Download Our currency, your problem? The global effects of the euro debt crisis

Our currency, your problem? The global e¤ects of
the euro debt crisis
Livio Stracca
January 8, 2013
Abstract
This paper is an event study focusing on the global e¤ects of the euro debt crisis
in 2010-2012. After identifying 17 key exogenous crisis events, I analyse the impact
on equity returns, exchange rates and government bond yields in 12 advanced and 13
emerging countries. The main e¤ect of euro debt crisis events is a rise in global risk
aversion accompanied by fall in equity returns, in particular in the …nancial sector, in
advanced countries (but not in emerging countries). The e¤ect on bond yields is not
statistically signi…cant for the whole set of countries, but is signi…cant and negative
for key advanced countries such as the US and the UK. The paper also analyse the
transmission channels by looking at how pre-crisis country characteristics in‡uence
the strength and direction of the spill-over.
Keywords: Euro debt crisis, contagion, spill-over, global risk aversion.
JEL: .
European Central Bank (ECB), DG International and European Relations. The views expressed in
this paper belong to the author and are not necessarily shared by the ECB. I thank participants in a
seminar at the ECB, in particular Arnaud Mehl and Roland Straub, and at the International Monetary
Fund.
1
1
Introduction
The euro sovereign debt crisis has made headlines the world over and attracted global
attention. Renewed turmoil in global …nancial markets has been associated with it by
observers and policy makers, and discussed in main international meetings such as the
G7 and the G20. Against this background, the main purpose of this paper is to try
and measure the impact of the crisis on global …nancial markets, using an event study
approach. More generally, this exercise may be useful to understand the external spillover
emanating from the euro area and the crisis may be interpreted, in this regard, as an
interesting pseudo-natural experiment for the global transmission of shocks. In this regard,
it is worth emphasising that the euro debt crisis is a particularly interesting example
in order to understand contagion and spillover, because it was punctuated by events
and decisions at the political level (at both national and European level, adding further
complexity), arguably more so than other crisis episodes in the past. It is therefore easier
to identify truly exogenous events driving its evolution.
In this paper we regards the "euro debt crisis" mainly as a phenomenon a¤ecting the
totality of the euro area and hence focus on the two largest high-yield countries, Spain and
Italy (as opposed to the smaller countries such as Greece, Ireland and Portugal). In this
respect, a worsening of the crisis may be interpreted as a shock increasing the distance
of Spanish and Italian government bonds from, and the substitutability with, German
bonds. Fluctuations in the spreads between sovereign bond yields are of course normal
and may well re‡ect country fundamentals as well as factors other than a worsening of the
euro debt crisis, such as increases in global risk aversion and shifts in sentiment elsewhere.
Nevertheless, sharp increases in government bond yield spreads within a monetary union
have been widely associated with a worsening of market sentiment on the viability of
the euro area in the medium to long term, and I therefore focus on this variable for the
identi…cation of crisis events.1
The period covered by the analysis goes from 2010 to 2012. We identify crisis events
based on three restrictions. First, on the event day there must be a very large increase
1
Note that in this paper I do not de…ne precisely what the “crisis” is. It may encompass the market
expectation of a sovereign default in a large euro area country, the simultaneous default in more than
one euro area country, possibly brought about by a self-ful…lling spiral of lack of con…dence, or event
redenomination risk or the prospect of a break up of the euro area. What the crisis scenario exactly
implies was probably unclear even in the minds of market participants.
1
in the government bond spread in both Italy and Spain vs. Germany (the assumption on
how large is based on the statistical distribution of daily changes in the spread). Second,
the movement must be motivated by a speci…c euro area event happening on that day
and widely reported in the …nancial press as the motivating factor for the jump in the
spread. I focus in particular on announcements by policy makers and decisions taken at
the European level, which have re‡ected repeated attempts at crisis resolution. Third,
no other important event should have intervened on the same day, in particular not from
outside the euro area (such as, e.g., a large data surprise from the US). Note that the
focus on days characterised by large movements in intra-euro area spreads is similar in
spirit to the identi…cation through heteroscedasticity proposed by Rigobon (2003).
After identifying crisis events, I examine their impact on daily equity returns, equity
returns for the …nancial sector, bond yields and exchange rates in 25 non-euro area countries (12 of which advanced) using a standard event study approach (see McLeod 1997
for a survey of event studies in economics). Generally speaking, there are advantages and
disadvantages associated with event studies. On the positive side, the identi…cation of
exogenous shocks is easier and cleaner than is usually the case in macroeconomic models.
On the other hand, event studies like the present one can only be used to analyse the
short term impact of exogenous shocks on …nancial market variables, and it is also di¢ cult
to give a structural connotation to the events without a theoretical model underpinning
the analysis. Hence, this study should be considered only as a …rst step towards the
understanding of the global implications of the euro debt crisis.
This paper is related to a large literature on contagion in …nancial markets; for a recent
survey, see Forbes (2012). As explained by Forbes, contagion (as opposed to interdependence) can be explained as the transmission of very large (in particular negative) events
such as crises, which is closely related to the event study in this paper. The spirit of the
analysis is especially close to the literature trying to identify the channels of contagion;
e.g., Van Rijckeghem and Weder (2001).
Two recent papers focused on the transmission of the euro debt crisis and are for
that reason also closely related to the present study. Aizenmann et al. (2011) analyse
the e¤ects of the global …nancial crisis (Lehman) and the euro debt crisis on stock and
bond market indices in developing countries, up to end-2011. In order to identify crisis
events, that paper looks at (i) daily news from the euro area, (ii) abnormal (very large)
2
returns in four …nancial indicators (the VIX, the EONIA 3-month EONIA swap, the 5year CDS index for Europe, the Fitch 1-year default probability of Western Europe).2
After identifying 23 euro crisis-related events in this way, the authors look at abnormal
returns around the event date and …nd that responses in developing countries to euro
crisis events are generally rather small, signi…cantly less than for the global …nancial crisis
(although they do not consider the 2012 events which arguably marked the peak of the
euro debt crisis). They also …nd that the e¤ect is larger in countries having a higher
trade exposure to the euro area (measured by exports to the euro area divided by the
country’s GDP). Claessens, Tong and Zuccardi (2011) look at 3 speci…c events, 10 May
2010 and 21 July 2011 (positive) and 8-10 June 2011 (negative), and is based on …rmlevel stock returns in EU and non-EU countries (for the most part advanced countries
– over one third from Japan). They run a cross sectional regression for each event and
…nd that …rms’ …nancial dependence matters for the impact of the events if interacted
with country-level bank exposure to euro area high-yield countries, and that the e¤ect
is positively related to trade linkages to the same countries. The present paper has a
somewhat a di¤erent focus compared with both Aizenmann et al. (2011) and Claessens
et al. (2011): its main objective is to understand the global implications of the crisis
events and the transmission to both advanced and emerging countries (di¤erent from
Aizenmann et al.) and on country-level variables (di¤erent from Claessens et al. 2011).
