The CLI is adjusted to ensure that its cyclical amplitude on average agrees with that of detrended reference series.
Business cycles are recurrent sequences of alternating phases of expansion and contraction in economic activity. The name 'business cycle' has some ambiguity, since it can refer to conceptually different economic fluctuation. Whenever the context does not eliminate ambiguity, the following qualifiers are used to distinguish the different concepts. The 'classical cycle' refers to fluctuations in the level of the economic activity (eg. measured by GDP in volume terms), the 'growth cycle' refers to fluctuations in the economic activity around the long-run potential level, or fluctuations in the output-gap (eg. measured by the de-trended GDP) and finally the 'growth rate cycle' refers to fluctuations of the growth rate of economic activity (eg. GDP growth rate). The OECD CLI is focusing on the 'growth cycle' concept with the amplitude adjusted CLI, but offers translations for the two other concepts with the trend restored CLI for classical cycles and the CLI 12-month rate of change (alternatively year-on-year growth rate) for the growth rate cycle.
The composite leading indicator (CLI) is an aggregate time series displaying a reasonably consistent leading relationship with the reference series for the business cycle in a country. CLI is constructed by aggregating together component series selected according to multiple criteria, such as: economic significance, cyclical correspondence and data quality. As a result of the multi-criteria selection process the CLI can be used to give an early indication of turning points in the reference series but not for quantitative forecasts.
Component series are economic time series which exhibit leading relationship with a reference series at the turning points. The component series are selected from a wide range of economic sectors. The number of series used for the compilation of the OECD CLIs varies for each country, tipically between 5 and 10 series. Selection of the appropriate series for each country is made according to the following criteria: Economic significance: there has to be an a priori economic reason for a leading relationship with the reference series; Cyclical behaviour: cycles should lead those of the reference series, with no missing or extra cycles. At the same time, the lead at turning points should be homogeneous over the whole period; Data quality: statistical coverage of the series should be broad; series should be compiled on a monthly rather than on a quarterly basis; series should be timely and easily available; there should be no break in time series; series should not be revised frequently.
The time span separating two turning points of the same nature (two peaks or two troughs).
De-trending is a procedure in which the long term trend, that may obscure cyclical variations in the component or the reference series, is removed. Up to December 2008 component series were de-trended with the Phase Average Trend (PAT) method. Starting from December 2008 the OECD has decided to replace the combined PAT/MCD approach with the Hodrick-Prescott (HP) filter to perform de-trending and smoothing in a single operation. The HP-filter is operated as a band-pass filter with frequency cut-off at 12 months for high frequency components (smoothing) and with frequency cut-off at 120 months for low frequency components (de-trending).
MCD (Months for Cyclical Dominance) is defined as the shortest span of months for which the I/C ratio is less than unity. I and C are the average month-to-month changes without regard to sign of the irregular and trend-cycle component of the series, respectively. There is a convention that the maximum value of MCD should be 6. For quarterly series, there is an analogous measure, quarters for Cyclical Dominance (QCD), which has a maximum value conventionally defined as 2. Starting from December 2008 the OECD has decided to replace the combined PAT/MCD approach with the Hodrick-Prescott filter, which makes the MCD smoothing obsolete.
This transformation of the detrended component series is required prior to aggregation into CLI in order to express the cyclical movements in a comparable form, on a common scale. The method used to calculate normalised indices is to subtract the mean from the observed value and then to divide the resulting difference by the mean absolute deviation. Finally the series is relocated to have mean of 100.
The time span between a peak and a trough.
Cyclical indicator systems are constructed around a reference series. The reference series is the economic variable whose cyclical movements the CLI intendeds to predict. In the OECD system, the index of total industrial production was used as the reference series up to April 2012. Starting from April 2012 Gross Domestic Product (GDP) is used as the reference series except for China.
SIX MONTH RATE OF CHANGE
This measure is used until December 2008. Starting from December 2008 the OECD has decided to replace the 6-month rate of change with the year-on-year growth rate. The annualised 6-month rate of change of CLIs is calculated by dividing the figure for a given month m by the 12-month moving average centred on m-6.5. Let R(t) and C(t) be respectively the 6-month rate of change and the CLI at t,
Smoothing eliminates the noise from the series, and makes the cyclical signal clearer. Up to December 2008 component series were smoothed according to their MCD (months for cyclical dominance) values to reduce irregularity. Starting from December 2008 the OECD has decided to replace the combined PAT/MCD approach with the Hodrick-Prescott (HP) filter to perform de-trending and smoothing in a single operation. The HP-filter is operated as a band-pass filter with frequency cut-off at 12 months for high frequency components (smoothing) and with frequency cut-off at 120 months for low frequency components (de-trending).
The trend restored CLI is composed of the trend of the reference series and the amplitude adjusted CLI. it is comparable with the original reference series.
In time series analysis, a given time series can be decomposed into: - A cyclical component, - A trend component, - A seasonal component, - An irregular component. Since December 2008 the OECD uses the Hodrick-Prescott (HP) filter to estimate the trend. Up to December 2008 the method of trend estimation adopted by the OECD was a modified version of the phase-average trend (PAT) method developed by the United States NBER.
A turning point occurs in a series when the deviation-from-trend series reached a local maximum (Peak) or a local minimum (Trough). Growth cycle peaks (end of expansion) occur when activity is furthest above its trend level. Growth cycle troughs (end of contraction/recession) occur when activity is furthest below its trend level. In addition, turning points should respect various censoring rules. In the simplified Bry-Boschan procedure, used in the OECD CLI system for turning point identification, these censor rules guarantee the alternation of peaks and troughs, while ensuring that phases last not less than 9 months and and cycles last not less than 2 years.
Component series are equally weighted in the aggregation process into a country CLI. On the other hand, GDP-PPP weights are used to estimate the CLIs for groups of countries, i.e. zone.
YEAR-ON-YEAR GROWTH RATES (YoY)
Alternatively called the '12-month rate of change', this rate is calculated by dividing the figure for a given period t (a month or a quarter in relation to the frequency of the data) by the value of the corresponding period in the previous year. Let R(t) and C(t) be respectively the Year-on-Year growth rate and the CLI at t is:
For monthly data:
For quarterly data:
Starting from December 2008 the OECD has decided to replace the 6-month rate of change with the year-on-year growth rate.
In addition to the individual country series OECD calculates zone aggregates for CLIs and reference series included in the OECD CLI framework. Five indicators (the amplitude adjusted CLI, the normalized CLI, original reference series, the trend of the reference series and the normalized de-trended reference series) are calculated using a chain linking formula, with country components weighted according to their GDP share (on a Purchasing Power Parity basis). Weights are changed every five years. Iceland is excluded from zone calculations since no CLI is published for this country. Four other indicators are calculated or derived from the five indicators above. This is done to preserve the relationship between the different indicator types. The derived indicators are: the trend restored CLI and its 12month rate of change, the ratio to trend of the original reference series and the 12 month rate of change of the reference series.
More detailed information is available in the CLI zone aggregation methodology document. The list of zones and their country composition is available on the OECD website.
Permanent urls for this page: www.oecd.org/std/cli/glossary
Composite Leading Indicators (CLI) Frequently Asked Questions (FAQs)
OECD Composite Leading Indicators: Reference Turning Points and Component Series
Composite Leading Indicators (CLI) for Zones - Weights