all series are not
permanent.
5
See Engle and Yoo (1987).
43
Table 1
Cointegration of Upstream and Downstream Prices
Price pairs
Upstream
Downstream
Significance
(β 1)
Wellhead
Electrical
1.401
.002
Wellhead
Industrial
1.453
.000
Wellhead
City Gate
1.186
.000
Wellhead
Commercial
1.294
.003
Wellhead
Residential
1.148
.120
City Gate
Commercial
1.079
.282
City Gate
Residential
.938
.447
the cointegrating relationship between pairs of
upstream and downstream prices.6 We find a linear
cointegrating relationship (of the form PD = βPU )
between all seven pairs of prices considered.
Because each estimated cointegrating relationship is stationary, the cointegrating terms provide
an efficient estimate of the long-run relationships
between upstream and downstream prices. If a
one-unit change in the upstream price occurs over
the long run, it will be met by a β change in the
downstream price over the long run. Conversely,
if a one-unit change occurs in the downstream
price over the long run, it will be met by a 1/β
change in the upstream price.
As Table 1 shows, the estimated β s are unequal, signifying that a one-unit change in an upstream price affects some downstream prices more
44
Cointegrating
relationship
6
The Johansen procedure is a maximum likelihood method.
We chose it over several other procedures because it
provides the most efficient estimates of the cointegrating
relationships. In addition, it provides estimates of the number of cointegrating relationships.
7
Let PE = βWE PW , PR = β WR PW , and PR = β ER PE. Then PR = ( βWR /
β WE )PE , and β ER = β WR / β WE. It follows that a value of one for
β ER implies β WE = β WR . A nonunitary value for β ER implies
β WE β WR .
than others. It is not possible, however, to directly
test whether the differences between the β s are
statistically significant. To assess whether the estimated β s differ significantly from each other, we
use an indirect test for which we estimate cointegrating relationships between all possible price pairs.
With these additional estimates, we judge which
β s are or are not equal to each other. For example,
to assess whether the β in the wellhead–electrical
price pair (βWE ) is different from the β in the
wellhead–residential price pair ( βWR ), we estimate
a cointegrating relationship between the electrical–
residential price pair (βER ) and check to see
whether it has a value equal to one. A value of
one for βER would imply βWE = βWR . A nonunitary
value would imply βWE βWR .7
We find the β s between the wellhead price
and the electrical and industrial prices are the same
as each other but greater than the β between the
wellhead price and the city gate price. We also
find that the β s between the wellhead price and
the commercial and residential prices are equal to
each other but less than the β s between the wellhead price and the electrical and industrial prices.
At best, the estimated β s only partially support public concerns about uneven changes in
natural gas prices. Differences in the estimated β s
do show that uneven changes in natural gas prices
are maintained in the long run. Specifically, a perFederal Reserve Bank of Dallas
manent change in the wellhead price is accompanied by more extreme changes in the electrical
and industrial prices for natural gas than in the
commercial and residential prices for natural gas.
Nonetheless, our estimates suggest that the recent
pattern, in which residential and commercial prices
for natural gas have fallen less than the wellhead
price, is unlikely to persist. The β s between the
wellhead price and the commercial and residential
prices are estimated at greater than or equal to
one, indicating that residential and commercial
prices for natural gas change by at least as much
as the average wellhead price in the long run.
For the three pairs of prices involving wellhead price with electrical, industrial, and city gate
prices, the β s are greater than one. These β s
mean that the markups over the wellhead price
taken by the pipeline companies on electrical,
industrial, and city gate prices increase as natural
gas prices rise and decrease as natural gas prices
fall. These β s can be consistent with either normal
responses to shocks in demand, or shocks to
supply coupled with increasing returns to scale.
Because price shocks can arise from either
shocks to supply or demand, further interpretation
of the β s requires additional information. For
natural gas, demand shocks can originate in factors
such as changing oil prices, economic activity,
weather, technology, and government regulation of
energy consumption. Supply shocks can originate
in factors such as changing production technology,
geophysical knowledge, and government policy.
Using this information for the period of analysis,
we view changes in demand to be a more important source of initial shocks to natural gas markets
than changes in supply (Brown and Yücel 1993).
Given our view, demand shocks account for longrun movements in natural gas prices (see “LongRun Sources of Variance,” below).
As such, the estimated β s between the wellhead price and the electrical, industrial, and city
gate prices are consistent with a normal response
to shocks in end-use demand. As end-use demand
is increased, the pipeline companies experience
rising costs and/or are able to increase profits. As
end-use demand is decreased, the pipeline companies experience falling costs and/or are forced
to reduce profits.
The likelihood of variable profitability is
consistent with the fact that pipeline companies
Economic Review — Second Quarter 1993
purchase more gas under long-term contract than
they sell under long-term contract. Given the existing contracts, prices at which pipelines sell natural
gas would vary more in the face of fluctuating
demand than the prices they pay for natural gas.
Data for the estimation period generally show
falling natural gas prices and declining profitability
for pipeline companies.8
The differences we see in the β s most likely
reflect how spot and contract prices respond to
changing market conditions. Spot prices adjust
more readily, and industrial users rely more heavily
on spot supplies than do LDCs. Differences in the
β s also may reflect a lack of incentive for LDCs to
pursue the cheapest sources of gas as prices are
falling because their rate of return is regulated,
and their customers cannot easily switch fuels.
The β s between the city gate price and the
commercial and residential prices for natural gas
are not significantly different from unity. These
estimates probably reflect a regulated rate of
return for LDCs and a direct pass-through of gas
price changes.
Causality. A causal relationship between two
variables implies that changes in one variable lead
to changes in the other. To test for a predictive
relationship between the variables, we perform
Granger causality tests on each of the seven pairs
of upstream and downstream prices.
Because all our price series are cointegrated,
we account for cointegration by specifying an
error-correction model in which changes in the
dependent variable are expressed as changes in
both the independent variable and dependent
variable, plus an error-correction term. For cointegrated variables, the error-correction term is the
deviations from the long-run cointegrating relationship between the variables. The coefficient on
the equilibrium error reflects the extent to which
8
An alternative explanation for declining pipeline profitability
is structural change. Since 1985, the Federal Energy Regulatory Commission has changed the role of pipelines toward
one of open access and contract carriage. These changes
helped foster the development of a spot market for natural
gas served by brokers and the pipeline companies. This
explanation is inconsistent with finding cointegration unless
one can view the structural change as endogenous to
market pressure brought to bear by falling demand.
45
Table 2
Causality and Adjustment to Equilibrium Error
(Significance)
Price Pairs
Upstream
Downstream
Causality
PU to PD
PD to PU
Adjusts to error*
PU
PD
Wellhead
Electrical
.025
.000
.014
.542
Wellhead
Industrial
.007
.000
.001
.285
Wellhead
City Gate
.005
.000
.000
.783
Wellhead
Commercial
.000
.166
.082
.001
Wellhead
Residential
.002
.320
.155
.002
City Gate
Commercial
.000
.758
.457
.006
City Gate
Residential
.005
.600
.570
.011
* The significance of errors in the cointegrating relationship in the respective upstream and downstream price equations indicates
adjustment to the equilibrium error.
the dependent variable adjusts during a given
period to deviations from the cointegrating relationship that occurred in the previous period.9
The tests involve estimating a reduced-form
vector-error-correction model comprising the
following set of equations for each pair of prices:
n
n
i =1
j =1
(2) ∆PU t = ∑ ai ∆PDt −i + ∑ b j ∆PU t − j + α1CI t −n −1 + µ1t ,
n
n
i =1
j −1
(3) ∆PDt = ∑ ci ∆PU t −i + ∑ d j ∆PDt − j + α 2CI t −n −1 + µ2t ,
where PU is the upstream price, PD is the downstream price for natural gas, CI is the errors in the
9
46
See Engle and Granger (1987).
10
The error-correction term is included because all pairs of
upstream and downstream prices are cointegrated with
each other.
11
The value of n was set in the Johansen procedure to assure
white noise in the residuals.
cointegrating relationship (PDt –β PUt ),10 ai , bj , ci ,
dj , α1, α2 are parameters to be estimated, and µ1t
and µ2t are white noise residuals.11 The coefficients α1 and α2 represent the adjustment to equilibrium error.
