The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
The Federal Reserve Bank of San Francisco’s Economic Review is published quarterly by the Bank’s
Research and Public Information Department under the supervision of Michael W. Reran, Senior Vice
President. The publication is edited by William Burke, with the assistance of Karen Rusk (editorial) and
William'Rosenthal- (graphics-)- Opinions-expressed in-the Economic Review-do-net necessarily reflect the
views of the management of the Federal Reserve Bank of San Francisco, or of the Board of Governors of the
Federal Reserve System.
For free copies of this and other Federal Reserve publications, write or phone the Public Information
Department, Federal Reserve Bank of San Francisco, P.O. Box 7702, San Francisco, California 94120.
Phone (415)544-2184.
2
Measurement and Policy
I. Introduction and Summary
II.
“ Shift Adjustments’ ’ to the Monetary Aggregates
Barbara A. Bennett
... Further adjustments to the monetary aggregates may be necessary to take account of the
many sweeping changes now taking place in the U.S. financial system.
III. Enforcing Anti-Redlining Policy
Under the Community Reinvestment Act
Alane K. Sullivan and Randall J. Pozdena
.. .In view of analytical limitations, as well as Congressional intent, CRA’s anti-redlining
provisions should center on the detection of irrational redlining—that is, arbitrary geo
graphic discrimination.
IV.
Pricing Federal Irrigation Water:
A California Case Study
Yvonne Levy
... Pricing reform is necessary to improve the efficiency of water usage, through the use of
more efficient irrigation methods and shifts to less water-intensive crops.
Editorial committee for this issue:
Brian Motley, John Scadding, and Barbara Bennett
L
3
asurement and Policy
ment of "other checkable deposit gro\x/th" that
occurred after the nationwide introduction of NOW
accounts at the end of 1980.
But Bennett continues, "We have not seen the
last of the sweeping changes recently taking place in
the U.S. financial system." Money-market funds
continue to grow rapidly. Increasing numbers of
brokerage firms and depository institutions are announcing deposit-sweeping services, while larger
numbers of banks and thrift institutions are offering
retail repurchase agreements and loophole accounts. In addition, the pressure to deregulate
deposit-interest rates continues to mount, and regulatory authorities have met that pressure by creating
short-term accounts designed to permit depository
institutions to compete more effectively with
money-market funds.
Because of these developments, Bennett argues,
"Observed MI growth may continue to give somewhat misleading policy signals." To the extent that
distortions in MI growth can be traced specifically
to the growth in certain financial instruments, shift
adjustments may be useful. But she cautions that
many of these changes cannot be quantified with
even the same degree of certainty as the NOW
account shifts.
Alane Sullivan and Randall Pozdena consider the
measurement problems involved in implementing
anti-discriminatory housing credit policy under the
Community Reinvestment Act (CRA). The act was
designed to encourage financial institutions to
"help meet the credit needs of the local communities in which they are chartered." To meet that
policy goal, the CRA directs each supervisory
agency to take into account a financial institution's
CRA record when ruling on branch, merger or other
applications. However, the affirmative orientation
of the CRA represents a significant departure from
standard bank-regulation procedures, which were
designed primarily to insure the safety and soundness of the banking system.
Proper analysis of public-policy issues depends
heavily on proper measurement of the economic
quantities involved. This issue of the Economic
Review demonstrates this obvious truth with examples taken from several widely different fields.
One article analyzes the shift adjustments taken to
improve the measurement of the monetary aggregates. A second article discusses ways of improving
the measurement of "redlining" in bank lending
practices. A third article proposes a change in the
pricing mechanism for irrigation water, as a means
of improving resource allocation in California's
Central Valley.
Barbara Bennett argues that changes in the public's demand for various types of financial instruments have aitered the meaning of the monetary
aggregates, making observed growth in these aggregates harder to interpret. The growth in MI, in
particular, has slowed considerably over the past
few years. Yet with the proliferation of higheryielding substitutes for the traditional M I-type
transaction instruments, slower observed growth
may not necessarily be associated with a slowdown
in the economy.
The Fede;ral Reserve has sought ways to minimize the effects of recent financial innovations and
regulatory changes upon the meaning of the monetary aggregates and their relationship to economic
activity. As Bennett notes, one part of the effort has
centered around the redefinition of the monetary
aggregates in 1980. In addition, the Federal Reserve has come to place greater emphasis in its
policy deliberations on broader aggregates, whose
growth rates and relationships to economic activity
are affected less by shifts of funds among financial
instruments. Again, the Federal Reserve has attempted to cope with the problem of measuring and
interpreting money growth by adjusting observed
growth rates of the aggregates to account for distortions caused by shifts of funds among financial
instruments. The obvious case is the Fed's treat4
The CRA has its origins in long-standing allegations by community groups that financial institutions discriminate against certain neighborhoods in
credit decisions. The practice called neighborhood
"redlining" allegedly contributes to, and even
causes, the decline of inner-city neighborhoods.
However, in view of analytical limitations, as well
as Congressional intent, the authors believe the
CRA's anti-redlining provisions should center on
detection of irrational redlining, or arbitrary geographic discrimination that is contrary to sound
business judgment.
With this in mind, Sullivan and Pozdena measured the usefulness of various analytical techniques and data sources in detecting the arbitrary
use of property location in mortgage-lending decisions. They found that simple index techniques
were unreliable, because they ignored the complexity of the economic decisions involved in the mortgage market. (These measures failed to account for
the sound business reasons or demand factors which
may be the cause of disparities in loan volumes
among neighborhoods,) They also found problems
with the "market model" approach used in more
sophisticated studies, because of the difficulty of
defining an individual lender's role in such a
complex context. "The most reliable technique for
evaluating charges of geographic discrimination appears to be loan applications analysis, which permits the scrutiny of a credit supplier's individual
lending decisions."
The authors argue that effective CRA enforcement may require substantive changes in the methodology used by regulators in evaluating allegations
of redlining. "In the absence of quantitative evaluation techniques, CRA assessments today largely
depend on the judgment of CRA examiners. Since
the detection of CRA violations is considered an
important regulatory responsibility, decisions
should be accurate and consistently applied, given
their far-reaching consequences. The use of formal,
objective methods of evaluation can make a positive
contribution to both of these goals. Among the
methods that probably should be considered are
those which analyze loan application records.
Turning to the area of rural development, Yvonne
Levy argues for a new approach to solving the
potential shortfall of water supplies in Southern
California. Most proposed solutions to the problem
have called for an expansion of supplies for pros-
pective water-short areas, primarily the construction of new dams and canals to bring more water
from Northern to Southern California. But Levy
argues for an alternative approach. "If water were
priced higher, final users would have a greater incentive to conserve, the projected demand would be
lower, and some or all of the proposed new water
facilities would not be required."
Levy notes that, in practice, the U.S. Bureau of
Reclamation charged on average about $5 per acrefoot of Central Valley Project (CVP) irrigation
water in 1981. She argues, however, thatthis represented a substantial subsidy to California farmers
because the Bureau's charge should have reflected
costs that would have been incurred by an investorowned utility.
With adjustments made for imputed property
taxes, amortization, and interest cost, the Central
Valley Project would have incurred an average unit
cost of almost $24 per acre-foot of irrigation water
in 1981, calculated on a historical accounting basis.
The calculations would yield a $48 acre-foot charge
if they took into account the replacement cost of the
CVP capital plant. And if efficiency of resource
allocation were the only criterion, the Bureau would
price all irrigation water on the basis of long-run
incremental cost-the cost of delivering an additional acre-foot of water from the next scheduled
block of new capacity. This approach, indeed,
would yield a $324 acre-foot charge for CVP irrigation water.
Levy argues that very high subsidies for Federal
irrigation water have had major consequences.
"The consumption of water and the size of the
Federal irrigation system have expanded beyond the
point where the net return to the last unit of water, in
terms of agricultural revenue, is equal to the cost of
supplying that extra unit. This suggests that more
resources have been devoted to the construction of
the Federal irrigation system in California than are
warranted by agricultural benefits." She calls for
increased emphasis on pricing reform to improve
the efficiency of water usage, through the use of
more efficient irrigation methods and shifts to less
water-intensive crops. "Indeed, Congress logically
should give more attention to the role of the price
mechanism in reducing the projected growth of
irrigation water demand not only in California, but
throughout the West."
5
Barbara A. Bennett*
the past. The virtual revolution in cash management
techniques of the past decade has permitted businesses and households to transact a greater volume
of transactions with a given level of transaction
balances than was true previously. Innovations involving the increased use of automation in bill collection (such as automated lockboxes) and in funds
transfer (such as automatic investment of idle
funds) have increased the rate of turnover of transaction balances and reduced the demand for narrowly defined money relative to income or spending.
Furthermore, the growing acceptance of new highyielding and highly liquid instruments, such as
money market mutual funds and retail repurchase
agreements, has profoundly affected the ways the
public chooses to hold its wealth and accommodate
its transaction needs. Likewise, regulatory and legislative changes aIlowing depository institutions to
pay interest on transaction balances that are held in
ATS (automatic transfer from savings) and NOW
(negotiable order of withdrawal) accounts have led
to substantial shifts of funds into these interestbearing accounts from traditional demand and
savings deposits.
Changes in the public's demand for various types
of financial instruments have altered the meaning of
the monetary aggregates, making observed growth
in these aggregates harder to interpret. The growth
in Ml, in particular, has slowed considerably over
the past few years; yet, with the proliferation of
higher-yielding substitutes for the traditional Mltype transaction instruments, slower observed
growth may not necessarily be associated with a
slowdown in the economy.
The Federal Reserve System has sought ways to
minimize the effects of recent innovations and regulatory changes upon the meaning of the monetary
aggregates and their relationship to economic activity. One part of this effort has centered around the
redefinition of the monetary aggregates announced
in early 1980. By including ATS and NOW ac-
The actions of the Federal Reserve System, as the
nation's central bank, have a
on economic activity and the level of prices. The impact of
these actions, however, cannot be observed apart
from the effect of other influences such as fiscal
policy decisions. As a result, the Fed must rely
on intermediate measures such as the money supply
or interest rates to gauge the effectiveness of its
policy actions. To be useful as an intermediate target, such a measure should, first, be related to
economic activity (GNP) in a stable and predictable
manner, and second, be susceptible to Federal Reserve control.
Since the mid-1970s, the Federal Reserve has
used the money supply as its intermediate target by
setting-and striving to meet-annual growth rate
targets for several monetary aggregates, including
Ml, M2, M3 and a measure of bank credit. Ml
comprises currency, demand deposits, other interest-bearing checkable deposits and travellers'
cheques. M2 includes Ml, savings deposits, smaIl
denomination time certificates, noninstitutional
money market mutual fund shares, overnight repurchase agreements and overnight eurodoIlar deposits
held by U.S. residents. M3 comprises M2, term
repurchase agreements, institutional money market
fund shares and large denomination certificates
of deposit.
Despite this wide array of measures of the money
supply, the public and the Federal Reserve have, in
practice, tended to focus on the narrower aggregates. These measures have the most stable and
predictable relationship to economic activity historicaUy, and they are most closely under the control of
the central bank. Recent changes in U.S. financial
markets, however, may be changing the nature of
that relationship, making the impact of monetary
policy decisions temporarily harder to gauge than in
* Economist, Federal Reserve Bank of San Francisco. Karen Vangsgard provided research assistance.
6
counts in the narrow definition of money and by
acknowledging the potential transaction characteristics of money market funds and overnight RPs and
eurodollars, the redefined aggregates correspond
more closely than did the old measures to the new
ways in which the public has chosen to hold its
transaction balances and liquid assets. This redefinition has helped to reduce the distortion in observed growth rates caused by shifts of funds among
financial instruments which are now viewed as
close substitutes but which, for whatever reasons,
were formerly classified as components of different
aggregates.
The Federal Reserve has also come to place
greater emphasis in its policy deliberations on
broader aggregates whose growth rates and relationships to economic activity are affected less by
shifts of funds among financial instruments. As
recently as 1980, the most prominent measure of the
money supply-then termed MI-A-included only
currency and demand deposits. Today, the policy
focus has shifted to MI-B-which includes ATS
and NOW accounts as well as traditional demand
deposits and currency-and, to a certain extent, to
M2. Reflecting this change of emphasis, MI-A is
now no longer published and MI-B has been renamed Ml. In 1981 the Federal Open Market Committee (FOMC) appeared, at times, to give considerable weight to M2 growth. Some Committee
members argued at that time that at least some of the
surprisingly sluggish growth of (shift-adjusted) Ml
may have been due to the public's ability to reduce
its holdings of traditional transaction balances by
Shifting funds to the higher-yielding substitutes
(such as money market funds) included in M2.'
The Federal Reserve System also has attempted
to cope with the problem of measuring and interpreting money growth by adjusting observed
growth rates (or growth-rate targets) of the aggregates to account for distortions that are caused by
shifts of funds among financial instruments resulting from readily identified changes in law. The
Fed's treatment of the growth in other checkable
deposits that occurred after the nationwide introduction of NOW accounts at the end of 1980 is an
example of this approach. Shifts offunds into NOW
accounts from sources other than demand deposits
"artificially" boosted Ml growth at that time.
To compensate, the Fed began using a "shiftadjusted" measure of M1 in its policy deliberations.
Although until now the Fed has used this approach
to accommodate specific regulatory changes, similar shift-adjustments to account for other types of
financial change may become possible and desirable in the future. For example, the explosive
growth of money market funds in 1981 may have
"artificially" reduced Ml growth, thus increasing
the desirability of a shift-adjustment to account for
that distortion. Likewise, the growing acceptance
of deposit-sweeping arrangements, whereby excess
balances in transaction accounts are automatically
"swept" into money funds, may also distort future
Ml growth and encourage consideration of a shiftadjustment.
This paper analyzes the shift-adjustment technique, both as it has been used to compensate for the
distorting effects of rapid NOW account growth
during a transition period and as it might be used to
compensate for the distorting effects of changes in
the demand for other financial instruments. The
first section presents the rationale for adjusting Ml
to compensate for growth in NOW balances. The
second section examines the alternative approach
of adjusting the growth rate target for Ml. Since
the shift-adjustment technique is essentially a
, 'sources-and-uses" concept, the third section provides an interpretation of the technique from a money demand context. The fourth section describes the
methodology used to calculate the published measure of shift-adjusted Ml, and the fifth section analyzes the sensitivity of that measure to alternative
assumptions about the sources of growth in NOW
balances. The sixth section examines the merits of
using the shift-adjustment technique to compensate
for other changes in financial markets, and develops
a shift-adjusted measure ofMl that would incorporate the impact of growth in money market funds.
The paper concludes with a discussion of potential
uses and limitations of the shift-adjustment technique in the future.
7
I. Rationale for Shift-Adjusted Measure
The nationwide introduction of NOW accounts at
the end of 1980 distorted the observed growth rates
of the narrow monetary aggregates, which were
identified then as Ml-A and Ml-B. (As noted earlier, Ml-A is no longer published and MI-B has
been renamed Ml.) By causing funds to shift to
NOW accounts from both transaction and nontransaction accounts, this development altered the
growth rates ofMI-A and MI-B relative to the rates
that would otherwise have prevailed. As a result,
the growth rates ofthe Fed's yardsticks-M I-A and
Ml-B-were distorted by these shifts of funds. Because the extent of the distortion was not directly
observable, analysts lost some ability to measure
the impact of monetary policy on economic activity.
As early as July 1980, the Federal Reserve acknowledged this problem, when in his monetary policy
report to Congress Chairman Volcker stated that:
The introduction of negotiable order of withdrawal (NOW) accounts on a nationwide basis in January will accelerate the shift from
regular demand deposits into interest-earning
transactions balances, thereby depressing
MI-A growth next year. On the other hand,
MI-B probably will be boosted somewhat
next year by shifts from savings deposits and
other interest-bearing assets into NOW accounts. The range for MI-B thus may have to
accommodate a period of abnormal growth as
the public adjusts to the availability of a new
instrument. 2
The extent to which MI-B was boosted and MI-A
depressed depended on first, the rate of growth in
NOW balances and second, the sources of that
growth. Had NOW balances grown only slightly,
the distortions in growth would likewise have been
minimal. However, the rapid growth in NOW balances that actually occurred during 1981, from
$28.1 billion to $77.0 billion, meant a distortion in
the growth of one or both of the narrow (MI-A and
MI-B) aggregates from these shifts offunds.
Conceivably, this distortion could have occurred
in the growth rate of only one of the two aggregates
-Ml-A, if the growth in NOW balances had come
entirely from demand deposits, or Ml-B, if that
growth had come entirely from savings deposits and
other non-Ml assets. In fact, however, NOW
growth affected both aggregates. MI-A growth was
affected to a greater extent because NOWs have the
transaction features of checking accounts-plus
5\4 percent intereSL However, NOWs were generally offered in connection with high minimum-balance requirements and offered almost the same rates
of return as savings accounts, and thus distorted
Ml-B growth also by inducing depositors to combine checking and savings funds to open accounts.
Faced with potential distortions in the MI-B
growth rate, then, the Federal Open Market Committee announced an annual growth target for that
aggregate which abstacted from any shifts of funds
related to the introduction of NOW accounts on a
nationwide basis. The target growth range was set,
in other words, as if NOW accounts had not been
introduced at the end of 1980. In order to evaluate
MI-B growth against its growth target, then, the
observed growth of this aggregate required an adjustment to account for shifts related to NOW
growth. The adjustment, in essence, involved subtracting from observed growth that proportion of the
growth in NOW balances which carne from transfers of funds from savings accounts and other nontransaction accounts.
II. Adjustments to Growth Rate Targets
introduction of Automatic Transfer from Savings
(ATS) accounts. In 1979, following the late-1978
debut of ATS accounts, the FOMC chose to lower
the target range for old MI, thereby widening the
difference between the midpoints of the annual targets for old Ml and M2-01d Ml comprised only
currency and demand deposits, while old M2 added
small denomination time and savings deposits and
As an alternative to adjusting the actual growth
rate of Ml, the Federal Reserve could have adjusted
the growth rate target for that aggregate in such a
way that observed growth could then be compared
directly to the target. These two approaches are
equivalent, in theory. The Federal Reserve, in fact,
had employed this second approach earlier to account for shifts of funds that were caused by the
8
February 1981 Policy Report to Congress that the
1980 growth targets for MI-A and MI-B should
have been adjusted to take account of these shifts:
"If the FOMC's [target] ranges are adjusted for
current estimates of the actual impact of shifting
into ATS and NOW accounts, the [observed] increases in both narrow aggregates are close to the
upper bounds of the FOMC's ranges for 1980."3
Although the two ways of compensating for distortions in aggregate growth are equivalent in theory, adjustments to the observed growth rates may be
preferable in practice, since less information about
the nature of the shifts of funds is required a priori
than is needed for adjustments to the targets. Also,
given the potentially greater impact on money
growth rates anticipated from the nationwide introduction of NOW accounts, the Fed decided to adjust
the observed growth rates in 1981. Raising the MI
growth rate target at a time when the Federal Reserve was anxious to demonstrate its commitment to
a gradual reduction in money supply growth might
have confused the general public. Furthermore, it
might have been more confusing to change publicly
announced growth rate targets as new information
on NOW sources became available than to change
the shift-adjustment of the levels of the monetary
aggregates.
other checkable deposits. By widening the spread
between the growth rate targets, the FOMC was
able to compensate for the divergence in the observed growth rates caused by shifts of funds from
demand deposits (included in Ml) to ATS accounts
(included in the non-Ml component of old M2).
Likewise, in 1980, the Federal Reserve adjusted
the growth rate targets of the aggregates to accommodate shifts of funds into ATS accounts. At the
beginning of the year, the FOMC set the spread
between the midpoints of the targets ranges for
Ml-A and Ml-B at 1/2 percentage point, on the
assumption that shifts into ATS accounts would
slow over the course of the year. However, with the
passage of the Depository Institutions Deregulation
and Monetary Control Act (March 1980), which
permanently authorized ATS accounts and authorized NOW accounts on a nationwide basis as of
year-end, commercial banks began to promote ATS
accounts more vigorously in order to get a "headstart" on the NOW account competition. (To the
depositor, ATS and NOW accounts are virtually
indistinguishable.) As a result, the growth of ATS
balances accelerated as depositors shifted funds
from checking and interest-earning assets. Hence,
MI-B grew more rapidly than it would have otherwise, while Ml-A grew more slowly. In light of
these shifts, Chairman Volcker stated in the Board's
III. Interpretation of Shift-Adjustment
are caused by the introduction of new instruments
and other legislative and regulatory changes.
