- Email: [email protected]
The effect of political communication on European financial markets during the sovereign debt crisis
ARTICLE IN PRESS EMPFIN-0875; No of Pages 6 Journal of Empirical Finance xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Journal of Empirical Finance journal homepage: www.elsevier.com/locate/jempfin
The effect of political communication on European financial markets during the sovereign debt crisis✩ Christian Conrad* , Klaus Ulrich Zumbach Heidelberg University, Germany
A R T I C L E
I N F O
Article history: Received 19 July 2015 Received in revised form 20 December 2015 Accepted 10 January 2016 Available online xxxx
A B S T R A C T We quantify all statements by major European politicians reported by Reuters during the August 2011 to December 2011 period and show that political communication significantly affects the EUR–USD exchange rate as well as European stock and bond markets. Communication with respect to Italy induces the strongest market reactions. Financial markets consider the German bond market a safe haven. © 2016 Elsevier B.V. All rights reserved.
JEL classification: E43 E62 G01 G12 C20 Keywords: Political statements High-frequency response Austerity measures Joint liability
1. Introduction We analyze whether the communication of European politicians with respect to the sovereign debt crisis affects the EUR– USD exchange rate as well as the European stock and bond markets. For the August to December 2011 period we quantify all statements by major European politicians that refer to the debt crisis and are reported by the news agency Reuters. We then explain the changes in the EUR–USD exchange rate as well as eight national stock and bond markets in the 15 minutes period following each statement by the content of the communication. The empirical results show that positive communication regarding the periphery countries or the Eurozone (EZ) as a whole leads to a significant appreciation of the EURO against the US dollar. More specifically, the exchange rate reaction is the strongest with respect to communication regarding the periphery countries and, in particular, in response to negative statements.
✩ This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 3202782013-2015. * Corresponding author at: Department of Economics, Heidelberg University, Bergheimer Strasse 58, 69115Heidelberg, Germany. Tel.: +49 6221 54 3173. E-mail addresses: [email protected] (C. Conrad), [email protected] (K. Zumbach).
http://dx.doi.org/10.1016/j.jempfin.2016.01.018 0927-5398/© 2016 Elsevier B.V. All rights reserved.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 2
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Further, the stock markets of the core European countries (Germany, France and Belgium) as well as the periphery countries (Greece, Ireland, Italy, Portugal and Spain) instantaneously increase in response to positive communication regarding the stance of the economy in the periphery countries or positive communication with respect to the EZ as a whole. In sharp contrast, the response of the national bond markets is asymmetric, whereby mainly the Italian and German bond markets are affected. While negative communication regarding the economic situation in Italy leads to an immediate increase in 10-year Italian government bond yields, German government bond yields decrease. This finding is in line with the view that financial market participants consider the German bond market a safe haven and highlights that Italy plays a pivotal role among the periphery countries. During our sample period, Silvio Berlusconi’s Italian government was under strong political pressure and finally had to resign in November 2011. Moreover, our results show that political communication concerning the periphery countries evokes stronger market reactions than statements on the EZ. While statements that suggest an expansion of the European Financial Stability Facility (EFSF) or shared liability for national debts do not lead to decreasing bond yields in the periphery countries, communication about the introduction of further austerity measures does reduce Italian yields. Our results extend upon and complement the recent work of Mohl and Sondermann (2013) and Beetsma et al. (2013) on the effects of political communication on sovereign bond spreads. In contrast to Mohl and Sondermann (2013) we distinguish between positive and negative communications. In addition, in contrast to Beetsma et al. (2013) we obtain bond market response coefficients for each individual country. Most importantly, while both studies employ daily data, we take a highfrequency perspective. Using high-frequency data allows us to monitor the effects of political communication on financial markets in real time and bypass problems with respect to identification and causality. In particular, we do not have to worry about potential control variables. Other related papers are Smeets and Zimmermann (2013), Ehrmann et al. (2014) and Kilponen et al. (2015). Smeets and Zimmermann (2013) investigate the effects of crisis meetings of European heads of state and government on stock and bond markets, while Ehrmann et al. (2014) study the determinants of the volatility of the EUR-USD exchange rate during the crisis. However, both studies again focus on daily data. Finally, Kilponen et al. (2015) focus on the effects of policy announcements on bond markets. Thus, all these studies can be seen as complementary to ours. 2. Data and methodology Our data set starts on August 01, 2011 and ends on December 06, 2011, i.e. it covers 92 trading days. We consider statements and the corresponding asset price changes lying within the trading hours from 9:00 a.m. to 5:30 p.m. Central European Time (CET). Although our sample period is relatively short, we believe that it allows for some unique insights from a subperiod of the sovereign debt crisis that was characterized by the severe uncertainties around the Italian government under Silvio Berlusconi. 2.1. Quantifying statements by European policy makers We search all news reports from Reuters for statements by the EZ’s 17 Heads of Government, their respective Finance Ministers, and the four leading EU representatives—the President of the EU Commission José Manuel Barroso, the President of the Euro Group Jean-Claude Juncker, the EU Economics Commissioner Olli Rehn, and the President of the European Council Herman Van Rompuy. We separate the statements into two groups. In the first group, we collect all statements that refer to the economic situation or austerity measures in the following periphery countries: Italy (IT), Spain (ES), Portugal (PT), Ireland (IE) and Greece (GR). We code each statement, CtP , as +1 if the statement implies a positive outlook for the specific country or the introduction of new austerity measures, and as −1 otherwise.1 For example, the Reuters report: “Italian PM Berlusconi says new agreement on austerity package confirms solidity of ruling coalition” on August 30, 2011 at 12:44 is coded as CtP = +1. The second group contains all statements that refer to the EZ as a whole, in particular to the EFSF, Eurobonds, the role of the ECB or the EURO as a currency. We code statements, CtEZ , as +1 when they – broadly speaking – suggest a shared liability for national debts within the EZ, e.g., statements that support the introduction of Eurobonds or the expansion of the EFSF. Statements with content opposed to such ideas are coded as −1. For example, the Reuters report: “Germany’s Merkel says Eurobonds are absolutely wrong” on September 15, 2011 at 10:47 is coded as CtEZ = −1. We disregard all statements that are either neutral or do not portray a clear message. Overall, Reuters reported 778 statements of which 164 were unanimously quantifiable.2 Table 1 provides a summary of the statements. Among the 164 statements, 77 (87) refer to the periphery countries (the EZ as a whole). 