Eurozone Sovereign Yield Spreads and
Diverging Economic Fundamentals∗
Alessandro Beber†, Michael W. Brandt‡, Maurizio Luisi§
June 2014
Abstract
We analyze the role of macroeconomic fundamentals for the term structure ofsovereign bond yields. We take a structured economic news flow approach to obtaina continuously updated measure of fundamentals and focus on a sample with largevariation in economic conditions, such as the Eurozone area between 1999 and 2012.We find a strikingly important role for macro fundamentals along a number ofdimensions. First, differences in economic growth explain 81% of the yield spreadsbetween AAA and non-AAA rated bonds. Second, levels and differences in economicgrowth are highly correlated with yield-based principal components, suggestingeconomic interpretations for latent factors. Furthermore, levels and differences ineconomic growth roughly triple the predictive ability of five yield-based factors forbond returns, capturing a large portion of time-varying risk premia. Finally, thispredictability cannot be captured with the standard headline GDP growth.
Keywords: sovereign yield, real-time economic growth
JEL classification: G12
∗We thank Daryl Caldwell, Robert Darwin, Ana-Maria Tenekedjieva, and seminar participants at theEuropean Central Bank, the Norges Bank, the Hedge Fund Conference at Imperial College, and the SystemicRisk, Contagion and Jumps conference at Cass Business School, for their comments and suggestions.†Cass Business School, City University London, and CEPR‡Fuqua School of Business, Duke University, and NBER§Bloomberg LP
1 Introduction
“This has little to do with fundamentals. There are plenty of money managers who
fear that part of Europe’s core are almost as rotten as the periphery.” (Financial
Times, Jan 2, 2013)
What is the role of macroeconomic fundamentals for determining the term structure of
sovereign bond yields? If we are studying a given issuer, across different bond maturities, there
is very little room for macroeconomic factors to matter (less the 1% of R2) once two or more
yield-based factors are used (e.g., Litterman and Scheinkman, 1991).1 If we are looking across
different sovereign issuers for a given maturity, macroeconomic factors seem to matter more, but
there is still a very strong yield-based factor structure (Longstaff et.al, 2011). If instead we are
studying yields through time, there is clearly a potential role for macroeconomic fundamentals
in determining the term structure of sovereign yields. In this case, Duffee (2011) finds that
more than two thirds of the variation in risk premia is not explained by yield-based factors.
However, there seems to be only weak correlation of this unspanned part of the risk premia
with macroeconomic variables.
The missing link between Treasury yields and macro fundamentals could be elusive for a
number of reasons. For example, fundamentals could be measured just too imperfectly, either
because they are observed at low frequency (e.g., the quarterly GDP releases), or because
they contain delayed and potentially restated information, quite unlike the continuous flow of
economic data that the bond market processes. Alternatively, past studies could have focused
on samples where fundamentals were fairly stable across time and/or issuers, thus hindering
identification.
Our paper addresses explicitly these two challenges. First, we take a structured economic
news flow approach. Specifically, we estimate a multifactor model for news about inflation,
output, and employment. Our model incorporates the entire cross-section of publicly released
economic data and produces a real-time macroeconomic index with continuous (at least daily)
updating. The real-time nature of our macroeconomic factors is especially important for our
empirical analysis, given that the large biases induced by reporting lags or data revisions seem
to largely affect bond predictability results (Ghysels, Horan, and Moench, 2012). Second, we
select a sample period and a sample of sovereign issuers characterized by substantial variation
in economic fundamentals, thus offering the best empirical conditions for identification. More
specifically, we focus on AAA versus non-AAA rated Euro area yields over the period 1999-2012,
which spans two recessions and the recent Eurozone sovereign debt crisis.
1Yield-based arbitrage-free term structure models deliver very similar low-dimensional factor representations(Dai and Singleton, 2000; Duffee, 2002). In their canonical forms, these models imply that macroeconomicshocks are perfectly spanned by the yield curve, that is, macroeconomic aggregates do not contain incrementalinformation beyond the cross-section of yields (Duffee, 2011; Joslin, Priebsch, and Singleton, 2014).
1
We focus on sovereign bond yields that, unlike corporate bond yields, have a tighter link
with macroeconomic conditions, as they affect directly the ability of a sovereign borrower to
repay. Studying yields and fundamentals of the Eurozone allows us to compare homogeneous
sovereign yields and sidestep the confounding effects of differential monetary policy or different
exchange rate dynamics that would typically arise in a cross-section of sovereign bonds.
We carry out our analysis on European sovereign spread dynamics at the broad regional level.
More specifically, we look at the aggregate sovereign spreads for all the Eurozone government
issuers with a credit rating other than AAA versus sovereign yields of AAA Eurozone governments.2
Measuring the yield spread based on credit ratings is a more appropriate approach than relying
on single indicators (e.g., the debt to GDP ratio), because in principle ratings incorporate
information on a whole range of factors that can affect the ability to repay, revolving around
two dimensions, the growth outlook and the country risk (e.g., Standard & Poor’s, 2011).
Consistent with the measurement of sovereign bond yield spreads, we construct real-time daily
macroeconomic factors conditional on country ratings. Our methodology allows us to obtain a
measure of macroeconomic growth divergence between AAA and non-AAA countries.
We find a strikingly important role for real-time macroeconomic fundamentals in explaining
sovereign yield spreads in the Eurozone, in both contemporaneous and predictive regressions.
More specifically, differences in economic growth between AAA and non-AAA countries explain
about 80% of the contemporaneous yield spreads between AAA and non-AAA rated sovereign
bonds at different maturities. Levels and differences in economic growth are highly correlated
with yield-based principal components, suggesting sensible economic interpretations for these
yield factors. Finally, real-time macroeconomic factors also explain up to 20% of the (albeit
very small) residuals from a yield-based factor model. When we turn to excess bond return
predictive regressions, we find that levels and differences in real-time economic growth roughly
triple the predictive ability of five yield-based factors for bond returns, capturing a large portion
of time-varying risk premia. Macroeconomic factors can also help, though not by as much, to
forecasts AAA versus non-AAA excess returns. A large part of the striking results on the
predictability of bond time-varying risk premia depends on our innovative measurement of
macroeconomic factors, as the traditional releases of headline GDP growth cannot capture this
degree of predictability.
Our paper blends three literatures: that on measuring the state of the economy based
on economic news data, commonly referred to as “nowcasting” (see Banbura et al., 2012,
for a survey), that on macro-finance fixed income asset pricing, and that on sovereign yield
spread determinants. There are two general approaches to nowcasting. The first approach is
to use a balanced panel regression, along the lines of the seminal paper of Stock and Watson
(1989), now the Chicago Federal National Activity Index (CFNAI). This first approach uses
2One advantage of using aggregate regional data, in contrast to country-level data, is that noise fromidiosyncratic country-level shocks is eliminated (Duffee, 1999; Driessen, 2005), thereby allowing more efficientestimation of the role of macroeconomic variables in the term structure. One disadvantage is that we are unableto assess the relative importance of country-level versus aggregate shocks in the pricing of individual bonds.
2
a large set of news releases but results in a relatively low measurement frequency because the
econometrician has to wait for the panel to be complete before the index can be constructed.
The second approach to nowcasting is to model macroeconomic data using a latent state-space
model (e.g., the ADS business conditions index of Arouba et al., 2009). The advantage of this
second approach is to produce an indicator at a higher frequency, since a state-space model
can effectively handle the sparse and delayed reporting of economic data, but this technique
is impractical for large cross-sections of news releases.3 Our approach to nowcasting uses a
large cross-section of news data but, by forward-filling missing data and making appropriate
correlation matrix adjustments, still produces high frequency indicators. We show in Beber et
al. (2014) that the resulting real-time indices are highly correlated with the CFNAI and ADS
business condition index, but that they appear to be more timely and informative about future
economic fundamentals.
We also add to the classical macro-finance models of the yield curve.4 Without imposing a
specific term-structure model and exploiting the large variation of fundamentals in our sample,
we demonstrate that macroeconomic variables are important determinants of risk-free yields and
credit spreads in both cross-sectional and time-series contexts. Other papers also use a large
panel of macroeconomic and financial variables (e.g., Ludvigson and Ng, 2009), with an emphasis
on return predictor selection and combination. Our focus, in contrast, is on documenting the
link between economic fundamentals and bond yields and returns. We do so by taking a two-
step approach. We first extract real-time indices that best capture the dynamics of economic
fundamentals conditional on rating classes and only then examine how these factors are related
to current yields and future bond returns. We thus explicitly exclude financial variables from
the factors. Furthermore, unlike the vast majority of predictability and nowcasting papers,
we exclusively work with precisely date- and time-stamped initial data releases, as opposed
to restated macroeconomic data. Ghysels et al. (2012) demonstrate the importance of using
unrestated data in the case of bond return predictability.
Finally, our paper contributes to the literature analyzing the determinants of sovereign
yields by highlighting the role of macro-based factors in addition to yield-based factors and
other financial variables (e.g., Longstaff et al. 2011). Specifically for Eurozone yields, a recent
literature explains sovereign yield spreads mainly with proxies for time-varying risk-aversion
(e.g., Favero, Pagano and vonThadden, 2010). We show that divergences in local macroeconomic
fundamentals, when they are measured frequently and in real-time, go a long way in explaining
sovereign yield spreads, contradicting the conventional wisdom synthesized in the opening quote
of the paper: it is precisely the differential economic growth of core and periphery Eurozone
countries that determines a large part of yield differentials.
