Content uploaded by Jesús Vázquez
Author content
All content in this area was uploaded by Jesús Vázquez on Jun 08, 2023
Content may be subject to copyright.
Economic Modelling 124 (2023) 106283
Available online 31 March 2023
0264-9993/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Contents lists available at ScienceDirect
Economic Modelling
journal homepage: www.journals.elsevier.com/economic-modelling
On the significance of quality-of-capital news shocks✩
Luis Herrera a,b, Jesús Vázquez b,∗
aIRES, Université catholique de Louvain, Collège L. H. Dupriez, B- 1348 Louvain-la-Neuve, Belgium
bUniversidad del País Vasco (UPV/EHU), Av. Lehendakari Aguirre 83, Bilbao, Spain
ARTICLE INFO
JEL classification:
E30
E32
E44
Keywords:
Quality-of-capital news
TFP news
Credit channel
Aggregate fluctuations
DSGE model
Bayesian estimation
ABSTRACT
This paper assesses the significance of quality-of-capital (QoC) news shocks and their transmission through
the credit channel in explaining aggregate fluctuations. Our framework is an estimated medium-scale DSGE
model augmented with a financial sector where two alternative sources of news shocks are considered. One is
a (standard) total-factor-productivity (TFP) news shock; the other is a QoC news shock. The latter has a clear
meaning that enables a close link to be built up between financial markets and the macroeconomy through
the credit and expectation channels, which greatly improves model fit and largely displaces TFP news shocks
as a source of the business cycle. The significance of pure (rather than realized) news underscores the role of
expectations.
1. Introduction
This paper builds on the expectation-driven business cycle hypoth-
esis, which has a long-standing tradition in macroeconomics. Thus,
Pigou (1927) argued that the business cycle was driven by variations
in the profit expectations of ‘business men’.1More recently, Beaudry
and Portier (2004) suggest a modeling approach for Pigou’s theory
of the business cycle, which suggests that anticipated (news) shocks
are a major source of business cycle fluctuations. Beaudry and Portier
(2006) provide further empirical evidence supporting Pigou’s view.
They identify two shocks using VAR methods; one shock drives short-
run fluctuations in stock prices and is orthogonal to innovations in total
factor productivity (TFP). This shock is closely correlated to a second
✩The authors thank the Editor (Angus Chu), two anonymous referees, Vincent Bodart, Mikel Casares, Ferre de Graeve, Paolo Gelain, Luca Pensieroso, Frank
Portier, Matthias Rottner, Rigas Oikonomou, and seminar participants at the EEA-ESEM 2022 Congress, 2021 Computation in Economics and Finance Meeting,
2021 IAAE Annual Conference, 2020 ASSET Meeting, III SMN Workshop, and DSE Winter School 2020 for useful comments and suggestions on preliminary
versions of this paper. This research was supported by the Spanish Ministry of Science and Innovation, MCIN/AEI /10.13039/501100011033/ under grant number
MCIN/AEI/PID2020-118698GB-I00, by the Spanish Ministry of Economy and Competition under grant number ECO2016-78749P, and by the Basque Government
under grant numbers IT-1336-19 and IT-1461-22. The first author also acknowledges financial support from the Spanish Ministry of Science, Innovation, and
Universities under scholarship grant FPU17/06331.
∗Corresponding author.
E-mail addresses: luisherrerabravo@hotmail.com (L. Herrera), jesus.vazquez@ehu.es (J. Vázquez).
1Pigou (1927) also claimed that changes in those expectations were triggered by two ‘impulses’: fundamental impulses, captured by news shocks that end up
realizing, and psychological impulses captured by revised and nonrealized news. These impulses have also been considered by recent macroeconomic literature
addressing the importance of news shocks.
2Gertler and Karadi (2011) first refer to them as quality-of-capital shocks, whereas Merton (1973) and Gertler et al. (2012) also call them asset price
shocks. Gertler et al. (2012) provide a sound micro foundation for QoC shocks based on the productivity of capital already installed. This literature views
QoC shocks as purely transitory, surprise shocks (i.e., they are described as independent and identically distributed processes). Our paper retains the assumption
of QoC shocks characterized by a stationary process, but we allow for some degree of persistence and the possibility of shocks being anticipated.
shock that drives long-run movements in TFP. Moreover, Beaudry and
Portier (2006) show that these two shocks anticipate TFP growth by
several years. This empirical evidence strongly supports the idea of an
expectation-driven business cycle in which the financial sector plays an
important role.
This paper suggests a novel approach for modeling the type of news
shocks described in Beaudry and Portier (2006) by considering quality-
of-capital (QoC) news shocks in the medium-scale DSGE model of Smets
and Wouters (2007) augmented with financial frictions à la Gertler and
Karadi (2011). Surprise QoC shocks have been considered in the related
literature (Gertler and Kiyotaki,2010;Gertler and Karadi,2011;Görtz
and Tsoukalas,2017); however, the importance of (anticipated) QoC
news shocks in the business cycle have not yet been assessed.2This
https://doi.org/10.1016/j.econmod.2023.106283
Received 9 February 2022; Received in revised form 18 February 2023; Accepted 14 March 2023
Economic Modelling 124 (2023) 106283
2
L. Herrera and J. Vázquez
paper contributes to that assessment, which stresses a close link be-
tween financial markets and the macroeconomy.
QoC shocks represent qualitative appreciations (depreciations) of
physical capital, which trigger an exogenous change in the productivity
of capital and also directly affect the balance sheet of financial inter-
mediaries whose assets are collateralized by that capital. News shocks
to the quality of capital thus have (arguably) a clearer interpretation
than TFP news shocks, where the latter are often measured as news
shocks to the Solow residual—which interprets any change in output
not explained by changes in factor inputs as a change in TFP (Solow,
1957). This clear interpretation of QoC news shocks (relative to TFP
news shocks) enables them to be connected with financial markets
through the credit and expectation channels. More precisely, a QoC
news shock affects the production function similarly to a TFP news
shock but also acts as an exogenous trigger of asset price dynamics.
For example, an anticipated upgrade in physical capital improves pro-
duction expectations and may immediately impact the balance sheets
of financial intermediaries whose assets are backed up by that capital.
Similarly, when sector-specific capital is expected to become obsolete,
production is expected to fall, and agents may also anticipate the
coming drop in capital (asset) value, making the level of debt excessive
relative to the stock of capital.3In short, the fundamental difference
between QoC and TFP news shocks lies in the direct effects on financial
variables induced by the former and amplified through the expectation
and credit channels. By estimating alternative model specifications,
we assess the relative contribution of QoC and TFP news shocks in
explaining aggregate fluctuations.
Turning to estimation results, we show that when TFP news and
QoC news shocks are included in the DSGE model, the latter be-
comes the main driver of aggregate fluctuations while the former play
a relatively minor role. An improvement in model fit supports this
finding. That improvement is especially large for hours, inflation, and
the investment growth rate. Thus, the data support a news shock
specification in which news directly affects the credit channel. The
estimation results also show three main differences between these
two alternative specifications of news shocks. First, the expansionary
responses of most real variables (output, investment, and labor) at
impact are more pronounced in response to a QoC news shock than to a
TFP news shock. This finding is explained by the direct impact of QoC
news shocks on the financial side of the economy and, in particular,
by the larger fall in the credit spread in response to an expansionary
QoC news shock. Second, the short-run consumption response is much
lower for QoC news shocks, underscoring the transmission of QoC news
through the investment/credit channel. Finally, a positive QoC news
shock triggers a mild negative response on the part of the inflation,
which is in contrast to the positive response of inflation to a positive
TFP news shock. Interestingly, the deflationary response of QoC news
shocks is in line with the deflationary response of TFP news shocks
found by Görtz et al. (2022), who use a VAR approach.
We further contribute to the recent literature analyzing news shocks
in a DSGE framework by addressing an important question. Does the
inclusion of QoC news shocks help to improve the characterization of
agents’ expectations? This question is important because identifying a
news shock must improve the fit of model expectations of forward-
looking variables (i.e., the expectation channel). We show that a DSGE
model that includes QoC and TFP news shocks outperforms one that
contains only the latter for all observable variables with counter-
parts reported in the Survey of Professional Forecasters, especially for
the investment growth rate. Including QoC news seems to enhance
the importance of the credit channel, thus helping to improve the
characterization of investment expectations, among others.
3The close link between TFP and financial shocks is also investigated
in Moran and Queralto (2018) and Queralto (2020), who emphasize demand-
driven factors determining medium-term dynamics in TFP. Under their
approach, financial shocks affect business innovation activities and future TFP.
The prominent role of QoC news shocks is further enhanced by the
decomposition analysis of news shocks suggested by Sims (2016) to
distinguish between pure and realized news shocks.4We find that pure
QoC news shocks are one of the main drivers of aggregate fluctuations.
This result somewhat contrasts with Sims (2016), which finds that
the empirical significance of TFP news shocks is due to their realized
component. In line with Görtz and Tsoukalas (2017), our estimation
results highlight the importance of considering a financial sector (which
is ignored in the DSGE framework used in Sims,2016) to assess the
relative importance of alternative sources of news shocks since financial
markets provide useful information that can help in identifying news
shocks.
The rest of the paper is structured as follows. Section 2connects the
contribution of this paper to the related literature, Section 3describes
the canonical DSGE model augmented with financial frictions, and
Section 4briefly describes the data set, the prior distributions, and cali-
brated parameters. Section 5discusses the estimation results, compares
QoC and TFP news shocks, assesses the importance of and differences
in investment-specific-technology (IST) news shocks compared to QoC
and TFP news shocks, and examines the relative importance of pure
and realized components of QoC news shocks. Section 6concludes.
2. Further related literature
A large body of news literature using DSGE and VAR approaches
suggests that the anticipation of future changes in production via
financial variables found in Beaudry and Portier (2006) is due to news
on TFP. For instance, Beaudry and Lucke (2010) use short- and long-
run restrictions to identify TFP news shocks as an important driver of
the business cycle. Barsky and Sims (2011) suggest another strategy
for identifying TFP news shocks in a VAR framework, also finding
them to be a significant source of fluctuations. Fujiwara et al. (2011)
and Schmitt-Grohé and Uribe (2012) are two seminal papers incor-
porating news shocks into a DSGE model. The former identifies TFP
news shocks in the US and Japan as an important source of aggregate
fluctuations in both countries, especially in the US. Schmitt-Grohé and
Uribe (2012) do not find TFP news shocks to be important5; how-
ever, Görtz and Tsoukalas (2017) show that findings in Schmitt-Grohé
and Uribe (2012) are due to their model’s specific assumptions, such
as the lack of transmission channels linking financial markets with real
economic activity. In particular, Görtz and Tsoukalas (2017) underlines
the importance of considering an endogenous financial sector (such as
the one suggested in Gertler and Karadi,2011) since financial markets
convey useful information that can help in identifying TFP news shocks.
