TrackingAsia was developed by researchers at the Asian Development Bank. The purpose in designing the framework was to complement toolkits of governments, businesses, investors, and analysts for monitoring current economic conditions and their likely direction.

The business cycle, represented by fluctuations in real GDP per capita around a long-term estimated trend, can gauge the state of an economy. But data on real GDP are often not timely and hence the grasp of overall cyclical conditions is not as current as required for real-time decision making. Even though GDP of the previous quarter is released with fairly long lags in many Asian economies, other statistical indicators are available every month that contain an abundance of information about activity in various sectors and the aggregate economy.

The TrackingAsia project designed the Economic Activity Index as an indicator to capture economic activity so that business cycles can be tracked monthly. The composite index draws data series for each economy from six categories and sectors—consumption, investment, trade, government, financial, and the external sector—and identifies the broad groups that are driving economic expansions and downturns, historically and currently.

Business cycles comprise upturns and downturns in aggregate measures of economic activity. In TrackingAsia, business-cycle fluctuations are measured in terms of the growth rate of real GDP around a long-term trend, called the growth gap.

A business cycle is characterized by four phases, depending on whether the growth rate is one of the following:

**Above trend and increasing****:**the economy is expanding and moves toward a peak.**Above trend and****decreasing:**the economy is still above trend but slowing; the cycle eventually crosses the line of long-term trend growth.**Below trend and decreasing****:**the economy is slowing and resources are underutilized; the economy moves toward a trough.**Below trend and****increasing:**rising from the trough, the economy is recovering and eventually crosses the line of long-term trend growth.

The duration of one cycle is the number of quarters from a peak to the next peak. A slowdown is a period from a peak to the next trough; an expansion is a period from a trough to the next peak. The height of the cycle from the peak or trough to the zero-trend line is called the amplitude.

Using historical quarterly real GDP per capita, the growth gap cycle is extracted and turning points—peaks and troughs—are identified. Based on historical patterns in this data, the average duration of a cycle, expansions, slowdowns, and amplitude are computed. The position on the cycle in the latest quarter can be compared with the summary statistics to gauge how many quarters expansions and slowdowns typically last, and how far an economy is from turning direction.

The EAI is a weighted average of multiple indicators of economic activity, where the weights capture the indicator’s relative importance to historical fluctuations. These weights are estimated using principal components analysis, applied individually to each economy. Indicators, which may differ by economy, are selected from the six categories and sectors of data—consumption, investment, trade, government, finance, and the external sector. This data are available at a monthly frequency and are detailed in Resources.

The following is the methodology used to compute the EAI:

In the first stage, monthly data that feed into the construction of the index are compiled from various sources for available years and transformed, using these steps:

- An indicator with multiple base years is spliced to create a longer time series. As the rule of thumb, the latest base year is used while splicing.
- Data that are seasonally adjusted by source are always preferred. If this data are not available and seasonality is detected, the data are treated using the X13-ARIMA package in EViews, a statistical software used mainly for time-series analysis.
- Series in year-to-date format are transformed into monthly values by deducting the values of the previous month in a year from the values of each month of the year.
- Series are converted to logarithm values. Variables that are in percentages, rates, and indexes are retained as is.
- The dataset is balanced; that is, ragged ends in variables are avoided by completing the indicator series that has different release dates using an AR(1) model.
- All indicators are rescaled to have mean zero and standard deviation of one before using principal components analysis.

*Generating the EAI*

Principal components analysis is applied to the monthly indicators selected in the first stage to identify common movements. The first principal components are used to create the index. They are first transformed into weights by taking the absolute value and dividing it by the sum of all the estimated components. The rescaled components are then multiplied with the monthly indicators and summed to generate the EAI. The index is a weighted average of the underlying indicators, where the weights capture the relative importance of historical fluctuations that arise in different categories and sectors of the data. To track periods of economic expansion and slowdown, represented by the business cycle, the EAI is projected onto GDP gap cycles.

In the second stage, and to improve the accuracy of the EAI in tracking economic cycles, statistical techniques are used to fine tune the selection of indicators that feed into the construction of the index.

*Variable selection for constructing the EAI*

To identify the relevant indicators for inclusion in the EAI using the principal components analysis method outlined earlier, three alternative selection methods are used. The first is a correlation method, where the correlation of each indicator with the GDP growth rate is computed using Pearson’s correlation. Indicators with less than 50% correlation are removed. The other two methods belong to model averaging techniques—Bayesian model averaging and weighted-average least squares.

Among the automatic variable selection processes to construct the EAI, the method which best tracks GDP gap cycles is identified. To improve the fit, an ad hoc examination—using economy-specific knowledge—of the existing set of indicators is conducted.

*TrackingAsia flowchart*

*Interpreting the EAI*

The following is an example of a business cycle dial for Indonesia.

TrackingAsia’s results on Indonesia’s business cycle based on quarterly data show the economy in the second quarter of 2020 was in the red zone, meaning below its long-term trend and slowing.

The monthly EAI suggests the economy in May 2020 was less unfavorable than in April. The index deteriorated from –1.77 in April to –2.24 in May. Being in the red zone, the economy was still deeply below its long-term trend and slowing. In the tracker, all sectors and categories are positioned below zero (purple).

For further details and information about TrackingAsia, email **trackingasia@adb.org**

ADB Headquarters: 6 ADB Avenue, Mandaluyong City 1550, Metro Manila, Philippines.

trackingasia@adb.org