Monetary policy spillovers: the impact of ECB conventional and unconventional monetary policies on the Swiss stock market

ABSTRACT

We examine the impact of conventional and unconventional ECB monetary policies on the Swiss stock market. Our results show that Swiss stock returns respond significantly to conventional and unconventional monetary policy measures, with conventional monetary policy having an impact only before the financial crisis. The response to unconventional policy shocks is less strong after the introduction of a minimum exchange rate regime by the Swiss National Bank (SNB). Our results are robust to simultaneity and endogeneity problems.

KEYWORDS:

• Event-study
• monetary policy shocks
• stock market
• unconventional monetary policy

JEL CLASSIFICATION:

• C32
• G12
• E44

I. Introduction

An extensive literature provides empirical evidence that both conventional and unconventional monetary policy surprises in the United States (US) and the Euro area exert substantial spillover effects on international bond and stock markets (e.g. Ehrmann and Fratzscher 2009; Wang and Zhu 2013; Bauer and Neely 2014; Rogers, Scotti, and Wright 2014; Neely 2015; Fratzscher, Lo Duca, and Straub 2016; Albagli et al. 2019; Ha 2021; Ferreira and Serra 2022). In this context, prior research has focused either on emerging economies (e.g. Bowman, Londono, and Sapriza 2015; Tillmann 2016) or large advanced economies (e.g. Bauer and Neely 2014; Rogers, Scotti, and Wright 2014; Neely 2015) but only limited evidence is available on the impact of monetary policy spillovers on stock markets in small advanced open economies (SAOEs). In this context, Switzerland is for at least two reasons of particular interest for such an analysis. First, the Swiss stock market is dominated by companies that derive their revenues from global activities with a tight link to the euro area. The weight of the euro area in the trade-weighted Swiss franc exchange rate index provided by the Swiss National Bank (SNB) varied between 40% and 60% from 2002 to 2022 (https://data.snb.ch/en). Second, the SNB enforced a minimum exchange rate (MER) of 1.20 Swiss franc per Euro between 6 September 2011 and 15 January 2015. This constellation provides an ideal setting to study the impact of a change in the domestic monetary policy regime on the size of spillover effects of ECB monetary policy shocks on the domestic stock market.

Moreover, our results complement the findings of Ter Ellen, Jansen, and Midthjell (2020) showing that ECB policy rate surprises have a heterogenous impact on stock markets in SAOEs Denmark, Norway and Sweden. They provide evidence that only Norwegian equity prices react to unexpected changes in ECB monetary policy, while the stock market of Sweden and Denmark is not affected by ECB policy spillovers at any common significance level. To the best of our knowledge, Bernhard and Ebner (2017) is the only paper that studies cross-border spillover effects of unconventional monetary policies (UMPs) conducted by major central banks on Swiss asset prices. In our empirical analysis we follow a similar approach as Bernhard and Ebner (2017) but extend and complement their work by considering three important differences. First, their focus is on UMP surprises, and they do not consider conventional monetary policy shocks. Second, we examine whether the impact of ECB monetary policy surprises is different in crisis and non-crisis years (e.g. Gregoriou et al. 2009; Wang and Mayes 2012; Kontonikas, MacDonald, and Saggu 2013). Third, as in Rogers, Scotti, and Wright (2014); Haitsma, Unalmis, and de Haan (2016); Fausch and Sigonius (2018) and Ferreira and Serra (2022) we use the change in the spread between 10-year Italian and German government bond yields as a proxy to identify UMP surprises. Empirically, we find that the Swiss stock market responds significantly to conventional and unconventional ECB monetary policy measures.

II. Data and econometric methodology

Econometric framework

We first apply an event-study method as used by Bernanke and Kuttner (2005); Haitsma, Unalmis, and de Haan (2016) and Fausch and Sigonius (2018) to measure the impact of conventional and unconventional monetary policy on the Swiss stock market. For that purpose, we estimate the following econometric model

(1) $r_t=\hspace{0.17em}\alpha +{\beta }_1\left(1-D_t\right)\mathrm{\Delta }{\mathrm{r}}_{\mathrm{t}}^{\mathrm{u}}+{\gamma }_1\left(1-{\mathrm{D}}_{\mathrm{t}}\right)\mathrm{\Delta }{\mathrm{r}}_{\mathrm{t}}^{\mathrm{e}}+{\beta }_2{\mathrm{D}}_{\mathrm{t}}\mathrm{\Delta }{\mathrm{r}}_{\mathrm{t}}^{\mathrm{e}}+{\gamma }_2{\mathrm{D}}_{\mathrm{t}}\mathrm{\Delta }{\mathrm{r}}_{\mathrm{t}}^{\mathrm{e}}+\phi \mathrm{\Delta }{\mathrm{r}}_{\mathrm{t}}^{\mathrm{u},\mathrm{c}}+\delta {\mathbf{X}}_{\mathrm{t}}+{\epsilon }_{\mathrm{t}}$(1)

