The Information Dynamics of Insider Intent: How Reporting Inversions (Form 144) Mask Informational Rents in Insider Sales (Form 4) Krishna Neupane kneupan@gmu.edu ORCID: 0000-0003-1689-0557 Abstract This study identifies and quantifies a significant informational friction embedded in the SEC Form 144 disclosure regime, characterized as predictive decoupling. Drawing on a theoretical foundation of welfare economics, the article argues that the current reporting inversion —where trade execution (Form 4) frequently precedes the public notice of intent (Form 144)—violates the conditions for Pareto efficiency by inducing non-symmetric pricing. Utilizing an event-study framework of intent-to-sell windows, the analysis examines cases where insiders file a notice of proposed sale but fail to execute within the statutory 90-day period. The machine learning audit reveals a persistent 52.4% opacity rate, where aborted signals remain statistically indistinguishable from routine executions, creating a structural information ceiling that prevents the market from exhausting the signal’s informational content. Contrary to the traditional small-firm effect, the study documents a large-cap significance paradox- while small-cap portfolios yield higher absolute abnormal returns (32.21 bps), statistically significant alpha is concentrated in large-cap firms (14.49 bps, p = 0.021 p=0.021 ). The results suggest that Institutional Salience enables more reliable processing of this negative non-event when reputational costs are maximized. Cross-sectional tests confirm that prior idiosyncratic volatility serves as a signal amplifier, with causal estimators identifying an illiquidity jump of up to 2.63 times. To mitigate this market failure, the study proposes a mandatory execution confirmation (Form 144-A) to transition the regime toward bilateral accountability, converting a predictive blind spot into a verifiable data stream and restoring the informational symmetry requisite for efficient capital allocation. © Krishna Neupane 2026. All rights reserved. This working paper is part of on going review for publication. No part of this publication may be reproduced without prior permission. https://krishpn.github.io Keywords: insider trading, Form 144, Form 4, welfare economics, informational rents, market failure, causal forest, machine learning JEL Classification: G14, G11, D61, K22, C45, C21 1 Introduction The First Fundamental Theorem of Welfare Economics (FTWE) posits that competitive markets achieve Pareto efficient allocations under the condition of symmetric information. Insider trading intent, generated privately and subsequently disclosed via SEC Form 144 (Notice of Proposed Sale of Securities), represents a fundamental violation of this condition. While the Form 144 mandate is intended to restore market symmetry, the temporal lag and execution uncertainty surrounding proposed sales prevent the immediate satisfaction of FTWE conditions. The resulting persistent information asymmetry induces a market failure characterized by non-Pareto efficient pricing. Consequently, equity prices during the filing window fail to reflect fundamental value, representing an inefficient allocation of capital. This theoretical breakdown is operationalized by the current regulatory environment. As documented by Franzen, Li, and Vargus (2013), the post-Sarbanes-Oxley (SOX) era is characterized by a reporting inversion; trade execution (Form 4) is mandated within two business days, whereas the notice of intent (Form 144) is frequently delayed. Historically, Franzen, Li, and Vargus (2013) found that while Form 144 preceded Form 4 in 93.7% of pre-SOX cases, this frequency declined to 17.5% post-SOX. This inversion serves as the empirical mechanism for market failure; at the moment of execution, the prevailing market price cannot incorporate the restricted nature of the securities being sold, resulting in a significant transparency deficit. Insiders may exploit this regulatory friction through strategic disclosure sequencing. Although post-SOX regimes sought to disincentivize reporting delays, insiders utilize the Filing Gap to mitigate market impact. Prior research suggests that the stringency of disclosure regimes directly influences opportunistic managerial trading (Cheng, Nagar, and Rajan (2007), Huddart, Hughes, and Levine (2001), Grossman (1981), Grossman and Stiglitz (1980)). Furthermore, heightened scrutiny in the post-SOX environment may increase perceived legal jeopardy, theoretically constraining the profitability of informative trades. This article argues that the reporting inversion facilitates a sophisticated circumvention of these constraints, manifested