Stock Market Asymmetry and Investors’ Sensation on Prime Minister: Indian Evidence

Abstract

This study empirically examines the growth of return, volatility shocks, market efficiency and investors’ sentiment on prime ministers during their administration as a prime minister. Thus, various volatility forecasting measures are applied. It is observed that BSE return does not follow a random walk and inefficient during their tenures as a prime minister. ARCH measure confirms about volatility clustering. According to the EGARCH measure leverage effect does not exist, but the presence of this effect based on TARCH during the tenure of few prime ministers. Finally, the investors are trustful to those prime ministers who are elected from the Indian National Congress according to the growth of return.

Keywords

BSE GARCH Indira Gandhi Rajiv Gandhi Narendra Modi

Introduction

The stock market movement around the globe partly depends on presidential or prime minister election. The role of a particular political party is important during their election camping and after their wining for stock market growth. Generally, investors have a positive emotion or faith on a particular political party or a leader which he/she assumes that this leader is trustable for the growth of the stock market if he/she becomes the prime minister. During the tenure of a prime minister, the stock market experiences several confrontations. Since independence, India has experienced 15 prime ministers (Narendra Modi presently holding PM office). But the Indian stock market (BSE) releases its value publicly from April 3, 1979 and thus, we have seen 13 prime ministers from that date. During this period, the stock market faces diverse information asymmetry.

At this ground, the present study tries to examine the growth of the stock market, market efficiency and various asymmetric effects during their tenures. Although, this study is restricted to those prime ministers who have completed their PM offices at least for three years.

The study is designed as follows: objective of the study; data and study period; methodology; result; and finally, a conclusion and recommendation.

Literature Review

The financial time series like stock and exchange rate tends to occur in volatility clusters due to changes in the market. This type of phenomenon is first observed by Mandelbrot (1963) and Fama (1965a, b), and further described by Baillie et al. (1996), Chou (1998), and Schwert (1989). Various models are employed to study this phenomenon. The empirical applications of the autoregressive conditional heteroskedasticity (ARCH) model introduced by Eagle (1982) or its extension generalized by Bollerslev (1986a) in GARCH model and its various extensions (EGARCH, TARCH, PARCH, etc.) by Engle et al. (1987), Glosten et al. (1993), Nelson (1991) tries to forecast stock returns’ volatility. Besides that, it is often observed in the stock returns that volatility is found to be higher after getting bad news (negative shocks) rather than good news (positive shocks) of the same magnitude. Hence, volatility is affected asymmetrically by positive and negative shocks. This phenomenon is called leverage effect which is pointed out by Black (1976) that means changes in stock prices tend to be negatively correlated with the changes in volatility (see Christie, 1982; Nelson, 1991). Engle and Ng (1993) explain the news impact curve (IMC) with an asymmetric response to both good and bad news. To test the leverage effect, many nonlinear extensions of the GARCH model are developed. Similarly, Threshold ARCH (TARCH), Threshold GARCH (TGARCH) and PARCH which are independently developed by Zakoian (1994) and Glosten et al. (1993). Besides these, a large numbers of recent studies have examined different aspects of volatility forecasting (see Ameur & Senanedsch, 2014; Brandt & Jones, 2006; Chen & Lian, 2005; Engle et al., 2007; Gazda & Vyrost, 2003; Goudarzi, 2011; Longin, 1997, etc.) and depicted many evidences by using a range of volatility measures. The risk-return relationship is another property widely examined in the past. In general, it is assumed that the risk-return trade-off is based on the unconditional distribution of return. In 1980, Merton criticizes about the failure of the previous studies in respect to changes in the level of risk when return is forecasted. Thus, it is essential to consider heteroskedasticity by using realized returns. Here, GARCH-M model allows the conditional variance to affect the mean (expected return). Generally, the GARCH model is based on the implicit assumption that the average risk premium is constant for the sample period. The GARCH-M model lightens up this assumption by allowing the velocity feedback effect to become operational. The evidences of risk–return relation are mixed with positive, negative or zero relation (see Glosten et al., 1993; Nelson, 1991; Poon & Taylor, 1992).

A lot of studies examine the various diverse asymmetric effects on stock markets around the globe. But research on this topic in poor and developing countries are very less. So, further research is needed in this counter to know the dynamics of information asymmetry. With this notion, the present study seeks to examine the growth of return, asymmetric effects, market efficiency and investors sentiment on the prime ministers. It is quite uncommon to analyze those aspects during the tenure of the prime ministers. At this ground, the present study is a little different and that adds value in the existing literature.

Objective of the Study

More specifically, the following objectives are achieved:

To examine the growth of BSE

To analyze the diverse asymmetric effect and market efficiency

To observe the investors’ sentiment on prime minister.

