The Credit Channel in Monetary Policy Transmission at the Zero Lower Bound. A FAVAR Approach Master Project Alexandru Barbu, Zymantas Budrys, Thomas Walsh Barcelona Graduate School of Economics Master in Economics June 7, 2014 Abstract This paper aims to provide a methodology for identifying the credit channel in US monetary policy transmission, consistent with periods at the zero lower bound. We follow Ciccarelli, Maddaloni and Peydro (2011) in identifying credit shocks through quarterly responses in the Federal Reserve’s Senior Loan Officer Survey, but augment their identification strategy in two key ways. First, we use the credit variables inside a Factor Augmented Vector Autoregression, to summarize the information contained in a set of 110 US macroeconomic and financial series. Second, we adopt the shadow rate developed by Wu & Xia (2013) as an alternative to the effective federal funds rate at the zero lower bound. We present our results through impulse response func- tions and carefully designed counterfactuals. We find that monetary policy shocks have considerably larger effects through the credit supply side than the credit demand side. Building counterfactual analyses, we find the macroeconomic effects arising from the supply side of the credit channel to be sizable. When focusing on the recent un- conventional policies, our counterfactuals show only very modest movements in credit variables, suggesting that the positive effects of unconventional monetary policy during the crisis may not have acted strongly through the credit channels. 1
35
Embed
The Credit Channel in Monetary Policy Transmission at the Zero Lower Bound. A FAVAR Approach
Barcelona GSE Master Project by Alexandru Barbu, Zymantas Budrys, and Thomas Walsh
Master Program: Economics
About Barcelona GSE master programs: http://j.mp/MastersBarcelonaGSE
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The Credit Channel in Monetary Policy Transmission at the
Zero Lower Bound. A FAVAR Approach
Master Project
Alexandru Barbu, Zymantas Budrys, Thomas Walsh
Barcelona Graduate School of Economics
Master in Economics
June 7, 2014
Abstract
This paper aims to provide a methodology for identifying the credit channel in US
monetary policy transmission, consistent with periods at the zero lower bound. We
follow Ciccarelli, Maddaloni and Peydro (2011) in identifying credit shocks through
quarterly responses in the Federal Reserve’s Senior Loan Officer Survey, but augment
their identification strategy in two key ways. First, we use the credit variables inside
a Factor Augmented Vector Autoregression, to summarize the information contained
in a set of 110 US macroeconomic and financial series. Second, we adopt the shadow
rate developed by Wu & Xia (2013) as an alternative to the effective federal funds
rate at the zero lower bound. We present our results through impulse response func-
tions and carefully designed counterfactuals. We find that monetary policy shocks
have considerably larger effects through the credit supply side than the credit demand
side. Building counterfactual analyses, we find the macroeconomic effects arising from
the supply side of the credit channel to be sizable. When focusing on the recent un-
conventional policies, our counterfactuals show only very modest movements in credit
variables, suggesting that the positive effects of unconventional monetary policy during
the crisis may not have acted strongly through the credit channels.
1
1. Introduction
The recent global financial crisis has heightened the interest of both academia and
policy circles in the empirical relevance of the credit channel for the effective trans-
mission of monetary policy to the real economy. In turbulent times, however, such
an analysis faces additional identification challenges. The breadth and heterogeneity
of the Federal Reserves unconventional measures have undeniably expanded the set of
macroeconomic variables relevant to monetary policy assessment. 1
A second consideration is the relative lack of variation in the conventional mone-
tary policy measure, the effective Federal Funds rate, since it had reached the ZLB in
January 2009. Consequently, economists have sought a single measure that can par-
simoniously capture the stance of the monetary policy at the zero lower bound while
possibly quantifying the impact of unconventional monetary policies on the macroe-
conomy.2
These arguments guide our choice of methodology and emphasise the relative impor-
tance of the identification of credit channel in monetary policy transmission mechanism.