where $r_t$ denotes the stock return on day $t$, $\alpha$ is a constant, $D_t$ is a dummy variable that takes values of 0 and 1 depending on the chosen model specification. $D_t$ is equal to 1 for the period after 2007, characterized by various crises and low policy rates, and 0 before. We define the ECB’s announcement of the first unconventional monetary policy on 22 August 2007 as the start of this policy regime. In a second specification the dummy takes the value of 1 for the time period where a minimum exchange rate regime (MER) was in place. $\mathrm{\Delta }{r}_t^u$, $\mathrm{\Delta }{r}_t^e$ and $\mathrm{\Delta }{r}_t^{u,c}$ denote the conventional monetary policy surprise, the expected policy rate change and the unconventional monetary policy surprise on day $t$. ${\mathbf{X}}_t$ is a vector of control variables consisting of a global stock market index excluding Europe to control for general economic developments in the rest of the world, the dummy variable $D_t$, a dummy variable that controls for spillovers from non-standard monetary policy announcements by the FED, a dummy that considers announced conventional and unconventional monetary policy measures by the SNB, and finally surprises related to the release of macroeconomic indicators in Germany, Italy, Switzerland and the United States.¹ To mitigate potential endogeneity concerns associated with the reverse causality between monetary policy and financial market conditions we use daily data (Rigobon and Sack 2003). To confirm the robustness of our results, we apply the heteroscedasticity-based identification method suggested by Rigobon (2003) and Rigobon and Sack (2003, 2004) which is robust to endogeneity and omitted variables problems. This approach uses the shift of the variance of the monetary policy shock between announcement days and non-announcement days as instruments for the identification. The dynamics of the short-term interest rates ($\mathrm{\Delta }{i}_t$) and the stock prices ($\mathrm{\Delta }{s}_t$) are assumed to be

(2) $\mathrm{\Delta }{i}_t=\beta \mathrm{\Delta }{\mathrm{s}}_{\mathrm{t}}+\gamma {\mathbf{z}}_{\mathbf{t}}+{\epsilon }_{\mathrm{t}}$(2)

(3) $\mathrm{\Delta }{s}_t=\alpha \mathrm{\Delta }{\mathrm{i}}_{\mathrm{t}}+{\mathbf{z}}_{\mathbf{t}}+{\eta }_{\mathrm{t}}$(3)

where ${\mathbf{z}}_{\mathbf{t}}$ is a vector of exogenous variables which affect both $\mathrm{\Delta }{i}_t$ and $\mathrm{\Delta }{s}_t$. The variable ${\epsilon }_t$ is the monetary policy shock and ${\eta }_t$ is a stock market shock. The shocks ${\epsilon }_t$ and ${\eta }_t$ are assumed to be serially uncorrelated, to be uncorrelated with each other and orthogonal to the vector ${\mathbf{z}}_{\mathbf{t}}$. In the above equation system, $\alpha$ is the coefficient of interest, measuring the impact of a change in $\mathrm{\Delta }{i}_t$ on $\mathrm{\Delta }{s}_t$. To determine $\alpha$ we follow the procedure outlined in Arai (2017) and use generalized methods of moments (GMM) estimation with White (1980) standard errors.

Data and identification of monetary policy

Stock returns are calculated as $r_t=ln(P_t)-ln(P_{t-1})$ where $P_t$ is the closing price at day $t$. We use the Swiss Market Index (SMI) as the main Swiss equity price index and the MSCI World Ex Europe as a global stock market index excluding Europe to control for general economic developments outside Europe. Our sample period is based on daily data and runs from January 1999 to December 2019 and covers 263 announcements of conventional and 39 announcements of unconventional monetary policy measures conducted by the ECB. For conventional monetary policy decisions, we use ECB Governing Council meeting dates. For the period January 1999 to February 2015 the announcement dates for unconventional monetary policy measures are provided by Haitsma, Unalmis, and de Haan (2016) while non-standard monetary policy announcements by the FED for the period January 1999 to February 2014 are taken from Falagiarda, McQuade, and Tirpák (2015). For the period up to and including December 2019 we use for both, the ECB and the FED, press releases to identify the announcements of unconventional policy measures. The most common method used in the literature to obtain the surprise element of a conventional monetary policy change is based on futures market data (e.g. Kuttner 2001; Bernanke and Kuttner 2005; Gürkaynak, Sack, and Swanson 2007). As shown by Bernoth and van Hagen (2004), 3-month Euribor futures rates are an unbiased and reliable predictor of Euro area policy rate changes. We follow Bredin, Hyde, and Reilly (2010) and Fausch and Sigonius (2018) and proxy surprise changes in the ECB policy rate ($\mathrm{\Delta }{r}_t^u$) by changes in the 3-month Euribor futures rate. To measure unexpected unconventional monetary policies, we use the approach in Rogers, Scotti, and Wright (2014); Haitsma, Unalmis, and de Haan (2016); Fausch and Sigonius (2018) and Ferreira and Serra (2022) and approximate the surprise component by the change in the spread between 10-year Italian (relatively risky) and German (relatively safe) government bond yields ($\mathrm{\Delta }{r}_t^{u,c}$). The reason for this choice is that the ultimate goal of the ECB’s unconventional monetary policies was aimed at decreasing sovereign spreads in the Euro area (Quantitative Easing). An increase (decrease) in the spread following an unconventional monetary policy announcement implies a tighter (more expansionary) monetary policy than expected. Following Fausch and Sigonius (2018) we control for various macroeconomic indicators for the USA, Germany, Italy and Switzerland in the event-study regressions. All macroeconomic variables are obtained from Bloomberg.

