Interest rates and other financial asset prices are based on expectations about economic developments. Asset prices react to new information. In this article, we explore the effects of news about key macroeconomic variables, external impulses, Norges Bank's interest rate decisions and the communication of monetary policy on Norwegian interest rates.
An overview of the days with the largest movements in markets rates since the beginning of 2001 shows that interest rates react strongly to certain news or surprises (see Table 1). These "large" changes, which all occurred in the period up to February 2004, reflected either surprising interest rate decisions, monetary policy signals in speeches or at monetary policy meetings, surprising CPI figures or in a few cases international interest rate movements. None of the observations on the list is from the past three years. Have market participants been exposed to fewer surprises over the past three years, or do they react less to surprising macroindicators or monetary policy signals than earlier? It is relevant to investigate this in the light of the changes in monetary policy frameworks in Norway and internationally through the period.
The monetary policy objective of most central banks is price stability. Monetary policy operates with a lag, i.e. it takes time for changes in official policy rates to influence real economic variables and prices, and central banks' interest-rate setting will therefore reflect the outlook for output, employment and inflation. Information about the current situation in the economy is an important source of information when assessing future prospects. Central banks and market participants follow current developments in macroindicators. If they contain new information about the outlook, this affects interest rate expectations. A number of international studies confirm that key figure releases influence interest rate expectations. US key figures have a particularly strong impact on US interest rate expectations, but also on interest rate expectations in other countries.
Market participants do not fully understand how central banks assess the economic situation or how they will react to new information. Market interest rate expectations are thus also influenced by actual interest rate decisions and by central banks' communication of the monetary policy strategy ahead. The monetary policy framework in Norway and many other countries has changed considerably. A common feature is the shift to greater monetary policy transparency. Mervyn King, Governor of the Bank of England, has described this as a development where "mystery and mystique have given way to transparency and openness". With greater openness about monetary policy the degree of asymmetric information between central banks and the public has been reduced. Monetary policy has become more predictable, which has reduced the uncertainty about future interest rate developments. In tandem with the emphasis on transparency, there has been an international tendency towards a more gradualist approach to interest rate setting (2), perhaps best illustrated by the Federal Reserve's rate hikes from June 2004 in 17 increments of 0.25 percentage point. These changes have contributed to reducing the volatility of short-term interest rates and market participants have been less surprised by central banks' interest rate decisions than earlier.
We examine how Norwegian interest rate expectations, as measured by implied forward rates, react to macroeconomic news and monetary policy decisions, and whether these relationships have changed as a result of the shift to greater transparency in the conduct of monetary policymaking. The analysis is based on data that we have collected about news and market reactions, an overview that includes the most relevant news since the beginning of 2001.
Section 2 provides a review of the relevant literature in the field. Our selection of data and modelling strategy is explained in section 3, while the findings are discussed in section 4. Section 5 concludes.
2 Existing literature
Several international studies have examined how interest rates and other asset prices react to the publication of key macroeconomic figures, central banks' interest rate decisions and communication.
An often cited finding in the litterature is that news about US key macroeconomic variables have a strong impact on financial data in the US and in other economies. Goldberg and Leonard (2003) find that news about the US labour market, GDP growth and consumer confidence influence US yields, while European key figures have little impact on US interest rates. In many cases, US key figures have a stronger impact on European interest rates than European key figures. Goldberg and Leonard argue that this phenomenon probably reflects a view among market participants that developments in the US are important for global growth and that the economic situation in different countries has become more synchronised. Moreover, the European Central Bank points out that US key figures are normally published earlier than European figures, giving the former the role of leading indicators for European financial markets (ECB, Monthly Bulletin, 2006).
Most studies analyse the effects of news on a single instrument, e.g. short-term forward rates or long-term forward rates. However, Fleming, Piazzesi and Remolona (2003) analyse the effect of macroeconomic news on the entire US yield curve. They find that strongest effects on interest rates in the maturity segment one to five years, with a peak at two to three years and declining thereafter. This has been referred to as the hump-shaped yield reaction with regard to term structure.
High-frequency data can be used to examine how quickly interest rates react to the release of key figures. Fleming and Remolona (1997) find that the most of the response is completed within two minutes. Most studies confirm that new information has a rapid effect on yields.
In recent years, central bankers and others have analysed how the link between information and yield reactions has changed as a result of the shift towards more independent and transparent central banks. Kohn and Sack (2003) find that for the US communication in connection with interest rate decisions and Congressional testimonies have a significant impact on US interest rate expectations, and that communication has a greater impact on interest rate expectations in the longer term than the actual interest rate decisions.
Conelly and Kohler (2004) investigate, among other things, how interest rates respond to communication by the central banks of Australia, Canada, the euro area, New Zealand, the UK and the US. They find that the predictability of actual interest rate decisions is about the same for all the countries. This indicates that the central banks are fairly similar in terms of communicating the monetary policy strategy ahead. They find that the main central bank communication channels are comments on interest rate decisions, monetary policy reports and testimonies before national parliaments.
Most studies assume that interest rate changes reflect changes in interest rate expectations and therefore disregard changes in risk premiums in markets. Using affine term-structure modelling on US rates, Beechey (2007) demonstrates that macroeconomic news announcements influence both forward rates and term premiums. At short horizons, changes in interest rate expectations account for most of the rate changes. At longer horizons, changes in term premiums account for most of the changes in forward rates.
3 Data and model
Our data for Norway comprises 1 637 daily observations between 1 January 2001 and 30 June 2007. For each day, the data set contains information about changes in Norwegian forward rates and any news released that day. News announcements include what is assumed to be the most important macroeconomic variables published monthly and all of Norges Bank monetary policy meetings. In addition, we have included the Governor's annual address and two additional speeches. (3) The data set is compiled using Norges Bank's ongoing internal reporting on market reactions to key macroeconomic variables, monetary policy meetings and other events of importance for market rates. We have also included euro-area interest rates as a representative of international news.
Key macroeconomic variables
Key macroeconomic variables include five variables for the Norwegian economy; the consumer price index adjusted for tax changes and excluding energy products (CPI-ATE), two unemployment measures, one figure for retail trade sales and a credit indicator. These are key figures that are published monthly and that market participants have been shown to monitor closely. The variables are further described in Table 2.
The news component (or surprise) of the release of key figures is calculated as the difference between actual outcomes and the anticipated value of the key aggregate. Expected value is set equal to the average survey-based market expectation, measured by expectations surveys. (4)
News components are standardised by dividing the difference between actual outcome and expected value by the series' standard deviation. (5) As a result, the series with the different key variables' surprises can be compared. Descriptive statistics for the key variables are shown in Table 3. In addition, a complete overview of all the deviations between expected and actual CPI-ATE through the period is provided in Chart 1.
In the period since the beginning of 2001, changes in the CPI-ATE and registered unemployment have on average been slightly lower than expected, while the changes in retail sales and C2 have been higher. Retail sales are considerably more volatile than the other key variables because the projections for retail sales are less accurate than for the other key variables.
In the time series for each key variable, the...