The remainder of the paper proceeds as follows. Section 2 describes the Eurozone yield
3For example, Evans (2005) only considers the set of different (preliminary, advance, and final) GDP releases.Arouba et al. (2009) construct their business condition index using four indicators at different frequencies,including a continuously observable financial markets variable.
4This literature includes Ang and Piazzesi, 2003; Bikbov and Chernov, 2010; Duffee, 2011; Joslin, Priebsch,and Singleton, 2014, among others.
3
data and the macroeconomic announcements used to measure fundamentals. Section 3 presents
our methodology to construct real-time macroeconomic factors. In Section 4, we carry out the
empirical analysis showing the strikingly important role of real-time macroeconomic fundamentals
for bond yields and bond risk premia. Section 5 concludes with a summary of our findings.
2 Data
In this Section, we first describe the data that we use to construct our risk-free rate benchmark
for the Eurozone. We then explain how sovereign yield spreads are obtained as a difference
between sovereign bond yields of all Eurozone governments and the risk-free rates. Finally,
we describe the announcement data used to construct the real-time Eurozone macroeconomic
indices conditional on rating categories.
2.1 Risk-free Benchmark Sovereign Yields
The starting point for modeling the term structure of Eurozone interest rates is to obtain risk-
free benchmark rates. Traditionally, these rates are obtained from central government bond
issues with negligible credit risk. In the euro area, some central governments currently have
AAA issuer rating or had this rating at times during our sample period. Table 1 contains a
breakdown of Eurozone countries and their rating during our sample period, with an indication
of the date when some of these countries were downgraded from AAA status.
The European Central Bank (ECB) releases a reference yield curve based on AAA-rated
government bonds, obtained with a sound methodology that takes proper care of liquidity
differences and interpolation for constant maturities. In the remainder of this paper, we consider
this yield curve as our risk-free benchmark and thus work with the assumption that term
structures based on AAA-rated instruments are free of credit risk, therefore providing the
reference rate for the borrowing costs of the Eurozone economy. Using a government bond curve
rather than a swap curve as the risk-free benchmark has several advantages. First, the swap
curve would incorporate counterparty risk, as it would need to be constructed from instruments
that are more vulnerable to default of the involved counterparts. Second, the liquidity of interest
rate swaps is largely affected by systemic risk premia reflecting balance-sheet exposure of market
makers during periods of heightened financial uncertainty. Finally, in addition to credit risk
considerations, government bonds can be used as collateral, unlike swap contracts.
ECB AAA yield curves are available daily, but they only start in late 2004. We extend the
sample backward to the beginning of 1999, the start of the euro currency, using the euro risk-free
yield curve Fair Market Curve Index (FMCI) from Bloomberg, which provides the composite
yield of outstanding securities around each maturity point. The ECB and FMCI series turn out
to be almost indistinguishable in the overlapping part of the sample.
Figure 1, upper panel, shows a plot of Eurozone risk-free benchmark yields in our sample
period. The top and central panel of Table 2 show summary statistics on risk-free yields at seven
4
maturity points, from 3-months to 10 years, for the ECB and the full sample, respectively.
2.2 Eurozone Sovereign Yield Spreads
Besides the AAA yield curve, the ECB releases an additional yield curve based on central
government bonds from all euro area countries, regardless of their credit rating. Yields are
obtained with the same robust methodology that tackles distortions due to differential liquidity
and issues related to interpolation. We extend the sample backward before 2004 up to 1999
using Bloomberg yield data, as we did before for risk-free rates. Also in this case, the ECB and
Bloomberg ALL Eurozone yields are very similar for the overlapping part of the sample.
We compute Eurozone sovereign yield spreads as the differences between the ALL and the
AAA Bloomberg (1999-2004) and ECB (2004-2012) yield curves. Figure 1, lower panel, shows
the yield spread between ALL and AAA countries. We notice the tremendous evolution of
yield spreads in the last one third of the sample and the resulting challenge for term structure
modeling techniques that are traditionally facing more stable dynamics. In the bottom panel
of Table 2, we also show summary statistics for Eurozone sovereign yield spreads reported by
ECB at seven maturity points, ranging from 3-months to 10 years.
2.3 Macroeconomic Announcements
We obtain data on the dates, release times, and actual released figures for 183 Eurozone macro
releases covering the period from start of the euro single currency in January 1999 through
December 2012, for a total of more than 30,000 announcements over about 3,000 working days.
This data is obtained from Bloomberg through the Economic Calendar screen, which provides
precisely time-stamped and unrestated announcement data.5 The Appendix describes in detail
the set of macroeconomic news in our sample, including their frequency, source, and units of
measurement.
Most macroeconomic indicators are released on different days and at different frequencies,
making it difficult to process the flow of information in a systematic and consistent way.
Furthermore, news on different indicators are frequently released simultaneously. Finally, the
release frequency varies across different economic aggregates. Data releases of different economic
indicators are usually observed at different frequency; e.g., GDP data are typically sampled
quarterly, the unemployment data is instead released monthly. These features of our large
cross-section of macroeconomic releases generate a sparse matrix of data that the methodology
will have to take up.
2.4 Categorizing the macroeconomic news flow
Our aim is to extract a set of real-time macroeconomic factors describing the state of the
Eurozone economy, in both its (time-varying) AAA constituent and ALL constituent.
5The importance of using real-time versus final data in macroeconomic forecasting has been discussedextensively in the literature (e.g., Koenig et al., 2003, or Ghysels et al., 2012).
5
Rather than relying on a statistical procedure to obtain orthogonalized factors that are
increasingly difficult to interpret with the order of the factor, we impose a specific economically
motivated structure on the macroeconomic news flow. Based on both empirical evidence and
economic rationale, we first separate the aggregate economy into two broad dimensions: the
nominal and the real side.6 In practice, we split the set of announcements into nominal inflation-
related announcements and news that relates to real growth. Growth data, in turn, come in two
flavors - objective realizations of past economic activity and subjective often forward-looking
views derived from surveys which we label “macroeconomic sentiment.” Finally, economic
activity can be split one last time into information relating to output versus employment.
Through this structure, we can potentially obtain two (inflation and growth), three (inflation,
economic activity, and macroeconomic sentiment), or four (inflation, output, employment, and
macroeconomic sentiment) factors:
• Inflation
• Growth
Economic Activity
Output
Employment
Macro Sentiment
where, for example, the Economic Activity factor is obtained from the combined information
relating to Output and Employment. In that sense, the information is nested from right to
left. An important innovation of our approach is to obtain these macroeconomic factors for
different subset of countries, AAA and ALL, with this subsets potentially time-varying during
our sample period.
We use information from different subsets of news to construct the different macroeconomic
factors. For example, we extract the Eurozone AAA macroeconomic sentiment factor from the
news flow generated by 18 surveys: Finland Industrial Confidence Indicator, Finland Sentiment
Indicator, France Manufacturing PMI, France Services PMI, France Business Confidence Manu-
facturing Sentiment Index, Bank of France Business Sentiment Indicator, France Business
Confidence General Production Expectations, France Business Confidence Personal Production
Expectations, France Business Confidence Manufacturing Industry Demand Past 3 Month, Ifo
Pan Germany Business Climate, IFO Pan Germany Business Expectations, ZEW Germany
Assessment of Current Situation, ZEW Germany Expectation of Economic Growth, IFO Pan
Germany Current Assessment, Germany Manufacturing PMI, Germany Services PMI, Netherlands
Producer Confidence, and Netherlands Sentiment.7 For completeness, the Appendix lists the
assignments of all macroeconomic announcements for the Eurozone to the four categories:
inflation, output, employment, and macroeconomic sentiment.
It is worth reiterating at this point that we do not include any market-based data (such
6The economy is often separated into nominal and real sides because shocks to the two should be treateddifferently from a policy perspective. For example, many argue, from the perspective of monetary policy, thatnominal shocks should be minimized, whereas real shocks should not be intervened upon.
7These are the relevant surveys after March 2009, that is, after Ireland was downgraded from AAA rating,as indicated in Table 1.
6
as stock prices, interest rates, credit spreads, or implied voltilities) in our analysis, unlike, for
example, Arouba et al. (2009) and Giannone et al. (2008). While such data are very timely and
undoubtedly informative about the state of the economy, they represent already the market’s
interpretation of the macroeconomic news flow. Our aim is to objectively summarize and
describe the macroeconomic news flow itself, so that we can relate the actual state of the
economy to the term structure of sovereign bond yields and expected bond returns.
2.5 Transformation and temporal alignment
We examine the stationarity of each data series in two ways. First, we conduct a Dickey-Fuller
test on each series. Second, we read the definition and description of each statistic to determine
from an economic perspective whether it is a non-stationary index or a stationary quarterly
growth rate, for example. In a few cases where the conclusions from the two approaches differ,
usually because the available data is too short to examining statistically, we rely more on
the description to determine whether the series is stationary. All series that are deemed non-
stationary are first-differenced in news release time. The Appendix contains more details.
The final data management task is to align the data temporally by moving from announcement
time to calendar time. We do this by populating the news releases in a T ×N matrix where
T denotes the total number of week days in our sample and N refers, for example, to the 18
announcement types for the Eurozone AAA macroeconomic sentiment. The data at this stage
looks like the top panel of Figure 2.