Data and Study Period

The study considers the daily closing prices of the Bombay Stock Exchange (BSE) that covers the tenure of six full-time prime ministers. The raw data is converted into a series of continuously compounded percentage returns. During (from January 14, 1980 to May 30, 2019) these periods, there were six full-time PMs including Narendra Modi who completes his first tenure as prime minister in India on May 30, 2019 but the second term is not considered here. Although there are 12 PMs during this study period, most of them do not complete their five years tenure so exclude them. Here, four prime ministers (Indira Gandhi, Rajiv Gandhi, P.V. Narasimha Rao, and Dr Manmohan Singh) are from Indian National Congress (INC) and the remaining two are (Atal Bihari Vajpayee & Narendra Modi) from Bharatiya Janata Party (BJP). The daily closing index price is obtained from the official website of BSE and the information regarding prime minister is obtained from the prime minister’s office.

Methodology

The daily return of the BSE index is computed as under:

R_{t} = \log (I_{t} / I_{t - 1})

where It is the index value at the current period t and It–1 is the price at the previous period.

Here, the Jarque–Bera (J–B) test statistic is used to observe whether the time series data is normally distributed or not. The J–B test statistic as follows:

J B = n [\frac{S^{2}}{6} + \frac{{(K - 3)}^{2}}{24}]

(1)

where n is the number of observations, S is the skewness, and K is the kurtosis. To test normality, the following hypothesis is formulated:

H₀: Return series is normally distributed.

H_a: Return series is not normally distributed.

It is assumed that the time series data is stationary that means its mean, variance and covariance are time invariant. To test stationary, a nonparametric approach proposed by Phillips and Perron (1988) is applied and based on the following statistic:

{\tilde{t}}_{α} = t_{α} {(\frac{γ_{0}}{f_{0}})}^{2} - \frac{T (f_{0} - γ_{0}) (s e (\hat{α}))}{2 f_{0}^{}_{}^{1 / 2} s}

(2)

where $\hat{α}$ is the estimate and t_α is the t ratio of α, $s e (\hat{α})$ is the coefficient of standard error, and s is the standard error of the test regression. γ₀ is a consistent estimate of the error variance (computed as, (T – K)s²/T, where k is the number of regressors), f₀ is an estimator of the residual spectrum at frequency zero. Here, the testable hypothesis for unit root may be written as under:

H₀: Return series is nonstationary or unit root.

H_a: Return series is stationary.

KPSS test seems more reliable, as it nuances the type of nonstationary. The ADF and PP tests only determine whether a time series is stationary or not, implicitly assuming in their null hypothesis that the time series contains a unit root. Thus, in case the ADF and PP test states that the time series is nonstationary, while the KPSS does not advocate this hypothesis, it is likely that the time series is level or trend stationary, rather than being nonstationary (Pfaff, 2008). The KPSS is, therefore, more delicate in distinguishing (non)stationary, and seems most appropriate and adequate for further analysis. Kwiatkowski & Shin (1992) propose the following model:

\begin{array}{l} y_{t} = ξ_{t} + r_{t} + e_{t} \\ r_{t} = r_{t - 1} + μ_{t} \end{array}

(3)

Here, r_t corresponds with a random walk process and for the error term, e_t, is assumed that it is independent and identically distributed (i.i.d.) with 0 mean constant standard deviation. The initial value of r_t, which is r₀, is the level of the time series and is fixed. The null hypothesis, H0, posits that e_t is stationary meaning that y_t is level-stationary in case ξ is 0 and trend-stationary otherwise.

H₀: Return series is stationary.

H_a: Return series is nonstationary or has a unit root.

Now the growth model is developed to estimate the rate of growth of return of BSE as well as the growth of the BSE index in value during the tenure of the full-time prime ministers. Here, the basic model is as under:

\log (R_{i t}) = α_{i} + β (T) + e_{i}

(4)

where R_it is the return of the BSE at time t, which is converted to log, T is the time period (duration of each PM), and e is the error term with its usual assumptions.

Equation (4) is a semi-log model because the only dependent variable is in the log shape. After estimating Equation (4), the residuals are considered for testing like a serial or auto-correlation test, heteroskedasticity test, stationarity test and normality test to make the semi-log model suitable. For testing stationarity, correlogram analysis is used and the testable hypothesis is as under:

H₀: Residual is stationary.

H_a: Residual is not stationary.

Generally, the prices of stocks swing widely for an extended time period followed by a period of relative calm which looks like stepping of a drunker person meaning that it follows random walk or nonstationary. To capture such varying variance autoregressive conditional heteroskedasticity (ARCH) model is applied (Engle, 1982). Although, heteroskedasticity has no autoregressive structure that means heteroskedasticity observed over different periods is auto-correlated meaning that presence of ARCH effect.

