Person:

Pritchett, Lant

Firms in Africa report "regulatory and economic policy uncertainty" as a top constraint to their growth. We argue that often firms in Africa do not cope with policy rules, rather they face deals; firm-specific policy actions that can be influenced by firm actions (e.g. bribes) and characteristics (e.g. political connections). Using Enterprise Survey data we demonstrate huge variability in reported policy actions across firms notionally facing the same policy. The within-country dispersion in firm-specific policy actions is larger than the cross-national differences in average policy. We show that variability in this policy implementation uncertainty within location-sector-size cells is correlated with firm growth rates. These measures of implementation variability are more strongly related to lower firm employment growth than are measures of "average" policy action. Finally, we show that the de jure measures such as Doing Business indicators are virtually uncorrelated with ex-post firm-level responses, further evidence that deals rather than rules prevail in Africa. Strikingly, the gap between de jure and de facto conditions grows with the formal regulatory burden. The evidence also shows more burdensome processes open up more space for making deals; firms may not incur the official costs of compliance, but they still pay to avoid them. Finally, measures of institutional capacity and better governance are closely associated with perceived consistency in implementation.

Firms in Africa report "regulatory and economic policy uncertainty" as a top constraint to their growth. We argue that often firms in Africa do not cope with policy rules, rather they face deals; firm-specific policy actions that can be influenced by firm actions (e.g. bribes) and characteristics (e.g. political connections). Using Enterprise Survey data we demonstrate huge variability in reported policy actions across firms notionally facing the same policy. The within-country dispersion in firm-specific policy actions is larger than the cross-national differences in average policy. We show that variability in this policy implementation uncertainty within location-sector-size cells is correlated with firm growth rates. These measures of implementation variability are more strongly related to lower firm employment growth than are measures of "average" policy action. Finally, we show that the de jure measures such as Doing Business indicators are virtually uncorrelated with ex-post firm-level responses, further evidence that deals rather than rules prevail in Africa. Strikingly, the gap between de jure and de facto conditions grows with the formal regulatory burden. The evidence also shows more burdensome processes open up more space for making deals; firms may not incur the official costs of compliance, but they still pay to avoid them. Finally, measures of institutional capacity and better governance are closely associated with perceived consistency in implementation.

"Doing Business" (DB) provides measures of the time and costs associated with fully complying with an array of business regulations. Enterprise Surveys (ES) ask a wide range of firms about their actual experiences in doing business. We use three comparable indicators in both: time to get an operating permit, time to get a construction permit, and time to import goods, to compare these distinct de jure (DB) and de facto (ES) approaches to assessing the "investment climate" in over 100 countries. Four patterns emerge in each of the three indicators. First, while the DB, of necessity, reports a single estimate of the days for compliance for each indicator, firms in the same country report wildly different times to complete the same transaction. For instance, the DB indicators says it takes 65 days to start a business in Ecuador, whereas the distribution of the 265 firms who reported getting an operating license was between a 10th percentile reporting 1 day while the 90th percentile reporting 60 days. Second, regulatory compliance appears to be "under water" as firms report actual times much less than the DB reported days. For instance, the median DB time to obtain a construction permit across all countries is 210 days, while the mean of the days firms reported was 59, the median DB-ES "under water" gap was 145 days. Third, cross-nationally there is very little association between the ES distributions and DB numbers. A naïve view of "full compliance" might suggest that actually reported days would rise one for one with DB days, but the patterns are much more complex. The de jure environment appears to only affect some firms. For instance, as the de jure DB time to get a construction permit rose across countries by 524 days (from 77 to 601) the predicted time reported by the 25th percentile "favored" firm rose by just 2.7 days. In contrast, the time reported by the "disfavored" 90th percentile firm rose by 130 days, but all of this rise comes in the lower range of the DB. Fourth, for those countries with multiple ES surveys we find little association over time, with reductions in DB days as likely to be accompanied by increases in ES days. Comparing these two measures suggests very different ways of thinking about policy versus policy implementation, about what "climate" means, and what the options for "policy reform" really are.