Social capital and household networks in the founding of the East India Company

This deposit contains the datasets, the name-linking decisions and the Stata code needed to reproduce every result in the article "Social capital and household networks in the founding of the East India Company". The article asks which of London's established Levant merchants took up the East India venture and which did not. A roster of Levant Company freemen drawn up in June 1600 and preserved among the Cecil Papers names eighty-one men, twenty-five of whom appear in the East India charter sealed six months later. That roster supplies a population at risk of a single entry decision, which prior work on the 1599 subscription has not had. Three sources underlie the data, all published. The subscription list of 22 September 1599, with the meeting of 24 September that named the directors and treasurers and set the £200 minimum stake, is taken from W. N. Sainsbury (ed.), Calendar of State Papers Colonial, East Indies, China and Japan, vol. 2, 1513-1616 (London, 1864), pp. 99-102, calendared from the East India Company Court Book, vol. I. Levant Company membership and the June 1600 roster of freemen with their sons and servants derive from David Wilson, Earliest Levant Company Members (September 2017), a compilation of the Turkey (1581), Venice (1583) and Levant (1592, 1600, 1605) charters together with the Cecil list printed in Calendar of the Cecil Papers in Hatfield House, vol. 10 (London, 1904), pp. 214-17. Wilson's compilation is his own copyright and is not redistributed here; what is deposited is the structured dataset derived from it, and his caution that he worked from printed editions rather than manuscripts, that spellings vary, and that men of the same name may in places have been conflated, applies to everything built on it. No manuscript archive was consulted. The estimation file holds eighty-one rows, one per freeman, with charter flags, household tie counts split into kin and apprenticed factors, deaths recorded abroad, and the link to the subscription list. Two do-files reproduce the results: replication_eic_levant.do for the main tables and eic_firth.do for the penalised estimation, the penalised likelihood-ratio tests, the permutation test and the leave-one-out jackknife. Both run on the files deposited here and issue no working-directory commands. The session log and the jackknife output are included. README.md documents the variables, the coding rules, the three codings under which the descriptive results are reported, and the two linkage decisions that required judgement. MANIFEST.txt gives the full file listing with checksums. The eighty-one freemen are the full recorded population rather than a sample, so estimates are read under a superpopulation interpretation following Gill (1999) and Abadie, Athey, Imbens and Wooldridge (2020), and results are discussed by magnitude and interval rather than by whether they cross a threshold.