Causality runs from the downstream price to
the upstream price if α1 and the ai jointly are
statistically different from zero. Similarly, causality
runs from the upstream price to the downstream
price if α2 and the ci are jointly statistically differ-ent
from zero. If both sets of coefficients are significantly different from zero, causality is bidirectional.
As shown in Table 2, shocks in the wellhead, electrical, industrial, and city gate prices
cause shocks in all other prices with which they
are paired. If we abstract from shocks that might
be initiated by pipeline companies or LDCs, shocks
originating in electrical and industrial prices most
likely reflect changes in the prices of competing
fuels. Shocks originating in the city gate price may
reflect the LDCs’ response to quantity shocks in
their end-use markets. For any particular downstream price, shocks to the wellhead price may
reflect changes in drilling technology, changes in
reserve estimates, contracts fixed to oil prices, and
changes in demand by other downstream buyers.
Federal Reserve Bank of Dallas
Price shocks in commercial and residential
prices do not cause shocks in other prices. This
finding most likely reflects how LDCs administer
natural gas prices for commercial and residential
users. In response to changing weather, fluctuating
economic activity, or changing prices for competing fuels, the customers adjust their consumption
of natural gas.12 Changes in consumption prompt
LDCs to seek smaller supplies at lower prices or
greater supplies at higher prices. Only as the city
gate price paid by LDCs is changed, however, are
price changes passed on to end users.
Adjustment to Equilibrium Error. If two variables
are cointegrated, any movement away from the
long-run cointegrating relationship will eventually be
corrected, and the variables will move back into
their long-run relationship. The adjustment could be
through one or both of the variables. Which variable
adjusts to the equilibrium error depends on many
factors, including elasticities and market structure.
In a cointegrated system (such as represented
by equations 2 and 3), the presence of an errorcorrection term implies that the dependent variable
adjusts to the equilibrium error. The coefficient on
the equilibrium error, α, reflects the extent to
which a given price variable reacts in the short
run to deviations from its long-run relationship
with another price variable. In the equations
where α is significant, the dependent variable
adjusts to deviations from the cointegrating relationship. In equations where α is not significant,
the dependent variable does not adjust to deviations from the cointegrating relationship.
As shown in Table 2, the electrical and
industrial prices do not adjust to errors in their
equilibrium relationship with the wellhead price.
Instead, the wellhead price adjusts. These findings
are consistent with electrical and industrial demand
being more elastic in the short run than is the
supply of natural gas. A high short-run elasticity
of demand reflects these end users’ ability to
switch fuels.
Similarly, the wellhead price adjusts to errors
in its equilibrium relationship with the city gate
price, but the city gate price does not adjust. The
resistance of the city gate price may indicate some
ability of commercial and residential customers to
switch fuels, the ability of LDCs to foster competition between suppliers, or an inelastic supply of
gas at the wellhead.
Economic Review — Second Quarter 1993
From our institutional knowledge of the
natural gas market, we can make a compelling
case for a very inelastic supply of natural gas in
the short run. Producers hesitate to vary production because doing so could disturb pressure in
the well, which could reduce its eventual output
or make production from it more costly.
The adjustment of commercial and residential
prices to errors in their equilibrium relationships
with upstream prices is consistent with administered prices.
Impulse Response. To examine the dynamic
properties of shocks to the price variables, we
calculate impulse response functions. The impulse
response function traces the effects and persistence
of a shock on both the upstream and downstream
prices. The persistence of a shock tells us how
fast the system adjusts back to equilibrium. The
faster a shock dampens, the quicker the adjustment.
We use the Choleski decomposition to
calculate impulse response functions for each of
the seven reduced-form vector-error-correction
models.13 For each model, we analyze the effects
of a one-time, standard deviation shock to the first
difference of each price in the pair.14 We trace the
effects of this impulse on the equilibrium error
and the upstream and downstream prices.
We find that the maximum impact generally
occurs within one to three months of the shock.
12
Econometric evidence suggests that residential and commercial natural gas consumption do respond to changes in
the prices of competing fuels (Bohi 1981). The response is
not through fuel switching in the existing energy-using
capital stock, however. The response comes through longterm changes in the energy-using capital stock.
13
The Choleski decomposition decomposes the residuals µ1t
and µ2t into two sets of impulses that are orthogonal to each
other. Orthogonalization allows one to take covariance
between the residuals into account.
The Choleski decomposition imposes a recursive structure on the system in which the ordering of the dependent
variables is specified. If the covariance between the residuals is sufficiently high, the ordering can affect the results.
We experimented using both changes in the upstream
price and changes in the downstream prices first in the
ordering.
14
This implies a permanent shock to the price.
47
Table 3
Impulse Responses in Upstream and Downstream Prices
(Based on Choleski decomposition)
Price Pairs
Upstream
Downstream
Month in which shock dampens to 5 percent of maximum
∆PD first shock to
∆PU first shock to
∆PD
∆PU
∆PD
∆PU
Wellhead
Electrical
8
19
16
20
Wellhead
Industrial
19
16
17
17
Wellhead
City Gate
*
*
*
*
Wellhead
Commercial
11
11
11
9
Wellhead
Residential
10
10
10
10
City Gate
Commercial
24
26
24
27
City Gate
Residential
14
16
15
16
* Shocks do not dampen.
In six cases, deviations from the long-run equilibrium relationship between the prices dampen
below 5 percent of their peak value within eight
to twenty-seven months after a shock, as shown
in Table 3.15
In four cases, shocks dampen in about one
and a half to two years. Deviations from the equilibrium relationships between wellhead price and
commercial and residential prices appear to dampen
in less than a year. This quick dampening appears
somewhat anomalous given the slower adjustment
to equilibrium in the relationships between city
gate price and commercial and residential prices.
Nonetheless, the quick adjustment is reasonable
because in the long run, wellhead, commercial,
and residential prices adjust to shocks originating
in city gate prices.16
48
15
In one case, the wellhead price-city gate price relationship,
the shocks did not dampen.
16
See “Long-Run Sources of Variance,” below.
17
See footnote 13.
Long-Run Sources of Variance. To find out
which price shocks are the most likely sources of
variance, we use the Choleski decomposition to
calculate the variance decomposition. For given
time horizons, the variance decomposition apportions the stochastic variability in a given price to
shocks in itself and the price with which it is
paired. Given our impulse response analysis, we
use sixty months to represent the long run.
As shown in Table 4, the calculated source of
variance is generally invariant to the ordering of
the variables in the Choleski decomposition.17 The
two models in which the wellhead price is matched
with electrical and industrial prices are an exception. If the change in wellhead price is placed
first, both shocks to the wellhead price and the
end-use prices are equal sources of variability. If
either the change in electrical or industrial prices
is placed first, shocks to the end-use price account
for more than 90 percent of the variance.
In these cases, we prefer the ordering in
which innovations in the end-use price are placed
first. This ordering presumes that shocks are most
likely to originate in the end-use prices, which fits
our view that changes in demand were a more
important source of shocks than changes in supply
Federal Reserve Bank of Dallas
Table 4
Long-Run Sources of Variance in Upstream and Downstream Prices
(Based on Choleski decomposition)
Price Pairs
Upstream
Downstream
Wellhead
Sources of variance
∆PD first
∆PU first
∆PD
∆PU
∆PD
∆PU
(Percent)
(Percent)
Electrical
50
46
50
54
3
5
97
95
Industrial
49
47
51
53
8
5
92
95
City Gate
27
25
73
75
25
4
75
96
Commercial
85
79
15
21
69
62
31
38
Residential
76
66
24
34
78
67
22
33
Commercial
94
92
6
8
99
94
1
6
Residential
96
87
4
13
100
88
0
12
Wellhead
Wellhead
Wellhead
Wellhead
City Gate
City Gate
during our period of analysis (Brown and Yücel
1993). Shocks to demand may come from fluctuations in economic activity, changing oil prices, or
other factors.
