Can Financial Incentives and Better Communication Improve the Last-Mile Delivery of Textbooks?

education with an allocation of about 20 percent of the government budget, but this investment has not translated into better student learning outcomes in primary education. Among the main reasons for this is the severe shortage of learning materials, particularly textbooks in both English and local languages. A public expenditure review of the Zambian education system has shown that 91 percent of schools lack textbooks and on average five to six pupils share less than one textbook, including textbooks in local languages.1 To address this shortage, the REACH Trust Fund provided a grant to investigate the factors related to the availability of textbooks in schools and to evaluate a set of financial incentive mechanisms for improving the last-mile delivery of textbooks. The questions that the research set out to answer were:

Zambia has invested heavily on education with an allocation of about 20 percent of the government budget, but this investment has not translated into better student learning outcomes in primary education. Among the main reasons for this is the severe shortage of learning materials, particularly textbooks in both English and local languages. A public expenditure review of the Zambian education system has shown that 91 percent of schools lack textbooks and on average five to six pupils share less than one textbook, including textbooks in local languages. 1 To address this shortage, the REACH Trust Fund provided a grant to investigate the factors related to the availability of textbooks in schools and to evaluate a set of financial incentive mechanisms for improving the last-mile delivery of textbooks. The questions that the research set out to answer were: • What is the current  The textbook distribution practices and timelines vary from year to year, making it difficult for the regional authorities and schools in Zambia to predict how many textbooks they will receive and when.

Zambia
There are multiple factors that make textbooks scarce in developing countries. These include: (a) poor planning, management, and monitoring of textbook production and distribution; (b) a lack of data on the availability and use of textbooks; (c) an inadequate storage and distribution system. 2 This study focused on the distribution aspect of textbook supply, in particular, the last-mile delivery of textbooks from regional education offices to schools in Zambia. Specifically, it evaluated the outcomes achieved by using result-based financing (RBF) to try to ensure that textbooks were delivered in a more efficient and timely way. Financial incentives used in RBF can take a variety of forms including pay for performance (P4P), pay for activities, bonuses, performancebased fee schedules, or grants. 3 Among these financial incentive schemes, P4P has been studied the most, but the findings regarding its effectiveness have been mixed. A recent review of the effect of P4P in the health sectors of 14 OECD countries concluded that its impact is unclear and attributed some positive program effects to improved monitoring and data recording rather than to the P4P scheme itself. 4 Several studies have examined the effects of teambased performance pay. One study found that the implementation of team-based performance payments for lower-level public sector workers in the UK had a positive impact on the productivity of small teams but a negative impact on the productivity of large teams. 5/6 This study evaluated a financial incentive scheme that combined "partial pay for activities" and "pay for performance" to improve the system for distributing textbooks in Zambia. The incentives, which consisted of pre-determined bonuses, were given to incentivize agents who were already a part of the last-mile delivery of textbooks-the DEBS offices and schools. The incentive schemes were designed to preclude the possibility of any cream-skimming in the DEBS offices by giving them larger bonuses if they served hard-to-reach clients (schools in remote locations). The study also evaluated whether giving incentives to schools or to DEBS was more effective in getting textbooks to schools faster and more efficiently.
Prior to this intervention, no clear procedure existed for informing schools about the availability of textbooks. It was left to the discretion of the staff of the DEBS offices who sometimes made phone calls to schools or made announcements in unrelated meetings with head teachers or even just posted a notice on their information board. As a result, many schools had no idea if the textbooks were available or not, meaning that the books were often left uncollected and undistributed in the DEBS warehouse or in the zone-center schools. Therefore, in addition to the financial incentive schemes, a new information and communication strategy managed by the MoGE was implemented in all intervention districts.

WHY WAS THIS INTERVENTION CHOSEN?
ZAMBIA 3

HOW DID THE INTERVENTION WORK?
The study evaluated two different incentive schemes to measure their impact on last-mile delivery of textbooks to schools. Scheme 1 provided payments to schools, while Scheme 2 provided payments to the DEBS. Each of the two schemes was implemented in nine randomly selected districts, while another nine districts were selected to be the control group. All 27 districts received an upfront grant of 7,800 ZMW to fund the distribution of textbooks.
In Scheme 1, each remote school was awarded an incentive payment of 700 ZMW (US$50) whenever staff from the school collected its allocation of textbooks from the DEBS office within four weeks after receiving an official notice that they were available. Alternatively, each remote school was awarded an incentive payment of 400 ZMW (US$28) if the school staff collected the textbooks between four and eight weeks after receiving the official notice.
In Scheme 2, each DEBS was awarded an incentive payment of 325 ZMW (US$23) for every remote school and 180 ZMW (US$13) for every non-remote school if the DEBS delivered textbooks to the assigned schools within one month after receiving the textbooks from the MoGE.
During the endline survey, data was collected on textbook availability and distribution for both pre-and post-treatment periods. Distribution of grade two textbooks (procured and distributed in 2016) was the outcome measure for pre-treatment period, and distribution of grade four textbooks (procured in 2018 and distributed in 2019) was the measure for the post-treatment period. The primary survey respondents were the teachers in charge of textbooks, and their answers were cross-checked against relevant MoGE and DEBS data as well as against any textbook records or inventory kept in the schools.
Two different constructs were measured-one for the availability of textbooks and one for the textbook delivery mechanism. The outcome variables for textbook availability were measured by constructing:  Figure 1 describes the lastmile delivery of grades two and four textbooks from DEBS to schools.
After implementation of the new communication strategy, 94 percent of schools reported receiving information about the availability of grade four textbooks. This is an improvement from 69 percent of schools that reported that they had received some type of communication from their DEBS about the availability of textbooks for grade two (Figure 2).
While almost of all schools (97 percent) reported receiving their grade four textbooks, only 49 percent of schools reported having received their grade two textbooks in 2016, which included books from both the new and old curricula because they were delivered after the 2014 curriculum revision. Only 28 percent of schools received the grade two books that were based on the new curriculum ( Figure 3). This low number was partly due to the fact that many DEBS had not yet received any books based on the new curriculum (63 percent). However, even after taking this into account, fewer than half of the schools (44 percent) reported receiving any textbooks. In terms of the textbook to student ratio, four or five students had to share one grade two textbook in one of the Zambian local languages (when combining the numbers of both new and old curriculum textbooks), while in grade four, every student had their own textbook in one of the Zambian local languages.
The distribution time was shortened dramatically for the grade four textbooks, with 90 percent of the schools receiving them within two months. Only 27 percent of schools reported receiving the grade two textbooks within a month after they were delivered to the DEBS offices from the MoGE's central warehouse.
Even after six months, only 31 percent of schools reported having received their books.