To test the ARCH effect the following regression equation (OLS) is formulated and estimated after making the series stationary by taking the first difference:

R_{i, t} = β_{1} + β_{2} R_{i, t - 1} + β_{3} R_{i, t - 2} + \dots + β_{p} R_{i, t - p} + e_{t}

(5)

It is assumed that e_t ~ N(0, α₀ + α₁μ²_t-1). Here, the variance of e at time t depends on squared distributions at time t – 1 that causes serial correlation and thus, the variance of “e” not only depends on one lagged squared disturbance term but also on several lagged squared disturbance terms which may be written as

V a r (μ_{t}) = σ_{t}^{2} = α_{0} + α_{1} e_{t - 1}^{2} + α_{2} e_{t - 2}^{2} + \dots + α_{p} e_{t - p}^{2}

(6)

Equation (6) is the ARCH model of order p and the ARCH effect is tested by examining the validity of the null hypothesis H₀: α₁ = α₂ = L = α_p = 0. To test this Engle proposed to run, the auxiliary regression (regressed squared standardized residuals on a constant) at p lags.

e_{t}^{2} = α_{0} + α_{1} e_{t - 1}^{2} + α_{2} e_{t - 2}^{2} + \dots + α_{p} e_{t - p}^{2}

(7)

If there is no ARCH effect in the residuals, then ARCH model is mis-specified. After testing for unit root and ARCH effect then the GARCH model may be applied.

The ARCH specification looks like a moving average specification as compared to the autoregression and thus, considers lagged conditional variance as an autoregressive term in the ARCH model (Bollerslev, 1986b). The GARCH model is based on the assumption that changes of variances depend on the lagged variances of the capital assets. Unexpected swings of stock prices generate more volatility in the upcoming periods. The GARCH (p, q) model may be written as follows:

σ_{}^{2}_{t}^{} = α_{0} + \sum_{i = 1}^{p} α_{i} e_{}^{2}_{t - i}^{} + \sum_{i = 1}^{q} β_{i} σ_{}^{2}_{t - i}^{} + v_{i}

(8)

where α₀ is the mean, p is the degree of the ARCH process (lagged terms of squared errors), and q is the degree of GARCH process (lagged terms of conditional variances) and v_i is the random error. Here, ARIMA technique is applied to determine the degree of p and q (see Box & Jenkins, 1970). The most widespread GARCH (1, 1) model can be written as

σ_{}^{2}_{t}^{} = α_{0} + α_{1} e_{}^{2}_{t - 1}^{} + β_{1} σ_{}^{2}_{t - 1}^{} + v_{t}

(9)

As the variance is expected to be positive then it is assumed that the regression coefficient α₀, β₁, and α₁ will be always positive, while the stationarity of the variance is preserved, if the coefficients β₁ and α₁ are smaller than 1. Here, (α₁ + β₁) expresses the influence of the variability of index or return from the previous period on the current value of the variability which is usually close to 1.

The GARCH models are unable to observe frequently asymmetric effects when different volatility is recorded systematically. According to the martingale model, decrease and increase of return/index price may be treated as bad and good news respectively. If decrease (negative shocks) in return is accompanied by an increase in volatility which is greater than the volatility induced by an increase in return is called leverage effect measured by EGARCH and TGARCH models.

Let R_t is the return of BSE index at time t

R_{t} = σ^{'} I_{t - 1} + ξ_{t}

(10)

ξ_{t} = σ_{t} Z_{t}

(11)

Z_{t} / Ω_{t - 1} ~ ψ (0, 1 ν)

(12)

The conditional variance may be written as follows:

\log σ^{2} = ω + \sum_{i = 1}^{p} ​ α_{i} | \frac{e_{t - 1}}{σ_{t - 1}} | + \sum_{j = 1}^{q} ​ β_{j} \log (σ_{}^{2}_{t - 1}^{}) + \sum_{k = 1}^{r} ​ γ_{k} \frac{e_{t - 1}}{σ_{t - 1}} + v_{t}

(13)

Equation (10) indicates that conditional variance is an exponential function of the BSE returns or values that automatically ensures its positive character. Where $σ_{}^{2}_{t}^{}$ is the conditional variance. Z_t is the standardized residual and ѵ denotes a vector of parameters that specifies the probability density function. ω, α, β, and γ are the parameters to be estimated. Here, α represents the symmetric effect or ARCH effect. β measures the persistence in conditional volatility or GARCH effect. An asymmetric effect is indicated by the nonzero value of γ and the presence of leverage effect is given by its negative value. When γ ˂ 0, then positive shocks (good news) generate less volatility than the negative shocks (bad news), and when γ ˃ 0, then positive innovations are more destabilizing than the negative innovations.

The positive and negative shocks in the stock market have diverse effects on volatility and to deal with this event, Glosten et al. (1993) and Zakoian (1994) independently introduce the Threshold GARCH or TGARCH model² that captures the possible asymmetric shocks to volatility by adding an additional term. The TGARCH (1,1) model is expressed as

σ_{}^{2}_{t}^{} = α_{0} + \sum_{i = 1}^{p} ​ α_{i} e_{}^{2}_{t - i}^{} + \sum_{j = 1}^{q} ​ β_{j} σ_{}^{2}_{t - j}^{} + \sum_{k = 1}^{r} ​ γ_{k} e_{}^{2}_{t - k}^{} I_{t - k}

(14)

where (a) I_t–₁ = 1, if e_t–₁ ˂ 0 or negative (bad news).