With our preferred ordering, we find shocks
to electrical and industrial prices to be the primary
sources of variance over the long run in their
pairings with wellhead prices. Our findings are
consistent with the long-run supply of gas at the
wellhead being fairly inelastic, and the long-run
demand for natural gas by industrial and electrical
users being fairly elastic.
Shocks to commercial and residential prices
are not the primary sources of variance in their pairings with the city gate price. We do find, however,
that shocks to the city gate price are the primary
source of variance over the long run in its pairing
with the wellhead price. This finding suggests that
LDCs drive city gate and wellhead prices in the face
of fluctuating sales to end users and that the longrun supply of gas at the wellhead is fairly inelastic.
Economic Review — Second Quarter 1993
Our evidence is consistent with administered
prices in the commercial and residential markets
for natural gas. As consumers lower (increase) their
consumption of natural gas, the LDCs pursue fewer
(greater) natural gas supplies at lower (higher)
prices. Only after the shock in the quantity of
natural gas demanded is transmitted to the city gate
price do commercial and residential prices for
natural gas adjust to changes in the city gate price.
Summary and Conclusion
Our econometric evidence indicates that
changes in natural gas prices are unequal in the
long run. Nonetheless, all downstream prices
change by at least as much as the average wellhead price. Statistically, residential and commercial prices change as much as the city gate price.
In the face of persistent shocks, however, market
institutions and market dynamics can lead to
lengthy periods in which the residential and com49
mercial prices of natural gas adjust less than the
wellhead or city gate prices.
Electrical and industrial users of natural gas
rely heavily on spot supplies and can switch fuels
easily. Their ability to switch fuels may be related
to the development of a spot market to serve
them. Reliance on the spot market may explain
why these end users have seen a greater reduction in natural gas prices than have the LDCs over
the past seven years. The ability to switch fuels
may account for electrical and industrial prices
being the source of shocks in their relationships
with the wellhead price. It also may explain why
prices in these end-use markets are quick to adjust.
Commercial and residential customers cannot
switch fuels easily and rely heavily on LDCs for
their natural gas. The inability of these end users
to switch fuels probably contributes to the reluctance of LDCs to purchase spot supplies of gas.
Reliance on contract supplies may explain why
the city gate price has not declined as much as
electrical and industrial prices of natural gas over
the past seven years.
Furthermore, the LDCs administer prices in
the commercial and residential markets under
state regulation. The administration of prices in
these markets leads to slower adjustment in commercial and residential prices. Only after city gate
18
50
prices can be reduced are commercial and residential prices for natural gas reduced.
Pipeline companies have been an integral
part of the uneven change in natural gas prices.
As natural gas prices have declined over the past
seven years, electrical, industrial, and city gate
prices have fallen more than the wellhead price,
while pipeline profitability has been reduced.
Our analysis suggests that the decline in pipeline
profitability is associated with reduced demand
for natural gas brought about by lower oil prices.
To summarize, public concern about recent
movements in natural gas prices, in which residential and commercial prices have fallen less than
wellhead prices, may be somewhat misplaced.
Although uneven changes in natural gas prices are
maintained in the long run, all downstream prices
change by at least as much as the wellhead price.
Uneven changes reflect differences in the end
users and the market institutions that serve them.
Our analysis indicates that if energy prices were
to rise, pipeline profitability would rise, and commercial and residential end users would see smaller
increases in natural gas prices than electrical and
industrial end users. Compared with other natural
gas prices, commercial and residential prices
would be slow to rise.18
Data limitations prevent us from testing for asymmetric
relationships.
Federal Reserve Bank of Dallas
References
Balke, Nathan S. (1991), “Modeling Trends in
Macroeconomic Time Series,” Federal Reserve
Bank of Dallas Economic Review, May: 19 –33.
Engle, Robert F., and Byung Sam Yoo (1987),
“Forecasting and Testing in Co-integrated
Systems,” Journal of Econometrics, 143–59.
Barcella, Mary L. (1992), “Natural Gas Distribution
Costs and the Economics of Bypass,” Annual
North American Conference, International
Association for Energy Economics, New
Orleans, October 25–28.
Farmer, Richard D., and Phillip Tseng (1989),
“Higher Old Gas Prices and the Staged Deregulation of the U.S. Gas Industry,” Energy
Policy (December): 567–76.
Bohi, Douglas R. (1981), Analyzing Demand
Behavior: A Study of Energy Elasticities (Baltimore: Johns Hopkins University Press for
Resources for the Future).
Brown, S. P. A. (1984), “Natural Gas Pipelines:
Rent Revealed,” in The Energy Industries in
Transition: 1984 –2000, John Weyant and
Dorothy Sheffield, eds. (Washington, D.C.:
International Association of Energy Economists).
——— and Mine K. Yücel (1993), “Market Structure and the Pricing of Natural Gas,” Federal
Reserve Bank of Dallas, work in progress.
Canes, Michael E., and Donald E. Norman (1984),
“Long-Term Contracts and Market Forces in
the Natural Gas Market,” Journal of Energy
and Development 10 (Autumn): 73–96.
Ellig, Jerome, and Jack High, (1992), “Social Contracts and Pipe Dreams,” Contemporary Policy
Issues 10 (January): 39–51.
Engle, Robert F., and C.W.J. Granger (1987), “Cointegration and Error Correction: Representation, Estimation, and Testing,” Econometrica,
(March): 251–76.
Economic Review — Second Quarter 1993
Johnson, Robert (1992), “Fueling Anger: Natural Gas
Prices Irk Both Producers, Users; Are Utilities
at Fault?” Wall Street Journal April 2: A1.
Lyon, Thomas P. (1990), “Spot and Forward
Markets for Natural Gas,” Journal of Regulatory Economics (September): 299–316.
Norman, Donald A., and F. Elizabeth Sowell (1988),
“Risk and Returns in the Interstate Natural Gas
Pipeline Industry.” American Petroleum Institute Research Study #044 (Washington, D.C.:
A.P.I., July).
Sims, Christopher A., James H. Stock, and Mark
W. Watson (1990), “Inference in Linear Time
Series Models with Some Unit Roots,” Econometrica, ( January): 113–44.
Stock, James H., and Mark W. Watson (1988),
“Variable Trends in Econometric Time Series,”
Journal of Economic Perspectives 2 (Summer):
147–74.
Yücel, Mine K. (1991), “A Closer Look at Natural Gas
Prices,” Federal Reserve Bank of Dallas Southwest Economy, November/December: 7–8.
51
52
Federal Reserve Bank of Dallas
Fiona D. Sigalla
Beverly Fox
Associate Economist
Federal Reserve Bank of Dallas
Assistant Economist
Federal Reserve Bank of Dallas
Investing for Growth:
Thriving in the World Marketplace
A Summary of the 1992 Southwest Conference
L
ong-term economic growth depends on investment today. That message reverberated throughout the Federal Reserve Bank of Dallas’ fifth annual
conference on the Southwest economy. “The
reality of free trade underscores potential weaknesses in the U.S. economy....As businesses in the
United States become even more interwoven with
businesses around the world, our success will be
critically affected by the way we manage problems
that reduce competitiveness,” cautioned Dallas
Fed President Robert D. McTeer, Jr.
Conference speakers expressed optimism
about opportunities for growth when and if reforms
under the North American Free Trade Agreement
(NAFTA) and General Agreement on Tariffs and
Trade (GATT) redefine the international marketplace. “The Southwest economy is in a position to
benefit disproportionately from free trade,” said
Gerald P. O’Driscoll, Jr.
The health of the domestic economy, however, will determine the nation’s future. To capitalize on the opportunities presented by a larger
market, the United States must overcome domestic
impediments that affect our growth potential. The
future, Admiral Bobby Inman explained, depends
on what we do to prepare this economy to compete more effectively.
Speakers identified several areas of the
economy that would pay future dividends from
investments in restructuring now. Ensuring free
enterprise and improving human capital were the
two most prominent examples. “For financial capital
and entrepreneurship to be productive, there has
to be investment in human capital, there has to
be investment in...physical capital, infrastructure.
