View original document

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

Home / Publications / Research / Economic Brief / 2022

Economic Brief
August 2022, No. 22-33

Introducing the Credit Market Sentiment Index
By Danilo Leiva-León, Gabriel Pérez-Quirós, Horacio Sapriza and Egon Zakrajšek

In a forthcoming paper, we develop a new signal-extraction statistical model to
estimate a factor summarizing conditions in U.S. credit markets. The factor
provides a real-time gauge of "sentiment" in credit markets, above and beyond
that attributable to contemporaneous economic conditions. Fluctuations in the
credit market sentiment factor are associated with strong asymmetric and
nonlinear effects on economic activity, depending not only on the magnitude and
sign of a credit market sentiment shock but also on the current economic
Understanding the sources and transmission of financial distress in the economy is
essential for macroeconomic stabilization policy. For example, policymakers and academics
have both pointed to excesses in credit markets — including abnormally low risk premiums,
misaligned incentives for risk taking, lax credit standards and excessive borrowing — as the
main culprit behind the 2008-09 financial crisis.
Since then, many questions have emerged regarding the role of credit factors in business
cycle fluctuations. Our forthcoming paper "Credit Market Sentiment: Estimation and
Macroeconomic Implications" addresses two such questions that are especially important
for policymakers:
Do "frothy" conditions in credit markets create risks to future macroeconomic
If so, how can we measure such froth?
To provide answers, we develop a signal-extraction model of joint nonlinear dynamics that
uses an array of economic activity and credit market indicators, with the latter including
both price-based and quantity-based indicators. We use the model to estimate a factor that
can be updated in real time and that summarizes "sentiment" in U.S. credit markets.

We call it a credit market sentiment factor because the signal is purged of its
contemporaneous correlation with economic activity. That is, our factor summarizes credit
market conditions beyond what can be attributed to the direct effects of economic activity
on credit markets. We document that fluctuations in this factor — which we term credit
market sentiment shocks — have economically and statistically significant effects on
economic activity. Moreover, these effects are nonlinear and asymmetric, depending not
only on the magnitude and sign of the shock but also on the current economic conditions.

Research on Credit Markets
Much of the empirical research that addresses the above questions has focused on the
growth in broad credit aggregates or balance-sheet measures of leverage. It has
documented that rapid and prolonged increases in credit outstanding presage economic
downturns and sharp declines in the stock market.1 Another strand of this literature
emphasizes time variation in sentiment on the part of investors in credit markets as an
important determinant of the credit cycle and the associated business cycle dynamics.2
This latter approach — which harks back to the narrative accounts of financial crises by the
1977 paper "The Financial Instability Hypothesis: An Interpretation of Keynes and an
Alternative to 'Standard' Theory" and the 1978 book Manias, Panics and Crashes — identifies
periods of frothy credit market conditions as those in which the objective expected returns
to bearing credit risk are driven too low because credit is being priced too aggressively. The
key takeaway from these studies is that periods in which credit risk was being aggressively
priced are predictably followed by a widening of credit spreads. The timing of this widening
reliably precedes the downturn in economic activity.
An important behavioral perspective linking the pricing of credit risk to economic outcomes
is related to the way investors in credit markets update their beliefs in light of incoming
data. In particular, investors may overreact to recent news and thus overweight future
outcomes that have now become a little more likely. For example, investors may become
complacent about default risk after a few years of economic expansion, which can lead to:
Compression in credit spreads
A loosening of bank lending standards and terms
A surge in the issuance of credit to risky borrowers
In such an environment, the arrival of unfavorable economic news — at least, unfavorable
by the recent overly optimistic standards — can lead investors to revise disproportionately
their assessment of default risk, precipitating a pullback in the supply of credit that triggers
an economic downturn. This reasoning implies that investor psychology can itself be a
cause of volatility in credit and investment, even in the absence of significant changes in
economic fundamentals. Several papers formulate psychological models of investor
confidence, in which investors' overly extrapolative expectations can lead to concordant
credit and business cycles, even without changes in economic fundamentals.3