(b) I_t–₁ = 0, if e_t–₁ ˃ 0 or positive (good news).

The leverage effect is captured by γ coefficient.

The study also uses a dummy variable to capture investors’ sentiment on prime minister regarding the growth of BSE return that may be shown as under:

R_{B S E} = α + β_{1} I G_{1 B S E} + e_{t}

(15)

R_{B S E} = α + β_{1} R G_{2 B S E} + e_{t}

(16)

R_{B S E} = α + β_{1} N R_{3 B S E} + e_{t}

(17)

R_{B S E} = α + β_{1} A V_{4 B S E} + e_{t}

(18)

R_{B S E} = α + β_{1} M S_{5 B S E} + e_{t}

(19)

R_{B S E} = α + β_{1} N M_{6 B S E} + e_{t}

(20)

where R_BSE represents the return of BSE index.

IG_1BSE = 1 if the investors’ positive sentiment on Indira Gandhi for BSE growth

= 0 otherwise

RG_2BSE = 1 if the investors’ positive sentiment on Rajiv Gandhi for BSE growth

= 0 otherwise

NR_3BSE = 1 if the investors’ positive sentiment on P.V.N. Rao for BSE growth

= 0 otherwise

AV_4BSE = 1 if the investors’ positive sentiment on A.B. Vajpayee for BSE growth

= 0 otherwise

MS_5BSE = 1 if the investors’ positive sentiment on Dr Manmohan Singh for BSE growth

= 0 otherwise

NM_6BSE = 1 if the investors’ positive sentiment on Narendra Modi for BSE growth

= 0 otherwise

Result and Discussion

The descriptive statistics of BSE is presented in Table 1. It is observed that the daily mean return of BSE is quite stable during the tenure of the prime ministers. The highest mean return is provided during the tenure of P.V. Narasimha Rao from Indian National Congress (INC) and the lowest return is provided during the occupancy of Narendra Modi from BJP. The BSE’s risk has reached to highest level during the tenure of Dr Manmohan Singh and lowest during the time of Narendra Modi, Atal Bihari Vajpayee. The JB statistics of the return distribution during the tenure of the prime ministers are very large and the probabilities of obtaining such statistics under the normality assumption are significantly zero meaning that rejection of null hypothesis (H0: normally distributed).

Table 1.

Descriptive Statistics of BSE Return

Prime Minister	OB	Mean	Max	Min	Std. Dev	Skew.	Kurt.	J-B
Indira Gandhi	906	0.0922	11.0530	–7.2100	1.1257	0.8834	17.1613	7688.36 (0.000)
Rajiv Gandhi	1063	0.1057	9.1329	–5.8220	1.6586	0.3238	4.531	122.47 (0.000)
P.V. Narasimha Rao	1065	0.1172	13.1353	–12.7680	1.0140	0.3072	9.0564	1644.43 (0.000)
Atal Bihari Vajpayee	1540	0.0320	8.2541	–11.1385	1.7334	–0.2489	6.0108	597.59 (0.000)
Dr Manmohan Singh	2494	0.0767	17.3393	–10.9564	1.5766	0.3152	12.1145	8674.08 (0.000)
Narendra Modi	1503	0.0292	8.9748	–13.1525	1.1134	–1.2828	25.8057	32983.55 (0.000)

Source: The author.

Note: Probabilities in parenthesis.

Table 2.

Test of Stationarity and Market Efficiency

Prime Minister	Unit Root Test
Prime Minister	ADF Test	PP Test	KPSS Test
Indira Gandhi	–16.77177**	–282.7538**	0.1810
Rajiv Gandhi	–19.91093**	–206.7896**	0.2403
P.V. Narasimha Rao	–16.13296**	–147.4540**	0.2620
Atal Bihari Vajpayee	–18.44720**	–385.3635**	0.1145
Dr Manmohan Singh	–22.95915**	–562.8931**	0.1733
Narendra Modi	–13.9228**	–40.3029**	0.0419

Source: The author.

Note: ** Significance at 5% level.

Table 2 provides information regarding stationarity and market efficiency. It is observed that the computed ADF and PP test statistics in level forms (absolute value) during the tenure of all the prime ministers are higher than the critical value and statistically significant at 5 percent level that means rejection of null hypothesis (H₀: δ = 0 or ρ = 1) and thus, the time series does not appear to have a unit root and do not follow a random walk and thus, inefficient at their weak forms. Based on KPSS test statistics, the LM-statistics also reject the null hypothesis as the LM-statistics are lower than the asymptotic critical value at 1 percent level of significance and may be said that the returns during the tenure of the PMs of BSE do not follow random walks and inefficiency is seen at weak forms during their occupancy.

