International Journal of Engineering and Management Research

1. Introduction

Calling a stock “defensive” can mean three quite different things in practice. It can mean that the stock moves less than the overall market on an average day (low market beta). It can mean that the stock's own price simply does not move around much (low volatility). Or it can mean that the stock protects an investor's capital when markets fall sharply (limited drawdown and positive returns in a crisis). A stock can satisfy the first definition while badly failing the other two, and the difference matters a great deal to how the label “defensive” is used in practice, including in discussions of Indian pharmaceutical equities.

Existing Indian research gives good reasons to examine these distinctions rather than assume the whole sector behaves in one uniform way. Studies of the pharmaceutical sector around COVID-19-related news and interventions report mixed, firm-specific responses rather than a single sector-wide pattern (Behera & Rath, 2021; Kumari, Rai, & Pandey, 2023; Lamba, Aggarwal, Gupta, & Joshipura, 2024). These studies examine short windows around specific announcements; on their own they say little about how a pharmaceutical holding behaves over a longer, multi-year investment horizon, or whether it improves several dimensions of portfolio risk at the same time.

This study examines five companies – Sun Pharmaceutical Industries Limited, Cipla Limited, Dr. Reddy's Laboratories Limited, Zydus Lifesciences Limited and Lupin Limited – over 1 April 2021 to 31 March 2025. The sample is fixed by that brief rather than drawn at random from the sector, and the paper does not claim the five firms represent the wider industry. The paper asks three plain questions: (i) does each stock move less than one-for-one with the market; (ii) does its sensitivity to falling markets differ from its sensitivity to rising markets; and (iii) does adding a pharmaceutical allocation to a market-index holding actually reduce realized portfolio risk, on more than one measure of risk.

Before addressing those questions, the price data itself required attention and was checked, line by line, against official exchange sources. I found that, from April 2022 onward, the Sun Pharma price series had been overwritten with the benchmark index's own values for roughly two-thirds of the study window – an error that, left uncorrected,

This is the reason the present paper reports standalone volatility, market beta and portfolio-level variance as three separate results rather than assuming that a high ranking on one measure implies a similar ranking on the others; a change in maximum drawdown, in particular, depends on the sequence of returns and does not follow automatically from a lower variance.

A related literature separates ordinary co-movement from co-movement specifically during market stress. Ang, Chen, and Xing (2006) find that US stocks with greater sensitivity to market declines earn a return premium, defining “downside” relative to the market's own average return. Baur and Lucey (2010) distinguish an asset that is broadly uncorrelated with the market (a diversifier) from one that specifically holds its value during a crisis (a safe haven), using gold as their setting; the distinction between the two is conceptual and does not, by itself, imply anything about Indian pharmaceutical shares. Working with Indian data, Janani Sri, Kayal, and Balasubramanian (2022) examine whether individual NIFTY 50 constituents behave as safe havens using formal dependence models. That work is a reminder that a “safe haven” claim requires its own dedicated test; the present paper's more limited stress-day comparison, described below, does not attempt to replicate that test and does not claim safe-haven status for any of the five stocks.

A second strand looks specifically at Indian pharmaceutical shares around COVID-19-related news. Behera and Rath (2021) report generally positive average pharmaceutical returns during the pandemic period alongside considerable firm-level variation. Kumari et al. (2023), covering 74 Indian pharmaceutical stocks, find that announcements were, on average, associated with negative abnormal returns and that smaller firms were more exposed. Lamba et al. (2024) study 77 intervention announcements by 46 listed firms and report a positive abnormal return in the days before an announcement. These are event studies of short, announcement-specific windows; they speak to firm-level news reactions rather than to a stock's ordinary sensitivity to the broad market over a multi-year period, which is the question this paper asks.