There are two important aspects of the data to discuss. First, there are a vast number of
missing values, as we can think of each news series as a continuously evolving statistic that
is observed only once per month or quarter. Second, not all announcements have a complete
history. Some announcements are initiated in the middle of the sample and/or are terminated
before the end of the sample. To solve the missing data problem, we simply forward fill the last
observed release until the next announcement. Forward filling can be rationalized as replacing
missing values with expected values under a simple independent random walk assumption for
each news series. Of course, both independence in the cross-section and random walk dynamics
through time are simplifying assumptions that are rejected by the data (in fact, the motivation
for our methodology described below is the cross-sectional correlation structure within news
category). A more sophisticated approach for filling in missing data would be to compute the
expectation of the missing values given the full cross-section of previous releases as well as the
cross-sectional and intertemporal correlation structure of the data. An optimal solution would
also allow for sampling error, which is the case in Kalman filter or Bayesian data augmentation
algorithms. However, there is a clear trade-off between statistical complexity and ability to
process a large cross-section of news series. Since the goal of our approach is to utilize the
entire cross-section of news, we choose a very simple statistical model for filling in missing
observations. After forward filling, the data looks like the bottom plot of Figure 2.
Note that the second data issue, the fact that some series do not span the entire sample
7
period, cannot be solved with missing values imputation. It is instead explicitly addressed in
our methodology below.
3 Methodology
3.1 Subset principal component analysis
Our goal is to extract from the cross-section of macroeconomic news releases a set of factors that
capture in real-time the state of inflation, output, employment, and macro sentiment, as well
as the two more overarching factors measuring economic activity and growth. As we already
discussed, the most obvious ways of accomplishing this, full data principal components analysis
(PCA) and forecasting regressions, do not appeal to us. First, with full data PCA we obtain
factors that are mechanically orthogonal, whereas the dimensions of the economic news flow
we want to capture are likely correlated (e.g., output and employment are both high at the
peak and low at the trough of an economic cycle). This orthogonalization makes is practically
impossible to assign an economic meaning to higher order factors. Second, trying to identify
the factors through predictive regressions on a candidate variables in each category, such as
final GDP for output, would require us being able to identify a single series that represents each
category. While this is a common approach in the nowcasting literature, it relies on ex-ante
knowledge of the key statistic to track and assumes that there is only one such statistic that
does not change over time (see also Stock and Watson, 1989).
Instead, we rely on our ex-ante categorization of the news and, within each category subset,
let the data speak for itself by extracting the first principal component of that subset of data.
Specifically, on each day of our sample t, we obtain for each news category i the first principal
component from the correlation matrix Ωt,i of the stationary news series in category i. We
work with the correlation matrix to abstract from arbitrary scaling of data. Moreover, in order
to obtain a real-time measure, we use a telescoping (meaning, with a common historical start
date and rolling end dates) correlation matrix starting in 1990.8 We denote the Ni×1 principal
component weights by ct,i, where Ni is the number of news series in category i. Consistent with
extracting principal components from a telescoping correlation matrix, we standardize the news
series using telescoping estimates of their means and standard deviations.
3.2 Economic new series correlation matrix
The key inputs to our methodology are the within news category correlation matrices Ωt,i.
Specifically, we need to calculate from historical data up through date t the correlation of all
news series of category i that are “active” on that date, where active means that the news
series was previously initiated and has not yet been terminated. There are two issues that need
to be addressed in computing these correlation matrices. First, the data is in the form of an
unbalanced panel due to some of the series being initiated after the start date of the estimation
8We also experimented with fixed window size rolling correlation matrices for 5, 10, 15, and 20 years. Theresults are qualitatively similar, particularly for the longer data windows.
8
window (e.g., series j = 5 in Figure 2). Second, the data is naturally persistent, partly due to
autocorrelation of the data in announcement time, partly due to the cross-sectional misalignment
of the news in calendar time, and largely due to the forward filling of missing data.
We address the first unbalanced panel issue by using a correlation matrix estimator along
the lines of Stambaugh (1997), who shows how to adjust first and second moments estimates
for unequal sample lengths. The intuition of his approach is to use the observed data on the
longer series, along with a projection of the shorter series onto the longer ones estimated when
both are observed, to adjust the moments of the shorter time series.
To correct for the persistence, we could use the standard approach of Newey-West (1987),
where due to the nature of the data we would like to account for up to one quarter of autocorrelation
and cross-autocorrelation. Unfortunately, the kind of persistence in our data is not ideally
captured by the non-parametric Newey-West approach for two reasons. First, we have daily
data, so adjusting for up to a quarter of autocorrelation would involve approximately 60 cross-
autocorrelation matrices. Second, the (cross-) autocorrelations are not exponentially decaying
as a typical autoregressive model might predict. Instead, the data is locally constant, due to
the forward filling, and over longer intervals only moderately (cross-) autocorrelated due to the
statistical nature of the news series.
This peculiar correlation structure of economic news forward filled onto a daily calendar
is actually identical to that found in high-frequency asset prices, where asynchronous and
infrequent trading creates a misaligned and locally constant panel of observations. In that
literature, Ait-Sahalia, Mykland, and Zhang (2005) propose a “two-scales realized volatility”
estimator to handle this specific structure of short-term constancy versus long-horizon weak
dependence. Specifically, their estimator subsamples the data at a sufficiently low frequency
that overcomes the local constancy and then averages over the set of all possible estimators that
start the subsampling schemes at different times.
We adopt exactly the same approach, except of course our application is very different.
Specifically, at date t we subsample the forward filled news series backward at a monthly
frequency and then compute a Newey-West estimate of the correlation matrix using four lags.
We repeat the same for monthly sampling starting at dates t− 1, t− 2, ..., t− d+ 1 (assuming
d days per month) and then average the resulting d correlation matrix estimates.
4 Results
We first describe empirically the dynamics of the real-time macroeconomic factors, conditional
on AAA versus non − AAA country ratings. We also relate our real-time Eurozone growth
factors to GDP releases to get a sense for how our macroeconomic indices compare to traditional
low-frequency measures of economic growth. We then examine the contemporaneous relation
between the Eurozone growth factors and sovereign yields. Finally, we examine the relation
between macroeconomic factors and bond risk premia.
9
4.1 Preliminaries
In panel A of Table 3, we present correlations between the five real-time macroeconomic indices
constructed for Eurozone AAA countries. There are a number of interesting observations. First,
employment tends to be the least correlated with the other macroeconomic indices. Its highest
correlation is 0.49 with economic activity. In contrast, output, economic activity, and sentiment
are highly correlated with each other (correlations ranging from 0.82 to 0.98) and are each even
more highly correlated with the composite indices for growth. The correlations with the growth
index range from 0.93 to 0.97. We conclude from these high correlations that the growth
index contains most of the information revealed by output, economic activity, and sentiment
of Eurozone AAA countries, and we therefore focus on examining the aggregated growth index
going forward.9
In panel B of Table 3, we present correlations between the five real-time macroeconomic
indices constructed for Eurozone non − AAA countries. The pattern is similar to Panel A,
with the employment index that tends to be the least correlated with the other macroeconomic
series. In this case, output, economic activity, and sentiment are still relatively highly correlated
with each other, although not as much as for AAA countries. The correlations with the growth
index range from 0.62 to 0.88, including the employment index. Again, we conclude from these
high correlations that the growth index contains most of the information revealed by the other
macroeconomic indices of Eurozone non−AAA countries, and we therefore focus on examining
the aggregated growth index going forward.
Figure 3 compares standardized headline GDP growth, broken down in Eurozone AAA and
non − AAA countries, with our real-time growth indices for the same group of countries and
observed on the date of the GDP release. This is an important comparison to understand
whether the output of our data-driven principal component approach resembles the traditional
GDP measurement of economic growth. We find that it does. GDP growth of AAA countries
is very closely approximated by real-time AAA economic growth, although the latter seems
to be more resilient in the 2008-2009 crisis. We observe a similar pattern for the non − AAApair. The real-time growth index tracks GDP growth closely, but tends to fall less during
the economic crisis, although it seems much less resilient during the most recent part of the
sample. In summary, the real-time growth indices seem to be well approximating traditional
headline GDP growth, but offer the important advantage of higher frequency observation for
our empirical analysis.
We now turn to the daily frequency real-time indices for the different rated Eurozone
sovereigns in Figure 4. Specifically, the top panel shows the real-time growth index for AAA and
non−AAA countries. As can be readily seen, non−AAA countries exhibit stronger economic
growth in some earlier part of the sample, more resilience during the 2008-2009 recession, but
far lower growth in about the last 3-years of our sample. This graphical evidence foreshadows
9We do not explore the potential implications of the lower general correlation of the real-time employmentindex in this paper.
10
the empirical analysis in the next section of the paper and is well summarized in the bottom
panel of Figure 4, which shows the striking correlation between the spread in economic growth
and the spread in yields.
