Investors’ rationality is the foundation of behavioural finance, which establishes a vital distinction against the assumptions of efficient market hypothesis (EMH). Further, it has been argued that the markets having only rational approach participants are impracticable to develop (Shleifer, 2000). EMH implies that investors are inclined toward active investment behaviour and then following the passive behaviour. However, investors follow or imitate the decision of other investors based on their irrational attitude. Behavioural imitation of investors is termed as herding, which leads the investors to a situation where they set aside their personal information to follow the crowd, even the realisation of their accurate personal information. Investors do it consciously or unconsciously or it can be a deliberate decision or an intellectual bias (Van Campenhout and Verhestraeten, 2010).

The financial models such as capital asset pricing model (CAPM) and arbitrage pricing theory (APT) describe the investors’ support toward EMH in terms of correct evaluation of financial securities by observing rational outlooks. In addition, these two models also assist in the best explanation of stock returns. These models further undertook the different economic variables that cause stock returns (Bessler and Opfer, 2004). The point of concentration of these models is on exploring factors that affect stock returns; i.e., the market return is solely responsible for influencing the stock returns, or some other factors are involved.

According to Sarwar et al. (2019) the stock markets are believed to be very much subjective toward macroeconomic variables. Considering the stock market of Pakistan, Securities and Exchange Commission of Pakistan (SECP) has the responsibility to regulate all the capital markets of Pakistan (Pakistan Economic Survey, 2012–2013). SECP monitors self-regulatory organisations (SRO), growth of capital markets, and intermediaries operative in capital markets. SRO consists of Pakistan Mercantile Exchange Limited (PMEX), Central Depository Company of Pakistan Limited (CDC), National Clearing Company of Pakistan Limited (NCCPL), and Pakistan Stock Exchange (PSX). The intermediaries and agents include ballotters, share registrars, underwriters, advisors, managers, and brokers. Previously, there were three stock exchanges of Pakistan, i.e., Islamabad Stock Exchange, Karachi Stock Exchange, and Lahore Stock Exchange. All these stock exchanges were merged on 11 January, 2016 and were given the name of Pakistan Stock Exchange. It is a sole platform to investors for performing the trading activities in Pakistan. PSX is considered to be one of the best performing stock exchange markets of Asia, and it secured 3rd and 5th positions worldwide in 2014 and 2016, respectively.

The researchers are aiming to examine the impact of herding on stock returns along with associated macroeconomic and fundamental control variables. As per the limited knowledge of researchers, herding has been thoroughly investigated in Pakistan’s context. Scholars have presented the mixed evidence of herding in the KSE-100 index or in the overall Pakistan stock market. Some have found its evidence (Malik and Elahi, 2014; Zafar and Hassan, 2016; Qasim et al., 2019), some have reported partial evidence (Shah et al., 2017; Khan and Rizwan, 2018; Yousaf et al., 2018; Jabeen and Rizavi, 2019), and some have highlighted its absence (Javed et al., 2013; Javaira and Hassan, 2015; Kiran et al., 2020).

Moreover, in the context of Pakistan, the researchers have found the impact of macroeconomic variables and fundamental variables on stock returns (Atiq et al., 2010; Butt et al., 2010; Mahmood et al., 2015; Afshan et al., 2018; Khan et al., 2018). The international researchers have also explored such relationship (Olugbenga and Atanda, 2014; Abdulmannan and Faturohman, 2015; Mitra, 2017; Ahmed, 2018; Ndlovu et al., 2018; Saragih, 2018). Some international studies have investigated the impact of herding on stock returns (Demirer et al., 2015; Chen and Demirer, 2018; Hwang et al., 2018; Rompotis, 2018). Also, researchers have found a study in the local context in which the impact of macroeconomic variables on herding has been investigated, followed by insignificant results (Javaira and Hassan, 2015).

