1. Impact Evaluation, Kenya Private School Support Program (KPSSP) Felipe Barrera-Osorio (HDNED, WB) Ilyse Zable (CHEDR, IFC)

2. The program IFC initiative to provide local currency financing and technical assistance (TA) to private K-12 institutions. IFC signed (Dec. 2006) a risk-sharing agreement with K-Rep Bank (K-Rep) of up to aprox. US$1.7 million on loans extended to private schools. IFC shares 50 percent of the risk on the pool of loans made to schools after an initial 5 percent first loss taken by K-Rep. Use of loans to finance construction projects, purchase educational materials, and cover other capital expenditures. A comprehensive technical assistance program is under preparation and is scheduled to begin in july of 2007. The TA program will be designed to improve schools’ financial, management, and educational capacities and strengthen K-Rep’s ability to evaluate and monitor loans to schools Target areas are Nyeri, Eldoret, Nakuru, Mombasa, and Nairobi.

3. The TA The TA delivered to schools includes the following:  An introductory workshop that will introduce the K-Rep facility and associated technical assistance, and provide comprehensive training on school self-diagnostics, strategic planning, and business plan development.  Following receipt of the loan, both individualized attention and workshops covering: the installation of and training in the use of a comprehensive EMIS, accounting and financial management, human resource and training management, curriculum and learning management, quality assurance (including self-diagnostic and evaluation processes) and student-level monitoring.

4. Research questions Does the program increase the sustainability of participating schools? Does the program increase access to education? Does the program improve educational outcomes? Is there heterogeneous effects of the program by income among students?

5. Channels of transmissions between program and learning outcomes Efficiency allocation of inputs –Better allocation of resources –Better allocation of teaching time –Better managerial control Direct effect of infrastructure (e.g., better classrooms, bathrooms, etc) Better pedagogical methods (small component of the TA) “Ownership” of the school: students / teachers will take better care of the (improve) school

6. Identification The nature of the risk-sharing agreement between IFC and K-Rep does not allow for randomization of loan recipients. We can use the TA program as the identification mechanism. –Encouragement model and ITT –Take-up rate is critical We can not separate the effects of the loan from the ones of the TA

7. Encouragement model Master list: all schools Potential participants Based on: Num. students Type of school Year of creation CRITERIA OF SELECTION: Based on: Randomization Invited to TANot invited Receive LoanNo Loan Bank criteria (non randomized)

8. Data Master list of schools One base-line survey, starting in July –Problem: TA and survey will be almost at the same time… It is possible to coordinate the collection of base-line information with the phase-in of the introductory workshops. We may want to carry out two follow-up surveys, one year after the base-line and two years after the base line. –We need resources!

9. Variables  Control variables: individual characteristics (gender, age, some proxy for income, demographics of the family, geographic region, etc.) and school characteristics (infrastructure and teacher characteristics)  Exposure to the program: use of EMIS, preparation of business plans, change in corporate governance, preparation of strategic plans, school diagnostics, HR policies.  Individual educational variables: repetition rates, drop-out rates, students absenteeism, standardized test performance (based on national test), self reported hours of study outside of school, self- reported assistance, grades on the last math test, student happiness.  School level variables: enrollment, allocation of teachers’ time, financial indicators (revenues and costs), teacher absenteeism, teacher happiness, teacher turnover, teacher qualification, interaction with parents

10. Issues Non-randomness of treated Sample size Surveys: –Schools –Teachers –Students Standardized test