Social Media Feeds Investor Biases When the Rupee Wobbles, Survey of Indian Mid-Cap Traders Finds

When the rupee slides against the dollar, India’s retail investors do not simply recalculate. According to a new survey-based study of 519 active mid-cap investors, many of them appear to lean harder on the crowd, on their own judgment, and on whatever dramatic currency headline crossed their feed most recently. The research, published in Discover Global Society by Devendra Kumar Shukla, Narendra Singh Bohra, and Vinay Kandpal of Graphic Era (Deemed to be University) in Dehradun, offers one of the first structured looks at how three classic behavioural biases—herding, overconfidence, and the availability heuristic—travel together through the fast-growing, digitally connected world of Indian mid-cap investing.

The timing matters. Individual investors’ net inflows into India’s National Stock Exchange cash market hit a record high in 2024 after years of volatile but steadily growing retail activity, and much of that money has flowed into mid-cap stocks. The mid-cap segment is widely regarded as more exposed to macroeconomic shocks than the large-cap universe, because mid-cap firms typically carry less international hedging and attract far less analyst coverage. That exposure became visible through 2025, when rupee depreciation against the US dollar coincided with foreign portfolio outflows and comparatively weaker mid- and small-cap performance. The result is a rapidly expanding, relatively inexperienced retail base trading in the segment most sensitive to exactly the kind of currency turbulence the study probes.

Methodologically, the researchers distributed a structured online questionnaire between January and March 2025 through investment forums, trading communities, and social-media investment groups, retaining 519 valid responses after removing incomplete submissions. All six core constructs—herding, overconfidence, availability-heuristic bias, financial literacy, social-media reliance, and investment decisions—were measured on five-point Likert scales adapted from previously validated instruments. The team then ran a confirmatory factor analysis followed by a structural path model in IBM SPSS Amos, using 1,000-sample non-parametric bootstrapping to generate confidence intervals for every direct path and every indirect, or mediating, effect. The measurement model showed strong psychometrics, with composite reliability above 0.70 and average variance extracted above 0.50 for every construct, and the structural model posted unusually strong fit indices, including a comparative fit index of 0.997 and an RMSEA of just 0.013.

The headline finding concerns social media. Heavier reliance on social platforms for investment information was positively associated with all three biases, not merely with herding, which is how earlier Indian survey work typically framed the relationship. The authors interpret this as consistent with social media functioning less as a single herding channel and more as a general amplifier of a fast, vivid, socially validated information environment—one in which collective signals, inflated self-assessments, and emotionally salient currency headlines can all flourish. Prior studies lend support to the mechanism: research on digital platforms has documented sentiment contagion that pushes prices beyond what fundamentals justify, and peer validation on investment apps has been linked to amplified overconfidence among millennial investors.

Financial literacy told a more complicated story. It was significantly and negatively associated with herding and with the availability heuristic, suggesting that knowledgeable investors are better at distinguishing transitory price moves from fundamental change and less swayed by whatever vivid news happens to be most memorable. But the path from literacy to overconfidence was not statistically distinguishable from zero, with a bootstrapped 95 percent confidence interval spanning −0.19 to 0.01. The authors are careful to note that this is a failure to establish an association, not proof that none exists. Still, the pattern fits what behavioural-finance researchers call the competence effect: investors who believe themselves knowledgeable tend to take on more risk regardless of whether that self-assessment is accurate, because overconfidence is rooted in self-perception rather than in information-processing skill.

All three biases, in turn, were positively associated with investment-decision scores under hypothetical currency-volatility scenarios, with herding showing the strongest association, followed by overconfidence and then the availability heuristic. Because the model treats each bias as a function of social-media reliance and financial literacy, the structure implies six indirect pathways to investment decisions. Bootstrapped confidence intervals excluded zero for five of the six, supporting a statistical mediation pattern through all three biases for the social-media pathway, and through herding and the availability heuristic—but not overconfidence—for the financial-literacy pathway. The authors stress that, in a cross-sectional design, these are statistical patterns consistent with the hypothesised sequence, not demonstrated causal mechanisms; establishing temporal precedence would require longitudinal or experimental data.

Perhaps the most theoretically provocative result is the one that did not appear. Prospect Theory’s reflection effect predicts that people become risk-seeking when outcomes are framed as losses and risk-averse when framed as gains. Applied to currency markets, a depreciating rupee should feel like a loss and an appreciating rupee like a gain, producing systematically different risk responses. The researchers tested this directly with a paired-samples comparison of responses to loss-framed depreciation scenarios against gain-framed appreciation scenarios. The difference was not statistically significant: t(518) = 0.60, p = .55, Cohen’s d = 0.03—an effect size close to nil. The authors offer three non-exclusive explanations: the reflection effect may hold more reliably for concrete, personally realised gains and losses than for hypothetical macro-level scenarios; investors who have already self-selected into a volatile segment may have a flatter risk-response range than the general population; or the four-item instrument may simply lack the sensitivity an incentive-compatible experiment would provide.

The authors are notably candid about the study’s limits. The sample was recruited purposively from online forums and social-media groups, meaning respondents are more digitally engaged and more actively trading than the broader Indian retail population, including semi-urban and rural investors who rely on traditional broker networks. Currency volatility was operationalised through perceptions of hypothetical scenarios rather than objective exchange-rate data, and the investment-decision measure captures stated intentions rather than observed trades. All constructs were self-reported at a single time point, and although two common-method-bias diagnostics—a Harman single-factor test, in which the largest factor accounted for only 34.3 percent of variance, and a full-collinearity test with all construct-level VIFs below 1.7—were reassuring, they are diagnostic rather than definitive. The team even flags its own near-ceiling model fit as unusually strong for applied survey work and asks readers to weigh it alongside the substantive results rather than treat it as proof of a uniquely correct model.

Even within those boundaries, the practical implications are pointed. The finding that social-media reliance relates to overconfidence and availability-driven reactivity, not just herding, suggests that regulatory attention to digital investment content—such as India’s securities regulator SEBI’s ongoing work on finfluencer disclosure—may need to address a broader set of behavioural risks than herding-specific messaging. The null literacy-overconfidence association hints that investor education built purely on declarative financial knowledge may be a poor fit for tackling overconfidence, and that metacognitive or self-assessment components could be a more relevant design target, though the study tested no intervention directly. For financial advisors guiding mid-cap clients through currency turbulence, the results argue for diagnosing which specific bias is most active for a given client rather than treating behavioural bias as a single undifferentiated problem. The authors frame all such applications as directions worth testing, not validated prescriptions, and call for longitudinal designs, objective exchange-rate data, probability-based sampling, and incentive-compatible framing experiments to determine whether the associations they report hold beyond this digitally active sample.

Subject of Research: Behavioural biases among Indian retail investors responding to perceived currency volatility in the mid-cap equity market

Article Title: Behavioural biases and retail investor responses to perceived currency volatility in the Indian mid-cap equity market

Article References: Shukla, D. K., Bohra, N. S., & Kandpal, V. (2026). Behavioural biases and retail investor responses to perceived currency volatility in the Indian mid-cap equity market. Discover Global Society, 4(1), Article 259. https://doi.org/10.1007/s44282-026-00618-w

Image Credits: AI Generated

DOI: 10.1007/s44282-026-00618-w

Keywords: behavioural finance, herding, overconfidence, availability heuristic, financial literacy, social media, currency volatility, Indian mid-cap equities, retail investors, Prospect Theory, structural equation modelling, rupee depreciation