Moreover, the identi…cation of the crisis events is di¤erent from both papers and covers
the year 2012, arguably the most important in the unfolding of the crisis. Fratzscher
(2009) looks at the global e¤ects of negative US-speci…c macroeconomic shocks during
the Lehman crisis and …nds them to have triggered a signi…cant strengthening of the US
dollar, rather than a weakening. Macroeconomic fundamentals and …nancial exposure of
individual countries are found to contribute to the transmission process of US shocks.
In particular, countries with low forex reserves, weak current account positions and high
direct …nancial exposure vis-à-vis the United States have experienced larger currency
depreciations during the crisis.3
Finally, a recent literature has analysed spillover and contagion during the crisis period
2
The VIX stands for the Chicago Board Options Exchange Market Volatility Index; the EONIA for
Euro OverNight Index Average; and CDS for Credit Default Swap.
3
Fratzscher et al. (2012) look at the global e¤ects of the non-standard monetary policy operations in
the euro area, in particular the very long term operations (VLTRO).
3
within the euro area, e.g. Amisano and Tristani (2012), Claeys and Vašíµcek (2012), De
Santis (2012), and Kalbaska and Gatkowski (2012); an earlier study on sovereign spreads
within the euro area in the pre-crisis period is Manganelli and Wolswijk (2009). These
papers however do not address the central question of this paper, namely the transmission
of the euro debt crisis to the rest of the world.
The main result of this paper is the euro debt crisis events have important e¤ects on
global …nancial markets, signi…cantly more so for advanced than for emerging countries
(the latter result in keeping with Aizenmann et al. 2011). The most important e¤ect of
the crisis events is a rise in global risk aversion, accompanied by a sizeable fall in advanced
countries’stock markets. The fall in equities is particularly sharp in the …nancial sector,
suggesting that the bank-sovereign nexus is not something which is limited to the euro area
countries only. Equity returns and excess returns for the …nancial sector go down even in
countries (such as Germany and the US) that are safe havens in terms of the reaction of
their government bond yields. Finally, we …nd that the euro depreciates across the board,
but other key exchange rates appear to be only marginally a¤ected.
In terms of the transmission channels, the countries’ risk rating appears to be an
important determinant, though not always with the expected sign. Financial openness
and trade integration with the euro area are amplifying factors for …nancial equities, but
not for the equity market more generally. For government bond yields, a higher public
debt to GDP ratio is also an amplifying factor. More generally, we …nd that the factors
in‡uencing the strength and direction of the transmission are quite highly asset-speci…c,
which should suggest caution in terms of not over-interpreting and over-generalising a
particular …nding, a lesson which may be useful for the literature on contagion more
generally.
The paper is organised as follows. Section 2 presents the data. Section 3 describes the
econometric approach. Section 4 presents the results. Section 5 contains the conclusions.
2
Data
This study is based on daily data for 10-year government bond yields and equity returns
for 25 advanced and emerging countries, available from Datastream (the full country list is
in Table 1 ) from January 2010 to November 2012. Moreover, I also use data for the VIX,
4
oil and gold prices, and key exchange rates, the BBB-AAA US corporate bond spread
at 7-10 year maturity. Since the country sample contains 12 advanced and 13 emerging
countries, it is well balanced and representative of the world economy (it also contains
virtually all systemically important countries with the possible exception of Saudi Arabia).
(Table 1 here)
Identi…cation of the crisis events. This is evidently the most important step in
an event study. As mentioned in the Introduction, I focus on the spread between the
average 10-year government bond yields in Italy and Spain and the corresponding rate in
Germany (henceforth "spread"; see Figure 1 ).
(Figure 1 here)
The …rst di¤erence in this spread exhibits fat tails and is highly non-Normal (with
a kurtosis close to 9), suggesting that it is characterised by large (and not so infrequent) ’jumps’. Based on this observation, following the general idea of identi…cation
by heteroscedasticity proposed by Rigobon (2003), I identify events by imposing three
conditions: (i) there must be a very large movement (jump) in the …rst di¤erence of the
spread, which I impose to be at least 2.5 standard deviations and which should be visible
for both Spain and Italy; (ii) the jump should be associated with a signi…cant political
event or euro area policy makers’ decisions (including the ECB); (iii) it should not be
(even potentially) explained by another non-euro area event occurring in the same day.
Following this approach, I identify 17 events (11 positive ones - namely implying an easing
of the crisis - coded “-1”; 6 negative ones coded “+1”).4 The full list of events and the
related explainations are reported in Table 2.
(Table 2 here)
After identifying the crisis dates, I then look at what country characteristics matter
for their international transmission. Generally speaking I use pre-crisis data, because the
objective is to understand what makes a country more or less vulnerable once the crisis
strikes and to avoid reverse causality problems. The data are mostly drawn from the IMF
4
Note that the sign is chosen so as to have the same sign of the spread following a crisis event.
5
World Economic Outlook (WEO) and International Financial Statistics (IFS) databases.
Table 3 reports the summary statistics for the variables that are used in empirical analysis,
Overall, we have around 19,000 daily observations.
In Table 4, I report correlations between daily equity returns (total market and …nancial sector) and daily changes in 10-year government bond yields. The correlation is
practically zero on both event and non-event days, but this masks considerable heterogeneity within the full sample. In advanced countries, in fact, the correlation is positive
since higher equity returns are accompanied by increases in bond yields and hence falls in
bond prices (especially so in event days, suggesting that bond markets in these countries
are normally seen as a safe haven), while the opposite holds true for emerging markets (not
reported for brevity), where equity and bond prices tend to move in the same direction
again both in event and non-event days.