4.2 Sovereign Bond Yields and Economic Fundamentals
As a natural starting point for the empirical analysis of Eurozone rates and the state of the
economy, we first look at simple linear unrestricted relations between the yield curve and
the real-time macroeconomic factors. More specifically, in Table 4 we further elaborate our
analysis of sovereign yield spreads illustrated in Figure 4 and regress spreads on the real-
time growth index of AAA Eurozone countries and the growth differential between AAA
and non − AAA countries. We complement the real-time macroeconomic indices with other
important explanatory variables, such as the volatility index VIX and the U.S. corporate default
spread, that extant literature has shown to capture the effect of time-varying risk aversion
on sovereign yield spreads (e.g., Longstaff et al., 2011; Favero, 2013). Furthermore, we also
add two set of news dummies for the days in our sample featuring important events related
mostly to institutional policy decisions. The first set of news items is from Zoli (2013), who
identifies good and bad news days for international and country-specific events for Italy, the
largest sovereign bond issuers outside the AAA rating group. The second set of news items
refers to a set of programs introduced by the ECB and aimed at lowering government bond
yields on the presumption that such a reduction would have beneficial macroeconomic spillovers
(Krishnamurthy et al., 2013). Finally, we also include the iTraxx credit default swap index for
subordinated debt of European financial institutions, available after June 2004, to capture the
intertwined relation between aggregate credit risk of the European banking sector and the credit
risk of the European sovereigns (e.g., Acharya et al., 2014).
Table 4, Panel A, reports the results for the 5-year maturity. We can readily observe that the
real-time growth divergence between AAA and non− AAA Eurozone countries is a significant
explanatory variable for sovereign yield spreads (column 1) and, combined with the growth
index of AAA Eurozone, generates a large R2 of more than 80%, even when the lagged yield
spread is not included. The economic effect is also striking, with a one standard deviation in the
economic growth differential generating about a 50 basis point spread between the AAA and
non−AAA sovereign yield, on a 28 basis point yield spread full sample average.10 The real-time
growth differential is even more statistically significant when it is considered in isolation and
not combined with the growth index of AAA countries.11
We then start including the other explanatory variables. The logarithm of VIX is statistically
significant (columns 2 and 4), consistent with earlier literature, but it does not drive out
the explanatory power of the real-time growth differential, which is, if anything, even more
10The economic effect is obtained using the estimates in the model that does not contain the lagged yieldspread among the regressors.
11We do not report this univariate results to conserve space. In this case, the t-statistic increases to 2.54 andthe explanatory power excluding the lagged yield spread is 63%.
11
statistically significant. Therefore, VIX seems to help make the relation between spreads and
fundamentals more direct. The U.S. credit default spread instead is never significant (column
3 and 5), suggesting that European sovereign yield spread dynamics have been somewhat
unrelated to the general pattern of U.S. corporate credit risk. We also control for different
set of news. The good news dummies are all statistically significant, as well as the international
bad news dummies, all with the expected signs (column 6). Still, the divergence in growth plays
a significant additional role beyond these specific news days (column 7). The set of news days
capturing the ECB programs turns out to be extremely important for Eurozone yield spread
dynamics, especially the news about the Security Market Program, which are directly related
to the demand of non−AAA Eurozone sovereign bonds (column 8). However, the explanatory
power of the real-time growth divergence is unaltered (column 9). Finally, in the last two
columns of Table 4, panel A, we include the iTraxx credit default swap index for subordinated
debt of European financial institutions using a later sample start date of June 2004, when this
information becomes available. As expected, this variable by itself has a very large explanatory
power for sovereign yield spreads. When it is added as a control to all the other regressors in
the last column of the table, VIX is no longer significant, as well as some of the news category
dummy variables. In contrast, the spread in economic growth is still a significant determinant
of sovereign yield spreads, suggesting that the effect of diverging fundamentals has effects on
sovereign yield spreads that go beyond the effects on the aggregate credit risk of the European
banking sector.
Table 4, Panel B, repeats the same empirical exercise for the 10-year maturity. We obtain
very similar findings. Diverging economic growth between AAA and non − AAA Eurozone
countries is always a statistically significant determinant of sovereign yield spreads, even when
we control for VIX, different set of news items, or the aggregate credit risk of the European
banking sector.
4.3 Yield-based Factors and Macro-based Factors
The cross-sectional evidence on the term structure of sovereign bond yields suggests that yield-
based factors explain almost the entirety of the yield variation (e.g., Litterman and Scheinkman,
1991). However, it is important to try and provide some economic intepretation for these latent
factors.
In Table 5, we thus present simple unconditional correlations between the first three principal
components of the Eurozone sovereign yield curve and the real-time macroeconomic factors. We
present the results for the 2004 to 2012 sample period, when sovereign yields from the European
Central Bank are available, but the results are very similar when we extend the series back to
the beginning of 1999 with Bloomberg approximated yields. More specifically, Panel A shows
the results for AAA sovereign yields. The AAA yield principal components correspond to the
classical level, slope, and curvature, as shown elsewhere in the literaure.The real-time growth
of AAA Eurozone countries is positively correlated with yields, as economic intuition would
12
suggest. The growth divergence between AAA and non − AAA countries is instead highly
negatively correlated with the first principal component of risk-free yields, suggesting a role
for flight-to-quality dynamics for risk-free sovereign bonds in times of large economic growth
divergence. The correlation with the second principal component is negative for both Eurozone
AAA growth and growth divergence. With lower economic growth, the slope of the yield curve
of the AAA Eurozone yield curve tends to be flatter, consistent with monetary policy being
more accomodating in recessions. Similarly, a more divergent economic growth between AAA
and non − AAA Eurozone countries tends to imply a more dovish monetary policy. Finally,
in contrast to level and slope, the third principal component is generally less correlated with
real-time macroeconomic indices.
Table 5, Panel B, provides the results of a similar analysis for the principal components of
sovereign yields of all the Eurozone countries. The yield principal components identify again
the traditional level, slope, and curvature of the term structure. In this case, the correlation
coefficients with real-time growth of AAA countries are comparable to Panel A, in that average
sovereign yields and the slope of the term structure tend to have a positive and negative
relationship with economic growth, respectively. In contrast, the relation of sovereign yields
of all countries to the divergence in economic growth tends to be muted, suggesting a potential
contrasting effect between the effect on risk-free rates and sovereign yield credit spreads.
In Table 5, Panel C, we thus repeat the same analysis using the first three principal
components of sovereign yield spreads. We confirm the indications from the previous panels.
We find a very large and positive correlation between the average level of yield spreads and the
real-time divergence in economic growth between AAA and non − AAA countries, suggesting
that the growth divergence is likely to act both on reducing risk-free yields (as shown in Panel
A) and on increasing non−AAA sovereing yields.
Table 6 departs from the classical yield-based factor cross-sectional analysis of term structure
of interest rates and analyze the
Table 6 shows the explanatory power of Eurozone real-time macroeconomic factors on the
yield curve beyond the first three principal components. This empirical exercise is suggestive
of the potential explanatory power of macroeconomic observable factors beyond latent linear
factors in an affine specification. More specifically, we test whether real-time macroeconomic
factors have any explanatory power for yield residuals from a regression of yields on three
principal components. This is admittedly a very tough test, as the first three principal components
typically explain almost the entirety of yield variation.
In Panel A, we perform this exercise with AAA yield residuals. The real-time overarching
growth factor of Eurozone AAA countries is remarkably strong, with explanatory power between
2% and 12% at different maturities. Among the constituent real-time factors, employment and
sentiment display the weakest and strongest explanatory power, respectively. In Table 6, panel
B, we carry out the same exercise for ALL yield residuals. In this case, the explanatory power
of real-time Eurozone macro factors is generally less striking and has larger influence at the very
13
short-end of the yield curve. Finally, Table 6, Panel C, shows the explanatory power of real-
time macroeconomic differentials between AAA and ALL Eurozone countries for the residuals
of yield spreads on the first three principal components. The spread in economic growth plays
an important role, especially for short maturities. Among the constituent factors, economic
activity differentials are much more important than macroeconomic sentiment differentials.
In summary, this preliminary empirical analysis shows that macroeconomic real-time factors
play an important role for Eurozone sovereign yields, over and beyond the traditional yield
principal components, and beyond predictors related to time-varying risk aversion or risk of
the banking sector. The overarching Eurozone macroeconomic growth factor impacts risk-free
yields. The differential in Eurozone growth between AAA and ALL countries has very strong
explanatory power for Eurozone sovereign yield spreads.
4.4 Sovereign Bond Yield Risk Premia
We now shift the focus on sovereign bond risk premia and examine the relationship between
yield-based factors, macroeconomic factors, and subsequent bond returns at different horizons,
for different bond maturities, and for both risk-free yields and sovereign yield spreads. The yield-
based benchmark empirical model in this setting is Cochrane and Piazzesi (2005), who show
that a simple linear combination of forward rates is a bond return forecasting factor with very
large explanatory power. The information in this forecasting factor is not completely captured
by the classic three yield factors and is related to higher-order yield principal components. For
this reason, we use the first five bond yield principal components to mimic the bond return
forecasting factor and examine whether real-time macroeconomic factors offer any improvement
beyond that.12 This empirical setting sets a high bar for macroeconomic information to matter,
as it must be absent from the cross-section of yields and thus not picked up by the five principal
components.
Figure 5 shows the adjusted R2 of regressing AAA yield returns for maturities of 1-year,
5-year, and 10-year, on the five yield-based factors (in blue) and on both the yield-based and the
real-time macroeconomic factors (in red). Specifically, the two macroeconomic factors are the
real-time growth of the AAA Eurozone countries and the real-time growth differential between
AAA and non − AAA countries. The forecast horizons are set at a quarter, six months, and
one year.