There also has to be investment in...social capital,
Economic Review — Second Quarter 1993
[in] developing the kind of society that enables an
enterprise system to emerge,” Ernesto Cortes said.
Investing in free enterprise
To maximize long-run U.S. economic growth,
U.S. firms must thrive, which means they must
take risks, innovate, and create. Rules and regulations, some speakers complained, have become a
burden that inhibits innovation and risk-taking by
distorting market incentives and encouraging rentseeking behavior. “We are one of the few countries
in the world not moving toward freer enterprise
these days,” McTeer said. He cited Mexico, on the
other hand, as a nation that sparked a dramatic
increase in economic growth by restoring free
enterprise (see the box titled “Mexico’s Investment
in Free Enterprise”).
To move back to free enterprise in the United
States, policymakers must rethink economic incentives and government regulation. Conference
participants proposed several strategies for reform:
let domestic markets work, redefine government,
make a commitment to long-term growth, establish
credible policies, and allow free trade to let
international markets work.
Let domestic markets work. Government regulations can be beneficial when they organize society
or force firms and individuals to be socially respon-
The authors thank Rhonda Harris, Jerry O’Driscoll, and
Harvey Rosenblum for their helpful comments and suggestions.
53
Mexico’s Investment in Free Enterprise
Mexico is reaping the rewards of its
commitment to free enterprise and its investment in long-run economic growth. “At a time
when our own economic policies leave much
to be desired, Mexico’s policymakers are setting an example for the world,” said Dallas Fed
President Robert D. McTeer, Jr.
In 1982, Mexico was plagued by large
fiscal deficits, unsustainable levels of external
borrowing, a highly protected economy, rampant inflation, and heavy government regulation, explained Ariel Buira of Banco de Mexico,
Mexico’s central bank. Mexico has rebounded
after restructuring its economy through such
market-based reforms as fiscal and monetary
discipline, privatization, deregulation, and
trade liberalization.
Monetary Discipline
Mexico’s success in reducing inflation
expectations and long-term interest rates required a commitment to fiscal restraint. McTeer
explained how Mexico benefited from international interdependence through the use of
“external exchange rate discipline to reinforce...sound but temporarily painful domestic policies.” Buira agreed that Mexico was
attempting to import some price stability by
pegging its currency to the dollar, which has
sible. The highway code, for example, organizes
traffic to safeguard the public. Environmental
regulations force firms to be accountable for pollution they may cause. Some regulations, however,
inhibit decision-making, distort investment, and
transfer resources from one group to another. In
many cases, firms and individuals are able to
profit by subverting market forces and lobbying
the government, in effect, to regulate competitors
out of business. Such rent-seeking behavior differs
from productive investment because it does not
54
helped establish a credible monetary policy.
The peso, Buira said, “now moves in a fairly
narrow band and within that band can fluctuate freely.”
Fiscal Discipline
Mexico has successfully reduced its
public debt and curbed its public spending.
“The control of public finances has been the
keystone of economic stabilization,” said Buira.
At the beginning of its debt crisis, Mexico’s
budget deficit was 17 percent of GDP. “Sustained efforts to correct a huge deficit led to an
unprecedented fiscal adjustment of more than
15 percentage points of GDP in the period
between 1982 and 1991.” According to Buira,
Mexico’s fiscal adjustment is equivalent to
five times that considered under the Gramm–
Rudman–Hollings Act.
Privatization and Deregulation
Mexico’s far-reaching privatization program shrunk the public sector and helped
reduce the country’s public debt burden. “The
process of divestiture of state enterprises has
been intensified since 1989, with the privati(Continued on the next page)
create wealth. “At best,” Harvey Rosenblum
explained, “rent-seeking behavior redistributes
wealth; at worst, it destroys wealth.” Rosenblum
advocated restructuring regulations to institute
market-based incentives.
In banking, he said, regulation has often
limited choice, increased costs, stifled innovation,
and distorted investment. “The banking system
has already felt the repercussions of regional
shocks, yet geographic and product-line restrictions have limited banks’ ability to diversify their
Federal Reserve Bank of Dallas
Mexico’s Investment in Free Enterprise—Continued
zation of the two major airlines, major mining
companies, sugar mills, fisheries, the telephone company, all the commercial banks
and steel companies, and a number of others,”
Buira said.
The country also undertook major regulatory reforms. “Deregulation has advanced on
a number of fronts,” said Buira, citing the
elimination of restrictions to entry and licensing
requirements for telecommunications and land
transportation and the simplifications of regulations in industries such as air transportation,
mining, petrochemicals, and automobiles.
Trade Liberalization
Mexico’s far-reaching program of trade
liberalization demonstrates the benefits of letting international markets work. In 1985, Mexico
joined GATT, opened its economy to foreign
investment, and reformed the regulatory framework for economic activity. “This was a major
change in philosophy. The idea was that we
practices.” Bankers risking their own capital will
make better financial decisions than examiners,
regulators, legislators, and other government
bureaucrats, he said.
“The increased level of regulation that we have
is not particularly conducive to better serving our
customers in all cases,” said Linnet Deily. Regulations add to the cost of the services people buy,
she explained, adding, “I think that we have gone
overboard in response to consumerist activity without keeping the broader range of public interest
in place.”
Redefining government. Regulations may not
operate effectively because government is not
working properly. “We have to get government to
work,” said Donald Shuffstall. Tom Luce explained
that government has yet to undergo the dramatic
restructuring that has kept many firms in business.
Economic Review — Second Quarter 1993
would not be able to export unless we were
able to import freely,” Buira explained.
“Liberalization has stimulated both imports and exports. In fact, the growth of total
imports is explained to a very considerable
extent by imports of capital goods and intermediate inputs for export-oriented industries.
Exports of manufactured goods have replaced
oil as the country’s main foreign exchange
earner,” he said.
Buira was optimistic about the avenues
that will be opened with NAFTA. “Since the
liberalization of trade policy, trade between
Mexico and the U.S. has more than doubled,
to our mutual benefit. This will be further
enhanced by the establishment of a free trade
area in North America,” he said.
“With implementation of this program of
stabilization, structural adjustment, and liberalization, the economy has resumed growth
after a decade of stagnation. These reforms
have created the internal conditions for sustainable economic growth.”
“If my organization and your organization had not
adapted, we would not be here today. The one
institution...in our country that has not gone through
that dramatic restructuring is government,” he said.
Rosenblum noted that government often fails
to focus on long-term economic growth: “Our
country waits until problems are out of control
before beginning to discuss solutions, then responds
by applying Band-Aids to deal with symptoms
rather than causes.” And, Luce added, “Too often,
compromise has lead to changes on the margins,
which basically means that nothing fundamentally
has been restructured and changed.”
Special interests may be distracting government from decisions that would benefit long-run
economic growth, Luce and Rosenblum suggested.
Politicians are hesitant to adopt long-run strategies
that require short-run sacrifice, Rosenblum said.
55
“The reason we have changes on the margin,”
Luce explained, “is because every special interest
is represented, but very seldom is the common
good represented in terms of a voice that says,
‘Hey, what about the kids?’ ” He challenged people
and business to get involved in the political process and to learn how to bring about change.
Government credibility. Governments that set
long-term policies and stick to them establish
credibility that benefits their economies. Credible
policies encourage firms and individuals to make
long-run plans. One important policy that must be
credible is a government’s stance on inflation.
Said Rosenblum: “Inflation distorts prices, worries
investors, and slows capital spending. Sound,
stable, and credible macroeconomic rules that
allow economic agents to take a long-term view
support economic growth. Low-inflation countries
with sound monetary policies tend to grow faster
than countries with high inflation.” (See the box
titled “Exporting Credibility.”)
The exchange value of a nation’s currency
depends on three things, Rosenblum explained:
“One is the current interest rates in each country,
relative to one another. Another is expected
prices—expected inflation in each country. The
third factor...is the expected growth rate of national
income. Foreign exchange traders are betting every
day on economic growth, on whether or not
governments have a commitment to credible growth
policies.” The measure of Federal Reserve credibility, Rosenblum added, “is the gap between longterm interest rates and short-term interest rates.