III. Empirical analysis and results

ECB monetary policy spillovers on the Swiss stock market

Table 1 reports ordinary least squares (OLS) estimates² with Newey and West (1987) standard errors, truncated at lag 9. For the pre-crisis period, without any unconventional monetary policy measures, we find a significant spillover effect of ECB conventional monetary policy surprises on the Swiss stock market.³ Our estimates imply that a 25 basis points surprise decrease in the ECB policy rate is associated with a 1.45% increase in stock returns. For the crisis period, the coefficient of conventional monetary policy is not significant. A Wald test shows that the pre-crisis parameter is significantly different from the crisis parameter.

Table 1. Influence of ECB monetary policy surprises on Swiss stock returns (SMI).

Constant	0.000
	(0.183)
Conventional surprise pre-crisis	−0.058**
	(−2.454)
Expected change pre-crisis	−0.143***
	(−4.001)
Conventional surprise crisis	0.045
	(1.163)
Expected change crisis	−0.020
	(−0.448)
Surprise unconventional	−0.023**
	(−2.641)
${\bar{R}}^2$	0.249
Wald test $F$-statistic	5.088**

This table presents the results of the event-study regression estimated in equation (2.1). t-statistics are based on Newey and West (1987) standard errors, truncated at lag $4{\left(5482/100\right)}^{2/9}=9$ and given in parenthesis. The sample period is January 1999–December 2019. The reported measure of fit, ${\bar{R}}^2$, R² is the adjusted . *, **, ***, indicate statistical significance at the 10, 5, and 1% level, respectively. Number of observations: T = 5482.

For our sample period, we find a significant negative relationship between UMP surprises and the stock market. More specifically, an average decrease of 0.013% on event days in the yield spread between Italian and German government bonds implies a 0.03% increase in stock returns. Our estimates for Switzerland are similar to the empirical findings reported in Ferreira and Serra (2022) for various European stock market indices but contradict the result of Bernhard and Ebner (2017) that a positively surprising UMP announcement by the ECB implies a decrease in Swiss equity prices. Finally, we use Cook’s distance (Cook 1977) to test if our findings are sensitive to outliers and find that our regression results are robust. Table 2 presents the results for the heteroscedasticity-based identification technique of Rigobon and Sack (2004) and confirms the similarity of our findings to the event study methodology.

Table 2. Influence of ECB monetary policy changes on SMI returns with identification through heteroscedasticity.

	Conventional surprise		Conventional surprise crisis		Unconventional surprise
	GMM est	$t$ stat	GMM est	$t$ stat	GMM est	$t$ stat
SMI	−0.064***	−4.517	0.061	0.599	−0.033*	−1.734

This table reports the results of the identification through heteroscedasticity methodology suggested by Rigobon and Sack (2004). Coefficient estimates are obtained using GMM as in Arai (2017). t-statistics are based on heteroscedasticity-robust standard errors. *, **, ***, denote statistical significance at the 10, 5, and 1% level, respectively.

Impact of SNB’s minimum exchange rate policy

We further investigate the impact of SNB’s MER policy on the size and scope of spillover effects of ECB monetary policy shocks on the Swiss stock market. As before, we apply an event-study approach and use the Rigobon-Sack method to check for robustness. Table 3 and Table 4 report the corresponding results.

Table 3. Influence of ECB monetary policy surprises on Swiss stock returns under a MER regime.

