Full text of Review (Federal Reserve Bank of Dallas) : Second Quarter 2001

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

•••••••••••••••

•••••••••••••

c. the Sf
Tell Bank Supervisors
Anything They Don't Already Know?
Jeffery W Gunther. Mark E. Levonian,
and Robert R. Moore

The Democratization of America's
Capital Markets
John V Duca

The Transition to Consumption
Taxation, Part 2: The Impact on
Existing Financial Assets
Alan D. Viard

This publication was digitized and made available by the Federal Reserve Bank of Dallas' Historical Library (FedHistory@dal.frb.org)

[eonomie dnd
findneidl Review
Federal Reserve Bank of Dallas

Robert D. McTeer, ]r.
President and Chiif!f Executive Officer

Helen E. Holcomb
First Vice President and
Chief Operating Officer

Robert D. Hankins
Senior Vice PreSident, Banking Supervision

Harvey Rosenblum
Senior Vice President and Director ofResearch

W. Michael Cox
Senior Vice President and ChiefEconomist

Editors
Stephen P. A. Brown
Senior Economist and Assistant VIce President

John V. Duca
Senior Economist and Vice President

Jeffery W. Gunther
Research Officer

Mark A. Wynne
Senior Economist and Assistant Vice President

Director ofPublications
Kay Champagne
Associate Editors
Jennifer Afflerbach
Monica Reeves
Art Director
Gene Autry
Design & Production
Laura ]. Bell
Economic and Financial Review (ISSN 1526-3940),
published quarterly by the Federal Reserve Bank of
Dallas, presents in-depth information and analysis
on monetary, financial, banking, and other economic
policy topics. Articles are developed by economists in
the Bank's Economic Research and Financial Industry
Studies departments. The views expressed are those of
the authors and do not necessarily reflect the positions
of the Federal Reserve Bank of Dallas or the Federal
Reserve System.
Articles may be reprinted on the condition that the
source is credited and the Public Affairs Department is
provided with a copy of the publication containing the
reprinted material.
Subscriptions are available free of charge Please
direct requests for subscriptions, back issues, and
address changes to the Public Affairs Department,
Federal Reserve Bank of Dallas, p.O. Box 655906,
Dallas, TX 75265-5906; call 214-922-5254; or subscribe
via the Internet at www.dallasfed.org. Economic and
Financial ReView and other Bank publications are
available on the Bank's web site, www.dallasfed.org.

Contents
Cdn t~e Stoc~ Mdr~et Tell
Bdn~ Supervisors Anyt~ing
T~ey Don't Alreddy Know?
Jeffery W. Gunther, Mark E. Levonian,
and Robert R. Moore
Page 2

T~e Demoudtizdtion of
Hmericd'~ Cdpitdl mdr~et~
John V. Duca
Page 10

T~e Trdnsition to Consumption
Tdxdtion, Pdrt 2: T~e Impdct on
[xisting findncidl Assets
Alan D. Viard
Page 20

This article provides evidence consistent with recent policy
proposals calling for a greater role for market forces in promoting
a safe and sound financial system. The authors' empirical results
indicate a measure of expected default probability distilled from
equity prices helps predict the financial condition of individual
banking organizations, as reflected in their supervisory ratings.
Moreover, the stock market data have predictive power over and
above the information in the quarterly financial statements available
to supervisors between inspections. These findings suggest financial markets can provide useful information to supplement supervisory assessments, particularly between inspections, and point to
the value of additional research to further clarify the information
content of market prices and quantities.

In this article, John Duca shows how financial innovations
have benefited the United States by increasing the availability of
financing for new firms and improving Americans' access to financial investments. Two dramatic examples are the explosive growth
of venture capital financing and the doubling of stock ownership
rates since the early 1980s. This democratization of America's capital markets stems from technological improvements that have cut
the transaction and information costs of investing and from a series
of deregulatory steps aimed at improving the availability of capital.

Replacing the income tax with a consumption tax is likely to
reduce the total value of the capital stock. Alan D. Viard reviews
how this decline is divided between bondholders and stockholders
and the effect on household borrowers and lenders. He explains
that the results depend on whether monetary policy accommodates
the tax through a higher price level. Without accommodation, the
decline in the value of capital is largely borne by stockholders and
there is little reallocation of wealth between household borrowers
and lenders. If the tax is fully accommodated, bondholders bear
heavier burdens than stockholders and household borrowers gain
at the expense of household lenders.

Various initiatives have been pursued in
recent years to enlarge market forces’ role in
promoting a safe and sound financial system.
Faced with dramatic increases in the size, scope,
and complexity of banking organizations, policymakers have increasingly considered the possibility that the forces determining prices and
quantities in the financial markets might be harnessed to supplement supervisory efforts aimed
at maintaining safety and soundness.
A primary example of the increased emphasis on market forces is the comprehensive
approach to capital adequacy recently developed
by the Basel Committee on Banking Supervision.
The new framework rests on three pillars—minimum capital requirements, supervisory review,
and market discipline. By including market discipline, the committee recognizes that market
forces can reinforce capital regulation and other
efforts to promote safety and soundness.
Another example of the new emphasis is
the Gramm–Leach–Bliley Act of 1999, which
directed the Federal Reserve Board and the
Treasury secretary to assess the appropriateness
and value of requiring large depository institutions to issue subordinated debt. While the resulting study did not recommend the immediate
establishment of such a requirement, the Board
and the Treasury nevertheless concluded that
the evidence supports use of subordinated debt
both in supervisory monitoring and to encourage market discipline.1
This article considers only one of several
avenues through which market forces might be
used to support safety and soundness —the idea
that investors’ views on the financial condition
and prospects of banking organizations can be
distilled from stock prices and that such views
can provide a useful supplement to supervisory
assessments. Despite its intuitive appeal, insufficient analysis has been undertaken to document
the empirical content of this basic idea. As a
result, controversy remains over whether the
financial markets can say anything about the
health and prospects of financial institutions
that supervisors do not already know.
Our empirical work uses supervisory
ratings as a benchmark for banking organizations’ financial safety and soundness, under
the assumption that the results of supervisory
inspections accurately reflect the financial condition of individual organizations. If after an inspection bank supervisors know everything about
an organization’s financial condition that investors
know, and perhaps more, the question becomes
whether market data can provide incremental
information in the periods between inspections,

Can the Stock Market Tell
Bank Supervisors Anything
They Don’t Already Know?
Jeffery W. Gunther, Mark E. Levonian,
and Robert R. Moore

S

tock prices provide useful

predictive information, even
after taking into account past
rating information and
information from the
quarterly financial statements
banking organizations file
between inspections.

Jeffery W. Gunther is a senior economist and research
officer and Robert R. Moore a senior economist and policy
advisor in the Financial Industry Studies Department of
the Federal Reserve Bank of Dallas. Mark E. Levonian is a
vice president in the Banking Supervision and Regulation
Division of the Federal Reserve Bank of San Francisco.

2

FEDERAL RESERVE BANK OF DALLAS

beyond that offered by past inspection results
and regularly reported data. It is important to
note that the issue here is not which of these
sources of information is better or more accurate. To be valuable, market indicators need not
be superior to standard supervisory indicators.
They just have to add a new perspective or
dimension that helps provide a more complete
picture of an institution’s financial health, as
Flannery (forthcoming) suggests. The tests reported in this article address this issue.
We find that a measure of financial viability based on stock prices helps predict the
financial condition of individual banking organizations, as reflected in their supervisory ratings.
Moreover, this measure provides useful information beyond that of past inspection results
and quarterly financial statements. To the extent
that these data are a reasonable proxy for the
full set of information supervisors use between
inspections, these findings indicate the financial
markets can provide useful information to supplement supervisory assessments. The equitymarket data give the right signals —or at least
they are in broad agreement with subsequently
assigned supervisory ratings —and they appear
to contain new, or more timely, information not
reflected by financing accounting statements.

insurance severely restricts or altogether eliminates the downside risk for depositors, leaving
a substantial degree of that risk with the deposit
insurance fund. However, models have been
developed to account for these factors.
One prominent model is the option-based
framework developed by Merton (1974, 1977).
This model relies on the fact that under limited
liability, equity is equivalent to a call option on
the issuer’s assets. With the analogy to options,
the technology of option pricing can be brought
to bear, and information on investors’ implicit
views of risk can be extracted from stock prices.
This model has often been used in the banking
context.2
KMV LLC has commercially implemented a
variant of this model and incorporated proprietary elements that extend the basic Merton
approach. Crosbie (1999) describes the KMV
framework, in which the EDF ™ credit measure
serves as a summary measure of default risk.
(EDF is an acronym for expected default frequency.) In essence, the EDF measure for a firm
represents an estimate of the percentage of firms
in the same financial condition that historically
defaulted on an obligation within the next
twelve months. We use KMV’s EDF credit measure to investigate market information’s capacity
to supplement supervisory assessments.
A number of studies address the issue of
whether market data can usefully supplement
supervisory monitoring efforts. Flannery (1998)
provides an overview of these and related studies, many of which focus on subordinated debt.
Relatively recent studies examining equitymarket data include that of Berger, Davies, and
Flannery (2000), who find that supervisory assessments are generally less accurate than equitymarket indicators in anticipating changes in
financial performance, such as earnings, except
when the supervisory assessments are based on
a very recent inspection. Elmer and Fissel (2001)
offer evidence that stock returns can help forecast bank failures. And finally, in the study most
similar to our own, Krainer and Lopez (2001)
find that equity-market information can help
forecast downgrades in the supervisory ratings
assigned to bank holding companies. Our
analysis is distinguished by the estimation of
statistical models based directly on downgrades
from various rating categories, whereas Krainer
and Lopez infer downgrade forecasts from a
single statistical model based on the level of
ratings. The estimation of downgrade models
may allow a sharper focus on the contribution
of equity-market data to the identification of
adverse financial changes and at a minimum

EQUITY-BASED MARKET SIGNALS
The consensus of investors regarding individual organizations is reflected in market prices
and price movements. The prices depend on
future payoffs to investors and so are inherently
forward-looking. With money at stake, investors
have a strong incentive to collect valid information, evaluate it, and accurately assess the potential risks and rewards. At least in principle, a
sense of what that assessment is can be extracted from the pricing of any risky claim on a
bank or bank holding company.
In practice, the equity claims of an organization’s owners have a number of advantages as
a source of this type of information. Compared
with other types of bank-related claims, markets
for common shares are fairly liquid, so the
quality of the price signals is reasonably high.
Moreover, equity values are sensitive to changes
in the condition of the issuing firm, making
those changes easier to observe in share prices.
Equity claims present some complications
because shareholders benefit if the issuer does
well but have limited downside risk should
losses occur, given the legal limits on their liability. Shareholders’ limited liability is a particularly prominent issue in banking, since deposit

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

3

Table 1

Definitions of Explanatory Variables
Equity-market data
EDF
Past supervisory assessments
BOPEC-1
CAMELS
Financial accounting data
SIZE
CAPITAL
RESERVES
PAST-DUE 30
PAST-DUE 90
NONACCRUAL
SECURITIES
LARGE CDs
PROVISIONS
ROA

mation about current conditions than can be
captured and conveyed by a single composite
measure, such as the BOPEC rating. Nevertheless, for banking organizations in general, the
BOPEC rating is a good summary indicator of
condition, and for the purposes of our analysis
we assume this rating accurately reflects the
financial condition of individual organizations.
With BOPEC ratings as our primary benchmark, we examine how these ratings are related
to other types of information. Specifically, we
assess the extent to which various types of information can help explain, or predict, the BOPEC
rating an organization will receive once it is
inspected. The variables we use to explain the
level of and changes in BOPEC ratings fall into
three main categories: equity-market data in the
form of EDFs, past supervisory assessments, and
financial accounting data. In combining the various data, an effort is made to reflect the flow
of information as it occurs in real time, after
accounting for reporting lags and other factors.
That is, to predict a BOPEC at any point, we
only use information that realistically would
have been available to bank supervisors.3 This
allows us to focus on whether market data can
provide incremental information to bank supervisors between inspections, beyond past inspection information and regularly reported accounting data. The explanatory variables themselves
are summarized in Table 1 and discussed below.

EDF ™ credit measure (EDF is an acronym for
expected default frequency.)
Composite BOPEC rating from the immediately prior
inspection of the holding company
Asset-weighted average composite CAMELS rating
from the most recent exams of individual banks
Log of total assets
Total equity capital
Loan-loss reserves
Loans past due 30 – 89 days
Loans past due 90 or more days
Nonaccrual loans
Investment securities
CDs of $100,000 or more
Loan-loss provisions in quarter
Net income for quarter

NOTES: EDF is proprietary and from KMV LLC. BOPEC-1 and CAMELS are confidential and from
the Federal Reserve Board. SIZE and the financial ratios are based on data from a
regulatory report, Consolidated Financial Statements for Bank Holding Companies (FR Y-9C),
issued by the Federal Reserve Board. EDF is as of the end of the month falling two
months prior to the month in which the corresponding holding company inspection was
opened. CAMELS is based on the most recent bank exam (one-bank holding company)
or exams (multibank holding company) closed prior to the month in which the corresponding holding company inspection was opened. Financial ratios are scaled by assets,
except for PROVISIONS and ROA, which are relative to average assets. These variables,
along with SIZE, are from the quarter-end two months prior to the three-month period in
which the corresponding holding company inspection was opened.

provides a separate vehicle for confirming the
positive results Krainer and Lopez document.
The following section details the types of data
our analysis uses.

Equity-Market Data
To incorporate information from the equity
market, the analysis includes the EDF credit
measure for individual banking organizations,
as constructed by KMV (EDF ). As described
above, EDF is an estimate of the probability a
firm will default within the next year. As a measure of credit risk, EDF should be positively
associated with problem BOPEC ratings; while
BOPEC ratings are not explicit estimates of the
probability of default or failure, we would
expect institutions in relatively weak financial
condition to have higher EDF values and higher
(worse) BOPEC ratings. KMV generally releases
data about two weeks after each month’s end,
so EDF is as of the end of the month falling two
months prior to the month in which the corresponding inspection was opened.

