Researchers and development practitioners have been making an effort to fill this gap, but they are faced with a series of challenges including: (1) safety concerns for field work, (2) changing circumstances on the ground, (3) lack of reliable macro numbers due to outdated or non-existent census or registry data, and (4) low response rates to any data collection efforts.
Haiti, a country whose development has been hindered by instability and disaster risk, is no exception. However, after several attempts over 15 years, the World Bank’s Enterprise Analysis team has completed and published its first ever survey of establishments in Haiti. The data provides invaluable insights into the challenges the private sector faces in an especially difficult operating environment.
Social unrest was widespread during the period of data collection. Still, a team of enumerators were able to complete 149 interviews with top managers on behalf of the World Bank.
How did we address these challenges?
Our team of interviewers and the businesses operating in Port-au-Prince were clearly affected by the unrest. Our data showed that 45% of businesses incurred costs for security and protection. Given the circumstances, the standard Enterprise Survey methodological approach had to be adapted to meet the dual goal of (1) prioritizing the safety of the team and (2) generating quality establishment-level data. These adjustments meant that the survey would not be fully comparable with all other Enterprise Surveys and the associated indicators were not published on the Enterprise Survey website, but the micro-data is available in the ES portal for download.
For implementation purposes, the sampling area was restricted to the Port-au-Prince Arrondissement. It soon became evident that certain parts of the city were too dangerous to work in. Fieldwork was consequently restricted to certain communes – Port-au-Prince, Delmas, Pétion-Ville, and Tabarre, as shown on the map above.
The data collection effort relied on a team of enumerators with extensive knowledge of the environment in which they operated. Given the unrest, the situation could deteriorate very quickly. About 28% of the interviewed businesses had experienced losses due to theft, robbery, vandalism, or arson. Therefore, fieldwork had to adapt as circumstances changed.
When it was safe, interviewers would go into the field to screen and interview businesses. For Port-au-Prince and Delmas, fieldwork was contingent on the level of violence of the day, while in Pétion-Ville, thought to be one of the wealthier communes, fieldwork was scheduled for mornings, before the arrival of protesters. Tabarre was considered relatively safe, with less violent protests even when they occurred.
Addressing lack of reference data
No data collection starts from scratch. Underlaying aggregate data is necessary for making inferences about the population. However, environments affected by fragility, conflict and violence (FCV), by their nature, are faced with continuously changing circumstances that affect the baseline. In business surveys, it is often the case that some businesses have ceased to operate while new ones have sprouted up. Under these circumstances, any census or registry data is likely to be out of date and not very useful.
To address this, we updated our sampling frame to reflect the situation on the ground. The sampling strategy began by mapping the locations of all firms in the geographically defined sampling frame. As the enumerators deployed to the field, they not only checked the existence of the businesses that appeared in the existing sampling frame, but also noted any firms that were missing. Such an update to the underlying data more accurately reflects the reality on the ground.
It is not uncommon for businesses in areas marred by violence and conflict to refuse to participate in a survey or close temporarily. Even among businesses in relatively safe parts of Port-au-Prince, 92% reported temporarily closing due to unrest during the past year. Getting this information required a qualified and knowledgeable local team.
Some issues of trust remained among businesses despite willingness to participate in the survey. Nearly 85% of respondents refused to provide any financial information.
The World Bank recently published Data Collection in Fragile States, which describes the challenges of collecting data in places like Haiti. We hope that this long-awaited Haiti firm-level dataset will be used by policy makers and researchers and demonstrate the paradigm shift that study envisions: moving from “there are no data” to “how do we collect data here?” The recently published dataset for Haiti is a good example of this mindset in practice.
The Haiti dataset is the latest in a recent series on businesses in FCV environments published by the Enterprise Surveys team. Similar data for The Gambia, Kosovo, Mozambique, and West Bank and Gaza were released in the past year. Data collection projects are underway in Iraq, Lebanon, Myanmar, and Somalia. New releases will be available on our website.