The sectoral effect of demonetization on the economy: Evidence from early reaction of the Indian stock markets

We investigate the impact of the Demonetization of 85% currency in circulation in India on the eve of 8 November 2016 on all the listed stocks spanning over 20 broad industry clusters (sectors) and their affiliation type from the Indian economy over the period of November to Mid-January 2016. Using the event study methodology, we assess the effects of Demonetization, relative to what had been anticipated, as measured by abnormal returns (ARs). The results indicate that Group Affiliated firms witnessed the highest negative abnormal returns both on the event days and during the event window period, while PSUs witnessed the least wrath. On the sectoral front, Demonetization shows a mixed effect in the early days which changes to positive for most of the sectors barring a few. Banking Sector was the worst hit in the early days with a CAAR of −1.74%, while many sectors like Pharma, Paper and Wholesale Trading witnessed a windfall gain in the long run. Subjects: Financial and Monetary Economics; Financial Markets; Regulatory Intervention; Cashless Transaction


ABOUT THE AUTHORS
Amol S. Dhaigudeis working as an Assistant Professor in T A Pai Management Institute, Manipal, India. He is Fellow (OM&QT) from IIM Indore, India. He is actively engaged in research in supply chain coordination, tourism management, and services operations. He is profound case-writer and teacher.
Archit Vinod Tapar is working as an Assistant Professor in IIM Rohtak, India. He is a Fellow (Marketing) from IIM Indore, India. His research interest includes brand management, services marketing, tourism marketing, and e_tailing. He has published and showcased his research work in journals & conferences of repute worldwide.

PUBLIC INTEREST STATEMENT
The impact of the historical decision of demonetization of Nov 2016 on Indian economy is investigated in this study. Data has been collected from BSE archives and CMIE Prowess database. Using event study methodology and parametric and non-parametric significance tests we found that on the whole, all the firms witnessed a significant negative CAAR on the vent day and event window period, which reversed for most of the sectors. Banking was the worst hit while Pharma, Paper and Mining & Minerals made the most windfall gains in the long run-post-Demonetisation.

Introduction
The government of India, in a historic decision, on the night of 8 November 2016 announced the replacement of its 86% currency in circulation by banning the existing currency of INR 500 and 1000 (RBI Report 2018, https://rbidocs.rbi.org.in/rdocs/AnnualReport/PDFs/2ECONOMIC88A5 CC5468FA4639A767862F5921304A.PDF). The rationale put forth by the Government sources for such a bold move was to tackle with the black money problem, counterfeit currency, corruption, checking the terror funding and money laundering and forced adoption of the online transaction by the citizens of the country (Singh, 2018). This was seen as one of the most prominent artificial disruptions to the flow of money in an emerging economy, which had a very high proportion of the informal economy. Due to the sudden replacement of old currency with new ones, and the execution challenges of replacing the 86% of the cash with new one both citizens and economy -especially the informal one had to face the wrath. Many citizens lost lives; there were severe hiccups in the day-to-day functioning of many facets of business operations. Moreover, to make the condition worse Government had imposed serious policing on the quantum of cash which one could withdraw on a given day. This not only severely impacted the day-to-day functioning of the Banks-which started to operate only for facilitating the cash conversion and cash delivery channel-leaving aside their usual money-making business of borrowing and lending.
The present study evaluates the impact of demonetization on Indian stock market-a barometer of the health of the economy-and also seeks to evaluate such impact on various Industry clusters, and firms based on their affiliation type in India.
Event study methodology is used for the study on all the listed firms in the largest Stock exchange of India-Bombay Stock Exchange (BSE) (Black & Khanna, 2007). We choose a large sample for the study as the implication of such an umbrella regulation is supposed to be very different for the nature of firms, their industry cluster, and affiliation type. The industry wise analysis is also warranted as many of industrial sectors-such as Pharma, Travel & Logistics (no collection of tolls for almost over a month across the country), petroleum retailing firms, amongst few others-got benefited through the Government permission of extended use of banned old currency notes for their business transaction. Also, due to the nature of the Indian corporate ecosystem which is heavily dominated by Business Groups and PSUs in terms of their breadth and depth in the market respectively (Srinivasan, 2003).
To the best of our knowledge and belief, no study has analysed the implication of Demonetization on all the listed firms in the country. Secondly, we base our study on the Industry clusters as per the definition of the National Industrial Classification code which classifies the Industries based on their fundamental characteristics and nature of the business. Our study is also the first one to analyse the impact of Demonetization on the firm affiliation type-which gives a good understanding of their interlinks with the informal and cash-based economy and dependencies-which is very important to understand in an emerging economy with weak policing and enforcement of legal framework.