DATA
At the holding company level, the primary
supervisory indicator is the BOPEC rating, derived from financial performance along five
dimensions: bank subsidiaries (B ), other (nonbank) subsidiaries (O ), the parent company (P ),
consolidated earnings (E ), and consolidated capital (C ). The composite rating forms the basis of
the dependent variables in the regressions
reported below. This rating is defined as follows: 1—basically sound in every respect; 2—
fundamentally sound but with modest weaknesses; 3 —financial, operational, or compliance
weaknesses that cause supervisory concern; 4—
serious financial weaknesses that could impair
future viability; and 5 —critical financial weaknesses that render the probability of near-term
failure extremely high.
Current BOPEC ratings serve as our benchmark for banking organizations’ financial safety
and soundness. For many of the largest banking
organizations, this clearly is an oversimplification. For these firms, continuous on-site supervision provides supervisors with far more infor-

Past Supervisory Assessments
To help predict BOPEC ratings for individual organizations, the analysis includes two variables reflecting supervisory assessments made
prior to the opening of the current inspection.
The first variable is the rating an organization

4

FEDERAL RESERVE BANK OF DALLAS

latter type of funding is often associated with
aggressive banking strategies and frequently
subjects an organization to added expenses.
Finally, two income-statement variables are
included to capture the effect of asset quality
problems and other factors on profitability: loanloss provisions (PROVISIONS ), which should
hurt BOPEC ratings, and net income (ROA ),
which should help the ratings. The financial
ratios and SIZE are as of the quarter-end two
months prior to the three-month period in
which the corresponding holding company inspection was opened. The two-month lag used in
the regressions compensates for lags in the submission and processing of financial statements.

received on its most recent prior inspection
(BOPEC-1 ), which may be positively related to
the organization’s current rating. In addition,
information is included from a separate bank
exam process, which complements supervision
at the organization level. Bank-level exam results can trigger changes in an organization’s
BOPEC rating. Ratings at the bank level range
from 1 (best) to 5 (worst), similar to composite
BOPEC ratings, and are referred to as CAMELS
ratings. Composite CAMELS ratings are derived
from the evaluation of six bank-level factors:
capital adequacy (C ), asset quality (A), management (M ), earnings (E ), liquidity (L ), and sensitivity to market risk (S ). The asset-weighted
average of the composite ratings for an organization’s bank subsidiary or subsidiaries (CAMELS )
is included to capture supervisory information
at the bank level. The variable CAMELS is based
on the most recent bank exam (one-bank holding company) or exams (multibank holding
company) closed prior to the month in which
the corresponding holding company inspection
was opened.

Sample
The sample is based on bank holding
company inspections opened in the period from
June 1996 through March 2000. Of the 11,450
inspections and corresponding BOPEC ratings
for this period, prior BOPEC ratings are available
for 10,315. While many banking organizations
have publicly traded stock, many more do not.
Largely because of this, equity-market data can
be obtained for only 948 of these 10,315 observations. Of the 948, supervisory financial reports
are available for 914. Given the lagged structure
of the regressions, the financial reports used are
for the period from first quarter 1996 through
fourth quarter 1999. CAMELS ratings are available for all of these remaining 914 observations.

Financial Accounting Data
The analysis also controls for the potential
predictive capacity of a number of indicators
based on the quarterly reports banking organizations file with the Federal Reserve. One basic
indicator is an organization’s size. The log of total
assets (SIZE ) may reduce the chances of a substandard BOPEC rating if largeness provides
financial strength, through either a greater ability
to diversify risk or a closer relationship with the
broader financial market.
The remaining nine variables are financial
ratios that reflect various aspects of financial
strategy and performance. The balance-sheet variables are scaled using total assets, and the income statement variables are expressed relative
to average assets. Total equity capital (CAPITAL)
and loan-loss reserves (RESERVES ) serve as measures of capital adequacy. Each of these variables
is expected to reduce the chances of a substandard BOPEC rating. Asset quality is measured
using loans past due thirty to eighty-nine days
(PAST-DUE 30 ), loans past due ninety or more
days (PAST-DUE 90 ), and nonaccrual loans
(NONACCRUAL ). These variables are expected
to raise the chances of a substandard BOPEC
rating. Liquidity is measured using two variables: investment securities (SECURITIES ) and
certificates of deposit of $100,000 or more
(LARGE CDs). SECURITIES should reduce the
chances of a substandard rating, while the
reverse is true for LARGE CDs. A reliance on this

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

RESULTS
Sample Means
Table 2 shows the sample means of the
explanatory variables for the different BOPEC
ratings.4 Reading across the columns in the first
row of the table, worse supervisory ratings are
associated with a higher EDF. Organizations
that are assigned worse BOPEC ratings tend to
have had worse ratings at the prior inspection
(BOPEC-1 ). Weak BOPEC ratings also tend to
be associated with previous supervisory problems at the bank level (CAMELS ), asset quality
problems (PAST-DUE 30, PAST-DUE 90, and
NONACCRUAL), a reliance on large CDs (LARGE
CDs), and high loan-loss provisions (PROVISIONS ). In addition, substandard supervisory
ratings are negatively related to capital adequacy (CAPITAL), profitability (ROA), and organization size (SIZE).
Two minor surprises are that the relationship between investment securities (SECURITIES )
and supervisory ratings is not statistically significant (p value ≤ .05) and that loan-loss reserves

5

Table 2

Sample Means
(Quarterly Data for 1996 – 99)

nificant in this regression, the variable conveys
useful predictive information beyond whatever
information may be contained in the other variables. As a first step, all the variables listed in
Table 1 are included in the regression. The least
significant variable is then dropped from consideration, and the regression is estimated again.
This process is repeated until all the included
variables are statistically significant.
The first column of Table 3 shows the
results. The key finding is that the equity-market
variable, EDF, is highly significant in explaining
BOPEC ratings, indicating stock prices provide
useful predictive information, even after taking
into account past rating information and information from the quarterly financial statements
banking organizations file between inspections.
The positive sign of the estimated coefficient on
EDF indicates supervisory problems are associated with higher values of EDF, as would be
expected. The other significant variables are
BOPEC-1, CAMELS, CAPITAL, RESERVES, PASTDUE 90, NONACCRUAL, PROVISIONS, and ROA.
The sign of the coefficient on each of these variables corresponds to expectations. For example,
lower values of ROA are associated with more
severe supervisory problems. SIZE, PAST-DUE 30,
SECURITIES, and LARGE CDs are not significant
in explaining BOPEC ratings.
To assess the extent of EDF ’s contribution
to the ability to predict BOPEC ratings, the second column of Table 3 displays the results of
estimating the regression in column 1 with EDF
excluded. The predictive capacity of the two
regressions can be compared based on the
measures of association shown in the last three
rows of the table. For the purposes of these
measures, pairs of observations are categorized
as concordant (loosely speaking, the model gets
it right), discordant (the model gets it wrong),
or tied. A high incidence of concordant pairs,
together with a low incidence of discordant
pairs, indicates superior predictive capacity in
the form of a close association between predicted and actual outcomes. The statistic gamma
is a summary measure based on the number of
concordant and discordant pairs; a high gamma
value reflects superior predictive performance.
The maximum value for gamma is 1. As the first
and second columns of Table 3 show, the measures of association register only slightly better
values for the regression including EDF than for
the regression excluding it.
However, the regressions shown in the first
and second columns apply to BOPEC ratings in
general, most of which are unchanged from the
prior inspection. Given the significant inertia in

BOPEC rating
Equity-market data
EDF
Past supervisory assessments
BOPEC-1
CAMELS
Financial accounting data
SIZE
CAPITAL
RESERVES
PAST-DUE 30
PAST-DUE 90
NONACCRUAL
SECURITIES
LARGE CDs
PROVISIONS
ROA
Observations

1

2

3

4

p value

.25

.39

1.33

2.89

.001

1.16
1.15

1.91
1.86

2.66
2.57

2.71
2.43

.001
.001

15.22
9.16
.97
.69
.12
.30
24.71
8.68
.06
.36

14.90
8.28
.99
.78
.17
.46
23.53
9.93
.07
.30

13.89
6.86
1.12
.86
.24
.90
21.89
10.61
.16
.14

12.81
6.15
1.62
1.34
.54
1.30
24.75
13.71
.77
–.21

.001
.001
.001
.001
.001
.001
.115
.007
.001
.001

449

428

29

7

—

NOTES: Dates are for the quarterly Y-9C data used to construct the financial ratios. Corresponding dates for EDF, CAMELS, and the current BOPEC rating are in the Table 1 notes.
Composite BOPEC ratings are defined as follows: 1— basically sound in every respect;
2 — fundamentally sound but with modest weaknesses; 3 — financial, operational, or
compliance weaknesses that cause supervisory concern; 4 — serious financial weaknesses that could impair future viability; and 5 — critical financial weaknesses that render
the probability of near-term failure extremely high. To preserve the data’s confidentiality,
the sample’s single 5-rated observation is not shown. P values less than or equal to
.05 are associated with statistical significance. For BOPEC-1, the p value is determined
by a likelihood ratio chi-square test, based on the ratios of observed and expected
frequencies, for the null hypothesis of no association with current BOPEC ratings. The
p values for the remaining variables are determined by the k-sample Van der Waerden
test (chi-square approximation) for the null hypothesis of the same location parameter
across BOPEC ratings. Financial ratios are multiplied by 100.

(RESERVES ) do not appear to ameliorate supervisory problems. However, the positive association between RESERVES and supervisory problems can be explained, as asset quality problems
are not held constant when comparing the average level of RESERVES across ratings. Problem institutions are likely to be so in part due to poor
asset quality, which is commonly addressed
through higher loan-loss provisions, leading to
higher levels of reserves. The multivariate statistical techniques used in the next section facilitate an assessment of the relationship between
RESERVES and BOPEC ratings that holds asset
quality constant.
Predicting BOPEC Ratings
While Table 2 reveals interesting patterns
involving potential relationships between an
organization’s current BOPEC rating and the explanatory variables, each variable’s importance
in explaining BOPEC ratings cannot be determined based on the differences in means alone.
To identify each variable’s incremental information content in predicting BOPEC ratings, we
estimate a statistical model, or regression, of
BOPEC ratings. If a variable is statistically sig-

6

FEDERAL RESERVE BANK OF DALLAS

Table 3

Probit Regressions of BOPEC Ratings
Ordered regressions,
all rating levels
α1

3.02**
(.37)

2.88**
(.36)

α2

6.89**
(.44)

6.47**
(.41)

α3

10.24**
(.71)

8.74**
(.61)

α4

18.02**
(1.85)

14.65**
(1.63)

Equity-market data
EDF
Past supervisory assessments
BOPEC-1

.55**
(.10)
1.21**
(.12)

1.65**
(.15)

1.60**
(.14)

–10.18**
(3.37)

–10.36**
(3.29)

RESERVES

– 52.71**
(17.70)

– 46.52**
(17.03)

PAST-DUE 90

117.7**
(31.25)

98.76**
(30.64)

NONACCRUAL

43.02**
(16.19)

35.99*
(15.98)

PROVISIONS

235.9**
(59.22)

190.6**
(56.08)

–141.8**
(39.65)

–126.1**
(35.08)

Financial accounting data
CAPITAL

ROA
Measures of association
Concordant
Discordant
Gamma

95.2
4.4
.91

2.95**
(.47)

2.52**
(.45)

.72**
(.19)

1.14**
(.13)

CAMELS

Binary regressions
Downgrades from 1
Downgrades from 2

1.97**
(.27)

.11
(.83)

.97**
(.23)

1.94**
(.26)
– 27.29*
(11.02)

148.7*
(66.74)

94.8
4.9
.90

85.2
14.4
.71

– 25.67*
(10.09)

129.6*
(64.57)
117.8**
(31.92)

– 338.9**
(100.2)

–.87
(.74)

– 375.1**
(99.64)
82.6
16.8
.66

91.95**
(28.60)

– 447.9**
(151.7)

– 556.8**
(130.6)

96.3
3.6
.93

91.4
7.8
.84

** Significant at the 1 percent level.
* Significant at the 5 percent level.
NOTES: Each type of regression is first estimated using all the variables in Table 1, except that BOPEC-1 is excluded from
the two types of downgrade regressions. The least significant variable in each type of regression is then dropped
from consideration, and the regressions are estimated again. This process is repeated until all the included
variables are significant at the 5 percent level. EDF is then dropped from the resulting regressions for comparison.
Standard errors are in parentheses. The current BOPEC rating is the dependent variable in the ordered
regressions. The likelihood contribution of an observation with a BOPEC rating of i is N(αi – β′X ) – N(αi –1 – β′X ),
where N (•) is the standard normal cumulative distribution function, α0 = – ∞, and α5 = ∞. The dependent variable is
1 for downgrades and 0 otherwise in the binary regressions. The likelihood contribution of an observation that is
not downgraded is N(α1 – β′X ), and the contribution of a downgraded observation is 1 – N(α1 – β′X ), where N(•)
is the standard normal cumulative distribution function. The regressions for downgrades from a rating of 1 are
based on 429 observations (BOPEC-1 = 1), including fifty-two downgrades. The regressions for downgrades from
a 2 rating are based on 453 observations (BOPEC-1 = 2), including seventeen downgrades. The predicted
probability of the dependent variable falling into the best category — a rating of 1 for the ordered regressions and
0 for the downgrade regressions — is grouped into intervals of length .002 and defined as the “event probability.”
A pair of observations with different values of the dependent variable is defined as concordant if the observation
with the best (lowest) value also has a higher event probability. The opposite case is defined as discordant. If a
pair of observations with different values of the dependent variable is neither concordant not discordant, it is
defined as a tie. Let n represent the number of pairs with different values of the dependent variable, c the number
of concordant pairs, and d the number of discordant pairs. Concordant observations are reported as 100 • c /n and
discordant as 100 • d /n. The summary measure of association is the Goodman – Kruskal gamma, an indicator of
rank correlation between the observed ratings and predicted probabilities, defined as (c – d )/(c + d ).

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

7

finding as evidence that investors’ views regarding the financial condition of individual banking organizations, as distilled from equity prices,
provide a useful supplement to supervisory
assessments. In essence, the markets give the
right signals, and the information they provide
is not redundant.
This interpretation of our results hinges on
whether past rating information and quarterly
accounting data together form a reasonable
proxy for the extent of supervisory information
between inspections. It is important to note that
the supervisory information produced between
inspections in some—and perhaps many—cases
almost surely extends beyond the types of information included in our statistical models. This
is especially true for the largest organizations,
where a continuous on-site presence provides
supervisors with more information about current
conditions than is reflected in the past rating
information and quarterly financial data we use
as standard supervisory indicators. Based on
these considerations, further work is needed to
incorporate additional supervisory information
into the analysis. Related to this work is the
important issue of the extent to which supervisors systematically quantify any assessments
formed between inspections.
An additional question is whether market
data can provide incremental information, even
when inspections are current. We have shown
market information is useful in tracking financial
conditions, as reflected in BOPEC ratings. But
are BOPEC ratings a comprehensive indicator of
organizations’ financial condition? BOPEC ratings
themselves may be only imperfect indicators of
risk levels. One possible avenue for exploring
this question would be further work along the
lines of Berger, Davies, and Flannery (2000) that
compares the ability of equity-market information and BOPEC ratings to predict additional
indicators of financial condition, such as default.

these ratings, the measures of association shown
in the first two columns may not reflect the
extent of EDF ’s help in predicting whether an
organization receives a different rating than it
received during its last inspection. A change
in rating is an event of considerable supervisory interest, particularly if the newly assigned
rating is worse than the previous one. Reflecting
these considerations, the next section focuses
on EDF ’s contribution to the ability to predict
supervisory downgrades.
Predicting BOPEC Downgrades
The third and fourth columns of Table 3
show the results of predicting which 1-rated
organizations are downgraded to a rating of 2
or worse; the fifth and sixth columns apply to
the downgrade of 2-rated organizations to a 3 or
worse. The number of observations involving
downgrades from a rating of 3 or 4 is too small
to support statistical estimation. In the downgrade regressions we estimate for 1- and 2-rated
organizations, we follow the same procedure
used in estimating the regressions for BOPEC
ratings in general, through which statistically
insignificant variables are sequentially eliminated from the list of variables in Table 1.
As Table 3 shows, a smaller number of
variables help predict BOPEC downgrades than
BOPEC ratings in general. Nevertheless, EDF is
identified as a statistically significant variable in
predicting BOPEC downgrades for both 1- and
2-rated organizations. In addition, the measures of
association indicate considerably better performance for the regressions including EDF than for
the regressions excluding EDF, suggesting EDF ’s
incremental contribution to the ability to predict
BOPEC downgrades is notable.5 The summary
gamma statistic shows a 7 percent reduction in
the association between predicted and observed
outcomes when EDF is excluded from the
regression predicting downgrades of 1-rated
organizations. And the exclusion of EDF results
in a 10 percent reduction in predictive association for downgrades of 2-rated organizations.6
These results indicate equity-market information
is valuable in identifying potential downgrades.

