Skip to content
sW:Sean Wang
Information economics in capital markets

When information breaks, who pays?

Sean Wang

I study the forces that distort information in capital markets — bias in analyst research, contractual control of clinical-trial disclosure, executive personality, and the intangible capital that accounting leaves off the balance sheet. The work spans accounting, finance, behavioral science, and public policy.
I am currently visiting Rice University for the 2026–27 academic year.

Featured Missing Intangible Capital Dashboard — look up a company and find the resources spent towards knowledge (R&D) and organizational capital (SG&A) that is currently missing from GAAP balance sheets. →
Audience

Showing identification, samples, and estimates.

Showing plain-English findings and quotable figures.

Research program

Three forces that distort markets

One question, pursued three ways: what happens when the information systems in capital markets break down?

01

Information bias & discrimination

Racial prejudice and executive personality distort what financial intermediaries produce. I document the distortions, validate new ways to measure them, and identify what makes them go away.

JAE 2025 · JAR 2017 · RAST 2018 · RAST 2025 · TAR (R&R)
02

Price discovery & attention

Attention is finite and inventory is costly. Processing firm news crowds out macro news; microstructure frictions make prices predictable around scheduled announcements.

JFE 2014 · RAST 2026 · TAR 2020 · AOS 2019 · JFQA 2018
03

Transparency & measurement

Some information is suppressed by contract; some is omitted by accounting rules. Pharmaceutical embargoes, regulatory attention that backfires, and the intangible capital missing from balance sheets.

Mgmt Science 2025 · JAR 2022 · JCF 2014 · JAE (R&R)
Selected work

How we know it

Everything here is quotable; every figure links to the paper behind it.

Working paper · Jun 2026New

Fragile Intangible Capital: Measuring the Real Cost of Exposed Disinformation

with J. Bai, Y. Cao, and C. Wan

When the market learns a firm's reputation was built on fabricated reviews, what does it cost — and does the damage heal?

Firms caught with fabricated reviews lose about 8% of foot traffic and 5–6% of transaction volume against demand-matched peers.

  • The loss lands within a month and is still there at eighteen
  • It outlasts both the warning banner and the advertising ban
  • Penalties roughly double when the alert carries hard evidence
Design & identification
Design — staggered DiD; Sun & Abraham (2021) interaction-weighted estimator for event studies.
Controls — never-treated demand-side peers from the intermediary's own "People Also Viewed" graph (revealed substitution, not industry code or geography).
Estimate — −0.0763 (t = −3.52), N = 16,837 firm-months; entropy-balanced −0.0772 (t = −4.95).
Pre-trends — max |t| = 1.24; no effect in the alert month itself (−0.6%, t = −0.47).
Data — Yelp alerts × SafeGraph GPS foot traffic × Consumer Edge transactions, 2019–2024.

Get caught faking your reviews and you lose roughly 8% of your customers — permanently.

Eighteen months later the traffic has not come back, long after the warning label is gone.

Paper (PDF) SSRN
Fig. 1 — Event study, foot traffic
Event study of foot traffic around a consumer alert Foot traffic is flat for twelve months before the alert, drops about 7 percent within one month, and remains 8 to 11 percent below matched peers through month eighteen. Δ FOOT TRAFFIC vs MATCHED PEERS 0 −5% −11% ALERT POSTED AD BAN ENDS banner ~90d −12 0 +9 +18 MONTHS FROM ALERT

Sun–Abraham estimates, month −1 omitted (◦). Filled = p<0.05. Months +8 and +9 are insignificant — a partial rebound while the ad ban is still active.
R&R · J. Accounting & EconomicsPolicy relevant

Disclosure Control as a Real Option: Contractual Governance in Clinical Trials

with O. Chen and Y. Zhang

When drug sponsors buy the contractual right to review and delay the trials they fund, do they use it — and when?

Three-quarters of industry-sponsored trials run by outside investigators carry an embargo clause — and the sponsors use it.