Demonetization: context & related literature
Demonetization is a process of eradicating old currency by introducing a new one. In India, demonetization happened on 8 November 2016 and INR 500, and INR 1000 currency notes lost their legal tender. This decision had a worth of nearly 15.4 trillion (85% of the currency notes in circulation).
Demonetization was a bold move from the Indian government and was intended to curb the issues of black money, corruption, terrorism funding and fake money issues that are predominant in India (Mali, 2016). Promoting the cashless transactions via the digitisation was highly endorsed by the government during the demonetization period. Economic rationale behind demonetization was generally accredited to mainly three factors-one controlling hyperinflation, two eradicating imitation currencies, and lastly broadening the tax base. Despite that people throughout the country had suffered for replacing old notes with the new one. Long queues in all banks and ATMs were a common scene during the demonetization period. Government also set limits on banks & ATMs for the daily transaction of money. The new currency notes of INR 2000 differed considerably from the existing notes in terms of size and shape. This lead to the need of recalibration of ATMs. As the demonetization was a sudden move, there was no time for recalibration of ATMs, leading to increase in the agitation of citizens of India who were in dire need of money. The demonetization is expected to bring structural differences in the longer run and leading to a better reinforcement of the current economy. June 2017 statistics given by the RBI have cast doubt on the "curbing black money" objective of the demonetisation. As per RBI's report, 99% of the open illegal tenders had returned to them, while 1% amounting to INR16,050 crores did not come back. Moreover, then RBI Governor also stated that the short-term pains of demonetization move will outweigh the long-term benefits. Table 1 shows the timeline of events of the demonetization of November 2016.

Related literature
Demonetization is not a new phenomenon, and world economies are witnessing the act of demonetization on a regular interval of time. For instance, Zimbabwean Government has opted for demonetization in 2015, to fight the country's record-breaking hyperinflation. In another case, adoption of Euro by the European Monetary Union in 2002 witnessed the act of demonetization. Moreover, the Coinage Act of 1873 demonetised silver in favour of adopting the gold standard as the legal tender of the United States and so on. The impact of demonetization on Indian economy was a huge one and worth scholarly attention. Researcher and practitioners across the globe have expressed varied views on this topic (e.g., Betz, Anderson & Puthanpura, 2017;Bhatnagar, 2017;Chelladurai & Sornaganesh, 2016;Jaggi, Jain, & Verma, 2018;Lawrence & George, 2018), however there is no consensus on the results (Bhavnani, 2018;Ganesan & Gajendranayagam, 2017;. This paper attempts to study the impact of demonetization on all the listed stocks spanning over 20 broad industry clusters (sectors) and their affiliation type from the Indian economy over the period of November to Mid-January 2016. Figure 1 and Table 2 highlights the literate on demonetization.

