Keywords

1 Introduction

The Indian automotive industry recorded the biggest fall in its vehicle sales (almost 40%) in August 2019 (Hindustan Times 2019). During the same time, Nifty Auto index, which tracks the performance of the automotive sector in the National Stock Exchange also fell by about the same rate in comparison to its previous year’s value. Generally, a vibrant automotive sector is considered as an important indicator of the economic performance of any country (Tambade et al. 2019). Past literature states that, macroeconomic variables measure the economic stability of a country and cannot be controlled by corporations (Mohi-u-Din and Mubasher 2013); but, might affect the volatility of their stock prices (Sheikh et al. 2020). Hence, by drawing on asset pricing theory, this paper seeks to examine the association between selected macroeconomic variables and the Nifty Auto Index during a time when the automotive industry in India recorded the sharpest fall in their vehicle sales post year 2000.

Many researchers have attempted to understand the relationship between the macro economy and the stock prices in general (Mishra et al. 2010; Tripathi and Seth 2014). Except for a recent study (Alexander and Al-Malkawi 2022), there is only limited discussion in literature regarding the impact of the macroeconomic factors on the movements of the auto indices in an Indian context. Therefore, we believe that this research study will make a significant contribution to the body of knowledge. Hence, this study seeks to investigate the following question:

RQ: How do the selected macroeconomic variables like crude oil price, gross domestic product (GDP), inflation, exchange rate, interest rate and gold price affect the Nifty Auto Index?

Our study identifies the specific macroeconomic variables (like crude price, exchange rate, GDP, inflation, gold price and interest rate) affecting the auto index of NSE and the extent of the impact in terms of long run and short run relationship using Autoregressive Distributed Lag (ARDL) co-integration technique. The findings from this study will have implications for researchers, corporations, portfolio managers, investors, and policy makers.

This paper is organized as follows. A brief review of literature is provided in Sect. 2, data and methodology is explained in Sect. 3, results are illustrated in Sect. 4, and finally, conclusion along with implications and limitations are discussed in Sect. 5.

2 Literature Review

The extant literature has identified several macroeconomic indicators that affect the automobile industry around the world. For example, Shahabuddin (2009) examined the impact of various economic factors like discount rate, GDP, GNP and other leading economic indicators on automobile sales in the US using regression analysis. Both the lagged and unlagged independent variables showed same strength of relationship with the selected response variables. In another research, Srivastava (2010) investigated the impact of change in interest rate, wholesale price index and industrial production on the Indian stock market using the Johansen’s co-integration test. This study revealed that Indian stock markets are influenced by domestic macroeconomic factors when compared to global factors in the long run. Gaspareniene and Remeikiene (2014) employed correlation and multiple regression analysis to determine the link between macroeconomic factors that influenced the EU automotive industry during the period of global financial crisis. The results indicated that automobile production is strongly influenced by its demand (i.e. new vehicle registration) and the GDP. It also found a moderate correlation between public debt and automobile production.

More recently, Büyük et al. (2018) studied the relationship between macroeconomic dynamics and the automobile sales among four topmost auto production countries viz China, USA, Japan, and Germany by using Ordinary Least Squares (OLS) and Fixed Effect Model (FEM). The findings revealed that real GDP, gasoline price, car production have positive linkage with car sales while change in exchange rate, GDP per capita and inflation and cause the opposite. In another research, Nanda & Panda (2018) examined the impact of firm-specific and macroeconomic indicators on the profitability of Indian manufacturing firms. The study claims that firm-specific variables and exchange rate can be considered as potential indicators of manufacturing firm profitability. However, exchange rate is no better predictor in the short run when compared to the long run. Furthermore, Misra (2018) sought to examine the possibility and strength of linkages between Sensex and a few macroeconomic factors using various statistical tests like co-integration, granger causality and vector error correction methods. They found that there is a long-run causal relationship between money supply, inflation, index of industrial production (IIP), gold prices, interest rates, exchange rate, foreign institutional investment, and BSE Sensex. Additionally, the study also identified a short run causality between inflation and Sensex.

As discussed above, there had been many studies done in the past connecting stock market performance and macroeconomic indicators in many countries including India. However, there exist a dearth of research specifically aimed at finding a possible nexus between these macroeconomic factors and the auto indices in the Indian stock market. Moreover, the above studies have reported inconsistent results. Thus, this research contributes to the existing literature by analyzing the possible long run and short run relationship between selected macro-economic variables and the Nifty Auto index by employing an ARDL cointegration method.

3 Data and Methodology

3.1 Data

The review of literature was followed by a discussion with an industry expert to identify crude oil price, exchange rate, gross domestic product, inflation, gold price and repo rate as the independent variables of the study. All variables except repo rate were converted to their logarithmic values. A brief discussion on the data and their sources are provided in Table 1. This covers the period January 2017 to August 2019.

Table 1 Description of data
Full size table

3.2 Methodology

Auto Regressive Distributed Lag (ARDL) co-integration method (Pesaran et al. 2001) is employed to empirically analyze the long run linkage and dynamic interaction between the macro variables and the Nifty auto index.