101 (or 62%) of the statements are coded as being positive (+1). Interestingly, for the EZ positive and negative statements are almost balanced. On the contrary, 75% of the statements referring to the periphery countries are positive. Since these countries are at the center of the debt crisis, the high number of positive statements might be surprising at first sight. However, this can be explained by the fact that a large share of these statements refers to the introduction of new austerity measures. Finally, among the statements that refer to the periphery countries the majority are related to Italy (37) and Greece
1 This approach of coding statements is commonly referred to as Content Analysis and has been used by, among others, Ehrmann and Fratzscher (2007) and Conrad and Lamla (2010) for analyzing the effects of central bank communication. 2 Although this means that only 21% of the statements are used in the empirical analysis, we thereby avoid problems of potential misclassification. Further, we excluded multiple communications that were released at the same time or communications that we pure announcements of, e.g, a further summit.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
3
Table 1 Summary of political statements. Positive Periphery (CtP )
IT GR ES IE PT Total
Eurozone (CtEZ ) Total
24 23 4 4 3
Negative 65% 82% 100% 100% 100%
13 5 0 0 0
Total 35% 18% 0% 0% 0%
37 28 4 4 3
58
75%
19
25%
77
43
49%
44
51%
87
101
62%
63
38%
164
Notes: The table shows the number and the tone (positive/negative) of the statements that refer to the periphery countries and the EZ, respectively.
(28). This reflects the fact that during our sample period the political debate was very much focused on these two countries. In contrast, Spain was not yet under such scrutiny. 2.2. Financial data For the stock market indices of three core EZ countries (DAX (Germany), CAC (France), BEL20 (Belgium)) and the five periphery countries (FTSE MIB (Italy), IBEX (Spain), PSI20 (Portugal), ISEQ (Ireland) and ASE (Greece)), we calculate 15-minute returns as rt,t+15 = 100×(ln(Pt+15 )−ln(Pt )). Similarly, we calculate the 15-minute change in the 10-year government bond yields of the respective countries, denoted by it,t+15 . Finally, we consider the returns, ext,t+15 , on the EUR–USD exchange rate. All financial data were obtained from Bloomberg. 2.3. Econometric methodology In order to measure the high-frequency response of the European capital markets to political communication, we follow Almeida et al. (1998) and regress the 15-minute change in the respective financial market, yt,t+15 ∈ {rt,t+15 , it,t+15 , ext,t+15 }, on the political communication Ct ∈ CtP , CtEZ : yt,t+15 = b0 + b1 Ct + et,t+15
(1)
Note that Eq. (1) is not a time series regression. Instead, each observation reflects the respective capital market’s 15-minute adjustment in response to the preceding communication.3 Finally, we also consider regressions in which we split the statements into positive and negative ones: Neg
yt,t+15 = b1 CtPos + b2 Ct
+ et,t+15 .
(2)
This allows us to test for asymmetric response coefficients. 3. Empirical results Table 2 provides a summary of the empirical results. As Panel A shows, political communication has a weak but significant influence on the EUR–USD exchange rate. Positive communication leads to a significant appreciation of the EURO within the 15 minutes following the statement. More specifically, positive statements about the periphery countries, CtP , lead to an appreciation of the Euro of about 0.06% in the 15 minutes after the statement. The statements explain a remarkable 21% of the total change during that 15-minute period. Panel B shows how the national stock markets respond to communication regarding i) the periphery countries, CtP , and ii) the EZ, CtEZ . First, all stock markets significantly increase in response to positive communication regarding the periphery countries. The size of the response is strongest in Germany and Italy with an R2 of 0.32 and 0.29, respectively. The R2 for Portugal, Ireland and Greece are considerably lower. Second, statements in favor of a shared liability for national debts within the EZ lead to significantly positive returns in all stock markets except Greece. Note that in all stock markets the response to CtEZ is weaker than the response to CtP .
3 More precisely, the dependent variable is observed every 15 min. The communication falls somewhere into the 15-minute interval, so that the response is measured at most for a period of 15 min and at least for a period of 1 min.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 4
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Table 2 Financial market response to political communication. Ct
# of obs.
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtP
77
CtEZ
R2 87
0.0563∗∗∗ (0.0123) 0.21 0.0322∗∗∗ (0.0082) 0.15
R2
Core countries GE
CtP
77
CtEZ
R2 87 R2
CtP
77 2
CtEZ
R 87 R2
Periphery countries FR
BE
IT
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2282∗∗∗ 0.2160∗∗∗ 0.1630∗∗∗ 0.2307∗∗∗ (0.0357) (0.0428) (0.0329) (0.0470) 0.32 0.29 0.26 0.29 0.1784∗∗∗ 0.1478∗∗∗ 0.0930∗∗∗ 0.1537∗∗∗ (0.0273) (0.0240) (0.0225) (0.0323) 0.33 0.31 0.17 0.21
ES
PT
IE
GR
IBEX 0.1785∗∗∗ (0.0398) 0.25 0.1321∗∗∗ (0.0268) 0.22
PSI20 0.1167∗∗∗ (0.0275) 0.18 0.0889∗∗∗ (0.0255) 0.13
ISEQ 0.0923∗∗∗ (0.0334) 0.14 0.0735∗∗∗ (0.0233) 0.10
ASE1 0.1120∗∗ (0.0472) 0.06 −0.0050 (0.0546) 0.00
0.1151 (0.1552) 0.00 −0.2773 (0.3229) 0.01
8.2820 (8.6568) 0.10 −1.4056 (4.0341) 0.00
−1.1924 (0.8722) 0.01 0.3810 (0.7649) 0.00
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.6929∗∗∗ 0.3608∗∗ 1.4409 −0.8470∗∗∗ −0.1863 (0.1523) (0.1731) (1.0245) (0.2289) (0.1867) 0.25 0.06 0.07 0.12 0.01 0.2376∗∗ 0.1623 −0.5044 −0.0761 −0.1829 (0.1110) (0.1247) (0.8002) (0.1605) (0.1672) 0.05 0.02 0.00 0.00 0.01
Notes: The table presents the estimates of b1 in Eq. (1). Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels. ∗∗ Indicates significance at the 5% levels. ∗ Indicates significance at the 10% levels. 1 Due to the Greek trading hours (930 –1600 CET), the Greek stock market regressions are based on 67/71 statements respectively. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based on 36/53 statements.
