2. Quick start

Please go quickly over Section 1.1 (Data) and Section 1.2 (Sample) before you proceed with this section.

2.1. Parsed tables

Tables W01_exp_main_parsed and E01_edu_main_parsed are the main tables in the database. They relate the id’s of each entrepreneur to the companies where they worked (as founders and/or employees) and their job title, as well as the educational institutions they attended and the programs studied there. As well, these tables list the starting and ending date at each place.

2.1.1. W01_exp_main_parsed

A row in table W01_exp_main_parsed can be read as:

Person <ind_id> worked for firm <org_id> in a role described by <jt_raw_id>, from <exp_sy> to <exp_ey>.

Where the fields <ind_id>, <org_id> and <jt_raw_id> can be looked up in tables F01_founders_info, T01_org_concordance and W02_job_titles_raw_parsed, respectively.

2.1.2. E01_edu_main_parsed

Similarly, a row in table E01_edu_main_parsed can be read as:

Person <ind_id> studied at <org_id> the program <edu_prg_id>, from <edu_sy> to <edu_ey>.

Where the fields <ind_id>, <org_id> and <edu_prg_id> can be looked up in tables F01_founders_info, T01_org_concordance and E03_edu_programs, respectively.

Note

The fields <org_id>, <jt_raw_id> and <edu_prg_id> are missing for some rows. That is the case when the entrepreneur reported her employer but not the job role, or a job role but not the employer (e.g., ‘freelance’), or a university but not the study program.

2.2. Flat tables

Given the database structure, a typical analysis will involve combining and merging multiple tables. To reduce such initial building costs, and to facilitate initial data manipulation and exploration, we include a set of ready-to-use tables that already combine the essential fields and indicators.

In the flat tables, U01_exp_flat and U02_edu_flat, a row represents an instance of an entrepreneur’s work or educational experience, respectively. Hence, each row contains an identifier for an entrepreneur, the organization that she founded or where she worked/studied, the job role or study program, the time spell and additional indicators whose construction would otherwise require to merge one or more tables.

Example:

The rows of W01_exp_main_parsed represent work experience entries. To know the entrepreneur name in any given row, this table must be merged with F01_founders_info through the field ind_id. To query more complex data, such as the employer’s sector, the table must be merged:

  1. with table T01_org_concordance through org_id, and

  2. with table T02_orbis_company_data through bvd_id.

The flat tables already contain the results of such basic operations.

As well, the tables retain identifiers that can be used to append further indicators in the future.

2.3. Other tables

2.3.1. U03_entrepreneur_dyads

This table allows to explore the relationship between entrepreneurs’ shared past (e.g., same schooling, same employer, similar jobs, etc.) and the probability that they team up and co-found a start-up. To achieve this, the rows of U03_entrepreneur_dyads contain distinct dyads (i.e., pairs) of entrepreneurs. There are two types of dyads: observed and counterfactual. Observed dyads involve two entrepreneurs that teamed up and co-founded a start-up in a given year. By contrast, counterfactual dyads involve two entrepreneurs that founded two different start-ups within a three-year period (and hence are not co-founders). The rationale is that counterfactual dyads represent co-founding opportunities that could have taken place, but did not.

We introduce additional variables to narrow down the counterfactuals set by imposing other restrictions. Variables counterf_win_2yr and counterf_win_1yr are dummies that indicate whether the startups in a dyad were founded within a two-year or a one-year period, respectively.

Additionally, we identify start-ups:

  1. in the same industry (using Crunchbase industry tags), or

  2. in the same sector (using the Orbis sector classification).

To use the set of all real and counterfactual dyads, use all the rows of U03_entrepreneur_dyads. To limit the counterfactuals using time window, industry- and/or sector-based restriction you can filter rows where counterf_same_industry==1, counterf_same_sector==1, or counterf_win_Xyr==1.

If you use industry and/or sector indicators, please pay attention to the warning box below.

For reference, the number of real dyads is 329. The number of counterfactual dyads under the broadest definition (i.e., pairs of entrepreneurs who founded two different start-ups the same year) is 61,539. Using industry and sector definitions, the number of counterfactuals are 5,792 and 11,468, respectively.

See U03_entrepreneur_dyads for the content of the table.

Warning

Please note that while nearly all start-ups have Crunchbase industry tags, not all companies in the Orbis dataset have a sector. Moreover, about 13% (n=159) of the start-ups in our dataset could not be matched to Orbis (e.g., some Crunchbase-listed start-ups were not incorporated, some may not show up in Orbis due to their vintage and some may not be matched because their trade and incorporation names differ significantly or are different altogether). As a result, sector data is missing for 15% (n=186) start-ups and, by construction, counterfactual dyads involving these firms are automatically dropped under the sector-based definition.

Warning

We assume all work/study entries without an end year are still ongoing.

2.4. The be_founders dashboard

We are working on a dashboard that allows to explore part of the existing tables and data. It is accessible here.

Comments and suggestions?

Have comments and suggestions about the dashboard, or ideas to improve it? Please send them to manuel.gigena@kuleuven.be