as market sentiment bias. Under the contrarian framework of Lakonishok and Lee (2001) and the decoding logic of Cohen, Malloy, and Pomorski (2012), insiders acting as behavioral contrarians typically execute sales during periods of price appreciation. Because Form 144 is frequently reported ex-post (Franzen, Li, and Vargus (2013)), insiders capitalize on signal misclassification: the market may interpret a high-signal opportunistic trade as a routine, liquidity-driven exit. By integrating with uninformative trading volume, the insider preserves trade profitability while minimizing the visibility that triggers regulatory or market scrutiny. The existence of this gap suggests delayed price discovery. As Kothari, So, and Verdi (2016) note, if the initial market reaction to a signal is biased or insufficient, subsequent price drifts or reversals become predictable. By formalizing the 90-day non-execution event—defined as the expiration of a Form 144 notice without a corresponding trade—this analysis shifts the focus from realized transactions to omitted trades. In this context, the market’s initial failure to interpret the intent ensures that informational content is not exhausted upon filing. Instead, it manifests as a predictable negative drift ( H_{1} ) or a subsequent reversal upon the trade’s abandonment ( H_{2} ). This study identifies a persistent non-execution frequency of 52.4%, where unexecuted notices remain statistically indistinguishable from realized trades, establishing a structural barrier to market efficiency. This allows for the quantification of informational rents: a window of market inefficiency where insiders operate during periods of delayed price correction. This concept extends the work of Zhang (2006), who demonstrates that information uncertainty—proxied by fundamental volatility—amplifies investor underreaction. Consistent with the prediction that behavioral biases are most acute when assets are difficult to value, empirical results show that prior idiosyncratic volatility significantly amplifies the abnormal returns of unexecuted signals ( p=0.021 ). Finally, the informational potency of these signals challenges the established small-firm effect. While traditional literature (Lakonishok and Lee (2001), Seyhun (1986)) suggests insider signals are only informative in small-cap stocks, this research identifies a statistically significant 14.49 bps abnormal return ( p=0.021 ) in large-cap sub-samples. This suggests that the 90-day non-execution event possesses higher signal reliability in transparent, professionalized environments. Market efficiency in large-cap equities does not preclude the existence of abnormal returns; rather, it shifts the signal’s nature from a profit-seeking event to a high-stakes reputational one, where the market eventually corrects the information contained within the unexecuted intent. To account for complex, non-linear dynamics, this manuscript implements a multi-stage empirical framework. The analytical progression initiates with baseline univariate OLS correlations, advances to high-dimensional predictive modeling via Gradient Boosted Decision Trees (XGBoost), and culminates in a Causal Machine Learning (CML) ensemble. This methodological sequence is designed to isolate the causal mechanisms underlying market salience, ensuring that the documented effects extend beyond simple predictive correlation. The remainder of the article is organized as follows. Section 2 delineates the central hypotheses and theoretical contributions. Section 3 describes the three-phase methodological strategy: establishing statistical baselines, evaluating the predictive determinants of market liquidity, and identifying causal factors through heterogeneous treatment effects. Section 4 details the data construction process, which integrates insider trading records, executive profiles, and market-level financial metrics. Section 5 presents the empirical findings and subsequent analysis. Finally, Section 7 offers concluding remarks. 