A large control wedge should make a trader demand more room in the price. Bangert, Myers and Schmardebeck find that future earnings news arrives later when insiders control more votes than cash flow rights. Their claim concerns timing. Institutional monitoring, outside directors and high-quality disclosure weaken most of the effect.
Measuring control without ownership
Bangert, Myers and Schmardebeck derive two figures from the share classes of each dual-class firm-year. InsiderVR measures the fraction of total votes controlled by officers and directors. InsiderCFR measures their fraction of dividend rights. The control wedge between them appears either as a ratio, WedgeRatio, or as a percentage-point difference, WedgeDifference.
Across 1,264 dual-class firm-years, insiders average 51.05% of the votes and 22.72% of the cash flow rights. WedgeRatio averages 2.85, while WedgeDifference averages 28.33 points. At the extreme, insiders command 100.00% of the votes with 12.35% of the cash flows. About 7.99% of dual-class firm-years have no wedge, because control remains proportional to the equity stake.
The authors interact this wedge with a returns-on-future-earnings regression. Current annual buy-and-hold return is the dependent variable. The future earnings response coefficient, or FERC, is the coefficient on aggregate earnings over the next three years. It captures how much future accounting news has already entered today's price. Future returns serve as a control. A parallel model separates earnings into operating cash flow and accruals, then uses the coefficient on future cash flows, the FCFRC.
The proposed channel is credibility. Insiders who command votes without carrying proportional economic exposure may release less information. Investors may instead receive the same information and trust it less, leaving less of it embedded in price. Untabulated tests report no association between dual-class status, or wedge size, and poorer disclosure across three measures. Disaggregation quality also has no significant univariate difference, at 0.6932 for dual-class firms against 0.6972 for single-class firms. Those results lead the authors toward the credibility account. They are unaware of any firm-year measure that captures the credibility of information about future performance, and their tests cannot rule out the disclosure explanation.
The sample covers S&P 1500 firm-years from fiscal 1995 through 2015, excluding regulated and financial industries. It contains 17,604 observations, including 1,264 (7.18%) dual-class firm-years. Dual-class records come from Gompers, Ishii and Metrick for 1995 to 2002, followed by hand collection from EDGAR.
The price learns later
Across the FERC specifications, the interaction between the wedge and future earnings is negative and significant. For WedgeRatio, the coefficient is -0.0590 (p=0.0097) in the dual-class-only sample and -0.0481 (p=0.0010) in the full sample. Adding controls to the full sample produces -0.0431 (p=0.0086). WedgeDifference coefficients range from -0.4016 (p=0.0043) to -0.9596 (p=0.0001).
The cash flow model follows the same pattern. The full-sample coefficient is -0.0641 (p=0.0011), moving to -0.0751 (p=0.0016) with controls. In the dual-class-only sample, the cash flow coefficients reach significance only at 10%, with -0.0449 and p=0.0809. Table 4 Panel B reports a WedgeDifference cash flow interaction of -0.7366 (p=0.0083) for dual-class firms alone and -1.0545 (p=0.0001) for dual-class firms alone with controls.
Consider the interquartile spread. WedgeRatio is 3.48 at the 75th percentile and 1.57 at the 25th, leaving about 1.9 units between them. Applying the full-sample-with-controls coefficient of -0.0431 gives roughly 0.08 less future earnings news in the current price. The future earnings coefficient in that column is 1.0880. It does not represent a firm-level FERC.
What the FERC can tell you
The FERC uses three years of realised future earnings and three years of realised future returns, which prevents its calculation at time t. The paper does not examine whether the wedge predicts risk-adjusted returns and makes no return-predictability claim. Current returns remain the dependent variable in both models.
The raw comparison is flat. Average current returns are 0.1199 for dual-class firms and 0.1219 for single-class firms, a difference the descriptive table reports as statistically insignificant.
Their accounting profile also argues against a naive short. Current earnings-to-price is 0.0546 for dual-class firms versus 0.0391 for single-class firms, a difference of 0.0155 with p<0.01. Current cash flows are 0.1267 versus 0.0969, a difference of 0.0298 with p<0.01. Dual-class firms are smaller, with log market cap of 7.2184 against 7.6692, and receive thinner coverage, with log of one plus analyst count at 1.9862 against 2.3250. The roughly 0.08 FERC gap across the WedgeRatio quartiles supports a wider margin of safety when a name prices information late.
As a trade, it gives you nothing yet.
Monitoring takes most of the effect away
Attenuation is the paper's other headline finding. Its abstract says monitoring by influential institutional investors, strong board oversight and high-quality disclosure significantly attenuate the effect. Monitoring and disclosure, it adds, can alleviate investor concerns about disproportionate insider control. The harder question follows directly: when the effect disappears across most cross-sections, how much remains of the unconditional headline?
