EVALUATION OF NORGES BANK'S PROJECTIONS.

• Author: Jore, Anne Sofie
• Position: Statistical Data Included

As a means to drawing up the most accurate projections possible for economic developments, Norges Bank regularly evaluates its model-based forecasts. Analyses of forecast errors may make an important contribution to improving projections. It is also desirable to compare Norges Bank's projections with those of other institutions.

The strength of the cyclical upswing was clearly underestimated in the projections for the period 1994-1996. Projections for employment growth were particularly low, also for the 1997 projections albeit to a lesser extent. On the other hand, forecasts for price and wage inflation were fairly accurate. The forecast errors were to a large extent ascribable to erroneous assumptions about economic policy, particularly growth in public demand. Petroleum investment was also much higher than projected. If the model-based projections are corrected for these factors, they are very close to the outturn, particularly for price and wage inflation. Over the past year, Norges Bank took steps to improve the accuracy of the exogenous variables used in the projections. Among other things, the estimate for public spending growth is supplemented by Norges Bank's own assessment of local government demand. Furthermore, data on oil-related activities are collected from a larger number of sources than earlier.

A preliminary analysis of forecast errors for 1998, based on figures from the national accounts figures published in February 1999, indicates that previous analyses of forecast errors may have improved the projections. Forecast errors seem to have been reduced in 1998 compared with the two previous years.

A comparison with projections from Statistics Norway and the Ministry of Finance for the period between 1994 and 1998 indicates that the degree of forecast errors from the three institutions has been fairly similar. On average, however, projections from Statistics Norway and Norges Bank have been more accurate than forecasts from the Ministry of Finance.

In this article, we focus on forecast errors stemming from erroneous estimates of economic policy and inaccurate projections for other exogenous variables. A more thorough analysis would also include a further disaggregation of errors stemming from the incorrect use of the model and those errors occurring due to model deficiencies. Such an analysis will be undertaken once the revision of the national accounts system that took place in the mid-1990s has been fully incorporated in the model data.

Introduction

Pursuant to the Norges Bank Act, Norges Bank is an executive and advisory body in the area of monetary and foreign exchange policy. Analyses of the macroeconomic situation, including forecasts for economic developments in the Norwegian and international economy, which are published in the Bank's quarterly Inflation Report, are an important basis for the formulation of monetary policy. In addition, the analyses are used as a basis for advice on the orientation of general economic policy.

Norges Bank aims to produce the best possible projections for the Norwegian economy. The macroeconomic model RIMINI, developed in Norges Bank's Research Department, has been the principal tool for Norges Bank's analyses since 1994. RIMINI is an econometric model with approximately 370 equations. About 70 of these equations are estimated on the basis of historical data, while the remaining equations are definitional relationships.

It must be possible to evaluate the outturn of Norges Bank's projections if they are to be credible over time. For this reason, Norges Bank has placed considerable emphasis on transparency and the availability of its forecast work, including adjustments of previous errors. Projections are based on a model that is publicly known, and the Bank's use of the model is published. The purpose is to provide others with the basis for evaluating how we have arrived at our projections. Systematic evaluation of the use of the model places greater demands on the Inflation Report and ensures that projections are not actively used in an attempt to influence the market in any way.

It is important that errors are revealed if projections are to improve and become more accurate. This will lay the foundation for better exogenous projections and for improving the model and the way the model is used. We also wish to compare Norges Bank's projections with those of other institutions in order to determine whether our projections are at least on a par with those of other institutions.

Analyses of Norges Bank's projections have been published previously. In an article published in Penger og Kreditt 1996/1 (Madsen 1996), Norges Bank's projections for the years 1987-1994 are compared with those of other institutions. An article published in Economic Buletin 1998/1 (Jore 1997) provides an analysis of Norges Bank's projections for the period 1994-1996 that focuses on the factors behind forecast errors in 1996. The article also provides some summary measures of forecast errors for the Ministry of Finance, Statistics Norway and Norges Bank, showing that these institutions' projections were almost equally accurate. The article also shows that the strength of the cyclical upswing was considerably underestimated by all the institutions.

The first section of this article briefly describes the main sources of forecast error. This is followed by an analysis of forecast errors in Norges Bank's projections for 1997. The evaluation was carded out using the same method employed in the evaluation of projections for 1996 (Jore 1997). Forecast errors in the projections for 1998 are examined briefly on the basis of preliminary national accounts figures published in February this year. We will publish a more thorough analysis of 1998 at a later date. The last section of the article presents an overview of projections for the entire period 1994-1998, comparing projections from Norges Bank with those of Statistics Norway and the Ministry of Finance.

...