	SMI
Constant	−0.000
	(−0.659)
Conventional surprise without MER	−0.045*
	(−1.714)
Expected change without MER	−0.119***
	(−3.706)
Surprise unconventional without MER	−0.077***
	(−3.686)
Conventional surprise with MER	0.026
	(0.666)
Expected change with MER	−0.034
	(−0.925)
Surprise unconventional with MER	−0.014
	(−4.637)
${\bar{R}}^2$	0.246
Wald test $F$-statistic	4.978***

This table presents the results of the event-study regression estimated in equation (2.1) where the dummy variable where the dummy variable $D_t$ takes a value of 1 during the period in which the MER was in place. t-statistics are based on Newey and West (1987) standard errors truncated at lag $4{\left(5482/100\right)}^{2/9}=9$ and given in parenthesis. The sample period is January 1999 - December 2019. The reported measure of fit, ${\bar{R}}^2$, is the adjusted R². *, **, ***, indicate statistical significance at the 10, 5, and 1% level, respectively. Number of observations: 5482.

Table 4. Influence of unconventional ECB monetary policy changes (MER regime) on Swiss stock returns with identification through heteroscedasticity.

	With MER		Without MER
	GMM est	$t$ stat	GMM est	$t$ stat
SMI	−0.022*	−1.662	−0.101***	−4.581

This table reports the results of the identification through heteroscedasticity methodology suggested by Rigobon and Sack (2004) during the period with and without a minimum exchange rate regime. Coefficient estimates are obtained using GMM as in Arai (2017). t-statistics are based on heteroscedasticity-robust standard errors. *, **, ***, denote statistical significance at the 10, 5, and 1% level, respectively.

Our results show that conventional monetary policy shocks are only weakly significant for the SMI in a policy regime without a MER and not statistically significant during the time the MER regime was in place. A Wald test allows us to reject the null hypothesis of equality of the coefficients with respect to conventional and unconventional monetary policy surprises before and during the MER regime at the 1% level. While unconventional surprises are highly significant in both policy regimes, the strength of the effect weakened by a factor of between 3 and 6 after the introduction of the MER. This finding is in contrast to the results of Bernhard and Ebner (2017), which found that the Swiss stock market (SMI) reacted more strongly to foreign UMPs after the MER introduction and that the effect became more significant. However, based on an $F$-test they could not reject the null hypothesis of equality of the before-MER and during-MER coefficients related to unconventional policy surprises.

IV. Discussion and concluding remarks

This paper extends the literature on monetary spillover effects by studying the impact of conventional and unconventional ECB monetary policy shocks on the Swiss stock market. We find that Swiss stock returns respond significantly to conventional and unconventional ECB monetary policy measures, with conventional monetary policy having an impact only before the financial crisis. Furthermore, we provide empirical evidence that the response to unconventional monetary policy shocks is less strong after the introduction of a MER regime by the SNB on 6 September 2011. The empirical results are robust and consistent whether we use an OLS event-study or the heteroscedasticity-based GMM methodology suggested by Rigobon and Sack (2004), which accounts both for simultaneity and omitted variables problems. Our findings suggest that expansionary monetary policy lowers risk premia and increases future expected dividends by stimulating output and profits which generates the empirically observed positive impact on stock prices. However, it is important to note that the sign of the stock market response is not clear ex-ante. The theory of rational bubbles studied in Galí (2014) and Galí and Gambetti (2015) implies contrary to the conventional view that expansionary monetary policy might act to decrease the expected bubble component of asset prices when the bubble is strong and thus decreasing stock market valuations. Moreover, as discussed in Laine (2023) monetary policy may also has a different impact on the term structure of risk premia used to discount future dividends. Laine (2023) shows empirically that expansionary monetary policy lowers short-horizon premia and tends to raise long-horizon premia, while Van Binsbergen et al. (2013) illustrate that the slope of the term structure of risk premia changes over the business cycle. In this context, an econometric analysis how monetary policy spillovers affect the term structure of Swiss stock market risk premia is left for future research.

Credit authorship contribution statement

Jürg Fausch: Conceptualization, Methodology, Code, Validation, Formal analysis, Writing – original draft, Writing – reviewing and editing. Daniel Sutter: Conceptualization, Data curation, Validation, Writing – reviewing and editing.

Acknowledgements

We would like to thank the anonymous referee for valuable comments and suggestions. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Surprises are defined as the difference between the actual announcement and the expected value, measured by the Bloomberg median forecast, of the respective macroeconomic indicator (e.g. Balduzzi, Elton, and Green 2001; Swanson and Williams 2014).

² We do not report OLS estimates for the control variables in (2.1). While the MSCI World excluding Europe is highly significant, the other control variables are all insignificant,

³ In our analysis, we use the SMI as the main Swiss equity index. However, our results are robust when we use instead the SPI as the relevant equity index.