Understanding the role of credit market sentiment for economic activity also matters for
policy. The 2013 paper "Issuer Quality and Corporate Bond Returns" emphasizes the role of
investor risk appetite and beliefs in driving credit cycles. It argues that: "[A] major challenge
for any countercyclical credit policy is identifying the existence of a sentiment driven credit
boom in the first place."
Relatedly, the 2013 speech "Overheating in Credit Markets: Origins, Measurement and
Policy Responses" argues that policymakers should pay close attention to the effects of
non-fundamental factors on the economy. It notes that "[O]ne of the most difficult jobs that
central banks face is in dealing with episodes of credit market overheating."

Estimating Credit Market Sentiment
Gauging sentiment in credit markets is not easy. While brisk credit growth, loose lending
standards, rising leverage and narrow risk spreads are all indicative of accommodative or
even frothy credit market conditions, they also partly reflect strong borrower fundamentals
and robust economic growth. The fact that credit market conditions are influenced
importantly by economic fundamentals implies a need for a methodology that accounts for
the effect of current economic conditions on credit markets. Moreover, to the extent that
the relationship between credit market sentiment and economic conditions may differ
between normal and crisis periods, the methodology should be able to accommodate
possible nonlinear dynamics.
Several credit market indicators can — and indeed have been — used as proxies for credit
market sentiment. Individually, however, each indicator provides only a partial
approximation that contains information about sentiment embedded in a specific indicator
or credit market segment.
Moreover, any credit market indicator contains information about current economic
conditions. Individual indicators can also be noisy and often difficult to reconcile with the
readings of other variables, making it difficult to arrive at a comprehensive assessment of
how "frothy" or "tight" credit market conditions are.
To identify the role of sentiment-driven shocks in business cycle fluctuations, we need a
measure that captures the underlying common signal from a variety of credit market
indicators. This measure also needs to be purged of its contemporaneous correlation with
real economic activity.
With this in mind, we develop a statistical model that efficiently incorporates information
from key economic activity and credit market indicators to produce a credit market
sentiment factor with high information content about the potential fragility of U.S.
economy. The construction of this factor entails extracting information about credit market
conditions from multiple credit market indicators, both price-based and quantity-based.

We consider price-based indicators such as credit spreads and associated risk premiums
based on the 2012 paper "Credit Spreads and Business Cycle Fluctuations" and related
studies, which point to the information content of these indicators for future economic
activity. Similarly, we also use quantity-based measures such as the growth of bank balance
sheets, following the 2012 paper "Credit Booms Gone Bust: Monetary Policy, Leverage
Cycles and Financial Crises, 1870-2008" and others that show that rapid and sustained
credit growth anticipates economic downturns.4
By considering the interaction between credit quantities and prices, our framework is
consistent with the 2021 paper "Can Policy Tame the Credit Cycle?," which emphasizes that
the interplay between leverage and mispricing is central to the 1977 paper "The Financial
Instability Hypothesis: An Interpretation of Keynes and an Alternative to 'Standard' Theory"
and in understanding the role of credit sentiment in the credit cycle and economic activity.
Our model also incorporates a range of real activity indicators, and when we extract the
signal about credit market conditions, we consider the effect of real activity — as
summarized by the economic activity factor — on credit market variables. The reason is
that we want our credit signal to capture only information from credit markets that is not
attributable to current economic conditions.
Credit market variables also exhibit a markedly nonlinear relationship with economic
activity.5 Therefore, we consider a nonlinear framework that allows for the economy to
move between "normal" (expansionary) and "adverse" (contractionary) states. The two
states are characterized by the joint behavior of credit market sentiment and economic
activity factors.6
Our model delivers three key indicators:
The economic activity factor, which summarizes current economic conditions
The credit market sentiment factor, a measure of how accommodative or restrictive are
credit conditions in the economy beyond what can be explained by the economic
activity factor
The probability that the economy is currently in an adverse state, determined by the
joint dynamics of the credit market sentiment and economic activity factors
Figure 1 shows these three indicators.



