|Page tools: Print Page RSS Search this Product|
1992 Feature Article - The Business Cycle in Australia: 1959 to 1992
This article is an attempt to date and measure the Australian business cycle. It is the first in a series to be published in the AEI to study the time relationships between the business cycle and the main economic indicators.
Business cycles were given the following definition by Burns and Mitchell from the National Bureau of Economic Research (NBER) of the United States in 1946:
“Business cycles are a type of fluctuation found in the aggregate economic activity of nations that organise their work mainly in business enterprises: a cycle consists of expansions occurring at the same time in many economic activities followed by similar general recessions, contractions and revivals which merge into the expansion phase of the next cycle; this sequence of changes is recurrent but not periodic...".
This definition has been quoted many times and is still appropriate. Monitoring the fluctuations in aggregate economic activity is of crucial importance to decision makers, both in economic policy making and in business activities. Dating the past turning points of the business cycle and measuring the relative sizes of the successive fluctuations permit the study of the time relationships between different economic indicators. The knowledge of these past relationships assists in the detection of current and future turning points. The research of the turning points in economic activity, called cyclical analysis, was pioneered in the early 1930s by Burns and Mitchell. Since then, this approach to economic analysis has been used and developed in many countries, including Australia.
Cyclical analysis has developed its own terminology which is worth detailing in this introduction. The succession of fluctuations in the aggregate economic activity is named the business cycle, sometimes the reference cycle. The chronology of turning points, which is the dates of the peaks and the troughs
delimiting expansions and contractions of the general economic activity, is generally referred to as the reference chronology. A cycle is composed of two phases, an expansion and a contraction (or recession).
The objective here is to have the broadest possible measure of aggregate economic activity, so the Burns and Mitchell definition has been applied to the average of the three estimates of constant price gross domestic product (GDP(A)). Some previous analyses of the business cycle, particularly in the United States and Australia, have focused on aggregate general economic activity represented by a cluster of economic indicators. This approach, justified for a monthly analysis or when only poor national accounts data are available, has not been followed here. The present study is focused on a quarterly analysis of the business cycle. GDP(A) is the quarterly series which has the broadest economic coverage possible across industries, economic agents, income, expenditure and output (ABS, 1990). Besides these qualities, GDP(A) has the advantage of being recognised by any user of economic statistics. Quarterly estimates of production-based GDP are only available from September 1974 onwards. Before this date GDP(A) is a weighted average of expenditure and income based GDP.
Another important element of this study is the length of cycles examined. Since the objective of this work is to provide insights into short-term fluctuations in the economy it has been decided to concentrate on cycles with total duration of two to eight years. This choice is consistent with the classical approach to measuring business cycles. Cycles of total duration shorter than two years are more likely to be associated with irregular movements including measurement errors and socio-economic shocks (Zarb, 1992). Fluctuations corresponding to cycles of durations longer than eight years, here known as long-term trend, originate from changes in structural factors such as technology, culture or demography.
Two approaches to cyclical analysis are commonly used. The “growth cycles” method is used by most of the institutions which have worked on business cycles determination (OECD, 1987) and is the method used in this study. The other commonly used method analyses the growth in the seasonally adjusted series. This implies, without any real justification, that the long-term trend is a deterministic function of time.
Using the “growth-cycle” method, time series can be broken down into seasonal and trading-day variations, irregular short-term movements, long-term trend and, finally, the business cycle components. Estimates of these different components are obtained by using filters. The solid line in Chart 1 shows the series of GDP(A) corrected for seasonal, trading day and irregular movements while the dotted line shows its long-term trend. The business cycle is obtained by removing the long-term trend element from the series already corrected for seasonal, trading day and irregular movements (ie. the deviations of the solid line from the dotted line in Chart 1). Chart 2 shows the result of this process: the deviation from the long-term trend of the smoothed GDP(A) from 1960. The vertical lines in Chart 1 and Chart 2 delineate the successive phases discussed below.