  • They clear the FDA’s p < 0.0001 bar 22.8% of the time versus 14.2%
  • Their abstracts erase 83% of the caution on near-miss results
  • The effect appears only where the option is worth something
Design & identification
Design — cross-sectional OLS with year × indication × firm fixed effects; SEs clustered by firm. Not a DiD; the paper reports associations.
Identification — a within-design placebo: the embargo effect on spin is +0.336 (t = 3.05) in the marginal zone (0.05 < p ≤ 0.15) but +0.097 (t = 1.45, n.s.) where results already cleared p ≤ 0.05. Difference tested, p = 0.02.
Magnitude — 0.336 on a LIWC tone index with SD 0.402, i.e. ≈0.84 SD; it closes 83% of the −0.407 caution non-embargoed investigators show.
Sample — 58,144 firm–trial–indication observations, ClinicalTrials.gov × PubMed, 2007–2023.
Downstream — the FDA discounts the approval premium on extreme significance by ≈35% (−0.227, t = −2.55); capital markets show no reaction at all.

Drug companies can pay for the right to review their own trial results before publication. Three out of four trials come with that clause.

Those trials cluster at exactly the p-value the FDA treats as strong enough to approve a drug on a single study. The FDA appears to notice; investors don't.
Not a fraud claim — the paper is explicit that sponsors don't fabricate data.

SSRN
Fig. 2 — Spin by outcome zone
Spin in published abstracts by statistical-significance zone Embargoed and non-embargoed trials show identical framing when results clearly pass or clearly fail. Only in the marginal zone, where results just missed significance, do embargoed trials show markedly more positive framing. OPTION IN-THE-MONEY OPTION NOT NEEDED 0 −.15 −.41 MORE POSITIVE FRAMING ↑ +0.097 n.s. — PLACEBO ✓ +0.336 t = 3.05 · 0.84 SD closes 83% of gap PASS p ≤ .05 MARGINAL .05 < p ≤ .15 already significant just missed Embargoed Not embargoed

Spin = LIWC tone, conclusions minus results, relative to trials that clearly failed. Table 5 Panel A, N = 7,907. Independent investigators write cautiously about near-misses; embargoed abstracts do not. The near-zero gap where results already passed is the placebo.
J. Accounting & Economics · 2025

Do Sell-side Analysts React Too Pessimistically to Bad News for Minority-led Firms?

with K. Rupar and H. Yoon

Do analysts price the same bad news differently when a firm's CEO is not White?

Analysts price bad news 57% harder when a firm’s CEO is not White.

  • No comparable effect for good news, and none in EPS forecasts
  • The asymmetry points to unconscious stereotyping, not a race effect
  • And it was wrong: those firms beat the target 2.7pp more often
Design & identification
Design — analyst-firm-date panel with analyst and year fixed effects, all 13 controls interacted with bad news; SEs two-way clustered by analyst and year-month. Plus a randomized experiment (N = 366) varying only the CEO's photo.
Estimate — Non-White × Bad News = −0.013 (t = −2.27), N = 97,624, S&P 1500, 2005–2019. Per one SD of bad news, implied returns fall 3.6% vs 2.3%; the 57% is the ratio of those sensitivities, not a level gap.
Ex-post — Beat = +0.027 (p < 0.01). The gloomier targets were wrong.
Moderators — amplified post-2016 and in high hate-crime areas; attenuates once an analyst has covered the CEO before.

Wall Street analysts mark a company down substantially harder for the same bad news when its CEO isn't white — and the gloomier forecasts turn out to be wrong.

No equivalent gap for good news, which is what points to an unconscious reflex. Confirmed by a randomized experiment where only the CEO's photograph changed.

Paper (PDF) DOI
Fig. 3 — Valuation sensitivity to bad news
Analyst target-price sensitivity to bad news by CEO race Both groups start at the same implied return. As bad news increases, valuations for non-White-led firms fall more steeply, a 57 percent larger slope. Good news shows no comparable gap. 10.8% 8.5% 7.2% TARGET-PRICE IMPLIED 12-MO RETURN White CEO Non-White CEO 57% STEEPER 0 +1 SD MAGNITUDE OF EARNINGS MISS → GOOD NEWS · no effect (n.s.) EX-POST — WAS THE PESSIMISM JUSTIFIED? Δ probability the price exceeds the 12-month target, per +1 SD of bad news WHITE CEO ≈ 0 — not significant NON-WHITE CEO +2.7 pp (p < 0.01) The gloomier targets were beaten more often — not less.

