UC Santa Cruz

My dissertation focuses on two important aspects that affect micro-entrepreneurs in developing countries: electricity outages and saving constraints. In my first chapter, I study the impacts of infrastructure deficiencies on small business owners in the context of the electricity supply in Malawi - a country where power is very unreliable and only 25% of small businesses are even connected to the grid. Access to reliable electricity is a requirement for development. Yet, most research focuses only on the effects of the first stage - establishing connection to the grid. However, the quality provision is quite poor and existing customers experience frequent outages. Using high frequency data collected on 391 small businesses across the city of Blantyre, I document the following: (1) power outages are very frequent, occur on 67% of days and last on average for 6.6 hours; (2) outages do not affect an average business due to the fact that the majority are not even connected; (3) however, the effects are large for the sub-sample of businesses highly reliant on power (those with electrical appliances), and as a result of blackouts, those businesses lose 25% of daily profits compared to non-outage days; (4) the profit losses are permanent - there is no evidence of profit increases in the following days; (5) the income shock results in reduced total weekly household-level expenditures, and affected households spend 4.8% less for every outage experienced during the week. My results suggest that outages have limited average impacts on small business owners, but businesses relying on electricity to generate income are heavily affected.

My second chapter is a pilot study for the first chapter, which informed the design and development of the main study. In this chapter, I study effects of electricity interruptions among market vendors across 5 markets in Lilongwe, Malawi. I overcome several empirical challenges associated with estimation of the effects. First, due to data limitations and heavy under-reporting of outage instances, I aggregate definition of an outage to a market level and use various constructions of outages definitions to verify the results. Second, I use a bootstrap method estimate clustered standard errors with a few clusters. To estimate the effects, I use the detailed logbooks kept by small businesses owners to study the short-term effects of blackouts on business outcomes. Since outages are so frequent, only 31% of market vendors select into electricity dependent operations. These businesses are affected by power outages and lose 7-16% of revenues on days of blackouts. Due to likely presence of measurement error in the independent variable, I conclude that the presented effects are lower bounds of the true impacts.

The third chapter, co-authored with Jonathan Robinson (Associate Professor of economics, University of California at Santa Cruz) and Shilpa Aggarwal (Assistant Professor of Economics and Public Policy at Indian School of Business), focuses on the effects of provision of savings accounts to micro-entrepreneurs in a developing country. Labeling savings accounts for goals might increase savings for those goals but limit liquidity for other uses. We conduct a field experiment with small business owners in urban Malawi, in which we randomly offer people one or multiple labeled savings accounts - mobile money or simple lockboxes. Since access to a secure place to save is the same for those with one as with multiple accounts, any effect of the incremental account should be due to labeling. We show that (1) the savings products were popular, especially the box (95% of respondents used the box at least twice within 4 months compared to 53% in the mobile money group); (2) takeup of multiple lockboxes was high (66\% of those offered multiple boxes actually used it at least once) compared to 6% in mobile money groups; (3) total deposits across all saving sources were higher for the group with multiple boxes compared to the single box holders; that (4) the savings products had a number of effects on downstream outcomes, including increased labor supply in farming activities, increased spending on certain

categories such as holiday spending and school supplies for children; but that (5) there is no evidence that the labeled accounts limited

liquidity of money for non-labeled purposes, and we do not detect that mental accounting deterred households from using savings account

to deal with health shocks, even though about half of respondents reported that they were reluctant to withdraw from the account. Lastly,

we conclude that mobile money technologies are not able to induce the same saving behaviors as lockboxes likely due to higher transactional costs.