Hm. Interesting analysis but I would be careful with inference. Rare disease isn't a random treatment assignment, it's usually a choice made by companies that typically have clearer biological targets (often monogenic), stronger IP positions, and regulatory considerations (orphan drug designation, smaller trial sizes, priority review).
The multinomial logistic regression did control for location, modality, and lead stage, but it almost certainly did not control for quality of underlying biology, strength of genetic validation, founding team. So I think in this case "rare disease focus" may be proxying for "well-defined mechanism with strong target validation" which would predict success regardless of disease prevalence.
The acquisition signal is particularly suspect? Big pharma has been systematically acquiring rare disease companies for a specific financial reason ie to get orphan drug pricing power combined with small, manageable commercial footprints. That's a market structure thing, not really evidence that rare disease programs are scientifically better or perform better in the end.
I.e., the article may make it seem like going into rare diseases is a lower risk endeavor than some other indications, but it couldn't be farther from the truth.
Nice work here. The flyover state finding is brutal but makes sense given how hard it is to recruit senior biotech talent outside hubs. Curious if the "Trading-" bucket might hide successful pivots that haven't recovered market cap yet, they'd look indistinguishable from slow failures in this data. Phase III IPO sweet spot probably reflects survivorship bias too.
Any thoughts on extending this analysis past just public companies? The large majority of biotech startups never go public, so there's likely significant upstream selection effects coloring this analysis.
How did you decide CA, MA, NY, NJ, PA as the hub states? Big pharma is all moving their facilities to be near NIH. I would have put MD in the top 3. But it looks like you did specifically consider "DC metro" and found it lousy?
I'm not sure what hub effect I would predict. Hubs should create more companies, but why better companies? People know that they aren't in a hub, but don't realize the cost? Why don't VCs know the value and provide pressure?
Hm. Interesting analysis but I would be careful with inference. Rare disease isn't a random treatment assignment, it's usually a choice made by companies that typically have clearer biological targets (often monogenic), stronger IP positions, and regulatory considerations (orphan drug designation, smaller trial sizes, priority review).
The multinomial logistic regression did control for location, modality, and lead stage, but it almost certainly did not control for quality of underlying biology, strength of genetic validation, founding team. So I think in this case "rare disease focus" may be proxying for "well-defined mechanism with strong target validation" which would predict success regardless of disease prevalence.
The acquisition signal is particularly suspect? Big pharma has been systematically acquiring rare disease companies for a specific financial reason ie to get orphan drug pricing power combined with small, manageable commercial footprints. That's a market structure thing, not really evidence that rare disease programs are scientifically better or perform better in the end.
I.e., the article may make it seem like going into rare diseases is a lower risk endeavor than some other indications, but it couldn't be farther from the truth.
"Did the founders go to MIT?" might be a decent control here.
Nice work here. The flyover state finding is brutal but makes sense given how hard it is to recruit senior biotech talent outside hubs. Curious if the "Trading-" bucket might hide successful pivots that haven't recovered market cap yet, they'd look indistinguishable from slow failures in this data. Phase III IPO sweet spot probably reflects survivorship bias too.
These are some really cool findings and I love the analysis methodology using Claude. Thanks!
Any thoughts on extending this analysis past just public companies? The large majority of biotech startups never go public, so there's likely significant upstream selection effects coloring this analysis.
How did you decide CA, MA, NY, NJ, PA as the hub states? Big pharma is all moving their facilities to be near NIH. I would have put MD in the top 3. But it looks like you did specifically consider "DC metro" and found it lousy?
I'm not sure what hub effect I would predict. Hubs should create more companies, but why better companies? People know that they aren't in a hub, but don't realize the cost? Why don't VCs know the value and provide pressure?
It might be good to turn this into like a biannual panel dataset + company fixed effects to capture trading+- with more resolution.