Table 4, N = 97,624. The 57% compares slopes; the level gap in target prices is not statistically distinguishable from zero.
Management Science · 2025Free data + codeDashboard: Missing Intangibles

Measuring Intangible Capital with Market Prices

with M. Ewens and R. Peters

If accounting keeps internally created knowledge and organizational capital off the balance sheet, what are the right parameters for putting them back on?

Across 2,004 acquisitions and liquidations, the market is forced to price exactly the intangibles that never reached the balance sheet.

  • R&D depreciates a third faster than official estimates imply
  • Overhead that is really investment: 20% consumer, 51% health
  • Purchase accounting makes that price public — that is the identification
Design & identification
Design — structural perpetual-inventory model fit by nonlinear least squares to observed exit prices; industry-year intangible market-to-book fixed effects; bootstrapped SEs (1,000 firm-level replications).
Identification — exit price P = identifiable intangible assets + goodwill, scrubbed of synergy and overpayment via announcement returns; matched against up to ten trailing years of R&D and SG&A.
Estimates — δG = 0.33 (0.034), γS = 0.28 (0.024), with δS fixed at 0.20. Pseudo-R² = 0.542, N = 2,004 (1,523 acquisitions + 481 bankruptcies), 1996–2017.
Validation — on a 1978–2017 panel excluding all estimation firm-years, the resulting stocks perform equal to or better than status-quo measures; the gains are significant for market enterprise value and human-capital risk, marginal for brand, patents, and trademarks.

Trillions in corporate assets never appear on a balance sheet. Acquisitions reveal what they're worth.

Research spending loses value about a third faster than the official government estimates assume. Parameters, data, and code are public; used in Morgan Stanley research.

Paper (PDF) DOI Data + code MS: Impact

These parameters, applied to every listed firm: the interactive dashboard lets you look up a company and see how much of what it owns never reaches its balance sheet — sortable, filterable by industry or index, 1975 to 2025.

Fig. 4 — Exit prices identify the parameters
How exit prices identify intangible capitalization parameters Trailing R and D and SG and A spending are accumulated with unknown decay rates, then set equal to the observed price paid for a firm's intangibles at acquisition or liquidation, which pins down the parameters. R&D FLOWS t−9 … t SG&A FLOWS t−9 … t PERPETUAL INVENTORY Σ (1−δ)ᵏ · Zₜ₋ₖ δG unknown → estimate γS unknown → estimate δS = 0.20 assumed = PRICE AT EXIT IIA + goodwill 1,523 acquisitions 481 liquidations N = 2,004 · 1996–2017 ESTIMATED δG = .33 γS = .28 knowledge decay · SG&A investment share PRIOR BENCHMARK (BEA-HH) δG = .23 γS = .30 flat across industries

Table 1, exits column. γS is an investment share, not a depreciation rate, and ranges 0.20–0.51 across industries.
J. Financial Economics · 2014

News-driven Return Reversals: Liquidity Provision Ahead of Earnings Announcements

with E. So

Do market frictions create predictable price patterns around scheduled news?

Short-term reversals grow more than six-fold at earnings announcements.

  • A long-short position earns 1.45% over three days, 0.22% on placebos
  • The cause is inventory: the other side is dangerous to hold
  • Market makers price that risk, so the concession widens
Design & identification
Design — conditional quintile sort on the market-adjusted return over t−4 to t−2, held t−1 to t+1, with breakpoints from the prior calendar quarter. The identifying comparison is a pseudo-announcement date drawn uniformly 10–40 trading days earlier.
Estimate — 1.448% (p = 0.00), Table 2 Panel A; pseudo-date counterpart 0.218%. Returns are market-adjusted raw, not factor-adjusted, and are gross of transaction costs. With size, book-to-market, momentum, and beta controls the spread is 1.16%.
Sample — 107,039 earnings announcements, 1996–2011, price ≥ $5.
Mechanism — inventory risk, not adverse selection: no size gradient, survives Nagel (2012) weights, scales with option-implied announcement volatility, and vanishes for unanticipated announcement dates.