3.3 Ardl Model Specification

The ARDL models constructed to empirically analyze the impact of macroeconomic variables on the movement of Nifty Auto index is given below (Joshi and Giri 2015; Alexander and Al-Malkawi 2022).

$$ \begin{array}{*{20}l} {\Delta {\text{LNiftyAuto }} = {\text{A}}_{0} + {\text{A}}_{{\text{1}}} {\text{LCRUDE}}_{{{\text{t}} - {\text{1}}}} ~~~ + {\text{A}}_{{\text{2}}} {\text{LER}}_{{{\text{t}} - {\text{1}}}} ~~~~ + {\text{A}}_{{\text{3}}} {\text{LIIP}}_{{{\text{t}} - {\text{1}}}} ~~} \hfill \\ {~ + {\text{A}}_{{\text{4}}} {\text{REPO}}_{{{\text{t}} - {\text{1}}}} ~ + \sum\nolimits_{{i = 1}}^{q} {a_{i} \Delta {\text{LNifty}}\,{\text{Auto}}_{{t - i}} + } \sum\nolimits_{{i = 1}}^{q} {b_{i} \Delta {\text{LCRUDE}}_{{t - i}} } } \hfill \\ { + \sum\nolimits_{{i = 1}}^{q} {c_{i} \Delta {\text{LER}}_{{t - i}} + \sum\nolimits_{{i = 1}}^{q} {d_{i} \Delta {\text{LIIP}}_{{t - i}} + } } \sum\nolimits_{{i = 1}}^{q} {e_{i} \Delta {\text{REPO}}_{{t - i}} + \varepsilon _{t} } } \hfill \\ \end{array} $$
(1)

Here, the first part of this equation with A1, A2, A3 and A4 refer to the long run coefficients and the second part with ai, bi, ci, di and ei refers to the short run coefficients.

The null hypothesis is stated as, H0: A1 = A2 = A3 = A4 = 0 (i. e no co-integration) and the alternate hypothesis is specified as, H1 = A1 \(\ne \) A2 \(\ne \) A3 ≠ A4 ≠ 0 (i.e. co-integration).

The prefix L indicates that the model uses data in the log form. As there was an indication of significant multicollinearity between REPO and LGoldPrice (ϱ = −0.84) from the pre-generated correlation matrix, the latter was eliminated from subsequent analyses.

3.4 ARDL Bounds Testing

In the ARDL procedure, the first step is to estimate the above Eq. (1) by an Original Least Square regression to test the possibility of a long run relationship between the corresponding variables. This is done by conducting an F-test which tests for the joint significance of the lagged levels of variables. After establishing co-integration, the next step is to estimate the conditional ARDL long run model.

Similarly, the conditional ARDL model for LNiftyAuto is specified as follows:

$$ \begin{array}{*{20}l} {\Delta {\text{LNiftyAuto}}_{{\text{t}}} = {\text{ A}}_{0} + \sum\nolimits_{i = 1}^{q} {{\text{A}}_{1} \Delta {\text{LNifty}}\,{\text{Auto}}_{t - i} + \sum\nolimits_{i = 1}^{q} {{\text{A}}_{2} \Delta {\text{LCRUDE}}_{t - i} + \sum\nolimits_{i = 1}^{q} {{\text{A}}_{3} {\text{LER}}_{t - i} + } } } } \hfill \\ { + \sum\nolimits_{i = 1}^{q} {{\text{A}}_{4} {\text{LIIP}}_{t - i} + \sum\nolimits_{i = 1}^{q} {{\text{A}}_{{5}} \Delta {\text{REPO}}_{t - i} + \varepsilon_{t} } } } \hfill \\ \end{array} $$
(2)

In the final step, the short run dynamic parameters are estimated by an error correction model with the long run estimates. The error correction version of the above model is as follows:

$$ \begin{array}{*{20}l} {\Delta {\text{LNiftyAuto}}_{{\text{t}}} = {\text{ const}} + \sum\nolimits_{i = 1}^{q} {a_{i} \Delta {\text{LNifty}}\,{\text{Auto}}_{t - i} + \sum\nolimits_{i = 1}^{q} {b_{i} \Delta {\text{LCRUDE}}_{t - i} + \sum\nolimits_{i = 1}^{q} {c_{i} \Delta {\text{LER}}_{t - i} + } } } } \hfill \\ { + \sum\nolimits_{i = 1}^{q} {d_{i} \Delta {\text{LIIP}}_{t - i} + \sum\nolimits_{i = 1}^{q} {e_{i} \Delta {\text{REPO}}_{t - i} + {\text{c}}_{{1}} {\text{ECM}}_{{{\text{t}} - {1}}} + \varepsilon_{t} } } } \hfill \\ \end{array} $$
(3)

ai, bi, ci, di and ei are the short run dynamic coefficients to equilibrium and c1 is the speed adjustment coefficients of LNiftyAuto.