Accordingly, this paper seeks to provide a framework for identifying the credit
channel in US monetary policy transmission, consistent with periods at the zero lower
bound. Following Ciccarelli et al.(2013), we trust the bankers and identify credit shocks
through quarterly responses in the Federal Reserves Senior Loan Officer Survey. We
augment their identification strategy in two important ways. First, we employ the
credit variables inside a Factor Augmented Vector Autoregressive model, to summa-
rize the information contained in a set of 110 US macroeconomic and financial series.
Second, we adopt the shadow rate developed by Wu& Xia(2013) as an alternative to
the effective federal funds rate for the period after January 2009. The shadow interest
rate provides a measure claimed to summarize the stance of US monetary policy at the
zero lower bound.
Our paper makes contributions to three different strands of literature. To our
knowledge, this is the first application of the shadow rate to the identification of the
credit channel in monetary policy transmission at the zero lower bound. Second, the
literature examining the implications of the credit channel in a Factor Augmented VAR
models is rather thin. Jimborean et al.(2013) provide such an analysis for the French
1For a more comprehensive review of unconventional policies, see, for example, Thornton (2012).2Bullard (2013) provides a brief summary of the ongoing research on this topic.
2
economy, but identification of credit shocks is based on bank-level balance sheet ratios.
Third, this is the first methodology able to disentangle and quantify the effect of broad
lending channel and credit demand channel, as defined by responses from Senior Loan
Officer Survey, on such a large set of macroeconomic series, as facilitated by our FAVAR.
The paper is organized as follows: Section 2 reviews the credit channel literature
and discusses its main identification challenges. Section 3 presents our methodology
and proposes a proper identification of the credit shock, of the monetary shock at the
zero lower bound and of the wider macroeconomic model. Section 4 summarizes our
data. Section 5 presents the main results and interpretation, section 6 presents our
evaluation and section 7 concludes.
2. Literature review
A common puzzle in business cycle analysis is the observation of large and persis-
tent business cycle fluctuations stemming from relatively small and temporary real
or monetary shocks (King and Rebello, 1999). The traditional view in the monetary
policy transmission literature is that monetary authorities leverage their control over
short term interest bearing securities to affect the cost of capital and subsequently real
spending on durable and investment goods (Bernanke and Gilchrist, 1995) Following
the literature, we interpret this transmission mechanism as the interest rate channel.
Bernanke and Gertler(1995) claim the interest rate channel fails to explain empir-
ical evidence in 3 important aspects: timing (some real variables, such as business
fixed investments, react long after the interest rate has reverted to trend), magnitude
(large output fluctuations come at odds with the relatively small cost-of-capital effects
predicted in empirical studies) and composition (large impulse responses in long lived
assets to shocks in short term rates.
Their findings point to the existence of a credit enhancement channel - a mecha-
nism that amplifies and propagates monetary policy shocks to the real economy. At its
root lies the concept of external finance premia: a wedge between the cost of internal
finance (liquid assets, retained earnings) and external sources of finance (debt, equity).3
3Conventionally, the external finance premium is rationalized through the presence of frictions such ascredit market imperfections, asymmetric information, principal agent problems, costly monitoring or costlystate verification.
3
Another theoretical result is the financial accelerator hypothesis: the presence of en-
dogenous dynamics in the external finance premia across the business cycle (Bernanke,
Gertler and Gilchrist (1996)). For identical financing needs, the external finance pre-
mia is inversely correlated with the firm’s net worth (liquid assets) and collateral value
on illiquid assets. A small negative shock to firms’ net worth damages firms’ credit-
worthiness, lowers access to capital, dampens investment expenditures, lowering future
net worth, which in turn has negative consequences on present net worth, and so on
and so forth.
Testing for the empirical relevance of credit channel in business cycle dynamics
exposes us to severe identification challenges. These problems stem from the fact
that fluctuations in credit demand and supply are by and large unobserved variables.
According to Bernanke & Gertler (1995), credit aggregates are largely unable to dis-
entangle the effects stemming from the credit channel from those generally associated
with the interest rate channel. As shown in Bernanke, Gertler and Gilchrist (1996),
following a monetary policy tightening, both the interest rate channel, through the
policy rate, and credit channel, through the external finance premia, predict similar
dynamics in lending volumes.
