Published on Development Impact

Insuring small firms against big political and economic risks: an experiment in post-revolution Egypt

This page in:
The Arab Spring brought about a wave of joy in many countries in the Middle East and North Africa as repressive regimes that had ruled for years were overthrown. But the aftermath brought about considerable turmoil and uncertainty as to what was going to happen in many countries. In Egypt, the immediate aftermath of the January 2011 revolution which toppled Hosni Mubarak included closing the stock market for 55 days, curfews of up to 18 hours a day, and an interim government under the control of the Armed Forces. The next year included almost weekly large-scale protests in Tahrir Square, three Ministers of Finance resigning, tourist arrivals falling 33 percent, prolonged negotiations with the IMF over an emergency package, and debate over when to hold elections and the disqualification of many leading candidates.

Consider the impact on a small business owner. In addition to the immediate direct effects of this fall in economic activity on his or her business, they suddenly faced much more uncertainty about the future. So if they had plans to potentially expand their business by buying new equipment or hiring new workers, they would likely put these plans on hold while they waited for the situation to stabilize. Even in “normal” times, policy uncertainty and macroeconomic instability are the two most common constraints to firm growth listed by firms in developing countries in the World Bank’s Investment Climate Surveys. This raises the question of whether there is anything we can do to help protect firms against such risks and enable them to go ahead with expansion plans even in an environment of uncertainty?

My new working paper with Matt Groh reports on an experiment we ran to test the idea of insuring microenterprises against this risk. If you want a 2-page version, here is a policy impact note. But before telling what we ended up doing, let me tell you how we got there.

Our abandoned experiment in Egypt
We started working in Egypt in October 2009 with the aim of evaluating an expansion of microfinance into the poorest villages in Upper Egypt under a project to be financed through a World Bank loan. At the time there were very few RCTs of microfinance, and we were excited about the potential to learn from a large microfinance expansion. Randomization was to take place at the village level, and we had hopes of measuring the impact on other financial providers in these villages as well as measuring impacts on the borrowers themselves. We were fortunate enough to raise funding from 3ie and from World Bank trust funds to pay for what was to be a major survey effort.

As often happens with government projects, there were considerable delays in getting the intervention going, and sure enough, we were just about ready to go when the revolution happened. Even after the revolution we were told the project was to go ahead, and a first set of villages were selected for a baseline survey. We did a baseline survey of 13,413 households and 2,525 microenterprises in Menya. Then two months after conducting this baseline, the intervention was cancelled. We have put all the data and survey instruments from this baseline in the World Bank’s Open data library and hope it may be useful to researchers for other purposes: I’m using it for one multi-country microenterprise project at the moment on the role of business practices, but I am sure there are other interesting uses.
A key reason for cancellation was that microfinance organizations were reluctant to consider moving into new areas with all this uncertainty going on, and unsure about even whether their existing clients would want to borrow. This led us to thinking about how to overcome this constraint, and talking with several microfinance organizations about what could be done.

Testing macroinsurance in Alexandria
We ended up partnering with Egypt’s largest microfinance organization, Alexandria Business Association (ABA) and designing a product tailored at their microenterprise borrowers who were finishing one loan cycle, and about to decide whether or not to take out a new loan. We worked with a sample of 2,961 clients who were randomly allocated to a treatment and a control group. The data for this second experiment are also out in the microdata library.

The treatment group was offered a new insurance product which would cost 0.5 percent of the loan value, and which would pay out if one of five specified economic or political shocks occurred in Egypt. 

A key challenge for us was how to price this product – 0.5% was chosen by ABA based on what they thought would be a sustainable long-term price for their clients, and would be easy to explain. But to know whether this would be actuarially fair or not would require us knowing the probability of events like “Egypt imposes a curfew of 14 hours or more a day in Alexandria for 5 or more days in a row at some point in the next year”. This is obviously not like pricing rainfall insurance where you can rely on years of history and stationarity to determine the probability. We elicited subjective probabilities of these events occurring from both the microfinance clients and from a number of economists. All suggest that the price charged was better than actuarially fair, but that there was considerable uncertainty about the probabilities.

We find:
  • High demand for insurance at this price – 36.7% of clients purchased it. It could only be purchased conditional on renewing a loan (which 68% of firms did), so the take-up was 55% among those renewing.
  • No impact of having this insurance on firm behavior – no change in the likelihood they take a loan, the size of a loan, how they spend this loan, or on their investment decisions. Revenues and profits do not increase, and we actually get a significant negative impact on revenues.
Despite the overthrow of the Morsi government and other shocks during our insurance time period, our macroeconomic triggers came close to being reached, but were not triggered. This had two consequences:
First, we could not measure the value of the insurance to borrowers in a state of the world when it pays out – so just as we don’t want to conclude fire insurance is useless by only observing the behavior of people whose houses don’t burn down, we can’t know how beneficial this insurance is to firms by only observing the no pay-out state.
Second, the fact that the product didn’t pay out despite this turmoil completely eroded trust in the product, and meant that when we attempted to sell it again in a second year we had very little take-up. A big factor here was that loan officers, who had recommended it to clients in the first year, were the ones who received client complaints, and then were averse to recommending it to clients in the second year, in order to protect themselves from future complaints.
Many of the first rainfall insurance pilots had rather disappointing results, with very low product uptake and no changes in farmer behavior, but more recent studies have shown the importance of building a reputation and trust over time, and have found this insurance to lead farmers to take more risks and earn higher incomes. This is the first pilot we are aware of to try out insurance for urban businesses, so I hope it isn’t the last and other insurance products can be developed and trialed to help such businesses deal with the enormous volatility in income they face on a daily basis, let alone in the face of large macro and political shocks.
As a final note, I’d like to thank the funders of this research, especially 3ie, for their flexibility in allowing us to adjust the impact evaluation in the light of all these events. Unfortunately many research funds have very rigid and short timeframes, and if we had only had 2 years to spend or end, this work would never have happened. So thanks for allowing us to make this happen!


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000