Econometric analysis of the Asian Paints stock in the National stock exchange
Updated: Feb 2, 2021
Econometric analysis of the Asian Paints stock in the National stock exchange
Written by: Tanvi Gabriel
Asian Paints is one of the largest paint companies within India and is the third largest paint company in the world. Today, it captures around 41% of the Indian market share (Punmiya, 2020) and has not faltered even with the ongoing pandemic. Their stock has not only revolutionised the paint industry within India, but has also placed India as a serious competitor in the world. Given Asian Paint’s continual rise in the 2nd quarter, which is more than impressive in the midst of this seemingly undying pandemic, it’s able to continue making a profit amongst its competitors. Through the foundations of this company rooted in a dismissal of the British rule within India, to the stock split and finally the ability to maintain a profit in middle of a crisis in 2020, this stock has been nothing less than shocking.
Analysing the trend of the Asian Paints stock
The main purpose of this article is to analyse the trend of the Asian Paints stock from 2000 – 2020 Aug, along with building a forecasting model to predict the trend in the future. Python has been used to conduct the time-series analysis along with the forecasting given the ease of the MATLAB library.
To begin with, the closing price of the Asian Paints has been used as a measure to understand the analysis of the stock. This can be seen in Figure 1, wherein the graph of the closing price can be seen.
Figure 1: Plot of the closing price of Asian Paints
The graph depicts the constant increase in the closing price. Although there is a sudden drop in the stock in 2013, this is due to a stock split which usually does not affect the overall price of the stock. Thus, overall, the graph does have a consistent increase in the closing price of the stock.
Augmented Dickey-Fuller Test
To continue with the investigation, it is important to confer whether the graph depicts a stationary time-series. One of the most popular statistical tests, Augmented Dickey Fuller test has been used as it detects the unit root that may be present in the time-series sample (Singh Chaun N, 2020). The following is the hypothesis that was taken:
Null Hypothesis: The series has a unit root
Alternate Hypothesis: The series does not have a unit root
In order for the series to be non- stationary, the null hypothesis must be “accepted”. This therefore implies that the mean and the standard deviation should not be a straight line and therefore should not display a constant mean and variance.
Figure 2: Rolling Mean and Standard Deviation
As in Figure 2, it is evident that the mean and the standard deviation are increasing. The fall in the graph, as previously mentioned, represent a split in the stocks and does not drastically affect the behaviour of the graph. Also, from the results of the Dickey-Fuller test below (Figure 3), it can be seen that the null hypothesis cannot be rejected due to the following 2 reasons:
1. The p-value is greater than 0.05 (Line 3)
2. The test statistic (Line 2) is greater than the critical values (line 6-8)
Therefore, it can be concluded that the data is non-stationary.
Figure 3: Dickey-Fuller Results
Forecasting is a common statistical tool that uses historical data as inputs to calculate informed estimates that are a predictive indicator of the future trends. As the stock prices are observed as sequentially, they can form a time series and thus prove as a viable application for forecasting. It is important to note that the main role of forecasting is to test how the sequence of observations will behave in the future (Tuovila, A & Anderson S, 2020). For the purpose of this article, the behaviour of the last 2 months within the dataset, July and August, will be forecasted and then compared to the actual data present in the dataset. This is done to ensure that the forecasting is validated and will thus improve the results of the test.
Training the Data
To begin with, the model will need to be converted into a stationary model in order for it to be forecasted. This is due to the fact that a stationary time-series has a constant mean and variance and thus ensures that it is easier for it be forecasted. For this reason, the ARIMA Model will be used which is often used to convert a non-stationary data set into a stationary data-set (Sangarshanan, 2018). An ARIMA model will be created and trained with the closing price of the stock. The dataset has been split to omit the last 2 months in order to predict their behaviour. This has been visualised in Figure 4, to gain a clear understanding of the data that is being trained and the data that is being tested (Prabhakaran, S, 2019).
Figure 4: Visual of splitting the Model
Autoregressive Integrated Moving Average, ARIMA for short, uses 3 predictive elements (Sangarshanan, 2018).