Aggregate credit measures fail to account for the amount of existing credit lines,
whose demand tends to be countercyclical (Ciccarelli et. al., 2013). Ultimately, statis-
tics of aggregate credit prices fail to control for such strategic behaviors as flight to
quality. This reported tendency of banks to optimally rebalance their portfolios to-
wards their most creditworthy borrowers during phases of financial fragility artificially
reduces the sensitivity of credit price aggregates to changes in monetary policy. Al-
ternatively, micro data takes into account actual credit granted, instead of total loan
demand, being forced to make restrictive assumptions over the latter.
There is an growing literature of the relevance of bank lending shocks in driving
fluctuations in macroeconomic variables. Amiti and Weinstein (2013) find that bank
lending shocks can account for around 40% of the variation in investment expenditure
in Japan. Chodorow-Reich(2014) finds that the contraction of credit can explain up
to half of the employment decline from a sample of SMEs following the collapse of
Lehman Bros. Kashyap, Stein and Wilcox (1993) provide more evidence of a loan
supply channel of monetary policy to the real economy.Following monetary tightening,
the mix of external finance changes such that firms rely more on other external sources
such as commercial paper, and less on bank loans. They find bank loan supply directly
affects firms’ investments, suggesting that firms cannot perfectly substitute bank lend-
4
ing. Kashyap and Stein (2000) have investigated the impact of monetary policy on
lending behaviour of banks using data on one million loans. Moreover, Kashyap, La-
mont and Stein (1994) show that monetary policy has significant impacts on firms
inventories through liquidity constraints.
3. Methodology
Given the aforementioned empirical challenges, we adopt the approach of Ciccarelli
et al (2013) in identifying the credit channel through responses in the quarterly US
Senior Loan Officer Survey. A breakdown in broad lending channel and credit demand
channel is done following definitions from Bernanke et. al. (1995).
Senior Loan Officer Survey
Regional Feds request quarterly information on the lending standards that banks apply
to customers and on the loan demand they receive from firms and households. The
survey applies to a representative sample of 60-70 insured, domestically chartered com-
mercial banks.4 Due to data availability, we consider only commercial and industrial
(C&I) loans. Our series starts in 1991Q4. Respondents are asked to assess the change
in lending standards they apply to business loans and credit demand they receive from
business customers. Responses are weighted on a scale, from eased considerably to
tightened considerably. Only credit changes in the last 3 months are considered. Re-
garding the identification of credit shocks, we follow Bernanke & Gertler (1995) and
denote an innovation to responses related to demand for loans as a shock to credit de-
mand and an innovation to total lending standards as a shock to credit supply (broad
credit channel). While the SLOS responses are qualitative, results are reported as
net percentages.5 Once again, we trust the bankers in the sense that we take their
responses to be true and accurate. A detailed description of the SLOS questions is
provided in the annex.6
4The number of Senior Loan Officer Survey respondents varies slightly across the series5For any given credit variable, net percentages are constructed as the difference between the number of
banks reporting that standards have eased somewhat or considerably and the number of banks reportingthat standards have tightened somewhat or considerably, divided by the number of banks in the sample.
6For a review of the relative performance of Senior Loan Officer Survey in identifying the credit channelin the US monetary policy transmission, see Lown&Morgan (2006)
5
The shadow rate and the zero lower bound
A common practice in the monetary policy transmission literature is to identify the
monetary policy shock as an unexpected standardized change in the overnight Federal
Funds rate.7 However, since December 2008, the Federal Funds rate has (effectively)
been at the zero lower bound. The ensuing lack of variation implies the Federal Funds
rate can convey little information about the changes in US monetary policy during
the ZLB period. Moreover, the structural break in the variation of Federal Funds
rate would cause significant identification challenges for a prolonged period, long af-
ter the policy would have exited the zero-lower bound. Moreover, as Williams (2014)
emphasizes, the frequency and duration of zero-lower bound events might be severely
understated.8
Consequently, the literature has sought to find a monetary policy measure consis-
tent with both normal and zero-lower bound periods. In a seminal paper, Black (1995)
defines the nominal interest rate as an option with a strike price at the ZLB and the
short term shadow rate as the value of its underlying asset. The nominal interest rate
will equal the shadow rate for any positive values rt ≥ 0, and zero otherwise. Wu &
Xia (2013) use this insight to model the shadow rate through a Gaussian Affine Term
Structure Model (GATSM). GATSM uses information from selected yields at different
maturities to construct the remainder of the yield curve.9. While GATSM are very close
approximators of the actual yield curves in normal times, they fail in zero-lower bound
periods, as they allow for the possibility of negative nominal rates, which is implausible.