NOTES

CONCLUSION
Our results show an indicator of financial
viability based on stock prices provides incremental information to bank supervisors during
the periods between inspections, beyond the
information contained in past supervisory ratings
and the quarterly financial statements routinely
used in the supervisory process. We see this

1

2
3

8

The authors would like to thank KMV LLC for providing
the proprietary data this study uses. They also would
like to acknowledge the helpful comments and suggestions of participants at the Conference on Using
Market Data in Banking Supervision, May 2 and 3,
2001, sponsored by the Federal Reserve Bank of
Minneapolis and the Federal Reserve Bank of San
Francisco.
See Board of Governors of the Federal Reserve System
and U.S. Department of the Treasury (2000).
For example, see Ronn and Verma (1986).
The data we use to construct financial ratios are from

FEDERAL RESERVE BANK OF DALLAS

4

5

6

quarterly financial reports, Consolidated Financial
Statements for Bank Holding Companies (FR Y-9C),
issued by the Federal Reserve Board. Insofar as these
data are subject to revision, the values of our financial
ratios may not reflect their assigned values when the
data were first reported, since we do not have access
to the original data.
The sole 5-rated observation is not shown to preserve
the data’s confidentiality.
These results are based on the degree of association
within the estimation sample. We have insufficient data
to assess how well the various models forecast events
outside the estimation sample.
When all the variables shown in Table 3 are included
in the downgrade regressions, whether they are significant or not, the reductions in association resulting from
the exclusion of EDF are 8 percent and 4 percent,
respectively.

Elmer, Peter J., and Gary Fissel (2001), “Forecasting Bank
Failure from Momentum Patterns in Stock Returns”
(unpublished manuscript, Federal Deposit Insurance
Corp., February).
Flannery, Mark J. (1998), “Using Market Information in
Prudential Bank Supervision: A Review of the U.S.
Empirical Evidence,” Journal of Money, Credit, and
Banking 30 (August, pt. 1): 273 – 305.
——— (forthcoming), “The Faces of ‘Market Discipline’,”
Journal of Financial Services Research.
Krainer, John, and Jose A. Lopez (2001), “Incorporating
Equity Market Information into Supervisory Monitoring
Models” (unpublished manuscript, Federal Reserve Bank
of San Francisco, April).
Merton, Robert C. (1974), “On the Pricing of Corporate
Debt: The Risk Structure of Interest Rates,” Journal of
Finance 29 (May): 449 –70.

REFERENCES
Berger, Allen N., Sally M. Davies, and Mark J. Flannery
(2000), “Comparing Market and Supervisory Assessments of Bank Performance: Who Knows What When?”
Journal of Money, Credit, and Banking 32 (August, pt. 2):
641– 67.

——— (1977), “An Analytic Derivation of the Cost of
Deposit Insurance and Loan Guarantees: An Application
of Modern Option Pricing Theory,” Journal of Banking
and Finance 1 (June): 3 –11.

Board of Governors of the Federal Reserve System and
U.S. Department of the Treasury (2000), The Feasibility
and Desirability of Mandatory Subordinated Debt,
December.

Ronn, Ehud I., and Avinash K. Verma (1986), “Pricing
Risk-Adjusted Deposit Insurance: An Option-Based
Model,” Journal of Finance 41 (September): 871– 95.

Crosbie, Peter (1999), “Modeling Default Risk,” KMV LLC,
January.

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

9

Since the early 1980s, financial innovations
have benefited the United States by increasing the
availability of financing for new firms and improving Americans’ access to financial investments.
Some innovations, such as the development of the
venture capital market, arose from deregulation
and efforts to find new ways of financing small
firms, whose risk profiles made traditional financial sources less appropriate. By making capital
more available, these financial-side innovations
have enabled such firms to take advantage of
new technologies by developing new products.
Other changes have stemmed from better
information technology that has cut the costs of
investing. In particular, lower mutual fund loads
have made it feasible for many households to
own diversified stock and bond portfolios. Partly
as a result, the share of U.S. households owning
stock has risen from less than 25 percent in the
early 1960s to about 50 percent by the late
1990s. In these ways, borrowers’ and investors’
access to capital markets has increased, and this
democratization of America’s capital markets
helped fuel the economic boom of the 1990s,
the longest economic expansion on record.

The Democratization of
America’s Capital Markets
John V. Duca

T

his article focuses on two

implications of the democratization
of U.S. capital markets: the possible
impact of more households being
exposed to stock market swings and
the possible increased sensitivity of
small and expanding businesses to

THE DEMOCRATIZATION OF SMALL FIRMS’
ACCESS TO CAPITAL

fluctuations in securities markets.

The ability of firms to finance start-ups and
expansions has greatly improved. Traditionally,
entrepreneurs financed a start-up from their savings or from capital provided by a few wealthy investors. After establishing a track record, a successful start-up could borrow from banks and, with
further success, issue stocks and bonds to fund
investments. The very best could even issue commercial paper to fund working capital (Prowse
1996). These traditional patterns have changed
since the late 1970s owing to the rise of the highyield, or “junk,” bond market and of venture
capital.
The High-Yield Bond Market
Before the 1980s, only large, well-established corporations issued bonds, typically bought
by large institutional investors (for example,
pension funds and life insurance companies)
that were primarily interested in bonds from the
most reputable and solid companies. Indeed,
many institutional investors face legal or fiduciary
constraints on whether and how much they can
invest in below-investment-grade bonds. Midsized firms generally were seen as having belowinvestment-grade credentials. Many bond investors viewed them as lacking creditworthiness
because information on such firms was limited

John V. Duca is a senior economist and
vice president in the Research Department
of the Federal Reserve Bank of Dallas.

10

FEDERAL RESERVE BANK OF DALLAS

Figure 1

and difficult to collect and analyze. Also, belowinvestment-grade, or junk, bonds lacked a track
record with which to assess their risk. Many
low-grade bonds in that era were issued by
firms whose credit ratings subsequently fell because of unexpected bad outcomes. As a result,
to fund investments, a mid-sized firm usually
borrowed from one bank, which, as the only
longer-term creditor, could spread out the fixed
costs of monitoring all the debt issued to the
firm, thus keeping the financing costs down.1
In the 1980s, three factors enabled midsized firms to issue bonds and shift away from
bank loans. One was the development of the
high-yield bond market fostered by a few pioneering investment banks that invested heavily
in new junk issues from a select group of midsized firms (Loeys 1986). Prior to this, most junk
bonds reflected the downgrading of bonds that
had been investment-grade when issued. As the
market gained experience with new junk issues,
their risks became better known and the bonds
became more acceptable to investors.
A second factor was a surge in mergers
financed with bonds (leveraged buyouts or LBOs).
This rise stemmed from fewer regulatory barriers to mergers, a greater need for consolidation among domestic firms because of more
globalized markets, and improved economies of
scale in back-office operations from lower computing costs. Corporate governance also shifted
as investors increasingly demanded that firms
cut costs and unlock value by divesting noncore
lines of business. The growing demand for merger
financing spilled over into the junk bond market because most merger-related issues had large
debt payments relative to the acquiring firm’s
cash flow. In turn, the increase in the junk market’s potential size spurred financial firms to
expand their capacity to issue and market junk
debt. As a result, the merger wave of the 1980s
created a deepening of this market (Loeys 1986
and Becketti 1990, p. 49).
A third boost to junk bonds came from
improved technology and analytical tools that
help investors obtain information about midsized firms and buy bonds at a lower cost. As a
result, information and transaction costs were
lowered, fostering junk bond issuance during
the economic expansion of the 1980s (Loeys
1986, p.11).
In the 1990s expansion, junk bond financing of mergers fell out of favor for two reasons.
First, many LBO bonds ran into trouble during
the 1990–91 recession. Second, the 1990s stock
market boom enabled firms to use stock instead
of debt to finance mergers. Nevertheless, the

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

Global Issuance of Below-Investment-Grade
Bonds, 1977–99
Billions of dollars
160
140

Total

120
100
Rule
144a

80
60
40
20
0
’77

’79

’81

’83

’85

’87

’89

’91

’93

’95

’97

’99

SOURCE: Merrill Lynch.

other factors behind the earlier growth of the junk
bond market remained. In addition, starting in
the early 1990s, Securities and Exchange
Commission Rule 144a allowed firms to file a preregistration form that enabled them to quickly
issue privately placed bonds (typically bought
by large investors) without the usual registration
and disclosure delays of traditional bond registration. As noted by Fenn (2000), although
many firms subsequently register 144a bonds,
they avoid the cost and uncertainty of delaying
bond issuance. He also points out that many
higher risk firms find Rule 144a issuance attractive not only to avoid delays but also because
many of them could not meet the registration
requirements of traditionally issued bonds.
Issuance of 144a bonds surged in the late 1990s
(Figure 1), especially for below-investment-grade
firms for whom issuance delays can be costly.
The Rise of Venture Capital
Firms’ ability to raise start-up funding has
improved greatly since the late 1970s (Figure 2 ).
One reason is the impact of regulatory changes
allowing pension funds and other institutions to
make venture capital investments through limited
liability partnerships, which overcome important hurdles to providing capital to new firms.
The development of the Nasdaq stock market
also aided venture capital by making it easier
for firms to issue stock. In addition, new information technologies have opened up many
opportunities for creating new products.
Limited Liability Partnerships, IPOs, and the
Nasdaq. Before the 1980s it was extremely diffi-

cult for institutional investors to fund start-ups.
Pension funds were limited by legal and fidu-

11

Figure 2

ness formation. This spurred a series of deregulatory moves, including one allowing institutional
investors to form partnerships that could invest
in several start-up firms and receive the benefits
of being delegated monitors.
These limited liability partnerships (LLPs)
are joint ventures of several investors who pick
a management team to select which start-ups to
finance and monitor. The LLP reaps the cost
advantage of a delegated monitor and the diversification gains to investing in a pool of startups. To encourage good performance, LLPs give
managers a share in the profits. LLPs also discourage managers from taking excessive risk
with others’ capital by restricting their investment choices and by overseeing them through a
board of directors. Other incentives for management include a limited lifetime of the LLP
and the potential to manage future LLPs once a
good reputation is established.2
When an LLP ends, it redeems the equity
in its venture investments. This equity value is
enhanced if the firms successfully conduct initial
public offerings (IPOs) of stock. The reason is
that traded equity can be held by more investors
and has more liquidity than private equity. The
development of the Nasdaq in the early 1970s
improved the prospects for IPOs. The requirements for a firm’s shares to be traded are generally easier on the Nasdaq than on the New
York (NYSE) and American (AMEX) stock exchanges. In addition, unlike these exchanges,
the Nasdaq is all-electronic, which lowers the
transaction costs of buying and selling smaller
batches of shares in newer or smaller companies.
As the Nasdaq made it easier and less costly for
firms to arrange IPOs through investment
banks, venture capital investing saw boosted
returns and volumes in the 1980s and 1990s.
The Role of New Nonfinancial Products. In
addition to the above innovations, which have
primarily expanded the supply of venture capital, nonfinancial innovations have increased the
demand for venture capital and, indirectly, its
supply. Advances in computer technology have
generated an array of information-related products. Indeed, most of the venture capital surge
in the late 1990s was concentrated in sectors
that used information innovations to spawn new
firms making new types of products (Figure 3 ).
Communications is an example of an
industry in which technological progress and
deregulation launched new firms. Likewise, the
service sector is more open to new entrants,
because the Internet facilitated the creation of
new business service, retailing and consumerrelated service firms (Figure 3). Computer advances

Venture Capital Disbursements, 1978 – 97
Billions of 1997 dollars
12

10

8

6

4

2

0
’78 ’79 ’80 ’81 ’82 ’83 ’84 ’85 ’86 ’87 ’88 ’89 ’90 ’91 ’92 ’93 ’94 ’95 ’96 ’97

SOURCE: Statistical Abstract of the United States.

ciary constraints to investing in only investmentgrade bonds and in stocks of well-established
corporations. Other investors faced informationrelated hurdles such as imperfect information
about how a new firm will fair. Will the firm use
start-up funding poorly, or take desperate gambles if it encounters trouble, or even defraud
investors? Gathering information on such questions is expensive. To recoup these fixed costs,
an investor must either demand a high return
that the firm may be unable to pay or take a big
stake in that firm that limits the ability to diversify against firm-specific risk. For these reasons,
outside seed capital came mainly from small
groups of wealthy individuals who often shared
information with one another.
Even if entrepreneurs surmounted this
limited base for start-up capital, their options for
financing working capital or new investments
were largely limited to internal cash flow, trade
credit, and bank loans. A bank, in contrast to
many securities investors, can spread the fixed
costs of monitoring and evaluating a firm over a
larger amount of debt. This enables the bank to
charge an affordable loan rate and monitor and
limit the firm’s risk-taking.
The ability of banks to perform this delegated monitor role was impaired during the
1970s, when high inflation pushed market interest rates above deposit rate ceilings. In response, households withdrew bank deposits to
invest in market securities, and banks had to
curtail lending. Particularly hard hit were smaller
firms, which, unlike large companies, lacked the
reputation needed to issue bonds or commercial
paper. Between this rationing of bank credit and
the limited base for seed capital, a lack of financing became a major impediment to small busi-

12

FEDERAL RESERVE BANK OF DALLAS

and the increased availability of venture capital
have also spawned new biotechnology firms. In
other industries, information advances created
fewer opportunities for new firms. For instance,
there is little venture capital financing of Old
Economy product firms, as shown in Figure 3.
Within the high-tech sector, the volume of
venture capital deals for software and networking firms has surged, while that for other types
of computer firms has been relatively flat. This
dichotomy fits with the general pattern of strong
venture capital growth in new industries and
less growth in industries dominated by established firms.

