What Volatility Tells Us About the Casablanca Stock Exchange and Investor Behavior

An empirical study conducted over three years shows that the MASI index is becoming more sensitive to emotions and external shocks, while still exhibiting a remarkable ability to maintain equilibrium. In other words, despite apparent volatility, the structure of the Moroccan market remains solid and aligns with long-term fundamentals. The empirical analysis of the MASI reveals a complex dynamic where volatility and a return to equilibrium coexist. This phenomenon, known as volatility cyclicity, is observed across various asset classes worldwide. Phases of high volatility are followed by calmer periods and vice versa. We studied the daily returns of the MASI—i.e., the daily fluctuations of the index—between October 2022 and October 2025. An initial test, the augmented Dickey-Fuller test, confirms their stationarity, a technical term indicating that prices naturally revert to their long-term trend. In simpler terms, the market corrects its excesses: extreme highs and lows eventually stabilize. However, the distribution of these returns does not follow a normal distribution, meaning it does not adhere to a classic Gaussian curve. The Jarque-Bera test reveals pronounced leptokurtosis (kurtosis = 8.3) and negative skewness. These technical terms indicate that there are more extreme downward events than upward ones. This reflects typical behavior in emerging markets, where reactions are more emotional and limited liquidity amplifies price variations. In essence, declines are more severe than increases. Another test conducted on daily returns, the Ljung-Box test, shows significant autocorrelation of returns up to the 18th day. This means that past performances still influence future ones: if the market declines for several consecutive days, this trend may continue before reversing. In theory, a fully efficient market should not exhibit this type of dependence; every new piece of information should be immediately reflected in prices. However, in Casablanca, as in many emerging markets, information circulates more slowly, and imitation behaviors amplify these cycles. Additionally, the Engle test (ARCH-LM) confirms the presence of conditional heteroscedasticity, meaning that volatility changes over time. Calm periods alternate with episodes of high instability, often linked to monetary policy announcements, corporate earnings results, or geopolitical tensions. The estimation of the GARCH(1,1) model, used to measure the persistence of fluctuations, shows that the volatility of the MASI remains high but stationary, thus contained in the long term. The ARCH (α₁ = 0.20) and GARCH (β₁ = 0.70) coefficients indicate that 90% of current volatility is explained by recent shocks and market memory. This phenomenon, known as volatility clustering, means that periods of high agitation tend to group together, just as calm periods do. Once nervousness sets in, it takes time to dissipate. In clear terms, after a shock—whether economic, political, or psychological—the stock market does not immediately return to normal. It retains a "memory" of fear or euphoria, a typical behavior in markets where investor confidence takes time to rebuild. These results also reflect the collective psychology of investors operating on the Casablanca Stock Exchange. The autocorrelation and persistent volatility indicate a tendency toward mimicry: many traders follow recent movements rather than anticipating fundamentals. The negative skewness of returns reveals a loss aversion: bad news triggers much stronger reactions than good news. This behavioral bias, well-known in finance, often leads to panic selling, even when fundamentals remain solid. The market's emotional memory, measured by the persistence of volatility in the GARCH(1,1) model, confirms that episodes of euphoria or fear continue to influence decisions long after their occurrence. Thus, the MASI behaves like a living organism: it learns, adapts, but does not forget. For a savvy investor, these findings open up concrete opportunities. The presence of autocorrelation and persistent volatility suggests that it is possible to exploit phases of market overreaction: buying when corrections are exaggerated and reducing positions after prolonged increases. While this is easier said than done, the slow dissipation of shocks calls for a highly polarized risk management approach. Investors must choose between being extremely reactive to bad news by quickly cutting positions or relying on the structural resilience of the MASI and its ability to return to equilibrium despite storms. In this case, they should remain passive and endure panics while waiting for them to pass. Ultimately, understanding market psychology and its internal dynamics becomes a competitive advantage. In an environment where fear and euphoria too often dictate prices, discipline and patience remain the best weapons for a rational investor. In conclusion, momentum strategies can be extremely profitable in such markets where mimicry is a pronounced bias. Choosing "in-fashion" stocks and holding them as long as possible would yield much more than selling stocks as soon as they start to seem expensive. This may be difficult for advocates of fair value to accept. However, these investors, if they are willing to hold their stocks long-term and not react to episodes of volatility, will find their performance improved. Volatility Models: ARCH(1) & GARCH(1,1) Formulations — relationships, hypotheses, conditions ARCH(1) — Engle (1982) ARCH Model: The ARCH model, introduced by Engle (1982), posits that the variance of the error at time t depends on the squares of past errors. In other words, current volatility is conditioned by previous shocks. Relationship: $$\begin{aligned} & r_t = \mu + \varepsilon_t,\\ & \varepsilon_t = \sigma_t z_t,\quad z_t \sim \mathcal{N}(0,1),\\ & \sigma_t^{2} = \alpha_0 + \alpha_1\,\varepsilon_{t-1}^{2} \end{aligned}$$ Hypotheses / conditions: Positivity: \(\alpha_0 > 0\), \(\alpha_1 \ge 0\). Stationarity: \(\alpha_1 < 1\). GARCH(1,1) — Bollerslev (1986) GARCH Model: The GARCH(1,1) model, proposed by Bollerslev (1986), generalizes the ARCH model by introducing a dependence of conditional variance on past shocks and volatility. Relationship: $$\begin{aligned} & r_t = \mu + \varepsilon_t,\quad \varepsilon_t|\Omega_{t-1} \sim \mathcal{N}(0,\sigma_t^2),\\ & \sigma_t^{2} = \omega + \alpha_1\,\varepsilon_{t-1}^{2} + \beta_1\,\sigma_{t-1}^{2} \end{aligned}$$ Hypotheses / conditions: Positivity: \(\omega > 0\), \(\alpha_1 \ge 0\), \(\beta_1 \ge 0\). Stationarity: \(\alpha_1 + \beta_1 < 1\).