The shift-adjustment technique is based on an
analysis of the shifts of funds among financial instruments that apparently arise from shifts in the
money demand function-i.e., from shifts of funds
into (or out of) a given monetary aggregate. Implicit
in this approach is the assumption that the introduction of a new instrument (such as NOWs) does not
change the way the monetary aggregate responds to
changes in income growth or the level of interest
rates. Of course, shifts of funds into a new instrument frequently alter the composition of the aggregate and, possibly, alter the income- and interest
elasticities of the demand for that aggregate as well
as the level of money demanded. To the extent,
then, that these shifts change elasticities, the use of
a shift-adjustment will not fully capture the change
The theoretical concept of a shift-adjustmentwhether to the observed growth rate or to the target
growth rate-has never been particularly welldefined in Federal Reserve publications. It has generally been described as a means of abstracting
from, or compensating for, shifts of funds which
temporarily produce an "abnormal" rate of growth
in a given aggregate. But what is "abnormal?"
Abnormal growth is not merely any deviation from
an aggregate's trend rate of growth, since that deviation may be due to changes in the levels of
interest rates or income which influence the demand
for money. Instead, abnormal growth is growth that
cannot be accounted for once adjustments are made
for such changes in income and interest rates. How,
then, should the shift-adjustment technique be interpreted? Essentially, it has been used to quantify
shifts in the level of the money demand function that
9
in the relation between the aggregate and the economic variables of ultimate concern-income,
prices, employment, etc.
This inability to measure fully the nature of the
shift in money demand may not represent a serious
shortcoming of the shift-adjustment technique in
the short-run, however. Initially, while the public
rearranges its portfolio of financial assets III response to the availability of a new instrument, the
effect on money growth of changes in elasticities
(the slope of the money demand curve) is likely to
be much less pronounced than the effect of changes
in the level of money demanded (the intercept of the
money demand curve). As a result, shift-adjusted
measures may be useful temporary measures of
money growth even if the income- and interest
elasticities of money demand have changed. The
Federal Reserve has, in fact, used shift-adjustments
as merely temporary yardsticks. The Fed stopped
calculating shift-adjusted MI, for example, when
the shifts of funds were apparently over. Once the
public has rearranged its portfolio, of course, shiftadjustments are no longer necessary. This is because either the level of measured money demand
has changed and the aggregate's long-term growth
rate has not been affected, or else the elasticities
have changed and the shift-adjustment is incapable
of capturing the change in the long-term money
growth rate. (In the latter case, the long-term
growth rate target must be changed to reflect the
new relationship between money growth and income and interest rates.)
The shift-adjustment to MI that was associated
with the nationwide introduction of NOW accounts
illustrates the way in which shift-adjustments attempt to cope with shifts in the slope as well as in the
intercept of the money demand curve. Clearly, the
introduction of NOWs increased the demand for
MI, causing the level of the money demand curve to
shift outward. Whether that shift was also characterized by changes in either the incorne- or interestelasticities of money demand depended on the way
that both the suppliers and demanders of NOW-account services responded to their availability. Prior
to the introduction of interest-bearing transaction
accounts, of course, depository institutions could
not pay explicit interest on transaction balances.
Although they were able to evade this restriction by
offering free services to their customers, the yield
on transaction balances tended to be low and unresponsive to changes in market interest rates. By
allowing depository institutions to pay up to 5 1,4
percent explicit interest, then, NOWs (and ATS)
undoubtedly raised the (j,verage yield on transaction
Chart 1
Alternative Assumptions About NOW Accounts'
Effect onthe Demand for Money (M1)
log i
Without
Change in Elasticity
M'
M~ =
M~ =
M~. =
log i
With
Change in Elasticity
M'
M'
10gM
10gM
money supply curve
initial money demand curve
new money demand curve after NOWs
assumed new money demand curve for shift-adjustment
10
balances substantially.
A one-time upward adjustment in yield would
increase the demand for M I, but would not affect
the income- and interest-elasticities. As illustrated
in Panel A of Chart I, the money demand curve has
shifted to the right, parallel to the old demand
curve. In this case, a shift-adjustment to MI would
produce an M I measure that would, in effect, return
the demand curve to its original position, as indicated by the arrow. The Federal Reserve would then be
able to use the old relation between the M I aggregate and income and interest rates to judge whether
its policies were appropriate.
The results are not so straightforward if NOW
accounts did, in fact, change the income- and interest-elasticities of money demand. This situation
may have arisen for two reasons. First, the introduction of NOW accounts changed the composition of
M I by raising the proportion of household deposits
contained in M I relative to the proportion of business deposits. Since the demand for household
deposits may respond somewhat differently to
changes in income and interest rates than would the
demand for business deposits, this change in composition could have changed the slope of the demand for M I. Second, because depository institutions are now permitted to pay up to 51,4 percent on
transaction balances, they have more flexibility
(and possibly more incentive) to vary the yield in
accordance with changes in market rates. Of
course, given current levels of market rates, most
institutions are offering the ceiling rate; however,
they are still able to vary the average yield (and have
done so on occasion) on NOWs by changing other
features such as minimum balance requirements.
The possibility that the yield on transaction balances now varies more than before with changes in
market rates may mean a decline in the elasticity of
the demand for money. Panel B of Chart I depicts
this new, steeper slope. Since the shift-adjustment
technique cannot measure this change in slope,
shift-adjusted MI will treat the new demand curve
as if it were parallel to the old curve at the new
equilibrium point. The lightest colored line and the
circle indicate that the shift-adjustment accurately
measures the impact of monetary policy at only one
point on the new demand curve.
Chart 2
$ Billions
445
Observed vs. Adjusted M1
435
Observed M1
...
430
425
420
j.
Shift-adjusted M1
415
11
To the extent that the introduction of NOW accounts altered the income- and interest-elasticities
of the demand for M I, then, distortions in the measurement of J;TIoney growth could have arisen. However, the post-NOW rearrangement of the public's
portfolio of financial assets apparently happened
very rapidly, thereby temporarily swamping the effects of any changes in demand elasticities that
might also have occurred. Distortions in shiftadjusted Ml, as a result, were probably minimal.
Problems with such distortions might have been
more serious, however, if the public's portfolio
adjustment had not occurred so rapidly and if NOW
accounts had been allowed to yield higher and more
flexible rates-since this clearly would have caused
major changes in the income- or interest-elasticities
of money demand. Similarly, distortions in shiftadjusted measures are likely to be more pronounced
when we consider some of the other, higher-yielding substitutes for traditional transaction accounts,
such as money market funds (Section VI).
IV. Calculation of Shift-Adjusted M1
The calculated NOW-account distortion in the
Ml growth rate was quite substantial during 1981,
particularly during the first four months of the year,
when the difference between the actual and shiftadjusted measures widened by several billion dollars each consecutive month. (See Chart 2 and Table
1). The spread widened at a much slower pace in the
latter half of the year, however, which suggests
a weakening of the shifts of funds creating such
a divergence.
As noted earlier, the calculation of shift-adjusted
MI involves a subtraction from the observed Ml
level of the increment in ATS and NOW balances,
or "other checkable deposits" (OCD), which originally came from accounts not included in Ml.
Though conceptually simple, the arithmetic involved in the Federal Reserve's calculation of shiftadjusted Ml was actually quite complicated because of seasonal adjustment factors. The basic
calculation can be described somewhat more simply
(Table 2). As a first step, the Fed estimates the level
to which OCD balances would have grown assuming normal, trend growth in such accounts, but
without the nationwide introduction of NOWs. According to these estimates, the trend growth rate
was $200 million per month during the first half of
1981 and $300 million per month in the latter half of
the year. Growth in NOW balances in the Northeast, where NOWs had been available prior to 1981,
would be included in this trend since, presumably,
such growth was not affected by the change in the
law. Likewise, some of the growth in ATS balances
was included in trend growth since ATS accounts
had been introduced by commercial banks in 1978.
(Still, the ATS trend was harder to measure because
banks began promoting such accounts more aggressively in late 1980 and early 1981 as an alternative to
the newly authorized NOW accounts.)
The above-trend growth in OCD-the growth
attributable to the introduction of NOWs-then
could be obtained by subtracting the trend level
from the observed growth of OCD. The amount of
growth in OCD balances that can be attributed to
Table 1
Monthly levels of M1 and Shift-Adjusted M1
(Billions of Dollars, Seasonally Adjusted)
Month
Jan.
Feb.
Mar.
April
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
M1
Shift-Adjusted M1
Difference
Change in ATSINOW Balances
417.9
419.4
424.4
433.3
429.2
428.4
429.4
431.1
431.2
432.9
436.4
440.9
414.4
413.4
416.8
423.6
420.1
418.8
419.5
420.9
420.7
422.2
425.0
428.7
3.5
6.0
7.6
9.7
9.1
9.6
9.9
15.1
10.2
10.5
10.7
11.4
12.2
12
10.1
6.2
7.0
-1.3
1.5
1.4
1.4
1.7
0.4
3.1
2.3
Table 2
Calculation of Shift-Adjusted M1
(Biilliclns of Dollars, Seasonally Adjusted)
Calculation
I) Observed level of OCD
2) Less trend level of OCD
3) Yields above-trend growth in OCD (cumulative)
4) Less previous month's above-trend level
5) Yields current month's above-trend growth
6) Times fraction of growth associated with shifts from nontransaction sources
7) Yields OCD from nontransaction sources
8) Cumulative nontransaction OCD
9) Observed level ofMI
10) Less cumulative nontransaction OCD (from line 8)
II) Yields shift-adjusted M I
January 1981
February 1981
43.2
53.3
0.0
14.9
14.9
9.9
3.4
417.9
6.1
419.4
Figures in this table may not agree precisely with lable I figures, due to slight differences in the calculation of seasonally adjusted totals.
come from
nontransaction sources.
shift-adjusted MI could be calculated by adding its individual components: currency, demand deposits,
travellers' cheques, cumulative trend OCD and
cumulative OCD estimated to have come from
demand delJOsits.
shifts of funds from savings and other non-M I deposits, then, is obtained by multiplying the abovetrend growth in OCD by the fraction of that growth
assumed to come from nontransaction sourcesestimated at .225 during January and .275 in subsequent months. Shift-adjusted MI is obtained,
finally, by subtracting from observed M I the cumu-
V.lmportance of Assumptions
These calculations reveal that the magnitude of
the shift-adjustment depends on several factorsthe growth in OCD balances outstanding, the assumed trend growth in those balances, the proportion of the above-trend growth attributable to shifts
from non-Ml sources and, finally, the duration of
the stock adjustment process. Only one factor-the
overall growth in OCD balances-actually can be
observed, and the other three factors can only be
estimated from indirect evidence. As a result, the
shift-adjusted measure of MI could be subject to
error, possibly providing the FOMC with misleadof its
about the
Estimates of the trend growth in ATS/NOW
ances and the proportion
associated with shifts from
and other nontransaction sources were based on econometric evidence and on surveys of depository institutions and
households. The econometric evidence came from
several
which .• related the !llcmtlhly
changes in NOW balances at individual banks to
reported changes in other deposit categories. The
underlying hypothesis was that, with the diversion
of funds to NOWs from other deposit categories
(either within each institution or intraregionaHy),
institutions with substantial growth in NOW balances would
a smaller increase in other deposit categories than would institutions with smaller
NOW increases. In each regression, the change in
NOW balances was the independent variable, while
the changes in the other deposit
demand, personal savings, nonpersonal savings,
personal time and nonpersonal time depositswere each treated in separate
as dependent variables. Each equation took the form:
t:::.DC/Z = -!3,(t:::.N/Z)+!3 2 +u, where
t:::.DC =
in
balance of each
specific deposit category
t:::.N =
in NOW balances;
Z
!32 = constant representing effects of "other
factors' '; and
u = error term.
statistica:Hy significaJlll coefficient on the NOW
account variable in each equation could be interpreted loosely as the proportion of NOW growth
coming from each deposit
However, only
the regressions for demand deposits and personal
13
from various types of assets.
Like the regression results, these surveys had a
number of weaknesses. The first was simply the
quality of the responses, particularly from the banks
-many of the respondents may have provided only
rough estimates. Furthermore, the original source
of each NOW deposit may not have been the account from which funds were actually transferred,
because this transaction would represent only one in
a chain of related portfolio-balance transactions.
And the surveys, by addressing only the new accounts, would not have captured any shifts to existing accounts. However, the agreement between survey and regression results lends added credence to
the estimates.
Nonetheless, small errors in estimating the
sources of NOW account growth could have significantly distorted the shift-adjusted measure of MI.
Table 3 shows the sensitivity of this measure to
alternative assumptions about the proportion of
NOW growth attributable to one-time shifts from
savings.
The growth rate of the shift-adjusted measure of
Ml varies widely, depending on the estimate of the
proportion of OCD growth associated with shifts of
funds from savings and other non-transaction accounts. If 17 percent of the above-trend growth in
OCD balances had come from nontransaction
sources, instead of 27.5 percent as assumed, "effective" Ml in 1981 would have grown 4.2 percent
instead of 3.2 percent-in the lower half of the
target range instead of below the lower bound of the
range. (See Chart 3). The disparity in the estimated
effective growth rates was even larger during the
early months of the year because of the rapid growth
in OCD balances at that time.
savings deposits yielded statistically significant
results-approximately 80 percent of NOW growth
in January outside of the Northeast came from
demand deposits, while nearly 20 percent of that
growth came from savings deposits (adjusted for the
relative market shares of banks and savings and loan
associations) .
This approach has some advantages for estimating the proportion of NOW growth attributable to
shifts from savings. By using a cross section of
institutions and a cross section of regions, we need
not specifically include general influences on deposit levels (such as interest rates and economic
activity) in the regressions, since most of those
influences presumably do not vary across institutions. However, to the extent that any omitted variables are correlated with specific variables, the
regression results may be biased. For example, institutions that report large changes in demand and
NOW balances may also have more "sophisticated" depositors who behave differently from depositors in general. These depositors are likely to
shift checking balances into NOWs and at the same
time continue to shift passbook savings balances
into higher-yielding instruments. Because these
shifts of savings will appear to be correlated with
NOW growth, the savings deposit regression will
tend to overstate the NOW effect.
In addition to regression data, the Federal Reserve obtained estimates of the sources of NOW
growth from a number of surveys of depository
institutions and households. Depository institutions
were asked to provide data on the percentages of the
total inflows to their new NOW accounts that came
from their own checking and other types of accounts. Likewise, consumers were asked what proportions of their NOW balances were transferred
Table 3
Shift-Adjusted M1 Under Alternative Shift Assumptions
(Billions of Dollars, Seasonally Adjusted)
January
Shift Assumption
(Percent from savings)
o
17
22.5 (Jan.)
27.5 (afterJan.)
~
Level
Annual
Rate"
$417.9
415.4
6.6%
-0.6
414.5
4m5
*Growth since December 1980 at simple annual rate
-3.2
-14.7
February
Annual
Rate"
Level
December
Annual
Level
Rate"
$419.4
415.2
$440.9
433.1
413.4
407.0
14
5.5%
-0.6
-3.2
-24.8
428.7
417.9
6.1%
4.2
3.2
0.6
Another source of possible distortion is the assumption regarding the duration of the stock adjustment process. Instead of extending throughout 1981
as assumed, the process may have been substantially over by June-as Table 1 possibly suggests.
Thus, if the Fed had stopped adjusting M1 at the end
of May, the calculated growth rate for the full year
would have been nearly 4 percent, instead of 3.2
percent.
Given the sensitivity of the shift-adjusted measure of MI to alternative assumptions about the
sources of OCD growth and the duration of the
stock adjustment process, the success of the Federal
Reserve's efforts to obtain a truly reliable measure
of effective money growth cannot be determined
with any degree of precision. Depending on the
assumptions one uses in calculating shift-adjusted
M1, monetary policy in 1981 could be viewed as
having been either moderately expansionary or
fairly contractionary. Clearly, then, the Fed's measurement problems did not disappear once shiftadjusted M1 had been calculated.
VI. Adjustments for Other Shifts
it is the sweeping changes in financial markets that
create the pressure for legislative and regulatory
remedies. For example, the pressure for payment of
interest on transaction balances and for deregulation
of deposit rate ceilings generally would be less
pronounced if financial markets were unable to offer depositors nondeposit alternatives.
Money market mutual funds, deposit-sweeping
arrangements, retail repurchase agreements, and
loophole accounts-all changes occurring largely
Although the Fed has used shift-adjustments primarily to account for shifts of funds engendered by
regulatory and legislative changes such as the introduction of ATS accounts (1978) and NOW accounts
(late 1980), the rationale for using a shift-adjusted
measure of money growth applies equally well to
shifts of funds that are related to other types of
financial change. In fact, shift-adjustments for regulatory changes probably address just a small part of
the whole money measurement problem. After all,
Chart 3
Adjusted M1 with Alternative Shift Assumptions
$ Billions
445
440
435
Observed M1
"
430
425
420
~
Shm-adjus~
Ml
(17% shift)
415
(Published)
4101..-....._ ....._ ..........._ ....._ .........._ ....._ .....--1........._
1981
1980
15
.....--1.........
with maturities as short as one day or as long as 89
days. In order to compete with MMFs, banks and
savings and loan associations frequently offer retail
RPs in connection with a checking or NOW account, permitting the depositor to order transfers of
funds between the two accounts by telephone.
Loophole accounts also permit depositors to earn
market interest rates on funds that might otherwise
have been held in a transaction account. They offer
a line of credit which can be drawn on by check, in
connection with instruments such as the six-month
money market
and thus grant delJositolrs
access to funds before the stated maturity date.
The current proliferation of high-yielding shortterm instruments with low minimum investment
requirements may be reducing the demand for Mltype balances. In the past, small savers did not have
many alternatives to low-yielding accounts, even
periods of high interest rates, and thus had
little incentive to reduce holdings of Ml-type balances. Now, however, they do have such an incentive. First, the public may be able to use these new
instruments-MMFs and loophole accounts especially-as substitutes for demand deposits and
other checkable deposits. Second, these new instruments yield market rates far in excess of ATS and
NOW rates, and thus induce depositors to limit their
outside the regulatory framework-may have contributed significantly to the current problem of interpreting money growth. Money market mutual
funds (MMFs), which are included in the M2 aggregate, are technically open-end short-term investment pools. They invest in a variety of highly liquid
money market instruments such as Treasury bills,
large negotiable certificates of deposit and commercial paper. However, minimum initial investment
requirements are generally low ($1,000 to $2,500),
and shareholders may write checks against their
accounts
transfer funds to third
wire, so that MMFs may be viewed as partial substitutes for the transaction accounts included in MI.
Deposit-sweeping arrangements also may create
money measurement problems, since they permit
depositors to keep their transaction account balances to a minimum, while automatically transferring idle funds to a highly liquid and higher yielding
instrument (usually MMFs).
_
With retail repurchase agreements and loophole
accounts, banks and thrift institutions have created
alternatives to traditional transaction deposits as
well. Retail repurchase agreements (RPs) are essentially short-term investments in denominations of
less than $1 OO,OOO-minimum investment requirements are usually in the $1,000 to $2,500 range-
Chart 5
Chart 4
Yield Spread and Money Markel Fund Changes
$ Billions
M
Noninstitutional Money Market Fund Assets
Percentage Points
5
150
12
10
4
1980
50
35
v:
J
F
2
!
!
I
!
I
I
I
M
A
M
A
S
0
N
I
D
-2
1980
16
1981
Table 4
Shift-Adjusted M1 Under Alternative
MMF-Shift Assumptions
(Billions of Dollars)
holdings of all types of transaction balances. As a
result, Ml growth may give misleading policy signals, particularly during the stock adjustment
period when the growth in these new instruments is
very rapid. Shift-adjustments to compensate for
these shifts out ofMl thus may be appropriate.
The phenomenal growth in money market funds
(from $75.8 billion to $184.5 billion in 1981) in
particular may reflect this adjustment in the public's
stock of assets. Of course, one could argue that this
growth was only a normal response to high interest
rates, which induced the
to shift funds into
MMFs (as one of several options) from lower-yielding assets. The 1980-81 experience does not support this argument, however. Although interest-rate
patterns were similar in both years, MMF balances
grew much more rapidly in 1981 than in 1980.
(Chart 4). Furthermore, the response to changes in
the spread between MMF yields and the six-month
T-bill rate was much more dramatic in 1981 than in
1980 (Chart 5).