CHART 1. GDP(A) AT AVERAGE 1984-85 PRICES
CHART 2. GDP(A) DEVIATION FROM TREND
The filters used in this study to eliminate the irregular and the long-term movements are, respectively, 7 and 33 term Henderson moving averages. When applied to a series corrected for seasonal and trading day movements they retain the fluctuations corresponding to cycles of 7 to 33 quarters length. The loss in terms of timeliness at end-points is minimal and no time-shift is induced. This methodology is also consistent with the general smoothing methods used by the ABS (ABS, 1987). This choice of filter offers the additional advantage of keeping the technique as simple as possible, from both users’ and statisticians’ points of view.
Using the deviation of smoothed GDP(A) from the long-term trend (Chart 2), turning points are selected visually amongst local minima and maxima. There may be an element of subjective judgement involved in the selection of turning points and two rules of thumb from the “Handbook of cyclical indicators” of the United States Bureau of Economic Analysis (BEA) are used for ambiguous cases (BEA, 1987). The first rule is that peaks and troughs must alternate. This rule enables the distinction between “double-turns” and full cycles. It also assists with the assessment of the validity of the most recent turning point. A turning point is clearly identified when either the next turning point has been identified or the corresponding phase has an amplitude greater than the smallest clearly recognised phase. The other rule of thumb used here is that the last value is chosen as the turning point in case of equal values.
Results and chronology
Table 1 gives the dates of turning points in the Australian business cycle together with the duration of each phase of cycle, expansions and contractions, and the total duration of each cycle. The relative amplitude of the deviation from the long-term trend at various turning points and the amplitude of each phase and of the total cycle are also presented in Table 1 . The amplitude of phase is simply measured by the sum of the absolute levels of amplitudes at each end of the phase. These measures are only meant to compare cycles in this study. The means of the different durations and amplitudes are also presented in Table 1. The last column shows the average annual percentage growth rate of the long-term trend during each cycle.
TABLE 1. THE AUSTRALIAN BUSINESS CYCLES
Eight full cycles can be identified from 1960 to now, nine contractions and eight expansions, with the dates of the last phase being preliminary. The turning point observed in Q3 1991 might become a trough, in the sense of growth cycle analysis, only when the current phase has reached an amplitude larger than the smallest amplitude of phases observed up to now or when the next peak is reached. Several conclusions can be drawn from Table 1 and Chart 2:
Given the nature of the estimation of the business cycle which involves a long-term trend estimate, the amplitude of the most recent phase is provisional. It will be revised until the long-term trend converges towards a more definite value. More data are also needed to confirm the last turning point of the series and the characteristics of the corresponding phase.
Table 2 contains data used for this study for long-term trend, smoothed and percent deviation of the long-term trend of the GDP(A) series.
TABLE 2. GDP(A) AT AVERAGE 1984-85 PRICES, SMOOTHED, LONG-TERM TREND AND DEVIATION FROM TREND
This article provides a dating of turning points of the Australian business cycle up to 1992 Q2, with the last turning point still being provisional. This chronology is used for the analysis of the time relationships of the cycles in individual economic indicators with those of general economic activity. The determination of leading and coincident indicators is derived from this work using the same techniques of filtering. The chronology of turning points in general economic activity can also be useful for analysing the behaviour of economic variables in the different phases of the business cycle.
This feature article was contributed by Gerard Salou and Cynthia Kim. Gerard Salou, of the OECD Statistics Directorate, was on a temporary assignment with the ABS. Cynthia Kim was an ABS officer.
Australian Bureau of Statistics, Australian National Accounts: National Income and Expenditure (cat. no. 5206.0), June quarter 1990.
Australian Bureau of Statistics, A Guide to Smoothing Time Series - Estimates of Trends (cat. no. 1316.0), 1987.
Bureau of Economic Analysis, Programmed Selection of Cyclical Turning Points, Handbook of Cyclical Indicators, 1987.
Burns, Arthur and Wesley Mitchell, Measuring Business Cycles, National Bureau of Economic Research, 1946.
OECD, OECD Leading lndicators and Business Cycles in Member Countries 1960-1985, Sources and Methods No. 39, January 1987.
Zarb, John, Smarter Data Use, Australian Economic Indicators, Australian Bureau of Statistics, March 1992.
These documents will be presented in a new window.