As shown in Figure 1a, the economic activity factor captures well the declines in economic
activity during recessions, as well as the quick rebound from the recent COVID crisis. Figure
1b shows the credit sentiment factor, which exhibits much wider swings than the economic
activity factor. The low-frequency movements in the credit sentiment factor are consistent
with the evidence presented by the 2012 paper "Characterizing the Financial Cycle: Don't
Lose Sight of the Medium Term" and represent comovements across an array of credit
market indicators, purged of the contemporaneous comovements that these indicators
have with the economic activity factor.
Note that our credit market sentiment factor clearly identifies the extended period of credit
market overheating leading up to the 2008-09 financial crisis and associated recession. It
also picks up a marked and sudden deterioration in credit market sentiment in the summer
of 2015, when concerns about global growth prospects (centered around China) sparked a
widespread repricing of risky assets and a sharp increase in financial market volatility. In
fact, according to our estimates, the ensuing gloom that engulfed credit markets was quite
persistent and started to lift only in 2019.

Lastly, Figure 1c shows the estimated time series of probabilities that U.S. economy is in an
adverse economic state. These probabilities accord well with recessions, an indication that
our model is able to distinguish between expansionary and contractionary periods of
economic activity, which (as is well known) differ notably in their macroeconomic dynamics.
In our framework, this distinction is informed by the dynamic interaction of the economic
activity and credit sentiment factors, an idea consistent with the extensive empirical
evidence that recessions accompanied by disruptions in credit markets tend to be longer
and deeper than other recessions.7
Figure 2 shows another output of the model, which can be used to gauge the vulnerability
of U.S. economy to sudden changes in credit market sentiment and real shocks and which
also illustrates the inherent asymmetries and nonlinearities of our framework.




In particular, Figures 2a and 2b show the model-implied changes in the probability of being
in an adverse economic state in month t + 1, conditional on a negative (Figure 2a) or
positive (Figure 2b) credit market sentiment shock hitting the economy in month t. Figures
2c and 2d show the corresponding changes in the same probabilities in response to real
economic shocks.8
According to Figure 2a, negative credit market sentiment shocks in normal times
significantly increase the probability that the economy will subsequently find itself in an
adverse state. Moreover, this sensitivity varies notably over the course of a business cycle,
generally peaking in the late stages of an economic expansion. This pattern is consistent
with a gradual buildup of financial excesses during economic booms that predictably end in
a credit bust accompanied by a severe economic downturn. By contrast, the same-sized
shock during a recession has very little effect on these probabilities because the economy is
already in an adverse state. As shown in Figure 2b, the impact of positive credit market
sentiment shocks is exactly the opposite: In adverse economic states, an improvement in
credit market sentiment significantly lowers the likelihood that the economy will remain in
that state during the following month.

Figure 2c shows the model-implied changes in the probability of moving from a normal to
an adverse economic state in month t + 1 in response to a negative shock in real activity in
month t. Not surprisingly, negative real activity shocks in normal times also tend to increase
the likelihood that economy will enter an adverse state. However, compared with adverse
credit market sentiment shocks, their effects on these probabilities are much smaller on
average and do not exhibit a clear cyclical pattern. As shown in Figure 2d, positive shocks to
real economic activity during recessions, as expected, significantly reduce the probability of
economy remaining in an adverse state. In fact, their impact during recessions is
comparable to the same-sized positive shocks in credit market sentiment.

All told, our results indicate that fluctuations in credit market sentiment — a comovement
in credit market indicators above and beyond the comovement induced by the
contemporaneous economic conditions — have economically and statistically significant
effects on real economic outcomes. The effects of credit market sentiment shocks on
economic activity are asymmetric and involve nonlinear dynamics, reflecting the slow
buildup of financial imbalances during economic expansions. Once the economy is in such
a vulnerable state, a negative sentiment shock can easily tip the economy into an adverse
state, where the interaction of falling asset prices, high debt burdens and balance-sheet
repair depresses growth for a protracted period.
Danilo Leiva-León and Gabriel Pérez-Quirós are senior economists at the Banco de España.
Horacio Sapriza is a senior economist and policy advisor in the Research Department at the
Federal Reserve Bank of Richmond. Egon Zakrajšek is a senior adviser at the Bank for
International Settlements.