Instead, some sort of stock adjustment apparently
increased the MMF growth rate substantially. Some
Shift Assumption
Year-End level
of M1" (adjusted)
$ 0
$428.7
3.2%
12
440.7
6.0
18
446.7
7.5
*This measure of M 1 also incorporates the Federal Reserve
estimate of the NOW shift-adjustmenl.
of this growth may have come from stocks of
demand and other transaction balances, causing
observed Ml to grow more slowly (after the NOWaccount adjustment) than if MMFs had not been
available. In fact, according to one source, MMFs
reduced Ml demand by more than $12 billion by
September 1981. 4 Therefore, just as the Fed adjusted Ml growth downward to account for NOWinduced shifts offunds from interest-earning assets,
so it should have made an upward adjustment to
account for MMF-induced shifts of funds out of
MI.
Chart 6
M1 Observed and Adjusted for MMF Shifts
$ Billions
445
Shift-adjusted ($18 billion MMF shift)
....
440
435
Observed M1
....
430
425
420
~
Shift-adjusted ($12 billion MMF shift)
415
410
I..-........_
........_
.............._
........_
Annual Growth
Rate Over 1981
..............""'-......_
1980
......-""............._
1981
17
......-""..........Il
resulting from MMF growth, the adjusted Ml measure would have grown at a 6.0-percent annual rate,
even after the NOW shift adjustment. But the $12billion shift assumption referred only to the first
nine months of 1981, and moreover, ,simulation
results showed the MMF impact increasing steadily
over that period. 5 Therefore, we could assume up to
an $18 billion shift from transaction accounts into
MMFs. In that case, shift-adjusted Ml would have
grown at a 7.5 percent rate-more than the 6. 1percent growth in observed M 1, offsetting the
downward NOW adjustment (see Chart 6). Relative
to its target range, then, MI growth may actually
have been somewhat expansionary in 1981. However, this result seems contrary to the sluggish economic growth observed in 1981. Furthermore, this
apparent inconsistency illustrates the difficulties
inherent in measuring demand shifts when the instrument involved, unlike a NOW account, pays
market interest rates.
Various measures of MI can be calculated,
depending on whatever assumption is made about
the percentage of noninstitutional MMFs (coming
from a reduction in Ml-type balances (see Table 4).
Incidentally, we exclude institution-only MMFs because such funds are generally regarded as substitutes for direct money-market investments. For this
reason, institution-only MMFs are included only in
M3, while other MMFs are included in M2. Shareholders of institution-only funds generally have
other options for earning market rates on transaction
balances, and these funds thus may not attract balances held for transaction purposes. Noninstitutional MMFs, by contrast, tend to be close substitutes for small-denomination deposit instruments
and, because of the lack of options their shareholders have for earning market rates, such MMFs
are more likely to attract demand and other transaction balances.
With a $12-billion reduction in Ml-type balances
VII. Conclusion
We have not seen the last of the sweeping
changes recently taking place in the U.S. financial
system. Money market funds continue to grow
rapidly. Increasing numbers of brokerage firms and
depository institutions are announcing depositsweeping services, while larger numbers of banks
and thrift institutions are offering retail repurchase
agreements and loophole accounts. The pressure to
deregulate deposit interest rates continues to mount.
Furthermore, the Depository Institutions Deregulation Committee (DIDC) has created a new market
rate 91-day account and is currently considering the
creation of other, more liquid accounts to permit
depository institutions to compete more effectively
withMMFs.
Thus, observed Ml growth may continue to give
somewhat misleading pplicy signals. To the extent
that distortions in Ml growth can be traced specifically to the growth in certain financial instruments, shift-adjustments may be useful. Many of
these changes, however, cannot be quantified with
even the same degree of certainty as the NOWaccount shift. We have insufficient data to make
shift-adjustments for certain innovations, such as
deposit-sweeping arrangements. Furthermore,
many of these new instruments pay market rates
(unlike NOW accounts), so that shifts of funds
become harder to classify, either as shifts in the
demand for money or as changes in the quantity of
money demanded due to interest rate changes. As a
result, the ability of shift-adjustments to compensate for these changes and to reduce uncertainty
about the effective growth rate of money may be
somewhat limited, compared to what could be
achieved with the NOW shift-adjustment.
FOOTNOTES
1. See, for example, the minutes of the August 18, October
5-6 and November 17, 1981, meetings of the Federal Open
Market Committee.
2. Federal Reserve Bulletin, July 1980, p. 535.
3. Federal Reserve Bulletin, March 1981, p.199
4. Michael Dotsey, Steven Englander and John C. Partlan,
"Money Market Mutual Funds and Monetary Control;' Federal Reserve Bank of New York Quarterly Review, Winter 1981-82, Volume 6 No.4, pp. 9-17.
5. Ibid., p. 17
18
.
rClng
I
ng Policy Under
Reinvestment
Alane K. Sullivan and Randall J. Pozdena*
On October 12, 1977, President Carter signed into
law the Community Reinvestment Act (CRA) as
Title VIII of the Housing and Community Development Act of 1977. The act was designed to encourage financial institutions "to help meet the credit
needs of the local communities in which they are
chartered." To meet that intent, the CRA directs
each federal financial supervisory agency to take
into account an institution's CRA record when ruling on branch, merger, or other applications.
The affirmative orientation of the CRA represents a significant departure from earlier bank regulation, which had been designed primarily to ensure
the safety and soundness of the banking system.
Regulators examine banks' financial structure and
portfolio quality, for example, to monitor their
overall soundness and thereby to minimize the incidence of bank failure and the disruptions to financial markets that might ensue. Similarly, they regulate competitive structure in banking marketsthrough chartering, branching, and merger regulation-presumably with the intention of preserving
vigorous rivalry without promoting" overbanking"
of individual markets.
Bankers and economists may not all agree that
such regulation is necessary (or even desirable) to
achieve the goal of a stable banking system. None-
theless, such regulation does not usually call into
question the basic ability of a competitive banking
market to make socially appropriate allocative decisions. The passage of the CRA, on the other hand,
indicates that Congress questioned the ability of the
market to produce desirable patterns of credit use.
Moreover, by linking the CRA to the regulatory
approval of merger and other applications, Congress has made the future development of banking
markets contingent on current patterns of credit
service to the community.
This paper traces the origins of the Community
Reinvestment Act and examines its aims and the
extent to which those aims are being met by the
current enforcement process. Section I sets forth the
legislative history of the CRA. Sections II and III
describe the law in more detail with specific focus
on its enforcement. Section IV examines the problem of detecting noncompliance with the antiredlining provisions of the CRA, with special attention to the agencies' evaluation methodologies and
the community group and academic studies of the
"redlining" phenomenon. Section V presents our
conclusions and discusses the policy implications of
a possible alternative evaluation method to those
currently used to enforce the Act.
I. Legislative History and Intent of the eRA
The CRA had its origins in long-standing allegations by community groups that financial institutions discriminate against certain neighborhoods in
credit decisions. The practice, called neighborhood
"redlining", allegedly contributes to and even
causes the decline of inner city neighborhoods.
Anti-discrimination and anti-redlining legislation was already in place at the time the CRA was
formulated, but community groups saw this earlier
legislation as ineffective in structure and application. The Home Mortgage Disclosure Act (1975),
for example, required financial institutions to disclose data on the volume of mortgage loans ona
census tract or zip code basis. Such disclosure of
geographic lending patterns was intended to provide an overt mechanism for detecting redliningbut provided no mechanism for imposing govern-
* Sullivan is a Research Coordinator, and Pozdena
is an Economist, with the Federal Reserve Bank of
San Francisco.
19
mental sanctions should such behavior be detected.
Similarly, the Equal Credit Opportunity Act (1974)
prohibited discrimination in credit transactions on
the basis of race, color, religion, national origin,
sex, marital status, or age, but did not address
problem of "geographic discrimination" im'ol1/ed
in redlining. Dissatisfaction with the thrust
plementation of
legislation led citizens'
groups to
and
pressure on
Congr'ess and the
agencies.
The response was an
bill (Senate
bill 406), which would have
the federal
regulatory agencies to encourage financial institutions to "help meet the credit needs of the local
communities."
banks already were required to serve the "convenience and needs" of
their
I the sponsors of the bill felt that
"convenience and needs" had focused traditionally
on the provision
facilities. The passage of
the
would ensure that, in
the "convenience and needs" consideration also included
credit services. The bill's proponents
felt
the "semiexclusive franchise" that government granted financial institutions obligated those
institutions to pursue "community" as well as private entrepreneurial goals. More specifically, the
draft bill emphasized that a financial institution's
first obligation was to the credit needs of its "primary
service area"---defined as an area
from which 50 percent of the institution's deposits
were derived. Thus institutions could not "export"
credit from the community from which deposits
were drawn without
first to that area's
credit demands.
saw in these
an unrealistic view
of the role of financial institutions and a challenge to
the traditional market mechanism of allocating
credit. As a
modified the initial bill
the very spesutlst,mtiially, removing, for
cific focus on the' 'primary
service area"
and
' undefined. In ad(iitl,on,
it deleted reporting requirements and inserted a prohibition against the imposition of any additional
administration burdens on affected financial institutions. Furthermore, the bill's sponsors argued repe,lte1jly in committee discussions that the bill was
not an
to allocate credit.
the final bill which became the Community
Reinvestment Act avoided the direct condemnation
of' 'exportation" of credit, because Congress clearly wished to avoid anocating credit or doing anythat
sacrifice the safety
and soundness of the banking system. At the same
time, however, the law retained the idea of "serving
the needs of the community," with an emphasis on
low- and moderate-income neighborhoods. The
reconciliation of these two potentially contradictory
aims was left to the regulatory agencies. 2
II. Regulatory Implementation
Congress gave the financial regulatory agencies
the task of drafting regulations which both reflected
Congressional intent and provided
compliance
for financial institutions. The agencies involved-the Federal Reserve Board of Governors, the Federal
Insurance Corporation
(FDIC), the Federal Home Loan Bank Board
of
(FHLBB), and the
(OCC)-completed the assigned task by October
1978. The following discussion covers the Federal
Reserve's Regulation BB, which is the same in all
res'pects as the
promulgated by
the other agencies.
Regulation BB reflects the apparent Congressional intent that the concept of "community"
errm!l)ve:d in CRA regulations be flexible enough to
accommodate the myriad markets and service needs
which banks confront. Regulation BB, in fact,
leaves the definition of geographic market and service orientation up to individual banks. The banks
must prepare a Community Reinvestment Statement and make it
available to the public.
The statement includes: 1) a clear definition of
market area, without
exclusions of lowand moderate-income areas; 2) a list of the types of
credit services offered; and 3) a copy of the public
eRA notice, a description of consumers' rights
under the CRA. In
each bank must maintain a file of aU comments received with regard to its
community lending practices and must include in
the file its replies to complaints and comments.
The evaluation process is based not only on com-
20
dis1cnrnmato:ry or other
pliance with these procedural requirements, but
also on a bank's behavioral compliance-whether
the bank's actual lending activity meets the "credit
needs of the community."
RegUlation BB
provides only general guidance for compliance, by
twelve factors the Federal Reserve will consider when
its CRA evaluation. These
include:
a. Activities conducted to ascertain a cOlmumnlity's
credit needs, induding the
efforts to communicate with community members
the credit services it orc>vic!es:
b. Extent of the bank's marketing and
credit-related programs to make community
members aware of the credit services it offers;
c. Extent of participation by the board of directors
in fonmulating bank policies and reviewing its
perfonmance with respect to CRA purposes;
d. Any practices intended to disCOllra!se alDDtllcations for types of credit set forth in the bank's
CRA statement(s);
e. Geographic distribution of the bank's credit exten:siorls, credit
and credit dellials;
sef1tice:s. includh. Partllcipation,
in local
and redevelopment
prcljects or programs;
i. Origination of residential
housloans, home
small business or small farm loans
within the bank's
or the
of
such loans
in its cmnmlunlty;
j.
in government-insured, guaranor subsidized loan programs for housing,
small businesses, or ~,m§lll
k. Ability to meet various community credit needs,
based on the bank's financial condition and
impeclimlen:ts, local economic conditions,
and other
l. Other factors
in the Board's
reasonably bear upon the extent to which a bank is
to meet the entire
credit
needs.
Legal Aspects of the eRA
into question if it has the effect of dl~icnmlnal:mg
against a certain neighborhood. In order to continue
such a practice, the bank must show that it is necessary to the business and that
nondiscrimiprElCtllce cannot be substituted. 4
Denial of merger or branch applications is the
most severe penalty imposed for noncompliance
with the CRA. However, the Federal Reserve (and
the other agencies) have the authority to condition
approval of an application on changes in the applicant's mode of
business. The Act aplJan;:ntly
is influential
somewhat vague in content,
since financial institutions
often
reach agreements independently. Most interested
agree that an accurate,
measurement method would add a great deal to the current
understanding and enforcement of the CRA.
21
IV. Problem of Detecting Noncompliance
The original legislation provided very little guidance for detecting noncompliance. The regulations
formulated by the regulatory agencies set forth general guidelines for' assessing lenders' behavior.
However, the agencies still had to devise an evaluation method which would yield an accurate detection of undesirable behavior, as is described below.
Regulatory Process
The regulatory agencies are involved in CRA
enforcement on two levels. First, the regular examination process involves routine evaluations ofCRA
compliance. Secondly, as the law states, the agencies must evaluate CRA performance every time a
financial institution applies to branch, merge, or
otherwise expand its operations. In 1980, for example, the Federal Reserve processed 917 applications
with CRA implications. Often, in these cases, the
CRA record is determined by studying the bank's
most recent examination report. These analyses are
expanded, however, when a protest arises.
Examiners conduct a CRA compliance exam as
one part of the overall examination which they
regularly make at financial institutions. The twelve
items listed in Regulation BB serve as a guide to the
examiner in determining whether the institution is
complying with CRA procedural requirements. In
addition, the examiner must study the bank's lending record and its public relations policy as well as
many other factors to determine the degree of behavioral compliance. From the observations made
and from contact with community groups, the
examiner then makes a final judgment regarding the
institution's overall record. A rating of 1 or 2, on a
scale of I to 5, means that the institution's CRA
record is above average, while a 3 represents a
"less than satisfactory" record. Standardized examination procedures include a weighting scheme
designed to cover all twelve assessment factors of
Reg BB, but examiners still have some latitude in
assessing performance. The agencies, therefore,
admittedly rely on the subjective analysis of experienced staff members.
The regulatory agencies have developed a joint
evaluation handbook as well as rigorous training
programs, but many observers remain uneasy about
the regulatory methods of evaluation and enforcement. 5 Even the examination handbook acknowl-
edges the lack of a standardized evaluation technique, when it notes that' 'the examiner is expected
to adjust the CRA procedures on a case-by-case
basis to accommodate institutions that vary in size,
expertise, and locale."6 In fact, a financial institution can attempt to serve its community's needs in
myriad ways, and somehow the examiner must determine whether the institution's effort is adequate.
Resolution of a protest also involves evaluation
of CRA compliance. The protest process begins
when a group claims that an institution has failed to
serve a community'S credit needs. The group then
submits a protest to the appropriate regulatory agency asking that the institution's application to expand
be delayed until after examination of its CRA record. When the Federal Reserve is involved, the
Board first determines whether the protest is nonsubstantive on its face or whether it warrants a
detailed investigation-and in the latter case, it
conducts a thorough analysis of the bank's CRA
performance.
In practice, the Federal Reserve first attempts to
create a constructive dialogue between the protesters and the bank to clarify the issues. Often a case is
then dismissed due to a prior misunderstanding of
the law or because of poor communication. At other
times, the two parties reach an agreement on their
own and the group drops the protest. However,
sometimes a thorough investigation is necessary,
and in such cases, the Board may hold a public
meeting where both sides may present their views.
The Federal Reserve's analysis entails the evaluation of the statements of the two parties and some
investigative research. The agency studies the
bank's marketing programs, along with other factors which may reflect the affirmative action it has
taken to serve community credit needs. To detect
whether actual lending behavior is in compliance
with the CRA, it also examines data available as a
22
result of the Home Mortgage Disclosure Act
(HMDA data), often along with real estate transfer
records and bank loan applications to account (at
least partially) for loan demand. Other relevant data
include information on neighborhood characteristics, compiled from such information sources as
U.S. Census data and city planning records. Using
all of these sources, the Board's Research staff
recommends whether the protest is substantiveand whether the bank's application to expand its
operations should be approved, approved subject to
certain conditions, or denied. The Board's decision
is based on some objective analysis coupled with a
subjective judgment of the bank's behavior and
management attitudes. Precedents are set on a case
by case basis.
One possible way of handling a protest case is
"conditioned approval," whereby the application
is approved subject to certain requirements. For
example, the Federal Home Loan Bank Board approved the application of Midwest Federal Savings
and Loan Association, Minot, North Dakota, to
establish a new branch-provided that the Association change its delineation of its local community
and withdraw its policy of refusing to make mobile
home loans on American Indian reservations unless
the policy could be shown to have a firm economic
basis.
On other occasions, a bank and a protesting
group have privately agreed on conditions, leading
the community group to drop its charges. Landmark
Bancshares Corporation of Clayton, Missouri, for
example, upon protest of its application to acquire
Ladue Bank and Trust Company, made an agreement with the Missouri Association of Community
Organizations for Reform Now (ACORN). The
agreement, which led ACORN to drop charges,
included a commitment of $1 million for home
improvement loans and mortgage loans to the
Wellston, Missouri community at below market
rates. Clearly, in cases such as these, conditioned
approval and private agreement raise concerns
about credit allocation, an activity not intended by
Congress. The Federal Reserve Bank of St. Louis,
reflecting the Federal Reserve System's opinion,
held, in the Landmark Bancshares case, that' 'since
the Board of Governors has stated that neither the
Bank Holding Company Act nor the Community
Reinvestment Act requires that the Board impose
commitments to allocate credit, the Reserve Bank
does not endorse any term of the agreement between
applicant and protestant which may have such a
result." 7
As these examples show, the costs involved in
lodging a protest are usually relatively low. On the
other hand, the protest process can be costly to the
institution involved. First, conditioned approval or
private agreements can result in significant financial commitments. Second, the application to
expand must be delayed until the CRA issue is
resolved. The bank must not only pay the costs
incurred in public meetings (legal fees, etc.) but
also those resulting from substantial delays in expansion plans, including the costs of affected competitive positions. In addition, protesters need not
be community groups, but can even include competing banks, which sometimes file CRA protests,
presumably with the hope of delaying or preventing
competitors' expansion plans.
In view of the high costs of an inaccurate decision
to all parties involved, the regulatory agencies
should attempt to devise the best possible methods
of detecting CRA violations and to encourage the
use of the best methods by protesters. In essence,
regulators have relied on a two-part approach. First,
regulators focus considerable attention on what
could be called affirmative marketing efforts. In
this regard, the law is designed to ensure that the
demand for loans is not adversely affected by a lack
of knowledge about availability. Since marketing
efforts such as advertising, community meetings
and discussions with realtors can enhance the flow
of information between potential loan applicants
and lenders, the monitoring of such efforts under
the CRA probably improves the efficiency of the
marketplace. It is probably impossible to measure
the optimum level of market information, so it is
reasonable to use only general criteria to formjudgmentson a bank's performance in this area.
A second important part of the CRA enforcement
process involves the examination of actual lending
activity to determine evidence (if any) of discrimination. Here, detailed objeCtive analysis is desirable, despite the difficulty of developing a good
evaluation method for detecting noncompliance.
23
The remainder of the paper, therefore, addresses the
problems associated with those CRA enforcement
procedures which focus on possible discriminatory
lending patterns, or redlining.
Definition of Redlining
Part of the trouble in this area stems from the lack
of a generally accepted definition of redlining.
Before choosing a method for detecting violations,
it is essential to decide on a correct legal definition
of redlining and determine what type of behavior
would be deemed unacceptable. From the begindifferences of opinion arose over the intent of
the CRA, and these differences naturally carried
over into the debate over the definition of redlining.
Community groups and other CRA proponents,
being concerned about urban "disinvestment,"
criticized as redlining any lending behavior resulting in an uneven distribution of loans across neighborhoods, regardless of the reason for this pattern.
In their view, lending policies that create uneven
distributions of mortgage credit have the effect of
discriminating. Many also argued that banks have
an obligation to make every effort to serve their
communities, even if this means lower profits than
could be earned elsewhere. By refusing to lend in a
neighborhood, for whatever reason, community
groups claim banks otherwise would create an externality effect: deterioration of the community.
