How to Use the Enterprise Surveys

Over the last decade the World Bank has invested heavily in developing the Enterprise Surveys. In the last 30 months, after the management of the database moved fully to the private sector and finance vice-presidency, all surveys have been put on the web and the methodology is now uniform across all regions. Previously, regions and even country teams could add whatever questions they thought were appropriate, without regard for the cross-country comparability of data. In short, the data are better and more accessible.

Now comes the hard part: making use of them. Traditionally, the enterprise surveys have fed into country-specific Investment Climate Assessments (in World Bank speak - ICAs). These are part analysis, part story-telling. They typically take a year to produce and often have suspect methodology. Some of the policy recommendations that come out of such work are, not surprisingly, naive. Last year, for example, I witnessed a meeting of the ICA team on Mauritius with a government delegation. An author of the report stated that the big issue in Mauritius was lack of reliable electricity, citing as proof the high percentage of businesses that have back-up generators. "These are needed during storms. No electricity grid can withstand the type of storms that sometime come through the island," answered a senior official. Somehow that had escaped the ICA team.

Be that as it may, how can the Enterprise Surveys be put to good uses? Here are 4 ideas:

1. Use them as purely descriptive sources, to show what the typical business in a developing country looks like. 'One day in the life of a business in ..." can teach us a lot about what it is like to be an entrepreneur. What you worry about, how you organize your production, how you deal with the government. One can ask some specific questions: why are there no large businesses in developing countries; do entrepreneurs separate their family finances from the business finances, or do they run a common pool; how are investment and hiring decisions made; how are new export markets opened. It is remarkable that no analysis like this has been done to-date.

2. Use the data to quantify the effect of specific reforms. This is done in tandem with Doing Business data on reforms. For example, Croatia has been a consistent top reformer in Doing Business lately. Use the 2002, 2005 and 2007 enterprise data to see whether businesses have noticed the effects of, say, the property registration reforms. If not, why? If yes, how have they benefited? This combination of data sources can lead to a new sub-field: reform impact studies, that can in turn inform the relative importance of policy advice in one area vs another.

3. Do analysis across regions within a country. In some of the large emerging economies, the Enterprise Surveys cover thousands of firms that can be separated by location and sector. This allows for analysis of the impact of regulation and institutions: doing so within a country eliminates many "alternative hypotheses." For example, one could study the effect of court efficiency across Mexican or Brazilian states on the number of new contractors a business has. A study of this type is Mohammad Amin's study on labor regulation in India, forthcoming in the Journal of Comparative Economics.

4. Study the divergence between perceptions and reality. This can be done in certain topics. For example, the survey asks businesses to rate the complexity of labor regulations; and then separately asks them to say how they would have behaved differently had regulatiosn been more flexible. If a manager said that labor regulation was a big burden, but then also said he would not change his employment startegy if the regulation were made more flexible, a problem arises. In essence, he doesn't know what he is talking about. If so, this analysis tells us something about the usefulness of the perceptions part of the Enterprise Survey, and more generally about the usefulness of all other perceptions surveys - like Transparency International, the Global Competitiveness Report, etc.

There must be some other obvious research opportunities. But this is a good place to start.