In the days before a company reports earnings, the cost of trading quietly rises — and it's predictable.

Reversals grow more than six-fold at earnings versus ordinary days. Note: 1.45% is a raw three-day return before trading costs, not an implementable net return.

Paper (PDF) DOI
Fig. 5 — Actual vs. placebo clock
Reversal strategy returns on actual versus pseudo announcement dates The identical three-day trade earns 1.45 percent around real earnings announcements and 0.22 percent when the same trade is run on a randomly drawn pseudo-earnings date, more than a six-fold difference. ACTUAL ANNOUNCEMENT earnings day SIGNAL HOLD (t−4, t−2) (t−1, t+1) versus PLACEBO DATE same trade, run on a random pseudo-earnings date drawn from [t−40, t−10] pseudo-date · no news SIGNAL HOLD 3-DAY LONG–SHORT RETURN 1.45% actual announcements 0.22% placebo dates > 6× larger, p < 0.01 MECHANISM ✓ inventory risk — scales with option-implied announcement volatility; absent when the announcement date is unanticipated. MECHANISM ✗ adverse selection — no size gradient; survives Nagel (2012) weights.

Table 2 Panel A, N = 107,039. Market-adjusted raw returns, gross of costs; 1.16% with risk controls.
J. Accounting Research · 201798th %ile cited

CFO Narcissism and Financial Reporting Quality

with C. Ham, M. Lang, and N. Seybert

Does a CFO's personality show up in the quality of the numbers the company reports?

CFOs who take up more space when they sign their name run more aggressive accounting.

  • Larger accruals, slower bad-news recognition, weaker controls
  • 25th to 75th percentile of signature size: 23% higher restatement odds
  • A lab experiment rules out handwriting. The CEO’s signature does not
Design & identification
Validation — n = 63 lab subjects; signature size correlates with NPI-40 (r = 0.30, p = 0.017), and narcissism fully mediates the path to misreporting: the direct effect falls from β = 0.36 (p = 0.035) to β = 0.27 (p = 0.124) once NPI-40 enters. Payoffs are deterministic, so overconfidence and risk aversion are inert by construction.
Archival — 512 notarized CFO signatures from SEC Order 4-460 (June 2002); firm-years with two-digit SIC fixed effects, SEs clustered by firm. Associational — no firm or CFO fixed effects are possible with one signature per CFO.
Estimates — accruals 0.009 (p<0.01); real EM 0.051 (p<0.05); conservatism −0.066 (p<0.05); ineffective controls 0.641 (p<0.10); restatement 0.455 (p<0.05).
Ruled out — options-based overconfidence, facial width-to-height ratio, expected letter size, CFO relative pay.

How big a CFO signs their name predicts how aggressively their company keeps its books.

A large signature is associated with 23% higher odds the year's financials get restated. The effect is specific to CFOs — the CEO's signature predicts almost nothing.
Covered in the FT, WSJ, NYT, HBR, and Washington Post.

Paper (PDF) DOI
Fig. 6 — Validate the proxy, then apply it
Mediation of signature size through narcissism, and outcomes by executive role In the lab, narcissism fully accounts for the link between signature size and misreporting. In the archival data, CFO signature size predicts all five reporting outcomes while CEO signature size predicts only real earnings management. STAGE 1 — VALIDATE THE PROXY (LAB, n = 63) NARCISSISM (NPI-40) SIGNATURE MISREPORTING β = 2.23 β = 0.05 direct β = 0.36 → 0.27, n.s. once narcissism is controlled SHARE OF PARTNER'S ALLOCATION TAKEN, BY SIGNATURE QUARTILE Q1 12% Q2 31% Q3 34% Q4 53% STAGE 2 — APPLY IT (512 CFOs) Does signature size predict worse reporting? CFO CEO Accruals EM ✓ ✗ Real EM ✓ ✓ Timely losses ✓ ✗ Internal controls ✓ ✗ Restatements ✓ ✗ ✓ significant ✗ no significant association +23% odds of restatement, 25th → 75th pctile

Lab mediation (Figure 2) and archival outcomes (Tables 4–8, 11). The CFO/CEO asymmetry is consistent with CFO oversight of accounting quality.
R&R · The Accounting Review

Migration Fear, Information Access, and Analyst Forecast Accuracy

with C. Wan, Y. Wang, and A. Yorulmaz

Does anti-immigrant sentiment impair minority analysts' ability to forecast earnings?

