Effects of State Expenditure on the Primary Completion Rate in Burkina Faso

The purpose of the paper is to identify and analyze the effects of budget expenditures (operation, salary, and investment) on the completion rate of primary education. The annual education budgetary data from 1970 to 2013 collected in Burkina Faso were used to estimate a parametric education production function. The results show a positive effect of unitary investment expenditure and a negative effect of unit staff salaries and budgetary instabilities. Pay mechanisms in public administration should be redesigned to reward proven performance and skills. Moreover, the control of budget programming would lead to better results in the education sector.


Introduction
The effect of inflow resources in the education sector and the results obtained in the sector have been a permanent concern in developing countries. Indeed, it is commonly observed that important resources are made available to the education sector that do not generate tangible results. In fact, the results obtained in the education sector regarding mobilized financial resources are mixed [1] [2] [3]. The combination of inputs to obtain an educational product is a black-box process [4]. The inefficiency of financial resources to induce effective outcomes in the education sector is the argument most often cited to explain this situation [5]. Yet, the availability of resources is recognized as a necessary condition for achieving education goals [6]. However, examining the growth dynamics of public budget resources and the evolution of education indicators raises budgeted public spending [3] [7]. This is even more worrying for indicators of universal S. Ouedraogo primary enrollment. In 2015, the median share of the gross domestic product (GDP) devoted to education was 4.7% worldwide and public education expenditure of the total public expenditure was 14.1%. In contrast, the completion rate for primary education was 83% and 56% of students did not attain the minimum level of reading skills [6]. On the other hand, budget expenditures are likely to increase, and financial spending on education should adapt to the growth of the school-age population in developing countries and to changes in enrollment [8].
Sub-Saharan Africa spent 4.1% of its GDP and 16.9% of the total public expenditure on education, with a completion rate of 59% [6]. As for Burkina Faso, spending on education accounts for about 5.9% of GDP and 18.27% of the national budget [9] [10]. In addition, more than 60% of the education budget and 3.8% of the GDP were destined for primary education [4]. The completion rate was 57.6% in 2013. Education objectives pursued by the country will be difficult to achieve, despite the fact that it has additional financial resources in the context of free primary education.
An education production function is generally used to understand the relationship between resources and academic achievement. The inputs are usually a combination of education and financial variables. The overall budget of the education sector or the total expenditure per student is the commonly used financial variable. Budget expenditures are capital expenditures and operating expenses. In addition, operating expenses are subdivided into strict salary and operating expenses. Schools operate in a political environment that is almost always defined by governments that set programs, fund schools, regulate labor laws, and set rules for certification and the hiring of teachers. The analysis of the effects of expenditures on educational outcomes would be consistent in the relationship between public budget expenditure categories and educational outcomes at the national level. Since the educational process is cumulative, the primary completion rate is the indicator to consider. Completion of primary education is a prerequisite for sustainable literacy [11].
This study is part of the empirical research on the effect of public budget spending on educational outcomes in Burkina Faso. Few studies in Burkina Faso have analyzed the effect of financial resources on educational outcomes in Burkina Faso. Except the study by [4], most research has made a comparative statistical description of the evolution of educational outcomes or goals and mobilized financial resources [10] [12]. To date, no study has been only carried out on the causal relationship between state expenditure and the Primary Completion Rate in Burkina Faso. Yet, this type of analysis is necessary in the context of durable development goal in which the education sector is involved more money to have quality education in equity. This paper aims to fill this gap. Using the education production function approach estimated by OLS procedure at the macroeconomic level, this study empirically establishes the causal relationship between teacher productivity measurement mechanisms and salary remuneration to in- The purpose of the article is to identify and analyze the effects of budgetary expenditures on the completion rate of primary education. We assume that salary expenditures have a positive effect on the completion rate of primary education.
In fact, salary expenditure is the largest part of the budget devoted to education.
In Burkina Faso, the proportion of the budget for staff costs has always exceeded 65% of the primary education budget. Disaggregated data on budget expenditures covering the period 1970-2013 are used to for econometric estimates.
The article is organized as follows. Section 2 presents a brief literature review.
Section 3 provides descriptive statistics of public expenditures in primary education. Section 4 outlines the methodology used in the study. Subsequently, Section 5 presents the empirical results of the research, while Section 6 provides a discussion of the results. The last section concludes the study.

