The use of projections and forecasts in the ECB s monetary policy AIECE Conference, Brussels 6 November 2007 Hans-Joachim Klöckers Director Economic Developments ECB The views expressed in this presentation are solely those of the presenter and do not necessarily reflect those of the European Central Bank 1
Overview 1 Introduction: some comments on the history 2 Publication of Eurosystem staff projections 3 The production of projections inside the Eurosystem 4 The link to monetary policy decisions 5 Eurosystem staff projection performance 6 Challenges ahead 7 Conclusion 2
1. Introduction: Forecast publication by Central Banks History: 20 years ago, forecasts played little (official) role in monetary policy; monetary and exchange rate targeting dominated. Over time, increase in communication and transparency, also largely related to move towards inflation targeting. Inflation targeting reacts to forecast: inflation forecast targeting. Growing number of central banks publishing forecasts: Bank of England (1993), Reserve Bank of New Zealand (RBNZ) (1996), Sveriges Riksbank (1997), Bank of Japan (2000). 3
1. Introduction: The EMI and the pre-ecb period Agreement in EMI: ESCB should, irrespective of its strategy, have at its disposal its own forecasts for inflation (and for other general economic developments) to assist in its policy decisions. In this context, a forecast of potential growth of the economy will be needed for defining a medium-term oriented monetary target or assessing the prospects for possible price pressures. It can be left to the ESCB to decide how much importance it wishes to attach to forecasts in the formulation of its policy. (Report February 1997, p. 13) Key argument: central bank needs to be forward looking, irrespective of monetary policy strategy. 4
1. Introduction: What does the ECB do? The ECB has produced forecasts (later renamed projections ) since the beginning (1998). Four forecast rounds per year: by the Eurosystem staff (ECB and NCBs) in June and December by the ECB staff in March and September Quote from January 1999 Monthly Bulletin: In parallel with the analysis of monetary growth.., a broadly based assessment of the outlook for price developments and the risks to price stability will play a major role in the Eurosystem s strategy. This assessment will be made using a wide range of economic indicators. The wide range of indicators will include many variables that have leading indicator properties for future price developments. These include.obviously, it will also be useful to look at inflation forecasts derived using all these variables In this respect, the Eurosystem will evaluate the full range of inflation forecasts produced by international organisations, other authorities, market participants etc., and will also produce its own assessment of the future inflation outlook. However, inflation forecasts will have to be interpreted with caution and care. etc. 5
1. Introduction: The ECB s monetary policy strategy (post evaluation 2003) Primary objective of price stability Economic analysis Analysis of economic dynamics and shocks Governing Council takes monetary policy decisions based on an overall assessment of the risks to price stability cross checking Monetary analysis Analysis of monetary trends Full set of information 6
2. Publication of Eurosystem staff projections Long controversy in the ECB on whether or not to publish forecasts. Advantages foreseen were: Enhance transparency; help the public to understand monetary policy decisions (given the role of internal forecast in the policymaking process); Emphasize key features of monetary policy: forward-looking orientation; 7
2. Publication of Eurosystem staff projections Concerns in publishing forecasts related to the impact on effectiveness and credibility of monetary policy: Concern that the publication of forecasts may obscure rather than clarify the presentation of policy decisions; Risk of misguiding private expectations if the published forecast signals risks to price stability (public may not understand conditionality); The unavoidable inaccuracy of forecasts, visible ex post, could risk undermining the credibility of the Eurosystem. 8
2. Publication of Eurosystem staff projections These fears have not materialised, thanks to appropriate communication: 1. Forecasts labeled projections to signal conditional nature (certain variables are based on technical assumptions); 2. Uncertainty of projections marked by use of ranges; 3. It was reconfirmed that projections are only one input into the decisions; 4. It was reconfirmed that Governing Council looks also at other factors; needs to understand nature of shocks; applies judgment when taking decisions; 5. It was reconfirmed that projections are prepared under the responsibility of staff, neither EB nor Governing Council are owners of projections. No evidence that public misunderstood nature of projection (or of adverse influence on inflation expectations). 9