(Tables 3-4 here)
Size. One possible …rst channel of international transmission is the size of the country
receiving the shock. It could be argued, in particular, that larger countries should be more
insulated from contagion, as it is harder "to rock a bigger boat". I proxy size by population
and world GDP weight, from the IMF WEO database, both in 2009 (hence this variable
only has a cross sectional variation).
Trade and …nancial openness. More open economies, both in trade and …nance,
could be more vulnerable to contagion. I proxy trade openness as the sum of exports
and imports to GDP; …nancial openness as the sum of external assets and liabilities to
GDP (sources are the IMF WEO for the former and IFS for the latter). Trade openness
is measured in 2009, …nancial openness in 2008 due to data availability constraints.
Net Financial Asset position (NFA). Contagion could also be transmitted di¤erently to countries that are net borrowers or net lenders at the international level. Habib
and Stracca (2012) show that the NFA position (relative to GDP) is the most reliable
predictor of the safe haven status of a currency. Data for all countries refer to end 2008.
Country risk rating. If the crisis events lead to changes in global risk aversion,
riskier countries may be more a¤ected than relatively safer ones. We use data on risk
rating from the International Country Risk Guide (ICRG), in particular in (i) economic
risk, (ii) …nancial risk, (iii) political risk and (iv) debt servicing risk. Again, the data refer
6
to end-2009.
Public debt to GDP. Since the euro debt crisis is mostly about public debt sustainability, it is straightforward to expect a stronger impact on non-euro area countries with a
worse …scal sustainability outlook (for example because market participants become more
attentive towards the …scal outlook). We proxy this variable with the public debt to GDP
ratio in 2009. The source of the data is the IMF WEO.
Trade exposure to the euro area. We measure this variable by dividing exports
to the euro area in 2009, divided by either total exports or real GDP (both in US dollars).
Again, the source of the data is the IMF WEO as well as the IMF Direction of Trade
(DOT) statistics.
Financial integration with the euro area. Here we want to measure whether a
certain country i is more or less …nancially integrated with the euro area. In this paper
we focus on price integration rather than quantity-based measures. Indeed, data on cross
border portfolio exposure (e.g., from the IMF Coordinated Portfolio Investment Survey,
CPIS) are incomplete due to non-reporting emerging countries and potentially distorted
by the presence of intermediation hubs such as the United Kingdom or Switzerland.5 We
therefore follow Bekaert et al. (2009) and estimate
rit =
it
+
glo glo
i Ft
+
EA EA
i rt
+ "it
(1)
where rit are monthly equity returns in country i (outside the euro area), and Ftglo is the
…rst principal component of equity returns in all countries including the euro area (but
excluding country i), and rtEA is equity returns in the euro area. We estimate equation
(1) on monthly equity returns between January 1999 and December 2009, hence again
on the pre-crisis period. For each country in Table 1 we derive two possible measures of
integration, (i) the
EA
i
coe¢ cient as such, (ii) its absolute value (in the understanding
that, e.g., even a strong negative co-movement with the euro area could still denote high
…nancial integration).
5
See also Dedola and Lombardo (2012).
7
3
Econometric approach
This is a classic event study, and we estimate a panel …xed-e¤ect equation
xit =
i
+ xi;t
1
+ It + zit + "it
(2)
where x is the variable of interest (e.g., equity returns), It is the euro area debt
crisis dummy, and z are controls. Standard errors are adjusted for serial correlation and
cross sectional dependence (Driscoll-Kraay). The coe¢ cient of interest in this part of the
analysis is .
We also want to understand the channels of transmission of the crisis events and hence
estimate
xit =
i
+ xi;t
1
+ It + zit + Xi;P RECRIS It + vit
(3)
where Xi;P RECRIS is a vector with the country variables of interest at their pre-crisis
value (e.g., the economic risk rating in 2009).6 In this case we are interested in the
parameters; if the parameter associated to a variable in the X vector is signi…cant, it
implies that that particular variable helps explaining the transmission of the shocks to
non-euro area countries.
4
Results
4.1
Preliminary evidence
Before turning to describing the multi-country analysis as in equations (2) and (3), we
…rst report illustrative evidence for selected individual countries in order to gain a …rst
understanding of the evidence. Impulse responses reported in Figures 2-6 are derived
from regressing the variable of interest on up to …ve lags of the crisis dummy, It . This
should also give a …rst idea of whether the e¤ect is concentrated on the …rst day or is
rather more drawn out. I mostly …nd the e¤ect to be con…ned to the …rst day, and for
this reason do not include lags of It in equations (2) and (3).
Figure 2 reports on the e¤ects of crisis events on government bond yields. In the
upper left corner of the …gure the e¤ect on the spread in Italy and Spain is reported.
6
Note that it is not needed to include the variable Xi;2009 separately because it is contained in the
…xed e¤ects.
8
By construction, this is positive and signi…cant, as the two spreads go up by about 40
basis points each. Interestingly, the French-German spread also goes up, but not due to
a rise in French yields (which remain practically unchanged) but rather due to fall, by
about 8 basis points, in the German yields (safe haven assets). This evidence con…rms
the conventional wisdom that France is an intermediate case between the euro high-yield
and low-yield countries. Bonds in other key advanced countries appear to be safe havens,
in particular the UK (around 8 basis points, similar to Germany), less so the US and
Switzerland (4 basis points) and Japan (1 basis point).
(Figure 2 here)
Figure 3 reports on the e¤ects on equity returns. These are negative in particular in
Italy and Spain, and to a lesser degree (but still signi…cantly negative) in Germany and
even in the US. The two measures of global risk aversion that we consider (the VIX and
the BBB-AAA corporate bond spread in the US at 7 to 10 years maturity) both go up
signi…cantly, suggesting a sizeable rise in market stress.
(Figure 3 here)
Figure 4 zeroes in on excess equity returns in the …nancial sector (de…ned as …nancial
equity returns minus equity returns in the whole market in each country). These are again
strongly negative especially in Italy, re‡ecting the bank-sovereign nexus that has been at
the core of the euro debt crisis. Excess returns also go down in Germany, suggesting that
German banks were also exposed to a worsening of the crisis, but signi…cantly less so in
the US.
(Figure 4 here)
Figure 5 reports on the e¤ect on exchange rates. The main message here, which
is hardly surprising, is a depreciation of the euro across the board, by about half a
percentage point.7 Do crisis events a¤ect global exchange rate con…gurations beyond the
euro? Figure 5 also reports on the USD-Swiss franc and the USD-yen bilateral exchange
rates. There is a slight appreciation of the yen vs. the USD and a slight appreciation of
7
The nominal e¤ective exchange rate of the euro also depreciates signi…cantly (not reported for
brevity).
