Communications Breakdown: The Transmission of Dierent types of ECB Policy Announcements

Size: px

Start display at page:

Download "Communications Breakdown: The Transmission of Dierent types of ECB Policy Announcements"

Rodney Summers
5 years ago
Views:

1 Communications Breakdown: The Transmission of Dierent types of ECB Policy Announcements Andrew Kane, John H. Rogers and Bo Sun April 27, / 27

2 Background I Large literature using high-frequency changes in bond prices around Central Bank announcements to capture monetary policy surprises I thus useful in identifying monetary policy shocks I Kuttner (21), Cochrane and Piazzesi (22), Gurkayanak, Sack and Swanson (25), Gertler and Karadi (215) I However, information on the economy released during these announcements complicates this interpretation. I \Fed information eect" I Romer and Romer (2), Nakamura and Steinsson (217), Campbell et. al. (21, 212) I forecasted (and perhaps realized) GDP response could be perverse I We explore theses issues for ECB announcements. 2 / 27

3 Outline of Our Paper I How are dierent types of ECB policy shocks transmitted? I Is there a central bank information eect, and if so how to identify it? I Where does it appear? I Asset Prices? I Private sector forecasts of GDP, ination? I Response to policy shocks? I What is its source? I Changes in the policy rate? I The statement? I Other communication? I Is the information eect related to uncertainty about monetary policy? 3 / 27

4 Identication of ECB Monetary Policy Shocks Periphery-Core bond yield spread our instrument I \Preserve the euro" objective Target Rate Window I Short statement released at conclusion of GC meeting I Decision on immediate monetary policy Communications Window I Press conference after statement release I Forward guidance, economic outlook 4 / 27

5 Figure: ECB Communication and Macroeconomic Data Forecast Timeline Target Rate Window Communications Window 1:3 2:25 3:2 Decision Released Press Conference Begins Private Forecast Made ECB Meeting (their outlook revealed) Data Release 5 / 27

6 Monetary Rubric policy Introductory statement of the March Governing Council Meeting Forward guidance on ECB interest rates: We continue to expect them [ECB interest rates] to remain at their present levels for an extended period of time, and well past the horizon of out net asset purchases. Net purchases: We confirm that our net asset purchases, at the current monthly pace of 3 billion, are intended to run until the end of September 218, or beyond, if necessary, and in any case until the Governing Council sees a sustained adjustment in the path of inflation consistent with its inflation aim. Reinvestment: The Eurosystem will continue to reinvest the principal payments from maturing securities purchased under the asset purchase programme for an extended period of time after the end of its net asset purchases, and in any case for as long as necessary / 27

7 Figure: Intraday Italian-German Spread Movements Italy Germany 2 Year Spread High Frequency Change Around Various ECB Communications Surprise Value Target Meeting Communication Intermeeting Communication Target Window: 13:3 14:25 Communication Window: 14:25 15:3 OMT introduced SMP for Italy and Spain / 27

8 Information Surprises I Examine changes in (i) actual data (GDP growth, 1-yr ahead), (ii) private sector expectations, and (iii) ECB outlook around ECB announcements. I Regress private forecast error on ECB forecast error I Residual captures what the central bank knew about the economy that the private sector did not: (data privateforecast ;t ) = + 1 (data ecbforecast ;t )+! t I : Release date of series being forecasted I t: Date of ECB forecast I t : Date of private forecast before ECB forecast at t I!: Information surprise. Negative values indicate CB pessimism surprise 8 / 27

9 Figure: Private Sector Forecast Change vs ECB Forecast Change GDP nowcasts GDP 1-year ahead forecasts Private sector Correlation: Private sector Correlation:.44 Changes in individual private sector forecasts vs changes in ECB forecasts. For each ECB meeting, the private sector change in the forecast around that meeting is the dierence between the rst forecast survey after that meeting and the last forecast survey before the meeting. Private forecast changes were computed monthly. ECB forecast changes were computed quarterly. 9 / 27

10 Figure: Information Surprises Information surprise / 27

11 \Preserve the euro" eect? High-frequency responses of exchange rate and bond yield spreads relative to Germany I narrowing spreads coincide with euro appreciation, especially in communications window I eect stronger for "more peripheral" countries I opposite relationship for Fed policy changes and the dollar 1 / 27

12 Figure: Change in 2-Yr IT-DE Spread and Exchange Rate Target Rate Window Communications Window Change in spread Change in USD/EUR Change in spread Change in USD/EUR / 27

13 Figure: Change in 2-Yr Interest Rate Spread and Exchange Rate Italy Ireland Change in spread Change in USD/EUR Change in spread Change in USD/EUR.29 France Austria Change in spread Change in USD/EUR Change in spread Change in USD/EUR.6 12 / 27

14 Estimate external instruments VARs I VAR in monthly data: A(L)Y t = " t I Y t = IP; CP I; spread; dollar=euro; ebp I 28:1 to 217:4 I Errors: " t = R t ; t = ( 1t ; 2t ) : I Dene Z t as intraday change in, e.g., two-year futures during communications window I set X t = " t for variables with only monthly data (do with macro data) I Implementation: estimate reduced form VAR and get residuals. Regress X t on Z t to get R 1 up to a sign and scale 13 / 27

15 External instruments identication, cont'd I Key Assumps: E( 1t Z t ) =, E( 2t Z t ) = I Identifying assumptions seem mild I no timing restrictions, dubious or otherwise, no sign restrictions I however, FRB-CH brethren, Nakamura-Steinsson, and others raise concerns I \Fed" information eect I suggests violation of 2nd assumption 14 / 27

