ලාංකීය සම්භවයක් සහිත මහාචාර්යවරයෙක් වන එමින් සිරිවර්ධන මහතා චිකාගේ විශ්ව විද්යාලයේ විශේෂනයක් සිදුකිරීමේදි කරගත් වරදක් සම්බන්ධව විවේචන එල්ලවෙමින් පවතී. හොඳ වෙලාවට මෙයා සිංහලයෙක් කියල තාම එච්චර highlight නොවුන එක.



ඇත්තටම සිදුවූයේ කුමක්ද?What actually happened
Emil Siriwardane (PhD NYU Stern 2015, now Finnegan Family Associate Professor of Finance at Harvard Business School) circulated a job-market paper, then titled something like “Concentrated Capital Losses and the Pricing of Corporate Credit Risk” (later published in the Journal of Finance 2019 as “Limited Investment Capital and Credit Spreads”). It used proprietary CDS data from the U.S. Treasury’s Office of Financial Research.
In at least one widely circulated version presented during 2014–2015 job talks (including at Chicago), tables listed log(NAICS) among the controls. NAICS codes are arbitrary categorical labels (e.g., 22 for utilities, 44–45 for retail). Taking their logarithm treats the numbers as if they have cardinal meaning, which they do not. This is a basic specification error, equivalent to logging zip codes or Social Security numbers. It was spotted in a seminar; later drafts replaced it with industry fixed effects. Results were reported as largely unchanged.
The paper still won awards (AQR Top Finance Graduate Award, NYU best dissertation and job-market paper prizes, WFA Nasdaq award) and helped land the HBS job. Robert Engle (2003 Nobel laureate) chaired the dissertation committee. The data’s proprietary nature limits independent replication.
This episode has been an inside joke and recurring topic on Econ Job Rumors since ~2015, with posters circulating the old PDF and screenshots of the tables. Chris Brunet’s 2026 video and Substack piece popularized it for a wider audience.
Points that are documented vs. inferred
Siriwardane did not respond to the video creator. HBS materials list him teaching MBA finance courses (Finance II, investment management), not a standalone econometrics class, though the distinction is secondary.
- The error in the draft and the subsequent correction are real and were known in parts of the profession at the time.
- Treating NAICS codes as a continuous logged variable is incorrect. Whether this reflects a fundamental misunderstanding, a late-night Stata/R coding slip (intending factor or dummy variables), or something else cannot be proven from public information.
- The “overnight hotel-room correction while data was supposedly only accessible on-site in DC” is alleged in contemporary forum posts and the video. Siriwardane was an OFR research associate (2012–2017), so he had official access. Brunet states he received no reply when he asked about the timeline. This remains unverified publicly and is treated as a red flag by critics.
- The published version uses standard industry controls. There is no public evidence that the main results were fabricated.
- Academic finance hiring is network-intensive; a high-profile advisor and strong placement record matter. The error did not prevent awards or a top job, which is the video’s main institutional critique.
In short, a sloppy and conceptually wrong control appeared in a job-market draft, was noticed, and was fixed. The rest—intent, data-access logistics, and how much “protection” occurred—is debated in anonymous academic forums and is the video’s editorial overlay. The published paper stands on the corrected specification and proprietary data that outsiders cannot easily re-run.