The improvement in explanatory power of adding macroeconomic information is striking.
For example, the five yield-based factors explain about 20% of the variation in 5-year AAA yield
expected returns at the one-year forecasting horizon. The adjusted R2 increases to about 70%
when we add the two real-time macroeconomic factors. We obtain similar remarkable evidence
for other yield maturities and for different predictability horizons. Macroeconomic information
tends to be relatively more influential at longer forecasting horizons.
12Cochrane and Piazzesi (2005) show that a 5-factor Gaussian affine model generates their regression resultsexactly.
14
Figure 6 shows the results of performing the same exercise with a different dependent
variable, namely the return on the yield spread between non−AAA and AAA sovereign yields
for bond of different maturities and at different forecast horizons. Yield spread returns are
more difficult to predict. The adjusted R2 of using all factors ranges between 10% and 30%,
compared to the 20% to 70% range of AAA yields. The breakdown into the yield-only based
factors and the macroeconomic factors shows a prominent role of the former. However, real-
time macroeconomic growth and growth differentials still provide a significant improvement of
explanatory power beyond yield-based factors, and more so for longer forecasting horizons. For
example, the five AAA yield-based factors explain about 15% of the variation in 5-year yield
spread expected returns at the one-year forecasting horizon. The adjusted R2 increases well
beyond 20% when we add the two real-time macroeconomic factors.
The established role for macroeconomic factors in predicting bond risk premia could depend
on our innovative methodology that allows to measure fundamentals in real-time and with a
daily frequency, or could be simply a results of the large variation in economic conditions in
our specific sample (both the sample period and the cross-section of countries). To have a
better understanding of our results, we repeat a similar exercise using the traditional headline
quarterly GDP releases and the quarterly time frequency. Specifically, we obtain an index of
economic growth for AAA and non − AAA countries using average logarithmic GDP growth
rates, weighted by 1999 first quarter GDP. We use un-restated GDP figures and measure yields
on release dates, so to avoid any look-ahead bias in our findings. The five yield principal
components are also obtained using the quarterly frequency.
Figure 7 shows that the general explanatory power in AAA bond return predictability is
severely impaired, with adjusted R2 that are always below 20% and in most cases below 10%.
The reduced forecasting power is partially the outcome of the lower frequency of observations
and also the resulting little overlap in the forecasting intervals. In any case, the role of
macroeconomic factors beyond yield-based factors is mixed. Headline GDP growth of AAA
countries and growth differential with non − AAA countries helps predictability only very
marginally, if anything, at forecasting horizons of one quarter or half a year, across all bond
maturities. In contrast, macroeconomic factors tend to matter for the longer forecast horizon
of one year.
Figure 8 shows the results of the same empirical analysis for the returns on bond yield
spreads between AAA and non − AAA countries. Again, the explanatory power is generally
low, with positive adjusted R2 in only five out of the nine displayed horizons and maturities.
The role of macroeconomic factors beyond yield-based factors is relevant only in a handful of
cases, without a consistent pattern except for the total lack of predictability for 10-year bond
spread returns.
In summary, our empirical analysis in this Section has demonstrated the importance of
macroeconomic factors beyond yield-based factors for the predictability of both the AAA bond
expected returns and bond yield spread returns. A large part of our findings depends on our
15
methodology of measuring the real-time macroeconomic news flow at the daily frequency and
is more difficult to identify when using quarterly frequency headline European GDP.
5 Conclusion
What is the role of macroeconomic fundamentals for determining the term structure of sovereign
bond yields? We take a structured economic news flow approach to measure macroeconomic
conditions. Using a multi-factor model for news about inflation, output, employment, etc., we
incorporate the entire cross-section of publicly released economic data and construct macroeconomic
factors with daily updating. We focus on AAA versus non − AAA rated Euro area economic
growth and sovereign yields over the period 1999-2012, spanning two recessions and the Eurozone
sovereign debt crisis. We find a strikingly important role for macro fundamentals in explaining
Euro area sovereign yield spreads that would not be uncovered with headline GDP data.
16
A Appendix: Macroeconomic News
This Appendix summarizes the main features of the Eurozone macroeconomic releases considered
in our sample. Category is either employment (EMP), output (OUT), or macroeconomic
sentiment (SEN). If the sample series is stationary in our sample, we make no adjustment
(Adj.=0), otherwise we use first differences with respect to previous period (Adj.=1) or previous
year (Adj.=12). We also indicate Units, Frequency (M for monthly, W for weekly, Q for
quarterly), and the source of the release.
17
Category Name Macro Release Adj. Units Freq Source
EMP Belgium Unemployment Rate SA 1 Rate M National Bank of Belgium
EMP Estonia Unemployment Rate 1 Rate M Estonian Labour Market Board
EMP Estonia Average Gross Monthly Wages (Quarterly figures) YoY 0 Percent Q Statistical Office of Estonia
EMP Finland Unemployment Rate 1 Rate M Finnish Statistics Office
EMP France Non-Farm Non-Government Payrolls Total Quarterly Per-Chge 0 Rate Q INSEE National Statistics Offi
EMP France Monthly Wage Index QoQ 0 Rate Q French Labor Office
EMP France Unemployment Rate ILO Method - Mainland France 1 Rate Q INSEE National Statistics Offi
EMP France Unemployment Rate ILO Method Net Change (000s) 0 Volume Q INSEE National Statistics Offi
EMP France Unemployment Rate ILO Method - Mainland & Overseas Const. 1 Rate Q INSEE National Statistics Offi
EMP France Jobseekers Total SA net change 1 Volume M French Labor Office
EMP Germany Unemployment Change SA 1 Rate M Deutsche Bundesbank
EMP Greece Unemployment Rate Monthly 1 Rate M National Statistical Service
EMP Ireland Unemployment Rate SA 1 Rate M Central Statistics Office Irel
EMP Ireland Total Persons on Live Register SA 1 Volume M Central Statistics Office Irel
EMP Ireland Total Persons on Live Register SA MoM 0 Rate M Central Statistics Office Irel
EMP Italy New Hourly Wages MoM SA 2005=100 0 Rate M ISTAT
EMP Italy Unemployment Rate SA 1 Rate Q ISTAT
EMP Netherlands Unemployment Registered SA Per 1 Rate M Dutch Statistics Office
EMP Portugal Unemployment Rate NSA 1 Rate Q Instituto Nacional de Estatist
EMP Portuguese Labor Cost Index: Year over Year Percentage Change 0 Percent Q Instituto Nacional de Estatist
EMP Slovakia Unemployment Available to Work Rate 1 Rate M The Center for Labor, Family a
EMP Slovakia Avg Monthly Real Wages Industry YoY 0 Rate M Statistical Office of the Slov
EMP Spain Unemployment Level MoM Net Change Latest Rev 0 Volume M Spanish Labour Ministry
EMP Spain Labor Costs Avg Monthly Labor Cost Worker YoY 0 Rate Q INE
EMP Spain Unemployment Rate 1 Rate Q INE
EMP Slovenia Unemployment Rate Unemployed of Active Population 1 Rate M Rep Statistical Office
EMP Slovenia Avg Gross Real Wages YoY 0 Rate M Rep Statistical Office
EMP Eurostat Unemployment Eurozone SA 1 Rate M Copyright Euro Communities
EMP Eurostat Eurozone Employment SA WDA QoQ 0 Rate Q Copyright Euro Communities
EMP Eurostat Labor Costs Nominal Values Eurozone YoY WDA 0 Rate Q Copyright Euro Communities
EMP Eurostat Eurozone Employment NSA YoY 0 Rate Q Copyright Euro Communities
OUT Austria GDP Constant Prices QoQ 0 Rate Q Austrian Institute of Economic
OUT Austria Industrial Production MoM SA 0 Rate M Statistik Austria
OUT Belgium GDP Constant 2008 Prices SA QoQ 0 Rate Q National Bank of Belgium
OUT Estonia Chain Linked GDP Seas Working Day Adj QoQ 0 Rate Q Statistical Office Estonia
OUT Estonia Retail Sale Enterprises Constant YoY 0 Rate M Statistical Office Estonia
OUT Finland GDP Constant Prices SA QoQ 0 Rate Q Finnish Statistics Office