(NSE, n.d.-a), and the market benchmark is the BSE SENSEX 50 price index (BSE Indices, n.d.). After adding the 31 March 2021 closing observation needed to compute the first return, the study window yields 993 aligned trading dates and 992 daily returns, split across four fiscal years: 248 in FY2021/22, 249 in FY2022/23, 246 in FY2023/24 and 249 in FY2024/25. The final trading observation is 28 March 2025, since 31 March 2025 fell on an exchange holiday (NSE, 2024); no return is invented for that date, and official special weekend trading sessions are retained rather than discarded. Both stock and index prices exclude cash dividends, so the reported growth rates describe price performance rather than an investor's total, dividend-inclusive return.

3.2 Data Checks and Corrections

Before any statistic was computed, the daily closing price was checked against the official sources above. Three material issues were found and corrected, summarized in Table 1. The Sun Pharma correction relies on the same official NSE price history already cited above (NSE, n.d.-a); the Dr. Reddy's split is confirmed independently by the company's own regulatory filing (Dr. Reddy's Laboratories Limited, 2025) and by NSE's corporate-action record (NSE, n.d.-b); and the Zydus ticker change is confirmed by NSE's own circular announcing the renaming (NSE, 2022). Every affected series was replaced or adjusted using these official sources.

The importance of this step is not merely procedural. Section 5.7 shows that repeating the full analysis on the original, uncorrected data reverses the paper's central conclusion about the portfolio's market beta. In this application, careful source verification mattered more to the result than any subsequent choice of statistical technique – a point worth stating plainly, since it is easy for a more sophisticated model to be applied to data that was never checked.

The “equal-weight pharma portfolio” holds all five stocks at a fixed 20% weight each, rebalanced daily; its daily return is simply the average of the five stocks' returns. The “80/20 blend” combines an 80% holding in SENSEX 50 with a 20% holding in the equal-weight pharma portfolio. Both are frictionless, paper calculations – chosen to illustrate a diversification question, not to describe an actual tradable fund or an optimized weight – and this limitation is returned to in Section 8.

Cumulative wealth is tracked by compounding daily returns from a starting value of one on 31 March 2021. Total return is the resulting terminal wealth minus one, and the compound annual growth rate (CAGR) restates that total return as an equivalent constant annual rate over the four fiscal years in the window. Annualized volatility is the standard deviation of daily returns multiplied by the square root of 252 (the approximate number of trading days in a year); downside deviation applies the same annualizing step to only the negative daily returns. Maximum drawdown is the largest percentage decline from any prior peak in the cumulative wealth series, capturing the worst peak-to-trough loss an investor holding the position would have experienced within the window.

4. Method

The methods used here are deliberately restricted to tools taught in an introductory finance or statistics course: descriptive statistics, Pearson correlation and ordinary least-squares (OLS) regression. This choice is made because the underlying research questions – is beta below one, does it differ across up and down markets, and does a blended allocation reduce realized risk – do not require a more elaborate econometric apparatus to answer meaningfully once the input data has been verified. Where a simplification carries a genuine limitation (for example, the possibility that daily returns are not independent of one another, or that their variance changes over time), that limitation is stated plainly in Section 8 rather than addressed with an additional layer of technical machinery.

4.1 Market Beta

For each stock and for the equal-weight portfolio, market beta is estimated by regressing the asset's daily return on a constant and the contemporaneous SENSEX 50 daily return, using ordinary least squares. The slope coefficient (beta) measures the asset's average linear sensitivity to the market:

50's own daily returns, expressed as a percentage change (variance of the blend divided by variance of the index, minus one); a negative number means the blend's realized daily variance was lower than the index's alone. Because a single full-period number could be an artefact of one unusual period, this comparison is repeated on two sub-samples – the first two fiscal years against the last two, and separately, each of the four fiscal years on its own – rather than relying on a single simulated resampling procedure. Consistency of the sign and rough magnitude of the variance reduction across these independent sub-periods is treated as the robustness check, in place of a bootstrap confidence interval; both approaches ask the same underlying question, but splitting the actual, observed sample into independent sub-periods is more directly interpretable and easier to reproduce by hand.