The Rise of Money Funds and
Money Market Deposit Accounts
In the early post-World War II era, there
were regulatory ceilings on bank deposit rates.
In periods of high inflation and high market
interest rates, these ceilings were binding and
many households earned below market interest
rates on deposits. As high market interest rates
became more persistent in the 1970s, some
households withdrew their funds from banks and
bought Treasury securities. With loanable funds
shrinking, banks restricted lending and encouraged their larger, more established commercial
borrowers to issue commercial paper backed by
bank lines of credit.3
In response, some mutual funds invested
in short-term securities (mainly Treasury bills
and commercial paper) and offered households
mutual fund shares with constant prices but
yielding market interest rates and featuring limited checking. In the high-interest period of the
late 1970s and early 1980s, these money market
mutual funds grew rapidly, while banks and
thrifts saw large deposit outflows. This led regulators to allow depositories to offer a new
instrument —money market deposit accounts—
that, like money funds, offered market interest
rates and limited checking. Also, deposit rate
ceilings ended in the early 1980s (Mahoney et
al. 1987). These changes made it easier for
households to invest in short-term money market instruments.

THE DEMOCRATIZATION OF HOUSEHOLD
INVESTORS’ ACCESS TO CAPITAL MARKETS
Between the mid-1970s and late 1990s,
household portfolios changed greatly as the
share of household financial assets in bank
deposits fell, while that in mutual funds and
securities jumped from 22 percent in 1975 to 42
percent in 1999. To a large extent, this shift
stemmed from several innovations that lowered
the cost of investing and broadened the menu
of investments. These include the rise of money
market mutual funds, the advent of Individual
Retirement Accounts (IRAs), and declines in
transaction costs.
Figure 3

Venture Capital Surges in High-Tech and
Nonhealth Services, 1995 –2001

The Advent of Individual Retirement Accounts
Starting in the early 1980s, Americans were
able to deposit up to $2,000 annually in IRAs.
The annual investments were tax deductible
and the principal and earnings not taxed until
withdrawn, presumably during retirement,
when taxpayers would likely be in lower
income-tax brackets. Since the mid-1980s,
Congress has altered the eligibility and annual
contribution provisions of traditional IRAs and
has created new types of IRAs, such as the Roth
IRA. The ability to compound investment
returns tax-free until withdrawal for all types of
IRAs and the deductibility of initial investments
in traditional IRAs encouraged many wealthy
Americans to shift existing assets into tax-preferred IRA accounts.
The advent of IRAs had four important
effects. First, the eligibility requirements encouraged people to use third parties, such as mutual
funds, to manage IRA assets and induced many
families to shift from directly held stocks and
bonds to indirect holdings through mutual funds
(Figure 4 ). Second, the rise of defined-contri-

Billions of dollars
16
High-tech
Nonhealth services
Health
Old Economy products

14
12
10
8
6
4
2
0
’95

’96

’97

’98

’99

’00

’01

NOTES: High-tech includes communications, other information
technology, semiconductors, and software. Nonhealth
services include consumer/business services and retail.
Health includes biopharmaceuticals, healthcare services,
medical devices, and medical information systems. Old
Economy products include consumer/business and
industrial products.
SOURCES: PricewaterhouseCooper’s MoneyTree Survey
(in conjunction with VentureOne) and author’s
groupings of the survey’s categories.

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

13

Figure 4

that they get used to a certain standard of living
from which they do not wish to deviate. Effectively, this assumption of “habit formation”
makes near-term consumption more important
relative to future consumption. As a result, transaction costs affect portfolio decisions. In calibration exercises, a decline in transaction costs
can induce a large rise in equity participation.
Theoretically, transaction fees have also been
shown in other models to be barriers to entry,
especially under uncertainty, as described by
Dixit (1989).
Because of their limited wealth, many
families are more apt to acquire a diversified
stock portfolio by buying mutual fund shares
rather than by directly buying stocks. For these
families, the relevant transaction costs for
investing in stocks are mutual fund fees, and if
these fees fall, stock ownership rates should
rise. This is consistent with Figure 5, which shows
large increases in overall stock ownership rates
accompanying large declines in equity mutual
fund costs (see Duca 2000 and 2001a).4 In addition, detailed data reveal that most of the rise in
overall equity ownership occurred in indirect
forms and that indirect ownership is also negatively correlated with equity fund loads. Furthermore, other data imply that the rise of indirect
ownership primarily occurred through increased
mutual fund ownership. The much higher fees
of the 1970s and early 1980s may thus account
for many households’ relatively greater resistance to owning stocks before the late 1990s
(Aiyagari and Gertler 1991, and Haliassos and
Bertaut 1995).5

Households’ Rising Reliance on Mutual
Funds to Own Equity, 1970 – 98
(Equity fund assets as a share of household equity assets)
Percent
30

25

With IRAs and 401(k)s
With IRAs only
Excluding IRAs and 401(k)s

20

15

IRA share of stocks given
by vertical distance

10

5

0
’70 ’72 ’74 ’76 ’78 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98

SOURCES: Flow of Funds, Federal Reserve Board; Investment
Company Institute; Duca (2001b).

bution pension plans encouraged many people
to incur the one-time costs of learning about
investing, which prompted many to shift their
nonretirement assets into mutual funds as well.
A third stimulus to non-IRA mutual fund assets
arose from the minimum balance requirements
of mutual funds, toward which both IRA and
non-IRA assets often count. Fourth, IRAs have
enabled many who switch jobs to accumulate
assets, whereas previously, job switchers lost
pension assets held in defined-benefit plans that
favored long-time employees.
Declines in Asset Transaction Costs
Declines in transaction costs have taken
three forms: falling mutual fund loads, declining
brokerage fees, and lower cost exchange-traded
funds.

Figure 5

Equity Fund Loads Fall and
Stock Ownership Rates Rise

The Falling Costs of Investing in Mutual Funds.

Percent of
households

One of the more dramatic changes affecting
household portfolios is the large decline in
mutual fund costs, which may have spurred
many middle-income families to begin investing in stocks. This possibility accords with the
impact of lower transaction costs in portfolio
choice models of Heaton and Lucas (2000) and
Saito (1995). In these dynamic optimization models, utility functions characterized by habit formation imply that transaction costs can deter
many families from investing in stocks. These
papers tweak the conventional intertemporal
framework by assuming that people’s utility
reflects not the sum of how they value consumption in separate periods but rather reflects

Average equity fund load
(percent)

70

9
Average load on equity funds

Only indirectly owning
Directly owning

60

8
7

Percent of households
owning equity

50
40
30

6

Total
(compositional
details
unavailable)

5
4
3

20
2
10

1

0

0
’64

’67

’69

’70

’77

’83

’89

’92

’95

’98

SOURCES: Survey of Consumer Finances; Duca (2001a, 2001b).

14

FEDERAL RESERVE BANK OF DALLAS

name Standard & Poor’s Depository Receipts, or
SPDRs. Mirroring the name abbreviation, shares
in this ETF are called spiders. Nine other S&Pbased ETFs (Select Sector SPDRs) have been
created that replicate the subcomponents of the
S&P 500.6 Other ETFs now include World Equity
Benchmark Series (WEBS), which duplicate
indexes of foreign stocks, and Diamonds, which
are based on the Dow Jones industrial average.
How do ETFs compare with index mutual
funds? ETFs are continuously traded, unlike
mutual fund shares, which can be bought or
sold once a day. Like index mutual funds, ETFs
buy and sell securities to match changes in the
composition of the stock index they mirror. As
a result, they have low costs like index mutual
funds and are arguably a close substitute (for
further discussion, see Malkiel 2000).
While ETFs compete with index mutual
funds, a new investing service offers a substitute
for actively managed mutual funds. In particular, some web-based companies offer investors
customized stock portfolios at costs that, for
investments of at least $30,000, are purportedly
below the expenses of purchasing actively managed mutual funds (McGeehan 2000). These
kinds of services, along with ETFs, broker innovations, and the potential for further declines in
mutual fund costs, will likely continue to reduce
household investors’ transaction and asset management costs.

In a related study using nearly three
decades of time series data, Duca (2001b) finds
that lower mutual fund loads and greater confidence in the future have boosted the relative
use of mutual funds as a way of owning equity.
He measures the relative reliance on mutual
funds with equity fund assets as a share of all
stocks owned by households. He argues that
lower fund fees spur some shareholders to shift
some assets from directly owned shares to
mutual funds. Also, the lower fees induce initial
stock ownership by households that are more
apt to hold shares in mutual funds for reasons
related to limited wealth and portfolio diversification. Recent calibration models of and empirical evidence on household portfolio behavior
together suggest that falling mutual fund costs
have boosted equity ownership rates in the
United States.
The Decreasing Costs of Trading Stocks.

Before 1974, the costs of trading stocks on the
NYSE were fixed to prevent price competition
among brokers. This placed small investors at a
disadvantage because there were discounts for
trades of large blocks of stock. Price competition was allowed for small trades on the NYSE
beginning in April 1974 and on all trades after
April 1975. These steps helped drive down
brokerage costs. Although continuous data are
unavailable, partial data from Schaefer (1997,
p. 13) show that broker fees fell just after deregulation. Since the early 1980s, the rise of discount brokers has likely pushed costs down further, giving investors the option of buying or
selling securities without professional advice.
The Internet has aided such low-cost investing
by partially substituting for broker services. Indeed, some old-line brokerage firms now offer
the option of buying or selling stocks without a
broker. While traditional broker services are still
important, direct investors in stocks now can
choose from a menu of services.
One interesting nondevelopment is that the
proportion of American households that directly
own stock has not increased despite declines in
the fees of buying and selling stocks. One reason
is that the limited wealth of many Americans
does not permit them to easily buy and maintain
a diversified portfolio of directly held stocks. As
a result, declining fees have likely benefited
wealthy households.
The Advent of Exchange-Traded Funds. Since
December 1998, a new type of stock has traded
on the American Exchange. Exchange-traded
funds (ETFs) offer the diversification of index
mutual funds at a lower cost. The first ETF
duplicated the stocks in the S&P 500, thus the

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

IMPLICATIONS
The increased openness, or democratization, of U.S. capital markets means that households have a wider array of investment choices
and small businesses have more sources of capital. For example, more households can feasibly
invest in stocks, and more small firms have
access to venture capital. In addition, better
diversification in their investment menus offers
protection from disruptions in particular markets. This article focuses on two implications of
the democratization of U.S. capital markets: the
possible impact of more households being
exposed to stock market swings and the possible increased sensitivity of small and expanding
businesses to fluctuations in securities markets.
A Possible Rise in the Sensitivity of
Consumption to Stock Wealth
According to many theoretical models, rising stock wealth boosts consumption by raising
the permanent or life-cycle income of households (see Ando and Modigliani 1963, and Friedman 1957). Stock market wealth has a role in

15

Figure 6

loads appear to be a good instrument for equity
participation. (This is suggested by Figure 5.)
Taking this tack, Duca (2001a) adds the product
of loads and the log of stock wealth as a separate variable to consumption regressions containing standard wealth variables to test whether
loads affect the stock wealth elasticity of consumption.7 If such interactive terms have negative signs, this is evidence that falling loads boost
the impact of stock wealth on consumption by
inducing more of the population to own equities. Using this approach, Duca (2001a) finds
that the overall stock market wealth effect has
grown in magnitude as equity mutual fund costs
have fallen. His estimates indicate that a 100
percent rise in stock wealth is associated with a
3 percent increase in annual consumption in
the late 1990s, up from about 1.5 percent in the
mid-1960s. He also finds that estimates of stock
market wealth coefficients vary less in rolling
regressions that account for the time-varying
effect of equity loads on wealth effects. As with
all relatively new research, readers should view
these results as providing some support for a
hypothesis rather than conclusive proof.

Wealth Gains Associated with a
Lower Personal Savings Rate
Ratio

Percent

7

14

Wealth/income

12

6
10
8

5

6
4

4
2

3

Savings rate
0
–2

2
’70 ’72 ’74 ’76 ’78 ’80 ’82 ’84 ’86 ’88 ’90 ’92 ’94 ’96 ’98 ’00

SOURCES: National Income and Product Accounts; Flow of
Funds, Federal Reserve Board.

many econometric models of consumption based
on the permanent income and life-cycle hypotheses (Board of Governors of the Federal Reserve
System 1999, Bosworth 1975, Brayton and
Tinsley 1996, and Mishkin 1977) and in models
deviating from the life-cycle hypothesis. One
example of the latter is Carroll’s (1992) buffer
stock model, in which utility-maximizing households alter their savings to hit a target wealth-toincome ratio. This implication is consistent with
the recent fall in the personal savings rate and
jump in the wealth-to-income ratio (see Board
of Governors of the Federal Reserve System
2000 and Figure 6).
One concern about the importance of
stock market wealth is that stock ownership is
concentrated among the very rich, whose consumption is probably not affected much by
swings in stock prices. Indeed, some studies
find that the savings behavior of the rich differs
greatly from that of the general populace
(Carroll 2000a, 2000b and Dynan, Skinner, and
Zeldes 2000). In particular, evidence reveals that
the rich save partly to acquire and preserve (primarily through bequests) power and status
(Carroll 2000b). These concerns imply that the
stock market wealth effect will be very limited.
Nevertheless, rising stock ownership rates
suggest that an increasing share of households,
whose consumption is affected by wealth, is exposed to the stock market. Unfortunately, equity
participation rate data are unavailable to directly
test whether stock market wealth effects have
become more important as stock ownership has
become more widespread. However, equity fund

Is Small Business Finance More Sensitive to
Securities Market Fluctuations?
Evidence suggests that innovations have
increased the availability of financing for small
and expanding firms. This, however, makes
financing for such firms more subject to swings
in financial market conditions. For example, if
the venture capital market should dry up, small
or new business financing would contract. Even
in such a case, small firms might still be able to
borrow from banks, an option they had before
the advent of the venture capital market. For
this reason, while innovations could make the
volume of financing more sensitive to financial
market swings, they likely have boosted the
absolute levels of such financing.
How sensitive to financial markets is the
availability of financing for small or expanding
firms? From the short history of venture capital,
Gompers and Lerner (1999) find that when
stock market prices fall substantially, the IPO
market tends to shrink for a while, as it did in
the late 1980s and early 1990s. In turn, the
decline in the near-term prospects of making a
successful IPO will likely reduce the expected
returns to investing in LLPs, which generally
have five- to ten-year lives. Consistent with this
hypothesis, Gompers and Lerner find that a
downturn in the IPO market is associated with
a decline in the volume of new venture capital
investments.