In such perspective, the current research is novel in the sense it has examined the impact of herd behaviour on stock returns with the macroeconomic (exchange rate, inflation rate, and interest rate) and fundamental variables (earnings per share and return on assets). Another contribution of this manuscript is the inclusion of 34 sectors of PSX instead of just targeting the KSE-100 index. The study has also focused on the extended time span, i.e., 1999–2017, whereas the previous studies have been conducted for some specific time (2004 onward). Besides, the significant contribution of the current research is the application of panel autoregressive distributed lag (ARDL) and pooled mean group (PMG) technique, which has provided novelty to this study as to date, and as per the limited knowledge of researchers, PMG technique has not been applied in herding context especially in Pakistan. Indeed, the non-stationarity panel has guided the adoption of such technique (Sare et al., 2018; Yao and Hamori, 2018).

Herd behaviour in the financial markets has been discussed in detail in the already available literature. The 1st empirical study to examine market-wise herd behaviour was carried out by Christie and Huang (1995). The study found no proof of herd behaviour in the market of the United States. According to the authors, the investors use their personal information to make decisions while acting rationally; however, during extreme market conditions, they follow the crowd. The authors found an indication of herd behaviour during the panic market conditions in Taiwan and South Korea. Afterward, a study conducted by Chang et al. (2000) also found assertive herd behaviour in Taiwan and South Korea. However, they found no evidence in Hong Kong and United States stock market and minimal effect in the Japanese market. Their conclusion about the presence of herding during extreme market conditions was based on a non-linearity test.

Chiang and Zheng (2010) investigated herding effects by utilising a modified version of CSAD in 18 global markets from 1998 to 2009. According to their research, herd behaviour was found in Asian markets and developed markets except for the United States. On the other hand, herd behaviour was found in Latin markets and the United States market during crisis periods. Herd behaviour was also investigated by Doğukanlı and Ergün (2011) in Borsa Istanbul, but no evidence of herding was found. Likewise, herd behaviour was not observed by Al-Shboul (2012) in the Jordanian market, before and after the financial crisis of 2008, using the cross section standard deviation (CSSD) methodology. Herding has been reported by the CSAD method only in the crisis time. Ahsan and Sarkar (2013) used the same method to investigate the herd behaviour in the Dhaka Stock Exchange, but no herding signs were found. The authors explained that investor’s decisions are based on the information available to them instead of following the market trend.

Furthermore, Messis and Zapranis (2014) observed herding and its impact on the market volatility in the Athens market by utilising Hwang and Salmon (2004) measure. Herding was investigated in the Spanish stock market (Ahmed et al., 2015) after, during, and before the global financial crisis of 2008 by utilising the above-discussed measures. Herd behaviour was also highlighted in China’s up and down stock market through CSAD method (Chong et al., 2016). Herding was attributed to risk factors; short-term investment aptitude of the investors, and analyst recommendation, but it was not related to the firm’s size. During the significant crisis in France, herd behaviour was examined by Litimi (2017) using the CSAD model in the stock market at the sectoral level. Mertzanis and Allam (2018) utilised daily and monthly data to find the herd behaviour in the Egyptian stock market during bear and bull market conditions, but no herd behaviour was found. They inferred that herd behaviour is a short-living aspect. Bui et al. (2018) utilised CSAD and CSSD models and found herd behaviour in the frontier stock market, especially in the stock market of Vietnam at both market and industry levels (up/down market). Furthermore, the stock market of Vietnam was influenced by the United States stock market, but Hong Kong stock market was only affected by the market information.

Vo and Phan (2019a,b) found herding in the equity market of Vietnam. Herding was investigated by Can and Dizdarlar (2019) in Istanbul’s stock market from 2011 to 2017. They found herding at 1%, but at 5%, it prevailed only during upper extremes. Herd behaviour in the Balkan Stock Exchanges (Solvenia, Romania, Croatia, and Bulgaria) from 2000 to 2016 utilises CSAD model with a primary focus on the Eurozone crisis and global financial crisis (Economou, 2019). The herd behaviour was exhibited in the Balkan region as a whole as compared to the individual stock markets, with the Romanian stock market exhibited the most substantial effect among all other markets. Using the quantile regression and CSAD model, Shantha (2019) studied the evolutionary nature of herd behaviour from 2000 to 2008 in Colombo stock market. Herd behaviour was found to be of twisted nature. Stavroyiannis and Babalos (2019) observed negative herding in the Eurozone stock markets from 2000 to 2016 by utilising CAPM-based procedure. Using the CSAD measure, Ju (2019) explained the prevalence of herding in A and B share markets. The investors of B share herd on various styles such as growth and value stocks, whereas A share stock market herd on growth stocks.