Panel C presents how political communication impacts on European bond markets. In general, the R2 s in the bond market regressions are lower than the corresponding ones in Panel B. We observe significant reactions to communication concerning the periphery countries in the German, French and Italian bond markets only.4 The sign of the response of the German and French bond markets is positive, while the sign of the response of the Italian bond market is negative. More precisely, bad news regarding future economic development in one of the periphery countries leads to a rise in the 10-year Italian government bond yield by 0.85 basis points while the German yield decreases by 0.69 basis points. This bond market response is in line with the interpretation that financial market participants view the German bond market as a safe haven. The fact that Italy is the only periphery country for which we observe a significant market response may be explained by the observation that due to its financial and political situation, the country was under the spotlight in financial markets during our sample period. In comparison to Greece, Ireland, and Portugal, the Italian debt level has been too high to allow the country to be taken under the European rescue umbrella. Therefore, Italy’s debt problem was considered to be the crucial issue in solving the European debt crisis. Since Italy plays such a pivotal role, we analyze the response in capital markets to communication with respect to Italy in more detail in Table 3. The result that the 10-year Greek government bond yields do not react to statements regarding the periphery countries (although a large amount of these statements refer specifically to Greece, see Table 1) may be explained by financial market participants who do not consider the rescue measures taken as being sufficient to solve the Greek debt problem. Interestingly, only the German bond market significantly reacts to communication regarding the EFSF, Eurobonds, the ECB or the EURO. Statements that are in favor of an expansion of the EFSF, for example, lead to rising German bond yields. The finding that the periphery countries’ government bond yields do not react at all to such statements is remarkable. It implies that the periphery countries do not ‘benefit’ from such statements. We also split the statements into positive and negative communication in order to check whether the effect is asymmetric. The results are presented in Table 4 in the Appendix (we present the coefficient estimates only). Most importantly, negative
4 Since Bloomberg reported the 10-year Irish government bond yields until October 11, 2011 only, the corresponding regressions are based on a shorter time period and, hence, include only 36/53 observations.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
5
Table 3 Financial market response to statements regarding Italy. Ct
# of obs.
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtI
37
0.0698∗∗∗ (0.0159) 0.38
R2
Core countries GE
CtI
37 R2
CtI
37 2
R
Periphery countries FR
BE
IT
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2361∗∗∗ 0.2485∗∗∗ 0.2007∗∗∗ 0.2857∗∗∗ (0.0445) (0.0572) (0.0404) (0.0609) 0.45 0.41 0.44 0.42
ES
PT
IE
GR
IBEX 0.2125∗∗∗ (0.0540) 0.35
PSI20 0.1155∗∗∗ (0.0307) 0.36
ISEQ 0.1403∗∗∗ (0.0305) 0.43
ASE1 0.0897 (0.0815) 0.03
0.4237 (0.3373) 0.02
20.8583 (19.0324) 0.26
−0.5587 (1.0818) 0.01
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.9595∗∗∗ 0.5702∗∗ 2.0974 −1.4918∗∗∗ −0.4827* (0.1992) (0.2195) (1.4842) (0.3325) (0.2580) 0.44 0.16 0.09 0.30 0.08
Notes: The table presents the estimates of b1 in the regression yt+15 = b0 + b1 CtI + et,t+15 , where yt,t +15 ∈ {rt,t +15 , it,t +15 , ext,t +15 }. Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels, respectively. ∗∗ Indicates significance at the 5% levels, respectively. ∗ Indicates significance at the 10% levels, respectively. 1 The Greek stock market regression is based on 28 statements. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based is 11 statements.