The growth rate of return of BSE (in percentage) during the tenure of the prime ministers is depicted in Table 3. It is seen that the time coefficients are not significant as the probabilities values are higher than 5 percent (in parenthesis) of all the PMs and the percentage of growth rate is negative during the tenure of Indira Gandhi (from January 14, 1980 to until March 24, 1984), Rajiv Gandhi (October 31, 1984 to until December 2, 1989), P.V. Narasimha Rao (June 21, 1991 to until May 16, 1996) and Dr Manmohan Sing (May 23, 2004 to May 25, 2014) who elected from INC. Similarly, the growth rate of return during the tenure of Atal Bihari Vajpayee (March 19, 1998 to until May 22, 2004) and Narendra Modi (May 26, 2014 to May 30, 2019) who are elected from BJP is positive. Although, the percentage rate of return during their tenure is not lucrative. The R² value during the tenure of all the PMs is not satisfactory and the F statistics are insignificant. The residuals are not serially correlated during the tenure of Indira Gandhi, Atal Bihari Vajpayee, and Narendra Modi but heteroskedasticity problem is found in the residuals during the tenure of all the prime ministers except Rajiv Gandhi. The residuals are also not normally distributed during the tenure of the PMs (JB statistics). But the residuals are found to be stationary during the tenure of Indira Gandhi and Atal Bihari Vajpayee.

Detection of ARCH Effect

Volatility clustering of daily return is examined by applying the ARCH test. Here, AR(1) model is used to generate squared residuals for testing ARCH effect. It is found (Table 4) that both the F-statistics and LM statistics are statistically significant at 5 percent level during the tenure of the prime ministers that means the presence of ARCH effect in return series. It is also observed that the return volatility is lower during the regime of Indira Gandhi (6.0951) who was elected from INC as compared to the other PMs and higher during the time of P.V. Narasimha Rao (7.7133) who was elected from INC and then Dr Manmohan Singh (7.0310) who was elected from INC and so on based on AIC criterion. But in a nutshell, the return volatility is almost same according to AIC criteria.

Table 3.

Growth Rate of Return of BSE

Prime Minister	Coefficient	Growth %	R²	F-stat	Breuch–Godfrey LM Test (Ser. Cor.)	Breuch–Pagan–Godfrey Test (Hetero.)	J–B	Q-Stat. (Stationary)
Indira Gandhi	–0.000144 (0.3141)	–0.014	0.0011	1.01466 (0.3141)	0.3626 (0.8342)	5.1577** (0.0231)	7668.233** (0.0000)	Insignificant
Rajiv Gandhi	–0.000197 (0.2345)	–0.019	0.0013	1.4152 (0.2345)	32.0795** (0.0000)	2.9751 (0.0847)	121.9854** (0.0000)	Significant
P.V. Narasimha Rao	–0.000324 (0.1065)	–0.032	0.0025	2.6105 (0.1065)	31.3032** (0.0000)	44.2694** (0.0000)	1635.526** (0.0000)	Significant
Atal Bihari Vajpayee	0.000056 (0.5730)	0.006	0.0002	0.3178 (0.5729)	4.9964 (0.0822)	19.1055** (0.0000)	603.0378** (0.0000)	Insignificant
Dr Manmohan Singh	–0.000045 (0.3020)	–0.005	0.0004	1.0647 (0.3022)	16.1025** (0.0003)	6.8703** (0.0088)	8708.093** (0.0000)	Significant
Narendra Modi	–0.000020 (0.5964)	0.006	0.0006	0.0989 (0.7531)	2.6363 (0.2676)	23.5660** (0.0000)	32830.86** (0.0000)	Significant

Source: The author.

Note: ** Significance at 5% level.

The estimated outcome of GARCH model is presented in Table 5. It is observed that the lagged squared residuals coefficients of the return during the tenure of the Indian prime ministers are significant as the probability values are less than 5 percent that means the existence of volatility clustering (ARCH effect) that also confirms about the volatility of risk, which is affected significantly by the past squared residuals during the tenure of the prime ministers. Similarly, the coefficients of the lagged conditional variance in the conditional variance equation (GARCH effect) are significant as the probability values are less than 5 percent meaning that the past volatilities of BSE return during the tenure of the prime ministers’ significantly influence current returns. Moreover, the summation of ARCH and GARCH effect measures the shock persistence which is very close to unity during their tenure.

Table 4.

Test of ARCH Effect

Prime Minister	F-Statistics	Obs*R²	AIC
Indira Gandhi	4.7561** (0.0232)	4.7538** (0.0109)	6.0951
Rajiv Gandhi	6.3019** (0.0122)	6.2764** (0.0122)	6.1746
P.V. Narasimha Rao	105.4367** (0.0000)	96.0869** (0.0000)	7.7133
Atal Bihari Vajpayee	204.5026** (0.0000)	180.7093** (0.0000)	6.5219
Dr Manmohan Singh	35.1090** (0.0000)	34.6487** (0.0000)	7.0310
Narendra Modi	65.3867** (0.0000)	62.7374** (0.0000)	6.3944

Source: The author.