16 Decomposing Monetary Policy Surprises I Regress raw high-frequency surprises on ECB and private sector forecast errors MP S m;t = + 1 ecbforecasterror ;t + 2 privateforecasterror ;t + t I m either target window or communications window I restrict 1 = 2 I private forecasts those made before ECB meeting I tted values are the Delphic Guidance surprises I residuals t are orthogonalized MP surprises 15 / 27

17 Figure: Orthogonalized Monetary Policy Shocks and Delphic Guidance Shocks Orthogonalized Target Rate Orthogonalized Communication Change in Spread Change in Spread Date Date Delphic Target Rate Delphic Communication Change in Spread Date Change in Spread Date 16 / 27

18 Figure: Percentage of \Perverse" outlook changes Target Rate Surprises Percent of forecasters with Delphic responses Communications Surprises Percent of forecasters with Delphic responses / 27

19 Figure: Impulse Responses Target Rate Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Communications Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Orthogonalized Target Rate Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Orthogonalized Communications Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread / 27

20 Figure: Impulse Responses where 1 = 2 Target Rate Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Communications Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Orthogonalized Target Rate Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread Orthogonalized Communications Shock IP CPI Italian German 2 Year Spread Dollar/Euro GM Credit Spread / 27

21 Figure 2: ImpulseFigure: responses to one Jarocinski standard deviationkaradi shocks, baseline (218) VAR. Median (line), percentiles (darker band), percentiles 5-95 (lighter band). surprise in 3m FF futures surprise in S&P5 1y govt. bond yield (%) S&P5 (1 x log) Real GDP (1 x log) GDP deflator (1 x log) A. Sign restrictions B. Standard HFI Monetary policy CB information Monetary policy (negative co-movement) (positive co-movement) (Choleski, 3m fff first) surprise in 3m FF futures surprise in S&P5 1y govt. bond yield (%) S&P5 (1 x log) Real GDP (1 x log) GDP deflator (1 x log) EBP (%).5.5 EBP (%) months months months 19 / 27

22 \Information eect" and monetary policy uncertainty Examine the relationship between Delphic FG shocks and measures of monetary policy uncertainty I market-based and news-based measures of uncertainty about ECB monetary policy I implied volatility of 3-month Euribor from estimating probability distribution functions using options prices. Variance of the estimated distribution for 3-month Euribor 18 days ahead. I news-based measures akin to Baker-Bloom-Davis and Husted-Rogers-Sun 2 / 27

23 Table: Monetary Policy Uncertainty Index Construction Country (i) (ii) (iii) Newspapers uncertainty monetary policy \Bank of Canada" Gazette uncertain interest rate \BOC" Globe and Mail Canada policy rate Ottawa Citizen overnight rate Toronto Star overnight lending rate Vancouver Sun uncertainty monetary policy \European Central Bank" Financial Times Euro uncertain interest rate \ECB" New York Times Area policy rate \Governing Council" Wall Street Journal renancing tender eonia uncertainty monetary policy \Bank of Japan" Financial Times Japan uncertain interest rate \BOJ" New York Times policy rate Wall Street Journal call rate uncertainty monetary policy \Bank of England" Daily Telegraph United uncertain interest rate \BOE" Financial Times Kingdom policy rate \Monetary Policy Committee" Guardian bank rate \MPC" Independent overnight lending rate Times of London Notes: Words in quotes are searched as exact terms. All other words searched over allow for plural forms. 2 / 27

24 35 Figure: Monetary Policy Uncertainty for the Eurozone Euro Crisis Liftoff Brexit US Election 3 25 Index (Avg = 1) Year 21 / 27

25 MPS Decomposition Estimates MP S m;t = + 1 (privateforecast ;t ecbforecast ;t )+ t I Target Rate window: 1 =.77, F-stat = (p=.4) I Communications window: 1 = -.28, F-stat =.542 (p=.467) 21 / 27

26 Figure: 3-Month Euribor Volatility vs Target Rate Surprises Orthogonalized Delphic Euribor 3 month volatility Euribor 3 month volatility Orthogonalized target rate MP surprises Delphic target rate MP surprises 22 / 27

27 Figure: 3-Month Euribor Volatility vs Communications Surprises Orthogonalized Delphic Euribor 3 month volatility Euribor 3 month volatility Orthogonalized communications MP surprises Delphic communications MP surprises 23 / 27

28 Figure: News-Based MPU vs Target Rate Surprises Orthogonalized Delphic Orthogonalized target rate MP surprises News Based MPU Delphic target rate MP surprises News Based MPU / 27

29 Figure: News-Based MPU vs Communications Surprises Orthogonalized Delphic Orthogonalized communications MP surprises News Based MPU Delphic communications MP surprises News Based MPU / 27

30 Conclusion I Analyze two types of ECB monetary policy shocks I ECB objective on periphery-core spreads, preserving the euro I Target rate shocks look conventional I Communication shocks suggestive of information eects I \Orthogonalized" shocks produce more conventional responses 26 / 27

Communications Breakdown: The Transmission of Different Types of ECB Policy Announcements

Communications Breakdown: The Transmission of Different Types of ECB Policy Announcements Andrew Kane Federal Reserve Board John Rogers Federal Reserve Board June 18, 18 Bo Sun Federal Reserve Board The