AD ALTA
JOURNAL OF INTERDISCIPLINARY RESEARCH
economic recovery and limited the fall in the inflation (see also
Engen et al., 2015).
Broader study of Hausken and Ncube (2013) examines both
effects of the QE on the interest rates and the yields and the
broader economic effects using event-study and Bayesian Vector
Autoregression (BVAR) models. Results of event studies from
Japan, the United Kingdom (UK), the United States (US) and the
EU are in line with others – the QE caused significant drop in
sovereign yields, affecting the whole yield curve. In addition,
authors provide evidence, that the FED and the BoE were more
effective in lowering yields compared to the BoJ and the ECB.
Authors argue, that in general for all four mentioned areas,
effects of the QE on the gross domestic product (GDP) were
rather limited, or insignificant. On the other hand, in order to
push the inflation higher, the QE is effective tool in ZLB terms
according to the BVAR results presented in this study.
Considering now only the ECB’s QE, in general we find similar
results (lowering yields along the whole yield curve) compared
to the US, although the magnitude of the effects is lower. This is
in our point of view caused by less developed and less
interconnected capital and credit markets in Europe compared to
the US (Suchy and Safar, 2017). We also argue that effects and
implications of the ECB’s QE are less examined (in comparison
to the FED’s QE programmes) because of shorter duration of the
programme. On the other hand, before the QE was launched,
there were other narrower-scale programmes, which are
commonly included in studies dealing with the QE.
Andrade et al. (2016) analyse – in line with other studies –
effects of the ECB’s APP on sovereign yields, additionally
explaining implications to macroeconomy. Using “New Area –
Wide Model” authors came to conclusion, that announcement of
the QE programme reduced significantly and persistently
sovereign long-term bond yields. To add on, authors present that
share prices of the banks with more sovereign bonds in their
portfolios soared – which is in line with portfolio rebalancing
channel theory, taking into account reducing risk of duration.
Study also provides some insights about effects on economy in
the Euro area. Using weighted average least squares (WALS)
and Bayesian model averaging (BMA), Afonso and Jalles (2017)
contributes by assessing the determinants of sovereign bond
yield spreads in the period from 1999 to 2016 with respect to
unconventional monetary policy measures. Besides identifying
such baseline determinants, authors points out that the CBPP
reduced yield spreads in all Euro area countries examined,
particularly in between 2011 and 2013. Authors express
conclusion, that long-term refinancing operations contributed to
reducing the yield spreads in most countries.
Urbschat and Watzka (2017) studied the short-term financial
markets reaction after APP press releases, analysing the
development of bond yields and spreads around these releases
estimating the different asset price channels by quantifying the
cumulative decrease of spreads using event regressions for
several Euro area countries. According to authors, effects in
yields and spreads reduction were most pronounced for the
initial announcement on the PSPP but declined afterwards for
additional announcements. They conclude that rather modest
effects from the portfolio rebalancing are present (for all
countries examined).
On the other hand, Boermans and Vermeulen (2018) investigates
whether the ECB’s PSPP affected Euro area investors’ demand
for bonds by using granular securities holdings data. Empirical
results of this study show that the PSPP did not affect the
coefficients of bond demand functions among the Euro area
investors. This suggests that investors’ preferred habitat for
certain bonds remained stable over the period 2013-2016 despite
the QE programme.
Gambetti and Musso (2017) describes the transmission channels
and provide empirical evidence of the macroeconomic impact of
the APP. Evidence from this study (using Vector Autoregression
– VAR model) suggests that the QE had significant upward
effect on both real GDP and inflation in Euro area during first
two years. Considering time frame, impact on real GDP appears
to be stronger in the short-term, while impact on inflation
appears to be more significant in the medium-term according to
authors. Such results find some support in Hutchinson and Smets
(2017), where main focus is put on successfulness of reaction
function and the way it is communicated, besides examining
reactions of the GDP and inflation to monetary policy
“package”. Results suggest, that policy package absenting,
inflation on average, would be almost half a percentage point
lower than currently projected in each year over 2016–2019.
From different perspective of view, Blattner and Joyce (2016)
examines how shocks to the net supply of government bonds
affect the Euro area term structure of interest rates and the wider
economy. Authors use the BVAR model, while results provide
evidence of significant lowering of Euro area 10-year bond
yields. According to the results, authors argue that the QE
propped up both the inflation and the output gap in the Euro
area.
Considering spillovers of the ECB’s QE, Falagiarda et al.
(2015), describes effects of the QE announcements on non-
eurozone members’ (Czech Republic, Hungary, Poland and
Romania) bond yields. Authors present results of their event-
study suggesting strong spillover effect via the portfolio
rebalancing and signaling channels (excluding Hungary) which
is in line with conclusions in studies examining the FED’s QE
spillovers.
However, considering the QE effects on the equity markets, we
find veer limited literature background. We point out study of
Henseler and Rapp (2018) where authors are interested in the
question of which businesses benefited from the ECB’s QE
stimulus, using the event-study, focusing on substantial cross-
sectional variation in a sample of 2,625 non-financial firms in
the Eurozone. Results show that the announcement returns are
positively correlated with leverage and negatively with size
(authors find it consistent with the credit channel). Furthermore,
the announcement returns are negatively correlated with the
market-to-book ratio, suggesting different exposures of the value
and growth stocks. These patterns are more pronounced once the
programme initiation announcements are only examined
according to this study. Authors also argue, that only “few
existing studies analyse aggregate portfolios or even stock
indices and report mixed results.“ For example Rogers et al.
(2014) find positive announcement returns for the German
equity market. Fratzscher et al. (2017) confirms that for banking
sector and country indices and Haitsma et al. (2016) for the
Eurostoxx50. In contrast, Hosono and Isobe (2014) find negative
returns for both the Eurostoxx Banks and Stoxx600. To sum up,
equity markets related research examining effects of the QE is
rather scarce, with only very limited sample examined.
3 Empirical Analysis
Based on literature linked to the QE, we find event-studies as
most used methodological approach. Papers most related to the
event-studies assume that the surprise part can be measured from
the excess movements in asset prices in a particular window
around the announcement time of the policy decision. Within
this method it is therefore assumed that the monetary policy
shock is fully captured within some time-window around the
chosen event. If this assumption does not hold, the method may
be biased (Rigobon and Sack, 2004). Too narrow window may
cause that part of the reaction to the announcement news would
be missed, too wide window on the other hand may contaminate
the announcement with other news. That is why several papers
(see e.g. Kholodilin et al., 2009; Sondermann et al., 2009; Hayo
and Niehof, 2011 or Rogers et al., 2014) apply the
heteroscedasticity-based identification approach of Rigobon and
Sack (2004). This approach is robust considering endogeneity
and omitted variables problems – therefore relies on weaker
assumptions than the event-study approach. Rosa (2011) then
provides evidence, that the event-study estimates (of the
response of asset prices to monetary policy) contain a significant
bias. But more importantly, Rosa (2011) also concludes that
- 84 -