Under the community groups' definition, successful CRA enforcement would mean a more equal
distribution of loans across neighborhoods. Evaluation methods devised under this approach simply
involve the examination of loan distribution patterns for inequalities, as discussed below. However, in its final form, the CRA falls far short of
mandating credit allocation or affirmative urban
renewal efforts if they are unprofitable. Instead, the
law seems to recognize that there may be sound
business reasons for an uneven pattern of loanspartly reflecting differences in demand across
neighborhoods, but also lenders' recognition of
higher risks in certain areas. In the economic literature, this type of lender behavior is referred to as
"rational" redlining. Lenders who operate efficiently will make loans to minimize risk and maximize profit, subject to regulations regarding the
overall quality of loan portfolios. We assume here
24
that the law was not meant to outlaw rational redlining-but rather irrational practices whereby a
lender avoids lending in a certain area despite the
fact that the activity would yield a normal balance
between risk and return. Under irrational redlining a
property's location remains a significant explanatory factor for a
even after
adjustment for all the factors which might explain
the pattern on the basis of sound business judgment.
This does not mean that a lender who uses a
property's location as a decision criterion necessarily has discriminatory (irrational) intent. To make
economically sensible decisions, lenders must use
the available information to evaluate individual requests for funds. At times, the cost of obtaining this
information is prohibitively high, so that lenders
may attempt to economize on information costs by
using proxies for certain variables. If these proxies
have statistically significant results, the quality of
the lending decision is probably enhanced. However, the law prohibits the use of certain variables
(such as zip code), assuming that their use would
have discriminatory effects. This practice may be
rational in a purely private decision-making process, but since the law outlaws it in a social context,
we must include such variables in our definition of
irrational redlining.
The use of the CRA's anti-redlining provisions to
address irrational redlining gives us the basis for
choosing an appropriate evaluation method. Detection of irrational redlining requires an understanding of the factors necessary to make a sound business decision. We will narrow our focus to the
factors that affect risk and return in the mortgage
market, since it is the behavior of lenders in this
market that has drawn
most criticism from
proponents. After discussing these factors, we will
examine the evaluation methods devised to detect
redlining by interested parties (community groups,
academicians and regulators) to determine whether
they account for the rational business reasons affecting lending decisions.
Demand for Mortgages
Outcomes in the mortgage market (as elsewhere)
result from the interaction of demand and supply
forces. Although the CRA directs its attention to
supply side (i.e., lender) behavior, it is also necessary to specify demand behavior to extract evidence
on supply behavior from the available data on mortgage activity.
According to a number of studies,8 the desired
stock of household debt is determined as an element
of a broader decision concerning the consumption
of housing and non-housing goods and services. In
the most general models, household wealth, current
income, prevailing interest rates, and market housing prices are found to determine the demand for
housing and mortgage debt. (Here wealth is defined
as the present value of lifetime earnings plus the
stock of savings.)
This relatively straightforward assumption is
complicated, however, by certain imperfections in
the credit and housing markets. First, the progressiveness of income tax rates and the tax deductibility of mortgage interest reduce the after-tax cost of
indebtedness to wealthier households. Second, the
lifetime earnings portion of household wealth is
fairly illiquid; households are typically not able to
borrow against their future income. Coupled with
the convention of minimum downpayment require-
ments, this fact may make initial savings-and not
simply total household wealth-independently important to housing and mortgage demand. A household with a lower level of initial savings would
display a lower effective demand for housing than
an equally wealthy household with a higher initial
level of savings.
In addition, the household's current income position (rather than its wealth alone) may be an important independent factor influencing its housing
ownership decision. With conventional mortgage
instruments, the borrower can encounter cash-flow
problems if the monthly loan payment is large relative to current nominal income. Lower current
income is likely to result in lower demand for housing, everything else being equal.
Finally, the variability of income may playa role
in the demand for housing and mortgage debt-the
more variable its income, the greater the risk that a
household will be unable to meet mortgage payments in the normal manner. Because of the high
legal and other costs of meeting (or avoiding) loan
delinquencies and defaults, a household with a variable income may have a lower effective demand for
debt and for housing than a similarly situated family
with a stable income.
These demand variables would suggest the weak-
Chart 1
Savings'
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
t
0.1
0.0 ..
_
o
....,j
'
10
'
30
Permanent income ($ thousands)
'
40
....1
'
50
'Savings is a dichotomous variable where O=households without savings greater than two
estimated here represents the
months' income and 1= all other households. The
probability of a household
savings greater than two
months' income. 10
25
Chart 2
Income variability"
040
.35
.30
.25
.20
.15
.10 ----.......--"""--_ _......_ _"'-_...L._......L.-_......_
o
10
20
30
40
......_ - - I
50
Permanent income ($ thousands)
"Income variability is defined as the standard deviation of each household's income from its
nine-year-average trend values. 10
gage demand on the part of less wealthy households
should be less than their wealth alone would predict. Therefore, we would expect to find poorer
households demanding fewer mortgages than richer"
households, even after adjustment for income and
wealth. Since neighborhoods tend to be relatively
homogeneous with respect to household wealth, an
uneven pattern of mortgage lending across neighborhoods may be explained, in part, by these differences in mortgage demand.
ness of wealth alone (or its reasonable proxy, permanent income) as an accurate predictor of the
demand for mortgages. Indeed, data on family
economic characteristics 9 suggest that savings as a
proportion of wealth tends to increase as wealth
increases over much of its range (see Chart 1).
Moreover, poorer households also tend to have
more variable incomes than all but the wealthiest
households (see Chart 2).
These nonlinear relationships suggest that mort-
Mortgage Supply
In addition to these demand-side influences, a
number of factors relating to the applicant and the
property will necessarily influence lenders' willingness to supply credit. One major factor is the lender's general inability to obtain security for the loan
by attaching the borrower's future income; thus, the
loan must be secured by the property itself. Factors
bearing on the likelihood and cost of foreclosure
and liquidation will thus influence the lender's willingness to supply credit. The borrower's ability to
handle the cash-flow burdens of a mortgage, of
course, would be paramount in a bank's assessment
of the risk of foreclosure. Thus, the borrower's
current income, liquid asset position and income
stability are all considered by lenders in this regard.
The lender's perceptions about the "quality" of
the real assets securing a loan also will affect the
lender's willingness to make a real estate loan. In
obtaining a secured loan, the borrower in effect
obtains an option to hand over the security (the
house) to the lender and abandon the loan. As option theory suggests, the more uncertain the future
value of the security (the house), the more valuable
this option becomes. Thus, lenders should charge
more or demand more security (by offering a lower
loan-to-value ratio, for example) for a loan on a
property with an uncertain future value. Such uncertainty typically will arise due to the lender's
assessment of the remaining economic life of the
house. This assessment, in tum, may be a function
of the house's current level of upkeep and of the
maintenance of nearby properties.
Consequently, we would expect fewer loans to be
supplied to those loan applicants with lower
26
consequences for mortgage supply. Lenders would
be willing to supply less mortgage money at any
given mortgage rate to variable-income households
because of potentially greater delinquency and default risks.
incomes, with lower savings/loan ratios, or with
intentions to buy homes with uncertain future value,
all else being equal. The greater income variability
of poorer households (see Chart 2) should also have
Resultant Lending Patterns
ular, households in poor neighborhoods are likely to
receive less mortgage money than households in
well-to-do neighborhoods. In fact, a comparison of
lending outcomes between inner city and suburban
neighborhoods probably would reveal a pattern of
fewer loans and lower dollar loan values in the
typical American inner city because of its general
pattern of household characteristics. The uneven
distribution can, therefore, often result from rational behavior on the part of both lender and potential
loan applicant. However, the evaluation methods
used traditionally by community groups generally
have led them to equate such uneven distributions
with redlining.
An accurate evaluation method for detecting noncompliance with the CRA, or "irrational" redlining, thus would necessarily incorporate variables
such as those discussed above. No available data set
would permit us to prove this point directly or to test
directly all the implications of our model on the
demand and supply of mortgages. However, some
of these variables are important to household financial decisions through their influence on the pattern
of home ownership. II
Since households of various wealth levels tend to
be concentrated geographically, these conclusions
about demand and supply factors may translate into
unevenness in the observed number, value, or price
of mortgage loans across neighborhoods. In partic-
Community Group Studies
Because of their limited resources, community
groups have tended to use the simplest analytical
procedures when providing evidence to support
their protests against financial institutions. Typically, their analysis of residential lending patterns
consists of construction of simple indices-such as
loans per census tract-to depict the geographic
pattern of mortgage lending. The type of indices has
depended upon the type of data available.
Prior to the passage of the HMDA, these groups
obtained their data from manual reviews of public
registers of real estate transactions-as seen, for
example, in the New York Public Interest Research
Group report on redlining in Brooklyn. The report
compared total value of Brooklyn mortgages made
by certain Brooklyn savings banks to these lenders'
total assets and total mortgage-loan volume. The
resultant ratios were small, and the report's authors
thus inferred that the lenders were redlining Brooklyn neighborhoods. 12
The passage of the HMDA considerably facilitated this simple index analysis, because it required
each financial institution to disclose the number and
value of all mortgage and home improvement loans
made, by type and by neighborhood (using census
tracts or zip codes to represent neighborhoods). 13
The HMDA provided much of the data used by
community groups to stimulate Congressional interest in the CRA. For example, in her March 1979
testimony, Gale Cincotta of National Peoples' Action used such data to show that three major cities
received a smaller share of urban financial institutions' loans than did their suburbs and non-metropolitan areas. 14 Ms. Cincotta used this example of
asymmetry between urban and suburban lending
patterns as evidence of the existence of redlining
and of the need for the CRA.
The Buckeye-Woodland Community Congress
(BWCC) in Ohio also used HMDA data and simple
index analysis to establish its standing in a eRA
protest. When AmeriTrust, a Cleveland bank holding company, proposed to acquire a new bank early
in 1980, BWCC alleged that AmeriTrust had a poor
record of real estate lending in predominantly black
27
areas of the community. These accusations were
based on simple indices developed from HMDA
data, real estate transfer data, and deposit data. 15
Although simple index analysis has
effective in
att€mtion,
suffers from serious
problems.
simple index analysis does not address the problelffi
of irrational
It cornp~tres
across
and thus cannot
lender is arbitrarily dis1crirninatirlg
neighborhood. Data on geClgI<lphic lendirlg p::tttems
alone cannot show whethl~r
of demand or
the behavior is Hn.v",,"
In
these
tion to the impact of risk variatlles
behavior. Without cOIltre1Uirig
legiltimlate]!y influence mortg,:tge dernaJ:1ld
possible to use
be occurring in those markets where the simple
index measures might imply the opposite.
Some groups have
the severe limitations of the simple index approach and have tried to
overcome them-for
by using additional
data to
for differences in demand. 16
Some have used real estate transfer activity and
other variables as proxies for mortgage demand.
these variables have limited value as
wen, since it is doubtful that
control for demand factors in a neighborhood. 17
index a'1!lal~{sis
prCIVirlg the existence of irrational redlining,
its frequent use in CRA protests. To isolate
patterns that involve
more than
economically "rational" behavior-namely, to
jr!",ntiJF" disclrimlimltOl"'j and irrational redliningfor the factors expected to
influence rational
behavior.
resear'ch(~rs, in their search for better measurement
methods, have come to employ one of two apI) market models or
applications
amUY:Sls. 18
HMDA data do not correct for population or
differences among
and
use
would be
even to
screen for
CRA violations.
redlining behavior could
Market models
(1) Mobserved = f
B, R).
This "reduced form"
is
estimated with
analysis,
data on the
number or value of mortgages made in various
census tracts or neighborhoods in a certain period as
measures of M. The borrower arid property characteristics are the average characteristics of the occuand properties in those tracts.
Within this
framework,
have
att(~mpted to obtain evidence of
in several
ways. Under one approach, certain characteristics
of the borrower (such as race) or the property (such
as the age of
in the
considered irrelevant to the banking decision. Thus
if these variables are statistically significant in explaining observed mortgage activity, the analysts
conclude that
is involved.
Hutchinson, Ostas and Reed,19 for example,
found the number of mortgages made in a crosssection of 120 census tracts in Toledo, Ohio, to be
negatively related to average housing age. They
In the market model approach, researchers
the
demand and
factors in the process that determines observed mortgage activity. Demand for mortgages is
typically assumed to have the form
Md = Md (i, P, X)
where Md is the demand for mortgages, i is the
mortgage interest rate and other loan terms, P is the
of
and X is a set of variablles
the scale of demand. (such
and
mortgage supply relationship takes the general fmm
=
where i is the terms of the mortgage, B is a set
borrower characteristics related to creditworthiness, and R is a set ofcharacteristics ofthe property.
in eq\lililtlrillm,
recoglliz(~d
Md
= M s = M observed
and the model can be solved for the relationship
between observed mortgage activity (Mobserved )· and
borrower and property characteristics:
28
took these findings as evidence of iredlining.
this ap]:)ro~lch,
MiichiJgan Panel Survey data suggest that
variatlles are
in
of
retrospective analysis.
In addition,
portion of the modg,age
tional mOltgalge lendill1g),
ity of
bankers
other
lenders. The exclusion from these studies of private
now
an increasmOltgalge supply,
may
any findings of redlining behavior,
since lenders may specialize in
kinds of
loans and Perhaps neighbdrhoods
Fdr these reasons,. market
solved and are unlikely to resolve
about
redlining. Moreover, the difficulties encountered in
verifying allegations of redlining for the market as a
whole are multiplied severalfold when a specific
lender's behavior is involved, since the market
mddel must then
the market.shares of varias well
what the
CRA is meant to address. Conceptually, market
models are far superior to the simple index analysis
practiced
analysts and some regulatory
market models have not
det~~ctirlg "redlingenerally arnbilguolls findings
"oh'"'''''''' sllpelficial e"ldence of
approal~h
is to estimate e(~Ua1tion (1) to
alk~gec:ny redlined
pUl:'pdrtel::Uy nonredallegedly
VUl'UlU1t:, the anat()rlsj(lerthi~. evidence ot redlinin,g. Using this
study
"mort-
York City
With this approach, however, valid comparisons
may not be pdssible becauSe not enough legitimate
factors' influencing mortgage volumes have been
the prediction
Richard:i()n
out the need for
and
areas typically differ dramatically in borrower and
property chmractlerisl:ics.
initial savings, and income variability in the
mOltga~;e demand (mainlv
29
Applications Analysis
Because ofthe difficulties encountered with market models, some academic analysts have tried to
simplify the problem by focusing only on the lender's loan evaluation process. If lenders reject loan
applications involving properties in certain locations more frequently than similar applications elsewhere, this could be taken as an indication of possible redlining behavior.
In essence, such studies are pure supply studies;
"demand" is given since an application has been
filed. 22 This alleviates the problem of modelling the
demand process. In addition, inferences about individual mortgage suppliers can potentially be made
by focusing on the applications process of the
specific lender.
The typical applications analysis model involves
estimation of a mortgage decision relationship of
the form
Prob(MD) = f(i, B, R)
where MD is the mortgage decision made concerning the application (such as "denial" ofthe application), and Prob = f( ... ) is a function describing the
probability of that decision as a function of i, the
terms of the loan request, B, the characteristics of
the applicant, and R, the characteristics ofthe property, (including location). This model indicates the
presence of irrational redlining if all characteristics
of the applicant and the property relevant to a wise
business decision are included and the property's
location is still independently important.
The most thorough study of this type was conducted by Schafer and Ladd for the Department of
Housing and Urban Development. 23 Since such
studies require detailed data on individual mortgage
applications, the researchers were limited to an
analysis of mortgage markets in New York and
California, where state laws require certain institu-
tionallenders to provide information in applications
registers. 24 The authors obtained information on
allegedly redlined areas, and then tested whether
such property locations were independently important in explaining mortgage decisions.
The results of the study are mixed; the location of
a property in an allegedly redlined area increases the
probability of adverse treatment by the lender in
some but not all of the cities studied. More surprisingly, there are statistically significant cases where
"nonredlined" areas appear to receive less favorable consideration than "redlined" areas. Indeed,
in California, there are only six cases in which an
adverse mortgage decision is statistically more likely for central city properties than for suburban properties; yet there are twenty-one cases where the
reverse is true. Similar, though less pronounced
ambiguities arise from the New York data.
The authors conclude from their results that
"some neighborhoods appear to be redlined and
others do not." An alternative explanation, however, is the omission of some locally important
variable(s) from the model specification. The applications data provide relatively good information on
the applicant's financial position (such as some savings and net worth data), but they lack most information that might bear on the market's perception of
the riskiness of the specific property (with the exception of age of house). The data thus had to be
augmented with census and other data, which might
have been insufficient to the task; indeed, the analyst really needs all of the data available to the
lender to discern accurately the "unbiasedness" of
the loan decision process. Nevertheless, applications analysis, by virtue of narrower focus, has
greater practical potential than market modelling or
index analysis for eRA evaluation.
Use of Applications Analysis
Most of the regulatory agencies have access to
the loan application register (LAR) maintained by
each institution. The LAR is a record ofloans made,
including details related to the applicant, the property and the loan terms. This record must be maintained for 25 months for every loan made by each
institution. 25
The Federal Home Loan Bank Board conducted a
pilot study of such data in 1978 to detect discrimination in the overall lending practices of savings and
loan associations. 26 Discrimination was defined as
the arbitrary use of applicants' age, race, sex, or
marital status, or property location, to make decisions regarding appraised value, loan acceptance or
30
denial, or mortgage terms. Violations of CRA
would have been found if property location alone
had made a difference in lending decisions, but
discrimination was not evident after controlling for
applicant and property risk variables. 27 The pilot
study demonstrates the agencies' concern about
finding appropriate objective measurements of
CRA compliance. It also suggests that detailed
studies of individual applications reveal more information than simple index models and can explain
much "suspect" activity.
v. Conclusions and Policy Implications
This paper has focused on some of the analytical
challenges posed by the Community Reinvestment
Act. Proponents of the CRA had many policy aims
in mind-inCluding the regulation of credit flows to
stimulate redevelopment of deteriorated urban
areas. But in view of analytical limitations as well
as Congressional intent in the ultimate CRA legislation, we believe that enforcement of the CRA's
anti-redlining provisions should center on detection
of irrational redlining, or arbitrary geographic discrimination. Indeed, the debate surrounding the
passage of the CRA makes it clear that geographic
credit allocation was not Congressional intent in the
Act's final form.
With this in mind, we have attempted to assess
the usefulness of various analytical techniques and
data sources in detecting the arbitrary use of property location in mortgage lending decisions. We
found that the simple index techniques commonly
used by community groups are likely to be unreliable, because they ignore the complexity of the
economic decisions involved in the mortgage market. These measures simply do not account for the
sound business reasons or demand factors which
may be the· reason for disparities in loan volumes
amongheighborhoods. We also found that the market model approaches used in more sophisticated
studies were also inappropriate, due to the difficulty
of defining an individual lender's role in such a
complex context. Arnorereliable technique for
evaluating. charges of g~ographic discrimination
may be loan applications analysis, because it
focuses on individual lendihg decisions, while at
the same time drawing on a more complete set of
data thanthe.simple index techniques.
These conclusions suggest that effective CRA
enforcement may require substantive changes in the
methodology used by regulators to assess a lender's
CRA performance and to evaluate allegations of
redlining. In the absence of quantitative evaluation
techniques, CRA assessments today largely depend
on the judgment of CRA examiners. Since the detection of CRA violations is considered ari important regulatory responsibility, decisions should be
accurate, and consistently applied, given their farreaching consequences. The use of formal, objective methods of evaluation can make a positive
contribution to both of these goals. Among the
methods that probably should be considered are
those which analyze loan application records.
At the present time, regulators must conduct
thorough analyses of CRA compliance when a protestant alleges the existence of redlining and also
during routine examinations. With a loan application register readily available for analysis, regulators could address complaints more quickly and
accurately. The burden of the analysis would rest
with the regulatory agencies, which have the appropriate staff and resources for the task. This should
serve the desire of Congress to keep enforcement
costs to a minimum. Also, it should save protestants
from the time-consuming, unreliable use of simple
index methods, and should reduce the burden of the
exi~ting CRA process on affected institutions.
However, the loan application register has drawbacks also. The need to maintain the necessary
standard-format applications data files would impose a non-trivial compliance burden on affected
institutions. (There is now no standard format nor
standard method of analysis, and lending institutions must only maintain a file oHoan applications
and make these records accessible to regulators.)
The. costs of maintaining the loan registers would be
high, especially since relatively few banks are faced
with proteSts or allegations ofpoor perfonnance. In
addition, unless institutions were also required to
maintain records of all requestsfor lending information (in addition to formal applications), this
method would not detect "pre-screening" forms of
31
lending discrimination.