Thus, they show that when a financial sector is considered, TFP news
shocks recover their role as the main news source. Gunn and Johri
(2018) use a calibrated model to show that news shocks to future
financial returns can create business cycles without recourse to other
news sources. More recently, Görtz et al. (2022) used VAR methods and
found that TFP news is highly associated with credit spread indicators
and that the dynamics of financial variables are critical for amplifying
TFP news shocks in a two-sector (consumption and investment) DSGE
model.
This paper’s main contribution is to stress the direct impact of news
on financial markets through QoC news shocks in a DSGE framework,
which induce effects amplified through the credit channel and are also
a source of real and financial fluctuations. The main difference between
4More precisely, Sims (2016) proposed a method for distinguishing be-
tween the effects of pure news and realized news shocks, with the former
seen as the effects at horizons before the realization of the news and zero at
horizons after that, i.e., realized news effects are just the effects of news shocks
at horizons after the realization.
5They find that other shocks, such as news for the wage markup, are crucial
in explaining the business cycle.
Economic Modelling 124 (2023) 106283
3
L. Herrera and J. Vázquez
our approach and that in Görtz and Tsoukalas (2017) is that they
consider the endogenous financial sector as an important amplifier,
whereas we consider a type of news shocks, (namely, QoC news shocks)
which themselves are a direct source of financial fluctuations. QoC
surprise shocks were also introduced in Gertler and Karadi (2011). They
used in a calibrated model to theoretically investigate the effects of an
exogenous source of variation in asset values; however, their paper is
mainly concerned with the empirical significance of QoC news shocks
having a direct impact on the financial sector and an anticipated effect
on production, like that of a standard TFP news shock.
Our paper also assesses the relative importance of the IST news
shocks posited in Ben Zeev and Khan (2015) as a major driver of
aggregate fluctuations. Further motivation for this assessment can be
found in several papers, suggesting that IST shocks can act as a veil,
hiding the response of investment to changes in asset prices (Kam-
ber et al.,2015;Afrin,2017;Görtz and Tsoukalas,2017). Justiniano
et al. (2010) also find evidence that IST shocks strongly correlate
with financial variables such as the interest rate spread. These findings
might therefore be viewed as additional evidence reported in the recent
literature that IST news shocks may act as a veil that may capture the
risk premium fluctuations that affect the price of capital. Our estimation
results confirm this view, as discussed below.
3. The model
This paper considers a medium-scale DSGE model with several
sources of rigidity and news and surprise (unanticipated) shocks. The
model is similar to the workhorse New Keynesian DSGE model sug-
gested in Smets and Wouters (2007), augmented with the financial
frictions suggested by Gertler and Karadi (2011). This model has been
widely used in recent macro finance literature (Sanjani 2014,Villa
2016,Afrin 2017,Gelain and Ilbas 2017;Görtz and Tsoukalas,2017).
This section provides a brief overview of the model. The demand
side of the model economy is formed by households that choose con-
sumption and hours worked and hold riskless assets, such as bank
deposits and government bonds. Hours worked are homogeneously
supplied by households to an intermediate labor entity that differen-
tiates and supplies labor to labor packers, who subsequently sell labor
services to the intermediate goods sector.
Intermediate goods firms choose their production inputs (labor
services and effective capital) and sell a differentiated good to the final
sector, which sells a homogeneous good to households in a perfectly
competitive market. The intermediate goods firms and the labor entity
supply differentiated inputs (goods/labor) used in producing the final
consumption good, so they are assumed to have some market power.
This assumption also enables the inclusion of nominal rigidities à
la Calvo (1983). Capital services producers acquire physical capital
produced by capital goods producers and assemble it into effective
capital, rented to intermediate goods firms. Capital services producers
finance their capital acquisition by borrowing funds from financial
intermediaries in a perfectly competitive market. Hence, financial fric-
tions are introduced from the credit supply through bank balance sheets
as suggested in Gertler and Karadi (2011).6Clearly, news on the quality
of capital services financed by banks directly impacts their balance
sheets, which further affects the credit supply.
The DSGE model with financial frictions considers that banks lend
funds obtained from household deposits to nonfinancial firms. There-
fore, they act as intermediaries that assist firms in channeling funds
from household deposits to investors; however, banks would like to
expand their assets by borrowing additional funds from households
indefinitely since the discounted risk premium they face is always
6This approach of introducing financial frictions contrasts with the ap-
proach suggested in Bernanke et al. (1999), which builds on the financial
accelerator.
positive by construction. A moral hazard problem is introduced to
restrict their ability to do this. The banks decide whether to divert a
fraction of their assets and transfer them to the households to which
they belong. The cost for banks of diverting assets is that the depositor
can force them into bankruptcy and recover the remaining fraction of
assets; therefore, households only deposit their savings up to the point
where the gain of banks from diverting assets is equal to the gain of not
doing so. This incentive constraint introduces credit supply rigidity.
Next, we describe how two types of news shock are included in the
DSGE model and their main differences. A brief description of the whole
model can be found in Appendix A.
Production channel
As is standard in the literature, we consider that intermediate goods
firms produce goods according to a Cobb–Douglas production function,
where the endogenous inputs are capital and labor. This production
function is affected by three different shocks. Two of them are the
stationary and nonstationary shocks that compound the standard TFP
shock, and it is assumed that news arises from the latter. Furthermore,
we consider stationary QoC shocks as in Gertler and Karadi (2011),
Gertler et al. (2012). As explained above, these represent qualita-
tive appreciation (or depreciation) of physical capital, so they trigger
exogenous changes in capital productivity, affecting the production
function similarly to a TFP shock. Formally, the production function
is as follows:
𝑌𝑡=𝑇 𝐹 𝑃𝑡𝑄𝑜𝐶𝑡𝐾𝑡−1𝑈𝑡𝛼𝐿1−𝛼
𝑡−𝐴𝑡𝜙𝑝,(1)
where 𝑇 𝐹 𝑃𝑡=𝜖𝑎
𝑡𝐴𝑡,𝜖𝑎
𝑡is the aforesaid transitory TFP shock. 𝐴𝑡is the
permanent TFP shock, and its growth rate is denoted by 𝑎𝑡=𝑙𝑛 𝐴𝑡
𝐴𝑡−1 .
𝑄𝑜𝐶𝑡captures exogenous shocks in the quality of capital, 𝐾𝑡−1 is capital
stock at the beginning of period 𝑡,𝑈𝑡is the capital utilization rate, 𝛼
is the capital share in production, and 𝜙𝑝is the share of fixed costs
involved in production.
Financial channel
The main difference between a TFP news shock and a QoC news
shock is that the latter has an amplifying effect on the price of assets
(which is equivalent to the price of capital in the model); thus, that
there is a distinctive, direct impact on the balance sheets of financial
intermediaries. The rationale is that stock investors’ valuation of as-
set prices is highly influenced by incoming information on transitory
capital quality upgrades (obsolescence).
Capital services firms purchase physical capital at the end of period
𝑡at a price 𝑄𝑡and sell the undepreciated component to capital good
producers at the end of period 𝑡+ 1 at a price 𝑄𝑡+1 . They also decide
on capital utilization by considering the adjustment cost and the rate
at which they rent the installed capital to the intermediate goods firms.
Capital services firms also finance their purchases of capital at the end
of each period with funds from financial intermediaries, considering
that the funding is obtained by issuing claims that are equal to the value
of the capital purchased, the price of which is the same 𝑄𝑡𝑆𝑡=𝑄𝑡𝐾𝑡.
Thus, the profit maximizing problem of these agents is
max𝐾𝑡{𝑟𝑘
𝑡+1𝑈𝑡+1 𝐾𝑡𝑄𝑜𝐶𝑡+1−𝑎𝑈𝑡+1 𝐾𝑡𝑄𝑜𝐶𝑡+1
+(1 − 𝛿)𝑄𝑡+1𝐾𝑡𝑄𝑜𝐶𝑡+1 −𝑅𝑘
𝑡+1𝑄𝑡𝑆𝑡}
𝑠𝑡. 𝑄𝑡𝑆𝑡=𝑄𝑡𝐾𝑡,
where 𝑟𝑘
𝑡is the rental rate of capital in period 𝑡,𝑎𝑈𝑡is the capital
utilization adjustment cost function, and 𝑅𝑘
𝑡is the return of each claim.
The optimal decision obtained from the above problem means that
the price of assets (capital) depends directly on QoC shocks:
𝑄𝑡=
𝑟𝑘
𝑡+1𝑈𝑡+1 −𝑎𝑈𝑡+1 + (1 − 𝛿)𝑄𝑡+1
𝑅𝑘
𝑡+1 𝑄𝑜𝐶𝑡+1.(2)
TFP and QoC shocks affect 𝑄𝑡through general equilibrium, but QoC
shocks also have a direct effect.
Economic Modelling 124 (2023) 106283
4
L. Herrera and J. Vázquez
Table 1
Calibration of fixed parameters.