Research design & methodology
Event study methodology is used for analysing the impact of promoters' equity dilution on the stock returns. We used the Market Model, 1 for predicting the normal returns.
Market Model: where, R i;τ is the stock return of firm i on the day τ. R M;τ is the market return on the day τ; and ε i;τ is the random error term. 2 BSE CNX 500 3 Index returns are used as a proxy for market returns. Demonetisation was announced on the night of 8 November 2016, and thus we have considered 9 November 2016 as the event day. Taking cues from previous studies, 4 we chose 220-days estimation window (leaving 20 days before the event as cool off period) and 11-day event window period. We also consider a smaller event window period of 3 days in our analysis. Normal returns were predicted for the event window based on the historical return values of the estimation window observations and an expected return was calculated for each firm which was then subtracted from the actual returns on the event window days to get abnormal returns in the events window using the standard event analysis methodology used in financial economics research.
Mostly the event studies in the financial literature rely on the parametric tests. However, the parametric tests have been reported to have one disadvantage that they necessarily require the assumptions of normal distribution of returns which has been time and again refuted by many-e.g., Brown and Warner (1985). If this necessary assumption is violated then parametric tests, yield misspecified test statistics. Researchers have reported the Non-parametric tests are well-specified and more potent at detecting false acceptance of the null hypothesis of nonexistence of an abnormal return. The most successful among these tests were the nonparametric sign and rank tests advanced in Corrado (1989), Cowan (1992). Each of these studies documents that sign and rank tests provide better specification and power than parametric tests.
Additionally, since our study has a single event day, we expect the issue of event clustering and presence of stocks' return variance around the event date indicating that simple crosssectional t-test may reject the null hypothesis too often. We, therefore, followed Boehmer, Masumeci, and Poulsen (1991) and Kolari and Pynnönen (2010) adjustments as suggested in many recent literary works on using financial event analysis (Fernando et al., 2012;Ricci, 2015). The method specified by Boehmer et al. (1991) relies on standardised values of abnormal returns, unlike the basic cross-sectional t-tests. Kolari and Pynnönen (2010) adjustment are done to take care of cross-correlation due to clustering of the event dates along with the presence of higher variance of returns during the event dates over the estimations window. Since there is a higher discrepancy in firm affiliation and characteristics in our samples along with the advantage of having a reasonably large and diverse dataset, we expect a different degree of impact of the regulation on the firms based on their affiliation type, nature of business-sector or industry clusters, and many more.
To sum up the discussion, we use two parametric tests, namely Cross-Sectional t-test and Standardized Cross-Sectional t-Test-famously known as the BMP test. Also, the two Non-Parametric tests, namely-Rank Test (Corrado, 1989) and Sign test (Cowan, 1992). The underlying econometrics has been outlined in the following sub-section. Impleme ntation (1,17) Digitiz ation (2,24,28) Sector Specific Theoretical / Conceptual (3,5,6,7,9,13,23) Positive (2,11,17,20) Negativ e Mixed (3,13,14,15,17,18,21) Figure 1. Literature on demonetization source: developed by authors.  Brown & Warner (1980) have proved that cross-sectional t-test has higher power over normal time series t-tests. It is robust enough to handle the increase in abnormal returns variance induced by the event itself. However, later it was proved by Boehmer et al. (1991) that standardised crosssectional test developed by them has greater power of the test statistics, which has been the workhorse for the analysis done in the first essay. However, to assure robustness of the test results, We have also used this method of testing the hypotheses.
Cross-sectional t-test assumes the null hypothesis that the average abnormal returns (averaged over all the firms) is equal to 0 and is calculated as under: Analysed the impact of demonetization on the Indian retail sector and proved that the losses in the short-term will outweigh the long-term benefits 28 Pal et al., 2018 Studied the adoption of mobile wallets aftermath of demonetization using seven constructs from existing technology adoption literature and three moderators.
29 Roy et al., 2018 Conducted a sentimental analysis using social media data exchanges and found out the temporal negative impact of the move.
Source: Developed by authors where,σ CAAR T 1 ;T 2 ð Þ is the estimated cross-sectional variance of the abnormal returns, calculated as under 3.1.2. Standardised cross-sectional test (BMP 1991) Boehmer et al. (1991) have shown that plain standardised residual test developed by Patell (1976) works well under conditions of no increase inthe variance of abnormal returns during the event window period. However, in cases where there is the occurrence of event-induced variance increase, then the standardised residual test rejects the null hypothesis too often. Boehmer et al. (1991) in their method modify the standardised residuals test developed by Patell (1976). They, based on the cross-section of the event window period abnormal returns, combine an empirical version of the event variance estimate. This test is assumed to be robust enough to eventinduced variance to stock returns and t-stat for the null hypothesis that CAAR = 0 is given by, where, CSARðT 1 ; T 2 Þ is the cross-sectional average of the abnormal returns cumulated over time (as calculated in the method specified by Patell's (1976) method of the standardised residual test), given by Where, Std CSAR À Á is the standard deviation of CSARðT 1 ; T 2 Þ calculated as,

Generalised sign test
Proposed by Cowan (1992), the generalised Sign test is based on the ratio of positive abnormal returns ratio over the vent window period. The null hypothesis is based on the assumption that the positive returns ratio of the event window does not deviate from the positive abnormal returns ratio during the estimation window. The test statistics is calculated as follows: where p þ event window is the ratio of positive abnormal returns in the event window period. Moreover, p þ estimation window is the ratio of abnormal returns in the estimation window period.

Corrado's rank test
Rank test, a relatively more robust test in caparison to the standard parametric tests was proposed by Corrado (1989). Following the same approach as the generalised sign test, the rank test is free from the necessity of the symmetry of the cross-sectional abnormal return distribution. To apply this test, we have transformed each firm's abnormal returns into their respective ranks. To do so, let ðKI0 À KbarÞ=SðKÞ Where S(K) is the standard deviation.
This statistic is distributed asymptotically as unit normal. Cowan and Sergeant (1996) document that if the return variance is unlikely to increase, then Corrado's rank test is better specified and more powerful than parametric tests. With the increase in variance, however, this test is misspecified.