Given the serious problems associated with existing evaluation methods, and given the high costs of
a more accurate, (but still imperfect) alternative
method, regulators might do well not to try to detect
redlining per se, but rather to concentrate on encouraging the affirmative marketing efforts of
financial institutions. Assuring the free. flow of
information to all market participants should
increase competitive pressures on lenders who discnmllflate, making them less able to continue such
pr2lcti,ces in the
run.
If the efforts to detect redlining are to continue,
however, improvements in evaluation methods may
be necessary. At the very least, analysts should
conduct a more thorough study of the costs and
benefits of alternative evaluation methods than we
have attempted here. Our analysis indicates that the
current method has severe limitations, and that a
more accurate method would involve incorporating
all the information lenders receive in loan applications. Although the substantial investment of time
and capital necessary to maintain these data may
exceed the explicit costs of current compliance regulations, an accurate cost/benefit analysis would
also have to consider the high, hidden costs incurred
under existing protest procedures-such as concessions made in private agreements, penalties incurred through conditioned approvals, legal fees,
and costs of jeopardized
When these hidden costs are taken into account, it is
not obvious that loan applications analysis would be
too costly to implement. Indeed, if eRA enforcement and elimination of discriminatory lending
practices continue to be desirable legislative goals,
a review of the current evaluation method clearly
would be in order.
FOOTNOTES
1. Whenever a bank applies to expand its operations, the
regulatory agency does an analysis of the competitive
effects of the proposed activity and an analysis of how the
expansion meets the "convenience and needs" of the COmmunity. Banks must, therefore, include in their applications
descriptions of how the expansion wiii benefit customerS by
improving services in their communities.
6. Comptroller of the Currency, Federal Deposit Insurance
Corporation, Federal Home Loan Bank Board and Federal
Reserve Board, Community Reinvestment Act Examination Procedures, November 1978, p. 5.
7. Federal Reserve Bank of St. Louis, news release,
November 30, 1979.
8. See Diewert, W. E. "Intertemporal Consumer Theory
and the Demand for Durables," Econometrica, May 1974,
pp. 497-516; Dunkelberg, W. L., and Stafford, F. P., "Debt
in the Consumer Portfolio: Evidence from a Panel Study,"
American Economic Review, September 1971, pp. 598613; Hess, A. C., "A Comparison of Automobile Demand
Equations," Econometrica, April 1977, pp. 683-701; Mishkin, F. S., "Illiquidity, Consumer Durable Expenditures, and
Monetary Policy," American Economic Review, September 1976, pp. 642-54; and Sandmo, A., "The Effect of
Uncertainty on SaVing Decisions," Review of Economic
Studies, July 1970, pp. 353-360.
2. For further details, see U.S. Senate, Hearings before
the Committee on Banking, Housing and Urban Affairs
01'1 S. 406, "Community Credit Needs," March 23-25, 1977,
95(1), and Consumer Bankers Association, A Compliance
Guide for the Community Reinvestment Act: Background and Implications.
3. Board of Governors of the Federal Reserve System,
Regulation SS (12 CFR 228), effective November 6, 1978.
4. For a thorough discussion ofthe effects test as it applies
to consumer credit legislation, see Sarah E. Burns, "Credit
Scoring and the ECOA:
Yale
law Journal, 88(7),
9. The data are from continual follow-up surveys of5,000
American families in each of the nine years 1968-1976
conducted by the Survey Research Centerofthe University
of
(the Michigan Panel Survey). Our sample,
ever, involves only families in which the married couple
5. The General Accounting Office, for example, in a recent
stUdy of the enforcement of several consumer credit lavvs,
was critical of the agencies' monitoring of sUbstantive compliance or compliance with the "basic principles of the law."
The GAO complained that few detailed analyses were conducted on the data available. Although the study refrains
from drawing conclusions on CRA enforcement (since eRA
was new at the time), we can probably apply the agency's
findings to CRA as well. Comptroller General of the United
States, Report to the Congress: Examinations of Financiallnstitutions Do Not Assure Compliance With COnsumer Credit laws, U.S. Government Printing Office, January 2,1981.
which headed the household rernained together overall the
years of the survey.
10. The equations for the charts took the forms s = a +
+ cy 2 (for Chart 1) and I = a + bY + cy2 (for
where S = the savings variable as described, I = income
variability, and Y permanent income. Source of the data
used is described in footnote 9.
11. The importance of the factors presented in tl1e th~)relti
cal discussion was demonstrated in a regression of home-
32
and does not seem to improve upon the market models and
applications analysis, we do not discuss it in the text.
ownership on various homeowner attributes (Michigan
f'anei Survey data). Forexample,levelof permanent
income and level of savings show a statistically significant
relationship to home ownership. Increased income variability, on the other hand, reduces the probability Of.· home
ownership for the families in the sample. Because of the
obvious relationship between home ownership and mortgage indebtedness,.the findings suggest that these same
variables would affect fhe.pattern. of mortgage. indebtedness, although we cannot distinguish whether the factors
are supply or demand related.
19.· Peter M. HutChinson, James R. 05tas, J. David Reed,
"A •Survey and Cortlparison of Redlining Influences in
Urban Mortgage Lending Markets," AREUEA Journal, 5,
Winter 1977, ppA63-472.
20.·.HarryW,RiChardson and Peter Gordon, "Measuring
MortgageDefiCiency and Its Determinants," The Annals of
Regional Science, November 1979, 13:3, pp. 25~34; and
RobertSchafer,Mortgage lending Decisions: Criteria
and Constraints, Cambridge, Mass. 1978.
12. New York Public Interest Research Group, "Take the
Money and Run," New York, 1976.
21. See footnote 11.
22. Ofcourse these studies assume that no pre-screening
the actual application process.
13. The HMDA was amended in 1980 and now requires
disclosure of the number and value of mortgage loans
made by census tract only. Zip code can no longer be used
as a designation of neighborhood.
23. Robert Schafer and Helen Ladd, Equal Credit Opportunity Accessibility to Mortgage Funds by Women and
by Minorities, Volume 1~3, U.S. Department of Housing
and Urban Development, U.S. Government Printing Office,
Washington, D.C., 1980.
14. See Hearings, footnote 2, pp, 132-147.
15. Federal Reserve Bulletin, "Bank Holding Company
and Bank Merger Orders issued by the Board of Governors," March 1980, pp. 238-242.
24. In California, only state-chartered savings and loan
associations were involved. In New York, data were Obtained from savings and loan associations, commercial banks
and mutual savings banks.
16. The Department of Housing and Urban Development,
in its CRA guidebook, Assessing Community Credit
Needs (August 1979, p. 13), recommends that community
groups use census tract data, coupled with. HMDA data, to
help determine demand as well as to explain possible discrepancies in loan volumes between two dissimilar census
tracts.
25. The Federal Reserve Bank does not require regular
reports on this subject, but the other three agencies must
collect such data on a regular basis as a result of a federal
lawsuit
26. A. Thomas King, "The Loan Application Register: A
Tool for Examiners," Federal Home loan Bank Board
Journal, August 1980, pp. 8-13.
17.. With regard to the use of real estate transfer records to
compensate for demand, evidently there is not always a
one-to-one relationship between mortgages and transfers
of real estate. The use of private or mortgage bank financing, the practice of assuming existing mortgages, and other
factors wit! make this link a loose one. It is also very possible
that real estate transfers are a reflection of the mortgage
market, not vice versa-people may demand fewer homes
because they have difficulty finding mortgages.
27. One exception was the higher rate of denials among
Blacks and Hispanics. However, even if this conclusion is
valid, it is a violation of the Equal Credit Opportunity Act
rather than CRA, and hence is not addressed in this paper.
Furthermore, we cannot say whether this finding is conclusive evidence of discrimination, since race could be correlated with certain risk factors not compensated for in the
study.
18. The survey method of analysis has also been tried (for
instance, a study of redlining in Rochester, N.V. by George
Benston). Because this method has not been used widely
REFERENCES
Barth, James R., Joseph J. Cordes and Anthony M. J.
Yezer. "Financial Institution Regulations, Redlining
and Mortgage Markets," The Regulations of Financial Institutions, Conference Series No. 21, Federal
Reserve Bank of Boston and the National Science
Foundation, October 1979, pp. 101-143.
Brown, Jonathan. "Will the New Law Rub Out Bank Redlining?" Business and Society Review, December
1978.
Burns, Sarah E. "Credit Scoring and the ECOA: Applying
the Effects Test," Yale law Journal, Volume 88(7),
June 1979, pp. 1450-1486·.
Benston, George J. "Mortgage Redlining Research: A
Review and Critical Analysis Discussion," The Regulations of Financial Institutions, Conference
Series No. 21, Federal Reserve Bank of Boston and
the National Science Foundation, October 1979,
pp. 144-195.
Canner, Glenn. "Redlining and Mortgage Lending Practices," Unpublished PhD dissertation, Brown University,1979.
____ . "The Community Reinvestment Act: A Second
Progress Report," Federal Reserve Bulletin, Volume
67, November 1981, pp. 813-823.
___ , Dan Horsky and H. Martin Weingartner. "An Empirical Study of Mortgage Redlining;' Monograph Series
in Finance and Economics, 1978-5, New York University,1978.
____ and Joe M. Cleaver. "The Community Reinvestment
Act: A Progress Report," Federal Reserve Bulletin,
Volume 66, February 1980, pp. 87-96.
33
Comptroller General of the United ~tates .. Rep0r:t t~ the
Congress: Examinations of F!nanClal Instltutlon~
Do Not Assure Compliance With Consumer Credit
Laws, U.S. Government Printing Office, January 2,
1981.
Mishkin, F. S. "Illiquidity, Consumer Durable Ex?endit~res
and Monetary Policy," American Economic ReView,
September 1976, pp. 642-654.
Nathan, Harold C. "Monitoring Mortgage Lending by
Banks," Issues in Bank Regulation, Winter 1979,
pp.7-16.
Comptroller of the Currency, Federal Deposit Insurance
Corporation, Federal Home Loan Bank Board, Federal
Reserve Board. Community Reinvestment. Act
Examination Procedures, November 1978.
Richardson, Harry W. and Peter Gordon. "Measuring Mortgage Deficiency and Its Determinants," The Annals of
Regional Science, November 1979,13:3, pp. 25-34.
Consumer Bankers Association, A Compliance Guide for
the Community Reinvestment Act: Background
and Implications.
_ _ . "The Construction of Mortgage Deficiency Indices,"
A Report for the California Department of Savings and
Loan, March 1978.
Dennis, Warren L. "How Do You Know You're Not Discriminating?" Federal Home Loan Bank Board Journal,
May 1977, pp. 15-24.
Sandmo, A. "The Effect of Uncertainty on Savings Decisions," Review of Economic StUdies, July 1970, pp.
353-360.
__ . "The Community Reinvestment Act of 1977: Defining 'Convenience and Needs of the Community,'''
Banking Law Journal, pp. 693-717.
Schafer, Robert. Mortgage Lending Decisions: Criteria
and Constraints, Cambridge, Massachusetts, 1978.
Diewert, W. E. "Intertemporal Consumer Theory and the
Demand for Durables," Econometrica, May 1974, pp.
497-516.
-
Dunkelberg, W. L. and F. P. Stafford. "Debt in the Consumer Portfolio: Evidence from a Panel Study," American Economic Review, September 1971, pp.
598-613.
and Helen Ladd. Equal Credit Opportunity: Accessibility to Mortgage Funds by Women and by M.'norities Volumes 1-3, U.S. Department of HOUSing
and U~ban Development, U.S. Government Printing
Office, Washington, D.C., 1980.
Segala, John P. "Redlining: An Economic Analysis," ~ed
eral Reserve Bank of Richmond Economic ReView,
Volume 66, Number 6, November/December 1980,
pp.3-13.
Federal Reserve Bulletin, "Bank Holding Company and
Bank Merger Orders Issued by the Board of Governors," Volume 66, March 1980, pp. 238-242.
State of California Department of Savings and Loan. Fair
Lending Report, No.1, Volumes 1 and 2, 1977, No.2,
1978.
Guttentag, Jack M. and Susan M. Wachter. "Redlining and
Public Policy," Monograph Series in Finance and
Economics, 1980-1, New York University, 1980.
Survey Research Center-Institute for Social Research.
Five Thousand American Families-Patterns of
Economic Progress, University of Michigan, 1978.
Hess, A. C. "A Comparison of Automobile Demand Equations," Econometrica, April 1977, pp. 683-701.
United States Department of Housing and Urban Development. Assessing Community Credit Needs, August
1979.
Hutchinson, Peter M., James R. Ostas, and J. David Reed.
"A Survey and Comparison of Redlining Influences in
Urban Mortgage Lending Markets," AREUEA Journal, 5, Winter 1977, pp. 463-472.
United States Senate. Hearings before the Committee
on Banking, Housing, and Urban Affairs on S. 406,
"Community Credit Needs," March 23-25, 1977, 95-1.
King, A. Thomas "Redlining: A Critical Review of the literature with Suggested Research," Federal Home Loan
Bank Board, Working Paper #82, February 1979.
_. "The Loan Application Register: A Tool for Examiners," Federal Home Loan Bank Board Journal,
August1980,pp.8-13.
34
Yvonne levy*
Forecasts of California water supply and demand
to the year 2000 suggest that overall supplies will be
ample. But according to the same studies, the state's
total water supplies are distributed so unevenly
geographically that a chronic shortage could develop in certain areas by the late 1980's. I Southern
California-which currently accounts for twothirds of the state's total water consumption-is
particularly vulnerable to a potential shortfall. Despite an increase in projected demand, that area by
the mid-1980's will lose over one-half of the 1.2
million acre-feet of surface water it currently receives annually from the Colorado River. 2
Most proposed solutions to the problem have
called for an expansion of supplies for prospective
water-short areas, primarily by the construction of
new dams and canals to bring more water from
Northern to Southern California. Recently, the most
intense debate has centered on the Peripheral Canal,
a proposed addition to the State Water Project which
would cross the Delta formed by the Sacramento
and San Joaquin rivers at the head of San Francisco
Bay and bring more water to Southern California.
(An overwhelming negative vote from Northern
California voters caused the plan to be rejected in
the June 1982 election.) But programs to expand the
Central Valley Project-the huge Federally-owned
water system-also have created considerable
controversy.
An alternative approach-an economic approach
-would solve the problem through pricing reform,
as a means of reducing the projected growth of
demand. 3 According to this view, the projected
supply-demand imbalance reflects the assumed
continuation of inefficient pricing practices followed by Federal, state and local agencies (utilities) in
pricing water at all stages of distribution. If water
were priced higher, final-users would have a greater
incentive to conserve, the projected demand would
be lower, and some or all of the proposed new water
facilities would not be required. The present arti~le
follows this approach in analyzing the pricing of
surface irrigation water supplied from the Central
Valley Project (CVP) and sold at wholesale by the
U.S. Bureau of Reclamation (Bureau).
Agriculture accounts for about 85 percent of the
total water consumed annually in California. The
Central Valley Project is the single largest supplier,
accounting for nearly 40 percent of total irrigation
water, with the State Water Project and groundwater sources accounting for the remainder. The
pricing practices followed by the U.S. Bureau of
Reclamation in pricing irrigation water at wholesale
thus significantly influence the general level of
water prices ultimately faced by the important agricultural sector.
Economic theory suggests that resources would
be allocated most efficiently if the Bureau based its
rates for CVP irrigation water on the "long-run
incremental cost" of supplying that water. This
concept refers to the cost of delivering an additional
unit (acre-foot) of water, taking into account the
need to add more fixed factors, namely new plant
facilities. Pricing all CVP irrigation water on the
basis of the cost of the last increment would be the
most efficient method of allocating scarce resources, because customers would then be aware of
the cost of the resources required to bring them
additional water.
In this paper an estimate of long-run incremental
cost based on the cost of building the next scheduled
block of capacity-namely, the proposed AuburnFolsom South Unit-is developed. This estimated
long-run incremental cost is far higher than the
"replacement average cost"-the average cost of
irrigation water from the existing plant (including
both old and new facilities), when this plant is
valued at its current replacement cost, i.e., the
* Senior Economist,
Federal Reserve Bank of San
Francisco. Alane Sullivan and Elaine Foppiano provided research assistance for this article.
35
demand for Federal irrigation water.
Section I discusses the economic-efficiency
argument for pricing on the basis of long-run incremental cost. As noted there, water utilities traditionally have followed other pricing methods
because their operations presumably have been
characterized by decreasing long-run replacement
average costs owing to economies of scale. Under
such conditions, pricing on the basis of incremental
cost would fail to recover average cost valued on a
replacement cost basis, and thus would result in a
loss. But as Section n indicates in the case of the
Central Valley Project, long-run incremental cost
today actually is far higher than the average cost of
CVP irrigation water, even when average cost is
measured on a replacement basis. This suggests that
water provision is no longer a decreasing cost industry. In Section HI, we show that the
realized
by the Bureau do not cover the actual costs of
supplying that water, partly because ofthe Bureau's
failure to recover full historical cost, and partly
because of the failure of the utility industry's historical accounting methods to reflect inflation over
time. In Sections IV and V, we discuss some of the
implications of higher Federal irrigation water
prices for the demand for water, and also for the
future development of the CVP irrigation system.
opportunity cost to society of the resources that are
currently tied up in supplying water. The differential is even greater between incremental .cost and
"historical average cost"-the average unit cost of
water from the existing plant, with the latter valued
on a historical (original) cost basis. Most regulatory
commissions use the historical pricing method for
investor-owned utilities under their jurisdiction.
Yet today, the Bureau of Reclamation is realizing
an average price for Central Valley Project irrigation water that does not even recover full historical
average cost let alone the
average cost.
In implementing reclamation law, which calls for
such practices as basing rates on farmers' ability to
pay and not charging interest on public funds invested in the CVP irrigation system, the Bureau is
requiring taxpayers and electrical users to pay a
substantial subsidy per acre-foot of Federal irrigation water supplied. The subsidy is even greater
when the realized average price is measured against
the "true" average cost-average cost on a replacement cost basis.
The purpose of this study is to describe and measure how Bureau prices for Central Valley Project
irrigation water deviate from the efficiency model
and to discuss the implications of higher relative
water prices for the California agriculture sector's
I. Rationale for Different Pricing Methods
Social objectives related to the development of
the arid West and the creation of a prosperous farm
sector have traditionally guided the Bureau of Reclamation in its pricing of irrigation water4 (see
Box). To foster these objectives, Federal reclamation law has limited the agricultural sector's repayment responsibility to its "ability to pay," a concept described later. But in the course of fostering
such objectives, the Bureau not only has failed to
recover the full historical average cost for irrigation water but has not even consistently utilized the
traditional av,~ra:ge-co~;t
by utility regulatory commissions for private
investor-owned utilities.
Under this standard historical average-cost
the
first determines its revenue
requirement for a particular function, for example,
electrical power or irrigation water. 5 This refers to
the total costs that must be recovered through rates
during a given period to compensate the utility for
all the expenses incurred in supplying the product,
including a return on invested capital. 6 Under present statutes, total revenues must exactly equal total
costs, a requirement known as the budgetary constraint. Dividing total costs by the number of units
expected to be sold in a given period yields the
average unit cost-and thus the price-of the product.
the "ability to pay" doctrine has led
the Bureau to set its price below the level implied by
this type of computation.
In economic theory, the value of the resources
embodied in each unit
is determined on the
basis of the replacement cost ofthe plant employed.
As we shall see, utility commissions have used
economic theory as the rationale for pricing on the
basis of average cost. But
prescribed an
accounting method for measuring average cost
which understates the true replacement average
cost.
36
But even if the Bureau had priced its water according to the true economic definition of average
cost, that method still would not have been the most
efficient in allocating resources. Theory also
demonstrates that for efficient resource allocation
the price per unit should be equal not to average cost
but rather to incremental cost. Incremental cost is
the change in total cost resulting from an additional
unit of output-that is, the cost of producing one
more unit of a good or service, or alternatively, the
cost that would be saved by producing one less unit.
A
of economics states that
optimum efficiency is achieved when the prices of
goods and services are equal to their marginal cost
of production. Under such conditions, resources
would be channelled into their most efficient uses. 7
This is because each price would reflect the value of
the resources required to supply each particular
good or service, and because consumers therefore
would be provided with the proper price signals to
make the choices that would yield society the most
efficient use of resources. If price were less than
marginal cost, consumers would be induced to consume an additional unit, even if the benefits were
less than the marginal commitment of society's
resources to produce that unit.
An important point is the distinction between
short and long-run, which is based on whether or
not plant size is fixed. Short-run calculations show
how a firm's costs will vary in response to variations
in output within the limits of a given amount of
fixed plant. Long-run calculations show how costs
vary during a planning period long enough to permit
adjustment of the scale of productive (or distribution) facilities.
