Use of Survey Data for Industry, Research and Economic Policy

Selected Papers Presented at the 24th CIRET Conference, Wellington, New Zealand 1999

Karl Oppenlander, Gunter Poser, Bernd Schips

• Publisher: Taylor & Francis
• Year: 2018
• Print ISBN: 9781138726451
• eBook ISBN: 9781351752008
• Edition: 1
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

This title was first published in 2000: This text offers a comprehensive collection of selected papers from the 24th Centre for International Research on Economic Tendency Surveys (CIRET) conference. Areas selected include leading indicators and turning points, classifications of business cycles, survey data and policy decisions, attitudes and behaviour of firms, and economic forecasting. The text aims to be of interest to all those concerned with the use of business and consumer surveys in a global context.

Part I
Leading Indicators and Turning Points

1

On the Use of Markov Switching Models Applied to Business Survey Data for the Prediction of Turning Points

Marlene Amstad¹

Abstract

This paper investigates the use of Markov Switching Models (MSM) applied to Business Survey Data (BSD) as a predictor of business cycle turning points. Results of the quarterly Swiss industry survey are converted to a dichotomous signal by using Hamilton’s Markov Switching Model. The derived series tends to coincide with phases of excess boom and the end of recessions, respectively. It thus leads the turning points in the growth cycle. After determining the potential contributors to a leading indicator, the corresponding filtered series are combined in a single variable (H-Signal), which can be considered to perform reasonably well. The H-Signal outperforms the barometer of the Swiss Institute of Business Cycle Research in terms of a longer lead and less false signals in forecasting growth cycle phases. The proposed method can easily be applied to other countries.

1 Introduction

This paper presents the first results of attempts at the KOF/ETH² to design a leading indicator for turning points using Markov Switching Models (MSM) applied to business survey data (BSD) for the case of Switzerland. The key feature of such an indicator is the possibility to derive a quantitative measure for probabilities of being in a recession or boom and, therefore, to predict switches between the regimes. As the business survey data used as input for the model is not subject to revisions, forecasts are bound to be more stable. Furthermore, as business survey data has become available for most industrial countries over several business cycles, the proposed method could prove to be applicable for other countries as well.

The structure of the paper is as follows: In section 2 the motivation for the design of an additional indicator at the KOF/ETH is given. Section 3 strikes the issue of different possible reference series indicating phases of recession and boom, which can be used as a benchmark for the new indicator. Section 4 gives a brief outline of the methodology of Markov switching models followed by a description of the data and model specification used in this paper (section 5). After applying a measure of prediction performance on the markov-switching-filters (MS-filters) in section 6, a signal for turning points in growth cycles is constructed and its forecasting power is evaluated (section 7). The paper ends with the summarized findings and an outlook for future research (section 8).

2 Motivation

The forecasting of economic activity in general and the advance perception of turning points is one of the key interests of business cycle research. Early signals of recession or recovery are of great interest to businessmen, policy makers and investors. Two elements of analyzing this topics will be stressed here: (a) the desire to gain further information on turning points to assist the interpretation of existing indicators and (b) getting more insight into business survey data by means of a nonlinear methodology.

2.1 Further Information on Turning Points

Currently, the KOF/ETH uses mainly two instruments³ to describe and predict the economic activity in Switzerland. Firstly, there is the ‘macro model’ which was developed by Stalder (1991) in the eighties. It uses input data from both the official statistics and business survey data and generates detailed quantitative forecasts on consumption, investment and other macro variables. Predictions are made in the form of quarterly growth rates for the following two years and are published twice a year. The second instrument is the ‘KOF/ETH-Barometer’ (Spörndli (1977), Spörndli (1978)), which aims to condense the information on the general economic situation into a single leading indicator. It was first introduced in the seventies, but has been subject to several revisions with respect to the input data and methods used.⁴ The results are published monthly as a qualitative number with the characteristics of an index.

While both models give information on expected turning points as a byproduct, it can be seen from experience that they suffer from past and ongoing revisions of the input data. It goes without saying that improving the quality of forecasts of turning points is not a specifically Swiss issue. McNees (1991) looked at the predictive performance of a number of reputable forecasting firms in the post-war period and found that the turning points derived from the analyzed forecasts were far from satisfactory. From this point of view it is desirable to gain access to additional information which is independent of the time-consuming revisions of official statistics and which helps to find a better understanding of already existing indicators. In this respect, this paper focuses on the task to find an additional indicator which is specifically oriented towards detecting turning points. The paper presents the first proposal for such an early warning indicator, which shall be used in the future as an additional source of information to support the turning point predictions made by the KOF-Macro-Model and the KOF-Barometer.

2.2 More Insight into the Statistical Properties of Business Survey Data

Business surveys have become increasingly popular and have been conducted in most industrialized countries covering several business cycles. In the endeavor to increase the benefits of this valuable source of information for forecasting economic activity, the statistical properties of business survey data and its behavior over time have been examined in different ways.⁵ Up to today, analysts often employed systems of leading and lagging indicators designed by means of linear methods such as correlations and spectral analysis.⁶ The non-linear Markov switching approach used in this w...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Contents
• List of Contributors
• Preface
• Introduction
• Part I Leading Indicators and Turning Points
• Part II Classification of Business Cycles
• Part III Survey Data and Policy Decisions
• Part IV Attitudes and Behaviour of Firms
• Part V Capacity Utilisation
• Part VI Stocks and the Business Cycle
• Part VII Survey Data and Economic Forecasts
• Ciret