Impact of the Incentive Schemes and Communication Strategy
The school incentive had the most significant impact in terms of increasing the probability of the schools receiving textbooks (0.126) compared to the control group (Figure 4), while the DEBS incentive had no statistically significant impact on the probability of schools receiving their textbooks and actually reduced the probability of schools receiving their textbooks within a month by 0.324.
The impact of the financial incentives on the textbook-pupil ratio was not statistically significant. This may have been because the MoGE distributes a pre-determined number of books to each district rather than basing the textbook allocations on the current number of students or the needs of the schools in each district.
The school incentive scheme reinforced the most common distribution practice in which schools collect their textbooks from zone-center schools or DEBS offices. The DEBS incentive scheme had no statistically significant impact on how books were either distributed or collected, but the school incentive scheme increased the probability of schools collecting the books from zone-center schools by 0.356 and increased the probability of them collecting books from DEBS offices by 0.158 at the 5 percent significance level. (Figure 5) Disseminating information to schools about the availability of their textbooks had a positive effect on almost all outcome variables.
The results of the study show that providing information and increasing communication to schools increased the probability of them receiving their textbook allocations by 0.334, increased the textbook-pupil ratio by 0.226, and increased the probability of schools receiving their textbooks within a month by 0.202 at the 1 percent significance level ( Figure 6). This is a bigger impact than the effect of the school incentive. This may indicate that not knowing that textbooks are available because of irregular textbook procurement and distribution schedules has been a vital barrier preventing schools from accessing textbooks. By removing this information barrier, schools are then motivated to collect their books from wherever they are located, either in zone-center schools or in the DEBS office.
Communicating with all schools using multiple channels cost about 2.8 ZMW per school, which is much more cost-effective than the school incentive scheme. The school incentive scheme cost an average of 513 ZMW per school, provided that all remote schools collected their books within a month after they became available at the DEBS level. This is slightly higher than the amount estimated by the DEBS for transporting textbooks to all schools within their districts (506 ZMW). However, improving communication without providing the necessary resources to fund the delivery of textbooks to schools is not a good solution in the long term because schools could be discouraged from collecting them without a proper incentive. It could also exacerbate any inequalities between schools. Providing financial incentives to encourage schools to collect textbooks from the DEBS level could be an effective result-based financing mechanism for increasing the availability of textbooks in schools in Zambia. However, providing financial incentives to DEBS does not seem to be effective in this regard. This is partly because the current textbook delivery practice in districts and communities relies heavily on the schools themselves taking the initiative to collect the books rather than on the DEBS delivering the books to the schools.
The amounts provided in the financial incentives should be carefully considered, given that the cost of collecting or distributing the textbooks is highly dependent upon the geographic accessibility of the schools. When deciding the incentive amounts, policymakers should also consider the intervention's longterm financial sustainability along with any possible potential negative consequences such as discouraging the intrinsic motivations of schools.
In order to successfully design and implement the suggested financial incentive scheme, policymakers would also have to consider the transparency and predictability of the book delivery process. MoGE would have to ensure transparency in textbook distribution and that all schools are informed about the availability of their textbooks at their local DEBS offices. They would also have to ensure that the textbooks are shipped to the DEBS from the central government in a predictable and organized manner.
The current practice results in every school receiving almost the same number of textbooks with no account taken of their actual needs. There are several reasons for this. Although the procurement of textbooks is based on the total number of enrolled students in the procurement year, it does not predict or accommodate any likely changes in textbook demand in the subsequent year. This, along with the prolonged procurement process and poor budget execution by the central government, distorts the supply and demand of textbooks at the regional and sub-regional levels. Furthermore, at present, the MoGE has no clear and consistent guidelines for textbook distribution. This may explain why the textbook-pupil ratio was not affected by any of the study's financial incentive schemes. The fact that textbook distribution takes place on an ad hoc basis makes the textbook-pupil ratio in schools only loosely correlated with the probability of schools receiving their textbooks.
It is recommended that the MoGE should establish a procurement plan for textbooks based on a reliable The school incentive had the most significant impact in terms of increasing the probability of the schools receiving textbooks, while the incentive to District Education Board Secretariats had no significant impact.

CONCLUSION
Providing financial incentives to schools to encourage them to collect textbooks from the DEBS level has the potential to be an effective results-based financing mechanism to increase the availability of textbooks in schools in Zambia. However, providing financial incentives to DEBS does not seem to be as effective in this regard. This is partly because the current textbook delivery system in districts and communities relies heavily on the actions taken by schools, not those taken by DEBS. As a result, providing incentives to DEBS does not change their behavior, but providing incentives to schools magnifies the positive effects of the current arrangements and yields better results. In addition, providing information to schools about the availability of textbooks at the DEBS level also significantly increases the number of textbooks available in schools at minimal cost.