To simulate the yield curve at the ZLB, Wu & Xia (2013) introduce a non-linearity
in their linear factor model. The short term nominal rate becomes a non-linear function
of the factors. Factors are extracted from the observed yields using principal compo-
nent analysis and regressed on the yields. Then the model parameters are estimated,
and the estimates are used to create a counterfactual shadow rate that is affine in the
factors. The shadow rate is the nominal rate that would prevail were there no physical
currency. (if the ZLB would not exist).
Wu & Xia (2013) further provide an approximation which allows for closed form
solutions in multiple factors models. Hence, it returns a model that is empirically
7A comprehensive review of monetary policy shock choices is provided by Christiano, Eichenbaum andEvans (1999).
8Williams (2014) argues that modelling the probability of ZLB occurring is chiefly based on historicaldata from a short enough period to indicate ZLB events would practically be non-existent
9Hamilton and Wu (2010) provide derivation and intuition behind GATSM
6
tractable, simulates the observed ZLB yield curve with an high level of precision and
returns a shadow rate that is robust to different specifications.10
Note, though, that the entire theoretical construct would be a risk in the event of
a very persistent zero-lower bound period.11 Since in the Wu & Xia (2013) model, the
shadow rate is a function of forward rates, and this forward rates summarize the ex-
pectations about the future short term rate, a long enough ZLB period could stabilize
investors expectations of future short rates to zero for long enough for the shape of the
yield curve to be impaired. However, Swanson and Williams (2013) provide evidence
that the sensitivity of longer term yields to news during the current ZLB period is not
significantly altered.
Evaluating the Shadow Rate as a Measure of Monetary Policy
Wu & Xia (2013) verify whether the shadow rate is a reliable representation of the US
monetary policy stance at the ZLB. To test for a structural break in the dynamics of
the monetary policy rate across pre and post crisis periods, they run a likelihood ratio
test. The restricted model requires that the autoregressive coefficients of the reduced
from model are not significantly different before and after the crisis. They cannot reject
the null hypothesis of no structural break for the shadow rate, but do reject the null
hypothesis for the effective federal funds rate. Given the shadow rate, by construction,
closely follows the effective federal funds rate in normal times (see fig. 1, annex), but
decouples and continues to exhibit reasonable variation at the zero lower bound, we
interpret Wu & Xia shadow rate as an alternative monetary policy measure consistent
with the ZLB.
Following the literature, we set the beginning of the ZLB period to Q1 2009. We
construct a continuous series of the monetary policy rate by appending the effective
Federal Funds rate before the ZLB period with the shadow rate estimates during the
ZLB period. We subsequently employ the shadow rate, as an alternative measure of
monetary policy rate at the ZLB, and the credit variables, as identified from the re-
sponses in the Senior Loan Officer Survey, in a Factor Augmented VAR model, as
10For a more extensive discussion on the effectiveness of the Wu & Xia (2013) shadow rate in summarizingmonetary policy stance at the zero lower bound, see Bullard (2012) and Hamilton (2013).
11In Black(1995) model, the prospect of non-positive longer term yields is excluded. This holds intheoretical cases with continuous time. In practice, though, with non-zero step intervals, longer rates canbe negative, since there is some probability, given the current level and volatility of the shadow rate process,the rate will remain negative for the length of the horizon.
7
detailed below.