Parameters Calibrated value
Discount factor 𝛽0.99
Capital depreciation rate 𝛿𝑘0.025
Wage markup 𝜖𝑤0.2
Price markup 𝜖𝑝0.2
S.S. government spending share 𝑔∕𝑦0.20
Fraction of capital that can be diverted 𝜆0.536
Transfer to the entering bankers 𝜔0.001
Survival rate of the bankers 𝜃0.96
Shock processes
The model considers eight types of purely unanticipated (surprise)
shock and two shock processes, including unanticipated and news shock
components. The unanticipated shocks are stationary TFP, price and
wage markup, monetary policy, preference, net worth, IST, and public
spending shocks. Each shock follows an AR(1) process:
𝜖𝑥
𝑡=𝜌𝑥𝜖𝑥
𝑡−1 +𝜂𝑥
𝑡,
where 𝑥=𝑎, 𝑝, 𝑤, 𝑚, 𝑏, 𝑛𝑤, 𝐼𝑆 𝑇 , 𝑔. Nonstationary TFP and QoC shocks
have two components: An unanticipated shock and a news shock. The
formulation of news shocks follows the seminal paper by Schmitt-Grohé
and Uribe (2012):
𝜖𝑧
𝑡=𝜌𝑧𝜖𝑧
𝑡−1 +
𝑖
𝜂𝑧
𝑡,𝑡−𝑖,
where 𝑧=𝑇 𝐹 𝑃 , 𝑄𝑜𝐶 ; and 𝑖= 0,1,4,8,12. Therefore, 𝜂𝑧
𝑡,𝑡−𝑖is a 𝑧news
shock which is expected to realize at time 𝑡but is forecast 𝑖periods
before (i.e., at period 𝑡−𝑖). For instance, 𝜂𝑧
𝑡,𝑡−8 is a 𝑧-innovation real-
ized at time 𝑡but anticipated eight periods in advance. Consequently,
agents react in advance to future forecast shocks (i.e., agents react to
newly obtained information about future shocks even though nothing
fundamental has yet changed). More precisely, agents forecast future
values of 𝜖𝑧
𝑡+𝑘as follows:
𝐸𝑡𝜖𝑧
𝑡+𝑘
=(𝜌𝑧)𝑘𝜖𝑧
𝑡+
𝜂𝑧
𝑡+𝑘,𝑡 +𝜂𝑧
𝑡+𝑘,𝑡−1 +𝜂𝑧
𝑡+𝑘,𝑡−4 +𝜂𝑧
𝑡+𝑘,𝑡−8 +𝜂𝑧
𝑡+𝑘,𝑡−12,
𝜂𝑧
𝑡+𝑘,𝑡−1 +𝜂𝑧
𝑡+𝑘,𝑡−4 +𝜂𝑧
𝑡+𝑘,𝑡−8 +𝜂𝑧
𝑡+𝑘,𝑡−12,
𝜂𝑧
𝑡+𝑘,𝑡−4 +𝜂𝑧
𝑡+𝑘,𝑡−8 +𝜂𝑧
𝑡+𝑘,𝑡−12,
𝜂𝑧
𝑡+𝑘,𝑡−8 +𝜂𝑧
𝑡+𝑘,𝑡−12,
𝜂𝑧
𝑡+𝑘,𝑡−12,
0,
𝑓 𝑜𝑟 𝑘 = 0,
𝑓 𝑜𝑟 𝑘 = 1,
𝑓 𝑜𝑟 1< 𝑘 ≤4,
𝑓 𝑜𝑟 4< 𝑘 ≤8,
𝑓 𝑜𝑟 8< 𝑘 ≤12,
𝑓 𝑜𝑟 𝑘 > 12.
(3)
This specification enables agents to revise their expectations about
future exogenous shocks, which provides additional flexibility by al-
lowing for anticipated future shocks that fail to materialize. We start
with a model specification in which QoC news shocks are muted for
the analysis presented here. In the second step we estimate a model
that considers TFP news and QoC news shocks. In Section 5.5 below,
IST news shocks are also included to assess their potential role as a
source of aggregate fluctuations once QoC news shocks are considered.
4. Data and estimation
The estimation procedure for the different model specifications
uses US data for nine macroeconomic variables: output growth, con-
sumption growth, investment growth, wage growth, hours worked,
inflation, the nominal interest rate, the spread (risk premium), and
the growth rate in the net worth of banks.7The set of observables is
7The observable for the interest rate spread is the credit spread estimated
by Gilchrist and Zakrajšek (2012) and the net worth observable is the total
equity capital for US commercial banks used in Görtz and Tsoukalas (2017).
the same as that in Smets and Wouters (2007), with the addition of
the credit spread and the net worth of banks, which seek to provide
information about financial reaction to alternative shocks. Financial
variables have shown a remarkable power to predict future economic
activity (Espinoza et al. 2012;Gilchrist and Zakrajšek,2012), which
in our case may help to distinguish the news component from the
unanticipated component of shocks. The predictive power of these
variables is due to their flexibility in adjusting more rapidly to shifts
in expectation than other (macroeconomic) observables that exhibit
a rather high degree of persistence (sluggishness). Moreover, given
that the sample period considered in the estimation includes the Great
Recession, which started around 2008, we have replaced those values of
the Fed funds rate that reach the zero lower bound by the shadow rate
constructed by Wu and Xia (2016).8The sample considered includes
the period 1987q1–2018q4, where the starting quarter is determined
by data availability for all the time series considered in the empirical
analysis. All the time series used in the estimation procedure are
transformed into (log) deviations from their respective means, so the
measurement equations are straightforward. Sample means and long-
term growth rates are removed because low frequencies may affect the
estimation of the business cycle dynamics.9The Bayesian estimation
procedure follows standard techniques (Fernández-Villaverde,2010)
and is implemented with the Dynare toolbox.10
Calibration and priors
The DSGE model seeks to reproduce business cycle features, so
several parameters that govern long-run growth are calibrated due to
a lack of identifiability. Table 1 shows the parameters calibrated and
their specific values. The discount factor 𝛽is 0.99, which implies a
quarterly real interest rate of one percent. Wage and price markup are
assumed to be 0.2. The quarterly depreciation rate is 0.025 and the
share of government spending is assumed to be 0.2. The parameters
associated with the financial sector – such as the steady-state fraction
of funds given to new bankers, and the fraction of funds that bankers
may divert – are set to hit the following two targets that correspond
to the data mean over the sample period: a steady-state (annualized)
interest rate spread of 200 basis points, and a steady-state leverage ratio
of 5.47. The survival rate parameter, 𝜃, is fixed at 0.96 as in Görtz and
Tsoukalas (2017).11
The prior distribution of the structural parameters estimated is the
same as in Smets and Wouters (2007). The prior distributions of all
8Recent papers (Wu and Zhang,2019;Mouabbi and Sahuc,2019;Aguirre
and Vázquez,2020) use the shadow rate instead of the federal funds rate
in the estimation of New-Keynesian frameworks. The estimation exercise was
also conducted with the Fed funds rate, and analogous results were obtained,
showing its robustness.
9Del Negro et al. (2007) suggest this low-frequency misspecification issue
and several other papers in the related literature also follow this data treatment
(Christiano et al.,2014;Görtz and Tsoukalas,2017). For instance, Christiano
et al. (2014) argue that they ‘‘remove sample means separately from each variable
in order to prevent counterfactual implications of the model for the low frequencies
from distorting inference in the higher business cycle frequencies that interest us.
For example, on average consumption grew faster than GDP in our dataset, while
our model predicts that the log of the consumption to GDP ratio is stationary’’.
Since we are also dealing with a relatively small sample in our paper, we
face issues similar to those pointed out in Christiano et al. (2014) in properly
identifying the low-frequency implications of the model regarding large ratios,
so we also follow their approach of removing sample means separately from
each variable.
10 We run 2 chains of 200,000 replications and performed the Brooks and
Gelman (1998) convergence diagnosis tests to ensure convergence.
11 A supplementary Appendix describes the calibration approach in more
detail. It also reports a sensitivity analysis conducted by estimating the baseline
model for the calibration used in Villa (2016), which uses lower values of 4 and
150 basis points for the steady-state leverage ratio and the spread, respectively,
and a higher value of 0.972 for the survival rate parameter. The estimation
results remain robust for this alternative calibration.
Economic Modelling 124 (2023) 106283
5
L. Herrera and J. Vázquez
innovations are also assumed to follow inverse gamma distributions,
with a mean of 0.1 and a standard deviation of 2.12
5. Estimation results
This section presents the results for estimating the DSGE model for
the alternative news shock specifications analyzed in this paper. The
first model specification mutes QoC news shocks, but the second in-
cludes them. This exercise of estimating alternative news specifications
lets the data determine whether considering a distinctive impact of
news shocks on the financial sector as implied by QoC news is a more
suitable assumption.
5.1. Model fit
The upper panel of Table 2 shows the (log) marginal data density
(MDD) associated with each model specification of news. The MDD
is based on the modified harmonic mean estimator (Geweke,1999).
Fernández-Villaverde and Rubio-Ramírez (2004) show that the MDD
favors the model specification closest to the true data generating pro-
cess. The specification that includes QoC news shocks outperforms the
specification that includes only TFP news shocks by almost 60 points.
This major improvement in model fit underscores the importance of
QoC news shocks.13 The major improvement found somewhat contrasts
with the small differences in the MDD between the QoC and TFP
news shock specifications found in Görtz and Tsoukalas (2017). We
argue that these contrasting results are mainly due to the different
approaches considered in the two papers to introduce QoC news shocks.
In our model, a positive QoC news shock triggers an appreciation of
the value of assets (capital) and positively affects the overall credit
market. In contrast, a sector-specific QoC news shock, as in Görtz
and Tsoukalas (2017), results in a credit reallocation effect. Thus, a
positive sector-specific QoC news shock has an expansionary effect on
that particular sector but a recessionary impact on credit in the other
sector. Hence, this credit reallocation effect partially offsets, from a
quantitative perspective, the shock transmission mechanism that this
paper suggests to be the most important for QoC news shocks, as
discussed below.14
To identify the sources of the major improvement in model fit,
the middle-left panel of Table 2 shows the RMSE statistics associated
with each filtered variable generated by the two specifications stud-
ied: (i) the specification including TFP news shocks alone; (ii) the
baseline specification including QoC news shocks in addition to TFP
news shocks. The improvement in model fit is observed to be espe-
cially large for hours, inflation, and the investment growth rate (the
RMSE statistics decrease by 22.2%, 17.4%, and 11.4%, respectively,
when QoC news shocks are included) but more modest for the rest
of the observable variables (the reduction in the RMSE statistic is
less than 10%). This table also contains a column showing the RMSE
statistics of the one-quarter-ahead forecast provided by the Survey
12 The results are robust to more conservative priors for news shocks, such
as those chosen in Christiano et al. (2014), which impose priors so that the
variance of the unanticipated component is 50% of the total variance of the
shock. The posterior estimates of standard deviations featuring news shocks
are much lower than those associated with surprise shocks, which means that
the data provide information about the low variability of news shocks relative
to other shocks.
13 We have also estimated a model including only QoC news shocks and
obtained an MDD of −986, suggesting that the exclusion of TFP news shock
does not worsen model fit.