OUT Finland GDP Working Day Adjusted 0 Rate Q Finnish Statistics Office
OUT Finland Industrial Production Volume MoM SA 2005=100 0 Rate M Finnish Statistics Office
OUT Finland Retail Sales Volume Index YoY Per 0 Rate M Finnish Statistics Office
OUT France GDP QoQ 0 Rate Q INSEE National Statistics Offi
OUT France Industrial Production MoM SA 2005=100 0 Rate M INSEE National Statistics Offi
OUT France Manufacturing Production MoM SA 2005=100 0 Rate M INSEE National Statistics Offi
OUT Germany GDP Chain Linked Pan German QoQ 0 Rate Q German Fed Statistical Off
OUT Germany GDP Chain Linked Investment in Construction QoQ 0 Rate Q German Fed Statistical Off
OUT Germany GDP Chain Linked Exports QoQ 0 Percent Q German Fed Statistical Off
OUT Germany GDP Chain Linked Imports QoQ 0 Percent Q German Fed Statistical Off
OUT Germany GDP Chain Linked Private Consumption QoQ 0 Percent Q German Fed Statistical Off
OUT Germany GDP Chain Linked Government Consumption QoQ 0 Percent Q German Fed Statistical Off
OUT Germany GDP Chain Linked Domestic Demand QoQ 0 Rate Q German Fed Statistical Off
OUT Germany GDP Chain Linked Gross fixed capital investment QoQ 0 Rate Q German Fed Statistical Off
OUT Germany Industrial Production MoM SA 0 Rate M Deutsche Bundesbank
OUT Germany Manufacturing Orders MoM SA 0 Rate M Deutsche Bundesbank
OUT Germany Retail Sales Constant 2005 Prices MoM SA 0 Rate M German Fed Statistical Off
OUT Greece Real GDP QoQ SA 0 Rate Q National Statistical Service o
OUT Greece Industrial Production YoY 0 Rate M National Statistical Service o
OUT Greece Retail Sales YoY 2005=100 WDA 0 Rate M National Statistical Service o
OUT Ireland GDP Constant 2005 Prices QoQ SA 0 Rate Q Central Statistics Office Irel
OUT Ireland Industrial Production SA MoM 2000=100 0 Rate M Central Statistics Office Irel
OUT Ireland All New Vehicle Registrations 1 Volume M Central Statistics Office Irel
OUT Ireland Retail Sales Volume All Businesses MoM SA 0 Rate M Central Statistics Office Irel
OUT Italy Real GDP QoQ SA WDA 0 Rate Q ISTAT
OUT Italy Industrial Production MoM SA 0 Rate M ISTAT
OUT Italy Industrial Orders MoM SA 2005=100 0 Rate M ISTAT
OUT Italy Industrial Sales MoM SA 2005=100 0 Rate M ISTAT
OUT Italy New Car Registrations YoY NSA 0 Rate M ANFIA
OUT Italy Retail Sales MoM SA 2005=100 0 Rate M ISTAT
18
Eurozone Macroeconomic News (continued)Category Name Macro Release Adj. Units Freq Source
OUT GDP at Real 2000 Prices Seasonally Adjusted in Euros QoQ 0 Rate Q Dutch Statistics Office
OUT Netherlands Industrial Production MoM 2005=100 SA 0 Rate M Dutch Statistics Office
OUT Netherlands Industrial Sales YoY 2005=100 0 Rate M Dutch Statistics Office
OUT Netherlands Retail Sales Turnover Index 2000=100 YoY 0 Rate M Dutch Statistics Office
OUT Portugal GDP Constant 2006 Prices QoQ 0 Rate Q Instituto Nacional de Estatist
OUT Portugal Industrial Production Index MoM 0 Rate M Instituto Nacional de Estatist
OUT Portugal Industrial Sales Index 2005=100 MoM 0 Rate M Instituto Nacional de Estatist
OUT Portugal Retail Sales Index MoM 0 Rate M Instituto Nacional de Estatist
OUT Slovakia GDP Constant Prices YoY 0 Percent Q Statistical Office of the Slov
OUT Slovakia Industrial Production Index Adjusted 0 Percent M Statistical Office of the Slov
OUT Slovakia Industrial Sales Constant Prices YoY 0 Rate M Statistical Office of the Slov
OUT Slovakia Industrial Orders MoM 0 Rate M Statistical Office of the Slov
OUT Slovakia Retail Sales Ex Motor Vehicles Constant YoY 0 Percent M Statistical Office of the Slov
OUT Spain GDP SA Chained Linked at Constant 2008 Prices QoQ 0 Rate Q INE
OUT Spain Industrial Production YoY 2005=100 0 Rate M Instituto Nacional de Estadist
OUT Spain Industrial Production Workday Adjusted YoY 0 Rate M Instituto Nacional de Estadist
OUT Spain Retail Sales Constant Prices 2005=100 YoY 0 Rate M INE
OUT Spain Retail Sales Constant Prices WDA YoY 0 Rate M INE
OUT Slovenia GDP Constant Prices YoY 0 Rate Q Statistical Office of the Repu
OUT Slovenia Industrial Production MoM 0 Rate M Statistical Office of the Repu
OUT Slovenia Retail Trade MoM 0 Rate M Statistical Office of the Repu
OUT Eurostat GDP cons prices Euro QoQ 0 Rate Q Copyright Euro Communities
OUT Eurostat GDP cons prices Euro Household CExp 0 Rate Q Copyright Euro Communities
OUT Eurozone Government Expenditure cons prices 0 Rate Q Copyright Euro Communities
OUT Eurostat GDP cons prices Euro Gross Fixed Cap Form 0 Rate Q Copyright Euro Communities
OUT Eurostat Ind Production Euro Ex Constr MoM SA 0 Rate M Copyright Euro Communities
OUT Eurostat New Orders Euro Manufact Ind Orders MoM SA 0 Rate M Copyright Euro Communities
OUT Eurostat Euro Monthly Prod Construction SA MoM 0 Rate M Copyright Euro Communities
OUT Eurostat Retail Sales Volume Eurozone MoM SA 0 Rate M Copyright Euro Communities
SEN Belgium General Index Business Confidence 0 Value M National Bank of Belgium
SEN Belgium SEN Indicator 0 Value M National Bank of Belgium
SEN Finland Industrial Confidence Indicator 0 Value M Confederation of Finnish Indus
SEN Finland SEN Indicator 0 Value M Finnish Statistics Office
SEN France Manufacturing PMI Markit Survey Ticker 0 Value M Markit
SEN France Services PMI Markit Survey Ticker 0 Value M Markit
SEN France Business Confidence Manuf Sent Index 0 Value M INSEE National Statistics Offi
SEN Bank of France Business Sentiment Indicator 0 Value M Banque De France
SEN France Business Confidence General Prod Expect 0 Value M INSEE National Statistics Offi
SEN France Business Confidence Personal Prod Expect 0 Value M INSEE National Statistics Offi
SEN France Bus Conf Mfg Industry Demand Past 3 Month 0 Value M INSEE National Statistics Offi
SEN Ifo Pan Germany Business Climate 0 Value M IFO Institute - Institut fuer
SEN IFO Pan Germany Business Expectations 0 Value M IFO Institute - Institut fuer
SEN ZEW Germany Assessment of Current Situation 0 Value M ZEW Zentrum fuer Europaeische
SEN ZEW Germany Expectation of Economic Growth 0 Value M ZEW Zentrum fuer Europaeische
SEN IFO Pan Germany Current Assessment 0 Value M IFO Institute - Institut fuer
SEN Germany Manufacturing PMI Markit Survey Ticker 0 Value M Markit
SEN Germany Services PMI Markit Survey Ticker 0 Value M Markit
SEN GfK SEN 0 Value M GfK AG
19
Eurozone Macroeconomic News (continued)Category Name Macro Release Adj. Units Freq Source
SEN Ireland Consumer Sentiment Index 0 Value M IIB Bank
SEN Italy Business Confidence 0 Value M ISTAT
SEN Italy Services PMI Markit Survey Ticker 0 Value M Markit
SEN Italy Manufacturing PMI Markit Survey Ticker 0 Value M Markit
SEN Italy SEN Indicator SA 0 Value M ISTAT
SEN Netherlands Producer Confidence 0 Price M Dutch Statistics Office
SEN Netherlands SEN Seasonally Adjusted 0 Value M Dutch Statistics Office
SEN Portugal SEN Indicator 3Mth Moving Average 0 Value M Instituto Nacional de Estatist
SEN Portugal Economic Climate Indicator 0 Value M Instituto Nacional de Estatist
SEN Slovakia Industrial Confidence Indicator 0 Yield M Statistical Office of the Slov
SEN Slovakia SEN Indicator SA 0 Yield M Statistical Office of the Slov
SEN Spain Business Confidence Indicator 1 Value Q Spanish Chamber of Commerce
SEN Slovenia Sentiment Indicator SA 0 Value M Statistical Office of the Repu
SEN EC Manufacturing Confidence Euro Ind Confidence 0 Value M European Commission
SEN EC Composite PMI Output 0 Value M Markit
SEN Eurozone Manufacturing PMI Markit Survey Ticker 0 Value M Markit
SEN Eurozone Services PMI Markit Survey Ticker 0 Value M Markit
SEN EC Economic Sentiment Indicator Eurozone 0 Value M European Commission
SEN EC Euro Area Business Climate Indicator 0 Value M European Commission
SEN EC Services Confidence Indicator Eurozone 0 Value M European Commission
SEN ZEW Eurozone Expectation of Economic Growth 0 Value M ZEW Zentrum fuer Europaeische
SEN Sentix Economic Indices Euro Aggregate Index 0 Value M SENtix Behavioral Indices
SEN EC SEN Indicator Eurozone 0 Value M European Commission
20
References
Acharya, Viral, Itamar Drechsler and Philipp Schnabl, 2014, A Pyrrhic Victory? - Bank Bailoutsand Sovereign Credit Risk, Journal of Finance, forthcoming.
Ait-Sahalia, Yacine, Mykland, Per A., and Zhang, Lan, 2005, How often to sample a continuous-time process in the presence of market microstructure noise, Review of Financial Studies 18,351–416.
Ang, A. and M. Piazzesi, 2003, A no-arbitrage vector autoregression of term structure dynamicswith macroeconomic and latent variables, Journal of Monetary Economics 50, 745–787.