Note: ** Significance at 5% level.

Table 5.

Estimation of GARCH

GARCH = C(3) + C(4)RESID(–1)² + C(5)GARCH(–1)
Prime Minister	C(3)	C(4)	C(5)	C(4) + C(5)	AIC
Indira Gandhi	0.2006** (0.0000)	0.2194** (0.0000)	0.6539** (0.0000)	0.8733	2.9629
Rajiv Gandhi	0.0558** (0.0003)	0.0609** (0.0000)	0.9205** (0.0000)	0.9814	3.7494
P.V. Narasimha Rao	0.0579** (0.0016)	0.1088** (0.0000)	0.8777** (0.0000)	0.9865	3.8525
Atal Bihari Vajpayee	0.1629** (0.0000)	0.1651** (0.0000)	0.7881** (0.0000)	0.9532	3.7587
Dr Manmohan Singh	0.0275** (0.0000)	0.0924** (0.0000)	0.8965** (0.0000)	0.9889	3.3657
Narendra Modi	0.0053** (0.0004)	0.1061** (0.0000)	0.8697** (0.0000)	0.9758	2.5864

Source: The author.

Note: ** Significance at 5% level.

Table 6 presents the estimated result of the EGARCH model. It is observed that the constant terms in variance equation are statistically significant and the persistence of conditional volatility (volatility clustering) exists in return during the tenure of the prime ministers as depicted by the probability values. The GARCH coefficients of the return series during their tenure are statistically significant meaning that past shocks persistence significantly influences current returns. The study also examines the existence of leverage effect on a return during their tenure. The gamma coefficients of BSE return during the occupancy of the prime ministers are nonzero meaning that presence of asymmetric effect in the volatilities. Here, the γ coefficients of BSE return during their regime are positive and significant. It may be said that leverage effect does not exist in the return during the prime ministers’ tenure that implies the presence of positive correlation between the past return and future volatility of the return or in other words, positive shocks (good news) generates less volatility than negative shocks (bad news). Thus, the return of the BSE is less volatile during the administration of the PMs.

Table 7 presents the outcome of the TGARCH estimation. Here, the constant term is significant during the tenure of the prime ministers. The GARCH coefficients are statistically significant that confirms about previous return volatility significantly influence current returns. The γ values during the tenure of the prime ministers are not zero that means the presence of asymmetric shocks in the return. If a fall in return is accompanied by an increase in volatility greater than the volatility induces by an increase in return then leverage effect exist or in other words, leverage effect means the negative correlation between past returns and future volatility of returns (bad news has more impact on the volatility of return than good news). Here, the C(5) coefficients during the time of Atal Bihari Vajpayee, Dr Manmohan Singh, and Narendra Modi are positive and statistically significant as the probabilities values are less than 5 percent that means leverage effect exists during their administration. Thus, it may be said that negative news has more impact on conditional volatility of return than the positive news during their tenure as compared to other prime ministers.

Table 6.

Estimation of EGARCH Model

LOG(GARCH) = C(3) + C(4)ABS(RESID(–1)/@SQRT(GARCH(–1))) + C(5)RESID(–1)/@SQRT(GARCH(–1) + C(6)*LOG(GARCH(-1))
Prime Minister	C(3)	C(4)	C(5)	C(6)	AIC
Indira Gandhi	–0.1915** (0.0000)	0.3014** (0.0000)	0.0457** (0.0029)	0.8812** (0.0000)	2.9511
Rajiv Gandhi	–0.0787** (0.0000)	0.1464** (0.0000)	–0.0190 (0.1480)	0.9654** (0.0000)	3.7432
P.V. Narasimha Rao	–0.1445** (0.0000)	0.2112** (0.0000)	0.0312** (0.0316)	0.9808** (0.0000)	3.8562
Atal Bihari Vajpayee	–0.1517** (0.0000)	0.2949** (0.0000)	–0.1301** (0.0000)	0.9155** (0.0000)	3.7356
Dr Manmohan Singh	–0.1376** (0.0000)	0.1947** (0.0000)	–0.0787** (0.0000)	0.9781** (0.0000)	3.3567
Narendra Modi	–0.0919** (0.0000)	0.1078** (0.0000)	–0.1446** (0.0000)	0.9742** (0.0000)	2.5275

Source: The author.

Note: ** Significance at 5% level.

The study uses various measures to forecast volatility of BSE return, and it is also a hard task to select the best performing model among them. Here, the AIC (Akaike Information Criterion) is used to select the best model which is presented in Table 8. It is observed that the EGARCH model is superior to forecast return volatility during the administration of Indira Gandhi (INC), Rajiv Gandhi (INC), and Narendra Modi (BJP). Similarly, TGARCH measure is appropriate to forecast BSE return volatility during the tenure of P.V. Narasimha Rao (INC), Atal Bihari Vajpayee (BJP), and Dr Manmohan Singh (INC).

