The Role of Intergovernmental Fiscal Transfers in Improving Education Outcomes - Annex

Jul 19, 2026

Annex material supporting the World Bank publication on how intergovernmental fiscal transfers influence education outcomes. It presents additional empirical approaches, including dynamic panel data models (DPD) that use lagged values to address endogeneity, alongside simpler fixed effects and OLS options when data constraints exist. The annex further summarizes an illustrative Uganda case study, including sector budget composition, grade transition patterns, and the structure and objectives of local government performance assessment-linked grants.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages42

Reading Time1 h 46 min

Introduction

Chapter 3: Alternative Empirical Approaches

Chapter 5: Uganda Case Study

Annex Figure 5.1: Budget by Sector, 2019/20

Annex Figure 5.2: Grade Transition, 2017-on-2016 Data

Annex Table 5.3: Main Transfer Mechanisms

Public Disclosure Authorized Public Disclosure Authorized The Role of Intergovernmental Fiscal Transfersin Improving Education Outcomes Annex Friday, June 25, 2021 EDU Introduction This annex contains additional information and results from the publication: Al-Samarrai, Samer, and Blane Lewis, eds. 2021. The Role of Intergovernmental Fiscal Transfers in Improving Education Outcomes. International Development in Focus. Washington, DC: World Bank. doi:10 . 1596/978-1-4648-1693-2. License: Creative Commons Attribution CC BY 3.0 IGO. The annexes are ordered by the chapters in the aforementioned publication. Not all chapters have annexes. Chapter 3 Alternative Empirical Approaches Lewis and Smoke (2017) and Lewis (2017) illustrate how Dynamic Panel Data (DPD) models can be used to estimate the effect of intergovernmental transfers on education spending and education spending impact on education outcomes, respectively. A positive aspect ofDPD models is that they can be used to estimate causal effects when some of the explanatory variables of concern are endogenous, by instrumenting those variables with their lagged values. The down-side is that the models are quite data intensive and somewhat complicated to estimate. Such models may also be quite sensitive to small changes in specification. Roodman (2007) provides an introduction to DPD models (both difference and systems GMM types) and their implemented in Stata. If such models cannot be employed, due to data deficiencies, for example, then it may be sufficient to estimate the desired impacts by using simple fixed effects/OLS methods. Consider the following two equations. Sit = 􀁄 + 􀁅1Tit + 􀁅2Xit +􀁋i +􀁊t +􀁈it (1) Oit = 􀁄 + 􀁅3Sit + 􀁅4Xit +􀁋i +􀁊t +􀁈it (2) In the above specifications, i and t are local governments and time, respectively; S is (log) per capita education spending; T represents (log) intergovernmental transfers per capita; O indicates education outputs, variously defined; X is a set of exogenous control variables; η and γ are unit and time fixed effects, respectively; ε is the error term; and α, β 1, β2 β3, and β4 are the parameters to be estimated. The above equations can be estimated using standard fixed effects techniques. If data are only available for a single time period, then the models can be estimated without fixed effects (η and γ) by OLS. If transfers and/or spending are endogenous, however, then the fixed effects/OLS estimation of relevant parameters will be biased. The extent of the bias cannot be determined a priori and will vary case by case. Sometimes the bias will be severe, rendering the estimates implausible; sometimes the bias may be acceptable, generating coefficient estimates with appropriate signs and reasonable magnitudes. Note that the lagged dependent variable may also be included on the right-hand side of the equation. The lagged dependent variable would also be endogenous, and its endogeneity must also be accommodated. References Lewis, B. (2017). Local government spending and service delivery in Indonesia: the perverse effects of substantial fiscal resources, Regional Studies, vol. 51, no. 11, pp. 1695-1707. Lewis, B. and Smoke, P. (2017) . Intergovernmental fiscal transfers and local incentives and responses: the case of Indonesia, Fiscal Studies, vol. 38, no. 1, pp. 111–139 Roodman, D. (2007) . How to do xtabond2: An introduction to difference and system GMM in Stata. Stata Journal, 9(1), 86–136. Chapter 5: Uganda Case Study Annex Figure 5.1: Budget by Sector, 2019/20 Ushs trillion | Education Interest Payments Water and Env., Works and Trans. Security Public Sector Mgmt., account'y, admin'n LG discretion. Energy Other Health Agriculture | Trillions 20 19/20 Ushs . | 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 | | | --- | --- | --- | --- | Note: the percentage indicates the percentage change in the budgeted amount since 2015/16 . Anne

FAQs

What empirical methods are discussed for estimating the effect of fiscal transfers on education outcomes?

The annex discusses dynamic panel data (DPD) approaches to estimate causal effects under endogeneity, and it also notes fixed effects/OLS as an alternative when DPD cannot be applied due to data deficiencies.

Why can fixed effects or OLS estimates be biased in this context?

If transfers and/or spending are endogenous, then fixed effects/OLS estimation of key parameters can be biased; the bias magnitude depends on the specific case.

What does the Uganda case study annex provide besides methods?

It provides sector budget composition (Figure 5.1), grade transition and enrollment outcomes (Figure 5.2), and a breakdown of main transfer mechanisms, including performance-based development grant components (Table 5.3).

Jul 19, 2026

Upload by Seraphina