2 Hypothesis Development To empirically evaluate the welfare implications of information asymmetry, this article utilizes the structural reporting windows established by SEC Form 144 filings. Within the framework of the FTWE, these filings serve as proxies for non-public signals that temporarily impede the conditions requisite for a Pareto efficient competitive equilibrium. To isolate these informational effects, the research design partitions the dataset into two cohorts based on the subsequent filing outcome: Hypothesis 1 ( H_{1} ): The interval between the filing of Form 144 and the subsequent execution reported in Form 4 represents a period of incomplete price discovery. During this window, securities are expected to exhibit a persistent negative price drift as the market incorporates the information contained in the proposed divestment. This drift constitutes a measurable welfare cost resulting from information-induced price distortion. Hypothesis 2 ( H_{2} ): The expiration of a Form 144 window without a trade execution represents the withdrawal of the initial selling signal. This non-execution event triggers a price reversal as market participants adjust to the absence of the expected divestment. This correction facilitates a return to informational equilibrium by removing the price distortion established during the filing window. Table 1 summarizes the data partitioning and the expected signal transitions used to test the central hypotheses: Table 1 : Framework for Testing Hypotheses: Mapping Filing Outcomes to Predicted Price Behavior Data Group Economic Role (Signal Logic) Hypothesis Executed Sales Filing adds a signal \rightarrow Price Drift : Measures price distortion and the delay in market clearing during the active filing window. H_{1} Unexecuted Notices Expiration removes signal \rightarrow Price Reversal : Serves as a counterfactual to observe price correction after the selling intent is withdrawn. H_{2} 3 Methodology The empirical strategy is organized into a three-stage hierarchy designed to move from descriptive association to causal identification. This structure ensures that the documented market failures are not merely predictive correlations but structural features of the regulatory environment. The analysis initiates with univariate OLS and high-dimensional fixed-effects models to establish the baseline relationship between reporting intervals and abnormal returns. This stage verifies the existence of the H 1 negative drift and the H 2 price reversal, providing the initial statistical signal of market mispricing during the Form 144 filing window. To determine if the observed pricing inefficiency can be resolved by market participants, Phase II implements a cross-paradigm audit. This stage stress-tests the decodability of aborted intents across ten distinct architectures, ranging from linear Elastic Net to state-of-the-art FT-Transformers and Neuro-Symbolic TabNet. Rather than identifying feature importance, this phase seeks to identify the information ceiling—the point at which increasing model complexity fails to improve predictive recall—thereby proving that the 52.4% Opacity Rate is a structural property of the public data. The final stage employs a causal machine learning ensemble, utilizing double machine learning and generalized random forest. This framework isolates the Conditional Average Treatment Effect of an aborted trade. By satisfying the unconfoundedness assumption through high-dimensional controls and honest inference, Phase III confirms that the 1.3% pricing differential is a causal result of the abrupt disconnect, establishing a formal link between the reporting inversion and Pareto inefficiency. 3.1 Phase I: Baseline Statistical Significance and Univariate Analysis The analysis begins with a univariate test to evaluate market activity during the reporting interval—the temporal window of unresolved information between the notice of intent (Form 144) and the trade execution (Form 4). This is quantified as the integer duration in Equation 1 : G_{\text{days},i}=\text{Date}_{\text{Execution},i}-\text{Date}_{\text{Filing},i}, (1) where G_{\text{days},i} represents the duration of the informational (filing) gap for filing i , while \text{Date}_{\text{Execution},i} and \text{Date}_{\text{Filing},i} denote the calendar dates of the trade execution and the initial notice of intent, respectively. This duration serves as the active window for calculating the Cumulative Market Excess Return (CMER). To capture the total wealth effect occurring while the insider’s intent remains unreflected in market prices, the methodology employs geometric compounding. The CMER is derived by aggregating daily market excess returns ( R_{m,t} ) from the day following the filing ( F+1 ) through the date of execution ( E ) (Equation 2 ): CMER_{i}=\left[\prod_{t=F+1}^{E}(1+R_{m,t})\right]-1, (2) where CMER i is the compounded excess return for the i -th filing, and R_{m,t} represents the market return in excess of the risk-free rate on day t . By defining the window as [F+1,E] , the research ensures that the CMER reflects market appreciation occurring after the intent is filed but before the transaction is finalized. This approach utilizes geometric compounding to avoid the upward bias inherent in the arithmetic summation of daily returns, ensuring