When long-term rates get well above short-term
rates, the Fed may be suffering from a credibility
gap....Any policy that lacks commitment and credibility, whether it be a government or corporate
policy, is bound to fail.”
The credibility of government policies is easily
measured in the international marketplace. “In this
world, central banks can run, but they can’t hide.
Their policies will be evaluated instantly on the
world’s computer screens and reflected instantly in
currency prices,” McTeer said. “That doesn’t imply a
loss of power based on fundamentals. It does imply
a loss of the power to fool some of the people
some of the time. The external value of a nation’s
currency will depend on the relative soundness
of domestic monetary policies no matter what is
said or done in the foreign exchange markets.”
56
He suggested that the information revolution
will lead to, “not a gold standard or a Bretton
Woods standard but to an information standard for
money where currency values instantly respond to
each bit of new information in a global plebiscite.”
Letting international markets work. A government’s commitment to free trade must be credible.
Countries have many incentives to open their
borders to foreign markets. Free trade will stimulate economic growth, increase domestic demand,
and allow firms and individuals to consume a
greater variety of goods and services. In fact, free
trade is “the only magic bullet on the horizon
with the potential...to raise our living standards
significantly,” McTeer said, explaining that soon
the world will operate as a single economy.
Quoting Walter Wriston in The Twilight of
Sovereignty, McTeer said, “National sovereignty,
including monetary sovereignty, is rapidly succumbing to the relentless pressure of computer
and communications technology and to satellites
and fax machines.” This technology, McTeer
noted, gives individuals all over the world instantaneous access to information and makes national
borders increasingly irrelevant.
“An economy is more likely to grow if it has
an open, competitive trade policy rather than high
protectionist barriers,” added Rosenblum. “Isolation is synonymous with poverty. Closing borders
can be done only at a very extreme cost,” he said,
comparing nations whose principal difference is the
openness of their trade policies, not their people or
location. North Korea vs. South Korea, the former
East Germany vs. West Germany, Hong Kong,
Singapore, and Taiwan contrasted with mainland
China are all examples.
Donald J. Carty agreed that free trade is
essential, but he said that the manner in which a
country opens its borders is also important for the
long-run health of an economy. Carty said the
United States has been slow to open trade, often
implementing policies that sacrifice long-term
competitive advantage for short-term political and
economic gains. “While few U.S. companies can
blame all their international problems on U.S. trade
negotiations, I think it’s fair to say that U.S. trade
policy has contributed to our worldwide competitive difficulties in virtually every industry,” he said.
The United States “has often given considerably more than we’ve gotten in trade negotiaFederal Reserve Bank of Dallas
Exporting Credibility
As national economies become more
integrated into a global economy, the Federal
Reserve can contribute to improved living
standards worldwide by providing an anchor
for nations pursuing price stability, suggested
Dallas Fed President Robert D. McTeer, Jr.
“Economies that are trying to pursue price
stability that may not have a long tradition of
internal monetary discipline may wish to import some of that discipline or borrow some
credibility by pegging to a more stable currency,” he said.
“As international trade becomes more
important to the U.S. economy, we have to
consider the impact of foreign economic conditions on domestic economic conditions....A
more integrated world economy with increased
capital mobility has potential implications for
our conduct of policy,” he said.
A more formal exchange rate mechanism could be put into place after free trade
tions,” said Carty, observing that the United States
approaches trade negotiations much differently
than other countries. Most countries focus on
creating a competitive advantage for their producers, while the United States has many other goals.
The United States often begins negotiations with a
sense of obligation to help less-fortunate countries.
Then, believing in the superiority of domestic producers, the United States will support an agreement that opens trade opportunities, regardless of
whether the agreement is imbalanced in favor of
foreign producers. What’s more, U.S. economic
interests often have been secondary to security
and geopolitical objectives, he said.
When entering bilateral agreements, the
United States usually already has a highly developed
industry that is ready for worldwide competition,
“while most other nations want to limit competition to ensure their fledgling industry sufficient
market share.” As a consequence, most bilateral
Economic Review — Second Quarter 1993
and capital movements have spread to most
of the Western Hemisphere. “After growth
rates and inflation rates have converged and
monetary and fiscal policies are reasonably
compatible, then the Americas will need to
have a look at alternative exchange rate
mechanisms. Perhaps a tighter and more
formal exchange rate relationship would be
indicated,” he said.
In the meantime, “Hopefully, the Federal Reserve System will conduct monetary
policy in the next few years in a way that it may
someday be regarded, as the Bundesbank
has been in recent years, as an anchor of
stability. If that could be the case, our contribution to world standards of living would dwarf
any contribution we might make with occasional quarter-point jiggles in the fed funds
rates or half-point changes in the discount
rate,” McTeer said.
agreements “limit rather than encourage competition,” Carty said.
Carty used the airline industry to illustrate the
consequences of slow, uneven trade deregulation.
The domestic airline industry, which was deregulated in 1978, has had difficulty integrating with
the still heavily regulated international aviation
market. Current trade agreements, according to
Carty, give an advantage to foreign carriers that
can partner with a domestic airline to create a
global network, while U.S. airlines have much
more difficulty acquiring access to international
markets. Carty cited an open skies agreement
recently negotiated with the Netherlands, “which
gives KLM unrestricted rights within the United
States in exchange for allowing the U.S. carriers to
have unrestricted rights in the Netherlands.” That
agreement “will strengthen KLM at the expense of
U.S. carriers,” he said. Further, U.S. carriers operating abroad must deal with a host of nontariff
57
barriers including, in some countries, requirements
that U.S. carriers hire their competitors to provide
customer service. “The restrictive instincts of most
of the world’s governments and the unwillingness
of the U.S. government to exert its negotiating
leverage have denied U.S. carriers not only dominance but even the level of leadership that should
be the natural benefit of being based in the world’s
largest aviation market,” said Carty.
The time has come, he said, for the United
States to call for a worldwide, multilateral open
skies agreement: “We can turn all the world’s
airlines loose to work the magic of competition in
every market. But if other countries are not willing
to support new opportunities for all, they must
not be allowed to buy their way into U.S. markets
to the detriment of U.S. producers....Most governments recognize...that workers and consumers are
simply the same folks in different clothes.”
Investing in human capital
Throughout most conference sessions, participants returned to the theme of investing in human
capital—the people whose skill and labor make
up a nation’s work force. Inman pointed out that,
while the United States is always at the forefront
of creating new technologies, we have become
increasingly laggard in turning that technology
into commercial goods and services. Part of the
reason, he suggested, is because we do not have
the highly trained work force to operate sophisticated technologies.
“There was a day, not too long ago,” said
Paige Cassidy, “when any average high school
graduate with basic mechanical aptitude could
expect to find employment in industry. That day
is gone. The value of unskilled labor is rapidly
disappearing. In our workplace of the future,
employees on the factory floor must be highly
literate and computer friendly. And if industry is
to be competitive and if our national economy
is to be viable, we must have a skilled, highly
trained work force.”
Speakers suggested several ways to maximize
the potential of America’s human capital: encourage immigration, capitalize on cultural diversity,
reinvigorate education, and reform health care.
The need to control costs, reduce bureaucracy, and
implement market-based reforms in the economy
58
was a common theme. Our economy, they said,
needs major structural reforms rather than tinkering at the margins.
Immigration. Opening our borders to both goods
and people will boost our nation’s human capital
and stimulate economic growth. As Julian Simon
explained: “Every time our system allows in one
more immigrant, on average, the economic welfare
of American citizens goes up, and every time we
keep out one more immigrant, on balance, our
economic welfare goes down....Additional immigrants, both the legal and the illegal, raise the
standard of living of U.S. natives and have little
or no negative impact on occupational or income
class.”
Since the turn of the century, the United
States has become much bigger, wealthier, and
better able to assimilate growing numbers of immigrants. But, Simon explained, both legal and illegal
immigration have declined significantly, both in
absolute numbers and as a proportion of the total
population in the United States: “We have this
phrase that we are a nation of immigrants; we are
not a nation of immigrants now. Indeed, there are
many countries in the world that we tend to think
of as homogeneous that have a much larger proportion of immigration than we do. For example,
Great Britain, Switzerland, France, and even Sweden
have a much larger proportion of immigrants in
the population than we do.”
