**3.1 Model specification**

It is broadly recognized that participatory development has played a prominent role in the achievement of projects. Despite increasing advocacy, it is still questionable whether the inclusion of the beneficiary community in project management could elongate the serviceable durability of community-driven development (CDD) projects.

The aim of this study is to assess the sustainability of CDD projects. This study intends to investigate a wide range of factors, potentially determining the durable existence of CDD projects. The outcome variable of the current study is dichotomous, coding one if a CDD project is well usable and zero if that project is not currently usable or broken. Given a binary response to the usability of the project as a dependent variable, there are several techniques applicable to estimate the equation. Linear Probability Model (LPM) is a straightforward approach that can be used in this context. This technique is a linear regression estimated by the Least Square method. Despite its simplicity, LPM is possibly subject to many shortcomings. The most critical constraint is that this model violates an important as sump that the predicted outcomes should bound in the restrictive range of zero and one [11].

Alternative approaches, in addition to LPM, are Logit and Probit Models. These two models are non-linear techniques estimated by the Maximum Likelihood method. While the Logit Model is reliant on logistic distribution, the Probit Model estimates the equation under a normal distribution. Since there is no convincing reason to justify the superiority of one to another, this research employs the Logit Model to investigate the extent to which community participation, monetary contribution, poverty rate, project types, and their locations have *Assessing the Sustainability of Community-Driven Development Projects in Lao PDR DOI: http://dx.doi.org/10.5772/intechopen.96406*

considerable impacts on the persistence of CDD projects. The structure of the Logit Model is shown and explained as follows:

$$\begin{split} L\_i = \ln \left[ \frac{Prob(Y=1)}{Prob(Y=0)} \right] &= \beta\_0 + \beta\_1 PRF\_i + \beta\_2 CF\_i + \beta\_3 PR\_i + \beta\_4 FS\_i \\ &+ \beta\_5 PO\_i + \sum\_{j=1}^{I-1} \delta\_j TP\_{ij} + \sum\_{k=1}^{K-1} \theta\_k LP\_{ik} + u\_i \end{split} \tag{1}$$

where *L*i denotes logit which is the logarithm of ratio between the probability.

that a CDD project is currently usable, *Y* = 1, and the probability that this project is not currently useable, *Y* = 0. *PRF*i represents the share of Poverty Reduction Fund's money contributed to the project *i*. *CF*i is the share of community's money contributed to the total value of project. *PR*i stands for participation rate which is the proportion of households participating in the project over total number of households in the village. *FS*i denotes projects selected by females in the village. *PO*<sup>i</sup> is poverty rate which is the ratio of poor villagers over the total number of villagers. *TP*ij represents project type *j*, including gravity-fed water system, Projects related to health, transportation, and projected related to education. *LP*ik denotes the location of projects in province *k*, including Phongsaly, Huaphan, Luang Namtha, Luang Prabang, Oudomxay, Xiengkhuang, Savannakhet, Saravan, Sekong, and Attapue. *β*<sup>0</sup> is constant term. *β*1 to *β*5 are the parameters of PRF's contributed money, community's contributed money, females' involvement in the selecting process, and poverty


*Note: Irrigation and energy projects are reference groups for types of community-driven projects. Projects in Attapue province are reference groups for the location of projects.*

#### **Table 1.**

*The description of variables in the logit model.*

rate, respectively. *δ* j and *θ*k are parameters of explanatory variables representing the types of projects and their provincial locations. *u*i is the stochastic disturbance of equation.

Before proceeding to analyze the sustainability of the CDD project, it is neces sary to draw particular attention to what project sustainability in the context of this study is. There is no consensus definition of project sustainability in the literature.

Since this study considers many types of CDD projects altogether, it is hard to define what the sustainability of the project exactly means. To overcome this indistinctness, this analysis uses a loose meaning of project sustainability. Based on a study of Chatterley et al. [12], the sustainability of CDD projects in this study is defined as if the project is not visibly dilapidated and still well workable. In other words, it means that the project functions appropriately without any significant repair needs, at least during the reference period of the survey. The description of this indicator and other variables attached in the empirical analysis are explained in **Table 1**.

### **3.2 Data descriptions**

Main data sources are from secondary data of suitability assessment in 2016 and 2019. In 2016, the assessment was organized in PRF's 10 targeted provinces for the


#### **Table 2.**

*The summary statistics of variables.*

#### *Assessing the Sustainability of Community-Driven Development Projects in Lao PDR DOI: http://dx.doi.org/10.5772/intechopen.96406*

project's establishment during 2012–2016 that includes 1,930 sub projects. In 2019, the assessment was organized in 10 provinces for the project's establishment during 017–2019 that include 1,169 sub projects. Therefore, total sub project during 2012–2019 are 3,099 (**Table 2**). More than 100 sub projects for each province has been evaluated. Approximately 696 projects or 22% of total have been assessed in Huaphan province.

In respond to a research question whether the contribution of the community, the involvement of female villagers, poverty rate, project types, and project locations by provinces do matter for the sustainability of CDD projects, this research is mainly reliant on a database of Poverty Reduction Fund (PRF). The dataset con tains the information of projects constructed from 2012 to 2019. The current study intends to emphasize CDD projects completed during 2012 and 2016. Those projects built and transferred to communities recently are not included in the analysis. This study solely focuses on all construction projects. Subprojects related to providing equipment and materials are excluded from the empirical analysis. After cleansing and removing missing data, the econometric analysis of this study is based on 1,574 projects.