2. Publication of Eurosystem/ECB staff projections Since December 2000, Eurosystem staff projections published twice a year (June / December). Quote from December 2000 Monthly Bulletin: As is usual at this time of year, several forecasts and projections have become available, including the projections by Eurosystem staff. The Eurosystem staff economic projections for the euro area are presented here for the first time. They are prepared twice a year as a technical input into the deliberations of the Governing Council. The staff projections are based upon a set of technical assumptions, including, inter alia, an assumption of unchanged short-term interest rates and exchange rates. Since mid 2004, projections are published four times a year (that is, also the ECB staff projections, in March / September): A further step in transparency; Also facilitating external communication (as reference to latest internal projection, and not artificially to the possibly outdated projection known to the public). 10
2. Publication of Eurosystem/ECB staff projections Publication focuses on selected key variables; example from September 2007 Monthly Bulletin: (average annual percentage changes) 2006 2007 2008 HICP 2.2 1.9-2.1 1.5-2.5 Real GDP 2.9 2.2-2.8 1.8-2.8 Private consumption 1.9 1.4-1.8 1.5-2.7 Government consumption 1.9 1.3-2.3 1.2-2.2 Gross fixed capital formation 5.2 3.6-5.2 1.4-4.6 Exports (goods and services) 8.2 4.6-7.4 3.9-7.1 Imports (goods and services) 7.9 3.7-6.7 3.6-7.0 Key technical assumptions are for fiscal policy (announced measures), oil prices (futures), exchange rates (random walk) and interest rates (constant / market rates). 11
2. Monetary policy assumption: from constant to market interest rates At the beginning: constant short-term interest rate Intuitively simple ( What happens if no policy change is made? ) Clearly technical assumption (no commitment; easy to explain why it is not followed) But: - Inconsistent method; - Difficulty to compare with other available forecasts; 12
2. Monetary policy assumption: from constant to market interest rates Since June 2006: market interest rates (based on Governing Council decision) Higher internal consistency of projections; Higher comparability with other forecasts; And: not a signal of future monetary policy intentions, (i.e. it remains a technical assumption). 13
2. Shall the projection be built on the central bank s preferred future policy path? Some central banks base published forecasts on own preference for future policy interest rate path (in New Zealand, Norway, Sweden, Iceland) ECB has seen no reason to do so; High level of uncertainty, notably at longer horizons; Risk that path is perceived as commitment; Not more transparent; What matters for longer-term interest rates is credibility of monetary policy (so as to ensure low and stable inflation expectations and low inflation risk premia). 14
2. Publication of Eurosystem staff projections Advantages of publication now well understood: Enhanced transparency (public sees an important input received by the Governing Council); Stressing the forward-looking nature of monetary policy. Monetary policy assumption has evolved over time, but no net benefits seen in moving to announcing path of future policy intentions. Appropriate communication helped to mitigate the early concerns related to publication and choice of monetary policy assumption (choice of a technical nature, without policy implications). 15
3. The production of projections inside the Eurosystem Eurosystem has specific needs 1. Special focus on inflation 2. Need to take into account monthly meeting frequency of decision-making body 16
3. Different projection exercises Quarterly exercises have two components: (Broad) Macroeconomic Projection Exercise (BMPE) looks at the macroeconomy. Quarterly projections at the 2-3 year horizon. Very broad exercise, also including fiscal and labour market. Narrow Inflation Projection Exercise (NIPE) looks at monthly movements in inflation 18-months ahead. It looks at all components of HICP. Mechanical update of both exercises at higher frequencies based on PUEs (projection update elasticities) (e.g. to update changes in oil prices, FX) and other information (e.g. weekly oil bulletin) Bridge and factor models for exploiting real-time information on GDP 17
3. Tools used Forecast is a mixture of models and expert judgement. Notably for (shorter-term) inflation projection: need to have detailed knowledge on institutional factors (administered prices, indirect taxes). Models used differ according to purpose. Large-scale macro-economic models used in BMPE: Area-wide (ECB). Multi-country (ECB). NCB models. Area-wide DSGE model is being phased in. 18
4. Link to monetary policy decisions Step 1: Projections need to be well communicated to the Governing Council Important to explain the difference between judgemental input and model in the baseline; Staff needs to build a clear story. Council needs to understand baseline to form its own assessment; Need to explain sources of revisions (New data, new assumptions or new judgment?); 19
4. Link to monetary policy decisions Step 2: Governing Council needs to consider other relevant forecast-related information Staff projections are compared with forecasts from other institutions, public and private. ECB conducts own Survey of Professional Forecasters. Cross-check with expectations embedded in financial market prices. News after the cut-off point of the projection are relevant. 20