Misperceptions about immigrants hurt the
economy, he said. “There is only one painless way
of dealing with the deficit that does not mean the
pain of...reducing services or increasing taxes,
and that is to bring in more immigrants....Immigrants
pay much more in taxes than the cost of the welfare services they use. Immigrants come when they
are young, when they are strong, and when they
are earning and contributing, not when they are
old and are taking in.”
Rather than displacing natives from jobs,
immigrants create jobs through their purchases
and by opening new businesses—small businesses,
a main source of the nation’s jobs. “Immigrants
earn and they spend, and their spending provides
jobs for others. Immigrants simply expand the
economy,” he said, explaining that immigration
helps raise productivity, improves our competitiveness, and provides an invaluable network of
communications abroad.
Federal Reserve Bank of Dallas
Simon suggested that immigration is often considered harmful because “it is very easy to identify
the losers, much harder to see the winners....It is
so commonsensical that immigrants push natives
out of jobs, [but] the virtue of economics is it gives
us anti-commonsensical, anti-intuitive answers to
many questions.”
Simon recommended a policy of phased,
incremental increases in immigration: “Increase
the level of immigration by a million people now,
and watch what happens in the next three years.
If anything unexpected happens, we could adjust
for it.” If the immigrants continued to assimilate
nicely into the economy, he said, the United States
could raise the limit by another million for another
three years and continue in a “systematic and
controlled way...to allow in more immigrants and
make us a better country.”
Capitalizing on our diversity. Racism and naiveté
about other cultures cause some people to oppose
immigration. But the key issue for policymakers
should be how many human beings, not which
ones, Simon said.
To benefit from the human capital open
borders provide, the country must also learn to
value fully all of the human capital that is already
here. “Our greatest strength is our cultural diversity, but we are not using the full capability of all
our work force,” cautioned Major General Hugh
Robinson.
Some corporations, Robinson explained,
undervalue workers by applying stereotypes to
women, African–Americans, Hispanics, Asians,
and others. When people act as if these stereotypes are true, the country does not use its work
force to its full capability. “We are talking about
using the full capability of our work force and
not devaluing some part of it for some unknown
reason that nobody can figure out....It is a business
decision to develop our work force so that all
have the same opportunity,” he said.
Cultural diversity is also an asset in the
global marketplace because corporations must
deal with the diversity that they will encounter in
other nations. “Should we not put our best foot
forward by utilizing the diversity we have here in
this country?” Robinson asked. “It is a business
decision to adjust the work force and to participate actively in this global environment.”
Capitalizing on our diversity will become
Economic Review — Second Quarter 1993
even more important as our population becomes
more diverse. According to Robinson, “85 percent
of the entrants into this country’s work force in
the year 2000 will be people of color and women.
Furthermore, by the year 2050, projections show
that whites will slip below 50 percent of the
national total. If you look at the world in which
we live, people of color are already and have
long been the majority.”
Robinson noted that while education is the
key to our success as a nation, who does the
teaching and what is taught are also very important: “We must really understand diverse peoples
and cultures, and we must do it in a way that
promotes a valuing of diversity. If we don’t value
diversity, it won’t be a plus for us....In American
education, as we move through the final decade
of the twentieth century, we have the responsibility to move closer to the fundamental goal that
underlies multiculturalism, pluralism and cultural
diversity—the transformation of the American
economy into a place where difference no longer
makes a difference.”
Education. Other speakers echoed Robinson’s
concern about the importance of education as a
factor in America’s competitiveness and long-term
growth potential. “Education,” McTeer said, “is
key because the United States has a competitive
advantage in producing goods and services that
use our abundant human resources. Therefore,
the more open we become to the world, the more
serious will be the consequences of a flawed
education system.”
The U.S. system of higher education is competitively supplied, and it is the envy of the world.
In contrast, McTeer pointed out, our primary and
secondary systems are basically “local monopolies
[that] not only are expensive; they don’t consistently produce a quality product.”
Education is one way to obtain “not only
higher incomes, but a fairer or more equal
distribution of income as well—clearly a win-win
situation,” suggested Rosenblum. Speakers criticized
the current educational system for not adequately
preparing our youth and for failing to implement
successful reform despite years of talk. They
called for fundamental change.
“Tinkering at the margins is not going to
work,” said Milton Goldberg. “If our schools are
to be competitive in the twenty-first century, then
59
nothing short of revolutionary change will have to
occur. Despite almost a decade of talk about education reform, education reform has been disappointing.” Often, educators have attempted only reforms
that are easily implemented, rather than working
on more complex, in-depth changes, he said.
“Look what happens when we don’t make
the investment in education,” Robinson said. “We
end up sending people to prison and having to
support them there at some ridiculous cost of
$30,000 or $40,000 per year. We send them to
hospitals because their health isn’t good. We send
them to other places where our public funds are
spent in a nonproductive fashion....If you think
education is expensive, try ignorance,” said
Robinson, quoting former Harvard University
President Derek Bok.
“If we merely put more money in the same
system”, Luce said, “we will get the same results.
We have to begin to change the system.” “How
do we break from tradition? What is it going to
take to break the mold to make a difference in
terms of doing what is right for our young people?”
asked Marvin Edwards.
Three strategies the speakers suggested to
improve education were emphasizing parental involvement, restructuring the educational system to
incorporate market-based incentives, and increasing the amount of time children are at school.
Parental involvement. According to Clint Bolick,
“The one thing almost everyone in education will
agree on is that parental participation is the single
most crucial factor in terms of educational outcomes.” Cassidy agreed: “If parents are not involved,
everything slows down. When the parents are involved, things speed up very quickly....[But] we
know many children in this country are not coming
to school ready to learn.” Many children are in an
environment that does not prepare them to be good
students, and some parents, Edwards said, “are not
prepared to be parents and need to be trained.”
Cassidy described several programs that
would involve parents and the community as a
team to help prepare children and facilitate the
educational process. Many of these break-the-mold
programs are being financed by the New American
Schools Development Corporation.
“I think we have to do a much better job in
American education of making it very clear...that
the parent is the child’s first and most influential
60
teacher,” Goldberg said. “I feel about parent
responsibility as I do about student expectation.
Our expectation for achievement in American
schools generally has been very low; students have
given us exactly what we have expected of them.
High expectations never hurt anybody.”
Incentives. Some speakers suggested that the education system, like the rest of government, would
operate more effectively if it incorporated marketbased incentives and were less regulated. “Our
classroom teachers are the most regulated professionals in America. We need to treat our teachers
as professionals. They must be empowered to
function like other professionals, with the opportunity to adapt, change, create, and modify as
they best judge,” Cassidy said.
Luce concurred: “What we need to do is treat
the teaching profession once again as a profession,
not a guild. We have to break the cycle of acrossthe-board pay increases....We have to change the
rules so that teachers who are not performing can
be discharged.” Luce suggested that the business
community use its skills to develop procedures for
the difficult task of evaluation and assessment.
“I understand why many teachers do not want to
be evaluated by the principal who, in their mind,
is a fired football coach. But let me tell you something. We do have assessment in football. If we
do not win in football, you are fired.”
The current education system operates based
on lobbying much like the rest of government, he
said. “Do you know in Texas public schools, the
last time I looked, there are 600,000 students taking
agriculture vocational educational courses? Folks,
there will not be 600,000 agriculture jobs in the
state in the next five lifetimes available to kids.
But we have 600,000 kids taking agriculture vocational education. Why? Because the agriculture
vocational education lobby is strong as horseradish....We give a school more money to teach a
vocational education course than we do to teach
math and English,” Luce said, adding that many
schools spend more money on one-day processing of their football films than on their English
department curriculum.
