ABSTRACT
Rigorous microfinance impact studies are rare and most fail to control for selection bias, undermining estimates of impact. It is argued that microfinance clients commonly self-select into a programme or are selected by their peers or the microfinance loan officers, biasing access to loans against the poorer. Further the selection or screening processes are driven by unobservable characteristics of clients such as entrepreneurial abilities, access to social networks, risk taking preferences and business skills (Coleman 1999). These characteristics are difficult to measure and are poorly dealt with or neglected by advanced measurement techniques resulting in misleading estimates of outcomes.