14 However, from a qualitative perspective, our results are in line with the
important insight put forward by Görtz and Tsoukalas (2017), who suggest
that an endogenous financial sector is key for the proper identification of
news shocks. In addition, our results suggest that the data support news shocks
directly affecting the credit flow.
of Professional Forecasters (SPF) concerning actual data.15 Comparing
the model-implied RMSE statistics with those implied by the SPF, we
conclude that the model with QoC news shocks outperforms the one
that ignores them for all observable variables with an SPF counterpart,
and especially for inflation and the growth rate of investment. In
short, these results suggest that the improvement in fit triggered by
the inclusion of QoC news shocks is mainly due to their ability to fit
macroeconomic variables.
The middle-right and bottom panels of Table 2 show several actual
and theoretical second moments derived from the posterior distribution
of the estimated parameters (namely, the standard deviation, the first-
order autocorrelation, and the correlation with output growth for each
observable variable obtained from actual data and the two estimated
specifications). The results for the second-moment statistics align with
those obtained by comparing the log density across the two news
shock specifications. The specification that includes QoC news shocks
performs better than the one with TFP news shocks alone in terms of
matching most of the second-moment statistics except the correlation
between output growth and consumption and investment growth since
the latter specification seems, in general, to induce too much volatility
across observed variables.
Along with the improvement in model fit and the matching of the
second-moment statistics provided by a specification that includes QoC
news shocks, we also contribute to the related literature by assessing
how QoC news shocks help to shape the expectations of forward-
looking variables (i.e., the expectation channel). This assessment is
important because the improvement in model fit must be closely related
to the ability of new shocks to characterize model expectations of
observed (forward-looking) variables used in the estimation procedure
of the DSGE model. The performance of expectations built on news
shocks can be further assessed using external information sources; thus,
the empirical validity of expectations based on news shocks can be
assessed by studying their ability to match the forecasts reported in the
SPF. The middle column in the second panel shows the RMSE statistics
of the one-quarter-ahead forecasts of the observable variables concern-
ing the forecasts reported in the SPF. We find that the expectations
generated by a model specification that amplifies the effects of the
credit channel via QoC news shocks are much closer to SPF forecasts.
This result reveals that this specification is better at capturing actual
agents’ expectations as reported in the SPF.
5.2. Parameter estimates
Table 3 shows the prior distribution, the posterior mean, and the
90% higher posterior density interval (between brackets) of the struc-
tural parameters and the estimated standard deviations of news shocks.
Notably, the estimated persistence of TFP shocks is greatly reduced
when QoC shocks are considered. This result suggests that the high
persistence of TFP shocks is due to the omission of an important source
of shocks, in the form of QoC shocks. Moreover, the reduction in the
persistence of TFP shocks explains their relative lack of importance
in the variance decomposition analysis carried out below when QoC
shocks are included in the DSGE model. Interestingly, the structural
parameter estimates are rather robust across the alternative specifica-
tions of the DSGE model with news shocks; however, there are a few
noticeable differences. Therefore, habit formation and the response of
the nominal interest rate to output are estimated as larger under the
specification that includes QoC new shocks. In contrast, the elasticity
of capital utilization adjustment cost and, as highlighted above, the
persistence of TFP shocks decrease greatly in this baseline specification
with QoC news shocks.
15 This survey is conducted by the Federal Reserve Bank of Philadelphia and
is publicly available on their website. The sample period considered for the
SPF matches that of the estimation sample.
Economic Modelling 124 (2023) 106283
6
L. Herrera and J. Vázquez
Table 2
Model fit assessment.
TFP QoC
Marginal data density −1051.70 −996.05
RMSE RMSE to SPF Standard deviation
TFP QoC SPF TFP QoC Actual TFP QoC
Output growth 0.58 0.56 0.50 0.19 0.09 0.59 1.08 0.99
Consumption growth 0.57 0.52 0.50 0.79 0.79 0.56 0.97 0.57
Investment growth 1.67 1.48 1.44 0.19 0.09 1.84 3.87 3.40
Hours 0.54 0.42 – – – 4.30 4.30 3.20
Wage growth 0.87 0.86 – – – 0.86 1.14 0.92
Inflation 0.23 0.19 0.19 0.03 0.02 0.24 0.42 0.33
Spread 0.17 0.16 – – – 0.25 0.56 0.43
Interest rate 0.09 0.09 – – – 0.79 0.61 0.52
Net worth growth 2.10 2.20 – – – 1.53 8.52 6.36
Autocorrelation Correl. with output growth
Actual TFP QoC Actual TFP QoC
Output growth 0.29 0.63 0.38 1 1 1
Consumption growth 0.33 0.74 0.34 0.66 0.67 0.51
Investment growth 0.68 0.64 0.60 0.66 0.66 0.71
Hours 0.99 0.98 0.97 0.21 0.46 0.18
Wage growth −0.15 0.28 0.14 −0.04 0.41 0.19
Inflation 0.62 0.74 0.71 0.05 0.25 −0.08
Spread 0.89 0.80 0.81 −0.57 −0.43 −0.36
Interest rate 0.98 0.98 0.97 0.13 0.40 0.20
Net worth growth 0.22 −0.05 0.02 0.04 0.30 0.35
Table 3
Selected parameter estimates.
Parameter Prior distribution Posterior Mean
Type Mean/Std TFP QoC
Structural parameters
Investment adjustment cost Normal 4/1.5 1.19[0.71, 1.64] 0.74 [0.47, 0.98]
Habit formation Normal 0.7/0.1 0.68 [0.62, 0.74] 0.94 [0.90, 0.98]
Calvo probability for wages Beta 0.5/0.1 0.77 [0.70, 0.85] 0.79 [0.72, 0.86]
Elasticity of labor supply Normal 2/0.5 1.09 [0.25, 1.88] 1.69 [0.91, 2.40]
Calvo probability for prices Beta 0.5/0.1 0.94 [0.93, 0.95] 0.94 [0.93, 0.95]
Indexation of past inflation in wages Beta 0.5/0.15 0.38 [0.15, 0.60] 0.21 [0.08, 0.33]
Indexation of past inflation in inflation Beta 0.5/0.15 0.21 [0.07, 0.34] 0.19 [0.07, 0.30]
Utilization adjustment cost Gamma 0.5/0.15 0.95 [0.91, 0.98] 0.69 [0.51, 0.88]
Fixed cost in production Normal 1.25/0.125 1.73 [1.58, 1.88] 1.65 [1.48, 1.81]
Capital share in production Normal 0.3/0.05 0.19 [0.15, 0.22] 0.24 [0.20, 0.28]
Monetary policy parameters
Interest rate smoother Beta 0.75/0.1 0.80 [0.75, 0.84] 0.80 [0.76, 0.84]
Response to inflation Normal 1.5/0.25 1.11 [1.00, 1.24] 1.19 [1.00, 1.64]
Response to output Normal 0.125/0.05 0.08 [0.04, 0.14] 0.36 [0.30, 0.42]
Response to output growth Normal 0.125/0.05 0.18 [0.11, 0.25] 0.15 [0.08, 0.22]
TFP news shocks
Persistence of TFP Beta 0.5/0.2 0.95 [0.92, 0.98] 0.31 [0.18, 0.44]
Std of TFP news shock - 1 quarter ahead Gamma 0.1/2 0.06 [0.03, 0.08] 0.10 [0.02, 0.19]
Std of TFP news shock - 4 quarter ahead Gamma 0.1/2 0.07 [0.03, 0.11] 0.06 [0.02, 0.10]
Std of TFP news shock - 8 quarter ahead Gamma 0.1/2 0.08 [0.03, 0.14] 0.07 [0.02, 0.11]
Std of TFP news shock - 12 quarter ahead Gamma 0.1/2 0.12 [0.05, 0.18] 0.17 [0.08, 0.27]
QoC news shocks
Persistence of QoC Beta 0.5/0.2 – 0.93 [0.87, 0.98]
Std of QoC news shock - 1 quarter ahead Gamma 0.1/2 – 0.05 [0.03, 0.08]
Std of QoC news shock - 4 quarter ahead Gamma 0.1/2 – 0.05 [0.02, 0.07]
Std of QoC news shock - 8 quarter ahead Gamma 0.1/2 – 0.06 [0.03, 0.10]
Std of QoC news shock - 12 quarter ahead Gamma 0.1/2 – 0.11 [0.03, 0.19]
5.3. News shocks as a driving force of the business cycle
Fig. 1 shows the proportion of the variance decomposition explained
by the two types of news shock for the set of observable variables con-
sidered in the estimation across alternative forecast horizons. Fig. 1(a)
shows the model where TFP news shocks are estimated alone, while
Fig. 1(b) shows the proportion of the variance decomposition explained
by QoC (solid black line) and TFP (red dashed line) news shocks when
both types are included in the DSGE model estimated. The results
shown in Fig. 1(a) align with those reported in the related literature,
where TFP news shocks are highlighted as a significant driving force
of the business cycle (Beaudry and Portier,2006;Fujiwara et al.,
2011;Görtz and Tsoukalas,2017).
The main finding of this analysis is that the data support the inclu-
sion of QoC news shocks in the estimated DSGE model in detriment to
TFP news shocks, whose importance as a driving force of the business
cycle is substantially reduced, as shown in Fig. 1(b). Thus, when TFP
news shocks are considered alone, they explain a substantial propor-
tion of the variability of all observed variables. More precisely, they
explain around 50% of output, investment, and nominal interest rate
fluctuations and one third of wage, inflation, and labor fluctuations.
They also explain a large proportion of the variability associated with
Economic Modelling 124 (2023) 106283
7
L. Herrera and J. Vázquez
Fig. 1. Conditional variance decomposition: Assessing the importance of TFP news vs. QoC news.
the two financial variables considered (approximately one-third across
medium- and long-term forecast horizons). In sharp contrast, the in-
clusion of QoC news shocks in addition to TFP news shocks results in
a large drop in the relative importance of the latter in explaining the
variability of many macroeconomic and financial variables. However,
they are still quantitatively very important in explaining inflation,
wage, and short-run consumption fluctuations. Nonetheless, QoC news
shocks are generally much more significant than TFP news shocks in
explaining aggregate fluctuations.