Data
Data needed for the empirical analysis was collected from secondary sources-like stock exchange data archives and PROWESS database provided by the Centre for Monitoring Indian Economy Pvt. Ltd. (CMIE-Prowess). We collected trading data (daily adjusted stock returns) from 15 November 2015 till the end of 15 January 2017 for all the firms listed on the Bombay Stock Exchange (BSE). 5 We also collect firm identity variables viz., ownership type and industry classification (National Industrial Classification name and code). The 13-digit NIC code classifies all the registered companies in India into specific industrial clusters. We group firms to 6-digit NIC codes into various sectors. Based on this classification we get 20 key industrial clusters/sectors. 6 The list of sectors and individual sample firms for analysis is outlined in Table 3. A total of 4959 firms are listed on the BSE. Of these many firms are highly illiquid and are very thinly traded. We drop all such firms who have less than 110 distinct returns data in the estimation window of 220 days. We also drop all the firms that do not have even a single day of missing returns data in the event window period [+5, −5 days]. We also look for firms, which had any earnings announcements or any major corporate events, which may have an impact on the trading volume and prices, coinciding with the event day and event window period, and drop them from our analysis. After the tedious data preparation and cleaning exercise, we were left with 2478 firms in our final sample for analysis.
We then segregated our data based on firm affiliation type into four major buckets-Standalone (Indian) Private-1295, Group Affiliates-1003, Foreign Subsidiaries/Affiliates-107, and Government Owned firms, also known as Public Sector Undertakings (PSUs)-81.

Empirical results & discussion
4.2.1. Short-run price impact of demonetization 4.2.1.1. Event study based on firm affiliation type. We segregate the data based on firm affiliation type. Since in India, business groups are a dominant force and enjoy the benefits of interlink between the listed group affiliates and unlisted subsidiaries of the group holding. The other dominant force in terms of size (market capitalisation) but not numbers is the Government-owned enterprises. They either operate in a monopolistic environment or are too large in their respective sectors compared to their industry peers. Managers appointed by the government and controlled by the governmentappointed board members run them. They are mostly plagued by inefficiencies and typical state-run organisational issues. Third dominant group is the Standalone Private (Indian) firms. Though large in number, they are mostly run by professionals or first generation entrepreneurs. Last dominant force is the subsidiaries of the foreign multinationals. These firms are very efficiently managed and darlings of the market. They have good corporate governance practices and also have superior technology over their peers. Thus, all four types of firms are very different in their interlocks, vendor-customer engagements, market competition, and corporate governance practices.
The results are reported in Tables 4 and 5. Table 4 reports the CAARs and t-statistics of Parametric tests while Table 5 report the statistics of Non-Parametric tests.
On the event day, all type of firms reported negative returns wherein Group Affiliates witnessed the highest negative CAAR of −2.96% while PSUs witnessed the least negative abnormal returns of −1.32%. However, the shorter event window period the PSUs witnessed a significant positive return while others continued their negative trail. By the 5 th day-in the 11-day event window period the CAARs mounted to almost −14% for the Group Affiliates, −12% each for the Private and Foreign firms and close to −7% for the PSUs. When we only look at the negative returns from the event day for a weeks' trade, we find that the negative abnormal returns are in tune of 8-10% for all the firm types except for the PSUs, which only witnessed −2.79% negative returns. These results are statistically significant for most of the conducted tests at a 1% level of significance.
The results indicate that the traders had a perception that PSUs by being Government undertakings would be shielded by the impact of the currency crunch in the markets. Also, since most of these PSUs deal with vendors and customers, which are government organisations, keeping their liquidity need at a much lower level. The most surprising results came from the Group Affiliates as one would expect that since they are conglomerates and have good interlinks with their subsidiaries both at the upstream and downstream shall not get impacted by the cash crunch so badlyhowever, traders and investors had a different opinion on the same. 4.2.1.2. Event study based on industry clusters/sectors. We then segregate our data based on industry cluster/sectors and analyse the short term impact of the regulation on the stocks. Since    the different industry has its characteristics and dealings with their vendor-customer and a different need for liquid cash for taking care of operations. Some are labour intensive while some deal in credits both at both sides of the supply chain. Moreover, certain sectors-especially the core/traditional sector have a good presence of informal economy which operates and command a sizable market-thus would change the dynamics of the product market at a very different level owing to the cash crunch caused by the regulation. We divide the firms into 20 major industry clusters and perform the same set of calculations and tests. We report the results of the analysis in Tables 6 and 7. The test statistics for Parametric tests are reported in Table 6, while Table 7 reports the tests stats of Non-Parametric tests. As a whole, on the event day most of the sectors witnessed negative abnormal returns barring a few, namely-Consumer Good (0.84%), IT/ITES (1.22%), Pharmaceuticals & Chemicals (0.25%), Textiles (0.26%) and Wholesale Trading (0.51%)-which mostly remained positive throughout the event window period and kept of increasing the respective initial gains on a cumulated basis. However, sectors like Banking showed a significant negative CAAR of −1.74% on the event day, which later reversed to +2.16% in the 11-day event window. These results were statistically significant as per the cross-sectional t-test and Corrado Rank test.
Certain sectors like Mining and Minerals, which showed early negative abnormal returns on the event day increased its negative CAARs to almost −2.57% in the 11-day event window. It was mostly owing to the risk owing to the day-to-day operations since they are a highly labourintensive industry on the informal side. Barring one or two cases like this most of the sectors reversed the early signs of negativity or at least reduced it in the 11-day event window. Most of the above-mentioned results were statistically significant by one of the Parametric or Non-Parametric tests. However, the BMP test mostly returned insignificant test statistics for short-run.