16

FEDERAL RESERVE BANK OF DALLAS

As with the IPO and venture capital markets, there is evidence that riskier firms’ ability
to issue high-yield bonds is more sensitive to
financial market conditions than is that of better
established firms. Indeed, the spreads of junk
bond yields over Aaa-rated (the highest grade of
corporate bond) yields jumped much more than
did the spread between Baa- and Aaa-rated
bonds during the 1990-91 recession.8 (Baa-rated
bonds are the lowest grade of investment-grade
bonds in which banks and most institutional
investors are able or willing to invest.) Mirroring
this jump in junk bond default spreads was a
relative rise in the default rates on junk versus
investment-grade bonds.
A decline in securities market conditions
can curtail the availability of high-yield bond or
venture capital financing for small and expanding firms in two ways. Reduced expectations
about the future economy will push stock prices
and investment down even without any feedback between them. These effects are typically
more pronounced for less established firms
whose investment prospects are more sensitive
to risk. Also, a drop in stock prices may increase
investors’ risk aversion, which may especially
boost the premiums on riskier investments, such
as high-yield bonds, IPOs, and venture capital.
This feedback effect from a declining economic
outlook further reduces the availability of small
and start-up business credit. Banks, however,
may partly offset such effects by lending more
to the affected firms. In late 1998 and early
1999, financial market disruptions drove many
large and mid-sized firms away from the bond
and commercial paper markets to banks, where
they had lines of credit. While smaller firms
tend to have less access to such lines, some
would be able to borrow from banks, which,
during periods of high risk, can more closely
monitor new credit extensions than can open
markets.
However, banks would likely be less willing to compensate for a dearth of financing for
start-up firms that lack credit and commercial
histories. In addition, the end of Regulation Q
has stabilized the availability of bank loans,
while credit scoring and other less expensive
ways of lending have helped banks lower their
costs to make small business loans. On balance,
the availability and stability of bank financing
have likely increased, while the long-run availability of volatile short-run sources of start-up
financing, such as venture capital, has also increased. In general, the availability of financing
has risen for firms, but the nature of short-run
fluctuations in availability has changed.

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

CONCLUSION
In the last quarter of the twentieth century, deregulation and technological advances
spawned several financial changes that have
increased the access of small investors and firms
to U.S. capital markets. For households, these
innovations have widened investment choices,
particularly for stocks and nonbank interestbearing assets. For less established firms, these
changes first made junk bonds a viable source
of capital; later innovations made venture capital increasingly available. The benefits of this
democratization of America’s capital markets
have spurred other developed nations, particularly in Europe, to bolster venture capital and
increase their citizens’ access to mutual funds
and other financial products.
NOTES

1

2

3

4

17

I thank Robert D. McTeer, Jr., for suggesting the article’s
topic; Jeff Gunther, Mark Wynne, and an anonymous
referee for offering helpful suggestions; and Daniel
Wolk for providing excellent research assistance. I also
thank Stephen Prowse for extensive discussions about
the private equity market. Any errors are my own.
See Rosengren (1990) for more discussion of how the
rise of the junk bond market opened up the securities
markets as a source of finance for less established
corporations.
LLPs are one form of private, that is, nontraded, equity.
For a more complete discussion of the role of private
equity in funding new and existing firms, see Fenn,
Liang, and Prowse (1997, 1998) and Prowse (1998).
See Post (1992) for more on the evolution of the
commercial paper market.
Overall ownership includes directly owning individual
stocks and indirectly owning stocks through mutual
funds and other means. Ownership rates from the
Survey of Consumer Finances (SCF) are not fully
consistent over time for three reasons. First, pre-1989
SCFs treat all mutual fund assets as indirectly owning
stock, but data since 1989 distinguish between bond
and equity funds. Second, Federal Reserve statistics
treat stock in IRA or 401(k) plans as indirect stock
ownership since 1989. Third, some early SCFs treat
privately held equity as owning stock, whereas later
SCFs do not. 1986 data are omitted because they
likely undercount broad stock ownership. This SCF
only asked a family if it owned stock or mutual funds,
whereas others also asked if households held stock in
their employer or in investment clubs. There are also
concerns about the quality of the 1986 SCF because
it was done by phone, without edit checks of unusual
answers. In addition, because the 1986 SCF recontacted 1983 respondents, the 1986 SCF could have

5

6

7

8

been distorted by selection effects from movers.
Ownership rates are the most up-to-date data from
Kennickell, Starr-McCluer, and Surette (2000); Katona,
Lininger, and Mueller (1968); Katona et al. (1970);
Katona, Mandell, and Schmiedeskamp (1971); and
Durkin and Elliehausen (1978).
Equity fund load data are from Duca (2001b), who
constructs estimates of asset-weighted loads from a
sample of large equity mutual funds used by Duca
(2000). These estimates are based on data from the
funds, IBC/Donoghue, the Investment Company
Institute, and Morningstar. Duca’s cost estimates,
which span 1960 – 2000, move together with more
comprehensive industrywide cost estimates from Rea
and Reid (1998) that are available only since 1980.
Although Haliassos and Bertaut (1995) show that
investment minimums at mutual funds are too low to
explain why most households do not own equity, their
findings do not rule out the possibility that mutual fund
fees were an important barrier to more widespread
stock ownership. This possibility is consistent with the
findings of Heaton and Lucas (2000).
“Standard & Poor’s Depository Receipts,” “SPDRs,”
and “Select Sector SPDRs” are trademarks of The
McGraw-Hill Companies.
This interactive load variable may also track a rise in
the liquidity of stock wealth. In particular, by reducing
the cost of going in and out of stocks, lower loads
could conceivably boost the magnitude of stock
wealth on consumption in the short run.
Aside from an increase in downside macroeconomic
risk, junk bond spreads were also probably boosted
by new regulations that forced many thrift institutions
to sell their substantial holdings of junk bonds. By
causing an inward shift of the demand for holding junk
bonds, these regulations conceivably pushed down
junk bond prices and thereby put upward pressure on
junk bond yields relative to other corporate yields.

——— (2000), “Monetary Policy Report to the Congress,”
February.
Bosworth, Barry (1975), “The Stock Market and the
Economy,” Brookings Papers on Economic Activity,
no. 2: 257 – 90.
Brayton, Flint, and Peter Tinsley, eds. (1996), “A Guide
to FRB/US: A Macroeconomic Model of the United
States,” Finance and Economics Discussion Series
working paper no. 1996 – 42 (Washington, D.C.:
Board of Governors of the Federal Reserve System,
October).
Carroll, Christopher D. (1992), “The Buffer-Stock Theory
of Saving,” Brookings Papers on Economic Activity,
no. 2: 61–156.
——— (2000a), “Requiem for the Representative
Consumer? Aggregate Implications of Microeconomic
Consumption Behavior,” American Economic Review 90
(May): 110 –15.
——— (2000b), “Why Do the Rich Save So Much?” in
Does Atlas Shrug? The Economic Consequences of
Taxing the Rich, ed. Joel B. Slemrod (Cambridge, Mass.:
Harvard University Press): 465 – 84.
Dixit, Avinash K. (1989), “Entry and Exit Decisions Under
Uncertainty,” Journal of Political Economy 97 (June):
620 – 37.
Duca, John V. (2000), “Financial Technology Shocks and
the Case of the Missing M2,” Journal of Money, Credit,
and Banking 32 (November): 820 – 39.
——— (2001a), “Mutual Fund Loads and the Long-Run
Stock Wealth Elasticity of Consumption,” unpublished
manuscript, Federal Reserve Bank of Dallas (June).

REFERENCES
Aiyagari, S. Rao, and Mark Gertler (1991), “Asset Returns
with Transactions Costs and Uninsured Individual Risk,”
Journal of Monetary Economics 27 (June): 311– 31.

——— (2001b), “Why Have Households Increasingly
Relied on Mutual Fund Loads to Own Equity?” unpublished manuscript, Federal Reserve Bank of Dallas
(July).

Ando, Albert, and Franco Modigliani (1963), “The ‘Life
Cycle’ Hypothesis of Saving: Aggregate Implications and
Tests,” American Economic Review 53: 55 – 84.

Durkin, Thomas A., and Gregory E. Elliehausen (1978),
1977 Consumer Credit Survey (Washington, D.C.: Board
of Governors of the Federal Reserve System).

Becketti, Sean (1990), “The Truth About Junk Bonds,”
Federal Reserve Bank of Kansas City Economic Review,
July/August, 45 – 54.

Dynan, Karen, Jonathan Skinner, and Stephen P. Zeldes
(2000), “Do the Rich Save More?” Board of Governors
of the Federal Reserve System (unpublished manuscript,
April).

Board of Governors of the Federal Reserve System
(1999), “FRB/US Equation Documentation for the ModelConsistent Expectations Version of the Model” (unpublished manuscript, September).

Fenn, George (2000), “Speed of Issuance and the
Adequacy of Disclosure in the 144A High-Yield Debt
Market,” Journal of Financial Economics 56 (3): 383 – 405.

18

FEDERAL RESERVE BANK OF DALLAS

Fenn, George, Nellie Liang, and Stephen Prowse (1997),
“The Private Equity Market: An Overview,” Financial
Markets, Institutions, and Instruments 6 (4): 1–105.

Mahoney, P. I., A. P. White, P. F. O’Brien, and M. M.
McLaughlin (1987), “Responses to Deregulation: Retail
Deposit Pricing from 1983 through 1985,” Staff Study
151, Board of Governors of the Federal Reserve System.

——— (1998), “The Role of Angel Investors in Financing
High-Tech Start-Ups” (unpublished manuscript).

Malkiel, Burton G. (2000), “Investors Shouldn’t Fear
‘Spiders’,” Wall Street Journal, May 30, Eastern edition,
A26.

Friedman, Milton (1957), A Theory of the Consumption
Function (Princeton, N.J.: Princeton University Press).

McGeehan, Patrick (2000), “The Unmutual Fund: An
Iconoclast Says He Has a Better Idea for Individuals,”
New York Times, May 18, C1, C13.

Gompers, Paul, and Josh Lerner (1999), The Venture
Capital Cycle (Cambridge, Mass.: The MIT Press).
Haliassos, Michael, and Carol C. Bertaut (1995), “Why
Do So Few Hold Stocks?” The Economic Journal 105
(September): 1110 – 29.

Mishkin, Frederic S. (1977), “What Depressed the
Consumer? The Household Balance Sheet and the
1973 –75 Recession,” Brookings Papers on Economic
Activity, no. 1: 123 – 64.

Heaton, John, and Deborah Lucas (2000), “Portfolio
Choice in the Presence of Background Risk,” The Economic Journal 110 (January): 1– 26.

Morningstar, Morningstar Mutual Funds (Chicago:
Morningstar Inc., various issues).

IBC/Donoghue, Mutual Funds Almanac (Ashland, Mass.:
IBC/Donoghue, various annual issues).

Post, Mitchell A. (1992), “The Evolution of the U.S.
Commercial Paper Market Since 1980,” Federal Reserve
Bulletin 78 (December): 879 – 91.

Investment Company Institute, Mutual Fund Fact Book
(Washington, D.C.: Investment Company Institute,
various annual issues).

Prowse, Stephen D. (1996), “A Look at America’s
Corporate Finance Markets,” Federal Reserve Bank of
Dallas Southwest Economy, Issue 2, March/April, 5 – 9.

Katona, George, William Dunkleberg, Gary Hendricks,
and Jay Schmiedeskamp (1970), 1969 Survey of
Consumer Finances (Ann Arbor: University of Michigan).

——— (1998), “The Economics of the Private Equity
Market,” Federal Reserve Bank of Dallas Economic
Review, third quarter, 21– 34.

Katona, George, Charles A. Lininger, and Eva Mueller
(1968), 1967 Survey of Consumer Finances (Ann Arbor:
University of Michigan).

Rea, John D., and Brian K. Reid (1998), “Trends in the
Ownership Cost of Equity Mutual Funds,” Investment
Company Institute Perspective 4 (November): 1– 15.

Katona, George, Lewis Mandell, and Jap Schmiedeskamp (1971), 1970 Survey of Consumer Finances (Ann
Arbor: University of Michigan).

Rosengren, Eric (1990), “The Case for Junk Bonds,”
Federal Reserve Bank of Boston New England Economic
Review, May/June, 40– 49.

Kennickell, Arthur B., Martha Starr-McCluer, and Brian J.
Surette (2000), “Recent Changes in U.S. Family
Finances: Results from the 1998 Survey of Consumer
Finances,” Federal Reserve Bulletin 86 (January): 1– 29.

Saito, Makoto (1995), “Limited Market Participation and
Asset Pricing” (manuscript, Department of Economics,
University of British Columbia).

Loeys, Jan (1986), “Low-Grade Bonds: A Growing Source
of Corporate Funding,” Federal Reserve Bank of Philadelphia Business Review, November/December, 3 –12.

Schaefer, Jeffrey M. (1997), “Negotiated Rates and the
U.S. Securities Industry,” Securities Industry Trends 23
(January): 1– 43.

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

19

In Part 1 of this article (Viard 2000), I discuss how replacing the income tax with a consumption tax generally reduces the value of the
existing capital stock. In Part 2, I examine the
allocation of this wealth decline between holders of different financial assets. How much of
the total decline in the value of firms’ capital is
borne by bondholders and how much by stockholders? Are there wealth transfers between
household lenders and household borrowers?
I consider three consumption tax designs
—a retail sales tax, a traditional value-added tax
(VAT), and a two-part VAT (“flat tax”)—and
show that they have similar potential effects.
However, the impact of the consumption tax
depends on how monetary policy responds to
the tax reform and what transition policies, if
any, are adopted.
If monetary policy keeps the consumer
price level unchanged, the wealth decline generally falls on stockholders rather than bondholders. Also, there is little reallocation of wealth
between household lenders and borrowers. It is
possible to alter these outcomes with a transition policy that aids stockholders and household borrowers at the expense of bondholders
and household lenders.
The effects are different if monetary policy
accommodates the consumption tax by allowing
the consumer price level to rise. Such accommodation is unlikely under a two-part VAT, but
observers disagree on whether it might be
adopted under a sales tax or traditional VAT to
ease possible labor-market rigidities. If the tax is
fully or largely accommodated, bondholders
bear a heavier burden than stockholders. Also,
wealth is transferred from household lenders to
borrowers.
I first describe the three consumption tax
designs and their effect on the total value of
capital and after-tax rates of return. I then examine the allocation of the wealth decline under
different monetary policy responses.

The Transition to Consumption
Taxation, Part 2: The Impact on
Existing Financial Assets
Alan D. Viard

R

eplacing the income tax

with a consumption tax is
likely to reduce the value of
the capital stock. The division
of this reduction between
bondholders and stockholders
and the effects on household
lenders and borrowers
depend on whether the tax is
accommodated through
higher consumer prices.

ALTERNATIVE TAX SYSTEMS
Economic Framework
I use a simple framework with no uncertainty and no international trade and investment. Firms produce consumer goods for sale
to households, as well as capital and intermediate inputs for sale to other firms or for internal
use. Capital depreciates at geometric rate δ,
while intermediate inputs are (by definition)
immediately used up in production.
Each firm issues bonds that promise fixed
future payments and stocks that are residual

Alan D. Viard is a senior economist and
policy advisor in the Research Department
of the Federal Reserve Bank of Dallas.