These results are clearly due to the financial impact of QoC news
shocks. Consider, for instance, that agents anticipate a positive QoC
four quarters in advance. This positive news shock affects the economy
through the production function and the credit channel. On the one
hand, positive QoC and TFP news shocks have an equivalent effect
on the production function since both types increase expected future
productivity (see Eq. (1)). On the other hand, in the financial market,
a positive realization of QoC news shock results in a rise in asset prices
since agents anticipate an improvement in the quality of capital, as
shown by Eq. (2). This rise in asset prices immediately impacts banks’
balance sheets since their held assets increase in value. Moreover,
banks’ expected profits increase further due to the expected rise in
capital value, which increases credit supply and investment. Some
Economic Modelling 124 (2023) 106283
8
L. Herrera and J. Vázquez
subtle differences aside, this view is largely consistent with the results
in Beaudry and Portier (2006), where news shocks are identified with
shocks impacting the financial market (stock prices) and anticipating
future movements in TFP.16
The supplementary Appendix to this paper shows the conditional
variance decomposition for all (surprise) shocks associated with the
baseline specification. Given the major importance of QoC news shocks
in explaining the fluctuations of real and financial variables, the rela-
tive importance of unanticipated shocks substantially decreases when
compared to the variance decompositions obtained in the literature
abstracting from news shocks (Smets and Wouters,2007;Villa,2016).
Nonetheless, price- and wage-markup shocks still explain a large pro-
portion of inflation and wage fluctuations when news shocks are in-
cluded.17
5.4. Impulse response functions
The previous sections provide evidence favoring QoC news shocks to
the detriment of TFP news shocks. This section provides further insights
into this result through an impulse-response function (IRF) analysis.
Fig. 2 shows the responses of output, consumption, investment, asset
prices (the price of capital), net worth, interest rate spread, inflation,
the nominal interest rate, and hours worked to alternative one-percent
news shocks. The solid black line represents the IRF of each variable
to a four-quarter QoC news shock, while the dashed red line shows the
IRF to a four-quarter nonstationary TFP news shock.
Notably, QoC and TFP news shocks can generate sound, positive co-
movements between output, consumption, investment, and labor; how-
ever, the transmission mechanisms are substantially different. Thus, a
QoC news shock results in greater responses of real variables (output,
investment, and labor) and the credit spread than those of a TFP
news shock at impact. In contrast, TFP news shocks produce a large
response of consumption at impact, whereas consumption reacts much
more slowly for QoC news shocks. This finding means that a positive
QoC news shock results in a greater boost for investment relative to
consumption than TFP news shocks. More precisely, a positive TFP
news shock leads agents to anticipate higher output in the future and,
consequently, to increase their consumption in advance. In contrast, a
positive QoC news shock has the same effect as a TFP news shock (recall
that TFP and QoC news shocks are tantamount when only looking at
the production function) but also leads to a much lower spread that
mainly affects the real side of the economy through an expansion in
credit supply. Moreover, notice that a QoC shock means a lower peak
in the impulse responses of all real variables than a TFP shock. This
result is rather intuitive for two reasons. First, the positive response of
the nominal interest rate is much stronger in the short/medium term
16 Many studies have stressed the predictability of future economic activity
using financial variables. Gilchrist et al. (2009) determine that credit market
factors from corporate bond spreads predict future output, employment, and
industrial output movements. Espinoza et al. (2012) show that shocks to finan-
cial variables influence real activity. Gilchrist and Zakrajšek (2012) construct a
new corporate bond credit spread index that robustly predicts future economic
activity. Aguilar and Vázquez (2021) and Vázquez and Aguilar (2021) show
that the term spread plays an important role in the characterization of adaptive
learning dynamics in DSGE models.
17 The supplementary Appendix also shows that estimation results are fairly
robust when the sample period is restricted to 2006 (i.e., ignoring the Great
Recession period). (i) There is a significant improvement of the model fit in
terms of the MDD when considering QoC news shocks; (ii) similar results
are found for the importance of QoC news in explaining the fluctuations
of observables. Finally, (iii) when QoC news shocks are considered, the
importance of TFP news shocks is greatly reduced in favor of the former. A
noteworthy difference when this reduced sample is considered is the drop in
the importance of QoC news shocks in explaining output fluctuations, whereas
estimation results are fairly robust for the rest of the observables.
under a QoC shock than under a TFP shock, which partially offsets
the expansionary effects of news shocks. Second, the IRF of the spread
reverts more rapidly to the steady state under a QoC shock than under
a TFP shock.
Another noteworthy difference between QoC and TFP news shocks
is the inflation response to them, which is mildly negative for QoC news
shocks, while there is an inflationary response to TFP news shocks. This
inflationary response to TFP shocks is due to two main effects: (i) the
greater reaction of marginal costs to a TFP news shock and (ii) the
milder reaction of the nominal interest rate to such shocks. The reaction
of marginal cost is larger for TFP news because real variables need to
overreact to produce high fluctuations in financial markets, triggering
an inflationary process. The greater reaction of the nominal interest
rate in the case of QoC news shocks also enables inflation expectations
to be anchored. Both effects incite a change from an inflationary to a
deflationary response when the effects of TFP and QoC news shocks
are compared. Importantly, this deflationary response of QoC news
shocks is in line with the VAR analysis carried out in Görtz et al.
(2022); thus, our findings contribute to the literature on the effects of
financial shocks where the inflation reaction to these shocks has been
part of an ongoing debate. We find evidence of a deflationary response
to financial shocks, which is in line with findings in Meh and Moran
(2010) and somewhat in contrast with Benes and Kumhof (2015), Ajello
(2016), Villa (2016). A positive QoC news shock acts as a supply shock,
leading to an expansionary response in output and a fall in inflation.
5.5. Why do QoC news shocks fit better than IST news shocks?
The previous sections show that by amplifying financial markets
through the credit channel, QoC news shocks induce a stronger propa-
gation mechanism than TFP news shocks. More precisely, this is due to
the more pronounced effect of QoC news shocks on interest rate spreads
and thus on the credit supply. IST and QoC news shocks are expected
to have similar effects on real macroeconomic variables. Indeed, using
a VAR approach, Ben Zeev and Khan (2015) also find that IST news
shocks reduce the importance of TFP news shocks, as QoC news shocks
do in our analysis based on DSGE modeling. We estimate a model
specification that includes QoC and IST news shocks and TFP news
shocks to shed light on this matter.
Fig. 3 shows the proportions of aggregate variability explained by
IST, QoC, and TFP news shocks. It is noteworthy that IST news shocks
do not explain aggregate fluctuations, while QoC news shocks remain
highly important. Moreover, the (log) MDD when IST news shocks are
included (−997.43) is roughly similar to the baseline case where they
are omitted (−996.05). These results indicate that IST news shocks add
nothing when QoC news shocks are already considered in the analysis.
In short, our empirical findings suggest that the results of Ben Zeev
and Khan (2015), showing that IST news shocks displace TFP news
shocks, can be viewed as a veil cast over the financial effects captured
by QoC news shocks. The reason why the data favors QoC news shocks
in DSGE modeling lies in the effect of IST news shocks on the price
of assets, which is ignored in a VAR analysis. Fig. 4 shows the IRFs of
asset prices for a one-percent positive (i) QoC news shock, (ii) TFP news
shock, and (iii) IST news shock, each anticipated 4-quarters in advance.
Notably, QoC and TFP news shocks positively affect asset prices, so
the supply of credit rises, thus pushing up investment (although the
response of investment is larger for QoC, as discussed above). By
contrast, IST news shocks negatively affect asset prices. Therefore, the
rise in investment triggered by IST news shocks is partially offset by the
credit supply contraction induced by the drop in asset prices. These
results confirm findings in Görtz and Tsoukalas (2017) in a DSGE
framework, including news and those of Kamber et al. (2015), which
only consider unanticipated shocks. Moreover, our findings shed light
on the importance of matching the comovements between financial and
Economic Modelling 124 (2023) 106283
9
L. Herrera and J. Vázquez
Fig. 2. IRF to QoC and TFP news shocks.
real variables (investment) in which a specification including QoC news
shocks performs better.18
5.6. The role of pure news
A distinctive feature of news shocks is that they affect aggregate
variables without changing fundamentals. They do so through agents’
expectations; however, in the standard specification (also followed in
this paper) all innovations realize but other news innovations can
offset (revise) their effects, characterizing the news revision process
and nonrealized news. Sims (2016) argues that an analysis of the
importance of news shocks through variance decomposition may be
biased because it accounts for the pure news effects of each innovation
but also for the effect of the shock once it is realized (in which case the
effects are not substantially different from those of a standard surprise
shock). Sims (2016) suggests a method for separating these two effects
to assess whether pure news shocks matter and whether news shocks
affect aggregate variables without changing fundamentals (i.e., through
the expectation channel). More precisely, he distinguishes between two
impulse response functions: those associated with pure news and those
based on realized news shocks. A pure news IRF is equal to the IRF
associated with a news shock at horizons before the realization of that
news and zero at horizons after that. Conversely, a realized news IRF
takes a value of zero before the realization of the news shock and takes
on the values of the IRF for news shocks at horizons after that.
18 Section 7 of the supplementary Appendix shows the results of a robustness
exercise where we allow all shocks to follow an AR(1) process augmented
with a news shock component. In addition to QoC news shocks, we find other
news shocks that seem rather important. The most noteworthy case is that
of net worth news shocks, which can explain roughly one-fourth of output
and interest rate fluctuations, and at least one-third of each financial variable.
It is also interesting to notice that the two (price and wage) markup news
shocks explain 32% and 20% of inflation and wage fluctuations, respectively.
Caution is advised in analyzing these results since considering such a large
number of news shocks without including additional observables may affect
their identification. We are further considering the possibility of analyzing the
roles of all types of news in future work by including more observables to
capture expectation changes, such as those reported in the SPF, which may
help to discriminate between alternative news sources.
We conduct the decomposition proposed by Sims (2016) to assess
whether pure QoC news shocks are a major source of macroeconomic
fluctuations or whether their importance in the variance decomposition
is due to realized news shocks. Fig. 5 shows the conditional variance
decomposition for alternative forecast horizons.
In the long-run pure QoC news shocks account for 31% of output
fluctuations, which make up 73% of the total contribution of news
shocks. Pure QoC news shocks account for roughly 20% of fluctua-
tions in investment, interest rate spread, and nominal interest rate. In
contrast, the proportion of pure QoC news shocks that explain long-
run consumption fluctuations is very modest. The news decomposition
suggests that pure news has an initial impact on investment through
the credit channel, and the effect on consumption is mainly due to the
investment reaction to the news. This result underscores the importance
of the credit channel in producing an expectation-driven business cycle,
as suggested by Pigou (1927).19
In short, our findings reveal the importance of considering a finan-
cial sector and QoC news shocks, which amplify the credit channel
in explaining aggregate fluctuations in the real economy and financial
markets. In contrast with the framework considered in Sims (2016), the
direct impact of QoC news shocks in financial markets is transmitted
smoothly, through the credit channel, to the rest of the economy.20
19 The importance of pure QoC news is somewhat in contrast to that found
by Sims (2016) on analyzing the importance of pure TFP news in a rather
different framework (i.e., using the real business cycle model of Schmitt-Grohé
and Uribe,2012). He finds pure TFP news is relatively unimportant, suggesting
that such news shocks do not qualitatively differ from surprise shocks. These
contrasting results do not come as a surprise since, as pointed out by Görtz and
Tsoukalas (2017), the lack of transmission channels linking financial markets
with real economic activity in a DSGE framework may substantially affect
the identification of news shocks and their relative contribution in explaining
aggregate fluctuations.