Table 7.

Estimation of TGARCH

GARCH = C(3) + C(4)RESID(–1)² + C(5)RESID(–1)²(RESSID(–1) < 0) + C(6)GARCH(–1)
Prime Minister	C(3)	C(4)	C(5)	C(6)	C(4) + C(5)	C(4) + C(5) + C(6)/2	AIC
Indira Gandhi	0.2075** (0.0000)	0.2715** (0.0000)	–0.0999** (0.0155)	0.6442** (0.0000)	0.1716	0.4079	2.9617
Rajiv Gandhi	0.0727** (0.0002)	0.0539** (0.0001)	0.0379 (0.0576)	0.9039** (0.0000)	0.0918	0.4979	3.7492
P.V. Narasimha Rao	0.0553** (0.0013)	0.1178** (0.0000)	–0.0297 (0.1881)	0.8816** (0.0000)	0.0881	0.4849	3.8531
Atal Bihari Vajpayee	0.2014** (0.0000)	0.0439** (0.0080)	0.2144** (0.0000)	0.7838** (0.0000)	0.2583	0.5211	3.7327
Dr Manmohan Singh	0.0332** (0.0000)	0.0396** (0.0000)	0.1076** (0.0000)	0.8929** (0.0000)	0.1472	0.5201	3.3508
Narendra Modi	0.0275** (0.0000)	–0.0168** (0.0253)	0.1919** (0.0000)	0.8887** (0.0000)	0.1751	0.5319	2.5343

Source: The author.

Note: ** Significance at 5% level.

Table 8.

Selection of Appropriate Volatility Forecasting Model

AIC
Prime Minister	GARCH	EGARCH	TGARCH	Selection
Indira Gandhi	2.9629	2.9511	2.9617	EGARCH
Rajiv Gandhi	3.7494	3.7432	3.7492	EGARCH
P.V. Narasimha Rao	3.8525	3.8562	3.8531	TARCH
Atal Bihari Vajpayee	3.7587	3.7356	3.7327	TARCH
Dr Manmohan Singh	3.3657	3.3567	3.3508	TARCH
Narendra Modi	2.5864	2.5275	2.5343	EGARCH

Source: The author.

Table 9.

Investors’ Sentiment on Prime Minister

Prime Minister	Coefficient	t-Statistic	Probability
Indira Gandhi	0.0922**	2.4656	0.0139
Rajiv Gandhi	0.1057**	2.0779	0.0380
P.V. Narasimha Rao	0.1172	1.8985	0.0579
Atal Bihari Vajpayee	0.0320	0.7251	0.4685
Dr Manmohan Singh	0.0768**	2.4325	0.0151
Narendra Modi	0.0349	1.3269	0.1650

Source: The author.

Note: ** Significant at 5% level.

The investors’ sentiment on prime minister regarding the growth of BSE return is presented in Table 9. It is observed that the investors are more sensitive during the administration of Indira Gandhi, Rajiv Gandhi, and Dr Manmohan Singh who are elected from INC because the prime ministers’ dummies are positive and statistically significant meaning that those prime ministers have a significant impact on the growth of BSE return and gaining the faith of the investors point of view. Similarly, during the tenure of P.V. Narasimha Rao (INC), Atal Bihari Vajpayee (BJP), and Narendra Modi (BJP) have failed to gain the faith of the investors’ as the coefficients are not statistically significant and therefore, cannot influence the growth of return significantly.

Conclusion

It is observed that the BSE return does not follow a random walk and inefficient at its weak form during the tenure of the prime ministers. Therefore, the investors cannot forecast future return during their occupancy. Moreover, volatility shocks present in the return. GARCH coefficient indicates that past volatility affects the current returns during their tenure that is also followed by the EGARCH measure. Leverage effect does not exist in the return during their time span. But leverage effect exists based on TGARCH measure during the administration of Atal Bihari Vajpayee, Dr Manmohan Singh, and Narendra Modi meaning that negative shocks generate more volatility than positive shocks during their tenure. The appropriate volatility forecasting model during their tenure is determined based on the AIC criterion. Finally, the investors are more biased on those prime ministers who are elected from the Indian National Congress for the growth of BSE return. Finally, it may be concluded that election of prime minister can influence stock market’s growth and also recommended that every elected prime minister should take appropriate measure and policy by which Indian stock market can flourish and investors can trust on them.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author received no financial support for the research, authorship and/or publication of this article.

Note

References

Ameur

H. B.

, & Senanedsch

(2014). Socially responsible investments: An international empirical study of time-varying risk premiums. Journal of Applied Business Research, 30(5), 1513–1524.

Baillie

T. R.

, & Mikkelsen

O. H.

(1996). Fractionally integrated generalized autoregressive conditional heteroskedasticity. Journal of Economics, 74(1), 3–30.