Beber, Alessandro, Michael W. Brandt, and Maurizio Luisi, 2014, Distilling the MacroeconomicNews Flow, Journal of Financial Economics, forthcoming.
Benzoni Luca, R. Goldstein, P. Collin-Dufresne, and J. Helwege, 2012, Modeling CreditContagion via the Updating of Fragile Beliefs, Working Paper, Federal Reserve Bank ofChicago.
Bernanke, B S, M Gertler and S Gilchrist, 1999, The financial accelerator in a quantitativebusiness cycle framework , in J B Taylor and M Woodford (eds), Handbook ofMacroeconomics, vol. 1, Amsterdam: North-Holland, 1341–93.
Cochrane, John, and Monika Piazzesi, 2005, Bond risk premia, American Economic Review 94,138–160.
Collin-Dufresne, P., R. Goldstein and J. Spencer Martin, 2001, The determinants of creditspread changes, Journal of Finance 56, 2177–2207.
Dai, Q and K Singleton, 2000, Specification analysis of affine term structure models, Journalof Finance 55, 1943–1978.
Di Cesare, Antonio, Giuseppe Grande, Michele Manna and Marco Taboga, 2012, Recentestimates of sovereign risk premia for euro-area countries, Working Paper, Bank of Italy.
Driessen, J., 2005, Is default event risk priced in corporate bonds?, Review of Financial Studies18, 165–195.
Duan, J and J Simonato, 1995, Estimating and testing exponentially affine term structuremodels by Kalman filter, Working Paper, McGill University.
Duffee, G., 1999, Estimating the price of default risk, Review of Financial Studies 12, 197–226.
Duffee, G., 2002, Term premia and interest rate forecasts in affine models, Journal of Finance57, 405–443.
Duffee, G., 2011, Information in (and not in) the term structure, Review of Financial Studies24, 2895–2934.
Duffie, D., and K. Singleton, 1999, Modeling term structures of defaultable bonds, Review ofFinancial Studies 12, 687–720.
Favero, C., 2013, Modelling and forecasting government bond spreads in the euro area: a GVARmodel, Journal of Econometrics, forthcoming.
21
Favero, C., M. Pagano, and Elu Von Thadden, 2010, How Does Liquidity Affect Bond Yields?Journal of Financial and Quantitative Analysis 45, 107–134.
Ghysels, Eric, Casidhe Horan, and Emanuel Moench, 2012, Forecasting through the Rear-ViewMirror: Data Revisions and Bond Return Predictability, Working Paper, University of North-Carolina.
Koenig, E., S. Dolmas, and J. Piger, 2003, The use and abuse of real-time data in economicforecasting, Review of Economics and Statistics 85, 618–628.
Krishnamurthy, Arvind, Stefan Nagel, and Annette Vissing-Jorgensen, 2013, ECB Policiesinvolving Government Bond Purchases: Impact and Channels, Working Paper, Universityof California at Berkeley.
Joslin, S., M. Priebsch, and K.J. Singleton, 2010, Risk premium accounting in macrodynamicterm structure models, Working paper, Stanford University.
Lando, D., 1998, Cox processes and credit-risky securities, Review of Derivatives Research 2,99–120.
Litterman, R and J Scheinkman, 1991), Common factors affecting bond returns, Journal ofFixed Income 1, 51–61.
Longstaff, F., J. Pan, L. Pedersen and K. Singleton, 2011, How Sovereign is Sovereign CreditRisk?, American Economic Journal: Macroeconomics 3, 75–103.
Ludvigson Sydney C., and Serena Ng, 2009, Macro Factors in Bond Risk Premia, Review ofFinancial Studies 22, 5027–5067.
Lund, J., 1997, Econometric analysis of continuous-time arbitrage-free models of the termstructure of interest rates, Working Paper, Aarhus School of Business.
Newey, Whitney K.; West, Kenneth D., 1987, A simple, positive semi-definite, heteroskedasticityand autocorrelation consistent covariance matrix, Econometrica 55, 703–708.
Stambaugh, Robert F., 1997, Analyzing investments whose histories differ in length, Journal ofFinancial Economics 45, 285–331.
Standard and Poor’s, 2011, Sovereign Government Rating Methodology And Assumptions,Global Credit Portal Working Paper.
Zoli, E., 2013, Italian Sovereign Spreads: Their Determinants and Pass-through to BankFunding Costs and Lending Conditions, Working Paper, International Monetary Fund.
22
Table 1: Rating Composition for Eurozone Member Countries
This table shows the breakdown of Eurozone countries in AAA or non-AAA Fitch rating categories, alongwith relevant change of rating dates.
AAA rating categoryEurozone Country Fitch Rating AAA Rating
End of sample From To
Austria AAA 1999 end of sampleFinland AAA 1999 end of sampleFrance AAA 1999 end of sampleGermany AAA 1999 end of sampleNetherlands AAA 1999 end of sampleLuxembourg AAA 1999 end of sampleIreland BBB+ 1999 06 March ’09Spain BBB 1999 10 Dec ’03
Non-AAA rating categoryEurozone Country Fitch Rating Non-AAA Rating
End of sample From To
Belgium AA 1999 end of sampleItaly A- 1999 end of samplePortugal BB+ 1999 end of sampleSpain BBB 10 Dec ’03 end of sampleIreland BBB+ 06 March ’09 end of sampleGreece CCC 2001 end of sampleSlovenia A- 2007 end of sampleCyprus BB+ 2008 end of sampleMalta A+ 2008 end of sampleSlovakia A+ 2009 end of sampleEstonia A+ 2001 end of sample
23
Table 2: Summary Statistics on Eurozone Risk-free Benchmark Yields andSovereign Spreads
This table reports summary statistics for Eurozone AAA risk-free benchmark yields and sovereign spreads.The yield spread is obtained as the difference between ALL Eurozone sovereign yield and AAA yield withthe same maturities. Data are at daily frequency. The sample period for the ECB curves is 2004:09-2012:10,while the Bloomberg FMCI curves cover the period 1998:12-2012:10.
Maturity
(Months) Mean SD Skew Kurt Min Max
ECB Benchmark Sovereign Yields - starting 2004/09/06
3 1.83 1.41 0.27 1.61 -0.02 4.33
6 1.89 1.44 0.24 1.56 -0.05 4.36
12 2.00 1.42 0.21 1.60 -0.10 4.54
36 2.45 1.16 -0.04 1.97 0.15 4.74
60 2.86 0.94 -0.30 2.39 0.65 4.73
84 3.19 0.77 -0.51 2.77 1.17 4.74
120 3.56 0.63 -0.69 3.06 1.76 4.78
Benchmark Sovereign Yields - Long Sample - starting 1998/12/09
3 2.40 1.43 -0.15 1.89 0.02 5.06
6 2.48 1.43 -0.13 1.90 0.01 5.21
12 2.58 1.41 -0.15 1.92 0.01 5.27
36 3.00 1.25 -0.32 2.27 0.12 5.35
60 3.38 1.07 -0.37 2.55 0.62 5.38
84 3.67 0.96 -0.36 2.71 1.11 5.55
120 3.95 0.84 -0.33 2.72 1.66 5.64
ECB Sovereign Spreads
3 0.14 0.28 3.22 17.60 -0.80 2.33
6 0.21 0.42 2.59 10.77 -0.62 2.83
12 0.32 0.55 2.20 7.60 -0.17 3.17
36 0.40 0.57 1.57 4.37 -0.03 2.46
60 0.41 0.56 1.41 3.68 -0.01 2.14
84 0.41 0.55 1.39 3.60 0.00 2.03
120 0.42 0.53 1.37 3.57 0.01 1.95
24
Table 3: Correlations between News Categories
We show unconditional linear correlation coefficients between different Eurozone real-time macroeconomic
factors for AAA countries (Panel A) and non−AAA countries (Panel B).
Panel A: AAA countries (1999-2012)
Sent Output Empl EconAct Growth
Sent 1.00 0.82 0.26 0.84 0.97
Output 0.82 1.00 0.40 0.98 0.93
Empl 0.26 0.40 1.00 0.49 0.33
EconAct 0.84 0.98 0.49 1.00 0.94
Growth 0.97 0.93 0.33 0.94 1.00
Panel B: non−AAA countries (1999-2012)
Sent Output Empl EconAct Growth
Sent 1.00 0.68 0.13 0.27 0.80
Output 0.68 1.00 0.53 0.69 0.88
Empl 0.13 0.53 1.00 0.97 0.62
EconAct 0.27 0.69 0.97 1.00 0.75
Growth 0.80 0.88 0.62 0.75 1.00
25
Table 4: Explaining Yield Spreads
We show estimates of the following regression:
yt − rt = α+ ρ (yt−1 − rt−1) + βXt + εt,
where here the yt − rt denotes the yield spread for the 5-year maturity (Panel A) and the 10-year maturity
(Panel B). Xt contains the set of explanatory variables that includes the real-time growth index of AAA
countries, the real-time growth spread between AAA and non − AAA countries, the log of the VIX index,
the U.S. credit spread, a set of bad/good international and Italian news from Zoli (2013), and the iTraxx
credit default swap index for investment grade European Financial entities. The sample period extends from
January 1999 (from June 2004 in columns 8 and 9) to December 2012. All of the regression are based on daily
observations. The constant and the lagged yield spread are always included but not reported to save space.
