To understand the differential impact of different economic indicators on the sub-segments of equities market

Abstract

What drives the stock markets has always been an intriguing question. There are many theories, many models of valuations, many trading algorithms and many modes of analysis, be it fundamental of technical. Through this study we tried to look at what the history says, what the last 8 years of real stock prices say. This study is about finding those relationships between different sectors of the stock markets and the well-known economic indicators and to identify how different sectors move together. The relationships between the different published macro-economic indicators and the stock markets are less understood, more so in the Indian context. Very few studies have been conducted on this subject in the Indian markets ecosystem. The macro-economic variables that we have considered for the study are based on their relevance in the Indian context and hence impact the Indian markets. The underlying idea is to check whether or not the changes in the macro-economic indicators are adequately captured by the stock markets. In conjunction with this, we will try to explore the interrelationships between the different sectors. To evaluate the relationship, the regression tests were performed between the variables. We observe that the regression results can often be spurious since these tests only work with the assumption of stationarity of the underlying variables which may not be the case. Therefore, to establish the causality we ran Engle-Granger co-integration test between the variables. Finally, we will try to understand and analyse the established causality on a subjective basis as to whether it makes sense. The differentiating factor of this study is the scale at which it was done. We worked on the real data of around 25 different sector indices of BSE for last 8 years. All the statistical procedures were performed by the Python code we developed exclusively for this study. This enabled us to run multiple statistical procedures like hundreds of regressions and co-integrations, batch data processing and transformation in couple of minutes. Few examples of such transformations would be converting all the prices to returns, periodicity transformation from days to months to quarters, cleaning the data and proper representation of the output results.