Restoring incentives would spur innovation
in the educational system, Luce said, suggesting
that, rather than pay schools for having children
in class, it may be more effective to pay schools
for graduating children. “Every child can learn, but
Federal Reserve Bank of Dallas
every child learns in a different, unique, special
way. And yet we are still conducting our classes
as if they were assembly lines for a manufacturing
model that disappeared many years ago.” With an
incentive program, Luce said, high school classes
could grow to as large as 250 students. “One year
later, they go to college and they have a class that
big. Wouldn’t it make more sense to do that and
take that same money and spend it on a 3-yearold?...We should have prekindergarten programs
so all children, of all races and financial groups,
will have an equal opportunity to line up at the
starting line in the first grade equally prepared
and equally equipped,” he said.
Choice. One way to introduce incentives into
America’s educational system is by making public
schools compete with private schools. Several
speakers noted that public schools are structurally
different from private schools. Competition has
helped keep down the size of private school
administration. In large urban school systems,
Bolick noted, about 50 cents of every dollar spent
on education makes it to the classroom in terms
of salaries and instruction; in private schools,
about 95 cents of every dollar spent makes it to
the classroom.
School choice, a proposal that would enable
parents to apply state funds to either public or
private schools, can help introduce incentives into
the public education system because schools would
have to compete for students. Bolick advocated a
system in which public schools and private schools
participate on a voluntary basis. School choice,
he said, could provide the impetus for radical
deregulation and decentralization of public schools:
“Public schools will compete for kids they previously could take for granted, since these kids had
nowhere else to go.” Bolick would like a public
school system “that exists because parents choose
to send their children there, not because they have
no alternative....Choice transfers power over basic
educational decisions from bureaucrats to parents,”
he said, adding that parents have the greatest
stake in their children’s success.
Choice has worked at universities, Bolick said,
because college students can take their financial aid
anywhere. With choice, “if the schools do not attract
enough students, they literally go out of business.”
But proposals for choice leave a lot of questions unanswered, according to Edwards. “We
Economic Review — Second Quarter 1993
have heard a lot about choice over the years....Is
choice the ability to have any child choose any
school anywhere? If so, are we going to provide
transportation for any choice any child makes anywhere? So then we have a major, major, financial
commitment.”
Bolick countered: “In most instances you
could have choice plans and provide transportation for low-income kids and still not spend as
much money as the public schools are currently
spending.”
“I think we can probably make anything a
success in a vacuum, in a small scale....Choice is
not realistic unless choice is available for everybody. We have a lot of experiments in choice, but
we do not have choice. No one has shown how
choice can work in an entire state or an entire
county or across lines of a school district,” Edwards
said. Choice “is an emotional response to a need
on the part of the American people, because it
sounds good, but no one is ready to define it. Until
someone has the courage to really define it beyond
emotion, we do not have an answer to this big
dilemma.”
Despite many good suggestions for reforming the education system, problems arise in implementing effective change, speakers noted. “We
need to guard against quick fixes—ideas that
simply sound good and capture people’s emotion
are not going to fix education,” said Edwards,
who recommended working within the current
system to find reforms that can be effective.
Goldberg, however, restated the need for
systemic reform that considers many variables if
we are to improve the achievement of American
children. He noted a problem in moving educational reform from pilot programs into mainstream
reform. “American education is a graveyard of
innovations that have worked. One of the most
serious problems in American education is our
inability to learn from experience.”
Time and learning. Goldberg suggested that the
United States rethink the way students spend their
time. American students, he noted, spend less
time on school work than students in most nations,
and the time that is spent in the classroom and on
homework is often used ineffectively.
Goldberg and Cortes recommend lengthening the school day to give teachers more time for
planning and to help students be more motivated
61
to learn. “Students will spend extra time in additional recess, additional lunchtime, and additional
extracurricular activities....When the youngsters
come into the classrooms, the work is intensive,
direct, and heavy. The youngsters have already
blown off steam,” said Goldberg, adding that
teachers will be held accountable for planning
and will have more time to work at their jobs.
“There are some very simple ways that time
can be increased in American schools. For example,
we have to reduce student absenteeism. A youngster who is not in school is not learning schoolwork. We can improve school management to
make distractions in the classroom far fewer. We
can improve classroom management so teachers
know how to do the instructional work more
efficiently. And we can do the most dramatic of
all; we can restructure the school day and think
about year-round schools. The research is very
clear: there is a very direct correlation between
time spent on homework and student achievement,
particularly at the junior and senior high level,”
Goldberg said.
Lengthening the school day will increase the
productivity of our past investment in education,
he said, explaining: “We have a capital investment
of enormous proportions in the school buildings
of this country. These fancy buildings that we
built all over the country for hundreds of millions
of dollars are being used for instruction 15 percent
of the available instructional time.”
Health care. Another way to invest in human
capital is to cure our nation’s sick health care
system. Spiraling public and private health care
costs have increased the expenses of employers
and government and are limiting the availability
of medical services. Taming health care costs is
crucial to containing business costs and ensuring
a quality work force.
Speakers addressed two problems: consumers
cannot obtain the information necessary to make
choices about the cost and quality of various health
care options, and medical care is overconsumed
because consumers do not pay the marginal cost
of health care services. “There are strong reasons
to believe that the additional medical services
patients receive are not worth the additional costs,”
William A. Niskanen suggested. Speakers called
for a structural reorganization of the health care
system to realign incentives and reduce costs.
62
“We need a major overhaul instead of marginal
changes...that do not lead to overall savings,”
Ron Anderson said.
Costs. Rising health care costs have placed a heavy
burden on our economy. In the past 30 years,
“total expenditures for medical care have increased
from about 5 percent of GDP [gross domestic
product] to now about 14 percent of GDP....The
cost of health insurance is now the most rapidly
increasing component of private payrolls, and
payments for public medical programs are the
most rapidly increasing component of government
budgets,” Niskanen said.
Two causes of rising health care costs cited
by participants were consumers’ and physicians’
lack of incentive to reduce expenditures in a
system subsidized by insurance, government payments, and tax deductions; and a lack of competition that stems from the difficulty consumers
experience in collecting information about the
cost and quality of health care providers.
Under the current system, individuals who are
most able to control costs have little incentive to
limit expenditures. “The share of medical costs
that are borne directly by the patient has declined
from about 50 percent to under 20 percent,” explained Niskanen. “Neither patients nor physicians
have an adequate incentive to control the costs of
medical care.” Health care is subsidized by tax
deductions and insurance; consumers do not pay
the true price of medical care. Frequently, consumers do not even see the true price because
a third party—government or private insurance—
pays the bills.
“The common phrase ‘health insurance’ is a
double misnomer. The event that is insured is not
some adverse change in health status but is the
payment for some specific medical service. The
basic concept of insurance is to reduce the variance
of cost among groups with the same prior risks.
Most plans, however, include people with very
different prior risks in the same premium pool.
What we call health insurance in this country would
be much better described as a medical prepayment plan. These plans redistribute income from
people who use few medical services to people
who use many medical services,” Niskanen said.
Subsidizing health care costs results in an
overutilization of health services beyond the level
that is optimal for society. The use of technology
Federal Reserve Bank of Dallas
in medicine illustrates the breakdown between
health care incentives and costs. “The technological revolution,” Douglas Werner explained, “be it
therapeutics or diagnostics or combinations thereof,
is the major cost driver [in medicine].”
“The people who make decisions on the
types of technology to use do not pay the bills. In
basically all other sectors, improvements in technology have led to reduction of relative costs, not
increases in relative costs,” Niskanen commented.
Hospitals can afford to purchase amenities such
as chandeliers or another CAT scan or another
MRI because “someone else is paying for that.
They are not really making the hard decisions
because they pass those costs on to someone
else,” Anderson said.
Consumers have no idea if they are buying
chandeliers or superior medical care because they
lack information about the cost and quality of
health care services. Individuals needing health
care rarely call and ask a hospital’s room rate or
mortality rate, for example. “We have a system
driven by utilization,” said Anderson, adding that
there is no measure for health care quality and
“people don’t know what they’re buying.” Harper
agreed and noted that hospitals and doctors respond
to these incentives: “The reward system has been
wrong. Rewards have been based on utilization,
on quantity and not on quality of care.”
The lack of incentive for cost or quality control
has encouraged the growth of an expensive health
care bureaucracy. According to Anderson, 23 percent of health care costs may be the result of the
bureaucracy, which, he believes, is making the
health care system incoherent. “It should not drive
patients and doctors and nurses and everyone
crazy because it is irrational,” he said.