Batmunkh et al. (2020) found herd behaviour in the Mongolian stock market from 1999 to 2019 during bullish and bearish market trends. Wanidwaranan and Padungsaksawasdi (2020) established the strong impact of return jumps on global equity markets. Kumar et al. (2020) investigated herd behaviour in major Asian commodity future markets. Heterogeneous herding effects were observed in these economies. Chauhan et al. (2020) followed aggregate herding signs in large-cap Indian stocks. Mand and Sifat (2021) found herd behaviour in Bursa Malaysia and explained that herding was regime-dependent and non-linear approach. Ukpong et al. (2021) did not observe herd behaviour in the overall United States stock market, but it was profound at the industry level.

Descriptive statistics of various sectors of PSX have been summarised in Table 2. The summary contains all the variables of study, i.e., return on equity, earning per share, herd behaviour, inflation rates, interest rates, exchange rates, and stock returns. Numbers of observations, maximum values, minimum values, standard deviation, and mean for each sector and each variable have also been reported. For macroeconomic factors, i.e., inflation rates, interest rates, and exchange rates, across all sectors, the data for these variables are same across all sectors. It is reported once in the table showing the summary of descriptive statistics. The standard deviation and mean of Inflation Rate (INR) follow the range of 4–7, whereas Exchange Rate (ER) lies in the range of 79–19, and Interest Rate (IR) falls in the range of 2–11. The higher values of mean indicate a significant surge in market differences across sectoral returns. On the other hand, the high values of standard deviation indicate abnormal cross-sectional disparities in various sectors of the stock market because of shocks or unanticipated events. The standard deviation and mean values of herd behaviour also suggest a similar trend across all the sectors. Furthermore, the statistics about the fundamental factors, i.e., Return on Assets (ROA) and Earnings per Share (EPS), show that different sectors follow different directions based on their financial performance. Some corporations earn a high return on their assets, and some corporations pay significantly high earnings to investors because of favourable earnings.

Checking of stationarity of variables is very important before running the required tests on the panel data. Referring Table 3, researchers have used the Fisher type augmented dickey fuller (ADF) test to check the stationarity of study variables in all PSX sectors. The alternate hypothesis considers the absence of unit root or stationarity of the panel, whereas the null hypothesis considers unit root or non-stationarity of the panel data. The results reveal that null hypotheses of unit roots for few variables were rejected, which indicates the stationarity of these variables. In contrast, non-stationarity problem was observed in few other variables, which was removed by taking the first difference. According to the stationarity results, stock returns and herd behaviour were not having stationarity problem at their levels in almost all sectors, i.e., integrated of order I (0). Whereas the macroeconomic variables, i.e., INR, IR, ER, and fundamental variables, i.e., ROA and EPS found stationarity after taking the first difference. Hence, it is finally concluded that all these variables are integrated at the order I (1). The use of the ARDL model has been led by this heterogeneous stationarity of various variables.

Pooled mean group 0 results have been shown in Table 4 for various sectors of PSX. The table indicates short-term and also long-term results. The PMG method has been applied to the herding results of the CSAD model. Furthermore, herding was not found in PSX; hence, PSX itself was not a part of this analysis. Only those sectors that show herding were considered suitable for applying this analysis. The error correction coefficient is required to be significant and also negative as per the PMG model. The error correction depicts the speed of adjustment, whereas the negative coefficient indicates a causal relationship between dependent and independent variables by exhibiting a degree of convergence and correction from short term to long term. A negative sign of the coefficient has confirmed the presence of a long-term relationship. All selected PSX sectors fulfil the criteria of the PMG model because of the negative coefficient of error correction. Differences in different sectors can be observed for the significance of variables in the long run and the short run. In the long run, it can be observed that herding has a significant and negative relationship with the stock returns of the majority of the sector except for textile weaving, textile composite, sugar and allied refinery, miscellaneous, leather and tanneries, leasing companies, investment banks, and food and personal care products.