20

FEDERAL RESERVE BANK OF DALLAS

Figure 1

Application of Consumption Tax Designs
Upstream
Capital income: 70 + K1 – D1
Investment: K1 – D1

50
Purchase
of K

Pu

C as

300
Purchase
of I

Downstream
Capital income: 60 + K2 – D2
Investment: 50 + K2 – D2

rc h 2 8 0
ase
of L
h flo 70
wd
istrib
utio
n

Households

140
Purchase of L
10
n
distributio
Cash flow
500
of C
Purchase

NOTE: Arrows denote direction of payment.

claims on all other cash flows generated by the
firm’s capital. By construction, the combined
value of the bonds and stocks equals the value
of the capital. The aggregate value of firms’
capital equals national wealth. Households also
make loans to each other. Since each loan is an
asset of the lender and an offsetting liability of
the borrower, these loans do not add to national
wealth.
For each of the consumption tax designs
and the income tax, I assume tax rates are the
same across firms or households. To address
regressivity, households may receive refundable
exemptions. Revenues are distributed to households as transfer payments.
I compare the application of the tax systems in an example with two firms, Upstream
and Downstream. Figure 1 shows the transactions between firms and households, with
arrows denoting the direction of payment. I
assume units are chosen such that consumer
goods, capital, and intermediate inputs have the
same per-unit marginal cost (at the observed
output level).1
During the year, Upstream produces 50 +
K 1 units of capital, selling 50 to Downstream
and retaining K 1. However, D 1 units of capital
depreciate. The firm also produces 300 + I1 units
of intermediate inputs, of which it sells 300 units
to Downstream and uses I1 units internally.
Upstream purchases labor from households at a
cost of 280 units. It distributes its remaining cash
flow of 70 to its bondholders and stockholders.
Figure 1 shows the capital income Upstream generates, 70 + K 1 – D 1, which is production (minus depreciation of capital and usage
of intermediate inputs) minus wage payments.
The figure also reports investment, K 1 – D 1,

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

which is the net increase in the firm’s capital.
Note that cash flow equals capital income minus
investment. I do not assign values to internaluse production and depreciation because these
values cannot be observed from payment flows.
Downstream produces 500 units of consumer goods and sells them to households. It
also produces K 2 units of capital and purchases
another 50 units from Upstream. D 2 units of
capital are used up through depreciation. Downstream produces I 2 units of intermediate inputs
and purchases 300 units from Upstream, all of
which are used in its production. Downstream
buys labor from households at a cost of 140
units and distributes 10 units of cash flow to its
bondholders and stockholders. This cash flow
equals the firm’s capital income, 60 + K 2 – D 2,
minus investment, 50 + K 2 – D 2.
The key difference between the various
consumption tax designs is whether the tax is
imposed on firms or on households. Table 1
explains the calculation of the bases on which
the firms are taxed under each system.2
Retail Sales Tax
Under the retail sales tax, each firm is
taxed on the consumer goods it produces. The
aggregate tax base is clearly national consumption. Firms are not taxed on production of capital or intermediate inputs (whether used internally or sold to other firms), and households are
not taxed. So Downstream is taxed on 500 units
of consumer goods, with no tax on Upstream or
the households.
Sales tax rates can be expressed in either
tax-inclusive or tax-exclusive terms. For example,
consider a consumer good for which a household pays a firm $100 and the firm pays $25

21

Table 1

Computation of Base on Which Firm Is Taxed

Retail sales tax
Traditional VAT
Two-part VAT
Two-part income tax

Sales of
consumer
goods

Sales of
capital

✔
✔
✔
✔

✔
✔
✔

Minus
purchases
of capital

Internal
production
of capital

Minus
depreciation
of capital

Sales of
intermediate
inputs

Minus
purchases of
intermediate
inputs

Minus
wage
payments

✔

✔
✔
✔

✔
✔
✔

✔
✔

✔
✔
✔

VAT base, like the aggregate sales tax base, is
national consumption.

sales tax and retains the other $75. The taxexclusive rate is 33.3 percent because the $25
tax payment is 33.3 percent of the $75 after-tax
amount retained by the firm. However, the taxinclusive rate is 25 percent because the $25 tax
payment is 25 percent of the household’s $100
pretax payment.
Although sales tax rates are usually expressed in tax-exclusive form, I use the taxinclusive form for the sales tax and the other
consumption tax designs. This is consistent with
the common practice of reporting income tax
rates in tax-inclusive form; if a household receives $100 pretax income, pays $25 income
tax, and retains $75, the rate is said to be 25 percent, not 33.3 percent. In this article, I generally
assume a 25 percent tax-inclusive rate for a new
consumption tax because this rate is roughly
sufficient to replace current U.S. individual and
corporate income tax revenues.3

Two-Part VAT
Unlike the tax designs outlined above, the
two-part VAT is partly imposed on households.
As the third row of Table 1 shows, the base on
which each firm is taxed is the same as under
the traditional VAT, except that wage payments
are deductible. Each household is taxed on its
wage income. So Downstream is taxed on 70
(350 value added minus 280 wage payment),
Upstream is taxed on 10 (150 value added minus
140 wage payment), households are taxed on
420 wages, and the aggregate base is still 500.
Since the combined tax base on which each
firm and its workers are taxed under the twopart VAT is the same as the base on which the
firm is taxed under the traditional VAT, the two
taxes have the same aggregate base.
It can be seen that the base on which each
firm is taxed—value added minus wages —
equals the cash flow distributed to bondholders
and stockholders. In the aggregate, firms are
taxed on national cash flow and households are
taxed on national wages. Many economists have
noted that national consumption equals national
cash flow plus national wages.4
Following the usage of its most prominent
supporters, Hall and Rabushka (1995), the twopart VAT is usually called the “flat tax,” a misleading name that often causes it to be confused
with a flat-rate income tax.5

Traditional VAT
The traditional VAT taxes each firm on its
value added, which is its sales (of consumer
goods, capital, and intermediate inputs) minus
its purchases (of capital and intermediate
inputs), as the second row of Table 1 shows.
Upstream’s value added is 350 because it sells
50 units of capital and 300 units of intermediate
inputs to Downstream and makes no purchases.
Downstream’s value added is 150 because it
sells 500 units of consumer goods and purchases 50 units of capital and 300 units of intermediate inputs from Upstream. The combined
tax base is still 500.
The VAT differs from the sales tax only
when one firm sells capital or intermediate
inputs to another firm. Neither firm owes anything under the sales tax; the VAT imposes tax
on the seller but reduces the purchaser’s tax by
the same amount. Since their combined liability
is zero, capital and intermediate inputs effectively remain tax-exempt. (As with the sales tax,
no tax is imposed on internal-use production of
capital or intermediate inputs.) The aggregate

Income Tax
The aggregate income tax base is net
national product, defined as national consumption plus net investment (the production of new
capital minus depreciation). The fourth row of
Table 1 describes a two-part income tax system
in which firms are taxed on the net capital
income they generate and households are taxed
on wages. To bring each firm’s net investment
into the tax base, the two-part VAT is modified
in three ways. The deduction for capital pur-

22

FEDERAL RESERVE BANK OF DALLAS

households must sacrifice to obtain one unit.
I now examine the equilibrium relationship between the value of capital and its production
cost under different taxes by considering a small
change, or perturbation, to the circular flow between a firm and its bondholders and stockholders. This analysis also explains the relationship between the marginal product of capital
and the after-tax rate of return savers receive.
In the initial year, a firm produces one
additional unit of capital, reducing its output of
consumer goods as required by its production
function. The firm issues bonds and stocks that
represent claims on the cash flow from the new
capital and sells them to a household.8 Since the
firm is indifferent to small changes around the
optimum, the equilibrium price at which it sells
these securities must equal the after-tax receipts
it would have obtained by selling the foregone
consumer goods. The household purchasing the
securities reduces its consumption by the purchase price. This reduction in consumption is
the value of capital.
The capital stock remains one unit higher
in each subsequent year. The output the additional capital produces, net of the portion required to replace depreciation, is sold as consumer goods. The firm distributes its after-tax
proceeds from these sales to the household
owning the securities. After paying any applicable taxes, the household consumes these proceeds. The after-tax rate of return equals the
ratio of the increase in the household’s consumption in each of these years to its consumption loss in the initial year.
Table 2 shows the results of this perturbation under the various tax systems.

chased from other firms is eliminated (but the
tax on selling firms is retained), capital produced for internal use is taxed, and depreciation
is deducted. The treatment of intermediate inputs
does not change; inputs used internally remain
tax-exempt, and purchased inputs remain
deductible (offsetting the tax on the selling
firm).6 Upstream is taxed on 70 + K 1 – D 1 and
Downstream is taxed on 60 + K 2 – D 2 , the capital incomes shown in Figure 1. Households continue to pay tax on their wage income of 420.
The aggregate tax base is 550 + K 1 + K 2 – D 1 –
D 2 , which is net national product.
The income tax base is more complex than
the consumption tax base because it requires
the measurement of depreciation and internaluse capital. Distinguishing capital from intermediate inputs and measuring depreciation of
the former are unavoidable complications of the
income tax.7
This hypothetical two-part income tax
omits many of the complicating features of the
actual U.S. individual and corporate income tax.
I now add a few of the omitted features to the
analysis. I assume firms are taxed at rate τf on
the tax base described above. (Since corporations
can deduct interest paid to bondholders but not
payments to stockholders, this approach implicitly assumes that firms issue a mix of bonds and
stocks and reduces the value τf to reflect the tax
savings from the interest deduction.) To represent the various investment incentives in the
U.S. income tax, I assume firms receive an
investment tax credit at rate Z for purchases and
internal production of capital and pay a recapture tax of the same rate on sales of capital.
Households are taxed at rate τp on both
wages and capital income from firms (although
the capital income is also taxed at the firm
level). This same tax rate applies to household
loans; household lenders are taxed on their
interest income and household borrowers
deduct their interest expense. Capital income
is measured in nominal terms; bondholders
and household lenders are taxed on the portion
of interest that compensates for inflation, and
household borrowers deduct this interest. Stockholders similarly pay tax on the portion of capital gains that reflects inflation. Capital gains are
taxed on accrual, and there are no front-loaded
savings incentives.

Consumption Tax
The three consumption tax designs operate identically in this context. Let Q and MPK
denote the equilibrium production cost of capital and its marginal product (both in terms of
consumption). The firm receives a tax savings of
τc Q by producing tax-exempt capital rather than
taxable consumer goods. The equilibrium value
of the securities and the reduction in the household’s consumption must be (1 – τc )Q, which is
the value of capital.9
In each subsequent year, the capital yields
MPK units of output, δQ units of which are
invested to replace depreciation. The firm sells
the remainder as consumer goods, paying tax of
τc (MPK – δQ) on these sales. The household
receives (1 – τc )(MPK – δQ) on its securities and
consumes this amount. Dividing this annual
consumption by the initial consumption reduc-

TOTAL VALUE OF CAPITAL AND RATES OF RETURN
I define the value of capital as the consumption owners gain when they liquidate one
unit of capital or, conversely, the consumption

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

23

Table 2

Value of Capital and After-Tax Rates of Return
Effects in Initial Year
(A)

(B)

(C)

(D)

Price of
foregone
consumer
goods

Firm’s
tax
savings

Price of
securities
(A) – (B)

Value of
capital
(C)

Sales tax and VATs

Q

τc Q

(1–τc )Q

(1–τc )Q

Income tax

Q*

ZQ*

(1–Z )Q*

(1–Z )Q*

Effects in Each Subsequent Year
(E)

Sales tax and VATs
Income tax

(F)

(G)

(H)

(I)

(J)

Additional
consumer
output

Firm’s
tax

Return on
securities
(E) – (F)

Household’s
tax

Additional
consumption
(G) – (H)

Rate of
return
(I) / (D)

MPK–δQ

τc (MPK–δQ)

(1–τc )(MPK–δQ)

0

(1–τc )(MPK–δQ)

(MPK /Q)–δ

MPK *–δQ*

τf [MPK * –
(1–Z )δQ*] – ZδQ*

(1–τf )
[MPK * – (1–Z )δQ*]

τp (1– τf )
[MPK * – (1–Z )δQ*]

(1–τp )(1–τf )
[MPK * – (1–Z )δQ*]

(1–τp )(1–τf )
[MPK */Q*(1–Z ) – δ]

tion of (1 – τc )Q reveals that the after-tax rate of
return is (MPK/Q ) – δ, which is the same as the
pretax rate of return. The consumption tax does
not drive a wedge between the pretax and aftertax rates of return and therefore does not distort
consumption –saving decisions.

However, it is more realistic to assume the
production cost of capital increases as more is
produced. Since tax reform is likely to increase
investment for most types of capital, Q > Q *.
Also, the current income tax includes many
front-loaded investment incentives, so Z > 0.
The proportional decline in value is then less
than τc and may vary across different types of
capital. Some types of capital may even rise in
value.
In the analysis below, I consider three
hypothetical firms, each of which holds capital
with a value of 400 prior to tax reform. The first
firm holds capital that is produced at constant
cost and receives no front-loaded incentives, so
replacing the income tax with a 25 percent consumption tax reduces its value by 25 percent,
from 400 to 300. The second firm holds capital
that is produced at increasing cost and receives
some front-loaded investment incentives. Tax reform reduces its value by 10 percent, from 400
to 360. The third firm holds capital for which
increasing costs and front-loaded incentives are
even more pronounced. The value of this capital is unchanged by tax reform.

Income Tax
The effect of the income tax is described
in the last row of each panel. Let Q * and MPK *
denote the equilibrium production cost of capital and marginal product under the income tax.
When the firm produces a unit of capital instead
of Q * units of consumption, it receives tax savings of ZQ * from the investment tax credit. The
securities must have an equilibrium value of
(1 – Z )Q *, which is the value of capital. In each
subsequent year, both the firm and the household pay tax on the output, as the table shows.
Since the after-tax rate of return is lower than
the pretax rate of return, the income tax distorts
consumption –saving decisions.
Tax Reform’s Impact on Value of Capital
An immediate, unexpected replacement of
the income tax with a consumption tax changes
the value of capital from (1 – Z )Q * to Q. The
result can be simplified by imposing two restrictive assumptions. First, an unlimited quantity of capital can be produced at constant cost,
so Q = Q *. Second, the income tax system provides no investment tax credit (or other frontloaded investment incentives), so Z is zero.
Under these assumptions, tax reform reduces
the value of capital by fraction τc ; the proportional decline equals the consumption tax rate.10

Tax Reform’s Impact on After-Tax Rate of Return
Before tax reform, the after-tax rate of
return is lower than the pretax rate of return.
Since the rates are equal after reform, either the
pretax rate must decline or the after-tax rate
must rise, or both. If the real pretax rate of
return was 6 percent per year and the real aftertax rate was 4 percent per year, what is likely to
happen after reform eliminates the 2 percent
wedge the income tax imposes?