Water pricing decisions thus depend upon
whether or not the scale of plant is to be increased. If
new plant is scheduled during the planning period
encompassed in the rate calculation, long-run incremental (marginal) cost is the appropriate basis for
efficient
I.e.,
be
to long-run incremental cost. 8 Long-run incremental cost equals the cost of water produced by the next
block of new storage and conveyance facilities
scheduled to be added. Under that
the
price per unit thus reflects only the cost of water
produced from new productive facilities-in contrast to the regulators' favored method of average
cost pricing, which also reflects the cost of water
from older facilities.
Regulatory agencies traditionally have not followed the incremental precept in establishing utility
rates because of their assumption, in their rate setting, that utility operations are characterized by
decreasing long-run average costs. Decreasing
long-run costs are the result of increasing returns to
scale, which mean that a larger plant has lower unit
costs than a relatively smaller plant. Average production costs decline for the individual firm with
any increase in the size of its
one or more
facilities). Economies of scale are internal to the
operation of the individual firm, in contrast to external economies which arise out of the growth of the
entire industry. 9
Most importantly, economies of scale are defined
for a particular point in time. At any given time, a
firm would be operating in an output range associated with decreasing long-run average cost if expansion to a larger-scale plant (or system) built from
scratch entailed lower average costs than a smaller
plant also built at that time.
With increasing returns to scale, the long-run
incremental cost associated with a given supply of
water is less than its long-run average cost. Hence,
if selling prices were established on the basis of
incremental cost (the cost of the last unit), average
cost would not be recovered, and the result would
be a loss. But this "loss," as measured by economists, exists because of the specific manner in
which average cost is defined, with reference to the
replacement cost of fixed plant. In practice, incremental cost pricing could yield accounting profits
because regulators traditionally have valued plant at
historic (original) purchase prices. However, in an
industry characterized by decreasing returns to
scale and increasing average costs -the CVP case
-incremental costs exceed average costs and thus
incremental cost pricing would result in a profit in
an economic as well as
sense.
Chart I-A shows the characteristics of decreasing
long-run average cost that originally led governments to grant utilities monopoly status and to institute
pricing. 10 The demand schedule
D, which shows the quantity that will be demanded
by customers at each price, intersects the long-run
average cost schedule (LRAC) at an output level
where further expansion in plant size (scale) will
37
reduce average unit cost, i.e., before the least-cost
size.
To achieve the most efficient allocation of resources possible under regulated-monopoly conditions, the utility would have to follow incrementalcost pricing. Under that method, the price (Pic)
would be determined by the cost of production of
the last unit, that is, by the intersection of the
demand schedule (D) and the long-run incrementalcost curve (LRIC). But setting the unit price at Pic
would generate losses for the firm (or agency) under
conditions of decreasing long-run average costs, in
that the cost of the last unit of output would be less
than the average replacement cost per unit. These
losses would be represented by the area, (PI-Pic) x
Qic'
To avoid the necessity for public subsidies to
offset these losses, rate-setting commissions originally selected average-cost pricing, incorporating
in the average cost a rate of return on invested
capital. Under this method, the maximum price per
unit is set at (Pac)' the intersection of the demand
schedule (D) and the long-run average cost curve
(LRAC). Under conditions of true decreasing longrun average cost, this method of pricing results in a
higher unit price and lower level of output than
would result from the more efficient incremental
cost method. This is because long-run average cost
is above long-run incremental cost under such
conditions.
Chart I-B illustrates the price and output combinations that would result under alternative pricing
methods if the utility were operating in a range of
increasing long-run average costs due to the exhaustion of economies of scale. This situation characterizes most utility systems today; for example, the
CVP is operating in an output range where further
expansion in size raises the average unit production
cost, that is, where the incremental unit cost is
above the average cost, measured on a replacement
basis. Under such conditions, pricing on the basis of
long-run incremental cost results in a price (Pie> and
output level (QiC)' That price would yield a profit
beyond the return incorporated in average cost, in
that the cost of the last unit of output would be more
than the average cost per unit. The excess profit
would be represented by the area, (Pic - PI) X Qic'
To avoid excess profits, regulators could follow
the replacement average-cost method, which would
result in price (Pac) and output level (QaC>. But
average-cost pricing, even under conditions of true
increasing long-run average costs, results in an
Chart 1
Pricing Alternatives in a Regulated Monopoly Situation
A. Utility Operating in a Range of
B. Utility Operating in a Range of
Increasing Long-Run Average Costs'
Decreasing Long-run Average Costs'
Price & Cost
Per Unit
Price & Cost
Per Unit
Q ao Q io
Output
Q", Q" Output
1 Describes behavior of costs at a given point in time.
'Based on plant valued at current prices, Le., prices prevailing at the given
point in time to ,which the cost schedules apply.
38
under-pricing of the product and a correspondingly
greater and uneconomic amount of resources devoted to its production. The use of historical rather than
replacement average cost results in a still lower
price and greater uneconomic amount of resources
devoted to its production.
II. Central Valley Project's Long-Run Incremental Cost
Congress has authorized a number of facilities to
expand the Central Valley Project, some of which
face an uncertain future due to environmentalist
opposition and uncertain funding. For that reason, it
is difficult to identify for analysis the next large
block of capacity likely to be added to the system.
The most likely candidate is the proposed AuburnFolsom South Unit, located between Sacramento
and Stockton. The project would consist of the
Auburn Dam, the Folsom South Canal and several
smaller structures, including the Sugar Pine Dam
and Reservoir, the County Line Dam and Reservoir
and associated conduits. II In addition to generating
as much as 450,000 kilowatts of electric power
annually, the project when fully operational would
supply about 440,000 acre-feet of water for irrigation and 300,000 acre-feet for municipal and industrial uses in the southern Sacramento and northern
San Joaquin Valley areas. 12
To date, only the foundation of the Auburn Dam
and some sections of the Folsom South Canal have
been completed. Congress has authorized (obligated) nearly $2.2 billion for construction, including about $1. 2 billion for irrigation purposes, but
the actions of environmentalists and the failure of
Congress to appropriate allocated funds have halted
further construction,I3
The project's long-run incremental cost would
equal the annual cost of adding an acre-foot of water
per year over the project's life. To compute this unit
39
Index Map
*
Project Headquarters
Water service area
Canal
Central Valley Basin Boundary
P Proposed
40
cost, we estimate the future stream of annual costs
in constant dollars-in this case, in 1981 dollars.
We then determine the present value of this future
stream of costs by discounting at an appropriate real
rate of interest. Multiplying the present value by the
real rate of interest yields the annual cost of investing resources in this project rather than using them
elsewhere in the economy. Finally, dividing this
annual cost by the project's expected annual output
yields an estimate of long-run incremental cost
Appendix
14
For
purposes, we have used the real
rate of return before taxes, i.e., a nominal rate of
interest minus the current rate of inflation. Also, we
have determined that rate on an opportunity-cost
basis-one that assures the general taxpayer a rate
of return on invested capital equal to that earned on
average in a private-utility sector financed solely
through long-term debt. This assumes little difference in risk between the Federal and private-utility
sectors, since the latter is regulated to ensure a
reasonable rate of return.
Specifically, we have selected a real discount rate
of 10 percent-the real rate prescribed by the Office
of Management and Budget for evaluating Federal
projects. 15 The rate is also consistent with a16
percent current nominal interest rate for new utility
bonds, minus an inflation rate of 6 percent. Employing these assumptions, we estimate the longrun incremental cost of irrigation water from the
proposed Auburn-Folsom South Unit project to be
around $324/acre-foot. 16
If it followed efficiency criteria, the Bureau
would price all CVP irrigation water on the basis of
incremental cost. In establishing rates for any given
future period, the Bureau would set the unit price
equal to the long-run incremental cost of the appropriate block of scheduled capacity. That practice
would make wholesale customers aware of the economic value of the resources required to supply
additional increments. Instead, in 1981, the Bureau
realized an average price of slightly over $5/acrefoot for CVP irrigation water-=-in contrast to the
$324/acre-foot price called for under purely economic criteria. This suggests that far more
resources are devoted to the Federal supply of irrigation water than are warranted by the value of the
agricultural commodities produced.
IU. Differential Between Long-Run Incremental and Average Cost
Three basic reasons can be found for the
differential between the estimated long-run incremental cost and the latest realized average price of
CVP irrigation water. First, the Bureau has not
followed the traditional utility pricing model, so
that its average realized price is far below the full
average cost determined on an historical accounting
basis. Ses;ond, the traditional utility model fails to
reflect replacement cost. Third, the long-run incremental cost of irrigation water would be higher than
"true" (economic) average cost, measured by average replacement cost.
Traditionally, a utility determines the capital
costs to be recovered through revenues on the basis
of its historical (original) cost of plant and equipment. These capital charges include such items as
depreciation, interest, and property taxes. During
periods of rapid inflation, when the cost of new
equipment rises far beyond the original cost of similar equipment acquired in the past, this historical
of
accounting method yields a
average cost than the replacement cost method. Yet
the Bureau does not even recover average cost determined under the historical accounting method.
The reasons are: I) reclamation law does not require
the Bureau to recover interest on Federal funds
invested in irrigation projects; 2) reclamation law
limits the repayment responsibility of farmers to
their "ability to pay"; 3) the Bureau supplies water
under long-term contracts at fixed rates which are
not adjusted upward to reflect the blending in of
new higher-cost capacity; 4) the Bureau pays no
property tax as would a private utility and 5) by
periodically extending the assumed lifetime of the
plant, the Bureau has reduced the amortization
charged on past investments.
1. Interest subsidy: The Reclfuliation Actof 1902
required beneficiaries to repay the construction
costs of Federal irrigation projects, but did not require payment of interest. 17 Congress has retained
that interest subsidy ever since.
Some critics claim that an "opportunity" interest
rate should be recovered on these CClngressional
appropriations, in the form of the prevailing aver41
age yield on long-term Treasury bonds at the time
the debt is incurred. 18 The author would go even
further and use the average rate paid by private
utilities for new bond issues. In other words, the
appropriate comparison should be between the Federal and private utility sectors, and not between the
Federal utility sector and the Federal government
sector in general. On that basis, the public would
earn as great a return on funds invested in the
Federal utility sector as it could earn from purehasingprivate-utility bonds. Over the 1948-81 period,
the average yield on Aaa public utility bonds ranged
from 2.6 to 15.6 percent.
2. Ability to pay: In a series oflaws passed in 1914
and 1926, Congress extended the repayment period
on irrigation projects from 10 to 40 years, 19 to help
provide relief to hard-pressed farmers during recession periods. Then, in the Reclamation Act of 1936,
Congress extended the repayment period to 50 years
and introduced the "ability to pay" concept. Under
that provision, farmers are required to repay only
that portion of irrigation water costs they can afford.
Their ability to pay (payment capacity) is measured
as a residual equal to the net increase in revenues
attributable to project water.
Table 1
Reconciliation of Realized and Imputed Unit Price for Central Valley Irrigation Water
Under the (Historical) Average Cost Method
(Dollars per acre/foot)
Fiscal Year!
Unit Price
As Realized 2
Actual
Operation &
Maintenance4
1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
2.83
3.48
1.93
2.37
3.32
2.14
3.33
1.88
7.87
2.47
4.16
3.47
4.46
3.29
3.70
4.20
6.84
5.41
4.02
5.83
4.55
5.09
2.40
3.72
1.78
1.95
1.12
2.29
3.89
1.52
.77
3.22
3.38
3.22
3.53
2.57
2.33
2.79
4.60
9.83
3.83
3.84
3.91
3.99
Imputed Costs3
Taxes Amortization
3.08
4.19
2.02
2.75
3.87
2.50
3.47
2.59
9.89
4.20
3.99
3.76
3.94
3.40
2.86
3.20
4.13
13.11
3.99
3.48
3.67
4.21
4.43
5.02
2.37
3.29
4.56
2.93
4.24
3.31
12.85
5.78
5.38
5.37
5.62
4.51
3.71
4.32
5.56
17.35
3.90
2.69
2.84
3.26
Imputed
Interest-Subsidized
Unit Price"
Imputed
Interest
Imputed
Full-Cost
UnitPrice6
9.91
12.93
6.17
7.99
9.55
7.72
11.60
7.42
23.51
13.20
12.75
12.35
13.09
10.48
8.90
10.31
14.29
40.29
11.72
10.01
10.42
11.46
3.96
5.48
2.59
4.18
5.96
3.78
5.16
3.78
19.68
8.21
7.71
7.20
7.37
6.84
5.67
7.03
9.14
29.73
8.92
7.62
9.66
12.31
13.87
18.41
8.76
12.17
15.51
11.50
16.76
11.20
43.19
21.41
20.46
19.55
20.46
17.32
14.57
17.34
23.43
70.02
20.64
17.63
20.08
23.77
Fiscal year ending June 30 until 1976, and ending September 30 for later years.
Derived for any given period by dividing revenues from water sales to irrigation districts, under 9-c water-service type contracts, as
reported by total sales to those districts. The recovery of costs associated with CVP-financed distribution systems under 9-d repayment
type contracts is excluded from this analysis. For derivation see Appendix B, Table I.
a For derivation of the various imputed-cost components, see Appendix B, Table 2. Note that 1968 and 1977 were drought years, i.e.,
years when water deliveries fell considerably, raising capital costs per unit. Also, in 1968 there was a large new investment in irrigation
capacity.
4 As reported by the Bureau of Reclamation. The Bureau is required by law to recover operation and maintenance costs incurred in
supplying irrigation water from the Central Valley Project. Yet in some years, the realized price did not even cover operation and
maintenance costs.
5 Excludes interest.
6 Derived on the basis of the average-cost pricing method, with costs determined on the basis of the original (historical) value of plant and
equipment, in keeping with traditional regulated utility practice. For derivation see Appendix B, Table 2 and technical notes.
!
2
42
3. Fixed-rate contracts: In contracts negotiated
before 1975, the Bureau established water rates for
each service area on an individual basis. That is, it
charged either an ability-to-pay price or a cost-ofservice figure for a service area's share of total CVP
costs, whichever was lower. 22 The Bureau also followed a standard practice of granting irrigation districts 40-year fixed rate contracts. But because of
this practice, the price realized during the life of the
contract failed to recover increased operational,
maintenance and new-facility costs.
In 1975, the Bureau introduced several modifications in its pricing policies for new contracts. It
began to utilize an average-cost pricing method, by
dividing total system costs for a given period by the
number of units expected to be sold. It also introduced adjustment clauses into its contracts to reflect
changes in costs. But these provisions called for rate
adjustments every five years to reflect only operation and maintenance costs, and every ten years to
reflect added capital costs. Finally, the first adjust-
More specifically, the Bureau detennines payment capacity by comparing the estimated gross
income from a representative small fann in a given
irrigation district under two different sets of dry and
irrigated fanning conditions. From the increase in
gross income attributable to project water, the
Bureau subtracts the increase in non-water costs
required to increase fann yields. These include
operating (variable) and capital (fixed) costs, plus a
projected rate of profit (return on investment) sufficient to encourage the fanner to increase fann
yields. 20 The Bureau then charges the irrigation
district the cost of service or ability-to-pay price,
whichever is lower. If the cost of service, excluding
interest, exceeds payment capacity, the remaining
costs are recovered from the sale of electric power
and municipal and industrial water. The Bureau
thus can legally shift a substantial portion of the
costs of supplying irrigation water to other beneficiaries of Federal water, beyond the costs shifted
through the initial cost-allocation process. 21
Table 2
Imputed Costs (Cumulative) as a Percent of Realized Unit Price l
1
2
Fiscal Year
Unit Price
As Realized
Operation
and Maintenance2
Including
Taxes
Including
Amortization
Including
Interest
1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
84.9
107.0
92.0
81.9
33.8
107.0
116.5
80.7
9.8
130.1
81.2
92.6
79.1
78.2
62.9
66.4
67.2
181.6
95.4
241.3
251.4
214.6
220.8
170.9
244.2
244.0
257.0
173.0
363.9
210.7
247.2
205.2
215.2
163.2
169.0
148.6
502.3
i92.4
112.0
148.5
142.4
350.2
371.7
319.0
336.5
287.5
361.0
348.0
395.2
298.7
533.8
306.7
355.6
293.6
318.6
240.5
245.1
208.9
744.4
291.8
171.6
229.0
225.1
490.1
529.4
452.7
512.5
466.7
538.1
502.7
596.5
548.7
865.9
492.4
562.7
459.0
526.5
394.0
412.5
342.5
1293.9
513.9
302.2
441.4
467.0
65.8
86.1
78.4
The imputed unit costs were calculated under the (historical) average-eost accounting method. In this table, each imputed cost item is
cumulatively added and expressed asa percent of realized price. For example, in 1981, the addition ofimputed taxes and amortization to
operation and maintenance costs equalled $11.46/acre-foot. This figure was 225 percent ofthe realized unit price, or 125 percent higher.
In some years, the Bureau of Reclamation realized a price greater than the cost of operation and maintenance alone. The imputed price
reflecting only that one cost would have been lower than the realized price.
43
ment was delayed until long after the initial delivery
of water. In 1981, the Bureau introduced further
reforms in this process, but the basic system still
had the same drawbacks as before.
4. Taxes: The Bureau of Reclamation pays no
local property taxes on lands occupied by the Central Valley Project. In contrast, private water utilities in California over the 1960-77
paid
annual property taxes averaging about 2.6 percent
of
total plant investment. Their property tax
then dropped to an average of 1.7 percent of
capital investment during the 1978-81
a
result of the passage of Proposition 13.
5. Amortization: The Bureau is required by
to
repay each dollar borrowed for investment in Federal irrigation facilities within 50 years after the first
delivery of water, but it has not repaid such borrowings on a systematic basis. With its low rates, in
fact, the agency frequently has failed to recover
even its annual operation and maintenance costs, as
required by law. And with its inadequate revenues,
the Bureau actually has extended the repayment life
for all CVP irrigation facilities each time new facilities have been added to the system. 23
The author has reestimated CVP irrigation costs
for the 1949-81 period on the basis of the methodology employed by privately-owned utilities. 24 The
adjustrpents for the "full cost" unit price, calcu-
lated under the historical average cost methodology, included the addition of an imputed property
tax and interest return on invested capital, as well as
the recalculation of amortization of Congressional
appropriations for the irrigation function. All these
costs were determined on the basis of capital investments valued at original prices. The author adopted
Bureau estimates of operation and maintenance
costs, and of the irrigation share of total CVP investment
1 and 2).
With adjustments made for imputed property
taxes, amortization and interest costs, the Central
Valley Project actuaUy incurred an average unit cost
of at least $23.77 per acre-foot of irrigation water in
1981, calculated on an historical accounting basis,
instead of the $5.09 per acre-foot actually realized
(Table 1). Had rates been raised to reflect this full
average cost, the price for CVP irrigation water in
1981 would have been 367 percent higher than the
amount actually charged (Table 2).
In any
year, the difference between the
imputed historic full-cost unit price and the price
actually realized represents the total subsidy paid by
the general taxpayer and electrical power users for
each acre-foot of irrigation water delivered. (In this
case, we used cost figures derived from plant and
equipment valued at original purchase prices.) Multiplying this subsidy by total acre-feet delivered, we
2
Central Val1ey Project Irrigation Water Costs
Under the Historical A
Cost Method, 1948-81
(Current ollars)
Price and Cost
(Dollars/Acre-foot)
100
Log Scale
50
As imputed
"
20
10
5
2
1
-
_
1948-60
_
1964
1972
1968
Annual Average
44
_...A...._...A...._--II
1976
1981
Table 3
obtain a total annual subsidy of $77 million for
1981, and of $966 million cumulative for the entire
1948~81period (Table 3).
The· interest subsidy is by far the largest single
contributorto the overall subsidy. In 1981, the subsidy artlounted to $51 million, or 66 percent of the
total subsidy. Over the entire 1948-81 period it
amounted to about $484 million or 50 percent of the
total subsidy.
Both the average realized and imputed price rose
over the post-World War II period (Table 4 and
Chart 2). Because of the use of the historical-cost
approach, both realized and imputed prices (especially the latter) trended downward over time in
constant dollars however (Table 4 and Chart 3).
Nevertheless, had the Bureau charged the higher
imputed price rather than the realized price, farmers
would have been encouraged to reduce their consumptionof irrigation water. Instead, irrigators
increased their annual deliveries of water from an
annual average of 0.78 million acre-feet during the
1948-60 period to 4.12 million acre-feet by 1981
(Appendix B, Table 1).