1. P which refers to the periods of lags
2. Q which depicts the error component of the lag that is not represented by any trends
3. D is the number of differencing required to make the time series stationary
The above 3 parameters are chosen by using the AUTO ARIMA function in python which seeks to obtain the most optimal parameters for an ARIMA model and returns a fitted AUTO ARIMA model.
Figure 5: Auto ARIMA Model Results
According to figure 5, it can be seen that the AUTO ARIMA has depicted the SARIMAX results. The reason behind this is that the AUTO-ARIMA function determined that as the dataset is missing multiple values, the SARIMAX model is more when training the data (Khandelwal, R, 2020). Also, it can be seen that the function has displayed the best values for p, d and q as seen in “ARIMA ((0,1,0)” and the model has been fitted accordingly.
To continue, it is imperative to initially ensure that the residuals of this fitted model do not display a specific pattern. For this the figure 6 illustrates the residual plots.
Figure 6: Residual Plots of the fitted model
Beginning from the top left, the residual plots seem to fluctuate around a mean 0 and have a constant variance of zero. Continuing from this, the top right graph, the density plot depicts a normal distribution around mean 0. As for the bottom 2 graphs, the “Normal Q-Q” graph shows that dots fall closely to the blue straight line. If there were any deviations in this case, it would suggest that the distribution is skewed. Lastly, the correlogram graph shows that the residual plots are not correlated.
Therefore, the residuals do not display any specific pattern and we can continue with this fitted model. Using the values of the p, d, q that was suggested by the AUTO ARIMA function, the stock prices can thus be forecasted. Below in figure 7, the ARIMA Model results can be shown which are based on the optimal p, d, q values.
Figure 7: ARIMA Model Results
The prices are forecasted on a 95% confidence level and the results can be seen below:
Figure 8: Asian Paints Inc. Stock Price Prediction
The model has been able to forecast the prices of the Asian Paint stock price quite well. However, another test will be used to ensure the accuracy of this model. For this, 4 different metrics have been used - Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE).
Mean Absolute Percentage Error is a common statistical test that is used to test the prediction of accuracy for forecasting methods (Everitt,B., & Skondral, A, 2002). In this case, the 10.9% MAPE illustrates that the model is 90.1% accurate.
Limitations to the Model
Although the model has performed reasonably well, with a 90% accuracy, there are various limitations that need to be considered.
1. Abstract Model
Given the predictive nature of the model, it is considered as an abstract model and is meant to be assumed as so throughout the article. The reason behind this is that the model is only using one single measure to understand the behaviour of the stock – the closing price. In order for the model to be more “holistic”, more stock-related parameters are required to be used such as opening price, PE, the value of the NSE Stock.
2. Subject to certain constraints
The model is also subject to a several number of constraints. One of the main constraints is that the seasonality of the model along with the associated trends have been excluded from this analysis. This would have provided a more accurate indication of the behaviour of the time series graph, however was omitted to ensure that the focus of the article was mainly on the forecasting.
3. Not a Universal Model
Lastly, it is important to note that this model should not be generalised and used as a template to forecast the behaviour of other stocks within the stock market. This model has been used in a very specific scenario, the behaviour of the Asian Paints stock, and thus may not be applicable to other stocks.
Economic implications of the constant increase in Asian Paints stock
Now, the implications of the consistent increase in the Asian Paints stock will be considered. The increase in the stock price has been compared to Asian Paints 3 competitors.
Figure 9: 1 Year Stock Comparison (The Economic Times Markets)
Figure 10: 5 Year Stock Comparison (The Economic Times Markets)
In the 1-year comparison, it can be seen that although there is a consistent increase in the stock prices for all the paint companies, Asian Paints maintains the highest stock price. The similar trend is being seen in figure 10, wherein the 5-year comparison, Asian Paints is depicting the highest stock price. Furthermore, between April’20 – July’20, the overall paint industry sees a decrease in their stock prices. As paint is seen as a discretionary spending item (according to the Asian Paints annual report), the pandemic weakened the renovation demand. However, given the shift to the upcoming festive season and a possible vaccine in the horizon, the demand for paints may increase in the near future.