20 Sims (2016) also argues that under this analysis the role of pure news
could be underestimated since the variance decomposition analysis does not
account for the effects of unrealized news (surprise shocks that offset news
shocks are interpreted as unrealized news but are accounted for in the variance
decomposition as surprise shocks).
Economic Modelling 124 (2023) 106283
10
L. Herrera and J. Vázquez
Fig. 3. Variance Decomposition of the DSGE model with QoC, TFP, and IST news shocks.
Fig. 4. IRF of the price of capital to QoC, TFP, and IST news shocks.
Fig. 5. Pure vs. realized QoC news shocks.
Economic Modelling 124 (2023) 106283
11
L. Herrera and J. Vázquez
6. Conclusions
The importance of news shocks as a major driver of economic
fluctuations has been stressed in recent literature (Beaudry and Portier,
2006;Fujiwara et al.,2011;Schmitt-Grohé and Uribe,2012;Görtz and
Tsoukalas,2017). We find that it is crucial to consider the financial
impact of such shocks. We provide evidence that actual data supports
a version of a standard DSGE model with financial frictions à la Gertler
and Karadi (2011). This finding suggests that QoC news shocks impact
financial markets by affecting the price of assets and the balance
sheets of banks, thus triggering an amplifying effect through the credit
channel.
More precisely, this paper contributes to two important strands of
the literature, namely news shocks and financial frictions. We show that
by amplifying financial markets, QoC news shocks displace standard
TFP (and IST) news shocks as a driving force of the business cycle.
This result can be understood through the different qualitative and
quantitative effects of each type of shock on real variables, such as
investment and consumption; TFP news shocks affect both variables on
impact, but QoC news shocks mainly affect the investment decision.
Moreover, the latter’s effects through the credit channel are much
larger than those of TFP news. This result is also noticeable in the
greater effects of QoC news shocks on financial variables; thus, TFP
news shocks need to be much larger than QoC news shocks to fit
financial data.
The paper also provides empirical evidence on the importance
of pure QoC news. We show that the effects of QoC news shocks
are mainly driven by pure news rather than realized news through
the methodology proposed by Sims (2016). This result stresses the
importance of the expectation channel in transmitting news shocks.
This paper provides robust empirical evidence suggesting that QoC
news shocks can provide a proper way to model expectations-driven
business cycles. This empirical evidence is in line with Beaudry and
Portier (2006), who find that news shocks are identified with shocks
impacting the financial market (stock prices) and anticipating future
movements in TFP, and more generally with Pigou (1927) by showing
that, by affecting business people’s expectations, news is an important
driver of aggregate fluctuations.
Declaration of competing interest
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared to
influence the work reported in this paper.
Data availability
We share the data and the code to run the baseline case considered
in the paper (see Appendix B).
Acknowledgments
The research of these authors was supported by the Spanish Min-
istry of Science and Innovation, MCIN/AEI /10.13039/501100011033/
under grant number PID2020-118698GB-I00, by the Spanish Ministry
of Economy and Competition under grant number ECO2016-78749P,
and by the Basque Government under grant number IT-1336-19. The
first author also acknowledges Financial support from the Spanish Min-
istry of Science, Innovation and Universities under scholarship grant
FPU17/06331.
Appendix A
This Appendix describes the DSGE model augmented with financial
frictions à la Gertler and Karadi (2011).
Households
The representative household 𝑖decides consumption, hours worked,
and savings in riskless assets to maximize a utility function that incor-
porates internal habit formation. Formally,
𝐸𝑡
∞
𝑘=0
𝛽𝑘𝜖𝑏
𝑡+𝑘𝑙𝑛 𝐶𝑡+𝑘(𝑖) − ℎ𝐶𝑡+𝑘−1−𝐿𝑡+𝑘(𝑖)1+𝜎𝑙
1 + 𝜎𝑙,(4)
where 𝛽is the household discount factor, ℎrepresents the degree of
habit persistence, 𝜎𝑙is the elasticity of labor supply (i.e. the Frisch elas-
ticity), and 𝜖𝑏
𝑡+𝑘is an exogenous process that affects the intertemporal
preferences of households. Deposit liabilities in banks and government
bonds represent household savings. These riskless assets, 𝐵𝑡+𝑘, are
perfect substitutes and pay the same nominal interest rate, 𝑅𝑛
𝑡+𝑘. House-
holds also obtain dividends from intermediate goods firms, capital
goods producers, and labor unions, 𝐷𝑡+𝑘. Hence, the budget constraint
is given by
𝐶𝑡+𝑘(𝑖) + 𝐵𝑡+𝑘(𝑖)
𝑅𝑛
𝑡+𝑘𝑃𝑡+𝑘
−𝑇𝑡+𝑘=𝑊𝑡+𝑘(𝑖)𝐿𝑡+𝑘(𝑖)
𝑃𝑡+𝑘
+𝐵𝑡+𝑘−1(𝑖)
𝑃𝑡+𝑘
+𝐷𝑡+𝑘
𝑃𝑡+𝑘
,(5)
where 𝑇𝑡+𝑘represents lump-sum taxes and 𝑊𝑡+𝑘is the nominal wage.
Labor unions and wage decision
As in Smets and Wouters (2007), households supply homogeneous
labor to intermediate labor unions that differentiate labor services.
Those intermediate labor unions then set wages to sell labor services
to a labor packer who aggregates the differentiated labor and resells it
to intermediate goods firms. Aggregation of labor services follows
𝐿𝑡=∫1
0
𝐿𝑡(𝑖)
1
1+𝜖𝑤
𝑡𝑑𝑖1+𝜖𝑤
𝑡
,
where 1 + 𝜖𝑤
𝑡is the desired markup of wages over the household’s
marginal rate of substitution, which is assumed to follow a stochastic
process around its steady-state value. Labor packers maximize profits
in a perfectly competitive market
𝑚𝑎𝑥𝐿𝑡(𝑖)𝑊𝑡𝐿𝑡−∫1
0
𝑊𝑡(𝑖)𝐿𝑡(𝑖),
where 𝐿𝑡is subject to the labor aggregation function, 𝑊𝑡is the ag-
gregate wage that intermediate firms pay for labor services, and 𝑊𝑡(𝑖)
is the wage that labor packers pay for the differentiated labor. This
optimization problem gives rise to the following labor demand function
𝐿𝑡(𝑖) = 𝑊𝑡(𝑖)
𝑊𝑡−1+𝜖𝑤
𝑡
𝜖𝑤
𝑡𝐿𝑡.
The labor demand function and the labor services aggregation function
jointly result in the wage aggregation function
𝑊𝑡=∫1
0
𝑊𝑡(𝑖)
1
𝜖𝑤
𝑡𝑑𝑖𝜖𝑤
𝑡
.(6)
Following Calvo’s lottery scheme, it is assumed that labor unions
can only adjust prices with probability 1 − 𝜉𝑤. The fraction of labor
unions 𝜉𝑤that cannot adjust prices is assumed to follow the indexation
rule, 𝑊𝑡+1(𝑖) = 𝑊𝑡(𝑖)𝑃𝑡
𝑃𝑡−1 𝜄𝑤
. Hence, the labor unions choose an
optimal 𝑊to maximize
𝐸𝑡
∞
𝑘=0
𝛽𝑘𝜉𝑘
𝑤𝛬𝑡+𝑘𝑊𝑡(𝑖)𝐿𝑡+𝑘(𝑖) − 𝜖𝑏
𝑡+𝑘
𝐿𝑡+𝑘(𝑖)1+𝜎𝑙
1 + 𝜎𝑙,(7)
subject to labor demand and the indexation rule.
Final goods firms
Competitive final goods producers buy intermediate goods and
assemble them to sell homogeneous goods to households. The inter-
mediate goods aggregation follows
𝑌𝑡=∫1
0
𝑌𝑡(𝑖)
1
1+𝜖𝑝
𝑡𝑑𝑖1+𝜖𝑝
𝑡
,
Economic Modelling 124 (2023) 106283
12
L. Herrera and J. Vázquez
where 𝑌𝑡is the homogeneous good, and 𝑌𝑡(𝑖)is the heterogeneous good
supplied by firm 𝑖.1 + 𝜖𝑝
𝑡is the desired markup of prices over firms’
marginal costs, which is assumed to follow a stochastic process around
its steady-state value. Final goods firms maximize profits in a perfectly
competitive market
𝑚𝑎𝑥𝑌𝑡(𝑖)𝑃𝑡𝑌𝑡−∫1
0
𝑃𝑡(𝑖)𝑌𝑡(𝑖)𝑑𝑖,
where 𝑌𝑡is subject to the goods aggregation function, 𝑃𝑡(𝑖)is the
price for differentiated goods, and 𝑃𝑡is the aggregate price index. The
optimal condition of this maximization problem results in the following
demand function for goods:
𝑌𝑡(𝑖) = 𝑃𝑡(𝑖)
𝑃𝑡−1+𝜖𝑝
𝑡
𝜖𝑝
𝑡𝑌𝑡.(8)
Hence, the goods demand function and the intermediate goods aggre-
gator result in the following price aggregator
𝑃𝑡=∫1
0
𝑃𝑡(𝑖)
1
𝜖𝑝
𝑡𝑑𝑖1
1−𝜖𝑝
𝑡.(9)
Intermediate goods firms
As in the labor market, it is assumed that intermediate goods firms
can only adjust prices with probability 𝜉𝑝. Those firms which cannot
adjust prices in period 𝑡simply reset their prices according to the
indexation rule: 𝑃𝑡+1(𝑖) = 𝑃𝑡(𝑖)𝑃𝑡
𝑃𝑡−1 𝜄𝑝
. Firms able to decide their
optimal prices 𝑃∗
𝑡at time 𝑡choose them by maximizing current and
future expected profits. Denoting the marginal costs by 𝑀𝐶𝑡and the
inflation rate by 𝜋𝑡, the price setting optimization problem faced by
intermediate goods firms is
𝐸𝑡
∞
𝑘=0
𝛽𝑘𝜉𝑘
𝑝𝛬𝑡+𝑘
𝑃𝑡
𝑃𝑡+𝑘𝑃∗
𝑡(𝑖)
𝑘
𝑙=1
𝜋𝜄𝑝
𝑡+𝑙−1 −𝑀𝐶𝑡+𝑘𝑌𝑡+𝑘(𝑖),(10)
subject to the price indexation rule and the demand function for goods.