statements appear to have stronger effects on the exchange rate than positive ones. For example, a negative statement on P,Neg the periphery countries, Ct , leads to an immediate depreciation of 0.07% while the exchange rate appreciates in response to a positive statement, CtP,Pos , by 0.04% only. In contrast, as the table shows, at least for the stock market there seems to be no significantly different response to positive and negative news. In addition, we experimented with splitting the statements according to the speakers, e.g. prime minister vs. finance minister. However, for such an analysis our dataset does not comprise a sufficient number of observations. Finally, Table 3 presents how financial markets respond to communication regarding Italy, CtI . Recall that 37 out of the 77 statements regarding the periphery countries are concerned with Italy. In comparison to Table 2, the estimated coefficients typically suggest a stronger market response and a higher R2 . Interestingly, the R2 for the exchange rate regression now increases to 38% and the response of the EUR–USD is about 0.07%. Thus, the EUR–USD exchange rate reacts more strongly to communication concerning Italy than to communication about the periphery in general. Similarly, the Italian stock market reacts most strongly to positive communication about Italy. More precisely, after a positive statement the FTSE MIB increases on average by 0.29% in the ensuing 15 minute period. In addition to the German, French and Italian bond markets, the Spanish bond market now shows a significant (at the 10% level) reaction as well. In response to positive communication the government bond yields of the core (periphery) countries increase (decline). For example, if a major European politician communicates that Italy will introduce new austerity measures, the Italian government bond yield decreases by 1.49 basis points. Overall, our results highlight the pivotal role that Italy plays among the periphery countries and confirm the safe haven interpretation of the German bond market. 4. Conclusion Our results show that the communication of major European politicians during the sovereign debt crisis has a significant impact on the European financial markets. We find that statements regarding specific periphery countries evoke stronger market responses than statements focused on the EZ as a whole. More specifically, the EUR–USD exchange appreciates more strongly in response to positive statements about the periphery countries than to statements about the EZ as a whole. Most importantly, negative statements regarding the periphery countries trigger the strongest response of the exchange rate. Further, while positive statements of either type lead to significant increases in the stock markets of all countries under analysis, only the German, French and Italian bond markets are affected by political communication. The bond market reactions imply that investors consider the German governments bonds to be a safe haven. Italian government bond yields decrease if political communication hints at improvements in the state of the economy or the introduction of new austerity measures, but they do not react to statements in favor of an expansion of the EFSF or the introduction of Eurobonds. Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 6
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Appendix A
Table 4 Financial market response to political communication. Ct
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtP,Pos
0.0395∗∗∗ (0.0126) −0.0730∗∗∗ (0.0211) 0.0234∗∗ (0.0106) −0.0411∗∗∗ (0.0126)
P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
Core countries GE
CtP,Pos P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
CtP,Pos P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
Periphery countries FR
BE
IT
ES
PT
IE
GR
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2045∗∗∗ 0.1754∗∗∗ 0.1492∗∗∗ 0.1751∗∗∗ (0.0392) (0.0362) (0.0305) (0.0392) −0.2520∗∗∗ −0.2565∗∗∗ −0.1765∗∗∗ −0.2862∗∗∗ (0.0597) (0.0776) (0.0583) (0.0855) 0.1828∗∗∗ 0.1504∗∗∗ 0.0819∗∗∗ 0.1424∗∗∗ (0.0367) (0.0346) (0.0308) (0.0454) −0.1741∗∗∗ −0.1452∗∗∗ −0.1042∗∗∗ −0.1649∗∗∗ (0.0404) (0.0332) (0.0327) (0.0460)
IBEX 0.1283∗∗∗ (0.0332) −0.2287∗∗∗ (0.0724) 0.1314∗∗∗ (0.0394) −0.1329∗∗∗ (0.0364)
PSI20 0.1089∗∗∗ (0.0290) −0.1246∗∗∗ (0.0467) 0.0865∗∗ (0.0416) −0.0912∗∗∗ (0.0296)
ISEQ 0.0734∗∗∗ (0.0218) −0.1112* (0.0631) 0.0707∗∗ (0.0295) −0.0763∗∗ (0.0360)
ASE1 0.1156* (0.0601) −0.1083 (0.0728) −0.0438 (0.0757) −0.0339 (0.0788)
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.5069∗∗∗ 0.2638 0.2397* −0.5414* (0.1278) (0.1619) (0.1312) (0.2832) −0.8789∗∗∗ −0.4579 −2.6421 1.1526∗∗∗ (0.2765) (0.3059) (2.0448) (0.3596) 0.2093 0.1837 −1.8860 −0.2000 (0.1519) (0.1576) (1.2997) (0.2104) −0.2659 −0.1409 −0.8773 −0.0477 (0.1618) (0.1934) (0.9339) (0.2424)
−0.1672 (0.1856) 0.2053 (0.3241) −0.2977 (0.2101) 0.0682 (0.2601)
0.2776 (0.2818) 0.0474 (0.1303) −0.3023 (0.5851) 0.2523 (0.2734)
−1.1931* (0.6657) −17.757 (17.301) −8.7613 (5.6828) −5.9500 (5.7272)
− 1.7638 (1.4162) 0.6211 (1.0183) 1.1302* (0.6372) 0.3682 (1.3907)
Notes: The table presents the response coefficients to positive and negative statements separately. Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels. ∗∗ Indicates significance at the 5% levels. ∗ Indicates significance at the 10% levels. 1 Due to the Greek trading hours (930 –1600 CET), the Greek stock market regressions are based on 67/71 statements respectively. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based on 36/53 statements.
References Almeida, A., Payne, R., Goodhart, C., 1998. The effects of macroeconomic news on high frequency exchange rate behaviour. J. Financ. Quant. Anal. 33, 383–408. Beetsma, R., Giuliodori, M., de Jong, F., Widijanto, D., 2013. Spread the news: the impact of news on the European sovereign bond markets during the crisis. J. Int. Money Financ. 34, 83–101. Conrad, C., Lamla, M.J., 2010. The high-frequency response of the EUR–USD exchange rate to ECB communication. J. Money Credit Bank. 42, 1391–1417. Ehrmann, M., Fratzscher, M., 2007. Communication by central bank committee members: different strategies, same effectiveness?. J. Money Credit Bank. 39, 509–541. Ehrmann, M., Osbat, C., Stráský, J., Uusküla, L., 2014. The euro exchange rate during the European sovereign debt crisis dancing to its own tune?. J. Int. Money Financ. 49 (Part B), 319–339. Kilponen, J., Laakkonen, H., Vilmunen, J., 2015. Sovereign risk, european crisis-resolution policies, and bond spreads. Int. J. Cent. Bank. 11, 285–323. Mohl, P., Sondermann, D., 2013. Has political communication impacted bond spreads?. Appl. Econ. Lett. 20, 48–61. Smeets, D., Zimmermann, M., 2013. Did the EU summits succeed in convincing the markets during the recent crisis?. J. Common Mark. Stud. 51, 1158–1177.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
Contents lists available at ScienceDirect
Journal of Empirical Finance journal homepage: www.elsevier.com/locate/jempfin
The effect of political communication on European financial markets during the sovereign debt crisis✩ Christian Conrad* , Klaus Ulrich Zumbach Heidelberg University, Germany
A R T I C L E
I N F O
Article history: Received 19 July 2015 Received in revised form 20 December 2015 Accepted 10 January 2016 Available online xxxx
A B S T R A C T We quantify all statements by major European politicians reported by Reuters during the August 2011 to December 2011 period and show that political communication significantly affects the EUR–USD exchange rate as well as European stock and bond markets. Communication with respect to Italy induces the strongest market reactions. Financial markets consider the German bond market a safe haven. © 2016 Elsevier B.V. All rights reserved.