4.5 Sensitivity to the Data Correction

Finally, to quantify how much the Section 3 data corrections mattered, the entire analysis is repeated once more on the original, uncorrected price series (that is, without removing the Sun Pharma duplication and without adjusting for Dr. Reddy's split), using the same starting baseline. This is a plain sensitivity comparison – the same simple calculations applied to two different input datasets – rather than a new technique in its own right.

5. Results

5.1 Returns, Volatility and Drawdowns

Table 2 and Figure 1 show wide differences in outcomes across the five stocks. Sun Pharma has the highest price CAGR (30.52%), while Dr. Reddy's has the lowest (6.09%), below SENSEX 50's own 12.40%. The equal-weight pharma portfolio grows at 19.30% a year on this price basis – a favourable outcome, but one that is not, by itself, evidence of lower risk, and one that Section 5.6 shows is concentrated in a single strong fiscal year rather than spread evenly across the period.

Every individual stock is more volatile than SENSEX 50: annualized volatility ranges from 20.84% (Sun Pharma) to 28.06% (Lupin), compared with 13.81% for the index. The equal-weight portfolio's volatility (17.62%) is lower than any single constituent, consistent with ordinary diversification across five imperfectly correlated stocks, but it still exceeds the index's own volatility. Its downside deviation (11.51%) is likewise higher the index's (9.85%).

5.2 Market Beta

Table 3 reports beta estimates from 0.308 (Cipla) to 0.669 (Zydus); all five confidence intervals sit entirely below one, and the one-sided test rejects a beta of at least one for every stock (p<0.001 in each case). The equal-weight portfolio's beta is 0.538, with a 95% confidence interval of 0.465–0.610. On this evidence, all five stocks and the resulting portfolio moved less than one-for-one with SENSEX 50 over the study window.

At the same time, the explanatory power of this one-factor comparison is limited: R² ranges from just 0.033 (Cipla) to 0.150 (Sun Pharma), and 0.177 for the portfolio. In plain terms, the great majority of each stock's day-to-day movement is unrelated to the broad market's own movement on that day – which is exactly why a low beta does not translate into low overall volatility, as Table 2 already shows.

Figure 2: Left: estimated market beta and 95% confidence interval for each stock and the portfolio, with the beta = 1 reference line. Right: drawdown from each series' own preceding wealth peak, for SENSEX 50, the equal-weight pharma portfolio and the 80/20 blend.

Table 3: Correlation with, and OLS market beta against, SENSEX 50. The 95% confidence interval and the p-value for the one-sided test of beta ≥ 1 use ordinary (non-adjusted) OLS standard errors; see Section 4.1 and Section 8 for the corresponding limitation.

Table 5: Average return and share of positive daily returns for each asset on the same 50 trading dates on which SENSEX 50 recorded its lowest returns (the empirical worst 5%).

5.5 The Diversification Comparison and its Stability

The 80/20 blend's own annualized volatility (12.93%) is below the index's (13.81%), and its variance is about 12.3% lower than the index's variance over the full period. This is not an isolated full-period artefact: splitting the sample into a first half (April 2021 to March 2023) and a second half (March 2023 to March 2025) gives variance reductions of 12.8% and 11.5% respectively, and repeating the comparison separately within each of the four fiscal years gives reductions ranging narrowly from 11.1% to 13.6% (Table 6). The equal-weight pharma portfolio held on its own, by contrast, has a variance roughly 63% above the index's – so it is specifically the blending of an imperfectly correlated pharmaceutical component (correlation of 0.421 with SENSEX 50) into a larger market holding, and not any property of the pharmaceutical portfolio in isolation, that produces the variance reduction.

Table 6: Variance of the 80/20 blend (and, where shown, the equal-weight pharma portfolio) relative to SENSEX 50's own variance,

More importantly for this paper's central claim, the uncorrected portfolio's estimated market beta is 1.16 – above one, the opposite of this paper's finding.