In addition to setting prices, intermediate goods firms decide on
the output of goods. They choose the number of production inputs by
maximizing the flow of discounted profits
𝐸𝑡𝛽𝛬𝑡+1 𝑌𝑡+1 (𝑖) − 𝑟𝑘
𝑡+1𝐾𝑠
𝑡+1(𝑖) − 𝑊𝑡+1
𝑃𝑡+1
𝐿𝑡+1(𝑖),(11)
where 𝛽𝛬𝑡+1 =𝛽 𝜆𝑡+1
𝜆𝑡
is the stochastic discount factor and 𝜆𝑡is the
marginal utility of consumption for households at time 𝑡,𝑟𝑘
𝑡is the rental
rate of capital, and 𝐾𝑠
𝑡=𝑄𝑜𝐶𝑡𝐾𝑡−1𝑈𝑡denotes capital services.
The production function is assumed to follow a Cobb–Douglas tech-
nology:
𝑌𝑡=𝑇 𝐹 𝑃𝑡𝑄𝑜𝐶𝑡𝐾𝑡−1𝑈𝑡𝛼𝐿1−𝛼
𝑡−𝐴𝑡𝜙𝑝,(12)
where 𝜙𝑝is the share of fixed costs involved in production, and the
disturbance 𝑄𝑜𝐶𝑡captures exogenous changes in the quality of capital.
Notice that QoC shocks are somewhat equivalent to TFP shocks. The
difference between them is made clear below. Shocks in the quality of
capital directly impact the real economy via the production process, as
TFP shocks do, and also have an amplifying effect through the credit
channel. The optimal inputs decision results in the following optimal
conditions:
𝑟𝑘
𝑡=𝛼𝑀 𝐶𝑡𝑇 𝐹 𝑃𝑡𝐾𝑠
𝑡𝛼−1 𝐿1−𝛼
𝑡,(13)
𝑊𝑡
𝑃𝑡
=(1 − 𝛼)𝑀𝐶𝑡𝑇 𝐹 𝑃𝑡𝐾𝑠
𝑡𝛼𝐿−𝛼
𝑡.(14)
Capital services firms
Capital services firms purchase physical capital from capital goods
producers and turn it into effective capital by choosing the utilization
rate, 𝑈𝑡:
𝐾𝑠
𝑡=𝑈𝑡𝐾𝑡−1 𝑄𝑜𝐶𝑡.(15)
Capital services firms decide the optimal capital utilization rate and
face a utilization cost. They solve the following maximization problem:
𝑚𝑎𝑥𝑈𝑡+1 𝑟𝑘
𝑡+1𝑈𝑡+1 −𝑎𝑈𝑡+1 𝐾𝑡𝑄𝑜𝐶𝑡+1 ,
where 𝑎𝑈𝑡is the utilization cost function. The optimal solution
implies
𝑟𝑘
𝑡=𝑎′𝑈𝑡.(16)
This equilibrium condition means that the degree of capital utilization
is a function of the rental rate of capital. It is assumed that the
utilization cost function features the following standard properties 𝑈=
1,𝑎(𝑈)=0, and 𝑎′′(𝑈)
𝑎′(𝑈)=𝜓in the steady state; hence, the parameter 𝜓
is a positive function of the elasticity of the capital utilization cost, and
is normalized to be between zero and one. A higher value of 𝜓implies
a higher cost of adjustment in capital utilization.
Capital services firms finance their physical capital acquisition by
borrowing from financial intermediaries. At equilibrium, the following
condition holds:
𝑄𝑡𝐾𝑡=𝑄𝑡𝑆𝑡,(17)
indicating that state-contingent claims, 𝑆𝑡, are equal to the number of
units of physical capital acquired, 𝐾𝑡, where firms price their claims
at the price of one unit of capital, 𝑄𝑡. Each claim pays the stochastic
return 𝑅𝑘
𝑡+1 over period 𝑡. Capital services firms operate in a perfectly
competitive market, so the revenue from renting effective capital must
equal the cost of purchasing physical capital. Hence, the optimal capital
demand satisfies
𝑅𝑘
𝑡+1 =
𝑟𝑘
𝑡+1𝑈𝑡+1 −𝑎𝑈𝑡+1 +𝑄𝑡+1(1 − 𝛿)
𝑄𝑡𝑄𝑜𝐶𝑡+1.(18)
This shows that the expected real interest rate on external funds equals
the marginal return on capital. Notice that the QoC shock also deter-
mines the return on financial claims.
Capital goods producers
Capital goods producers turn out physical capital and sell it to
capital services firms at price 𝑄𝑡. Investment goods are purchased from
final good producers. Capital goods producers are assumed to face
quadratic adjustment costs, 𝑆(𝐼𝑡∕𝐼𝑡−1). This adjustment costs function
is assumed to be a strictly increasing twice differentiable function; thus,
the optimization problem of the capital goods producers is
𝑚𝑎𝑥𝐼𝑡𝐸𝑡∞
𝑘=0
𝛽𝑘𝛬𝑡+𝑘𝑄𝑡+𝑘𝐼𝑡+𝑘𝜖𝑖
𝑡+𝑘−𝐼𝑡+𝑘−𝑄𝑡+𝑘𝐼𝑡+𝑘𝜖𝑖
𝑡+𝑘𝑆𝐼𝑡+𝑘
𝐼𝑡+𝑘−1 ,(19)
where 𝑆(.)is assumed to have the properties 𝑆(1) = 𝑆′(1) = 0,
𝑆′′(1) = 𝜑 > 0. Therefore, the parameter 𝜑captures the degree of
investment adjustment cost, and the disturbance 𝜖𝑖
𝑡is the investment
specific-technology shock. Capital accumulation evolves following the
standard equation
𝐾𝑡= (1 − 𝛿)𝐾𝑡−1 𝑄𝑜𝐶𝑡+1 − 𝑆𝐼𝑡
𝐼𝑡−1 𝐼𝑡.(20)
Financial intermediaries
Görtz and Tsoukalas (2017) find that the financial sector is crucial
for identifying TFP news shocks. We closely follow their characteriza-
tion of financial intermediaries, based on Gertler and Karadi (2011). A
fixed fraction of households is assumed to comprise bankers, who do
not supply labor but act as financial intermediaries. They face a survival
probability, 𝜃, and households become bankers in each period to keep
the proportion of bankers constant.
The financial intermediaries finance the acquisition of physical capi-
tal by purchasing claims 𝑆𝑡. Those purchases are funded through house-
hold liabilities. Hence, the balance sheets of financial intermediaries
are
Economic Modelling 124 (2023) 106283
13
L. Herrera and J. Vázquez
𝑄𝑡𝑆𝑡=𝑁𝑡+𝐵𝑡+1,
where 𝑁𝑡is the bankers’ net worth. Given that the return on financial
claims is 𝑅𝑘
𝑡+1 and the cost of liabilities is 𝑅𝑡, the net worth of the
intermediaries evolves as follows:
𝑁𝑡+1 =𝑅𝑘
𝑡+1𝑄𝑡𝑆𝑡−𝑅𝑡𝐵𝑡+1 =𝑅𝑘
𝑡+1 −𝑅𝑡𝑄𝑡𝑆𝑡+𝑅𝑡𝑁𝑡.
Let 𝛽𝛬𝑡+1 be the stochastic discount factor of the financial intermedi-
aries. The bankers’ decisions are endogenously determined in the model
through a problem in which they maximize future expected terminal
wealth
𝑉𝑡=𝑚𝑎𝑥 𝐸𝑡
∞
𝑖=0
(1 − 𝜃)𝜃𝑖𝛽𝑖𝛬𝑡+1+𝑖𝑁𝑡+1+𝑖=
𝑚𝑎𝑥 𝐸𝑡
∞
𝑖=0
(1 − 𝜃)𝜃𝑖𝛽𝑖𝛬𝑡+1+𝑖𝑅𝑘
𝑡+1+𝑖−𝑅𝑡+𝑖𝑄𝑡+𝑖𝑆𝑡+𝑖+𝑅𝑡+𝑖𝑁𝑡+𝑖.
However, a moral hazard issue arises in this maximization problem
because 𝛽𝑖𝑅𝑘
𝑡+1+𝑖−𝑅𝑡+𝑖≥0. Otherwise, bankers would not be willing
to purchase assets. Thus, bankers are incentivized to keep borrowing
additional funds indefinitely from households. An enforcement cost is
introduced to restrict their ability to do this. At the beginning of the
period bankers can divert a proportion 𝜆of the funds available. In
that case, the depositors can recover a fraction (1 − 𝜆)of the assets;
hence, for lenders to be willing to supply funds to bankers the following
incentive constraint must be satisfied:
𝑉𝑡≥𝜆𝑄𝑡𝑆𝑡,
where 𝑉𝑡, the gain from not diverting assets, can be expressed as follows
𝑉𝑡=𝜈𝑡𝑄𝑡𝑆𝑡+𝜂𝑡𝑁𝑡,
with
𝜈𝑡=𝐸𝑡(1 − 𝜃)𝛬𝑡+1 𝑅𝑘
𝑡+1 −𝑅𝑡+𝛽𝜃𝑥𝑡,𝑡+1𝜈𝑡+1 ,(21)
𝜂𝑡=𝐸𝑡(1 − 𝜃)𝛬𝑡+1𝑅𝑡+𝛽 𝜃𝑧𝑡,𝑡+1 𝜂𝑡+1,(22)
where 𝜈𝑡is the marginal gain from expanding assets with net worth held
constant. 𝜂𝑡is the expected value of one additional future unit of wealth
net worth with assets held constant, 𝑥𝑡,𝑡+1 =𝑄𝑡+1𝑆𝑡+1 ∕𝑄𝑡𝑆𝑡is the gross
growth rate of assets, and 𝑧𝑡,𝑡+1 =𝑁𝑡+1∕𝑁𝑡is the gross growth rate of
net worth.