The AdjustedR2 shows explanatory power excluding the lagged spread. Robust Newey-West t-statistics are
reported in parentheses.
Panel A: 5-year maturityGrowth -0.09
(-1.12)
Growth Spread 0.29 0.36 0.38 0.18 0.15 0.48
(1.90) (4.07) (3.95) (2.70) (2.24) (2.87)
Log(VIX) 0.20 0.55 0.52 0.12
(1.91) (3.95) (3.28) (0.37)
Credit Spread 0.02 0.06 -0.13
(0.23) (0.62) (-0.55)
Bad News 3.35 3.31 2.76
(1.82) (1.80) (1.53)
Good News -6.14 -6.16 -4.96
(-3.37) (-3.39) (-3.69)
ITA Bad News 3.14 3.20 2.56
(1.24) (1.26) (1.02)
ITA Good News -4.23 -4.23 -2.34
(-1.86) (-1.87) (-1.18)
SMP News -15.37 -15.33 -11.42
(-17.24) (-17.07) (-9.29)
OMT News -14.85 -14.76 -12.34
(-4.58) (-4.54) (-4.29)
LTRO News -1.95 -1.85 -1.98
(-0.80) (-0.77) (-0.88)
Bank Itraxx 0.01 0.01
(3.71) (3.59)
Adj −R2 0.81 0.01 0.03 0.73 0.77 0.04 0.63 0.01 0.63 0.85 0.95
26
Panel B: 10-year maturityGrowth -0.05
(-0.88)
Growth Spread 0.18 0.26 0.26 0.13 0.10 0.47
(1.61) (3.66) (3.59) (2.24) (1.80) (2.94)
Log(VIX) 0.19 0.45 0.46 0.03
(1.94) (3.80) (3.25) (0.10)
Credit Spread 0.01 -0.03 -0.32
(0.11) (-0.29) (-1.46)
Bad News 0.74 0.72 0.14
(0.36) (0.35) (0.07)
Good News -7.12 -7.14 -6.15
(-3.33) (-3.35) (-3.64)
ITA Bad News 2.76 2.82 2.36
(0.89) (0.90) (0.78)
ITA Good News -6.41 -6.42 -4.29
(-2.51) (-2.52) (-1.89)
SMP News -17.85 -17.85 -12.68
(-25.72) (-25.58) (-8.76)
OMT News -9.54 -9.49 -6.73
(-1.65) (-1.64) (-1.63)
LTRO News -0.03 0.01 -0.61
(-0.01) (0.01) (-0.30)
Bank 0.01 0.01
(3.85) (4.50)
Adj −R2 0.78 0.01 0.02 0.73 0.74 0.04 0.63 0.01 0.62 0.83 0.94
27
Table 5: Correlations of Macro Factors with Yields and Spreads PCs
We show unconditional linear correlation coefficients between real-time macroeconomic indices and sovereign
yield risk-free Principal Components (Panel A), sovereign ALL yield Principal Component (Panel B), and
sovereign yield spread Principal Components (Panel C). The sample period is from September 2004 (when
the European Central Bank sovereign yields become available) to December 2012.
Panel A: AAA yields PCs and macro
GROaaa GROdiff
PCAAA1 0.35 -0.52
PCAAA2 -0.51 -0.49
PCAAA3 0.25 0.08
Panel B: ALL yields PCs and macro
GROaaa GROdiff
PCALL1 0.39 -0.19
PCALL2 -0.40 0.10
PCALL3 -0.16 -0.18
Panel C: Sovereign yield spreads PCs and macro
GROaaa GROdiff
PCSPR1 -0.13 0.75
PCSPR2 0.02 0.25
PCSPR3 -0.33 -0.11
28
Table 6: Macro Factors and Principal Component Residuals
We show the explanatory power of macro factor for AAA risk-free rates and yield spreads beyond the first
three Principal Components. Panels report the R2 of univariate regressions where yield or yield spread
residuals are regressed on macroeconomic factors. Sample period is 2004-2012.
AAA resid SPR resid
GROaaa GROdiff GROaaa GROdiff
3m 9.68 3.08 0.46 3.82
6m 3.74 0.64 0.02 10.30
1y 11.70 4.77 0.02 12.35
3y 6.52 1.05 0.02 16.06
5y 4.11 2.96 0.11 19.74
7y 1.06 0.09 0.22 20.83
10y 0.96 0.85 0.59 20.87
29
1999 2001 2003 2005 2007 2009 2011 2013-1
0
1
2
3
4
5
6
3mo6mo1-y3-y5-y7-y10y
1999 2001 2003 2005 2007 2009 2011 2013-100
0
100
200
300
3mo6mo1-y3-y5-y7-y10y
Figure 1: In the upper panel, we plot AAA yields for different maturities expressed as annualrates. In the lower panel, we plot the yield spread between non-AAA and AAA issuers (in basispoints) for different maturities.
30
j=1 j=2 … j=5 j=6 … j=N
1 ... ... ... missing ... ... ...
… ... ... ... missing ... ... ...
… ... ... ... missing ... ... ...
t-22 At-22,1 not released ... missing not released ... ...
t-21 not released At-21,2 ... missing At-21,6 ... ...
… not released not released ... missing not released ... ...
t At,1 not released ... At,5 not released ... ...
t+1 not released At+1,2 ... not released At+1,6 ... ...
… not released not released ... not released discontinued ... ...
… ... ... ... ... discontinued ... ...
T ... ... ... ... discontinued ... ...
j=1 j=2 … j=5 j=6 … j=N
1 ... ... ... missing ... ... ...
… ... ... ... missing ... ... ...
… ... ... ... missing ... ... ...
t-22 At-22,1 E[At-22,2]=At-43,2 ... missing E[At-22,6]=At-43,6 ... ...
t-21 E[At-21,1]=At-22,1 At-21,2 ... missing At-21,6 ... ...
… E[A ...,1]=At-22,1 E[A...,2]=At-21,2 ... missing E[A...,2]=At-21,6 ... ...
t At,1 E[A t,2]=At-21,2 ... At,5 E[A t,2]=At-21,6 ... ...
t+1 E[At+1,1]=At,1 At+1,2 ... E[At+1,5]=At,5 At+1,6 ... ...
… E[A ...,1]=At,1 E[A ...,2]=At+1,2 ... E[A ...,5]=At,5 discontinued ... ...
… ... ... ... ... discontinued ... ...
T ... ... ... ... discontinued ... ...
Figure 2: This figure shows a stylized example of the macroeconomic announcement data, forN announcement types over a daily sample period between 1 and T . The releases j = 1 andj = 2 are monthly indicators released on two different days of the month. The macroeconomicindicator j = 5 is a news release that did not exist at the beginning of the sample, but wasincluded in the sample from day t onwards. The macroeconomic indicator j = 6 did exist atthe beginning of the sample, but was subsequently discontinued. The top panel represents thematrix of the actual macroeconomic releases in real-time as it is constructed from the data. Thebottom panel shows how our simple forward filling algorithm is used to fill in the expectationof the indicator, based on a local random walk assumption, when it is not released.
31
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012-5
-2
0
2
GDP AAAGrowth Index AAAGDP Non-AAAGrowth Index Non-AAA
Figure 3: We plot the standardized headline GDP growth of Eurozone AAA countries(continuous blue line), the real-time growth index for AAA countries observed on the samerelease dates (dotted blue line), as well as standardized GDP growth of non − AAA countries(continuous red line) and the corresponding real-time growth index for non−AAA countries.
32
1999 2001 2003 2005 2007 2009 2011 2013-4
-3
-2
-1
0
1
2
3
GrowthAAAGrowthNoAAA
1999 2001 2003 2005 2007 2009 2011 2013-50
0
50
100
150
200
250
5-Y
ear Y
ield
Spr
ead
(bp)
1999 2001 2003 2005 2007 2009 2011 2013-2
0
2
4
Gro
wth
Spr
ead
Figure 4: In the upper panel, we plot the real-time growth index for AAA and for non-AAAEurozone countries. In the lower panel, we plot
33
1-year 5-year 10-year0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1AAA
Figure 5: Adjusted R2 of forecasting AAA sovereign bond returns with yield-based factors(first five Principal Components, in blue) and macroeconomic factors (AAA real-time growthand non−AAA-AAA growth differential, in red).
34
1-year 5-year 10-year0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1SPREAD
Figure 6: Adjusted R2 of forecasting non−AAA vs. AAA excess sovereign bond returns withyield-based factors (first five Principal Components of AAA yields, in blue) and macroeconomicfactors (AAA real-time growth and non−AAA-AAA growth differential, in red).
35
1-year 5-year 10-year
0
0.2
0.4
0.6
0.8
1AAA
Figure 7: Adjusted R2 of forecasting AAA sovereign bond returns with yield-based factors(first five Principal Components, in blue) and macroeconomic factors (AAA GDP growth andnon−AAA-AAA GDP growth differential, in red).
36
1-year 5-year 10-year
0
0.2
0.4
0.6
0.8
1SPREAD
Figure 8: Adjusted R2 of forecasting non−AAA vs. AAA excess sovereign bond returns withyield-based factors (first five Principal Components of AAA yields, in blue) and macroeconomicfactors (AAA GDP growth and non−AAA-AAA GDP growth differential, in red).
37