A Factor Augmented Vector Autoregressive Model
Following Sims (1980) critique of incredible identifying restrictions in dynamic simulta-
neous equations models, structural vector-autoregressions have become a powerful tool
in monetary policy transmission analysis. As Bernanke, Boivin&Eliasz (2004) explain,
the VAR approach requires only a plausible identification of the monetary policy shock
and not necessarily of the remainder of the macroeconomic model. To the extent to
which the monetary authority sets policy based on variables that are excluded from
the model, the resulting impulse responses are likely to be biased. If Sims(1992) jus-
tification of the price puzzle as a response of the monetary authority to inflationary
pressures not captured in the VAR model was right, the reasoning can be generalized
for other omitted variables.
We follow Bernanke, Boivin & Eliasz (2005) in specifying a Factor Augmented
Vector Autoregressive model of the US economy. One benefit of a FAVAR is it can
summarize the information contained in a large set of observed macroeconomic vari-
ables Xt in a relatively compact vector of latent factors Ft. Moreover, it ameliorates
degrees-of-freedom problems, mimics the large information set monetary authorities
might actually use in setting policy and obtains impulse responses for a large set of
macroeconomic variables of interest.
Following Bernanke et. al.(2005), we specify the following factor augmented vector-
autoregressive model:[Ft
st
]=
[µF
µs
]+ Ψ1
[Ft−1
st−1
]+ ...+ Ψp
[Ft−p
st−p
]+
[eFt
est
](1)
where the latent factors Ft and the shadow rate st load on the macroeconomic series
Xt according to : [Xt
Yt
]= L
[Ft
Yt
]+ εt (2)
where Xt is a N × 1 vector of observed macroeconomic series, Ft is a K ′times1 vector
of latent factors which summarize the dynamics in Xt, with K << N , st is a vector of
observed shock variables, which includes the monetary policy rate but might include
8
also other macroeconomic or credit variables, and mt is a measurement error. Since
the factors Ft are unobserved variables, we cannot use OLS to estimate the dynamic
equation in (2). Following Bernanke et.al.(2004), we adopt a two step principal com-
ponent analysis.
The argument behind principal component analysis is that observable variables
tend to be correlated, therefore have certain degree of redundancy in estimation. It
should be, therefore, possible to extract a smaller sample of orthogonal factors which
capture most of the variance in the observed series. The factors would then be linear
combinations of the weighted series.
The principal component analysis is performed as follows: From the measurement
equation, we estimate the matrix of coefficients and extract eigenvalues and eigen-
vectors. We order eigenvalues from largest to smallest and extract the factors corre-
sponding to the largest eigenvalues. We then use these factors in the dynamic equation.
Since conventional factor selection and lag selection criteria are shown to be less
reliable for FAVAR specification, we are guided by the literature. Following Bernanke
et. al. (2005) and Wu et al. (2013), we set the number of factors to 3. However, results
are robust for different factor specifications. Following Boivin & Giannoni (2009), we
set our optimal number of lags for quarterly series to 1.
In the resulting reduced VAR model, we impose the following variable ordering: We
set the monetary policy variable last. Following Bernanke et.al(2004), we differentiate
between slow moving and fast moving variables. Slow moving variables are assumed
not to react contemporaneously to monetary policy shocks. We extract factors from
the slow moving variables and place them before the policy rate.F1,t
F2,t
F3,t
st
=
µF1
µF2
µF3
µs
+ Ψ1
F1,t−1
F2,t−1
F2,t−1
st−1
+
eF1,t
eF2,t
eF3,t
est
(3)
At first, we run our baseline specification, with the vector of contemporaneous vari-
ables Yt containing the factors and the monetary policy rate. Subsequently, we add,
along the monetary policy rate, our two credit variables. In ordering credit supply and
credit demand, we follow Ciccarreli et. al.(2010) argument that credit supply adjusts
quicker to monetary policy shocks. Accordingly, we order the credit supply variable
last, but before the monetary policy variable.
9
Results are presented through impulse response functions, historical decompositions
and counterfactual analyses. To obtain the impulse response functions, we convert our
VAR(1) process into a VMA(∞) and compute the response from a monetary policy
shock today ust to the s step ahead forecast of our macroeconomic variable of interest,
through the factor loadings.
