←
Shanthi Manian835 posts
Shanthi Manian avatar

Shanthi Manian

@shanthim

Assoc. Prof., WSU Pullman. Development econ, global health, behavioral. she/her

◷ Joined September 2008▧ 13 Photos and videos
VoxDev avatar
⟳ Shanthi Manian retweeted
VoxDev@vox_dev·Nov 16

Do loan officers discriminate against women-owned businesses? Last month on VoxDev, @ketkidsheth (@UTKnoxville), @shanthim (@WorldBankGroup) & @ayalew_shibiru (Adama Science and Technology University) discussed evidence from Ethiopia: voxdev.org/topic/finance/do-…

The Review of Economics and Statistics (REStat) avatar
⟳ Shanthi Manian retweeted
The Review of Economics and Statistics (REStat)@restatjournal·Nov 6

Loan officers do not discriminate against female business owners in Ethiopia. Just Accepted new paper by Ketki Sheth @ketkidsheth Shanthi Manian @shanthim and Shibiru Ayalew @ShibiruAyalew zurl.co/2pKJ0

VoxDev avatar
⟳ Shanthi Manian retweeted
VoxDev@vox_dev·Oct 19

An experiment in Ethiopia finds no evidence that loan officers discriminate against female business owners. Their decisions reflect the fact that once detailed info about the business is available, accounting for gender does not help identify which businesses will perform well.

Post attachment

🆕 Do loan officers discriminate against women-owned businesses? Today on VoxDev, @ketkidsheth (@UTKnoxville), @shanthim (@WorldBankGroup) & @ayalew_shibiru (Adama Science and Technology University) discuss evidence from Ethiopia: voxdev.org/topic/finance/do-…

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

3. Gender doesn’t predict performance - no tradeoff between equity and effective capital allocation.

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Takeaways: 1. In structured, info-rich evaluation of existing businesses like lending, discrimination is unlikely. 2. Decisions match loan officer beliefs that female- and male-owned businesses perform similarly.

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Where gender does matter (exploratory): when key business info was missing, officers were more likely to forward male-owned firms. Rich, salient information reduces reliance on gender as a proxy for performance.

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Result 3: Gender barely registers. Profits and assets are the most important predictors of endline outcomes. So there is no trade-off between gender equity and profit maximization.

Post attachment
Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Did not discriminating come at a cost to performance? We used ML algorithms (LASSO, random forest, XGBoost) to predict actual business outcomes.

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Result 2: Beliefs don't differ either. Loan officers expected the same survival, profits, assets, and repayment likelihood for male- and female-owned firms. This implies no taste-based (prejudice) or belief-based discrimination.

Post attachment
Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Result 1: No discrimination. Randomized gender had no effect on scores (<0.03 SD) or loan forwarding. The CI rules out a male advantage larger than 2.9 pp in loan consideration. The null result remains across pre-specified subgroups.

Post attachment
Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

Loan officers made two capital allocation decisions: a score that determined competition grants, and whether to forward the applicant to their institution to be considered for a loan. We also elicited incentivized beliefs about each firm's future outcomes.

Shanthi Manian avatar
Shanthi Manian@shanthim·Oct 19

The setup: a field experiment embedded in a real business plan competition in Ethiopia. 84 loan officers from 13 banks and MFIs scored 916 real applications that mimicked initial loan forms. We randomized owner gender on each evaluation, giving us a sample of 3,696 evaluations.