How ECB Announcements Affect Financial Markets

Ralf Fendel - 22. January 2021

Tips for practitioners

Economic Research is currently focusing on the effectiveness of the European Central Bank's (ECB) bond purchase programs. The channels of market reactions are very complex and the intended effects on economicactivity and ultimately on the inflationrate emerge only with a strong time lag. At the beginning of this so-called transmission mechanism are the effects on yields in the bondmarkets and, in particular, the markets for governmentbonds. It turns out that the mere announcement of bond purchases triggers reactions among market participants. Since reactions are expected to occur immediately and without delay, high-frequency data are increasingly used to measure yields on a minute-by-minute basis.

A recently published research paper (Fendel and Neugebauer, 2020) shows, however, that it is still appropriate and necessary to look at a lower frequency - such as daily data. At least for some asset classes, high-frequency identification may not be appropriate because market participants, and hence market prices, need some time to process monetary policy news. By means of an event study, such lags are explicitly identified. To this end, we first identified key announcement dates related to the ECB's purchase programs and then analyzed their impact on daily 10-year government bond yields for eleven euro area member countries. The following chart shows the development of yields and the most important announcement dates of the respective programs.

Delayed effects

When daily rather than high-frequency data are used, the majority of event studies published to date use a two-day window after an associated announcement (e.g., Altavilla et al., 2015; Szczerbowicz, 2015; Christensen and Krogstrup, 2019). However, this relatively large time window cannot explicitly identify potentially delayed announcement effects because it is unclear whether responses occur on the same announcement day (possibly also revealed by high-frequency data) or with a delay only on the following day, because both are covered by the two-day window. In contrast, Fendel and Neugebauer (2020) use a lagged one-day event window, i.e., they explicitly test the extent to which significant return effects occur only on the subsequent day of an announcement. Based on this lagged time window, statistically significant lagged reactions of individual euro area member countries' government bond yields to ECB announcements are identified. Only one day after the announcement there is a statistically significant decline in yields, while no impact is measurable on the announcement day itself. In another paper, the lagged effect is also confirmed for the entire yield curve for maturities up to 30 years (Fendel et al. (2020)).

The delayed reaction has not been explicitly documented before in the relevant literature, and it is surprising because financial market participants are often assumed to react immediately, given their information processing capabilities. However, two arguments explain the delayed effect. First, government bond trading - more than other securities - is subject to some transaction constraints. The bond market has a relatively larger volume and is less volatile compared with the stock market, for example. Trading in government bonds is more complex, as maturity, coupon payment and ratings play a role.

The second argument relates directly to the speed of information processing. Institutional investors such as pension funds usually trade in government bonds. They have a long-term strategic planning horizon and will therefore not adjust their portfolio very quickly (i.e. on the same day or even in just a few minutes) when new information is published by the ECB. For one thing, regulatory requirements prevent them from doing so. For example, a commercial bank holding government bonds as collateral must first find a replacement before liquidating this position. Second, the internal decision-making process within a large institution takes longer than for a small investment trust or a private individual investor, as such decisions are often made in investment committees and require formal meetings. In addition, government bonds are typically traded manually 'over the counter' on the trading floor by dealers - in stark contrast to automated transactions on a central exchange that are driven by computer algorithms. Overall, government bond yields will therefore react less quickly to news than, for example, stock prices, due to such decision and execution delays.

Interaction of Yield change and Credit Ratings

Fendel and Neugebauer (2020) also show that the extent of the yield reduction depends on the respective euro area country (as the corresponding debtor of the government bonds) and roughly corresponds to the familiar core/periphery pattern in the euro area. Overall, the yields of the peripheral countries (especially Greece, Spain, Italy Portugal) are affected three times as much as the yields of the core countries. However, the announcements work in the same direction for all euro area member countries: ECB purchase announcements lower euro area government bond yields. Therefore, the observed narrowing of the "spread" (i.e. interest rate differential) across economies in response to ECB announcements is due to a larger reduction in yields in the peripheral countries compared to the core countries (and here mainly Germany) and not to different directions of yield changes as is often presumed.

The following figure illustrates this relationship. The reaction coefficients estimated for individual countries or their government bond yields (for the one-day lagged effect) are compared with the credit rating of the country in question. A clear inverse relationship emerges.

Accordingly, the research shows that credit ratings can be linked into the consideration of the impact of monetary policy announcement effects. This is also supported in the study by the use of so-called interaction terms in the econometric analyses.

Literature reference

• Fendel, R., Neugebauer, F. (2020): Country-specific euro area government bond yield reactions to ECB’s non-standard monetary policy program announcements, in: German Economic Review (21/4), 417-474.
   
• Fendel, R., Neugebauer, F., Kilinc, M. (2020): ECB’s communication and the yield curve: core versus periphery effects, forthcoming in: Applied Economics Letters.