9
the USD vs. the Swiss franc with a delay of one day, but the e¤ects are quantitatively
small and hence not economically signi…cant. Finally, Figure 6 shows the impact of crisis
events on oil and gold prices. Quite surprisingly, the e¤ect is statistically insigni…cant in
both cases. For gold prices, this evidence suggests that this precious metal is not always
a safe haven asset and cannot be relied on as a hedge against a worsening of the crisis
(and perhaps more generally against rises in global risk aversion).
(Figures 5-6 here)
4.2
Baseline multi-country evidence
After presenting …rst illustrative evidence we now turn to the estimation of equation (2)
in a multi-country setting. The baseline results for equation (2) are reported in Table 5. I
…nd ( coe¢ cient) that equity returns go down by about 0.7 percentage points on impact,
and excess equity returns for the …nancial sector by an additional 0.2% (note that total
market returns are now included as a control variable). The e¤ect on bond yields is not
statistically signi…cant on the whole sample.
I also analyse whether the impact of crisis events is symmetric, by distinguishing
between positive and negative events (i.e., coded +1 and -1); see columns (4)-(5). I …nd
that the e¤ects appear to be broadly symmetric and there is no statistically signi…cant
di¤erence between the two
coe¢ cients associated to positive and negative events.
All in all, then, euro debt crisis events lead to (i) a global fall in equity markets and a
rise in global risk aversion, (ii) a fall in excess returns in the …nancial sector, (iii) no change
in long term government bond yields, though bond yields in key advanced countries such
as the US and the UK do go down and can therefore be considered as safe haven assets.
Moreover, the euro depreciates against other major currencies.
(Table 5 here)
4.3
Transmission channels
As the last step in the analysis we now turn to the transmission channels based on the
estimation of equation (3) and the parameters .
Equities. In Table 6 we report on total market equity returns. The main …nding is
that the spillover from euro crisis events is larger for advanced countries (also see columns
10
(9)-(10) where we consider advanced and emerging countries separately). A higher political risk rating dampens the negative e¤ect, but surprisingly a higher …nancial risk rating
has the opposite, aggravating, e¤ect (the latter result may be due to high collinearity between the risk rating measures). Trade openness in general also has a dampening (rather
than amplifying) e¤ect, while trade exposure to, and …nancial integration with, the euro
area appear to play no signi…cant role.
(Table 6 here)
Financial equities. Table 7 shows results for excess returns in the …nancial sector
(again, I control for total returns in the regression). Also for excess returns the impact
of euro debt crisis events is negative and stronger in advanced countries, while it is insigni…cant for emerging countries. Therefore, the spillover to equity markets is largely an
advanced countries phenomenon, which con…rms results in Aizenmann et al. (2011) about
the limited impact of this crisis on developing countries.8 A higher …nancial risk rating
is surprisingly found to have a negative (amplifying) e¤ect, as are …nancial openness and
trade exposure to the euro area (again, for advanced countries only). It is noteworthy
that trade exposure towards the euro area is an amplifying factor for the …nancial sector
in particular, but not for the stock market as a whole.
(Table 7 here)
Long term government bond yields. Finally, in Table 8 I report on the possible
contagion e¤ects of the debt crisis on government bond yields outside the euro area. Note
that a minus sign here indicates that a bond yield falls and hence the considered bond
market acts as a safe haven asset. We …nd that a higher rating in political and debt service
risk is more likely to make a bond market a safe haven, and so does …nancial openness
(contrary to the results for equities). Surprisingly, a higher NFA position has the opposite
e¤ect, for advanced countries only, as have a trade exposure vs. the euro area (signi…cant
only at the 10 per cent level) and a higher public debt to GDP ratio (not signi…cant in
8
Forbes (2012) …nds that countries with large external balance sheets are more vulnerable to contagion
during crises. For the euro debt crisis, we only …nd this result for …nancial equities, not for all equities
and we actually …nd the opposite result for bond yields (see below).
11
all speci…cations). Overall, for bond yields di¤erences between advanced and emerging
countries are somewhat less visible than for equity markets.
(Table 8 here)
Summing up. Table 9 below provides a synoptic overview of the results on the
transmission channels. The general message arising from it is that the channels are highly
asset-speci…c, and it is di¢ cult to make a general statement about a particular transmission channel, which should suggest caution about not over-interpreting the results.
Financial openness, for example, appears to be good for bond yields (bonds of a more
…nancially open country is more likely to be a safe haven after a crisis shock) but have
the opposite e¤ect on …nancial equities.
(Table 9 here)
5
Conclusions
The euro debt crisis has been in the global spotlight in the past couple of years and
has been heavily a¤ected, arguably more than past crises, by events and decisions at
the political level, both national and European. This paper has aimed at measuring the
impact of the crisis on global …nancial markets by conducting an event study around key
political events and decisions in the 2010-2012 period. We look at data on equities, bonds,
exchange rates and commodities from a large sample of advanced and emerging countries.
The main result of this study is that euro debt crisis events which led to large increases
in the spread between Italian and Spanish vs. the German government bond yields have
determined a sizeable rise in global risk aversion and a sell-o¤ of equities, in particular
…nancial equities and more in advanced than in emerging countries. The latter is a
notable …nding given the nature of the shock and suggests a structural break compared
with past crises. Another clear e¤ect of crisis events is a depreciation of the euro across
the board, while the impact on bond yields is generally less clear-cut. However, we …nd
that the government bond market in key advanced countries (most notably the UK and
the US) acts as a safe haven in the wake of crisis events. We also look at the possible
transmission channels and we …nd a bunch of pre-crisis country characteristics to matter
for the e¤ect of crisis events on non-euro area countries, such as the country risk rating,
12
trade and …nancial openness, the trade exposure vs. the euro area, but we also …nd that
the in‡uence of these variables is highly asset-speci…c and sometimes not straightforward
to interpret. This suggests some caution in not drawing generalised lessons from empirical
exercises of this type, an indication that may be useful in the literature on contagion more
generally and which could be usefully taken up in future research.