Conversely, in the short run, herding has a positive relationship with all sectors’ stock returns except some sectors, which negatively associates with the stock returns. The possible reasoning for this negative relationship is the price destabilisation caused by herding. Furthermore, herding negatively predicts long-term returns and positively predicts short-term returns. However, some sectors have shown an insignificant relationship between stock returns and herd behaviour. Furthermore, in few industries, short-run results are insignificant, and long-run effects are significant, i.e., textile weaving, refinery, leasing companies, and close-end mutual funds. Conversely, the results of gas development and the power generation sector demonstrate long-run insignificant and short-run significant results.

As far as the long-run exchange rate is concerned, a significant positive relationship was found between stock returns and exchange rate in various sectors of PSX except for few financial sectors (oil and gas exploration, leasing companies, investment banks, insurance companies, and commercial banks). An inverse relationship was observed in these sectors between stock returns and exchange rate. However, in the short run, most of the sectors showed a positive relationship except fertilisers. The frequency of this relationship was observed to be similar across all sectors.

Inflation rate and interest rate have an inverse relationship with stock returns. As far as the long run is concerned, most sectors indicated inverse connection between stock returns and interest rates except oil and gas exploration and commercial banks, where these two sectors showed a positive relationship. On the other hand, in the short run, a positive relationship was found in few sectors (sugar and allied industries, pharmaceuticals, commercial banks, cement, cable and electrical goods, and automobile parts and accessories), and few other sectors also showed a negative relationship. However, in the short run, most of the sectors showed an insignificant connection. Considering the inflation rate, most sectors indicated a negative relationship between stock returns and interest rates, but some sectors also showed a positive relationship, i.e., transport, sugar, and allied industries, power generation and gas distribution, modarabas, and cable and electrical goods. Furthermore, insignificant results were found for the short run (interest rate) and long run (inflation rate) for these mentioned sectors. Other than the long run, a positive relationship was found in the short run in many sectors. However, an insignificant relationship was also found in few sectors.

As far as the results of fundamental factors are concerned, for every sector, either the results of ROA or EPS are shown at one time based on their significance in the short run or long run or at both times. Various sectors showed a significant relationship between stock returns and EPS, and few sectors showed significant values for ROA in the short run and long run, as reported in the table. Positive ROA and EPS provide strength to the stock returns. The enhanced level of earnings can give a positive signal to the investors while boosting their investment behaviour. High return on assets allows a corporation to achieve good returns on their investments by utilising raising profits in the stock market and by providing an opportunity to investors for purchasing more stocks of the corporation.

The mixed evidence has been observed for the association between herding and stock returns, and the results of current research have been fully supported by previous researchers. The academicians have found the positive and negative impact of herding on stock returns (Jeon and Moffett, 2010; Celiker et al., 2015). The stock returns assume negative effect of herding due to price destabilisation (Sias and Starks, 1997). Moreover, in the short run, investors’ limited information processing capacity leads them to increase their stock investments, thereby generating high returns. Based on artificial spark, the bullish trend can also be observed in the short run. On the other hand, the investors may find sufficient time to process information in the long run, which leads to avoid investments or seems the stock investment as a favourable avenue. Literature also revealed that herd behaviour negatively forecasts long-run returns and positively predicts short-run returns (Dasgupta et al., 2011). The mixed results of current research are also backed by Zheng et al. (2015), who discovered the positive short-run and long-run linkage between stock returns and herd behaviour.

Previous researchers that include the researchers of Pakistan also supported the positive relationship between exchange rate and stock returns (Khan, 2014; Afshan et al., 2018; Khan et al., 2018) and negative relationship (Menike, 2006) in both long run (Mitra, 2017; Türsoy, 2017; Ndlovu et al., 2018) and short run (Bashir et al., 2016), thereby providing support to the results of this study. However, Gokmenoglu et al. (2021) explained that stock market returns are not affected by exchange rates. Some theoretical backgrounds are also available in explaining such a relationship; i.e., portfolio balance theory (1983) describes the negative connection, whereas theory of goods market (1980) presents the positive linkage between exchange rate and stock returns. It has been observed that the exchange rate stability reinforces the stock returns, and fluctuations in exchange rate leave unfavourable effects on stock returns in both long run and short run.