A one-standard-deviation rise in societal migration fear raises non-White analysts’ forecast errors by $0.04 EPS.

  • No effect on White peers
  • The mechanism is lost access, not lost skill
  • Their conference-call participation drops 5.7 percentage points
Design & identification
Design — 1.3 million EPS forecasts, 1990–2023, against a news-based migration-fear index; the 2015–16 election is used as an inflection point in sentiment, corroborating rather than identifying.
Mechanism — 5.7pp reduction in conference-call participation during high-fear periods, coinciding with the increase in forecast errors.
Moderators — the penalty is largest in high-idiosyncratic-volatility firms and is attenuated under non-White CEO leadership and in sanctuary jurisdictions.

When anti-immigrant sentiment rises, minority analysts get less accurate — because they get less access.

No comparable effect for White analysts. The gap closes under non-White CEOs and in sanctuary jurisdictions, which points to information channels rather than ability.

Paper (PDF) SSRN
Fig. 7 — Effect and mechanism
Forecast error response to migration fear, by analyst race A one standard deviation rise in migration fear raises non-White analysts' forecast errors by four cents per share with no effect for White analysts, alongside a 5.7 percentage point drop in their conference call participation. Δ ABSOLUTE FORECAST ERROR PER +1 SD MIGRATION FEAR NON-WHITE WHITE +$0.04 no significant effect MECHANISM — ACCESS, NOT ABILITY MIGRATION FEAR news-based index CALL PARTICIPATION −5.7 pp FORECAST ERROR non-White only ATTENUATED under non-White CEO leadership and in sanctuary jurisdictions. 1.3M forecasts · 1990–2023

The 5.7 is the participation mechanism, in percentage points — not a multiple of the main effect.
The complete record

Every paper, one click away

Twelve peer-reviewed articles, eleven in FT50 journals. Full text for every one — no paywall, no request form.

01 — Information bias & discrimination

2025Do Sell-side Analysts React Too Pessimistically to Bad News for Minority-led Firms? Evidence from Target Price Valuationswith K. Rupar, H. Yoon · valuations 57% more sensitive to bad news; no good-news effect; randomized experimentJ. Acct. & Econ. PDF · DOI
2025CFO Narcissism and the Power of Persuasion Over Analysts: A Mixed-Methods Approachwith C. Ham, M. Piorkowski, N. Seybert · argumentative language and euphemism on calls; lab study of stated tacticsRev. Acct. Studies PDF · DOI
2018Narcissism is a Bad Sign: CEO Signature Size, Investment, and Performancewith C. Ham, N. Seybert · higher R&D and M&A spending, similar capex; lower ROA and operating cash flowRev. Acct. Studies PDF · DOI
2017CFO Narcissism and Financial Reporting Qualitywith C. Ham, M. Lang, N. Seybert · 5th most-cited at JAR since publication (Web of Science, 3/2024)J. Acct. Research PDF · DOI

02 — Price discovery & investor attention

2026Firm-specific Information Processing and the Delayed Discovery of Macroeconomic News: Evidence from Earnings Announcement Returnswith J. Pan, E. Sul · day-0 market return predicts announcers' returns over days +1 to +3, rising with surprise extremityRev. Acct. Studies PDF
2020Asymmetric Timeliness and the Resolution of Investor Disagreement and Uncertainty at Earnings Announcementswith M. Barth, W. Landsman, V. Raval · conservatism is associated with slower resolution of disagreement and uncertaintyThe Acct. Review PDF · Companion note
2019Informational Environments and the Relative Information Content of Analyst Recommendations and Insider Tradessolo-authored · analysts appear to hold relative industry expertise, insiders relative firm-specific expertiseAcct. Org. & Society PDF · DOI
2018Know Thy Neighbor: Industry Clusters, Information Spillovers and Market Efficiencywith J. Engelberg, A. Ozoguz · denser clusters show 1–3% lower price delay per SD; suggestive evidence from 194 relocationsJFQA PDF · DOI
2014News-driven Return Reversals: Liquidity Provision Ahead of Earnings Announcementswith E. So · reversals more than six-fold larger at earnings than on placebo datesJ. Financial Econ. PDF · DOI