The reversal happens because the corrupted Sun Pharma segment briefly tracked the benchmark almost exactly, mechanically inflating both that stock's apparent market sensitivity and its measured volatility. This single comparison illustrates the paper's central methodological point directly: no amount of statistical sophistication applied to the uncorrected series would have produced a reliable answer to this study's research questions, whereas even the simple tools used throughout this paper give a stable, plausible answer once the data itself is verified.

Table 8: The same calculations applied to the uncorrected original price series versus the source-verified series used throughout this paper.

6. Discussion

The results support one specific, narrow meaning of “defensive”: lower average linear sensitivity to SENSEX 50. They do not support the broader claim that these pharmaceutical stocks are, in general, low-risk holdings. The two claims are logically distinct – beta compares an asset's co-movement with the market to the market's own variance, and says nothing directly about the asset's total volatility or its exposure to company- or sector-specific events. In this sample, correlations with the market are modest and R² values are low, which is exactly the condition under which a below-one beta can coexist comfortably with high total volatility and deep individual drawdowns, as it clearly does here.

The stress-day results sharpen this same distinction. The pharmaceutical portfolio loses less than the index, on average, on the market's worst days, yet it still loses money on the great majority of them, and its own maximum drawdown across the full period is considerably larger than the index's.

7. Implications for Investors, Advisers and Policy

The distinction this paper draws between market beta, standalone volatility and drawdown is not only an academic one. India's retail equity investor base has grown rapidly in recent years: SEBI reported roughly 194 million demat accounts by mid-2025, and a substantial share of new accounts continues to open through direct-trading apps and thematic or sector-focused mutual fund schemes, among which pharmaceutical funds are commonly marketed using language such as “defensive” or “stable” (Business Standard, 2025). The results in Sections 5 and 6 speak directly to what such a label can, and cannot, support, and this section restates the findings for four audiences who are unlikely to read a regression table but who make decisions based on exactly this kind of evidence.

7.1 Retail investors and their Financial Advisers

A retail investor choosing a pharmaceutical stock or fund because it is described as defensive should understand precisely what that word is likely to mean here: a tendency to move less than the broad market on an average day (an estimated beta of 0.31 to 0.67 across the five stocks in this sample). It does not mean the investment has low volatility in absolute terms – every stock's own annualized volatility (20.8% to 28.1%) is well above the market index's (13.8%) – and it does not mean the investor is protected from large losses: four of the five stocks experienced a peak-to-trough decline of more than 25% within this four-year window, and two exceeded 50%. A financial adviser using the word “defensive” to describe a pharmaceutical holding could reasonably be asked, and should be able to answer, which of these three distinct things is meant, ideally with a specific number attached to each rather than a single descriptive label.

7.2 Mutual Fund and Portfolio Managers

For a fund or portfolio manager weighing a modest pharmaceutical allocation alongside an index-tracking core holding, this study offers a concrete, if narrow, benefit: a 20% allocation reduced realized portfolio variance by roughly a tenth to an eighth of the index's own variance, and this reduction held up consistently across both halves of the sample and every individual fiscal year (Table 6).

8. Limitations and Future Research

The five-stock sample is inherited from an existing research brief rather than drawn using an explicit selection rule, so it cannot support claims about prevalence across the wider Indian pharmaceutical sector, and the possibility of survivorship or selection effects in how the five firms came to be chosen cannot be ruled out. The four-fiscal-year window is also finite: FY2023/24's unusually strong pharmaceutical returns and the subsequent drawdowns shown in Table 7 indicate that results vary meaningfully over time, and the sub-period checks in Table 6 describe that variation without constituting a genuine out-of-sample forecast test.

All reported price returns exclude cash dividends. Matching the stocks to a price index (rather than a total-return index) avoids one obvious mismatch, but dividend omission still limits any conclusion about an investor's actual realized wealth, particularly around ex-dividend dates. A useful extension would reconstruct dividend-inclusive returns and compare them with a matching total-return benchmark, and would also compare results against a pharmaceutical sector index (such as NIFTY Pharma) alongside the broad-market index used here, to separate general market exposure from sector-specific exposure.