24

FEDERAL RESERVE BANK OF DALLAS

reform, denoted P *, is the present discounted
value of its after-tax payments, which is unity.

In the long run, the rate of return (pretax
and after tax) may be close to 4 percent. Even a
modest rise in the 4 percent after-tax return
would probably prompt additional saving that
eventually results in a large expansion of the
capital stock. The expansion of the capital stock
drives down its marginal product, lowering the
pretax rate of return well below 6 percent.
However, because this expansion of the capital
stock is the cumulative effect of the increased
flow of savings each year, it occurs only gradually. In the short run, therefore, the capital stock
has not expanded significantly and the rate of
return is likely to be close to 6 percent. The
after-tax rate of return rises by nearly the full 2
percentage points.11
Asset owners’ well-being depends on the
equilibrium after-tax rate of return, as well as
the value of their assets. An asset’s value merely
measures the current consumption the owner
would receive from immediately liquidating it.
However, an owner who intends to consume
over an extended period benefits from the increase in after-tax rates of return, to an extent
that depends on the length of his or her consumption horizon.12
Conversely, the well-being of a borrowing
household depends (negatively) on the after-tax
rate of return, as well as on the value of its
liabilities. A liability’s value merely measures the
current consumption the borrower would have
to sacrifice to immediately retire it. However, a
borrower who intends to repay over an extended period is harmed by the increase in the
after-tax rate of return that must be paid on the
liability until it is repaid.

(1)

∑ r * (1 + r *)
M

t =1

−t

(

)

+ 1+r*

−M

= 1.

I now examine how an immediate, unexpected tax reform changes the value of bonds
and household loans. Subtracting the change in
the value of each firm’s bonds from the change
in the total value of its capital yields the change
in the value of its stock. The change in the value
of household loans controls the allocation of
wealth between lending and borrowing households.
Since nominal debt payments are fixed, an
important issue is whether the consumer price
level changes when the reform occurs. Since the
price level depends on money supply and demand, it remains unchanged if the money supply is adjusted to offset any changes in money
demand resulting from tax reform.13 I now consider the effects of tax reform on asset values,
under the assumption this monetary policy is pursued and the consumer price level is unchanged.
ASSET VALUES WITH UNCHANGED
CONSUMER PRICE LEVEL
What happens to the value of outstanding
debt immediately after a 25 percent consumption tax unexpectedly replaces the income tax?
The reform has two conflicting effects. First, the
price rises to reflect the tax savings bondholders
receive (the elimination of income tax on interest payments). Second, the price is reduced
because future payments are discounted at a
higher equilibrium after-tax interest rate. In general, the new price of the debt instrument is

Debt Structure
I assume each of the three hypothetical
firms issues bonds with a value of 100, onequarter of its capital. The value of each firm’s
stock is 300, or three-quarters of capital. (These
ratios are close to the averages for nonfinancial
corporations.) I also consider a household that
makes a loan of 100 (Lender) and one that
receives the loan (Borrower).
Let r * denote the nominal after-tax rate of
return prior to reform. Each debt instrument
(firm bond or household loan) provides a nominal pretax interest payment of r */(1 – τp ) each
year until it matures and a nominal principal
repayment of unity at maturity. Since nominal
interest payments are taxed at rate τp under the
income tax, the holder of each instrument receives net payments of r *. Consider a debt instrument that matures M years after the reform.
Its price immediately before the unexpected

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

P* =

(2)

P =

M

t =1

(1 + r ) + (1 + r )
)
−t

r*

∑ (1 − τ

p

−M

,

where r is the nominal after-tax (and pretax)
interest rate after reform.
For simplicity, I assume the pretax interest
rate is constant during the M years after tax
reform, so the after-tax interest rate rises by the
amount of the former tax on interest income.14
In other words, r equals r */(1 – τp ), the nominal pretax interest rate before reform.
Substituting into Equation 2 reveals that under
this assumption, the debt instrument’s price
remains equal to unity because the tax savings
offset the loss from the higher discount rate.15
As Table 3 shows, the value of each firm’s
bonds remains equal to 100. The decline in the
value of each firm’s capital falls entirely on
stockholders as residual claimants. As the value

25

Table 3

Impact of Tax Reform with Unchanged Consumer Price Level
Transition Relief: Grandfather Existing Debt
Depending on their consumption horizons,
stockholders may be either better or worse off
because of tax reform. Although their initial
wealth declines, they benefit from earning the
higher equilibrium after-tax rate of return. Bondholders and household lenders, however, clearly
gain because they avoid any loss of initial wealth
—under the assumption r = r */(1 – τp )—and
benefit from the higher equilibrium after-tax
rate of return if they do not immediately liquidate their assets and consume. Household borrowers are harmed because their initial liability
is unchanged but they face higher future aftertax borrowing rates.16
Some believe these results are too generous to bondholders and household lenders and
too harsh on stockholders and household borrowers.17 These observers often advocate grandfathering debt instruments that are outstanding
on the reform date. Bondholders and household
lenders would continue to pay tax on interest
income from these instruments, although at rate
τc rather than τp , and firms and household borrowers would deduct their interest expense
from these instruments at the same rate. In principle, total revenues would be unchanged.18
Under this policy, the value of outstanding
debt would be

Allocation of Decline in Value of Capital
Before tax reform
Each firm

Value

After tax reform
First firm
Second firm
Third firm

Value (% change)
Value (% change)
Value (% change)

Capital

Bonds

Stock

400

100

300

300 (– 25%)
360 (– 10%)
400 (0)

100 (0)
100 (0)
100 (0)

200 (– 33%)
260 (– 13%)
300 (0)

Wealth Reallocation Between Household Lenders and Borrowers
Before tax reform
After tax reform

Combined

Lender

Borrower

0
0

100
100

–100
–100

Table 4

Impact of Tax Reform with Unchanged Consumer Price Level:
Existing Debt Grandfathered
Allocation of Decline in Value of Capital
Before tax reform
Each firm

Value

After tax reform
First firm
Second firm
Third firm

Value (% change)
Value (% change)
Value (% change)

Capital

Bonds

Stock

400

100

300

300 (– 25%)
360 (– 10%)
400 (0)

89 (–11%)
89 (–11%)
89 (–11%)

211 (– 30%)
271 (– 10%)
311 (+ 4%)

Wealth Reallocation Between Household Lenders and Borrowers
Before tax reform
After tax reform

Combined

Lender

Borrower

0
0

100
89

–100
– 89

(3 )

P =

M

∑
t =1

( ) + (1 + r )

r *(1 − τc )
1+r
(1 − τ p )

−t

−M

.

Substituting in r = r */(1 – τp ) yields the price
1 – τc [1 – (1 + r ) –M ]. For debt nearing maturity,
the value would still be close to unity. (The treatment of interest is unimportant because most of
the present value consists of the imminent principal repayment.) For debt far from maturity, the
proportional decline in value would approach
the tax rate. If the nominal pretax interest rate r
were 6 percent, with a 25 percent consumption
tax rate the decline in the price of debt with ten
years to maturity would be 11 percent.
The top panel of Table 4 shows the impact of debt grandfathering on bondholders and
stockholders, using this 11 percent estimate.
Bondholders suffer an 11 percent loss at all
firms. Stockholders at the first firm suffer a 30
percent rather than a 33 percent loss, and those
at the second firm suffer a 10 percent rather
than a 13 percent loss. The third firm’s stockholders now gain wealth. Although the impact
differs across firms holding different types of
capital (or with different degrees of leverage),
debt grandfathering tends to make the proportional losses of bondholders and stockholders
more similar in the aggregate.

of the first firm’s capital declines by 25 percent,
from 400 to 300, the value of its stock declines
by 33 percent, from 300 to 200. As the value of
the second firm’s capital declines by 10 percent,
from 400 to 360, the value of its stock declines
by 13 percent, from 300 to 260. Because stockholders own only three-quarters of the firm but
bear the full wealth decline, their proportional
loss is four-thirds times the proportional decline
in capital value. For the third firm, whose capital value remains unchanged, neither bondholders nor stockholders experience any wealth
decline. The impact on stockholders varies, depending on the type of capital their firms hold.
Also, although not shown in the table, the impact is more (less) severe if firms are more (less)
leveraged.
Because the value of debt is unchanged
when r = r */(1 – τp ), there is no redistribution
of wealth between lending and borrowing households. As the second panel of Table 3 shows,
both the wealth of Lender and the liability of
Borrower remain equal to 100.

26

FEDERAL RESERVE BANK OF DALLAS

Table 5

Equilibrium Wage Rates
(A)
Payment
received from
households
MPL
MPL
MPL*

Sales tax, traditional VAT
Two-part VAT
Income tax

(B)
Firm’s
tax

(C)
Wage payment
to worker
(A) – (B)

Worker’s
tax

(E)
Worker’s
net wage
(C) – (D)

τc MPL
0
0

(1– τc )MPL
MPL
MPL*

0
τc MPL
τp MPL*

(1– τc )MPL
(1– τc )MPL
(1– τp)MPL*

As the bottom panel of Table 4 shows, the
household lender now suffers an 11 percent
wealth decline while the household borrower
enjoys an 11 percent decline in its liability.
Wealth is reallocated from lenders to borrowers.
If this policy is desired, it can be administered most easily under a two-part VAT because
working households can report interest income
and expense on the same tax returns they use
to report wages. The policy is less convenient
under a sales tax or traditional VAT because
households must file new tax returns solely to
report interest income and expense.19

be indifferent to the perturbation, the equilibrium value of its wage payment to the household must be (1 – τc )MPL. Since the household
pays no tax, its consumption increases by this
amount.
Under the two-part VAT (which taxes the
firm on value added minus wage payments), the
perturbation does not change the firm’s tax liability because its value added and wage payments rise by the same amount. For the firm to
be indifferent to this perturbation, the equilibrium value of its wage payment must be MPL.
However, the household pays tax of τc MPL on
its wage income, so its consumption increases
by only (1 – τc )MPL.
Under the income tax, this perturbation
does not change the firm’s tax liability because its
receipts and wage expenses increase by the same
amount. For the firm to be indifferent to the
perturbation, the equilibrium value of its wage
payment must be MPL *. However, the household pays tax of τp MPL * on its wage income, so
its consumption increases by only (1 – τp )MPL *.
Each of the three consumption tax designs
distorts the labor –leisure decision by driving a
wedge at rate τc between the marginal product
of labor and the after-tax wage rate. The income
tax also distorts this decision by driving a similar wedge at rate τp.
To consider the transitional impact of tax
reform on labor markets, I make a few simplifying assumptions. Although tax reform is likely
to greatly increase the marginal product of labor
in the long run by significantly expanding the
capital stock, the short-run change in marginal
product should be small because (as discussed
above) the capital stock expands only gradually.
So I assume the marginal product of labor is
initially unchanged, MPL = MPL *.20 I consider a
worker with a marginal product of twelve consumer goods per hour, who earns $12 per hour
when these goods sell for $1 each, and I assume

COULD TAX REFORM INDUCE AN
INCREASE IN CONSUMER PRICES?
The above analysis assumes the consumer
price level is unchanged by tax reform. The
asset pricing implications are different if the
consumer price level rises in response to tax
reform. Some argue that the labor-market effects
of adopting a sales tax or traditional VAT may
prompt a monetary policy response that raises
the consumer price level.
Labor Markets under Alternative Tax Systems
To understand this argument, I consider a
simple perturbation to the circular flow between
firms and workers (Table 5 ). A firm purchases
one additional unit of labor from a household
and produces additional consumption. Let MPL
and MPL * denote the marginal product of labor
under a consumption tax and the income tax,
respectively. For the firm to be indifferent to this
perturbation, its after-tax receipts from selling
the additional consumption must equal its wage
payment. After paying any applicable taxes, the
household consumes its additional wages.
Under the sales tax and the traditional VAT,
the firm pays an additional tax of τc MPL on its
additional consumption output. For the firm to

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

(D)

27

wage reductions achieved through inflation.
Similarly, since the FLSA minimum wage is not
indexed, its real value can be reduced through
inflation. If nominal-wage rigidity otherwise
impedes the necessary adjustment, a possible
response is a monetary policy that rapidly increases the consumer price level by a factor of
1/(1 – τc ).22 I refer to this policy as “full accommodation” of the consumption tax and to a
smaller price increase as partial accommodation.
Full accommodation raises the price of
consumer goods to $1.33. Since firms retain $1
after paying sales tax or VAT, they can profitably
hire workers at a $12 wage rate if their marginal
product is twelve units, the same as before reform. The necessary real-wage adjustment occurs
without any reduction in nominal wages.
It is unclear whether nominal-wage rigidity poses a problem. Workers who resist isolated
reductions in their nominal wage rate may accept
an economywide reduction made in response to
a highly visible change in tax policy, particularly
when take-home pay is unaffected. Congress
can also amend the FLSA when it adopts tax
reform. Accommodation may be unnecessary.
I make no prediction about whether monetary policy would accommodate all, some, or
none of a sales tax or traditional VAT. Instead, I
compare the asset-price effects of tax reform
under the different possibilities. I have already
described the effects without accommodation
and now describe them with full accommodation. Intermediate effects occur under partial
accommodation.

the consumption and personal income tax rates
are both 25 percent.
The replacement of the income tax with a
two-part VAT has little impact on labor markets
because these taxes are similar in form as well
as substance. Table 5 implies that under either
tax, the firm pays the worker $12 and the worker
pays $3 tax. Under the current income tax, firms
withhold the worker’s wage tax, so the paycheck
is actually $9, with a stub noting that the worker
is being paid $12, $3 of which is withheld for
taxes. Hall and Rabushka (1995, 145) propose
similar withholding under a two-part VAT.
The treatment of wages under a sales tax
or traditional VAT differs in form, but not in
substance, from that under an income tax
because the tax is imposed on the firm rather
than the worker. In accordance with Table 5,
the firm now pays a wage rate of only $9
because it retains only 75 cents after tax for
each of the twelve goods the worker produces.
The paycheck amount is still $9, but the stub is
now different, showing $9 as the wage with no
tax withheld. The worker owes no additional tax,
and disposable income is still $9. This wage-rate
adjustment illustrates the public-finance principle that imposing a tax on the seller rather than
the buyer has no real economic effect because
the equilibrium price adjusts by the amount of
the tax.
Potential Nominal-Wage Rigidity
and Monetary Accommodation
Some argue that this adjustment may not
easily occur. They note the longstanding literature suggesting that workers resist nominalwage reductions and argue that this resistance
applies to reductions in the wage rate listed on
the paycheck stub, not reductions in take-home
pay.21 Even if such resistance does not exist,
they note that the Fair Labor Standards Act
(FLSA) prohibits the necessary adjustment for
lower-paid workers. This law prescribes a minimum value (currently $5.15 per hour) for the
wage rate shown on the paycheck stub.
If the listed wage rate remains rigid, the
adoption of a sales tax or traditional VAT has
problematic labor-market consequences. After
paying sales tax or VAT, firms retain only 75
cents for each unit produced and they can profitably hire workers at a $12 wage rate only if
their marginal product is fifteen units, rather
than the twelve units possible before reform.
The implied hiring reduction and unemployment could be substantial.
The literature on nominal-wage rigidity
generally assumes workers do not resist real-

ASSET PRICE EFFECTS WITH
FULL ACCOMMODATION
With accommodation, the nominal interest
rate (and the inflation rate) is extremely high
during the brief period when the price level is
rising. If r = r */(1 – τp ) thereafter, full accommodation reduces the real value of debt by factor τc .
As the top panel of Table 6 shows, bond
values fall from 100 to 75 at each firm. At the
first firm, as total value falls from 400 to 300,
stock values fall from 300 to 225, which is also
a 25 percent decline. At the second firm, however, as total value declines from 400 to 360,
stock values fall only from 300 to 285, a mere 5
percent. At the third firm, where total firm value
is unchanged at 400, stock values actually rise,
from 300 to 325. Overall, bondholders now bear
heavier burdens than stockholders, reversing
the pattern seen in Table 3.