JEL classification: E43 E62 G01 G12 C20 Keywords: Political statements High-frequency response Austerity measures Joint liability
1. Introduction We analyze whether the communication of European politicians with respect to the sovereign debt crisis affects the EUR– USD exchange rate as well as the European stock and bond markets. For the August to December 2011 period we quantify all statements by major European politicians that refer to the debt crisis and are reported by the news agency Reuters. We then explain the changes in the EUR–USD exchange rate as well as eight national stock and bond markets in the 15 minutes period following each statement by the content of the communication. The empirical results show that positive communication regarding the periphery countries or the Eurozone (EZ) as a whole leads to a significant appreciation of the EURO against the US dollar. More specifically, the exchange rate reaction is the strongest with respect to communication regarding the periphery countries and, in particular, in response to negative statements.
✩ This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 3202782013-2015. * Corresponding author at: Department of Economics, Heidelberg University, Bergheimer Strasse 58, 69115Heidelberg, Germany. Tel.: +49 6221 54 3173. E-mail addresses: [email protected] (C. Conrad), [email protected] (K. Zumbach).
http://dx.doi.org/10.1016/j.jempfin.2016.01.018 0927-5398/© 2016 Elsevier B.V. All rights reserved.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 2
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Further, the stock markets of the core European countries (Germany, France and Belgium) as well as the periphery countries (Greece, Ireland, Italy, Portugal and Spain) instantaneously increase in response to positive communication regarding the stance of the economy in the periphery countries or positive communication with respect to the EZ as a whole. In sharp contrast, the response of the national bond markets is asymmetric, whereby mainly the Italian and German bond markets are affected. While negative communication regarding the economic situation in Italy leads to an immediate increase in 10-year Italian government bond yields, German government bond yields decrease. This finding is in line with the view that financial market participants consider the German bond market a safe haven and highlights that Italy plays a pivotal role among the periphery countries. During our sample period, Silvio Berlusconi’s Italian government was under strong political pressure and finally had to resign in November 2011. Moreover, our results show that political communication concerning the periphery countries evokes stronger market reactions than statements on the EZ. While statements that suggest an expansion of the European Financial Stability Facility (EFSF) or shared liability for national debts do not lead to decreasing bond yields in the periphery countries, communication about the introduction of further austerity measures does reduce Italian yields. Our results extend upon and complement the recent work of Mohl and Sondermann (2013) and Beetsma et al. (2013) on the effects of political communication on sovereign bond spreads. In contrast to Mohl and Sondermann (2013) we distinguish between positive and negative communications. In addition, in contrast to Beetsma et al. (2013) we obtain bond market response coefficients for each individual country. Most importantly, while both studies employ daily data, we take a highfrequency perspective. Using high-frequency data allows us to monitor the effects of political communication on financial markets in real time and bypass problems with respect to identification and causality. In particular, we do not have to worry about potential control variables. Other related papers are Smeets and Zimmermann (2013), Ehrmann et al. (2014) and Kilponen et al. (2015). Smeets and Zimmermann (2013) investigate the effects of crisis meetings of European heads of state and government on stock and bond markets, while Ehrmann et al. (2014) study the determinants of the volatility of the EUR-USD exchange rate during the crisis. However, both studies again focus on daily data. Finally, Kilponen et al. (2015) focus on the effects of policy announcements on bond markets. Thus, all these studies can be seen as complementary to ours. 2. Data and methodology Our data set starts on August 01, 2011 and ends on December 06, 2011, i.e. it covers 92 trading days. We consider statements and the corresponding asset price changes lying within the trading hours from 9:00 a.m. to 5:30 p.m. Central European Time (CET). Although our sample period is relatively short, we believe that it allows for some unique insights from a subperiod of the sovereign debt crisis that was characterized by the severe uncertainties around the Italian government under Silvio Berlusconi. 2.1. Quantifying statements by European policy makers We search all news reports from Reuters for statements by the EZ’s 17 Heads of Government, their respective Finance Ministers, and the four leading EU representatives—the President of the EU Commission José Manuel Barroso, the President of the Euro Group Jean-Claude Juncker, the EU Economics Commissioner Olli Rehn, and the President of the European Council Herman Van Rompuy. We separate the statements into two groups. In the first group, we collect all statements that refer to the economic situation or austerity measures in the following periphery countries: Italy (IT), Spain (ES), Portugal (PT), Ireland (IE) and Greece (GR). We code each statement, CtP , as +1 if the statement implies a positive outlook for the specific country or the introduction of new austerity measures, and as −1 otherwise.1 For example, the Reuters report: “Italian PM Berlusconi says new agreement on austerity package confirms solidity of ruling coalition” on August 30, 2011 at 12:44 is coded as CtP = +1. The second group contains all statements that refer to the EZ as a whole, in particular to the EFSF, Eurobonds, the role of the ECB or the EURO as a currency. We code statements, CtEZ , as +1 when they – broadly speaking – suggest a shared liability for national debts within the EZ, e.g., statements that support the introduction of Eurobonds or the expansion of the EFSF. Statements with content opposed to such ideas are coded as −1. For example, the Reuters report: “Germany’s Merkel says Eurobonds are absolutely wrong” on September 15, 2011 at 10:47 is coded as CtEZ = −1. We disregard all statements that are either neutral or do not portray a clear message. Overall, Reuters reported 778 statements of which 164 were unanimously quantifiable.2 Table 1 provides a summary of the statements. Among the 164 statements, 77 (87) refer to the periphery countries (the EZ as a whole). 101 (or 62%) of the statements are coded as being positive (+1). Interestingly, for the EZ positive and negative statements are almost balanced. On the contrary, 75% of the statements referring to the periphery countries are positive. Since these countries are at the center of the debt crisis, the high number of positive statements might be surprising at first sight. However, this can be explained by the fact that a large share of these statements refers to the introduction of new austerity measures. Finally, among the statements that refer to the periphery countries the majority are related to Italy (37) and Greece
1 This approach of coding statements is commonly referred to as Content Analysis and has been used by, among others, Ehrmann and Fratzscher (2007) and Conrad and Lamla (2010) for analyzing the effects of central bank communication. 2 Although this means that only 21% of the statements are used in the empirical analysis, we thereby avoid problems of potential misclassification. Further, we excluded multiple communications that were released at the same time or communications that we pure announcements of, e.g, a further summit.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
3
Table 1 Summary of political statements. Positive Periphery (CtP )
IT GR ES IE PT Total
Eurozone (CtEZ ) Total
24 23 4 4 3
Negative 65% 82% 100% 100% 100%
13 5 0 0 0
Total 35% 18% 0% 0% 0%
37 28 4 4 3
58
75%
19
25%
77
43
49%
44
51%
87
101
62%
63
38%
164
Notes: The table shows the number and the tone (positive/negative) of the statements that refer to the periphery countries and the EZ, respectively.