The average-cost figure of $24/acre-foot, as calculated by the traditional private-utility accounting
method, is still only a fraction of the estimated
long-run incremental cost of $324/acre-foot. This
does not necessarily mean that the Central Valley
Estimated Annual Subsidy to Users of
Central Valley Project Irrigation Water
(Historical Accounting Method)
Subsidy ($ Millions)
Fiscal Year
Interest'
Other"
Total"
Total. 1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Total,1948-1981
36.96
5.39
5.28
9.03
9.79
9.86
9.72
9.55
19.70
19.42
19.49
19.41
19.06
22.63
22.62
27.87
28.96
31.43
31.18
30.80
42.32
50.69
481.16
66.24
9.28
8.64
12.13
10.24
14.53
15.58
14.01
15.66
25.37
21.68
23.95
22.31
23.79
20.69
24.16
23.61
36.86
26.93
16.88
25.71
26.23
484.48
103.20
14.67
13.92
21.16
20.03
24.39
25.30
23.56
35.36
44.79
41.17
43.36
41.37
46.42
43.31
52.03
52.57
68.29
58.11
47.68
68.03
76.92
965.64
Derived by multiplying the imputed interest per acre-foot by the number
of acre-feet sold.
Derived by subtracting the realized price from the interest-subsidized
imputed price and multiplying by the total acre-feet of water sold.
o Derived by subtracting the realized price from the full-cost imputed price
(historical accounting basis) and multiplying by the total acre-feet of
water sold.
Source: Computed by the author.
1
2
Chart 3
Central Valley Project Irrigation Water Costs
Under the Historical Average Cost Method, 1948-81
(Constant Dollars)
Price and Cost
(Dollars/Acre-foot)
100
,
Log Scale
As imputed
50
20
10
5
2
1r-.._""-_......_ ....._ ....._ _r-.._....._ ....._ ....._ ....._ _.....
1948-60
1964
1972
1968
Annual Average
45
1976
1981
Project faces decreasing returns to scale and increasing long-run average costs. Those theoretical
concepts depict cost and output alternatives facing a
firm (or agency) at a moment of time under the
assumption of constant technology and factor prices
(Chart I-B). A firm would be operating in an output
range associated with increasing long-run average
costs if expansion to a larger scale plant (or system)
built from scratch today entailed higher average
costs than a smaller plant built today. In that case,
long-run incremental cost would be above average
cost, with both determined on the basis of plant and
equipment valued at today's prices. This contrasts
with the traditional private-utility practice of determining average cost. Under that method, long-run
incremental cost would be above average cost simply because of the failure of the utility industry's
average-cost methodology to reflect the effects of
inflation on equipment prices.
To determine whether the CVP may in fact be
facing increasing long-run average costs due to the
exhaustion of economies of scale, we have estimated the average cost of irrigation water with
capital costs valued at current replacement prices
rather than historical prices. Under the replacement
accounting method, we have valued the entire plant
in 1981 dollars, and have then compared the incremental cost of water from new plant with the average cost from the existing system, both valued at
today's prices (Table 5).
The aggregate value of annual plant investment
in 1981 dollars, $1. 7 billion, represents the replacement value of the entire system. After calculating
that value, we next calculated the average or unit
cost of irrigation water from this sytem by dividing
the total annual cost (capital, operation and maintenance, and taxes) by the number of acre-feet delivered. 25 This procedure yielded a $48/acre-foot
average replacement cost for irrigation water, and a
Table 4
Constant Dollar Realized and Imputed Unit Prices
(Dollars per acre-foot)
Fiscal Year
1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Current Dollar Unit Price
Imputed
Imputed
Unit Price
Interest-Subsidized
Full-Cost
As Realized
Unit Price
UnitPrice'
2.83
3.48
1.93
2.37
3.32
2.14
3.33
1.88
7.87
2.47
4.16
3.47
4.46
3.29
3.70
4.20
6.84
5.41
4.02
5.83
4.55
5.09
9.91
12.93
6.17
7.99
9.55
7.72
11.60
7.42
23.51
13.20
12.75
12.35
13.09
10.48
8.90
10.31
14.29
40.29
11.72
10.01
10.42
11.46
Producer
Price Index'
13.87
18.41
8.76
12.17
15.51
11.50
16.76
11.20
43.19
21.41
20.46
19.55
20.46
17.32
14.57
17.34
23.43
70.02
20.64
17.63
20.08
23.77
38.82
3918
39.08
39.15
39.13
39.46
40.55
41.44
42.23
43.56
45.35
46.67
48.02
50.60
56.82
65.41
70.20
74.44
79.96
88.04
100.00
109.23
Constant Dollar Unit Price"
imputed
Imputed
Unit Price
Interest-Subsidized
Full-Cost
Unit Price I
As Realized
Unit Price
7.28
8.88
4.95
6.07
8.49
5.42
8.22
4.53
18.64
5.68
9.17
7.44
9.29
6.50
6.51
6.43
9.74
7.27
5.02
6.63
4.55
4.66
, Calculated on the basis of the historical average-cost accounting method.
2
Producer price index. all finished goods. 1980= 100.
" Dollars per acre/foot in constant dollars; derived by dividing current dollar prices by producer price index.
46
25.53
32.99
15.79
20.41
24.42
19.56
28.62
17.91
55.67
30.31
28.11
26.46
27.26
20.71
15.66
15.76
20.35
54.12
14.66
11.37
10.42
10.49
35.72
46.99
22.41
31.09
39.63
29.15
41.33
27.03
102.28
49.16
45.13
41.88
42.62
34.23
25.65
26.52
33.37
94.06
25.81
20.02
20.08
21.76
total 1981 subsidy of nearly $175 million. 26 The
difference between this figure and the much higher
incremental cost suggests that the Central Valley
Project is operating in an. output range of increasing
long~run average cost, reflecting decreasing returns
to scale.
Economists would argue that, for efficient allo~
cation of resources, the Bureau of Reclamation
should price all CVP irrigation water on the basis of
long~run
incremental cost, estimated here at about
$324/acre-fooL A second-best option would be for
the Bureau to price water at least on the basis of the
estimated $48/acre-foot replacement average cost.
Only that approach would permit recovery of the
true. cost to society of the resources tied up in supplying water. Recovery of the estimated $24/acrefoot historical average cost, although an improvement over current Bureau practice, would not
recover this' 'true" average cost.
IV. Impact of Higher Water Prices
Would higher prices for CVP irrigation water
lead to a reduction in the quantity demanded? Some
analysts argue in the negative, because of agriculture's essential need for water. They maintain that a
given amount of water is required to produce a
given yield for any crop, that the amount is dictated
by soil and climatic conditions, and that it is invariant to higher water prices. Thus, because of the
limited technical substitutability between water and
other productive inputs,27 demand for irrigation
water is price inelastic, i.e., relatively unresponsive
to a higher (or lower) price. 28 Proponents of this
TableS
Imputed Unit Price of Central Valley Irrigation Water,
Replacement Average Cost Method
Fiscal Year
1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
Investment
in Irrigation Plant I
($ Millions)
Conversion Factor"
Investment
in Irrigation Plant
($ Millions, 1981 Dollars)
178.74
26.73
.41
90.82
21.83
6.62
1.66
1.19
166.96
2.15
6.17
4.83
1.48
53.12
6.13
64.31
2.73
2.30
2.30
2.29
2.28
2.26
2.22
2.16
2.09
2.03
1.96
1.88
1.82
1.79
1.67
1.44
487.96
61.47
.94
207.98
49.77
14.96
3.69
2.57
348.95
4.36
12.09
9.08
2.69
95.08
10.24
92.61
W~
I~
n.~
38.69
5.89
4.07
100.06
64.M
1.30
1.22
1.14
1.07
1.00
50.30
7.16
4.64
107.06
~M
866.57
1,665.69
Replacement Cost Per Acre-Foot (1981 Dollars) = (Capital Costa + Operation & Maintenance Cost + Property Taxes)/Acre-Feet of
Irrigation Water Delivered
($166.57 m. + $16.45 m. + $13.43 m.)/4.12 m.
= $47.69
1 Excluding distribution-system investment from canalside to farmgate under CVP, 9-c water service contracts.
2
Calculated by dividing the 1981 implicit price deflator for producer durable equipment by the actual deflator for each year.
a Capital cost of the entire project (1981 dollars) multiplied by a IO-percent real interest rate.
Source: Annual Investment Data: U.S. Bureau of Reclamation, Mid-Pacific Regional Office, Sacramento. Computation by author.
47
thesis consequently maintain that resorting to higher prices to allocate available supplies would be
ineffective.
Examination of the literature shows that the elasticity of demand for irrigation water varies according to the price range being considered. 29 That is,
the responsiveness of quantity demanded to a given
percentage change in price varies along a given
demand schedule. Early studies covering the 1950's
and early 1960's supported the argument for inelastic demand. 30
But those results were biased by the absente of a
wide range of observable prices over which to test
the demand responsiveness. More recent studies,
based either on observed water-use and price combinations or on linear-programming estimation
techniques replicating the cost-minimizing behavior of California farmers, show that water demand is
price elastic at a price of over $20 per acre-foot. One
of the latest studies, which utilizes an even more
advanced programming model of Central Valley
agriculture, shows an elasticity coefficient of -1.5
for a price range of $25-35/acre-foot. 31 For prices
above $35/acre-foot, the elasticity coefficient then
drops to -0.5.31
In summary, at water prices prevailing currently,
agricultural water demand is not very price responsive, but it should respond substantially at higher
prices. Indeed, given a demand of 3.8 million acrefeet at a price of $25/acre-foot, and given an elasticity coefficient of -1.5, a ten-percent price increase
would reduce the quantity demanded by 570,000
acre-feet. This would be sufficient to eliminate the
need for the proposed Auburn-Folsom project.
To maximize profit from any given crop on a
given parcel of land, a farmer will purchase and
apply additional units of water until its marginal
revenue product equals its price. 33 The "marginal
revenue product" refers to the net addition to total
revenue resulting from the last increment of water,
after subtraction of all other non-water operating
(variable) and capital (fixed) costS. 34
Farmers might react in three different ways to
sharply higher prices of CVP irrigation water. 35
Table 6
Indicators of Impact of Higher Water Price
on the California Agricultural Sector
Leading Crops (1980)1
Value of
Production
($ Thousands)
California as Percent
of Total Domestic Production
Water Cost as Percent
of Total Production Cost2
Cotton
Grapes
Hay
Tomatoes
Almonds
Rice
Lettuce
Wheat
Oranges
Strawberries
Sugar Beets
Peaches
Walnuts
Potatoes
Com, for Grain
1,389,342
1,215,585
723,316
490,310
473,340
423,612
382,563
357,945
224,548
201,266
182,930
176,438
168,300
157,590
151,268
28.2
91.6
5.9
79.2
95.0
24.3
74.4
3.6
18.9
75.3
24.8
66.4
95.0
6.2
0.5
2.72
1.49
4.81
0.78
1.29
1.55
0.96
4.82
2.72
4.31
2.76
0.34
1.39
2.26
3.24
I
Crop ranking based on value of production, 1980.
2
Water cost excludes cost of application. Total production costs include all variable and fixed cultural and harvest costs (including water
application), imputed rent on land and return to management. The latter two returns are actually part of profits and should not be
included in costs, but were included here because of lack ofrelevant data.
Source: California Department of Food and Agriculture, California Agriculture Statistical Review, 1980. Giannini Foundation of
Agriculture Economics, Agricultural Water Use and Costs in California, Information Series 80-2, Bulletin 1896.
48
First, if a given crop still represents the most profitableopportllnity on a· given parcel of land, the
farmer might continue to grow the same crop but cut
back water usage to the point where its marginal
revenue product equals the higher price. This cutback would be accompanied by SOme reduction in
crop output or introduction of more efficient irrigation methods. Alternatively, the farmer might shift
to another crop that yields a higher net return per
acre·foot of water, which could mean a shift away
from low-valued field crops to higher-valued specialty crops. Or again, the farmer might simply
withdraw land from irrigation, if irrigated crops fail
to yield a positive profit or if they yield less profit
than dry-land farming. In summary, if the price of
water is raised, farmers may react by reducing output, changing the input mix (for example, using
more capital intensive irrigation methods), and/or
shifting cropping patterns.
The elasticity of demand for irrigation water
varies significantly, depending On type of crop.
Price elasticity increases, in general, the greater is
the substitutability of other factor inputs for water,
and the greater is the water share of total production
costs. But elasticity varies inversely with the ability
of farmers to pass on higher water costs to cansumers in the form of higher food and fiber prices.
For all 15 of California's major crops, water
currently comprises a small percentage of total production costs·(Table 6). This reflects the low level
of currentwaterprices,and suggests why the price
elasticity of demand at current ranges is generally
low. Nevertheless, the data also show considerable
variability in the importance of water costs among
various products.
For certain field crops--especially, hay, wheat
and corn--water costs comprise a relatively large
proportion of total production costs. This factor
alone suggests that any given increase in water
prices would affect those products significantly. On
the other hand, water comprises a relatively small
percentage of total costs for tomatoes, peaches,
lettuce, grapes and nuts. In reaction to higher water
prices, farmers thus might tend to switch away from
field crops to specialty crops such as those.
California farmers also account for a relatively
small share of total domestic production of field
crops. For com,wheat and hay, for example, their
shares of the national market amount to only 0.5,
3.6 and 5.9 percent, respectively (Table 6). This
suggests a relatively elastic demand for water, due
to farmers' inability to influence the price of the
final product and thereby pass on higher water costs
to consumers. .
V. Summary and Conclusions
If efficiency of resource allocation were the only
criterion, the Bureau of Reclamation would price all
irrigation water from the Central Valley Project on
the basis of long-run incremental cost-the cost of
delivering an additional acre-foot of water from the
next scheduled block of new capacity. This
approach would be the most efficient because it
would make· customers aware of the cost of the
resources required to bring them additional water.
But this approach also would return huge annual
profits to the Bureau of Reclamation or to the U.S.
Treasury. This is because long-run incremental cost
is far higher than the system's average cost of irrigation watet, detennined as· econoIllists would· measure average cost orr the basis of replacement value
of plant. In addition, a switchover to strict incremental cost pricing could cause a major shrinkage in
both water usage and the size of California's agri. ... . •..
cultural sector.
The second-best option from an efficiency stand-
point would be pricing of irrigation water on the
basis of average cost, determined on a replacement
cost basis. This method would at least recover the
cost of the resources already embodied in the system, valued at today's prices. It would also enable
the Bureau to generate sufficient revenue to perpetuate the existing capacity. A case also could be
made for the Bureau to include all cost elements that
would be .incurred bya debt-financed private investor-owned utility-annual operation and maintenance costs,plus interest expense and property
taxes (opportunity-cost basis), plus steady amortization of capital bdrrowedfrom theU.s. Treasury.
Pricing on this basis would at least make customers aware of the resources already expended in the
system, and would provide fanners with a strong
incentive to reduce water consumption. In Iact,
empirical studies suggest that agricultural demand
is elastic above a retail price of around $20 per
acre-foot.
49
have expanded beyond the point where the net
return to the last unit of water, in terms of agricultural revenue, is equal to the cost of supplying that
extra unit. This suggests that more resources have
been devoted to the construction of the Federal
irrigation system in California than are warranted
by agricultural benefits.
The social objectives that justified the earlier
granting of subsidies-namely, the development of
the arid West-may no longer be appropriate.
Today's environment is dominated by intense competition for water among competing users-households, energy producers, and farmers. To some
observers, the correct policy issue remains the
perennial one-what size farms should get the subsidy? Should the 160-acre limit be enforced or expanded? Others would argue, however, that there
should be no water subsidy at all. In this view, the
focus should be on pricing reform, to improve the
efficiency of water usage through the use of more
efficient irrigation methods and shifts to less water
intensive crops. Indeed, Congress logically should
give more attention to the role of the price mechanism in reducing the projected growth of irrigation
water demand, not only in California but throughout the West.
When Congress passed the Reclamation Act of
1902, it clearly intended beneficiaries. of Federal
water projects to repay their original construction
costs as well as operating costs. It provided a subsidy, however, by not requiring repayment of interest on capital invested, and later increased that
subsidy through the ability to pay concept and other
measures. In some years indeed, irrigation water
rates have even failed to recover operation and
maintenance costs, despite the legal requirement to
do so. Moreoever, had the Bureau priced irrigation
water to reflect all costs measured on a replacement
accounting basis, the average realized price by 1981
would have been nearly ten times as high as the
price actually realized.
By charging highly subsidized rates for Federal
irrigation water, the Bureau has spurred the growth
of consumption beyond the growth that would otherwise have occurred had it priced water to reflect
the true average cost of service. Had it priced water
on the basis of long-run incremental cost,. the
amount of resources devoted to the construction of
Federal irrigation projects in California would have
been still smaller. Instead, the consumption of
water and the size of the Federal irrigation system
Appendix A:
Calculation of Incremental Cost of Irrigation Water,
Proposed Auburn - Folsom South Unit
The following technical note describes the methodology and assumptions employed by the author to
estimate the incremental cost of irrigation water from the proposed Auburn-Folsom South Unit. The
incremental cost of irrigation water from this project includes the capital costs (depreciation and interest),
taxes, and operation and maintenance expenses to be recovered over the project's life. Each of these costs is
expressed as a stream over time and then discounted to determine present value. (However, the present value
of the capital costs can be shown to be equal to the initial construction costs). The annual cost is then equal to
the real interest rate multiplied by this present value. We add these annual costs and then divide by the
expected average annual output inacre-feet.
The real interest rate is used to discount capital costs. Taxes and operation and maintenace costs are
discounted by the nominal interest rate, because they are assumed to reflect inflation in the future. We
assumed both the annual property tax and depreciation rates to be 2 percent of plant value, based on the 1981
tax rate and on a 50-year service life for plant and equipment. We used a real discount rate of 10 percent and a
nominal interest rate of 16 percent as the discount factors.
Variable List:
r
Real Interest Rate
= Initial Capital Outlay
k
Nominal Interest Rate
Stream of Capital Costs
K
Annual Operation & Maintenance Costs
IT
Inflation Rate
om
Depreciation Rate
OM
Stream of Operation & Maintenance Costs
Tax Rate
Stream of Tax Costs
t
T
Annual Water Production
Incremental Cost
p
CI
o
50
Each of the cost streams can be expressed as follows:
K
T
OM
= (r+S)k, (r+ S)k(1-S),(r+S)k(1-S)2 ,(r+ S)k(l-S)3
= tK+tK(1-S),tK(1-S)2,tK(1-S)3, . . .
= om(l +II),om(1+IW ,om(1 + IT)3 ,
Taking the present value of each cost stream:
(r+8)k
= -l+r
-+
PV(K)
tk
0+
PV(T)
PV(OM)
=
(r+S)k(l-S) (r+S)k(I-8)2 (r+S)k(l-6i
+
+
+.
(l+r)2
(l+r)3
(l+r)4
k
tk(l-S) tk(I-6)2 tk(l-6)3
+ (l+i)2 + (1+i)3 + (1+it +
tk
H6
am + om(l + II) om(l + IT)2 om(l + IT)3
(1+i)
(l +i)2 + (1 +i)3
+ (1 +i)4
+
=
am
i-II
am
= -r-
Multiplying the present value of each cost stream by the real rate of interest and dividing by production of
440,000 acre-feet yields an incremental cost of$324/Acre Foot of Water
CI
=
(Real Interest Rate (Capital Cost + Taxes + Operation and Maintenance) )/Annual
Water Production
tk
am
(r(k +~ + - ) )/p
l+u
C
1
C1
r
6
6
=
(.10(1217X106 +24.34X10 + 7.6X10
.16+.02
.10
=
$324/Acre-Foot
)
)/440x103
Appendix B:
Adjustment of Central Vaney Project's Realized Average Price
(Private Utility Basis)
was calculated by applying the average life of service of the equipment to the total value of the plant
in service, measured on an historical (original) cost
basis. This amortization procedure follows that
used by most private utilities. The average service
life of the CVP's total irrigation plant is estimated to
be 75 years. For any given year, depreciation thus
was calculated as 1/75th of the total value of irrigation plant in service. Since depreciation is calculated on a 75-year basis, compared with the
Bureau responsibility to recover borrowings within
a 50-ye peri()d, depreciation charges thus. calculated would fall short of meeting the CVP's
payment responsibilities. A reconciliation charge
therefore was calculated, representing the difference between l/50th and 1/75th of the value of the
plant in service.