The authors first triple-interact the wedge and future earnings with institutional ownership. They measure ownership four ways: total institutional, activist pension, all pension and quasi-indexer. The sign turns positive in 14 of 16 specifications. For high total institutional ownership, the interaction is 0.1325 (p=0.0010) in the dual-class-only sample and 0.0836 (p=0.0076) in the full sample. Activist pension ownership gives 0.1120 (p=0.0033) for dual-class firms alone and 0.1176 (p=0.0001) for the full sample.
Joint F-tests of the wedge term plus the triple interaction are generally insignificant. In the total-ownership columns, p=0.2141 and p=0.8529. The authors interpret those tests as complete attenuation. Even in the full-sample columns, however, the wedge terms are identified from the 1,264 dual-class firm-years. Low power is therefore another reading of the insignificant joint tests.
Quasi-indexer ownership attenuates the effect. Untabulated tests find no significant attenuation from dedicated and transient institutions.
Board results come down to one variable. Outside director proportion attenuates the FERC interaction, at 0.0724 (p=0.0846) for dual-class firms alone and 0.0967 (p=0.0055) in the full sample. For the FCFRC, the corresponding figures are 0.1317 (p=0.0425) and 0.1311 (p=0.0028). CEO duality, board co-option and director busyness have no significant moderating effect. Dual-class firms also have fewer outside directors, 0.7412 versus 0.8199 with p<0.01, and more co-opted boards, 0.5687 versus 0.4742 with p<0.01. They are short of the board characteristic that mitigates the wedge.
Disclosure has a similar effect. The triple interaction is positive in all 12 disclosure specifications and significant in 11. Two dual-class cash flow tests illustrate the result: 0.1167 (p=0.0267) for 10-K disaggregation and 0.2204 (p=0.0049) for annual-plus-quarterly guidance. Readability in the full sample is the lone insignificant cell, at 0.0417 (p=0.2012). The joint effect is insignificant in 10 of the 11 significant cases. The paper describes this as significant attenuation of the negative impact and, often, its elimination.
Results are less consistent under WedgeDifference. Mitigation is significant in 10 of 16 institutional specs, 2 of 4 outside-director specs and 10 of 12 disclosure specs. The authors explicitly decline to judge the relative merits of the two wedge measures. Readers still need to know which measure carries the cross-sectional findings.
A usable data warning
We did not attempt a replication. Building the wedge requires point-in-time voting rights and dividend rights for each share class, together with insider holdings, and we have no dual-class share-class dataset.
One finding can be used immediately. When the authors checked ISS's dual-class flag against filings, 11.39% of observations labelled dual-class by ISS were actually single-class. The paper traces these cases to Class A labelling, staggered director classes, or firms that had already unified. Among 200 randomly selected ISS single-class observations, the authors found zero overlooked dual-class firms. ISS's flag therefore produces false positives in one direction at better than one in ten.
Where identification gets thin
The first-stage selection probit for dual-class status has a ROC area of 0.8594. Its exclusion restrictions are seven IPO-era determinants of dual-class status drawn from Gompers, Ishii and Metrick. These include a family name in the firm name, a media-industry indicator, local market share, profit and sales rank among that year's IPOs, and regional sales concentration.
The outcome is observed an average of 28 years after the IPO. The authors argue that this distance is why the variables should have no direct second-stage effect. Relevance is tested in the first stage; the exclusion restriction remains the assumption. Their own check adds the exclusion variables to Models (2) and (3). Two significant exceptions appear: the media indicator interacted with future earnings and local market share interacted with future cash flows. Heckman, propensity matching using caliper 0.03 and nearest neighbour without replacement, and entropy balancing on the first three moments address observables only.
Firm fixed effects absorb time-invariant selection that the Heckman stage can only model. The within-firm estimates consequently provide the stronger identification. After firm fixed effects, 33.43% of WedgeDifference variation and 50.42% of WedgeRatio variation remains. Within a firm, increases in the wedge are associated with lower FERCs and FCFRCs. The association is concentrated among firms with weaker monitoring and low-quality disclosure.
The sample ends in 2015 by design, keeping COVID returns outside the analysis. Yet dual-class firms rose from 6% of U.S. market cap in 1999 to 16% in 2023, and reached 60% of IPO market cap in 2020. Most of the period driving the current debate falls beyond the paper's window.
A study covering 2016 to 2024 would change my reading if the result survived among the largest dual-class names, where quasi-indexer ownership is high and the wedge is extreme. If attenuation persists there, the wedge becomes a screen for identifying which dual-class names price information late, rather than a general warning about the structure.