03 — Transparency & measurement

2025Measuring Intangible Capital with Market Priceswith M. Ewens, R. Peters · δ_G = 0.33, γ_S = 0.28 from 2,004 exits; free data and code; applied in Morgan Stanley researchManagement Science PDF · DOI
2022The Relationship Between Non-GAAP Earnings and Aggressive Estimates in Reported GAAP Numberswith R. Guggenmos, K. Rennekamp, K. Rupar · informal SEC attention increases GAAP aggressiveness; the formal Reg G requirement did the oppositeJ. Acct. Research PDF · DOI
2014The Prevention of Excess Managerial Risk TakingVan Wesep & Wang · theory; severance contingent on results, with complete failure nullifying paymentJ. Corporate Finance PDF · DOI

Under review & working papers

2026Disclosure Control as a Real Option: Contractual Governance in Clinical Trialswith O. Chen, Y. Zhang · embargo clauses as an option exercised only when results are marginalR&R · JAE SSRN
2026Migration Fear, Information Access, and Analyst Forecast Accuracywith C. Wan, Y. Wang, A. Yorulmaz · +$0.04 EPS error for non-White analysts; 5.7pp drop in call participationR&R · TAR PDF · SSRN
2026Fear Cautions, Anger Commands: Information in Managerial Vocal Emotionwith E. Sul, V. Zhu · distinct informational content in executives' vocal emotion on earnings callsR&R · CAR SSRN
2026Information from Implied Volatility Comovements and Insider Tradeswith R. Bushman, V. RavalR&R · Rev. of Finance SSRN
2026Fragile Intangible Capital: Measuring the Real Cost of Exposed Disinformationwith J. Bai, Y. Cao, C. Wan · ~8% foot traffic loss with no sustained recovery through eighteen monthsWorking paper PDF · SSRN
2026Are Analyst Forecast Errors Really Kinky?with M. Barth, J. Jeong, W. Landsman · bundling closes ~66% of the excess kink vs ~32% for discretionary accruals; a substantial kink remainsWorking paper PDF · SSRN
Public goods

Data, code, and the missing balance sheet

Parameters and intangible capital stocks for any public firm, plus a dashboard for what GAAP leaves out.

Intangible capital parameters

Exit-price estimates of knowledge- and organizational-capital parameters, freely downloadable, with the code to rebuild adjusted balance sheets for any Compustat firm. Used by researchers worldwide and applied in Morgan Stanley Investment Management research.

# EPW exit-price estimates
delta_G = 0.33 # knowledge capital decay
gamma_S = 0.28 # SG&A investment share
delta_S = 0.20 # org. capital decay (assumed)
# prior benchmark: 0.23 / 0.30, flat by industry
$ python construct_stocks.py
→ adj_balance_sheet.csv ✓
Parameters + stocks (GitHub) Working paper See them applied →

Don't want to run the code? The interactive dashboard has already applied these parameters to 2,000 companies — search a firm by name or ticker and read the answer off the table.

Missing Intangible Capital Dashboard

Look up any of 2,000 public companies and see the research and organizational capital it has built up but never reported. Sort by which firms hold the most, or by how much of the business is invisible on the balance sheet. Filter to an industry, or to the S&P 500, Nasdaq 100, or Dow 30. Everything is built from the published EPW parameters and runs from 1975 to 2025.

Reported versus intangible-adjusted balance sheet A reported balance sheet shows only tangible assets; the adjusted version adds knowledge capital and organizational capital on top. GAAP REPORTED TANGIBLE INTANGIBLE-ADJUSTED TANGIBLE ORG. CAPITAL KNOWLEDGE CAPITAL missing from GAAP Illustrative. Stocks built from trailing R&D and SG&A at the published parameters.
Open the dashboardExtension data (CSV) ↓
Writing & press

Research, translated for people who act on it

Evidence turned into arguments practitioners and regulators can use.