13
References
[1] Aizenman, J., Jinjarak, Y., Lee, M. and D. Park (2012): "Developing Countries’
Financial Vulnerability to the Euro Crisis: An Event Study of Equity and Bond
Markets", NBER Working Paper No.18028.
[2] Amisano, G. and O. Tristani (2012): "Cross-Country Contagion in Euro Area Sovereign Spreads", Mimeo, European Central Bank.
[3] Beetsma, R., Giuliodori, M., de Jong, F. and D. Widijanto (2012): "Spread the
News: How the Crisis A¤ected the Impact of News on the European Sovereign Bond
Markets", CEPR Discussion Paper No. 9043.
[4] Bekaert, G., Hodrick, R. J. and X. Zhang (2009): "International stock return comovements", Journal of Finance, 64, 6, pp. 2591-2626.
[5] Claessens, S., Tong, H. and I. Zuccardi (2011): "Did the euro crisis a¤ect non…nancial …rm stock prices through a …nancial or trade channel?", IMF Working
Paper No. 11/227.
[6] Claeys, P. and B. Vašíµcek (2012): "Measuring sovereign bond spillover in Europe and
the impact of rating news", Mimeo.
[7] De Santis, R. A. (2012): "The euro area sovereign debt crisis: save haven, credit
rating agencies and the spread of the fever from Greece, Ireland and Portugal", ECB
Working Paper No. 1419.
[8] Dedola, L. and G. Lombardo (2012): "Financial frictions, …nancial integration and
the international propagation of shocks," Economic Policy, 27, 70, pp. 319-359.
[9] Forbes, K. (2012): "The "Big C": Identifying and mitigating contagion", paper
presented at the 2012 Jackson Hole Symposium at the Federal Reserve of Kansas
City.
[10] Habib, M. and L. Stracca (2012): "Getting beyond carry trade: What makes a safe
haven currency?", Journal of International Economics, 87, pp. 50-64.
14
[11] Kalbaska, A., and M. Gatkowski (2012): "Eurozone sovereign contagion: Evidence
from the CDS market (2005-2010)", Journal of Economic Behaviour and Organization, forthcoming.
[12] Jaramillo, L. and A. Weber (2012): "Bond yields in emerging economies: it matters
what state you are in", IMF Working Paper No. 12/198.
[13] Fratzscher, M. (2009): "What explains global exchange rate movements during the
…nancial crisis?", Journal of International Money and Finance, 28, 8, pp. 1390-1407.
[14] Fratzscher, M., Lo Duca, M., and R. Straub (2012): "The global impact of the ECB’s
non-standard measures", ECB, Mimeo.
[15] MacKinlay, C. (1997): "Event Studies in Economics and Finance", Journal of Economic Literature, 35, 1, pp. 13-39.
[16] Manganelli. S. and G. Wolswijk (2009): "What drives spreads in the euro area government bond market?," Economic Policy, 24, pp. 191-240.
[17] Ribogon, R. (2003):
"Identi…cation through heteroskedasticity", Review of Eco-
nomics and Statistics, 85, 4, pp. 777-792.
[18] Van Rijckeghem, C. and B. Weder (2001): "Sources of contagion: is it …nance or
trade?", Journal of International Economics, 54, 2, pp. 293-308.
15
TABLE 1. List of non-euro area countries
Advanced
Emerging
Australia
Canada
Denmark
Japan
Korea
New Zealand
Norway
Poland
Sweden
Switzerland
United Kingdom
United States
Argentina
Brazil
Chile
China
India
Indonesia
Malaysia
Mexico
Russia
South Africa
Thailand
Turkey
Venezuela
TABLE 2. Crisis events
Date
Coding
Change in b.p. in the
average spread ES-IT
vs. DE
Description
10 May 2010
-1
-57
Leaders of the Eurozone
countries resolved in
Brussels to take drastic
action to protect the euro
from further market
turmoil after approving a
$100 billion bailout plan
for Greece.
1-2 December 2010
-1
-50 (over two days)
ECB reported to be
purchasing government
bonds in large scale.
11 July 2011
+1
+48
The crisis engulfs Italy;
public disagreement
between Italian Prime
Minister Berlusconi and
Finance Minister
Tremonti.
21 July 2011
-1
-29
Euro area summit.
8 August 2011
-1
-77
ECB reported to
'actively' buy bonds to
fight the debt crisis.
5 September 2011
+1
36
Bond markets reacting to
perceived U-turn on
austerity by Italy's
Berlusconi.
9 November 2011
+1
+43
High tension on the
markets: reports that
Germany and France had
begun preliminary talks
on a break-up of the
Eurozone.
1 December 2011
-1
-34
Coordinated
announcement on dollar
swap lines by six central
banks.
5 December 2011
-1
-52
Sarkozy-Merkel
agreement that no losses
will be automatically
imposed
to
private
investors in the European
Stability
Mechanism
(ESM).
8 December 2012
+1
+40
ECB President Draghi
rules out more bond
buying ahead of the
summit.
19-20 June 2012
-1
-51 (over two days)
G20 summit in Los
Cabos; reports that
European leaders are
poised to announce a 750
billion euro deal to bail
out Spain and Italy by
buying the countries’
debts.
29 June 2012
-1
-52
EU Leaders Summit:
reports of Germany
accepting demands made
by Italy and Spain for
immediate aid to bring
down rising borrowing
costs.
26 July 2012
-1
-45
ECB President Draghi’s
pledge to do “whatever it
takes” to protect the
single currency.
2 August 2012
+1
+39
Market observers
disappointed by outcome
of ECB Governing
Council amid high
expectations; (view
partially reversed on the
following day).
6-7 September 2012
-1
-71 (over two days)
ECB Governing Council
announcement of the
Outright Monetary
Transactions (OMT).
26 September 2012
+1
+25
Heavy and violent
protests in Spain and
Greece.
17 October 2012
-1
-30
EU Leaders Summit;
promise to reform the
euro area, establish a
banking union, attribute
supervisory powers to the
ECB.
Note: Events are identified by imposing that, on the selected date, (i) there is a large movement
(jump) in the average government bond yield spread of Spain and Italy over Germany, affecting both
countries and larger than 2.5 standard deviations of the series, (ii) there is a specific euro areaoriginated political and policy-making driven event which is reported by the financial press as driving
market developments on that day and (iii) there is no particularly important news elsewhere
justifying a jump on Spanish and Italian bond spreads. Data on events are drawn from several online
sources, in particular the series “Debt crisis: As it happened” maintained by The Telegraph.