In equilibrium the incentive constraint holds with equality
𝑄𝑡𝑆𝑡=𝜂𝑡
𝜆−𝜈𝑡
𝑁𝑡=𝜙𝑡𝑁𝑡,(23)
where 𝜙𝑡is the leverage ratio of bankers. Thus, from the net worth
evolution equation and the incentive constraint, net worth can be
rewritten as
𝑁𝑡+1 =𝑅𝑘
𝑡+1 −𝑅𝑡𝜙𝑡+𝑅𝑡𝑁𝑡.
Based on this equation, the gross growth rates of assets and net worth
can be expressed as
𝑧𝑡,𝑡+1 =𝑁𝑡+1∕𝑁𝑡=𝑅𝑘
𝑡+1 −𝑅𝑡𝜙𝑡+𝑅𝑡,(24)
and
𝑥𝑡,𝑡+1 =𝑄𝑡+1𝑆𝑡+1 ∕𝑄𝑡𝑆𝑡=𝜙𝑡+1∕𝜙𝑡𝑁𝑡+1∕𝑁𝑡=𝜙𝑡+1 ∕𝜙𝑡𝑧𝑡,𝑡+1.(25)
Finally, the law of motion of bankers’ net worth is given by the law
of motion of the net worth of existing bankers plus the net worth of
households that become bankers in this period:
𝑁𝑡=𝑁𝑒
𝑡+𝑁𝑛
𝑡,(26)
with
𝑁𝑒
𝑡=𝜃𝑅𝑘
𝑡+1 −𝑅𝑡𝜙𝑡+𝑅𝑡𝑁𝑡−1,(27)
𝑁𝑛
𝑡=𝜔𝑄𝑡𝑆𝑡−1,(28)
𝑁𝑡=𝑁𝑡𝜖𝑛𝑤
𝑡,(29)
where 𝜔is the fraction of the total assets that households transfer to
new bankers, which enables them to start operating in the banking
sector. The disturbance 𝜖𝑛𝑤
𝑡captures exogenous variations in the net
worth of bankers (due, for instance, to exogenous changes in bank
profits).
Market clearing condition
The market clearing condition is
𝑌𝑡=𝐶𝑡+𝐼𝑡+𝑎(𝑈𝑡) + 𝜖𝑔
𝑡,(30)
where 𝜖𝑔
𝑡is an exogenous process that captures government spending
and exogenous net export shocks.
The central bank
The model is completed with a Taylor rule in which the nominal
interest rate set by the central banker reacts to inflation, output, and
output growth (where all variables are measured in deviations from
their steady-state values):
𝑅𝑛
𝑡
𝑅𝑛=𝑅𝑛
𝑡−1
𝑅𝑛𝜌𝜋𝑡
𝜋𝑟𝜋𝑌𝑡
𝑌𝑟𝑦1−𝜌𝑌𝑡
𝑌𝑡−1 𝑟𝛥𝑦
𝑒𝑥𝑝(𝜖𝑟
𝑡).(31)
Appendix B. Supplementary data
Supplementary material related to this article can be found online
at https://doi.org/10.1016/j.econmod.2023.106283.
References
Afrin, S., 2017. The role of financial shocks in business cycles with a liability side
financial friction. Econ. Model. 64, 249–269.
Aguilar, P., Vázquez, J., 2021. An estimated DSGE model with learning based on term
structure information. Macroecon. Dyn 25 (7), 1635–1665.
Aguirre, I., Vázquez, J., 2020. Learning, parameter variability, and swings in US
macroeconomic dynamics. J. Macroecon 66, 103240.
Ajello, A., 2016. Financial intermediation, investment dynamics, and business cycle
fluctuations. Amer. Econ. Rev. 106 (8), 2256–2303.
Barsky, R.B., Sims, E.R., 2011. News shocks and business cycles. J. Monetary Econ. 58
(3), 273–289.
Beaudry, P., Lucke, B., 2010. Letting different views about business cycles compete.
NBER Macroecon. Ann 24 (1), 413–456.
Beaudry, P., Portier, F., 2004. An exploration into Pigou’s theory of cycles. J. Monetary
Econ. 51 (6), 1183–1216.
Beaudry, P., Portier, F., 2006. Stock prices, news, and economic fluctuations. Amer.
Econ. Rev. 96 (4), 1293–1307.
Ben Zeev, N., Khan, H., 2015. Investment-specific news shocks and US business cycles.
J. Money Credit Bank. 47 (7), 1443–1464.
Benes, J., Kumhof, M., 2015. Risky bank lending and countercyclical capital buffers. J.
Econom. Dynam. Control 58, 58–80.
Bernanke, B.S., Gertler, M., Gilchrist, S., 1999. The financial accelerator in
a quantitative business cycle framework. Handbook of Macroeconomics 1,
1341–1393.
Brooks, S.P., Gelman, A., 1998. General methods for monitoring convergence of
iterative simulations. J. Comput. Graph. Statist. 7 (4), 434–455.
Calvo, G.A., 1983. Staggered prices in a utility-maximizing framework. J. Monetary
Econ. 12 (3), 383–398.
Christiano, L.J., Motto, R., Rostagno, M., 2014. Risk shocks. Amer. Econ. Rev. 104 (1),
27–65.
Del Negro, M., Schorfheide, F., Smets, F., Wouters, R., 2007. On the fit of new
Keynesian models. J. Bus. Econom. Statist. 25 (2), 123–143.
Espinoza, R., Fornari, F., Lombardi, M.J., 2012. The role of financial variables in
predicting economic activity. J. Forecast. 31 (1), 15–46.
Fernández-Villaverde, J., 2010. The econometrics of DSGE models. SERIEs J. Spanish
Economic Assoc 1 (1-2), 3–49.
Fernández-Villaverde, J., Rubio-Ramírez, J.F., 2004. Comparing dynamic equilibrium
models to data: a Bayesian approach. J. Econometr 123 (1), 153–187.
Fujiwara, I., Hirose, Y., Shintani, M., 2011. Can news be a major source of aggregate
fluctuations? A Bayesian DSGE approach. J. Money Credit Bank. 43 (1), 1–29.
Economic Modelling 124 (2023) 106283
14
L. Herrera and J. Vázquez
Gelain, P., Ilbas, P., 2017. Monetary and macroprudential policies in an estimated
model with financial intermediation. J. Econom. Dynam. Control 78, 164–189.
Gertler, M., Karadi, P., 2011. A model of unconventional monetary policy. J. Monetary
Econ. 58 (1), 17–34.
Gertler, M., Kiyotaki, N., 2010. Financial intermediation and credit policy in business
cycle analysis. In: Friedman, B.M., Woodford, M. (Eds.), Handbook of Monetary
Economics. 3, Elsevier, pp. 547–599.
Gertler, M., Kiyotaki, N., Queralto, A., 2012. Financial crises, bank risk exposure and
government financial policy. J. Monetary Econ. 59, S17–S34.
Geweke, J., 1999. Using simulation methods for Bayesian econometric models:
Inference, development, and communication. Econometric Rev. 18 (1), 1–26.
Gilchrist, S., Yankov, V., Zakrajšek, E., 2009. Credit market shocks and economic
fluctuations: Evidence from corporate bond and stock markets. J. Monetary Econ.
56 (4), 471–493.
Gilchrist, S., Zakrajšek, E., 2012. Credit spreads and business cycle fluctuations. Amer.
Econ. Rev. 102 (4), 1692–1720.
Görtz, C., Tsoukalas, J.D., 2017. News and financial intermediation in aggregate
fluctuations. Rev. Econ. Stat. 99 (3), 514–530.
Görtz, C., Tsoukalas, J.D., Zanetti, F., 2022. News shocks under financial frictions. Am.
Econ. J.: Macroecon. 14, 210–243.
Gunn, C.M., Johri, A., 2018. Financial news, banks, and business cycles. Macroecon.
Dyn 22 (2), 173–198.
Justiniano, A., Primiceri, G.E., Tambalotti, A., 2010. Investment shocks and business
cycles. J. Monetary Econ. 57 (2), 132–145.
Kamber, G., Smith, C., Thoenissen, C., 2015. Financial frictions and the role of
investment-specific technology shocks in the business cycle. Econ. Model. 51,
571–582.
Meh, C.A., Moran, K., 2010. The role of bank capital in the propagation of shocks. J.
Econom. Dynam. Control 34 (3), 555–576.
Merton, R.C., 1973. An intertemporal capital asset pricing model. Econometrica 41 (5),
867–887.
Moran, P., Queralto, A., 2018. Innovation, productivity, and monetary policy. J.
Monetary Econ. 93, 24–41.
Mouabbi, S., Sahuc, J.-G., 2019. Evaluating the macroeconomic effects of the ECB’s
unconventional monetary policies. J. Money Credit Bank. 51 (4), 831–858.
Pigou, A.C., 1927. Industrial Fluctuations. Macmillan, London.
Queralto, A., 2020. A model of slow recoveries from financial crises. J. Monetary Econ.
114, 1–25.
Sanjani, M.T., 2014. Financial Frictions and Sources of Business Cycle. (14–194),
International Monetary Fund.
Schmitt-Grohé, S., Uribe, M., 2012. What’s news in business cycles. Econometrica 80
(6), 2733–2764.
Sims, E., 2016. What’s news in news? A cautionary note on using a variance
decomposition to assess the quantitative importance of news shocks. J. Econom.
Dynam. Control 73, 41–60.
Smets, F., Wouters, R., 2007. Shocks and frictions in US business cycles: A Bayesian
DSGE approach. Am. Econ. Rev 97 (3), 586–606.
Solow, R.M., 1957. Technical change and the aggregate production function. Rev. Econ.
Stat. 39 (3), 312–320.
Vázquez, J., Aguilar, P., 2021. Adaptive learning with term structure information. Eur.
Econ. Rev. 134, 103689.
Villa, S., 2016. Financial frictions in the euro area and the United States: A Bayesian
assessment. Macroecon. Dyn 20 (5), 1313–1340.
Wu, J.C., Xia, F.D., 2016. Measuring the macroeconomic impact of monetary policy at
the zero lower bound. J. Money Credit Bank. 48 (2–3), 253–291.
Wu, J.C., Zhang, J., 2019. A shadow rate New Keynesian model. J. Econom. Dynam.
Control 107, 103728.