28

FEDERAL RESERVE BANK OF DALLAS

Table 6

Impact of Tax Reform with Full Accommodation
The value of the household loan also falls
from 100 to 75. As the bottom panel of Table 6
shows, the household lender suffers a 25 percent wealth decline while the household borrower enjoys a 25 percent decline in the value
of its liability.23
Although this analysis generally ignores
consumer-owned capital, it should be noted
that accommodation benefits homeowners with
mortgages at the expense of mortgage lenders,
in the same way it benefits stockholders of leveraged firms at bondholders’ expense. Consider
a homeowner with a $120,000 house and an
$80,000 mortgage, and assume that tax reform
reduces the home’s real value to $110,000.24 If
the consumption tax is not accommodated, the
real value of the mortgage is still $80,000 and
the homeowner suffers a 25 percent decline in
home equity, from $40,000 to $30,000. But if the
tax is fully accommodated, the real value of the
mortgage falls to $60,000 and the homeowner’s
equity rises from $40,000 to $50,000, as the
mortgage lender bears more than the full burden of the decline in value. (Of course, accommodation is irrelevant if the homeowner does
not have a mortgage.)
Some believe an accommodated consumption tax would be too harsh on bondholders
and household lenders and too favorable to
stockholders and household borrowers, the
opposite of the concerns expressed with no
accommodation. If accommodation is considered undesirable, it can be avoided by adopting
a two-part VAT. Or if a sales tax or traditional
VAT is adopted, steps can be taken to facilitate
rapid nominal-wage adjustment. Or if nominalwage rigidity is considered inevitable, the need
for accommodation can be avoided by phasing
in the sales tax or VAT and phasing out the
income tax over an extended period. With a
smooth ten-year phase-in, for example, nominal
wages need fall only 2.5 percent per year from
their prereform path; if they had been growing
4 percent per year, they can still grow 1.5 percent per year during the phase-in, avoiding outright reductions. However, since firms have an
incentive to delay investment while a consumption tax is phased in, it may be better to immediately replace the income tax with a two-part
VAT and then phase in a sales tax or traditional
VAT and phase out the two-part VAT.

Allocation of Decline in Value of Capital

Value

After tax reform
First firm
Second firm
Third firm

Value (% change)
Value (% change)
Value (% change)

Capital

Bonds

Stock

400

100

300

300 (– 25%)
360 (– 10%)
400 (0)

75 (– 25%)
75 (– 25%)
75 (– 25%)

225 (– 25%)
285 (– 5%)
325 (+ 8%)

Wealth Reallocation Between Household Lenders and Borrowers
Before tax reform
After tax reform

Combined

Lender

Borrower

0
0

100
75

–100
– 75

tax”)—is likely to reduce the total value of the
capital stock. The division of this reduction
between bondholders and stockholders and the
effects on household lenders and borrowers
largely depend on whether the tax is accommodated in the form of a higher consumer price
level. Accommodation is unlikely under a twopart VAT, but observers disagree about its likelihood under a sales tax or traditional VAT.
If the consumption tax is not accommodated, the real value of debt changes little. The
decline in the value of capital is largely borne
by stockholders, and there is little reallocation
of wealth between household borrowers and
lenders. With full accommodation, the real value
of debt is sharply reduced. Bondholders bear
heavier burdens than stockholders, and household borrowers gain at the expense of household lenders.
These transitional effects deserve careful
attention in the evaluation and implementation
of tax reform.
NOTES

1

2

CONCLUSION
Replacing the income tax with a consumption tax —whether in the form of a sales
tax, traditional VAT, or two-part VAT (“flat

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

Before tax reform
Each firm

29

I am grateful to Mark Wynne, Mine Yücel, Greg
Huffman, and V. Brian Viard for helpful comments.
For pedagogical purposes, I compare the application
of the various tax systems to a fixed level and composition of output. This does not imply that equilibrium
output is the same under the different taxes. As discussed elsewhere in this article, the income tax equilibrium differs from the consumption tax equilibrium.
Koenig and Huffman (1998, 25 – 26), Congressional
Budget Office (1997, 7– 16), Gravelle (1996, 1422 – 28),
Auerbach (1996, 43 – 46), and Joint Committee on
Taxation (1995, 51– 52, 57– 58) also describe these tax
designs. Another tax system is a personal consumption
tax in which each household is taxed on its consumption
and firms are not taxed. This design, which has received
little attention in the past few years, is not considered here.

3

4

5

6

7

8

9

10

11

See Viard (2000, 6, 19 note 5) and the references
cited therein.
Viard (2000, 7), Bradford (2000, 68), Koenig and
Huffman (1998, 26), Congressional Budget Office
(1997, 66), Gravelle (1996, 1425), Auerbach (1996,
30 – 31), and Joint Committee on Taxation (1995, 55)
note this equality.
The name is particularly misleading because this
tax has the potential to be less flat than the sales tax
or traditional VAT. Hall and Rabushka (1995, 55)
propose that the household component include a
nonrefundable exemption, and another prominent
academic supporter, Bradford (2000, 67–70), suggests
that it have a progressive rate structure. Bradford calls
this tax the X-tax. Koenig and Huffman (1998) and
Joint Committee on Taxation (1995, 57) call it the
Hall – Rabushka tax, while Hall (1996) calls it the
Hall – Rabushka VAT. Congressional Budget Office
(1997) calls it the bifurcated VAT.
Intermediate inputs remain outside the tax base
because they do not add to net national product.
Since they are immediately used up, depreciation
always equals production.
Sections 263 and 263A of the Internal Revenue Code
prohibit firms from deducting purchases of capital or
the costs of producing internal-use capital, while
section 162 allows a deduction for intermediate inputs.
Litigation has increased in the past decade as the IRS
has become more vigilant, albeit selectively, in denying deductions for intangible capital. Section 168 sets
fixed depreciation schedules for tangible capital, and
section 197 sets a fifteen-year schedule for many
purchases of intangible capital. However, section 167
requires that internally produced intangible capital be
depreciated over its “useful life,” which is another
source of litigation.
If the “new view” of corporate financial policy is valid,
firms are at a corner solution in which they issue no
new stocks. The perturbation in the text must be
modified to have the firm reduce dividends rather than
issue securities. Since the imposition of a higher
personal tax rate on dividends than on capital gains
then has effects similar to those of front-loaded investment incentives, Z can be viewed as including this tax
rate differential. Viard (2000, 14 –16), Congressional
Budget Office (1997, 67), Gillis, Mieszkowski, and
Zodrow (1996, 748), and Auerbach (1996, 37, 69)
discuss the new view and its implications.
The perturbation reduces production of consumer
goods by one unit, but the securities-owning household reduces its purchases by only (1 – τc )Q units.
A household that receives lower transfer payments
(due to the revenue loss) reduces consumption
purchases by τc Q, closing the circle.
Viard (2000, 8 –11) presents this simplified analysis in
greater detail.
Huffman and Koenig (1998, 25), Auerbach (1996,

12

13

14

15

16

17

18

19

20

21

22

30

58–59), and Kotlikoff (1996, 174 –76) provide more
detail on how tax reform may affect rates of return.
Viard (2000, 10), Congressional Budget Office (1997,
67), Gillis, Mieszkowski, and Zodrow (1996, 748),
Auerbach (1996, 60), and Joint Committee on Taxation
(1995, 87) make this point.
Tax reform is likely to initially reduce the real quantity
of money demanded by raising after-tax interest rates
and temporarily lowering consumption. So keeping
consumer prices unchanged may require slowing
money growth.
The assumption that the pretax interest rate is
unchanged shortly after tax reform is inspired by the
previous conclusion that the marginal product of
capital is little changed at that time. However, this
reasoning is simplistic. Before reform, the pretax
interest rate may not equal the net-of-depreciation
marginal product because bonds and stocks receive
different firm-level tax treatment. Congressional
Budget Office (1997, 33 – 34) and Auerbach (1996,
48 – 49) discuss this difficult issue.
Municipal bonds, which this article does not consider,
clearly decline in value because they receive no tax
savings (they are already tax-exempt) but are subject
to the higher discount rate.
Recall the assumption that the current income tax
allows borrowers to deduct interest expense. If interest
is nondeductible (as it is for some loans), tax reform
does not change borrowing costs.
For example, Bradford (2000, 111), Koenig and
Huffman (1998, 26), and Pearlman (1996, 421)
describe this outcome as a “windfall” for bondholders
and household lenders. Bradford (2000, 101) and
Gravelle (1996, 1445) note that leverage magnifies the
impact on stockholders.
Variants of this policy are proposed or discussed by
Bradford (2000, 110 –11), Koenig and Huffman (1998,
26), Pearlman (1996, 420 – 23), and Hall and Rabushka
(1995, 79 – 80). Pearlman argues that in practice, this
policy could reduce total revenue.
Pearlman (1996, 408, 413) notes that transition relief is
less likely to be offered under a sales tax. Another form
of transition relief allows firms to deduct depreciation
on existing capital. This relief is most beneficial to shortlived types of capital, as Viard (2000, 11) and Bradford
(2000, 110) note. The tax savings flow to stockholders
as residual claimants, with no effects on bondholders
or household lenders and borrowers. The resulting
revenue loss raises the revenue-neutral tax rate.
With endogenous labor supply, tax reform may initially
reduce the marginal product of labor. See Huffman
and Koenig (1998, 25), Auerbach (1996, 57– 58), and
Kotlikoff (1996, 175).
Taylor (1999, 1013 – 21) surveys the literature on
nominal-wage rigidity.
Bradford (2000, 100 –102), Congressional Budget
Office (1997, 65 – 66), Gillis, Mieszkowski, and Zodrow

FEDERAL RESERVE BANK OF DALLAS

23

24

(1996, 752 – 53), Hall (1996, 77–78), Auerbach (1996,
44 fn. 29), Gravelle (1996, 1441– 42), and Viard (1994)
discuss these issues. Most, but not all, of these
authors suggest that nominal-wage rigidity will require
some accommodation of a sales tax or traditional VAT.
Accommodation also lowers the real value of the $3
trillion of nominal debt the U.S. Treasury owes to the
public, thereby lowering the tax rate required for longrun budget neutrality.
Tax reform is likely to reduce the value of consumerowned capital, such as owner-occupied homes, along
with the value of business capital, but for different
reasons, reflecting the distinctive tax treatment of
consumer capital. See Viard (2000, 11– 13), Bradford
(2000, 107), and Congressional Budget Office (1997,
66 – 67).

Hall, Robert E., and Alvin Rabushka (1995), The Flat Tax,
2nd ed. (Stanford: Hoover Institution Press).
Huffman, Gregory W., and Evan F. Koenig (1998), “The
Dynamic Impact of Fundamental Tax Reform, Part 2:
Extensions,” Federal Reserve Bank of Dallas Economic
Review, Second Quarter, 19 – 31.
Joint Committee on Taxation (1995), Description and
Analysis of Proposals to Replace the Federal Income Tax,
JCS-18-95 (Washington, D.C., June).
Koenig, Evan F., and Gregory W. Huffman (1998),
“The Dynamic Impact of Fundamental Tax Reform,
Part 1: The Basic Model,” Federal Reserve Bank of
Dallas Economic Review, First Quarter, 24 – 37.
Kotlikoff, Laurence J. (1996), “Saving and Consumption
Taxation: The Federal Retail Sales Tax Example,” in
Frontiers of Tax Reform, ed. Michael J. Boskin (Stanford:
Hoover Institution Press), 160 – 80.

REFERENCES
Auerbach, Alan J. (1996), “Tax Reform, Capital Allocation,
Efficiency, and Growth,” in Economic Effects of Fundamental Tax Reform, ed. Henry J. Aaron and William G. Gale
(Washington, D.C.: Brookings Institution), 29 – 81.

Pearlman, Ronald A. (1996), “Transition Issues in Moving
to a Consumption Tax: A Tax Lawyer’s Perspective,” in
Economic Effects of Fundamental Tax Reform, ed. Henry
J. Aaron and William G. Gale (Washington, D.C.:
Brookings Institution), 393 – 427.

Bradford, David F. (2000), Taxation, Wealth, and Saving
(Cambridge: MIT Press).
Congressional Budget Office (1997), The Economic
Effects of Comprehensive Tax Reform (Washington, D.C.:
Government Printing Office).

Taylor, John B. (1999), “Staggered Price and Wage Setting
in Macroeconomics,” in Handbook of Macroeconomics,
vol. 1B, ed. John B. Taylor and Michael Woodford
(Amsterdam: Elsevier Science, North-Holland), 1009 – 50.

Gillis, Malcolm, Peter Mieszkowski, and George R.
Zodrow (1996), “Indirect Consumption Taxes: Common
Issues and Differences Among the Alternative
Approaches,” Tax Law Review 51 (Summer): 725 –74.

Viard, Alan D. (2000), “The Transition to Consumption
Taxation, Part 1: The Impact on Existing Capital,” Federal
Reserve Bank of Dallas Economic and Financial Review,
Third Quarter, 2 – 22.

Gravelle, Jane G. (1996), “The Distributional Effects of
Fundamental Tax Revisions,” San Diego Law Review 33
(January): 1419 – 57.

——— (1994), “Who Would Really Get Bitten by the
BAT?” Tax Notes, November 7, 784 – 85.

Hall, Robert E. (1996), “The Effect of Tax Reform on
Prices and Asset Values,” in Tax Policy and the Economy,
vol. 10, ed. James M. Poterba (Cambridge: MIT Press for
the National Bureau of Economic Research), 71– 88.

ECONOMIC AND FINANCIAL REVIEW SECOND QUARTER 2001

31