A Brief Literature Review
The unexpected results found regarding the role of educational resources on student performance [13] led to the intensification of the analysis of the relationship between public spending and educational outcomes. Several arguments have been put forward to describe the effect of education spending on educational outcomes and have led to conflicting conclusions [3]. Even though the relationship between education expenditures and educational outcomes has been the subject of intense studies, economists are still unable to agree on the type of causal relationship existing between financial resources and student knowledge acquisition (Table 1). For example, while [5] found that financial resources do not influence student knowledge acquisition or have little effect on education indicators, [16] found that ordinary spending on education per student and total expenditures have negative effects on educational outcomes. Lee and Barro [18] also found that ordinary expenditures of education per student have no significant effect on the test scores of students. Similar results were found for China, Sri Lanka, Zimbabwe, Colombia, Ghana, and Senegal in the study by [22]. They found that ordinary public expenditures on primary education have no significant effect on education indicators in these countries. Regarding school enrollment rates, expenditures per pupil in primary school have significant negative effects on this indicator [15]. McMahon [19] used the ordinary least squares  [12]. The study by [4] analyzed the effects of budget Teachers are more closely related to student outcomes.
Ouédraogo [4] Burkina Faso Ordinary least squares (OLS) in the error correction model Positive effect of student-teacher ratio.
Short-term negative and long-term positive effect of percentage of qualified teachers.
Long-term negative effect of overall repetition rate.
Not significant in the short-and long-term positive effect on the literacy rate.
Negative effect of unit costs of budgetary public expenditure.
Positive effect in the short-and long-term total operating expenditure.
Positive effect of salary unit expenditure.
Positive effect of unit investment expenditure.
Al-Samarrai [15] UNESCO countries OLS Expenditure per pupil in primary school has a positive significant effect on the survival rate only up to Grade 5 and a negative significant effect on school enrollment rates. The other coefficients are not significant.
Hanushek and Kimko [16] Countries of the International Association for the Evaluation of Educational Achievement (IEA) and International Assessment of Educational Progress (IAEP) OLS Student-teacher ratio is not significant.
Ordinary spending on education per student has a negative effect.
Total expenditure on education in percent of GDP has a negative effect.
Wößmann [17] United States of America, developing countries, OECD countries, East Asian countries Weighted least squares International differences in student performance are not caused by differences in schooling resources but are mainly due to differences in educational institutions. Class size has a positive effect.

Continued
Lee and Barro [18] Ten countries of the Middle East/North Africa, 23 countries of Sub-Saharan Africa, 23 countries of Latin America/Caribbean, 10 countries of East Asia/Pacific, 7 countries of South Asia, 23 OECD countries, 9 countries that formerly and centrally planned economics OLS Pupil-teacher ratio has a significant negative effect on scores in math, science, and language.
Average treatment of teachers has a significant positive effect on test scores.
Ordinary expenditure of education per student is insignificant.
McMahon [19] World and regional (Africa, Latin America, Asia, and OECD) OLS Ordinary public expenditure on primary education in percent of GNP has a significant positive effect on gross enrollment rates of boys and girls in primary school Grade 5 and the completion rate for boys and girls.
Ordinary public expenditure per elementary student has a significant negative effect.
Gupta et al. [20] Sub-Saharan Africa, Asia Latin America, Caribbean transition countries OLS and two-stage squares (2SLS) Education spending on primary and secondary education (as a percent of total education expenditure) has a significant positive effect on the gross enrollment rates in primary and secondary education, retention in 4th year of studies.
Education expenditure (in percent of GDP) has a positive and significant effect only on the secondary school enrollment rate; Effects on the other two variables are not significant. Difficulties of assessing the effect of resources on educational outcomes are related to methodological problems [23]. Most studies use the education production function or perform statistical analyses [5]. The conceptual framework considers that schools produce knowledge using educational resources and teacher characteristics as inputs. The acquired knowledge is influenced by the duration of schooling, school resources, characteristics of teachers and learners, characteristics of the household, and school inputs that are controlled by parents (textbooks, daily attendance, etc.). Frequently used independent variables are education facilities, teaching materials, characteristics of the teacher or the principal, school organization or management, incentives or motivations, competition between schools, and the amount of expenses per student. The dependent variable is the gross enrollment rate, number of years of schooling, or test scores [24]. An implicit assumption of homogeneity of skills is accepted over time and within countries.
The education production function approach is more microeconomic. Yet, current applications and interpretations go beyond this theoretical framework.
In addition, the use of scores tests does not integrate the cumulative process of Most of salary expenditures consist of teacher payments. It is usually stipulated that the completion of primary education is hindered by the high costs of teacher salaries and generally, by the costs of operating expenses [9]. In doing so, the tendency is to reduce average teacher remuneration to the level of the reference value of 3.5 times the GDP per capita. The goal is to reduce relative wages in order to generate additional resources for the education sector [24]. There is a focus on the possibility of increasing the supply of education to achieve successful educational outcomes [25].