Black

(1976). Studies of stock price volatility changes. Proceedings of the 1976 Meetings of the American Statistical Association, Business and Economical Statistics Section, 177–181.

Bollerslev

(1986a). Generalized autoregressive conditional heteroscedasticity. Journal of Econometrics, 31, 307–327.

Bollerslev

(1986b). A conditional heteroscedasticity time series model for speculative prices and rates of returns. Review of Economics and Statistics, 69, 542–547.

Bollerslev

, Chou

R. Y.

, & Kroner

K. F.

(1992). ARCH modelling in finance. Journal of Econometrics, 52, 5–59.

Box

G. E. P.

, & Jenkins

M. G.

(1970). Time series analysis: Forecasting and control. Holden-Day.

Brandt

W. M.

, & Jones

S. C.

(2006). Volatility forecasting with range-based EGARCH models. Journal of Business & Economic Statistics, 24(4), 470–486.

Christie

A. A.

(1982). The stochastic behaviour of common stock variances: Value, leverage and interest rate effect. Journal of Financial Economics, 10, 407–432.

10.

Chou

R. Y.

(1998). Volatility persistence and stock valuations: Some empirical evidence using GARCH. Journal of Applied Econometrics, 3, 279–294.

11.

Chen

Y. W.

, & Lian

K. K.

(2005). A comparison of forecasting models for ASEAN equity markets. Sunway Academic Journal, 2, 1–12.

12.

Engle

R. F.

(1982). Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation. Econometrica, 50, 987–1008.

13.

Engle

R. F.

, & Bollerslev

(1986). Modelling the persistence of conditional variances. Econometric Reviews, 5, 1–50.

14.

Engle

R. F.

, Lilien

D. M.

, & Robins

R. P.

(1987). Estimating time varying risk premia in the term structure: The ARCH-M model. Econometrica, 55, 391–407.

15.

Engle

R. F.

, & Ng

V. K.

(1993). Measuring and testing the impact of news on volatility. Journal of Finance, 48, 1022–1082.

16.

Engle

R. F. Focardi

, & Fabozzi

F. J.

(2007). ARCH/GARCH models in applied financial econometrics. JWPR026-Fabozzi c114-NP (pp. 1–14).

17.

Fama

E. F.

(1965a). The bahaviour of stock market prices. Journal of Business, 38, 34–105.

18.

Fama

E. F.

(1965b). Tomorrow on the New York stock exchange. Journal of Business, 38, 285–299.

19.

Godfrey

, Granger

C. W. J.

, & Morgenstern

(1964). The random walk hypothesis of stock market behaviour. Kyklos, 17, 1–30.

20.

Glosten

L. R.

, Jagannathan

, & Runkle

D. E.

(1993). On the relation between the expected value and the volatility of the national excess return on stocks. Journal of Finance, 48, 1779–1801.

21.

Gazda

, & Vyrost

(2003). Application of GARCH models in forecasting the volatility of the Slovak share index (SAX). BIATEC, XI, 17–20.

22.

Goudarzi

, & Ramamarayanan

C. S.

(2011). Modelling asymmetric volatility in the Indian stock market. International Journal of Management, 6(3), 221–231.

23.

Kwiatkowski

, & Shin

(1992). Testing the null hypothesis of stationarity against the alternative of a unit root: How sure are we that economic time series have a unit root? Journal of Economics, 154(1/3), 159–178.

24.

A. W.

, & MacKinlay

A. C.

(1987). Stock market prices do not follow random walks: Evidence from a simple specification test. Review of Financial Studies, 1, 41–66.

25.

Longin

M. F.

(1997). The threshold effect in expected volatility: A model based on asymmetric information. The Review of Financial Studies, 10(3), 837–869.

26.

Mandelbrot

(1963). The variation of certain speculative prices. Journal of Business, 36, 394–419.

27.

Nelson

D. B.

(1990). Stationarity and persistence in the GARCH(1,1) model. Econometric Theory, 59, 318–334.

28.

Nelson

D. B.

(1991). Conditional heteroscedasticity in asset returns: A new approach. Econometrica, 59, 347–370.

29.

Phillips

P. C. B.

, & Perron

(1988). Testing for a unit root in time series regression. Biometrika, 75, 335–346.

30.

Poon

H. S.

, & Taylor

J. S.

(1992). Stock returns and volatility: An empirical study of the UK stock market. Journal of Banking & Finance, 16(1), 37–59.

31.

Poon

S. H.

, & Granger

(2003). Forecasting volatility in financial markets: A review. Journal of Economic Literature, 41(2), 478–539.

32.

Schwert

W. G.

(1989). Why does stock market volatility change over time? The Journal of Finance, 44(5), 1115–1153.

33.

Zakoian

J. M.

(1994). Threshold heteroscedasticity models. Journal of Economic Dynamics and Control, 18, 931–944.