On the statistical side, this paper deliberately uses ordinary least-squares standard errors and does not adjust for the possibility that daily stock returns are not fully independent of one another or that their volatility changes over time (both plausible features of daily equity data). Where these features are present, ordinary standard errors can understate true uncertainty to some degree; the paper's headline finding – a below-one beta for every stock, with confidence intervals well clear of one – is unlikely to be overturned by a more technical correction given how far the point estimates sit from the boundary, but a reader should treat the exact reported p-values and interval widths as approximate rather than exact. Because five closely related beta-below-one tests are reported side by side without a formal adjustment for testing several stocks at once, individual p-values in Table 3 should likewise be read as indicative evidence for each stock rather than as controlling one combined error rate across the whole set of five.

individually or as a portfolio, as low-risk or capital-protective holdings.

The paper's contribution beyond this specific sample sits in three places. For the literature on Indian equity data practice, it offers a documented, reproducible case of how an unverified pricing error can reverse a headline finding, and a corresponding argument for treating source verification as a mandatory first step in any Indian equity study. For retail investors, financial advisers and fund managers, it translates that evidence into a concrete, numeric account of what a “defensive” pharmaceutical holding can and cannot be expected to do. And for investor-education and disclosure policy, it points to a specific, low-cost step – reporting volatility and maximum drawdown alongside the existing composite risk label – that could be adopted without new regulation or new data infrastructure, and that would let an investor see precisely where a low market beta does, and does not, protect them.

References

[1] Ang, A., Chen, J., & Xing, Y. (2006). Downside risk. The Review of Financial Studies, 19(4), 1191–1239. https://doi.org/10.1093/rfs/hhj035

[2] Baur, D. G., & Lucey, B. M. (2010). Is gold a hedge or a safe haven? An analysis of stocks, bonds and gold. The Financial Review, 45(2), 217–229. https://doi.org/10.1111/j.1540-6288.2010.00244.x

[3] Behera, C., & Rath, B. N. (2021). The COVID-19 pandemic and Indian pharmaceutical companies: An event study analysis. Bulletin of Monetary Economics and Banking, 24(Special Issue), 1–14. https://doi.org/10.21098/bemp.v24i0.1483

[4] BSE Indices. (n.d.). BSE SENSEX 50: Index information and historical values [Data set]. Retrieved September 9, 2026, from https://www.bseindices.com/indices-details/code/98

[5] Business Standard. (2025, July 15). Retail participation in mkts increase; demat accounts surge to 194 mn: Sebi. https://www.business-standard.com/markets/news/retail-participation-in-mkts-increase-demat-accounts-surge-to-194-mn-sebi-125071500934_1.html

[15] National Stock Exchange of India. (n.d.-a). Security-wise price-volume data: SUNPHARMA, CIPLA, DRREDDY, ZYDUSLIFE and LUPIN, EQ series, March 31, 2021–March 31, 2025 [Data set]. Retrieved September 9, 2026, from https://www.nseindia.com/report-detail/eq_security

[16] National Stock Exchange of India. (n.d.-b). Corporate actions: SUNPHARMA, CIPLA, DRREDDY, ZYDUSLIFE and LUPIN [Data set]. Retrieved September 9, 2026, from https://www.nseindia.com/companies-listing/corporate-filings-actions

[17] Saraf, M., & Kayal, P. (2023). How much does volatility influence stock market returns? Empirical evidence from India. IIMB Management Review, 35(2), 108–123. https://doi.org/10.1016/j.iimb.2023.05.004

[18] Securities and Exchange Board of India. (2021, August 5). Disclosure of risk-o-meter of scheme, benchmark and portfolio details to the investors (Circular SEBI/HO/IMD/DF3/CIR/P/2021/024). https://www.sebi.gov.in/legal/circulars/aug-2021/disclosure-of-risk-o-meter-of-scheme-benchmark-and-portfolio-details-to-the-investors_52262.html