Panelists disagreed on the most effective
reform. Managed care, in which the public or
private sector decides which doctors consumers
should use and which illnesses will be paid for,
could help control health care costs. Market-based
reforms, including the elimination of subsidies to
reduce demand, were also suggested. Panelists
disagreed on how reforms should be implemented
but generally agreed that consumers need more
information about the cost and quality of health care
providers, along with incentives to control costs.
Increasing information. Measuring health care
quality and cost and making that information
Economic Review — Second Quarter 1993
accessible to the public could help individuals
make decisions about health care purchases.
“I think competition will work if the purchasers
of health care really know what they are buying,”
said Robert Shoemaker. “The system needs information to identify high-quality, cost-efficient providers of services,” explained Dwain Harper.
“If you look at the literature, [there is little] agreement on what is a measurement of quality....There
is going to have to be some investment put into
developing the technologies to measure these
results,” he said.
The Cleveland Health Quality program, the
state of Oregon, and others have attempted to
improve health care productivity by using technology to bridge the information gap and help consumers measure quality. The Cleveland group
abstracts information manually off the records of
thirty-one hospitals and places it on a database.
Insurance companies can then easily compare information about hospitals that is risk-adjusted for
outcomes and patient satisfaction. According to
Harper, this program helps determine the “highest
quality outcomes and the most reasonable costs.”
Providing incentives to control costs. Even if more
information about the cost of health care services
were available, consumers would still need an
incentive to base decisions on what they know.
One suggestion for creating such an incentive is
to require consumers to pay the marginal cost of
consuming medical care. For example, curbing tax
deductions for private and public health insurance
could reduce the stimulant on the demand for
medical services. According to Niskanen, “The
reduction of tax-subsidized medical prepayment
plans is a necessary condition to reduce the
growth of demand for medical care.”
He advocated an income test to limit tax
deductions for medical services and insurance and
adding deductions for set levels of preventative
care. “It is important to recognize that a substantial part of tax-subsidized health insurance accrues
to higher income people. Higher income people
are more likely to be privately insured, and the
value of that insurance is a function of their marginal tax rate. Similarly, the people on Medicare
with the highest incomes are the ones who likely
live the longest, and the value of Medicare increases
with the marginal tax rate. Clearly, the amount of
tax-subsidized health insurance could be substan63
tially reduced without much change in the insurance available to the poor.”
Niskanen also recommended that the current
prepayment form of medical plans be restructured
as indemnity insurance, similar to auto insurance.
“Patients would be paid a fixed amount above
some deductible per illness or accident but would
bear the sole marginal cost of whatever medical
services they elect,” he said.
Managed care. Managed-care programs reduce
costs by providing a fixed level of care at a reduced
price. Under managed care, a company or government agency decides which types of illnesses to
treat and where consumers can obtain medical
care. Costs can be reduced by having a network
of physicians under contract with agreements on
rates, patient volume, and quality, Werner said.
He noted that managed-care programs are estimated
to reduce costs by 15 to 25 percent. “It is time for
somebody to address the costs, even if it means
capping reimbursement,” Anderson commented.
The state of Oregon and several private
insurance companies are attempting to control
costs through managed care. The Oregon plan,
Shoemaker explained, categorizes groups of people
and determines which medical procedures will be
covered and at what price for each group. The
program estimates the providers’ costs and adds
“a reasonable level of compensation.” Shoemaker
stressed that “emphasis is placed on preventive
care,” but “people will get an adequate, but not
an excessive, level of care.”
Anderson suggested that a managed-care
program should be combined with indemnity
64
insurance and operated with a single-reimbursement system. Such a system, he said, would be
understandable, portable, and would stop costshifting. “I believe it can be a social insurance program with competition for quality, delivering care
that is patient-centered,” he said. Werner, however, complained that a single-payer system would
be expensive because it would lack competition
on the administrative costs of health care.
Werner suggested managed care as a viable
strategy to use in a transition period, until “we get
better control of measuring quality, providing
quality, and understanding the economic benefit
of quality medical services.” Niskanen countered
that while managed care costs less than care in
which people select their own physicians, it is a
different standard of service that “does not prove
to be effective in reducing the rate of increase in
total cost.”
Conference participants, concerned about
the availability of health insurance, noted that it is
cheaper to provide preventative care than emergency care for people without insurance. Shoemaker advocated that employers be required to
either make health insurance available to employees
or pay a tax that will give them access to an insurance pool, a system frequently referred to as play
or pay. Anderson and Niskanen expressed concern
that a play-or-pay system would force employers
to cut back on employees or wages. “Pay or play
is merely a transition to national health insurance,”
Niskanen said.
Federal Reserve Bank of Dallas
Investing for Growth: Thriving in the World Marketplace
A conference sponsored by the Federal Reserve Bank of Dallas
October 29 –30, 1992
The Role of the Federal Reserve in the International Economy
Opening Address
Robert D. McTeer, Jr., President, Federal Reserve Bank of Dallas, Dallas, Texas
Investing in International Trade and Growth
Chairman
Gerald P. O’Driscoll, Jr., Vice President and Economic Advisor, Federal Reserve Bank of Dallas,
Dallas, Texas
Speakers
Structural Reform in Mexico
Ariel Buira, Director of International Affairs, Banco de Mexico, Mexico City, Mexico
The Positive Consequences of Immigration
Julian Simon, Professor of Business Administration, University of Maryland, College Park, Maryland
The Future of the Southwest in a Global Market
Admiral Bobby R. Inman, U.S. Navy (Retired), private investor, Austin, Texas
The Diversity Imperative: Looking to the Year 2000
Major General Hugh G. Robinson, U.S. Army (Retired), Chairman and Chief Executive Officer,
The Tetra Group, Inc., Dallas, Texas
Investing in Human Capital: Revitalizing Education
Chairman
Marvin E. Edwards, General Superintendent, Dallas Independent School District, Dallas, Texas
Speakers
It’s About Time to Learn
Milton Goldberg, Executive Director, National Education Commission on Time and Learning,
Washington, D.C.
The School Choice Imperative
Clint Bolick, Vice President and Director of Litigation, Institute for Justice, Washington, D.C.
New Ideas in Education
Paige S. Cassidy, Director of Communications, New American Schools Development Corporation,
Arlington, Virginia
Investing in Human Capital —Reforming Health Care (A Panel Discussion)
Moderator
Michael R. Levy, Publisher, Texas Monthly, Austin, Texas
Panelists
Ron J. Anderson, President and Chief Executive Officer, Parkland Memorial Hospital, Dallas, Texas
Economic Review — Second Quarter 1993
65
Panelists — Continued
Dwain L. Harper, Executive Director, Cleveland Health Quality Choice Coalition, Cleveland, Ohio
William A. Niskanen, Chairman, CATO Institute, Washington, D.C.
Robert C. Shoemaker, Jr., State Senator and Chairman—Health Insurance and Bioethics Committee,
Portland, Oregon
Douglas C. Werner, Senior Vice President—South Central Market Area, Aetna Health Plans,
Richardson, Texas
Economic Principles vs. Political Agendas
Harvey Rosenblum, Senior Vice President and Director of Research, Federal Reserve Bank of
Dallas, Dallas, Texas
Investing in the Southwest Economy (A Panel Discussion)
Moderator
Michael R. Levy, Publisher, Texas Monthly, Austin, Texas
Panelists
Ernesto J. Cortes, Jr., Southwest Regional Director, Industrial Areas Foundation, Austin, Texas
Linnet F. Deily, Chairman, President and Chief Executive Officer, First Interstate Bank of Texas,
N.A., Houston, Texas
Tom Luce, Partner, Hughes and Luce, Dallas, Texas
Donald C. Shuffstall, President, Don Shuffstall International, El Paso, Texas
International Aviation: Opportunities If...
Donald J. Carty, Executive Vice President—Finance and Planning, AMR Corp. and American
Airlines, Inc., Dallas–Fort Worth, Texas
66
Federal Reserve Bank of Dallas
@FedFRASER