TABLE 3. Summary statistics
Variable
Obs
Mean
Std. Dev.
Min
Max
Equity returns
18950
0.02
1.17
-10.11
31.89
Financial equity returns
18950
0.02
1.39
-11.28
36.91
Change in bond yields
18932
0.00
7.98
-777.53
775.34
Advanced country
25
0.48
0.50
0.00
1.00
Emerging country
25
0.52
0.50
0.00
1.00
Size (world GDP weight, %)
25
2.76
4.39
0.17
19.92
Population (millions)
25
166.28
330.31
4.32
1334.50
Public debt to GDP (%)
25
51.14
38.64
5.81
210.25
Economic risk rating rating
25
33.11
4.44
24.00
44.00
Political risk rating rating
25
75.78
10.42
57.00
89.00
Financial risk rating rating
25
40.54
4.61
31.00
48.00
Debt servicing risk rating
25
9.07
1.29
5.50
10.00
Trade openness (imports and exports over GDP)
25
0.63
0.32
0.22
1.63
Financial openness (external assets and liabilities over GDP)
25
2.38
2.73
0.59
12.04
Net Financial Asset position to GDP
25
0.00
0.37
-0.67
1.18
Trade exposure to euro area (scaled by GDP)
25
0.06
0.05
0.01
0.18
Trade exposure to euro area (scaled by total exports)
25
0.16
0.11
0.03
0.45
Financial integration with the euro area (beta)
25
0.01
0.08
-0.13
0.28
Financial integration with the euro area (absolute beta)
25
0.06
0.06
0.00
0.28
Sample period: 1 January 2010 to 30 November 2012. See text, Section 2 on how financial integration
with the euro area is calculated.
TABLE 4. Correlations between daily financial variables
(1)
(2)
(3)
On non crisis days:
Equity returns (1)
1
Financial equity returns (2)
0.88
1
Change in bond yields (3)
0.04
0.04
1
On crisis days:
Equity returns (1)
1
Financial equity returns (2)
0.95
1
Change in bond yields (3)
0
0
1
On non crisis days, adv. countries:
Equity returns (1)
1
Financial equity returns (2)
0.87
1
Change in bond yields (3)
0.37
0.36
1
On crisis days, adv. countries:
Equity returns (1)
1
Financial equity returns (2)
0.96
1
Change in bond yields (3)
0.61
0.63
1
Note: Sample period is 1 January 2010 to 30 November 2012, 25 countries (see Table 1). Crisis dates
are identified as reported in Table 2.
TABLE 5. Baseline results
(1)
Equity
return
Financial equity return, t-1
Equity return
Crisis event
Equity return, t-1
-0.72***
(0.26)
0.03*
(0.02)
(2)
(3)
Financial Change
equity
in
return
bond
yield
0.00
(0.01)
1.05***
(0.01)
-0.18***
(0.04)
Change in bond yield, t-1
(5)
(6)
Financial Change
equity
in
return
bond
yield
0.00
(0.01)
1.05***
(0.01)
2.82
(1.84)
0.03*
(0.02)
-0.00
(0.00)
Crisis event: Worsening
Crisis event: Improvement
Observations
Number of countries
(4)
Equity
return
18,925
25
18,925
25
18,907
25
-0.87***
(0.21)
-0.63
(0.40)
-0.17***
(0.05)
-0.18***
(0.06)
-0.00
(0.00)
3.85
(3.63)
2.22
(2.04)
18,925
25
18,925
25
18,907
25
Pooled fixed-effect OLS regression of equation (1) in text. Sample period of daily data from January
2010 to November 2012; the list of countries is in Table 1. Equity returns are in percentage points,
changes in bond yields in basis points. Driscoll-Kraay standard errors in parentheses; *** p<0.01, **
p<0.05, * p<0.1.
TABLE 6. Transmission channels for daily equity returns
Dependent variable: Daily equity returns, total market
Equity return, t-1
Crisis event
Advanced country*Crisis event
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
0.03*
(0.02)
-0.72***
(0.26)
-0.14***
(0.05)
0.03*
(0.02)
-0.72***
(0.26)
-0.17***
(0.05)
0.03
(0.02)
-0.71***
(0.26)
-0.42***
(0.08)
0.07
(0.04)
-0.06
(0.08)
0.29**
(0.12)
-0.16*
(0.09)
0.03
(0.09)
0.03
(0.02)
-0.72***
(0.27)
-0.45***
(0.10)
0.13***
(0.05)
0.03
(0.02)
-0.72***
(0.27)
-0.40***
(0.09)
0.10**
(0.05)
0.03
(0.02)
-0.72***
(0.27)
-0.39***
(0.09)
0.10**
(0.04)
0.03
(0.02)
-0.75***
(0.27)
0.02
(0.03)
-1.16***
(0.36)
0.03
(0.02)
-0.29
(0.32)
0.04
(0.04)
0.03
(0.02)
-0.72***
(0.27)
-0.39***
(0.09)
0.10**
(0.04)
0.12
(0.08)
0.10
(0.12)
0.28***
(0.09)
-0.20***
(0.04)
0.27***
(0.09)
-0.17**
(0.07)
0.24**
(0.11)
-0.17***
(0.05)
-0.10*
(0.05)
-0.15***
(0.04)
0.24**
(0.11)
-0.17***
(0.05)
0.22**
(0.09)
-0.22**
(0.10)
0.29**
(0.12)
-0.16
(0.14)
0.14**
(0.06)
-0.02
(0.07)
-0.01
(0.05)
0.17**
(0.07)
0.13**
(0.05)
0.11**
(0.06)
0.13**
(0.05)
0.10
(0.11)
0.15**
(0.07)
9,084
12
8,327
11
Debt to GDP ratio*Crisis event
Economic risk rating*Crisis event
Political risk rating*Crisis event
Financial risk rating*Crisis event
Debt service risk rating*Crisis event
Country size*Crisis event
-0.02
(0.07)
-0.10
(0.09)
Population*Crisis event
Trade openness*Crisis event
Financial openness*Crisis event
NFA position to GDP*Crisis event
Trade exposure vs. the euro area (scaled by exports)*Crisis event
-0.11
(0.11)
0.01
(0.05)
Financial integration with the euro area (beta)*Crisis event
Trade exposure vs. the euro area (scaled by GDP)*Crisis event
Financial integration with the euro area (absolute beta)*Crisis event
Observations
Number of groups
18,925
25
18,925
25
17,411
23
17,411
23
17,411
23
-0.09
(0.09)
0.00
(0.07)
17,411
23
(9)
(10)
Advanced Emerging
countries countries
-0.09
(0.09)
0.00
(0.07)
17,411
23
17,411
23
Notes: See notes to Table 5. The variables that are interacted with the crisis events vary only cross section and refer to the pre-crisis period (2008 or 2009, depending on data availability).