The following technical note describes the methodology used by the author to adjust the Central
Valley Project's realized average price for irrigation
water delivered at canalside for the 1948-81 period,
to include the major cost items and historical accounting methodology employed by private-owned
water utilities. The realized and imputed prices appear in Appendix B, Tables I and 2 respectively.
Taxes: Annual
were imputed by applying the average property-tax rate for
two privately-owned California water utilities in
any given year to the Central Valley Project's total
irrigation plant in service of that year, valued on
an historical cost basis.
Amortization: Amortization costs were imputed
annually for the 1948-81 period by developing a
straight-line depreciation schedule. Depreciation
ar
51
interest rate, plus any unamortized old debt multiplied by the rate(s) in effect when the debt was
incurred. The first debt was assumed to be incurred
in 1948, the earliest date for which data were available. Each increment in debt was amortized on a
straight-line basis by 1/50 each year after it was
incurred, in line with the 50-year payback period
specified by law. Note that Moody's Investor Service refers to private investor-owned utilities as
public utilities, using that term in a general sense.
A consistent series showing annual Congressional appropriations to the CVP was not available.
A proxy for "new debt" was developed by taking
the total value of the plant in service, i.e., the
capital stock, and calculating the annual change,
i.e., the new investment added each year. That
proxy was used under the assumption that borrowing was for capital investment.
Interest: Interest payments on an opportunitycost basis were imputed for any given year n by the
formula:
n
= 2: i y A y
Y = 1948
where: P n = total interest payment in year n
iy = Moody's Aaa interest rate on
Pn
Ay =
public
(private investor) utility issues in
year y
unamortized portion of appropriations
received in year y as of year n
This formula simply states that total interest payments in any given year Pn' equal the sum of all
interest payments on outstanding CVP debt in that
year. In other words, total interest payments equal
new debt (for irrigation plant) times the prevailing
Appendix B, Table 1
Average Price for Central Valley Irrigation Water, As Realized
Fiscal Year
Revenues'
Water Sales 3
Realized
UnitPrice4
1948-1960 1
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
2.20
3.42
3.95
5.13
5.46
5.57
6.28
4.75
7.88
5.85
10.49
9.37
11.52
10.89
14.73
16.65
21.68
5.72
14.04
23.58
19.93
20.99
.78
.98
2.04
2.16
1.64
2.60
1.88
2.53
1.00
2.36
2.52
2.70
2.58
3.31
3.98
3.96
3.17
1.06
3.50
4.04
4.38
4.12
2.83
3.48
1.93
2.37
3.32
2.14
3.33
1.88
7.87
2.47
4.16
3.47
4.46
3.29
3.70
4.20
6.84
5.41
4.02
5.83
4.55
5.10
1
Annual average computed from cumulative totals for the 12-year period 1948-1960.
2
Millions of dollars. Revenues from irrigation sal~s u~der 9-c water-service contracts as reported by the U.S. Bureau of Reclamation.
3
Millions of acre-feet.
4
Dollars per acre-foot. Derived by dividing revenues from water sales to irrigation districts by acre-feet of water sold.
Source: Revenues and sales data from U.S. Bureau of Reclamation, Mid-Pacific Regional Office. Realized unit price derived from that
data by author as described in footnote 4.
52
Appendix 13, Table 2
Average Price for Central Valley Irrigation Water,
As Imputed on a Private-UtmtyCost Basis 1
(Cost data in millions of dollars)
Variable
Costs
FlscalYeer
Operation
Maintenance
Property
Tax'
Depreciation"
Fixed Costs
Reconciliation
DeprecIation 8<
Amorllzatlon'
1948-1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1.87
3.65
3.63
4.20
1.85
5.96
7.32
3.83
.77
7.61
8.52
8.68
9.12
8.51
9.26
11.05
14.56
10.39
13.40
15.53
17.15
16.45
3.45
4.93
4.84
7.12
7.49
7.64
8.01
8.36
12.87
13.67
13.59
14.48
14.52
14.91
14.79
17.09
17.65
18.35
13.61
10.89
12.44
13.43
1.60
2.74
275
3.96
4.25
4.34
4.36
4.37
6.60
6.63
6.71
6.77
6.79
7.50
7.58
8.44
8.72
9.23
9.31
9.36
10.70
11.55
.80
1.37
1.37
1.98
2.12
2.17
2.18
2.19
3.30
3.31
3.36
3.39
3.40
3.75
3.79
4.22
4.36
4.62
4.65
4.68
5.35
5.78
Total Costs
ExcludIng
Interest
In_I'
Full Costs'
7.72
IM9
12.59
17.26
15.70
20.10
21.87
18.76
23.54
31.22
32.17
33.32
33.83
34.68
35.43
40.81
45.28
42.58
40.97
40.46
45.63
47.21
3.08
5.39
5.28
9.03
9.79
9.86
9.72
9.55
19.71
19.42
19.49
19.41
19.06
22.63
22.62
27.87
28.97
31.43
31.18
30.80
42.32
50.69
10.80
18.08
17.86
26.28
25.49
29.96
31.59
28.31
43.24
50.64
41.66
52.73
52.89
57.31
58.05
68.68
74.25
74.01
72.15
71.26
87.95
97.91
InterestFull-Cost
SUbsIdized
UnitPrlce
UnltPrlce
lDOllanl/acre-'oot)
9.91
12.93
6.17
7.99
9.55
7.72
11.60
7.42
23.51
13.20
12.75
12.35
13.09
10.48
8.90
10.31
14.29
40.29
11.72
10.01
10.42
11.46
13.87
18.41
8.76
12.17
15.51
11.50
16.76
11.20
43.19
21.41
20.46
19.55
20.46
17.32
14.57
17.34
23.43
70.02
20.64
17.63
20.08
23.77
1
These costs represent the author's interpretation of the amounts that should have been recovered directly by the U,S. Bureau of Reclamation in the form of revenues for water
delivered to irrigation districts if the Central Valley Project had been operating as a private investor~owned utility. using the historical average~cost accounting method to determine
unit price. The costs consist of the variable costs as actually measured and reported by that agency. plus computationsoffix.ed costs to include imputed property-tax payments, interest
charges reflecting the opportunity cost of capital. and a straight·line depreciation and amortization charge to repay all outstanding debt on a consistent and continuous basis,
:l
Derived by applying an estimated California property·tax rate for private investor·owned water utilities (property taxes paid as a percentage of total plant in service) to the Central
Valley Project's total irrigation plant (excluding CVP·financed distribution facilities from irrigation districts to the fann gate), valued on an historical-cost basis.
:1
Private water utilities recover their long-tenn borrowings for capital investment through their depreciation charges. The average service life of the Central Valley Project's total
irrigation plant is estimated to be 75 years. Straight·line depreciation has been used so that depreciation is 1/75th of the total value of the plant in service. measured on an
historica!·cost basis.
4
Depreciation is calculated on an average 75·year basis, whereas the Central Valley Project is required to amortize (pay back) its horrowings within a maximum of 50 years, The
.'reconciliation" charges represent the difference between 1/50th and 1175th of the value of irrigation plant in service.
~, Derived on an "opportunity cost" basis: total interest payments in each year equal the product of new debt and the current Moody's average Aaa interest rate for public (private
investor~owned) utilities,
plus the product ofold amortized debt and the interest rate in effect when the debt was incurred. Debt is reduced (amortized) on a straight~line basis by 1/50th
each year after it is incurred. Total value of irrigation plant in service was used as a proxy in detennining outstanding debt, under the assumption that borrowing was for capital
investment.
Ii
Purchasers of Central Valley Project irrigation water have been allowed an interest subsidy by law (Le., the Bureau of Reclamation is not required to recover through its rates any
interest on public funds appropriated by Congress for Central Valley inigation projects), Therefore, we calculate a price without interest (i.e" an interest~subsidized imputed price) in
addition to the full-cost unit price. Prices are in dollars per acre·foot.
Source: For data pertaining to the private·utiJity sector: Moody's Investors Services. Moody's Public Utilities Manual. Average property·tax rates per year derived from data for
California Water Service Company and Southern California Water Company. For reported data pertaining to the Central Valley Project. U.S, Bureau of Reclamation. Mid
Pacific regional office, Sacramento.
FOOTNOTES
deep. The measure is equal to 325,851 gallons of water.
California's loss of Colorado River water will occur as a
result of a
Arizona is entitled to over one-half of the Colorado
that has been corning to California, The diversion will
place as soon as the Central Arizona Project is completed,
making the re-routing possible.
1. For the most comprehensive recent assessment of the
long-term outlook for U.S. and California water supplies
and demands see, U.S. Water Resources Council,The
Nation's Water Resources 1975-2000 (Washington,
D.C.: U.S. Water Resources Council, 1978). See especially, Volume 3: Analytical Data Summary and Volume 4:
California •Region, pp. 17-30. Also, Governor's Commission to Review California Water Rights Law, Final
Report (Sacramento, December 1978).
3. The literature On the pricing of water is relatively sparse
compared with thatfor other important resources such as
energy and non-fuel minerals. Important. contributions
include: Jack Hirshleifer, James C. De Haven and Jerome
W. Milliman, Water Supply Economics, Technology and
Policy (Chicago: The University of Chicago Press, 1960);
2. Southern California refers to the area south of the
Tehachapi Mountain Range, the natural barrier that sets the
south apart from the rest of the state. An acre-foot of water
is the amount of water required to cover one acre one foot
53
amount. Incremental cost refers to the average additional
cost of a larger finite addition to production. Since rate
changes are relatively infrequent, additions to output where
costs must be recovered are of an incremental rather than
marginal magnitude.
Joseph Bain, Richard Caves and Julius Margolis, Northern
California's Water Industry (Baltimore: Resources for the
Future, 1966); Charles E. Phelps, Morlie H. Graubard,
David L. Jaqquette et.al., Efficient Water Use in California (Santa Monica: Rand Corporation, November
1978); and Donald Erlenkotter, Michael Haneman, Richard
E. Howitt and Henry J. Vaux, Jr., "The Economics of Water
Development and Use in California," California· Water
Planning and Policy, Selected Issues (Berkeley: University of California, June 1979), pp. 169-207.
9. The cost curves for an individual firm are drawn under
the assumption that the firm has no influence on the prices
of the factors of production it uses. Internal economies
therefore are those enjoyed by a firm apart from any change
in factor prices. When an industry as a whole expands its
output, the prices of factor inputs may be affected. External
economies affect the slope of the industry supply curve.
4. For a discussion of the social objectives embodied in
early reclamation law, see E. Phillip Le Veen, "Reclamation
Policy at the Crossroads," Public Affairs Report, Vol. 19
(Berkeley: Institute of Governmental Studies, October
1978). Also, Alan R. Dickerman, George E. Radosevich
and Kenneth C. Nobe, Foundation of Federal Reclamation Policies; an Historical Review of Changing Goals
and Objectives (Fort Collins: Colorado State University,
1968); William E. Warne, The Bureau of Reclamation
(New York: Praeger Publishers, 1973).
10. This chart, to emphasize, depicts the economic model
of decreasing long-run average costs that originally characterized the operations of individual utility firms and led regulators to prescribe average rather than incremental-cost
pricing. The cost schedules shown in Chart 1A depict the
behavior of long-run average and incremental costs at a
given point in time. Capital costs-I.e., amortization and
interest-are determined on the basis of the current cost of
plant and equipment valued at the time of the planning
decision. This conforms with the economist's definition of
average and incremental cost. Although this model provided the rationale for pricing on the basis of average cost,
regulatory commissions have prescribed the historical accounting method for valuing plant and equipment. This
method differs from the economic model in that average
costs are determined on the basis of plant valued at original
cost. As we shall see, most utility commissions continue to
prescribe average-cost pricing even though utilities are
currently characterized by increasing long-run incremental
costs, even in the static sense as defined in economic
theory.
5. For a description of the average-cost pricing methodology followed by private investor-owned water utilities in
establishing the level of rates, see American Water Works
Association, Water Rates Manual (Denver: American
Water Works Association, 1972).
6. For private investor-owned utilities, the return on invested capital consists of three components: 1) interest
payments on bonded indebtedness, 2) dividends on preferred stock, and 3) a return to common-equity holders, a
residual amount which becomes available to these owners
only after all other legitimate claims of the company have
been settled. The first two are specified on the bond indenture and the preferred-stock certificates. At present, Federal reclamation law does not require the Bureau of Reclamation to recover any return on long-term borrowings for
investment in the Central Valley Project irrigation system.
We argue in this article, however, that reclamation law
should be changed to require the return of interest to the
U.S. Treasury for funds appropriated for such investment,
and that the rate of interest should be determined on an
opportunity-cost principle.
11. For a physical description of this project see, U.S.
Department of the Interior, Bureau of Reclamation, A
Financial Analysis of the Authorized Central Valley
Project, Past, Present, Future (Sacramento: Bureau of
Reclamation, May 1972), pp.7-9. For a summary of the
official cost-benefit analysis of the project see, U.S. Department of the Interior, Water and Power Resources Service,
A Summary of Economic Reanalysis Related to the
Auburn-Folsom South Unit Central Valley Project, California (Sacramento: Water and Power Resources Service,
September 1980). This analysis was challenged by the U.S.
General Accounting Office, Federal Charges for Irrigation Projects Reviewed Do Not Cover Costs (Washington, D.C.: Comptroller General of the United States, 1981),
pp. 23-27 and 44-72.
7. For proof that marginal-cost pricing of all goods and
services leads to optimum welfare, see Edward Berlin,
Charles J. Cicchetti and William J. Gillen, Perspective on
Power, A Study of the Regulation and Pricing of Electric Power, A Report to the Energy Policy Project of the
Ford Foundation (Cambridge: Ballinger Publishing Company, 1975), pp. 127-130.
12. In its cost-benefit analysis of the Auburn-Folsom South
Unit project, the Bureau of Reclamation estimated the average annual output of the project to be 550,000 acre-feet of
irrigation water annually. Discussion with the staff of the
General Accounting Office and the California Department
of Water Resources indicated that the 440,000 acre-foot
estimate is more realistic.
8. In a perfect-competition model, there is one situation in
which short and long-run marginal (incremental) costs are
equal-that is, in long-run competitive equilibrium. In this
situation, plant capacity has been adjusted to its optimum
size for achieving a given level of output. It is assumed that
a firm starts from scratch in planning its optimal-size production facility. In reality, this optimum is never realized.
Instead, firms operate with plants of various ages, and must
make decisions with regard to adding new capacity, either
for replacement or growth purposes. Pricing on the basis of
short-run costs would not necessarily recover the capital
costs associated with this new plant.
13. These figures are as of the beginning of 1982 (January
1, 1982) and therefore really reflect costs as of 1981. See,
U.S. Department of the Interior, Bureau of Reclamation,
Project Data Sheet (Sacramento: Bureau of Reclamation,
January 1, 1982). In developing our estimate of long-run
incremental cost, we subtracted out the estimated distribution cost from canalside to farmgate.
Marginal cost, strictly speaking, refers to the additional cost
of supplying a single, infinitesimally small additional
54
example, the return to equity owners was excluded because the Central Valley Project is financed solely through
Congressional appropriations.
14. For the methodology for determining the long-run incremental cost of water, see Hirshleifer et. aI., op. cit.
pp. 152-165. Due to the absence of additional literature on
long-run incremental costs, the author had to use literature
available in the electric utility area. See, for example,
Charles R. Cichetti, William G. Gillen and Paul Smolensky,
The Marginal Cost and Pricing of Electricity: An Applied Approach (Cambridge: Ballinger Publishing Company, 1977; Charles R. Scherer, Estimating Electric
Power System Marginal Costs (Amsterdam: North HoIland Publishing Company, 1977; and Ralph Turvey, Optimal Pricing and Investment in Electricity Supply, An
Essay in Applied Welfare Economics (Cambridge: Massachusetts Institute of Technology, 1968).
25. Capital costs equal the present value of the total investment in plant, valued in current dollars. Again, we used a
real discount rate of 10 percent for this calculation.
26. The difference between this replacement cost estimate
($47.69/acre foot) and the average price actually realized
by the Bureau for CVP irrigation water in 1981 ($5.09/acrefoot), multiplied by the number of acre-feet of water delivered (4.12 million acre-feet) yields an estimated total subsidy
to irrigators of nearly $175 million for that year.
27. For a discussion of this view, characterized as the
"water-is-different syndrome," see Maurice Kelso. "The
Water is Different Syndrome, or What is Wrong with the
Water Industry?" Paper Presented at the Third American
Water Resources Conference, American Water Resources
Association, San Francisco, California, 1967.
15. Office of Management and Budget, Executive Office of
the President, "Discount Rates to be Used in Evaluating
Time-Distributed Costs and Benefits," Circular No. A-94,
Revised (Washington, D.C.: Office of Management and
Budget, March 27, 1972). The 10-percent real discount rate
called for in this policy memorandum is still in effect.
28. The formula for arc elasticity of demand is percentage
change in quantity divided by percentage change in price.
The resultant numerical value is the coefficient of price
elasticity. When the elasticity coefficient exceeds one, demand is said to be elastic. When the value of the elasticity
coefficient is less than one, demand is said to be inelastic.
And when the value of the coefficient is one, demand has
unitary elasticity.
16. The General Accounting Office estimated the long-run
incremental cost of irrigation water from the Auburn-Folsom
South Unit project at canalside to be $73.17 in 1978, with
interest at 7V2 percent. Our figure is much higher because
of the use of later cost data and methodological changes.
See U.S. General Accounting Office, Federal Charges for
Irrigation Projects Reviewed Do Not Cover Costs, op.
cit., page 58.
29. For a summary of this literature see, Larry D. Schelhorse, et. aI., The Market Structure of the Southern California Water Industry (La Jolla: Copley International Corporation, June 1974, pp. 167-175.
17. For affirmation of this point see, Ibid, pp. 1, and 9-10.
Also, E. Phillip Le Veen, op. cit., page 1 and U.S. General
Accounting Office, Reforming Interest Provisions in
Federal Water Laws Could Save Millions (Washington,
D.C.: Controller General of the United States, October 22,
1981), page 5.
30. For example, based on a cross-section sample of 38
irrigation districts in California in 1958, Bain estimated a
price elasticity of demand of -0.64. See Joseph Bain, et.
aI., op. cit., page 176.
18. The U.S. General Accounting Office reached this conclusion in a recent study of the interest subsidy. Reforming
Interest Provisions in Federal Water Laws Could Save
Millions, op. cit., page 19. That agency argues for the use
of the "constant maturities yield" rate series.
31. This estimate was reached by Richard E. Howitt, William D. Watson and Richard M. Adams, "A Reevaluation of
Price Elasticities for Irrigation Water," Water Resources
Research (August 1980), page 623. These authors used a
quadratic programming model.
19. U.S. General Accounting Office, Federal Charges for
Irrigation Projects Reviewed Do not Cover Costs, op.
cit., pp. 4-10.
32. Ibid.
33. For a detailed analysis of the concept of the demand for
irrigation water see, Joseph Bain, et. aI., op. cit., pp. 675686. These authors refer to marginal revenue product as
"net value of marginal product."
20. Ibid, pp. 10-11 and 15-19.
21. The Bureau of Reclamation uses the "separate costsremaining benefits" method of allocating total project costs
to various functions, Le., water, electric power, flood control, etc. Critics charge that the Bureau does not allot a
sufficient proportion of total costs to the irrigation function.
For a description of that process see, U.S. Department of
the Interior, Office of Audit and Investigation, Review of the
Central Valley Project Bureau of Reclamation (Washington,. D,C,: Office of Auditan<:llnvestigation; January
1978), Appendix III. For a critical evaluation see, Ralph
Nadar Associates, Damming the West.
34. The Bureau of Reclamation uses a marginal principle
in calculating payment capacity but calculates the measure
incorrectly. The agency determines the additional gross
revenue attributable to a new increment of irrigation water,
but the agency then incorrectly subtracts out all additional
variable and fixed· costs plus a return on investment to
arrive at a residual value that represents the amount the
farmer can afford to pay for water. As correctly measured,
marginal revenue product should be the additional profit
yielded by the last increment of water. Profit should not be
treated as a cost, as is the practice ofthe Bureau. Instead, it
should constitute the residual value that measures ability to
pay. By treating the return to management and investment
as a cost, the Bureau underestimates "ability to pay."
22. U.S. Department of the Interior, Water and Power
Resources Service, Central Valley Project Water Service
Rate Policy (Sacramento: January 8, 1981), pp. 1-3.
23. U.S. Department of the Interior, Office of Audit and
Investigation, op. cit., page 63.
35. Joseph Bain, et. aI., op. cit., pp. 679-681.
24. Major cost items were included only if appropriate. For
55
@FedFRASER