1 See, for example, the 2009 paper "The Aftermath of Financial Crises," the 2011 paper "Financial

Crises, Credit Booms and External Imbalances: 140 Years of Lessons," the 2012 papers "Credit
Booms and Lending Standards: Evidence From the Subprime Mortgage Market" and "Credit
Booms Gone Bust: Monetary Policy, Leverage Cycles and Financial Crises, 1870-2008," the 2013
paper "When Credit Bites Back," the 2015 paper "Leveraged Bubbles," the 2016 paper "The Great
Mortgaging: Housing Finance, Crises and Business Cycles," and the 2017 papers "Credit
Expansion and Neglected Crash Risk" and "Household Debt and Business Cycles Worldwide."
2 See, for example, the 2012 paper "Credit Spreads and Business Cycle Fluctuations," the 2013

paper "Issuer Quality and Corporate Bond Returns," the 2017 working paper "How Credit Cycles
Across a Financial Crisis," the 2017 paper "Credit-Market Sentiment and the Business Cycle," the
2018 working paper "Lending Standards and Output Growth" and the 2022 paper "Predictable
Financial Crises."
3 Examples of papers include the 1998 paper "A Model of Investor Sentiment," the 2010 paper

"The Gambler's and Hot-Hand Fallacies: Theory and Applications," the 2017 paper "Diagnostic

Expectations and Credit Cycles" and the 2019 paper "Reflexivity in Credit Markets." For studies
focused on the role of consumer sentiment in economic fluctuations, see, for instance, the 2022
paper "Sentimental Business Cycles."
4 Specifically, the quantity-based credit market indicators include the monthly growth of real

bank credit and high-yield corporate bond issuance, expressed as a share of total corporate
bond issuance. The price-based indicators include the Moody's Baa-Aaa corporate bond credit
spread, the excess bond premium and the at-issuance spread on B-rated leveraged syndicated
loans. The set of economic activity indicators is comprised of the monthly growth rates of
(nonfarm) payroll employment, industrial production, real manufacturing and trade sales, and
real total personal income (less transfer payments).
5 For example, see the 2013 paper "Intermediary Asset Pricing" and the 2017 paper "How Credit

Cycles Across a Financial Crisis."
6 For example, see the 2012 paper "How Do Business and Financial Cycles Interact?"
7 Formally, our statistical signal-extraction model is a dynamic factor model, in which the credit

market sentiment and economic activity factors follow a two-state Markov-switching vector
autoregression. To identify credit market sentiment shocks, we impose restrictions on the system
of measurement equations, whereby the credit market sentiment factor has no
contemporaneous effect on the economic activity indicators, but the economic activity factor can
have a contemporaneous effect on all the variables in the system.
8 The size of both shocks is always two standard deviations.

To cite this Economic Brief, please use the following format: Leiva-León, Danilo; Pérez-Quirós,

Gabriel; Sapriza, Horacio; and Zakrajšek, Egon. (August 2022) "Introducing the Credit Market
Sentiment Index." Federal Reserve Bank of Richmond Economic Brief, No. 22-33.
This article may be photocopied or reprinted in its entirety. Please credit the authors,
source, and the Federal Reserve Bank of Richmond and include the italicized statement
The views expressed in this paper are solely the responsibility of the authors and should not be
interpreted as reflecting the views of the Banco de España, Bank for International Settlements or
the Federal Reserve Bank of Richmond.

Subscribe to Economic Brief

Receive an email notification when Economic Brief is posted online:
Email address


Contact Us
RC Balaban
(804) 697-8144

© 1997-2022 Federal Reserve Bank of Richmond