4 Results and Discussion

4.1 Descriptive Statistics

A descriptive summary of all the variables is provided in the Table 2 below. No major discrepancies are observed.

Table 2 Descriptive statistics of all variables
Full size table

4.2 Augmented Dickey Fuller (ADF) Test

Before examining the association between selected macroeconomic factors and the Nifty Auto index, their stationarity properties were assessed using unit root tests like ADF tests and Philips Perron (PP) tests. The results of ADF tests are provided in Table 3 below.

Table 3 ADF test results
Full size table

From the above results it is seen that variables like LNiftyAuto, LCRUDE, REPO, LER, LCPI and LGoldPrice are integrated of order 1(i.e., stationary at first difference) whereas LIIP is integrated of order 0 (i.e., stationary at level). Hence it is ascertained that there is no problem in going forward with the ARDL tests. The PP test results also confirmed the same but are not provided here to ensure brevity.

4.3 Ardl Cointegration Test

Having established that the data series under study are a combination of I(0) and I(1) in the previous sections, now we proceed to analyze the co-integrating relationship using ARDL bounds testing approach (Adeleye et al. 2018). The step-by-step ARDL test results are discussed in the following sections. It is important to note that Stata reported a multicollinearity error with LCPI in the model and was automatically eliminated when ARDL regression was done.

Lag Length Selection

The optimal lag(s) selected by Akaike Information Criterion (AIC) for the variables LNiftyAuto, LCRUDE, LER, LIIP and REPO were 4, 4, 2, 4 and 1 respectively.

ARDL Bounds Test

$$ {\text{H}}_{0} :{\text{ no }}\,{\text{levels }}\,{\text{relationship}} $$

The bounds test results are provided in Table 4 below. As the F statistic value (7.036) is greater than the critical value for I (1) regressors, the null hypothesis is rejected. Thus, it suggests a co-integrating relationship between the variables (Pesaran et al. 2001).

Table 4 Bounds test results
Full size table

ARDL and ECM Results

Table5 given below presents the long run estimates.

Table 5 ARDL Long run results
Full size table

The long run results indicate that LER (−4.3025) and LIIP (−2.0583) have a negative impact on the NSE auto index. At the same time, CRUDE (0.958) & REPO (0.2461) seems to have a positive influence on the movement of the NSE auto index. The short run estimates along with the adjustment coefficients are given in the Table 6.

Table 6 ARDL short run results
Full size table
Table 7 Summary of diagnostic test results
Full size table

The error correction or adjustment term (i.e. the first lag of LNiftyAuto) is negative (−1.2318) and is found to be statistically significant as the p-value, 0.008 < 0.05. In this analysis, ECMt-1 = −1.2318 indicates that 123.18% of the disequilibrium of the previous month’s shocks are corrected back to the long run equilibrium in the present month via the explanatory variables in the model. The coefficient is less than −1 but lies within the dynamically stable range as it is not lower than −2 (Pesaran et al. 1999; Adeleye et al. 2018).

From the short run coefficients, it is seen that LER and LIIP has a positive influence whereas CRUDE has a negative influence on the auto index. These results are consistent with past studies (Joshi & Giri 2015; Sinha & Kohli 2015). Moreover, the first lag of CRUDE (−0.687) seems to be a significant predictor of Nifty auto index in the short run. For the first lag of CRUDE (i,e previous month value), this can be explained as, a 1% increase of last month’s crude price will lead to approximately 0.687% decrease in Nifty auto index in the current month, ceteris paribus. Further, R2 = 0.9465 indicates that 94.65% of the variations in Nifty Auto index is explained by the regressors in the model (Koop 2013).

Diagnostic Test Results

The diagnostic and stability tests results are given in Table 7.

5 Conclusion

The aim of the study was to analyze the association between selected macroeconomic factors such as crude price, exchange rate (USD/INR), index of industrial production, repo rate, inflation, gold price and the Nifty auto index. ARDL cointegration method was used to analyze the data.

Exchange rate and IIP were seen to be significant negative predictors of Nifty auto index in the long run. In addition, crude price and interest rates were seen to have a significant positive relationship with the index in the long run. On the other hand, in the short run, first lag of crude price revealed a significant negative association with the auto index while IIP showed a significant positive relationship. The above results have implications for investors, portfolio managers, corporations, and policy makers.

This study recommends that corporations may adopt appropriate hedging strategies to mitigate the exchange rate risk. Furthermore, they must also see whether the industry is up to date with the market trends and adopt appropriate strategies. Additionally, policy makers should monitor the current situation of the auto industry and implement appropriate macroeconomic policies (like scrappage policies, tax reduction) to foster growth.

Like with other empirical research, this study has some limitations that must be addressed in future research. The single equation model that we adopted for analyzing the association between the variables might not be suitable in such situations when there is inter-relationship among the chosen variables. Hence, future researchers should consider using simultaneous equation models. Even more, additional macroeconomic variables and longer time frames can be incorporated to have a better outlook on the association between the variables under study.