4. Link to monetary policy decisions Step 3: Governing Council assesses risks around the projection Given high uncertainty about point forecasts, risk analysis plays a key role in central banks. At the ECB, published risk analysis reflects view of Governing Council. Issues typically discussed are: Will oil prices follow futures prices? Will exchange rates remain unchanged? Will there be changes to fiscal policy announcements? What are risks to the global growth projection? What are risks to wages? What are risks coming from housing markets? Important to distinguish risks to inflation projections from risks to price stability. 21
4. Link to monetary policy decisions Step 4: Monetary analysis Projections focus on short-to medium-term horizon Projections typically have problems embedding the information content from monetary analysis (money, credit, asset prices) Monetary analysis focuses on medium- to longer-term risks to price stability 22
4. Link to monetary policy decisions To sum up: Staff projections are one input into policy process; No mechanical link between staff projections and monetary policy decisions; Reasons: Other forecast-related information matters Risk analysis matters Monetary analysis matters Important to be fully transparent to GC so that it can make best-informed judgment on monetary policy GC does not need to produce its own forecast for taking policy decisions 23
5. Published projection ranges and outcomes for GDP - current year GDP growth current year projections 4.0 Realised value 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 December 2000 (2000) March 2001 (2001) June 2001 (2001) September 2001 (2001) December 2001 (2001) March 2002 (2002) June 2002 (2002) September 2002 (2002) December 2002 (2002) March 2003 (2003) June 2003 (2003) September 2003 (2003) December 2003 (2003) March 2004 (2004) June 2004 (2004) September 2004 (2004) December 2004 (2004) March 2005 (2005) June 2005 (2005) September 2005 (2005) December 2005 (2005) March 2006 (2006) June 2006 (2006) September 2006 (2006) December 2006 (2006) Note: Years in parenthesis specify the year which was projected. Source: ECB 24
5. Published projection ranges and outcomes for GDP - one year ahead GDP growth one year ahead projections 4.0 Realised value 3.0 2.0 1.0 0.0 December 2000 (2001) March 2001 (2002) June 2001 (2002) September 2001 (2002) December 2001 (2002) March 2002 (2003) June 2002 (2003) September 2002 (2003) December 2002 (2003) March 2003 (2004) June 2003 (2004) September 2003 (2004) December 2003 (2004) March 2004 (2005) June 2004 (2005) September 2004 (2005) December 2004 (2005) March 2005 (2006) June 2005 (2006) September 2005 (2006) December 2005 (2006) Note: Years in parenthesis specify the year which was projected. Source: ECB 25
June 2006 (2006) September 2006 (2006) December 2006 (2006) 5. Published projection ranges and outcomes for HICP current year HICP inflation current year projections 3.0 Realised value 2.5 2.0 1.5 1.0 0.5 0.0 Note: Years in parenthesis specify the year which was projected. Source: ECB 26 December 2000 (2000) March 2001 (2001) June 2001 (2001) September 2001 (2001) December 2001 (2001) March 2002 (2002) June 2002 (2002) September 2002 (2002) December 2002 (2002) March 2003 (2003) June 2003 (2003) September 2003 (2003) December 2003 (2003) March 2004 (2004) June 2004 (2004) September 2004 (2004) December 2004 (2004) March 2005 (2005) June 2005 (2005) September 2005 (2005) December 2005 (2005) March 2006 (2006)
5. Published projection ranges and outcomes for HICP one year ahead HICP inflation one year ahead projections 3.0 Realised value 2.0 1.0 0.0 Note: Years in parenthesis specify the year which was projected. Source: ECB 27 December 2000 (2001) March 2001 (2002) June 2001 (2002) September 2001 (2002) December 2001 (2002) March 2002 (2003) June 2002 (2003) September 2002 (2003) December 2002 (2003) March 2003 (2004) June 2003 (2004) September 2003 (2004) December 2003 (2004) March 2004 (2005) June 2004 (2005) September 2004 (2005) December 2004 (2005) March 2005 (2006) June 2005 (2006) September 2005 (2006) December 2005 (2006)
5. Reasons for projection errors (I): Spot and futures prices of Brent crude oil (USD per barrel) Spot price Futures price 90.00 80.00 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Oil price assumptions for Eurosystem macroeconomic projections are derived as the average of future market prices over 10 days prior to the projections cut-off date. Source: ECB 28
5. Reasons for projection errors (II) Other possible candidates: Productivity/potential growth overestimated? Implications for GDP of 2001/02 financial turmoil underestimated? Inflation impulse from indirect taxes and administered prices underestimated? 29
6. Challenges ahead Can we find a better way to forecast oil prices? Can we improve the forecasting of fiscal behaviour? Can we better incorporate financial developments, balance sheet effects? Can we better incorporate asset prices and their implications? Methodological challenges: Introduction of DSGE models 30
7. Conclusion Forecasts now widely used in monetary policy around the world. Also key role in ECB (inside broader 2-pillar framework) However, important to keep in mind limitations: Risks, uncertainty - Monetary policy will never react mechanically to forecast only. 31