Panel Paper: Letting the Driver Steer: Country Context and Organizational Design in Delivering Better Aid

Thursday, November 6, 2014 : 1:00 PM
Dona Ana (Convention Center)

*Names in bold indicate Presenter

Dan Honig, Harvard University
This work examines an understudied component of aid effectiveness, the organizational features of international development organizations (IDOs). This paper examines whether, when, and how organizational autonomy affects project success.  It employs regression analysis of a novel dataset—evaluations of over 14,000 projects from nine international development organizations—using self-evaluated project outcomes as a measure of success, the State Fragility Index as a measure of environmental unpredictability, and both expert surveys and a measure constructed from organization-level responses to Paris Declaration monitoring surveys as measures of aid organization autonomy.  

The key finding is that organizational autonomy matters to project success, with increasing returns to autonomy in fragile states and in project domains where it is more difficult to externally observe (and thus contract on) outcomes.  Comparing recipient-country environments one standard deviation above and below the mean, a relatively high-autonomy development organization would see a difference of about .05 points in performance on a six-point scale, while a relatively low-autonomy development organization would see more than 10 times the difference.  High-autonomy organizations, then, see more consistent performance across countries.  This effect is concentrated in sectors in which it is difficult to contract on accurate output measures (such as capacity building) rather than in sectors in which such measurement is relatively straightforward (such as road construction).  Inasmuch as measurement (particularly legitimacy-seeking output measurement) is a constraint on organizational autonomy, this augurs for less organizational navigation by measurement and more organizational navigation by judgment in more unpredictable environments and less contractible task domains.