(28). This reflects the fact that during our sample period the political debate was very much focused on these two countries. In contrast, Spain was not yet under such scrutiny. 2.2. Financial data For the stock market indices of three core EZ countries (DAX (Germany), CAC (France), BEL20 (Belgium)) and the five periphery countries (FTSE MIB (Italy), IBEX (Spain), PSI20 (Portugal), ISEQ (Ireland) and ASE (Greece)), we calculate 15-minute returns as rt,t+15 = 100×(ln(Pt+15 )−ln(Pt )). Similarly, we calculate the 15-minute change in the 10-year government bond yields of the respective countries, denoted by it,t+15 . Finally, we consider the returns, ext,t+15 , on the EUR–USD exchange rate. All financial data were obtained from Bloomberg. 2.3. Econometric methodology In order to measure the high-frequency response of the European capital markets to political communication, we follow Almeida et al. (1998) and regress the 15-minute change in the respective financial market, yt,t+15 ∈ {rt,t+15 , it,t+15 , ext,t+15 }, on the political communication Ct ∈ CtP , CtEZ : yt,t+15 = b0 + b1 Ct + et,t+15
(1)
Note that Eq. (1) is not a time series regression. Instead, each observation reflects the respective capital market’s 15-minute adjustment in response to the preceding communication.3 Finally, we also consider regressions in which we split the statements into positive and negative ones: Neg
yt,t+15 = b1 CtPos + b2 Ct
+ et,t+15 .
(2)
This allows us to test for asymmetric response coefficients. 3. Empirical results Table 2 provides a summary of the empirical results. As Panel A shows, political communication has a weak but significant influence on the EUR–USD exchange rate. Positive communication leads to a significant appreciation of the EURO within the 15 minutes following the statement. More specifically, positive statements about the periphery countries, CtP , lead to an appreciation of the Euro of about 0.06% in the 15 minutes after the statement. The statements explain a remarkable 21% of the total change during that 15-minute period. Panel B shows how the national stock markets respond to communication regarding i) the periphery countries, CtP , and ii) the EZ, CtEZ . First, all stock markets significantly increase in response to positive communication regarding the periphery countries. The size of the response is strongest in Germany and Italy with an R2 of 0.32 and 0.29, respectively. The R2 for Portugal, Ireland and Greece are considerably lower. Second, statements in favor of a shared liability for national debts within the EZ lead to significantly positive returns in all stock markets except Greece. Note that in all stock markets the response to CtEZ is weaker than the response to CtP .
3 More precisely, the dependent variable is observed every 15 min. The communication falls somewhere into the 15-minute interval, so that the response is measured at most for a period of 15 min and at least for a period of 1 min.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 4
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Table 2 Financial market response to political communication. Ct
# of obs.
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtP
77
CtEZ
R2 87
0.0563∗∗∗ (0.0123) 0.21 0.0322∗∗∗ (0.0082) 0.15
R2
Core countries GE
CtP
77
CtEZ
R2 87 R2
CtP
77 2
CtEZ
R 87 R2
Periphery countries FR
BE
IT
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2282∗∗∗ 0.2160∗∗∗ 0.1630∗∗∗ 0.2307∗∗∗ (0.0357) (0.0428) (0.0329) (0.0470) 0.32 0.29 0.26 0.29 0.1784∗∗∗ 0.1478∗∗∗ 0.0930∗∗∗ 0.1537∗∗∗ (0.0273) (0.0240) (0.0225) (0.0323) 0.33 0.31 0.17 0.21
ES
PT
IE
GR
IBEX 0.1785∗∗∗ (0.0398) 0.25 0.1321∗∗∗ (0.0268) 0.22
PSI20 0.1167∗∗∗ (0.0275) 0.18 0.0889∗∗∗ (0.0255) 0.13
ISEQ 0.0923∗∗∗ (0.0334) 0.14 0.0735∗∗∗ (0.0233) 0.10
ASE1 0.1120∗∗ (0.0472) 0.06 −0.0050 (0.0546) 0.00
0.1151 (0.1552) 0.00 −0.2773 (0.3229) 0.01
8.2820 (8.6568) 0.10 −1.4056 (4.0341) 0.00
−1.1924 (0.8722) 0.01 0.3810 (0.7649) 0.00
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.6929∗∗∗ 0.3608∗∗ 1.4409 −0.8470∗∗∗ −0.1863 (0.1523) (0.1731) (1.0245) (0.2289) (0.1867) 0.25 0.06 0.07 0.12 0.01 0.2376∗∗ 0.1623 −0.5044 −0.0761 −0.1829 (0.1110) (0.1247) (0.8002) (0.1605) (0.1672) 0.05 0.02 0.00 0.00 0.01
Notes: The table presents the estimates of b1 in Eq. (1). Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels. ∗∗ Indicates significance at the 5% levels. ∗ Indicates significance at the 10% levels. 1 Due to the Greek trading hours (930 –1600 CET), the Greek stock market regressions are based on 67/71 statements respectively. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based on 36/53 statements.
