Ch01: Creating Harmony out of Noisy Data

Chapter Summary:

Basically, any economic time series have four components, i.e. trend, cycles, irregular patterns and seasonality. Graphical analysis is one of the best way to identify these components of the economic time series. For the accurate analysis, interpretation and decision making, we have to identify these specific component of the series. Specifically, trend and cycles, because, trend have long behaviour and cycles of the series have short run behaviour.

If there are two different type of series under analysis, then we have to identify graphically that these two series move with same behaviour in different time span or not. If they are behaving same through different time span, then it means they are co-integrated, otherwise it will make the analysis critical. There could be different issues such as; trend is linear or non-linear, cycles and sub-cycles, structural breaks, unit root. These different type of issues have different type of econometric models for solution. For example, (i) ARIMA are the model for cycles, (ii) co-integration and error correction, is the solution when there is problem of unit root, (iii) we use causality model when we cannot identify what drives what, (iv) for measuring volatility we have ARCH/GARCH family, (v) when there is issue of structural breaks then we have model of regime switch models, and (vi) finally we have forecasting models for predictability.