ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Coincident Indicators and Forecasting

Index of Industrial Production

Small business organisations need short-run estimation and forecasting, and a model that has limited data requirements. Statistical techniques currently used are linear in approach, depend on the choice of the data set’s start–end period, and have low statistical reliability. The ensemble empirical mode decomposition approach is not constrained by these limitations or by the non-stationarity and non-linearity attributes of data. As an illustration, the Indian Index of Industrial Production time series is used to develop a coincident indicator of movements in the index that is simple to model, uses real-time data, and makes accurate forecasts.

The history of forecasting models in economics is long. It dates to the famous “Harvard ABC curves,” which predate World War I. Broadly, forecasting models can be divided into long-term analytical macromodels and short-term indicator models. One of the best examples of long-term analytical macromodels is the so-called National Institute Global Econometric Model of the National Institute of Economic and Social Research, United Kingdom. This model, like others, describes the major activities of countries (like consumption, investment, and government expenditure). The various economies are then linked through the export–import sector of each, while trade equations are used to link these economies to the global economy. Typically, such dynamic general equilibrium models are used to forecast economic activity for individual economies over the following four to five decades. The second class of short-term “indicator models” is individual economy-based, and has the more limited objective of identifying “leading indicators” of cyclical behaviour of important macroeconomic variables like gross domestic product (GDP).

The “Harvard ABC” model holds that the best “leading indicators” of economic cycles were stock prices, volume of bank cheques, and interest rates—in that order—where the upturn (downturn) in the first is followed by the corresponding upturn (downturn) in the others, all culminating in an upswing (downswing) of an economy’s growth cycle. In other words, all the cycles should move together to reflect the true picture of the economy as a whole. Formal work on such models was continued by various researchers at the National Bureau of Economic Research and the Economic Cycle Research Institute. Authors like Sédillot and Pain (2003) and Mourougane and Roma (2003) conducted another set of studies at the Economics Department, Organisation for Economic Co-operation and Development. For an excellent survey of such indicator models, see Banerji and Dua (2011). The idea is that movements of macroeconomic aggregates like GDP and employment are themselves a function of cyclical behaviour of micro-level variables like raw material and other inputs and of the production of items like steel and manufactures. The movements in these micro-variables may reinforce or oppose each other at different times. Only when these movements in each are in the same direction would a persistent cyclical behaviour in aggregates like GDP or employment be observed. Hence, the best predictor of cyclical economic behaviour is one or more of these variables, which then become leading indicators. These indicators can be divided into “coincident” indicators (whose behaviour mimics aggregate economic activity) or “leading” indicators (whose movements precede movements of macroeconomic aggregates).

Dear Reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Updated On : 27th Apr, 2018
Back to Top