Descriptive Analysis of Budget Expenditures in Primary Education in Burkina Faso
Public effort for education in Burkina Faso shows an increase in public resources allocated to this sector. Budget allocations for education reached 5.46% of the GDP in 2014 against 0.83% in 1970. Thus, Burkina Faso is within the average of African countries in terms of the resources allocated to education today. The effectiveness of the transformation of resources into educational outcomes is seen through expenditures carried out under the implicit assumption that resources provided for education were effectively devoted to the latter regarding the rules and principles of budget execution and control. Figure 1 reveals an increasing trend of the gross enrollment ratio, net enrollment rate, primary completion rate, and gross admission rate. However, unit costs of total budget expenditures of the Ministry of Education are decreasing. Thus, the growth of education indicators is achieved simultaneously with a decrease in the unit costs of primary education. The disaggregation of the Ministry of Education total unit costs into unit expenditures yields similar results, as displayed in Table  1

Specification of the Econometric Model
For the econometric estimations, an education production function will be used.
According to Glewwe and Kremer [28], the structural function of knowledge production is given by the following: where A stands for knowledge acquired schooling, S is schooling, Q is a vector of school inputs and teacher characteristics, C is the vectors of student characteristics, including inherited skills, H is vectors of household characteristics, and I is the vector of school inputs under the control of parents (school frequency, acquisition of textbooks, etc.). Used both at the micro and macroeconomic levels after transformation, the education production function is more of an empirical approach to the educational production process than a genuine theoretical framework for study [29].
The empirical analysis of the relationship between education expenditures and outcomes is portrayed by the following equation:

( )
, , where Y is the education output (primary completion rate or gross enrollment ratio or score tests); X denotes the education expenditure (unit expenditure or

S. Ouedraogo
share of GDP allocated to education); W represents school characteristics (pupil-teacher ratio, teachers' level of education, teachers' experience, class size, and peer group); Z indicates the family characteristics (parents' level of education, number of household members, family income, and family location); and μ is the error term.
Our research is conducted at the macroeconomic level. The data are annual education budget data. Data on family characteristics are unavailable. Since the data are essentially those related to the financial dimension of the analysis, the model (2) must be adapted to take this constraint into account. This gives the following: where Y t stands for the completion rate of primary education at time t; X it denotes the budget variables (operating budget, salary, and investment expenditures of the Ministry of Education); β is the column vector of unknown coefficients of the estimated parameters, and ε t indicates the error terms. Variables and data sources are presented in Section 4.2.

Variable Specification and Data Source
Along with the endogenous variable and the primary completion rate, the va-

Unit Root Test
The augmented Dickey-Fuller (ADF) test, Philipps-Perron (PP) test, and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) tests were applied to test for the existence of unit roots. In addition, p, the optimal lag length, is determined by the Aikaike or Schwarz and Hannan-Quinn information criterion. The estimations give an optimal lag of 1. The results of unit root tests are displayed in Table 3.
Data in this table indicate that operating expenditures (opex), investment, and salary expenditures are stationary at the level. The primary completion rate is integrated at an order of 1.

Cointegration Test
Since the stationary test revealed the existence of variables integrated at the order of 1 and others that are stationary at the level, the presence of a long-term relationship between the series must be tested. The Johansen-Juselius test is performed. The results in Table 4 show the absence of cointegration relationships.
At the 5% level, the trace test shows that there are no cointegration equations that could be derived from this model. The absence of cointegration relationships is confirmed by the 95% maximum eigenvalue test. Section 5.3 presents the different tests for assessing the quality of the model.

Model Quality Assessment Test
In addition to these two tests, we conducted quality assessment tests of the model. The results of the various quality tests of the models show that the coefficients are individually different from zero and are all significant at the 5% level. The  The estimation of this model gives the results presented and discussed in Section 6.

Results and Discussion
In view of the results of the various specification tests, an OLS procedure is adopted to conduct econometric analyses. Also, the unavailability of disaggregated data of budget expenditures executed in municipalities in Burkina Faso does not allow for estimation analysis in panel data. The results of the estimation are presented in Table 5. Data in Table 5 [31]. The negative effect of salary treatment on the completion rate of primary education could be explained by various factors. In Burkina Faso, there are many substitute teachers in urban areas who, while contributing to the increase of the wage bill in the education sector, have a low participation in the production of education services. In addition, it should be noted that the wage bill for education includes the salaries of administrative or clerical staff. Beyond these elements, the phenomenon of teacher absenteeism and the non-implementation of the total number of hours assigned to teachers greatly contribute to reducing the productivity of education. Moreover, many teachers are assigned to tasks other than teaching and are always paid by the budget of the Ministry of Education. This poses the unfairness of the use of teachers and the lack of a reward system based on actual professional performance.
The instability of budget execution has a negative and significant effect on the completion rate of primary education. A 1% increase in budget instability decreases the primary school completion rate by 0.0552%.
In the light of all the above, what conclusions have been drawn?

Conclusions
This study aimed to identify and analyze From a policy perspective, the results highlight the importance of the fact that the teacher remuneration structure must be based on productivity with a financial incentive policy. Since salary is a response to the merits and efforts provided, remuneration should be indexed on productivity and the obtained results. Given the fact that investment has no significant effect on the primary completion rate,

S. Ouedraogo
the investment process needs to be transformed into a genuine investment strategy that ensures the effectiveness of the investments made so that resources spent are used for activities for which they are intended. For school buildings and equipment, flexibility needs to be introduced in budget execution (procurement). The involvement or empowerment of grassroots communities will be an asset.