TABLE 7. Transmission channels for daily financial equity returns
Dependent variable: Daily equity returns for the financial sector
Financial equity return, t-1
Equity return
Crisis event
Advanced country*Crisis event
(1)
(2)
(3)
(4)
(5)
(6)
(7)
0.00
(0.01)
1.05***
(0.01)
-0.18***
(0.04)
-0.11***
(0.04)
0.00
(0.01)
1.05***
(0.01)
-0.18***
(0.04)
-0.11***
(0.04)
0.00
(0.00)
1.07***
(0.01)
-0.14***
(0.04)
-0.24***
(0.09)
0.03
(0.04)
0.02
(0.04)
0.08
(0.05)
-0.17***
(0.04)
0.06
(0.03)
0.00
(0.00)
1.07***
(0.01)
-0.14***
(0.04)
-0.08**
(0.03)
0.00
(0.00)
1.07***
(0.01)
-0.14***
(0.04)
-0.03
(0.03)
0.00
(0.00)
1.07***
(0.01)
-0.14***
(0.04)
-0.05
(0.03)
0.00
(0.01)
1.05***
(0.01)
-0.18***
(0.04)
-0.01
(0.01)
1.06***
(0.01)
-0.16***
(0.05)
0.01*
(0.01)
1.04***
(0.01)
-0.02
(0.13)
-0.14**
(0.06)
0.10***
(0.04)
-0.04
(0.05)
0.02
(0.06)
-0.04
(0.06)
-0.00
(0.07)
-0.01
(0.02)
-0.15***
(0.04)
0.00
(0.04)
-0.11***
(0.03)
-0.10***
(0.03)
-0.11***
(0.04)
-0.12***
(0.03)
0.20
(0.29)
-0.14***
(0.03)
-0.15***
(0.02)
-0.10
(0.15)
18,925
25
9,084
12
9,841
13
Debt to GDP ratio*Crisis event
Economic risk rating*Crisis event
Political risk rating*Crisis event
Financial risk rating*Crisis event
Debt service risk rating*Crisis event
Country size*Crisis event
0.04
(0.03)
-0.00
(0.03)
Population*Crisis event
Trade openness*Crisis event
Financial openness*Crisis event
NFA position to GDP*Crisis event
Trade exposure vs. the euro area (scaled by exports)*Crisis event
-0.13***
(0.03)
-0.01
(0.03)
Financial integration with the euro area (beta)*Crisis event
Trade exposure vs. the euro area (scaled by GDP)*Crisis event
Financial integration with the euro area (absolute beta)*Crisis event
Observations
Number of groups
18,925
25
18,925
25
17,411
23
17,411
23
17,411
23
(8)
(9)
Advanced Emerging
countries countries
-0.13***
(0.03)
-0.02
(0.04)
17,411
23
Notes: See notes to Table 5. The variables that are interacted with the crisis events vary only cross section and refer to the pre-crisis period (2008 or 2009, depending on data availability).
TABLE 8. Transmission channels for daily changes in 10-year bond yields
Dependent variable: Daily changes in 10-year government bond yields
Change in bond yield, t-1
Crisis event
Debt to GDP ratio*Crisis event
(1)
(2)
(3)
(4)
(5)
(6)
(7)
-0.00
(0.00)
2.82
(1.84)
-0.00
(0.00)
2.82
(1.84)
-0.07***
(0.02)
-0.01***
(0.00)
0.00***
(0.00)
-0.00
(0.01)
-0.03***
(0.00)
-0.00
(0.00)
-0.01***
(0.00)
-0.07***
(0.02)
-0.01***
(0.00)
0.00*
(0.00)
-0.07***
(0.02)
-0.01***
(0.00)
0.01***
(0.00)
-0.07***
(0.02)
-0.01***
(0.00)
0.00***
(0.00)
-0.07***
(0.02)
-0.01***
(0.00)
0.00**
(0.00)
-0.00
(0.02)
0.01
(0.01)
0.00
(0.00)
-0.09***
(0.03)
0.01
(0.02)
0.01
(0.01)
-0.02***
(0.00)
-0.02***
(0.00)
-0.02***
(0.00)
-0.02***
(0.00)
-0.04***
(0.01)
0.00
(0.01)
-0.01***
(0.00)
-0.02***
(0.01)
-0.01***
(0.00)
-0.01***
(0.00)
-0.03***
(0.01)
-0.02**
(0.01)
-0.02***
(0.00)
0.01*
(0.00)
0.01*
(0.00)
-0.00
(0.01)
-0.02***
(0.00)
0.01*
(0.00)
-0.01***
(0.00)
0.01***
(0.00)
-0.02***
(0.00)
0.02***
(0.00)
-0.03
(0.04)
0.00
(0.01)
17,393
23
17,393
23
17,393
23
9,084
12
8,309
11
Economic risk rating*Crisis event
Political risk rating*Crisis event
Financial risk rating*Crisis event
Debt service risk rating*Crisis event
Country size*Crisis event
Population*Crisis event
Advanced country*Crisis event
Trade openness*Crisis event
-2.75
(1.77)
-1.17
(0.75)
-0.50
(0.33)
-0.00
(0.00)
-0.01***
(0.00)
0.01***
(0.00)
Financial openness*Crisis event
NFA position to GDP*Crisis event
Trade exposure vs. the euro area (scaled by exports)*Crisis event
Financial integration with the euro area (beta)*Crisis event
Trade exposure vs. the euro area (scaled by GDP)*Crisis event
Financial integration with the euro area (absolute beta)*Crisis event
Observations
Number of groups
18,907 18,907
25
25
17,393
23
17,393
23
(8)
(9)
Advanced Emerging
countries countries
0.01*
(0.00)
-0.00
(0.00)
Notes: See notes to Table 5. The variables that are interacted with the crisis events vary only cross section and refer to the pre-crisis period (2008 or 2009, depending on data availability).
