Analysts’ forecasts in financial market inflation expectations questionnaires: A look back at the past ten years

The CNB launched its survey of financial market inflation expectations in May 1999. This box describes the CNB’s ten-year experience of gathering and assessing financial market analysts’ expectations regarding key macroeconomic indicators.

The CNB requests forecasts for the following indicators:

• the annual change in the consumer price index at the one-year and three-year horizons;
• the values of the 2W repo rate, 1Y PRIBOR, 5Y IRS and 10Y IRS at the one-month and one-year horizons;
• the koruna-euro exchange rate at the one-month and one-year horizons;
• annual real GDP growth in the current year and the following year;
• annual nominal wage growth in the current year and the following year.

The period of 1999–2009 was characterised by relatively volatile monitored variables. The ongoing financial and economic crisis has also partly affected the survey results in recent months. Over the whole period under review, the forecasts for some variables fluctuated within relatively wide uncertainty bands, defined as the ranges between the maximum and minimum values, but the actual values of the individual indicators were often even more volatile than the forecasts. This is visible, for instance, in inflation. Its forecasted values were overestimated on average at both the one-year and three-year forecast horizons, but the analysts showed estimation errors in both the positive and negative directions. As regards monetary policy, it is important that changes in inflation expectations at the one-year horizon remain almost totally unreflected in changes in inflation expectations at the three-year horizon. Longer-term inflation expectations have stayed below 3% since the end of 2002 and have been converging gradually to the CNB’s new inflation target of 2% valid from the start of 2010. Despite the ongoing financial and economic crisis, the CNB’s new inflation target can be interpreted as credible.

The interest rate forecasts are quite significantly affected by current developments and are optically similar for all the monitored maturities. The forecasts at the shorter horizon logically fluctuate within narrower uncertainty bands. The average expected rates for the entire period under review at the one-month horizon are almost identical to the actual values, while the forecasted values at the one-year horizon are overestimated on average.

The exchange rate forecasts also changed in relation to the current situation. The forecasted values of the exchange rate were weaker on average than the actual ones at both monitored horizons. The appreciation of the koruna was thus even faster than the analysts expected.

Real GDP growth seems to be the most difficult-to-forecast indicator. The GDP forecasts were systematically underestimated, with the actual values fluctuating outside the ranges of the analysts’ forecasts in almost all the years under review. Moreover, the dispersion of the estimates tended to grow over time. An analysis of the published data reveals that the underestimations of GDP forecasts were most affected by CZSO revisions of the data, as they reflected statistical changes that could not have been predicted. By contrast, nominal wage growth can be forecasted very well. However, shorter time series are available for nominal wage growth forecasts.