Results of this study are also in line with the previous studies where the positive effect of interest rate on stock returns (Kibria et al., 2014; Ndlovu et al., 2018) and also the negative impact (Khan, 2014; Afshan et al., 2018) has been established. The increased cost of debts negatively influences stock returns; however, increased interests sometimes optimise stock returns. Interest rate and inflation rate have similar effects on stock returns. Inflation rate also influenced the stock returns in both ways, i.e., positive and negative in both long run and short run, and literature supports these findings (Butt et al., 2010; Mahmood et al., 2015; Khan et al., 2018; Ndlovu et al., 2018; Ifeanyi et al., 2021).

The fundamental variables, i.e., return on assets and earnings per share, have dual effects on stock returns for both long and short run, and the findings are backed by previous researchers (Mirfakhr-Al-Dini et al., 2011; Hatta, 2012; Zulkarnaen et al., 2012; Abdulmannan and Faturohman, 2015; Ahmed, 2018; Saragih, 2018). The higher earnings and the good EPS of a firm boost investors to make stock investments, which result in optimised returns. Agrawal and Bansal (2021) explained that EPS positively affect stock returns.

The current research has some implications for investors, policymakers, and managers. Minimisation of herd behaviour is possible if information is appropriately provided to the investors. Furthermore, investors should be very much clear about the firm-specific information. Moreover, to make rational and appropriate decision, they must know the herding contents. With respect to stock returns, herding can be useful to generate the returns; therefore, investors should keep in mind that herd behaviour is not something bad or undesirable phenomenon. In this situation, concreate efforts should be made by the investors for taking the thorough information about stock specifications, because when investors are motivated toward herding, they usually make investments in the security in which other investors get higher returns. On the other hand, when herd behaviour causes negative returns, the above-discussed precautions must be followed by the investors to minimise such losses. This study also suggests a couple of guidelines to regulatory authorities as to how to maximise stock returns. The policymakers must thoroughly understand the mechanism with which herding affects the stock returns. They can place an effort in magnifying the stock returns even in the presence of herd behaviour. They can ensure the best possible working of macroeconomic variables in generating higher returns. The efficient performance of inflation rate, interest rate, and exchange rate paves the way for the effective working of the economy and leads investors to consider the stock investment as a favourable avenue. Besides, substantial investment in stock generates positive returns. In such a perspective, managers can try their level best to enhance the functioning of fundamental factors (earnings per share and return on assets). The auspicious earnings create robust signals for the investors by boosting their investment attitude. Besides, preeminent return on assets optimises the stock investment potential of investors.

This study has provided some valuable insights into the current body of knowledge; however, it also has few limitations. These limitations can be exploited by the future researchers to produce more valuable and efficient research. The existing research has relied totally on secondary data to investigate the herd behaviour. Herding can be measured in Pakistan by the usage of primary data. Primary data can be very useful to measure the herd behaviour as it may be based an investor’s reallife experiences, and for this, the future researchers can also devise a herding scale. Moreover, future researchers can investigate herding in the South Asian stock markets, and herding phenomenon in these markets can be compared with each other.

This study investigates the impact of herd behaviour on stock returns with some macroeconomic and fundamental control variables at the industry level in Pakistan Stock Exchange. Herding has been detected in most of the sectors of PSX, which further positively and negatively influence stock returns. However, findings revealed that some sectors have an insignificant association between herd behaviour and stock returns. The connection between control variables (earnings per share, return on assets, exchange rate, inflation rate, and interest rate) and stock returns is significant in some sectors. On the other hand, some sectors exposed insignificant relationship. These diverse findings have been experienced in both the long run and short run.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

All authors have contributed equally to complete the research manuscript and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.