Overall, it can be seen that there is a consistent increase in the stock prices for all of the paint companies which may imply that the paint industry as a whole is creating a strong competitive oligopoly market. A research conducted on the outlook of the paint industry in India has identified that this increase in the stock prices is due to the increase in the availability of disposable income of the average middle class (Research and Markets, 2020). This is further supplemented by the increase in the access to education, urbanization and the overall development of the rural market. Furthermore, the paint industry’s ability to meet the shift in consumer demand from traditional white-wash paints to odour-free, dust and water-resistant and overall, environmentally friendly paints has suggested that in 2021-2022 the Indian paint industry will see a CAGR increase of 12%.
This continual increase in their stock prices and for the paint industry to survive the pandemic, will depend on the decisions taken by the Indian government. According to the report published by the IBEF on the outlook of the Indian Real Estate Industry, the government will be taking several initiatives to increase the development within this sector. The Union cabinet has approved an Alternate Investment Fund (AIF) of Rs 25,000 Crore (US $3.8 Billion) to revive over 1,600 stalled housing projects in many cities. Furthermore, according to the India Brand Equity Foundation (2017), government has sanctioned 1.12 crore (11.2 million) houses under the Pradhan Mantri Awas Yojana (Urban). This therefore depicts that the government is moving in the right direction by increasing the investment in the real-estate sector and thus will create a positive impact on the stock prices of the paint companies.
In conclusion, an econometric analysis has been conducted of the Asian Paints Stock on the National Stock Exchange of India. Python was used to analyse the stock followed by training the data set and conducting a forecasting of the stock. The forecasting model was found to have a 90.1% accuracy. Finally, the economic implications of the overall paint industry were discussed and concluded that in order for the Paint industry to continue thriving in a post-pandemic world, the Indian government will be required to invest in the real estate sector, thus having a positive impact on the Indian paint industry.
365 Data Science (2020). What Is a SARIMAX Model? | 365 Data Science. 365 Data Science. Retrieved from 365 Data Science.
Asian Paints Limited. (2020.). Annual Report 2019-2020 - MANAGEMENT DISCUSSION AND ANALYSIS. Retrieved from Asian Paints.
Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: principles and practice. Retrieved from OTexts.
India Brand Equity Foundation. (2017). Indian Real Estate Industry: Overview, Market Size, Growth, Investments...IBEF. Retrieved from India Brand Equity Foundation.
Khandelwal, R. (2019). Time Series Prediction using SARIMAX - Data Driven Investor - Medium. Retrieved from Medium.
Prabhakaran, S. (2019). ARIMA Model - Complete Guide to Time Series Forecasting in Python | ML+. ML+. Retrieved from Machine Learning Plus.
Research and Markets. (2020). Indian Paint Industry Insights Report, 2018-2022 - Growing Industrial Base Propelling the Growth of Industrial Paint, Rising Disposable Income in Rural Areas Accelerating Demand. GlobeNewswire Newsroom. Retrieved from GlobeNewswire.
Sen, J., & Chaudhuri, T. (2016, January). Decomposition of time series data of stock markets and its implications for prediction–an application for the Indian auto sector. In Proceedings of the 2nd National Conference on Advances in Business Research and Practices (ABRMP 2016).
Sangarshanan. (2018). Time series Forecasting — ARIMA models - Towards Data Science. Retrieved from Towards Data Science.
Singh Chauhan, N. (2020). Stock Market Forecasting Using Time Series Analysis - KDnuggets. Retrieved from KDnuggets.
TUOVILA, A., & ANDERSON, S. (2020). Forecasting Definition. Investopedia. Retrieved from Investopedia.
Varghese, A., Tarhen, H., Shaikh, A., Banik, P., & Ramadasi, A. (2016). Stock Market Prediction Using Time Series. International Journal on Recent and Innovation Trends in Computing and Communication, 4(5), 427-430.
Punmiya, V. (2020, September 01). Paint Sector. Retrieved from Nirmal Bang.
Everitt,B., & Skondral, A. (2002). The Cambridge dictionary of statistics (Vol.106). Cambridge: Cambridge University Press.