Capital markets

Mispricing, volatility anomalies, and information gaps — connecting published evidence to live market dynamics.

Research notes

Accessible walkthroughs of the papers: what we found, how we know it, and what it means for investors, firms, and regulators.

Policy

Where disclosure failures meet the public interest — FDA transparency, accounting standards, and financial regulation.
Research covered in  Financial Times · The New York Times · The Wall Street Journal · The Washington Post · Harvard Business Review · Bloomberg Businessweek · The Atlantic

Press kit

Working on a deadline? Everything you need is here: high-resolution headshot, short and long bios, the three most quotable current findings with their sources, and a direct line. I answer press email quickly and am happy to explain a result in plain English before you write.

Email for press Headshot (colour, print res) Full CV
Teaching

Peer-reviewed by 1,355 students

Financial Accounting II, every MBA section since Fall 2018. Official end-of-semester evaluations; raw response files on request.

14Semesters
F18 – Sp26
39Sections
1,355Students
enrolled
700Respondents
4,008 ratings
4.5Mean / 5
89% rate 4–5
For students Rice FSA Tutor — a ChatGPT assistant trained on the Financial Accounting II material. Work through a problem, check a journal entry, or have a concept explained a different way, at any hour. profsean.wang/FSAtutor →
Distribution of all student ratings
5
63.1%
4
26.2%
3
6.6%
2
2.8%
1
1.3%

N = 4,008 individual question ratings from 700 respondents across 14 semesters, 5-point scale. Median is 5 on every survey question across all sections.

The most common request, across roughly 14% of improvement comments over a decade, is for more practice problems and worked examples. It is the only suggestion to appear in every academic year, and I have been adding pre-class problem sets and TA-led review sessions in response.

I have my undergraduate and masters in accounting so I have taken my fair share of accounting course in my life. Sean Wang could be in my top 2 accounting professors that I have ever had.Spring 2025
He was constantly challenging us to look at accounting not just as debits and credits, but what did they mean. He spent a lot of time trying to help us understand the WHY of accountingFall 2019
Makes class exciting and actually ties it into real life. This class was the most useful course I've taken this year.Fall 2022
Sean is a fantastic teacher because he genuinely wants us to understand the accounting concepts and why he's teaching the material. I'm in awe and impressed and wish more teachers would care as much as Sean does.Fall 2018

Verbatim; student spelling and punctuation preserved. Selected from 464 open-ended responses.

About

A chemist who became an information economist

Sean Wang

Before markets, molecules. Organic and polymer chemistry at Duke and South Florida taught me to think in systems, reactions, and equilibria — and to insist that evidence lead. The pivot came at NYU Stern, where equity valuation gave me a front-row view of how information moves through markets, and how often it gets distorted along the way.

Cornell supplied the econometric toolkit. Since then the research has followed the information — wherever it is produced, wherever it is corrupted, and wherever it fails to reach the people who need it. Today that spans racial bias in analyst valuations, the contracts pharmaceutical sponsors use to control clinical-trial disclosure, and the trillions in intangible assets that accounting systematically omits.

The costs of information failure fall hardest on those with the least power to see them coming.

Best Paper Award, 2023 AAA Accounting Behavior & Organizations Conference · FARS Midyear Outstanding Discussion Award, 2021 · Track Chair, AAA Annual Meeting 2025

DukeB.A. Chemistry
South FloridaM.A. Chemistry
NYU SternM.B.A. Finance
CornellPh.D. Accounting · econometrics, finance · 2009
2008 — presentUNC Kenan-Flagler → Rice Jones → SMU Cox → Rice Jones
Get in touch

Research collaboration, advisory work, and media.

Happy to hear from co-authors, seminar organizers, and doctoral students working on information economics, disclosure, or analyst behavior.

Sean@profsean.wang Curriculum vitae ↓ Google Scholar SSRN Substack