Panel C presents how political communication impacts on European bond markets. In general, the R2 s in the bond market regressions are lower than the corresponding ones in Panel B. We observe significant reactions to communication concerning the periphery countries in the German, French and Italian bond markets only.4 The sign of the response of the German and French bond markets is positive, while the sign of the response of the Italian bond market is negative. More precisely, bad news regarding future economic development in one of the periphery countries leads to a rise in the 10-year Italian government bond yield by 0.85 basis points while the German yield decreases by 0.69 basis points. This bond market response is in line with the interpretation that financial market participants view the German bond market as a safe haven. The fact that Italy is the only periphery country for which we observe a significant market response may be explained by the observation that due to its financial and political situation, the country was under the spotlight in financial markets during our sample period. In comparison to Greece, Ireland, and Portugal, the Italian debt level has been too high to allow the country to be taken under the European rescue umbrella. Therefore, Italy’s debt problem was considered to be the crucial issue in solving the European debt crisis. Since Italy plays such a pivotal role, we analyze the response in capital markets to communication with respect to Italy in more detail in Table 3. The result that the 10-year Greek government bond yields do not react to statements regarding the periphery countries (although a large amount of these statements refer specifically to Greece, see Table 1) may be explained by financial market participants who do not consider the rescue measures taken as being sufficient to solve the Greek debt problem. Interestingly, only the German bond market significantly reacts to communication regarding the EFSF, Eurobonds, the ECB or the EURO. Statements that are in favor of an expansion of the EFSF, for example, lead to rising German bond yields. The finding that the periphery countries’ government bond yields do not react at all to such statements is remarkable. It implies that the periphery countries do not ‘benefit’ from such statements. We also split the statements into positive and negative communication in order to check whether the effect is asymmetric. The results are presented in Table 4 in the Appendix (we present the coefficient estimates only). Most importantly, negative
4 Since Bloomberg reported the 10-year Irish government bond yields until October 11, 2011 only, the corresponding regressions are based on a shorter time period and, hence, include only 36/53 observations.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
5
Table 3 Financial market response to statements regarding Italy. Ct
# of obs.
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtI
37
0.0698∗∗∗ (0.0159) 0.38
R2
Core countries GE
CtI
37 R2
CtI
37 2
R
Periphery countries FR
BE
IT
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2361∗∗∗ 0.2485∗∗∗ 0.2007∗∗∗ 0.2857∗∗∗ (0.0445) (0.0572) (0.0404) (0.0609) 0.45 0.41 0.44 0.42
ES
PT
IE
GR
IBEX 0.2125∗∗∗ (0.0540) 0.35
PSI20 0.1155∗∗∗ (0.0307) 0.36
ISEQ 0.1403∗∗∗ (0.0305) 0.43
ASE1 0.0897 (0.0815) 0.03
0.4237 (0.3373) 0.02
20.8583 (19.0324) 0.26
−0.5587 (1.0818) 0.01
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.9595∗∗∗ 0.5702∗∗ 2.0974 −1.4918∗∗∗ −0.4827* (0.1992) (0.2195) (1.4842) (0.3325) (0.2580) 0.44 0.16 0.09 0.30 0.08
Notes: The table presents the estimates of b1 in the regression yt+15 = b0 + b1 CtI + et,t+15 , where yt,t +15 ∈ {rt,t +15 , it,t +15 , ext,t +15 }. Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels, respectively. ∗∗ Indicates significance at the 5% levels, respectively. ∗ Indicates significance at the 10% levels, respectively. 1 The Greek stock market regression is based on 28 statements. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based is 11 statements.
statements appear to have stronger effects on the exchange rate than positive ones. For example, a negative statement on P,Neg the periphery countries, Ct , leads to an immediate depreciation of 0.07% while the exchange rate appreciates in response to a positive statement, CtP,Pos , by 0.04% only. In contrast, as the table shows, at least for the stock market there seems to be no significantly different response to positive and negative news. In addition, we experimented with splitting the statements according to the speakers, e.g. prime minister vs. finance minister. However, for such an analysis our dataset does not comprise a sufficient number of observations. Finally, Table 3 presents how financial markets respond to communication regarding Italy, CtI . Recall that 37 out of the 77 statements regarding the periphery countries are concerned with Italy. In comparison to Table 2, the estimated coefficients typically suggest a stronger market response and a higher R2 . Interestingly, the R2 for the exchange rate regression now increases to 38% and the response of the EUR–USD is about 0.07%. Thus, the EUR–USD exchange rate reacts more strongly to communication concerning Italy than to communication about the periphery in general. Similarly, the Italian stock market reacts most strongly to positive communication about Italy. More precisely, after a positive statement the FTSE MIB increases on average by 0.29% in the ensuing 15 minute period. In addition to the German, French and Italian bond markets, the Spanish bond market now shows a significant (at the 10% level) reaction as well. In response to positive communication the government bond yields of the core (periphery) countries increase (decline). For example, if a major European politician communicates that Italy will introduce new austerity measures, the Italian government bond yield decreases by 1.49 basis points. Overall, our results highlight the pivotal role that Italy plays among the periphery countries and confirm the safe haven interpretation of the German bond market. 