varied over time in the database. For example, if one attempted to identify “startups formed in New York in 2011”, say, one would find considerably fewer startups in the database in 2012 than in 2022. We used these calculations of historical lag, together with estimated firm death rates, to estimate how many startups may be missing from current data – and hence how many startups in total (what we term the Startup Output) exist in a given ecosystem at any time. Some time data also required correcting. Many startups were recorded as having been formed on January 1st of a given year; in our experience, this is invariably incorrect since New Year's Day is a public holiday in both the United States and Canada; rather, it is indicative of the original data entry having year data only, and defaulting to the first day of the year. To correct this, we shifted all such data points to the mid-year. Additionally, we caution that some of the 2024 data is less reliable because of the long lag in seed data capture by funding databases.
Startup Genome | San Francisco • Berlin • London • Delhi • Tokyo • Abu Dhabi • Bangalore • São Paulo
p. 15
Powered by FlippingBook