Jmp,is =
∂Xit+s
∂ump= by.i
∂Y it+s
∂ump+ bst.i
∂st+s
∂ump(4)
For historical decomposition, we make use of the Wald Theorem (Hamilton, 1994) and
write our VMA process as a sum of p initial conditions and the series of shocks for all
subsequent periods, where p is the number of lags.
[21] Sims, Christopher A. (1992)”Interpreting the Macroeconomic Time Series Facts:
The Effects of Monetary Policy,” European Economic Review, Elsevier, vol. 36(5),
pages 975-1000, June.
[22] Soares, Rita (2011) ”Assessing monetary policy in the euro area: a factor-
augmented VAR approach,” Working Papers w201111, Banco de Portugal, Eco-
nomics and Research Department.
[23] Swanson, Eric T. and John Williams (2012) ”Measuring the Effect of the Zero
Lower Bound On Medium and Longer-Term Interest Rates”, Working Paper Series
2012-02, Federal Reserve Bank of San Francisco
[24] Williams, John, (2014) ”Monetary Policy at the Zero-Lower Bound: Putting The-
ory into Practice,” Hutchins Center on Fiscal & Monetary Policy, Brookings In-
stitution.
[25] Wu, Jing Cynthia, and Fan Dora Xia (2014)”Measuring the Macroeconomic Im-
pact of Monetary Policy at the Zero-Lower Bound”, NBER Working Papers 20117,
National Bureau of Economic Research, Inc.
[26] Yellen, Janet L. (2011) ”Unconventional monetary policy and central bank com-
munications : a speech at the University of Chicago Booth School of Business U.S.
Monetary Policy Forum, New York, New York, February 25, 2011,” Speech 604,
Board of Governors of the Federal Reserve System (U.S.)
Annex
The following tables list the mnemonics, short names and transformations for the 97
macroeconomic series used in the paper. The transformation codes are: 1 no trans-
formation; 2 first difference; 4 logarithm; 5 first difference of logarithm. All series
are from the Federal Reserve of St. Louis Economic Database (FRED). Slow-moving
variables are marked with 1.
Results are presented through impulse response functions, historical decompositions
and counterfactual analyses. All impulse response functions are standardized, and
result from a 25 basis points monetary policy tightening.
21
Table 1: Macroeconomic dataNo. Mnemonic Short name Transformation Slow
Real output and income
1 CBI Change in Private Inventories 1 12 GDPC96 Real Gross Domestic Product 5 13 FINSLC96 Real Final Sales of Domestic Product 5 14 CIVA Corporate Inventory Valuation Adjustment 1 15 CP Corporate Profits After Tax 5 16 CNCF Corporate Net Cash Flow 5 17 GDPCTPI Gross Domestic Product: Chain-type Price Index 5 18 FPI Fixed Private Investment 5 19 GSAVE Gross Saving 5 110 PRFI Private Residential Fixed Investment 5 111 INDPRO Industrial Production Index 5 112 NAPM ISM Manufacturing: PMI Composite Index 1 113 HCOMPBS Business Sector: Compensation Per Hour 5 114 HOABS Business Sector: Hours of All Persons 5 115 RCPHBS Business Sector: Real Compensation Per Hour 5 116 ULCBS Business Sector: Unit Labor Cost 5 117 COMPNFB Nonfarm Business Sector: Compensation Per Hour 5 118 HOANBS Nonfarm Business Sector: Hours of All Persons 5 119 COMPRNFB Nonfarm Business Sector: Real Compensation Per Hour 5 120 ULCNFB Nonfarm Business Sector: Unit Labor Cost 5 1
Employment and hours
