In empirical research econometric concerns are often raised
with regard to two areas: sample selection bias and simultaneity.
Selection bias is not a concern here because we are analysing only
borrowers and controlled for the choice of the lender. However,
social ties are the basis for our social capital measures and may
be endogenous to good credit performance. Loans are supposed to
widen the personal network of borrowers (a vast amount of literature
exists, for example, on empowerment of women and formal
and semi-formal loans). This is particularly true for group loans
but may also apply to individual borrowers. Thus, the strength of
the relationship may grow or fade in intensity as a result of good
or bad credit performance. It might also be argued that bad loan
performance could have an effect on network ties. Bad loan performance
may cut loose one or the other tie, for instance due to
shame or social punishment. Our measures of social capital could
thus be endogenous to bad loan performance. Finally, a badly performing
loan may reduce the social status of the respondent. This
would leave our measurement variable for linking social capital
(that is, assessment of social status) endogenous to loan performance,
and hence the measures of linking social capital may be
endogenous. We have excluded all relationships from the personal
network of the borrowers which were created after the disbursement
of the oldest current or within the last year matured credit.
Furthermore, all ties which have been created within one year