4. Conclusion Our results show that the communication of major European politicians during the sovereign debt crisis has a significant impact on the European financial markets. We find that statements regarding specific periphery countries evoke stronger market responses than statements focused on the EZ as a whole. More specifically, the EUR–USD exchange appreciates more strongly in response to positive statements about the periphery countries than to statements about the EZ as a whole. Most importantly, negative statements regarding the periphery countries trigger the strongest response of the exchange rate. Further, while positive statements of either type lead to significant increases in the stock markets of all countries under analysis, only the German, French and Italian bond markets are affected by political communication. The bond market reactions imply that investors consider the German governments bonds to be a safe haven. Italian government bond yields decrease if political communication hints at improvements in the state of the economy or the introduction of new austerity measures, but they do not react to statements in favor of an expansion of the EFSF or the introduction of Eurobonds. Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018
ARTICLE IN PRESS 6
C. Conrad, K. Zumbach / Journal of Empirical Finance xxx (2016) xxx–xxx
Appendix A
Table 4 Financial market response to political communication. Ct
Panel A: exchange rate (dependent variable: ext,t +15 , change in %) EUR/USD
CtP,Pos
0.0395∗∗∗ (0.0126) −0.0730∗∗∗ (0.0211) 0.0234∗∗ (0.0106) −0.0411∗∗∗ (0.0126)
P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
Core countries GE
CtP,Pos P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
CtP,Pos P,Neg
Ct
CtEZ,Pos EZ,Neg
Ct
Periphery countries FR
BE
IT
ES
PT
IE
GR
Panel B: stock markets (dependent variable: rt,t +15 , change in %) DAX CAC Bel20 FTSE MIB 0.2045∗∗∗ 0.1754∗∗∗ 0.1492∗∗∗ 0.1751∗∗∗ (0.0392) (0.0362) (0.0305) (0.0392) −0.2520∗∗∗ −0.2565∗∗∗ −0.1765∗∗∗ −0.2862∗∗∗ (0.0597) (0.0776) (0.0583) (0.0855) 0.1828∗∗∗ 0.1504∗∗∗ 0.0819∗∗∗ 0.1424∗∗∗ (0.0367) (0.0346) (0.0308) (0.0454) −0.1741∗∗∗ −0.1452∗∗∗ −0.1042∗∗∗ −0.1649∗∗∗ (0.0404) (0.0332) (0.0327) (0.0460)
IBEX 0.1283∗∗∗ (0.0332) −0.2287∗∗∗ (0.0724) 0.1314∗∗∗ (0.0394) −0.1329∗∗∗ (0.0364)
PSI20 0.1089∗∗∗ (0.0290) −0.1246∗∗∗ (0.0467) 0.0865∗∗ (0.0416) −0.0912∗∗∗ (0.0296)
ISEQ 0.0734∗∗∗ (0.0218) −0.1112* (0.0631) 0.0707∗∗ (0.0295) −0.0763∗∗ (0.0360)
ASE1 0.1156* (0.0601) −0.1083 (0.0728) −0.0438 (0.0757) −0.0339 (0.0788)
Panel C: bond markets (dependent variable: it,t +15 , change in basis points) 10-Year government bonds2 0.5069∗∗∗ 0.2638 0.2397* −0.5414* (0.1278) (0.1619) (0.1312) (0.2832) −0.8789∗∗∗ −0.4579 −2.6421 1.1526∗∗∗ (0.2765) (0.3059) (2.0448) (0.3596) 0.2093 0.1837 −1.8860 −0.2000 (0.1519) (0.1576) (1.2997) (0.2104) −0.2659 −0.1409 −0.8773 −0.0477 (0.1618) (0.1934) (0.9339) (0.2424)
−0.1672 (0.1856) 0.2053 (0.3241) −0.2977 (0.2101) 0.0682 (0.2601)
0.2776 (0.2818) 0.0474 (0.1303) −0.3023 (0.5851) 0.2523 (0.2734)
−1.1931* (0.6657) −17.757 (17.301) −8.7613 (5.6828) −5.9500 (5.7272)
− 1.7638 (1.4162) 0.6211 (1.0183) 1.1302* (0.6372) 0.3682 (1.3907)
Notes: The table presents the response coefficients to positive and negative statements separately. Heteroskedasticity-robust standard errors are reported in parenthesis. ∗∗∗ Indicates significance at the 1% levels. ∗∗ Indicates significance at the 5% levels. ∗ Indicates significance at the 10% levels. 1 Due to the Greek trading hours (930 –1600 CET), the Greek stock market regressions are based on 67/71 statements respectively. 2 Since Bloomberg reported the yields on 10-year Irish government bonds until October 11, 2011 only, the corresponding regressions are based on 36/53 statements.
References Almeida, A., Payne, R., Goodhart, C., 1998. The effects of macroeconomic news on high frequency exchange rate behaviour. J. Financ. Quant. Anal. 33, 383–408. Beetsma, R., Giuliodori, M., de Jong, F., Widijanto, D., 2013. Spread the news: the impact of news on the European sovereign bond markets during the crisis. J. Int. Money Financ. 34, 83–101. Conrad, C., Lamla, M.J., 2010. The high-frequency response of the EUR–USD exchange rate to ECB communication. J. Money Credit Bank. 42, 1391–1417. Ehrmann, M., Fratzscher, M., 2007. Communication by central bank committee members: different strategies, same effectiveness?. J. Money Credit Bank. 39, 509–541. Ehrmann, M., Osbat, C., Stráský, J., Uusküla, L., 2014. The euro exchange rate during the European sovereign debt crisis dancing to its own tune?. J. Int. Money Financ. 49 (Part B), 319–339. Kilponen, J., Laakkonen, H., Vilmunen, J., 2015. Sovereign risk, european crisis-resolution policies, and bond spreads. Int. J. Cent. Bank. 11, 285–323. Mohl, P., Sondermann, D., 2013. Has political communication impacted bond spreads?. Appl. Econ. Lett. 20, 48–61. Smeets, D., Zimmermann, M., 2013. Did the EU summits succeed in convincing the markets during the recent crisis?. J. Common Mark. Stud. 51, 1158–1177.
Please cite this article as: C. Conrad, K. Zumbach, The effect of political communication on European financial markets during the sovereign debt crisis, Journal of Empirical Finance (2016), http://dx.doi.org/10.1016/j.jempfin.2016.01.018