21 UEMPLT5 Civilians Unemployed - Less Than 5 Weeks 5 122 UEMP5TO14 Civilian Unemployed for 5-14 Weeks 5 123 UEMP15OV Civilians Unemployed - 15 Weeks and Over 5 124 UEMP15T26 Civilians Unemployed for 15-26 Weeks 5 125 UEMP27OV Civilians Unemployed for 27 Weeks and Over 5 126 NDMANEMP All Employees: Nondurable Goods 5 127 MANEMP Employees on Nonfarm Payrolls: Manufacturing 5 128 SRVPRD All Employees: Service-Providing Industries 5 129 USTPU All Employees: Trade, Transportation & Utilities 5 130 USWTRADE All Employees: Wholesale Trade 5 131 USTRADE All Employees: Retail Trade 5 132 USFIRE All Employees: Financial Activities 5 133 USEHS All Employees: Education & Health Services 5 134 USPBS All Employees: Professional & Business Services 5 135 USINFO All Employees: Information Services 5 136 USSERV All Employees: Other Services 5 137 USPRIV All Employees: Total Private Industries 5 138 USGOVT All Employees: Government 5 139 USLAH All Employees: Leisure & Hospitality 5 140 AHECONS Average Hourly Earnings: Construction 5 141 AHEMAN Average Hourly Earnings: Manufacturing 5 142 AHETPI Average Hourly Earnings: Total Private Industries 5 143 AWOTMAN Average Weekly Hours: Overtime: Manufacturing 1 144 AWHMAN Average Weekly Hours: Manufacturing 1 1
22
Table 2: Macroeconomic dataNo. Mnemonic Short name Transformation Slow
Housing starts and sales
45 HOUST Housing Starts: New Privately Owned Housing Units Started 4 046 HOUSTNE Housing Starts in Northeast Census Region 4 047 HOUSTMW Housing Starts in Midwest Census Region 4 048 HOUSTS Housing Starts in South Census Region 4 049 HOUSTW Housing Starts in West Census Region 4 050 HOUST1F Privately Owned Housing Starts: 1-Unit Structures 4 051 PERMIT New Private Housing Units Authorized by Building Permit 4 0
Credit aggregates
52 NONREVSL Total Nonrevolving Credit Outstanding, SA, Billions of Dollars 5 053 USGSEC U.S. Government Securities at All Commercial Banks 5 054 OTHSEC Other Securities at All Commercial Banks 5 055 TOTALSL Total Consumer Credit Outstanding 5 056 BUSLOANS Commercial and Industrial Loans at All Commercial Banks 5 057 CONSUMER Consumer (Individual) Loans at All Commercial Banks 5 058 LOANS Total Loans and Leases at Commercial Banks 5 059 LOANINV Total Loans and Investments at All Commercial Banks 5 060 INVEST Total Investments at All Commercial Banks 5 061 REALLN Real Estate Loans at All Commercial Banks 5 062 BOGAMBSL Board of Governors Monetary Base 5 063 TRARR Board of Governors Total Reserves 5 064 NFORBRES Net Free or Borrowed Reserves of Depository Institutions 1 0
Monetary aggregates
65 M1SL M1 Money Stock 5 066 CURRSL Currency Component of M1 5 067 CURRDD Currency Component of M1 Plus Demand Deposits 5 068 DEMDEPSL Demand Deposits at Commercial Banks 5 069 TCDSL Total Checkable Deposits 5 0
100 CPIAUCSL Consumer Price Index For All Urban: All Items 5 1101 CPIUFDSL Consumer Price Index for All Urban: Food 5 1102 CPIENGSL Consumer Price Index for All Urban: Energy 5 1103 CPILEGSL Consumer Price Index for All Urban: All Items Less Energy 5 1104 CPIULFSL Consumer Price Index for All Urban: All Items Less Food 5 1105 CPILFESL Consumer Price Index for All Urban: All Items Less Food & Energy 5 1
Commodity prices
106 OILPRICE Spot Oil Price: West Texas Intermediate 5 1
New orders, sales, inventories
107 NAPM Purchasing Managers Index (SA) 1 1108 NAPMNOI NAPM New Orders Index (%) 1 1109 NAPMSDI NAPM Vendor Deliveries Index (%) 1 1110 PMNV NAPM Inventories Index (%) 1 1
Table 4: Senior Loan Officer Questions
Channel SLOS Question
Broad credit channel Over the past three months, how have your banks credit standards changed for approvingapplications for C&I loans or credit lines other than those to be used to financemergers and acquisitions to large and middle-market firms changed?