Long run price reversal
Since, demonetization was a vanilla regulation on all the firms, according to the Government shall strengthen the economy in the long run, as more and more capital would return to the formal sector. Thus, one school of thought-going by which the GoI took this initiative-that it would be a temporary shock to the economy which shall revert to normal sooner and shall make it better off in long run. However, another school of thought had a contrarian view on this matter. They believed that such a move coupled with a poor implementation would result in destroying the informal economy-some estimated which to be more than 20% of the GDP-which would have a trickledown effect on the formal sector as they interact in one way or the other. Since, the stock prices in a efficient market demonstrates the intrinsic value of the firm which is nothing but the cumulated present value of all the future cash flows, expected the sectors or industry clusters to suffer a lot who had a higher dealing with such an informal sector-either at the supply or demand side of the business. Secondly, it was also believed that the sectors that had a higher proportion of daily wage labourers would face operational hick-ups leading to production delays and cost overruns. Thus, we also look at the long run price impact of the policy on the stock prices by calculating the CAARs of the next 40 days after the demonetization [0, 40]-for firms based on their affiliation type and their respective industry clusters/sectors. The results are reported in Table 4-7.
On the firm-affiliation type analysis, we find that almost all type of firms better their CAARs in the next two months of the trading~40-day window. Group affiliates maintain their lead in the quantum of negative CAARs over others. However, PSUs were the only one which showed a complete reversal and gives a + 4.31% CAAR in the long-run. The results are mostly significant using various parametric and non-parametric tests, even after controlling for the event-induced variance and clustering effects.
On the industry cluster/sector level analysis of the data, we find that most of the sectors under study     losses to get into the positive zone. These results were mostly significant even after controlling for the event clustering and event-induced variance effects-especially for the sectors like Paper & Print, Pharma & Chemical, and Textiles.

Conclusion
Views on the effects of Demonetization on the Indian economy differ widely and reflect ideological differences, with those in favour, for whatever reason, anticipating positive effects, and vice versa. Though the implications of such a decision would reflect in long-term-the short-term reaction of the stock market to Demonetization provides early indicators as to what the effects have been or was anticipated to be.
The results presented in this study, on the basis of the event study methodology, confirm the proposition that Demonetization have varying sectoral effects, although most sectors reacted negatively as indicated by negative ARs in the early days; however, the trend reversed in the medium term -next two months of trading~40 days post event window. As expected, the banking and finance sector was the worst affected while the certain sectors like IT/ITES, Pharma and consumer durables witnessed a windfall gains in the early days which continued in the event window period. Surprisingly sectors like Travel & Tourism did not get affected adversely despite early hiccups. While from the firms' affiliation perspective PSUs looked to be mostly shielded by the adversaries of the cash crunch while Group Affiliates, which are considered to have superior bargaining power in their upstream and downstream because of their diversification benefits and interlinks of the group affiliated firms and subsidiaries, surprised with the most negative CAAR on the event day and event window period which could recover the least till the next 40 days of trading amongst others.