TABLE 9. Synoptic table with the results on the transmission channels
Equity returns
Financial equities (excess returns)
Advanced country
A
A
Political risk rating
D
Government bond yields
D
Debt servicing risk rating
D
Financial risk rating
A
Trade openness
D
A
Financial openness
A
D
Trade vs. euro area
A
A
Public debt to GDP
A
NFA
A
Note: See Tables 6-8 for a detailed presentation of the results. ‘A’ stands for ‘amplifier’ that implies that the variable contributes to (i) increase
the negative effect on equity and financial equity returns in absolute size and (ii) increase a positive effect on government bond yields. ‘D’
stands for ‘dampening’ and is the opposite concept to A.
FIGURE 1. Average spread in 10-year government bond yields, average Spain-Italy vs. Germany
Source: Datastream.
FIGURE 2. Effect of crisis events on selected bond yields
Ch an ge in b on d sp re ad vs Ge rm an y, I taly
Ch an ge in b on d sp re ad vs Ge rm an y, S p ain Ch an ge in b on d sp re ad vs Ge rm an y, F ran c e
.5
.5
.4
.4
.3
.3
.2
.2
.1
.1
.10
.08
.06
.04
.02
.00
-.02
.0
.0
-.1
-.1
1
2
3
4
5
-.04
-.06
1
Ch an ge in b on d yie ld , F ran c e
2
3
4
5
1
Ch an ge in b on d yie ld , Ge rm an y
.04
.04
.04
.03
.02
.02
.02
.00
.01
2
3
4
5
Ch an ge in b on d yie ld , U n ite d S tate s
.00
-.02
.00
-.02
-.04
-.01
-.04
-.06
-.02
-.03
-.08
-.04
-.10
1
2
3
4
5
.005
-.08
1
Ch an ge in b on d yie ld , Jap an
.010
-.06
2
3
4
5
Ch an ge in b on d yie ld , S witz e rlan d
.03
.04
.02
.02
.01
2
3
4
5
.00
.00
.000
1
Ch an ge in b on d yie ld , U n ite d Kin gd om
-.02
-.01
-.005
-.04
-.02
-.06
-.03
-.010
-.08
-.04
-.015
-.05
1
2
3
4
5
-.10
1
2
3
4
5
1
2
3
4
5
Note: The impulse responses are based on a regression of daily data for changes in bond yields on the crisis event dummy (see Table 2) and up to give lags.
Data are in absolute values (e.g. 0.1 is 10 basis points). The sample period goes from 1 January 2010 to 30 November 2012.
FIGURE 3. Effect of crisis events on stock returns and risk aversion measures
St ock ret urn, It aly
St ock ret urn, Spain
.01
.01
.00
.00
-.01
-.01
-.02
-.02
-.03
-.03
-.04
-.04
1
2
3
4
1
5
2
3
4
5
St ock ret urn, Unit ed St at es
St ock ret urn, Germany
.008
.005
.004
.000
.000
-.005
-.004
-.010
-.008
-.015
-.012
-.020
-.016
1
2
3
4
1
5
Change in t he VIX
2
3
4
5
Change in t he BBB-AAA corporat e bond spread, US
.04
.12
.10
.03
.08
.06
.02
.04
.02
.01
.00
-.02
.00
-.04
-.06
-.01
1
2
3
4
5
1
2
3
4
5
Note: The impulse responses are based on a regression of daily data for equity returns on the crisis event dummy (see Table 2) and up to give lags. Data are
in absolute values (e.g. 0.1 is 10%). The sample period goes from 1 January 2010 to 30 November 2012.
FIGURE 4. Effect of crisis events on excess equity returns for the financial sector
Ex cess stock return of the financial sector, Italy
.008
Ex cess stock return of the financial sector, Spain
.004
.004
.000
.000
-.004
-.004
-.008
-.008
-.012
-.016
-.012
-.020
-.024
-.016
1
2
3
4
5
1
2
3
4
5
Ex cess stock return of the financial sector, Germany Ex cess stock return of the financial sector, United States
.004
.003
.002
.002
.001
.000
.000
-.002
-.001
-.004
-.002
-.006
-.003
-.008
-.004
-.010
-.005
1
2
3
4
5
1
2
3
4
5
Note: The impulse responses are based on a regression of daily data for excess equity returns for the financial sector on the crisis event dummy (see Table 2)
and up to give lags. Data are in absolute values (e.g. 0.1 is 10%). The sample period goes from 1 January 2010 to 30 November 2012. Excess returns are
defined as returns for financial stocks minus total equity returns.
FIGURE 5. Effect of crisis events on selected exchange rates
Exchange rate, EUR/USD
Exchange rate, EUR/Swiss franc
.004
.012
.002
.008
.000
.004
-.002
.000
-.004
-.004
-.006
-.008
-.008
1
2
3
4
5
1
Exchange rate, EUR/yen
2
3
4
5
Exchange rate, EUR/British pound
.004
.002
.002
.001
.000
.000
-.002
-.001
-.004
-.002
-.006
-.003
-.008
-.010
-.004
1
2
3
4
5
1
Exchange rate, USD/Swiss franc
.004
2
3
4
5
Exchange rate, USD/yen
.003
.002
.002
.000
-.002
.001
-.004
.000
-.006
-.001
-.008
-.010
-.002
1
2
3
4
5
1
2
3
4
5
Note: The impulse responses are based on a regression of daily data for log changes in nominal exchange rates on the crisis event dummy (see Table 2) and up
to give lags. Data are in absolute values (e.g. 0.1 is 10%). The sample period goes from 1 January 2010 to 30 November 2012. A negative value denotes a
depreciation for the EUR rates and an appreciation for the USD rates.
FIGURE 6. Effect of crisis events on gold and oil prices
Gold price growth rate
.004
.002
.000
-.002
-.004
-.006
-.008
1
2
3
4
5
Brent oil price growth rate
.008
.006
.004
.002
.000
-.002
-.004
-.006
1
2
3
4
5
Note: The impulse responses are based on a regression of daily data for log changes in gold and oil prices in USD on the crisis event dummy (see Table 2) and
up to give lags. Data are in absolute values (e.g. 0.1 is 10%). The sample period goes from 1 January 2010 to 30 November 2012.