Credit demand channel Apart from normal seasonal variation how has demand for C&I loans changed over the pastthree months?
24
Figure 1: Impulse responses to a 25 basis points monetary policy tightening for a selectionof macro series, pre-crisis, no survey variables included, 3 factors, 1 lag, 90% confidencebands
25
0 12 24−0.04
−0.02
0
0.02
0.04
0.06
0.08
0.1Turbo Rate
0 12 24−0.8
−0.6
−0.4
−0.2
0
0.2Industrial Production
0 12 24−0.5
0
0.5CPI
0 12 24−0.5
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2GDP Deflator
0 12 24−0.5
0
0.5
1
1.5
2Unemployment
0 12 24−0.06
−0.04
−0.02
0
0.02
0.04
0.06Housing Starts
0 12 24−0.1
−0.08
−0.06
−0.04
−0.02
0
0.02Change in Private Inventories
0 12 24−0.8
−0.6
−0.4
−0.2
0Fixed Private Investment
Figure 2: Impulse responses to a 25 basis points monetary policy tightening for a selectionof macro series for the entire sample, no survey variables included, 3 factors, 1 lag, 90%confidence bands
26
Figure 3: Quarterly changes in credit demand (blue) and net tightening (red),as reported bythe Senior Loan Officer Survey, for the period 1991Q4-2013Q4. Figures in net percentages
Figure 4: Impulse responses to a 25 basis points monetary policy tightening for creditdemand and credit supply variables, pre-crisis, 3 factors, 1 lag, 90% confidence bands
27
Figure 5: Impulse responses to a 25 basis points monetary policy tightening for a selectionof macro series, pre-crisis, including survey variables, 3 factors, 1 lag, 90% confidence bands
Figure 6: Impulse responses to a 25 basis points monetary policy tightening for creditdemand and credit supply variables, including crisis, 3 factors, 1 lag, 90% confidence bands
28
Figure 7: Impulse responses to a 25 basis points monetary policy tightening for a selectionof macro series, including crisis, including survey variables, 3 factors, 1 lag, 90% confidencebands
Figure 8: Impulse responses to a 25 basis points monetary policy tightening for creditdemand and credit supply variables,shutting down demand channel pre-crisis, 3 factors, 1lag
29
Figure 9: Impulse responses to a 25 basis points monetary policy tightening for selectedmacro variables,shutting down demand channel pre-crisis, 3 factors, 1 lag
Figure 10: Impulse responses to a 25 basis points monetary policy tightening for selectedmacro variables,shutting down demand channel, including crisis, 3 factors, 1 lag
30
Figure 11: Impulse responses to a 25 basis points monetary policy tightening for creditdemand and credit supply variables,shutting down supply channel, pre-crisis, 3 factors, 1lag
Figure 12: Impulse responses to a 25 basis points monetary policy tightening for selectedmacro variables,shutting down supply channel pre-crisis, 3 factors, 1 lag
31
Figure 13: Impulse responses to a 25 basis points monetary policy tightening for selectedmacro variables,shutting down supply channel, including crisis, 3 factors, 1 lag
32
Shadow Rate
Demand
Supply
Figure 14: Counterfactual series using Historical decomposition: Credit demand and creditsupply - actual values (blue); counterfactuals (pink) by setting the shadow rate at the zerolower bound st = 0.25, Date 0 = start of ZLB period, Dec. 2008.
33
Figure 15: Counterfactual series using Historical decomposition: Credit supply - actualvalues (blue); counterfactuals (pink) by setting the shadow rate at the zero lower boundst = 0.25
34
−5
0
5Loadings on Factor 1
−5
0
5Loadings on Factor 2
−5
0
5Loadings on Factor 3
−1
0
1
Loadings on Demand for Loans variable
−1
0
1
Loadings on Broad Credit Supply variable
−1
0
1
Loadings on Turbo Rate
Real OutputEmployment and HoursHousing Starts and SalesInventories and OrdersExchange RatesInterest Rates and SpreadsMoney and Credit aggregatesPrice IndexesHourly Earnings