that the CMER accurately reflects the cumulative wealth effect during the reporting interval (Lakonishok and Lee (2001)). A one-sample t -test is performed against the null hypothesis H_{0}:\mu_{CMER}=0 to confirm the existence of significant market activity during the transaction cycle. A statistically significant positive CMER suggests that insiders strategically time executions to coincide with periods of market appreciation. This behavior potentially masks opportunistic intent (Cohen, Malloy, and Pomorski (2012)), as the idiosyncratic signal of the trade is subsumed by the broader market performance during the reporting interval. This behavior is further evaluated via the Divestment Intensity (SM i ), which is defined as the ratio of shares sold to total shares held (Equation 3 ): SM_{i}=\frac{Q_{\text{sold},i}}{Q_{\text{held},i}}, (3) where Q_{\text{sold},i} represents the number of shares sold in transaction i (adjusted for stock splits), and Q_{\text{held},i} represents the total number of shares held by the insider prior to the transaction. SM i distinguishes between routine liquidity rebalancing and high-conviction divestments. A high SM i coupled with a positive CMER identifies trades where informational opacity is most critical to preserving the insider’s execution price. 3.2 Carhart Four-Factor Event Study While the univariate CMER establishes the presence of price drift during the reporting interval, it does not isolate the idiosyncratic alpha generated by the insider’s private information. To address this, a firm-specific Carhart Four-Factor Event Study is employed to isolate the Abnormal Return (AR)—the portion of the return attributable specifically to the informational signal. Foundational studies, such as Seyhun (1986), demonstrate that simpler models like the CAPM introduce systematic biases in insider trading contexts. Furthermore, Lakonishok and Lee (2001) and Cohen, Malloy, and Pomorski (2012) caution that insider activity is often concentrated in size and value stocks and exhibits a strong contrarian bias. Consequently, the inclusion of the Momentum (UMD) factor is critical; as insiders frequently execute sales following price appreciation, the UMD factor serves to control for the “routine” component of the trade. By controlling for these factors, the methodology isolates the residual opportunistic signal that the market fails to fully incorporate during the reporting interval. The expected return, E [R_{i,t}] , is estimated by fitting the Carhart (1997) four-factor model for each security i in the database (see Equation 4 ): R_{i,t}-R_{f,t}=\alpha_{i}+\beta_{1,i}(R_{m,t}-R_{f,t})+\beta_{2,i}SMB_{t}+\beta_{3,i}HML_{t}+\beta_{4,i}UMD_{t}+\epsilon_{i,t}, (4) where R i,t is the return on security i at time t , R f,t is the risk-free rate, and SMB, HML, and UMD represent the size, value, and momentum factors, respectively. The coefficients ( \alpha_{i},\beta_{1\dots 4} ) are estimated over a 100-day window from T -120 to T -20 relative to the initial intent filing. The market reaction to both the intent (E 1 ) and the execution (E 2 ) is measured using a short-window event study centered on the respective filing dates. Cumulative Abnormal Returns ( CAR ) are calculated for the (0,+2) window to capture the immediate informational resolution: CAR_{i}(0,2)=\sum_{\tau=0}^{2}\left(R_{i,\tau}-\text{E}[R_{i,\tau}]\right), (5) where E [R_{i,\tau}] is the expected return derived from the pre-event calibration in Equation 4 . This dual-event structure allows for the calculation of abnormal returns for two distinct informational milestones: CAR 144 (Intent): The market reaction following the Form 144 filing date. CAR 4 (Execution): The market reaction following the Form 4 execution date. The central empirical test is a paired t -test on the difference between CAR 144 and CAR 4 . This comparison quantifies the Information Premium—the degree to which the market differentially processes the initial notice of intent versus the final confirmation of trade execution. 3.3 Phase II: Analysis of Non-Execution and Informational Resolution The second phase of the empirical strategy transitions from the analysis of realized transactions to a framework designed to evaluate the market effects of intent non-execution. The methodology constructs a binary outcome variable, Non-Execution, by mapping trade-level Form 4 data to unique Form 144 filings. A filing is classified as a non-execution ( y=1 ) if the cumulative volume executed within the 90-day statutory window (V executed ) is less than the proposed volume (V proposed ), as defined in Equation 6 : \text{Non-Execution}_{i}=\begin{cases}1&\text{if }\sum V_{executed,i}