Page 9 -

ื
ิ
ิ
ิ
์
โครงการหนังสออเล็กทรอนกสด้านการเกษตร เฉลมพระเกียรตพระบาทสมเด็จพระเจ้าอยู่หัว
ิ
vii

ABSTRACT

The objective of this study is fivefold: 1) to apply Latent-class analysis to cluster
rural households into distinct livelihood strategies based on incomes and assets; 2) to rank
livelihood strategies using stochastic dominance analysis; 3) to examine the transition of

household’s livelihood strategies; 4) to determine factors that affect the strategy outcomes
using multinomial logistic regression; 5) to analyze factors that affect the transition of
household’s livelihood strategy using ordered logistic regression. The balanced-panel data
of rural households obtained from the Townsend Thai Project from 2000-2017 was used
in the analysis. The results show that households can be grouped into six distinct strategies.

The households in the strategy 1-3 and 4-6 were classified as high-income and low-income
households, respectively. The change in livelihood strategy was more pronounced for the
former. Many households had repeated the same transition in livelihood strategy overtime.

Key factors that affect the probability of each strategy include primary occupation of
household head, highest level of education and position in the village of household
members, size of farmland with ownerships, size of land rented in and rented out, and
ratio of household members with rights to the Social Security Scheme. Factors that

positively affect the probability of observing an upward transition include size of farmland
with ownerships, whether households had guarantor loans, and whether households used
loans for agricultural production or purchasing productive assets. Factors that reduce the
probability of an upward transition are female-headed households, highest education level

of household members, the presence of short-term loans, and whether households
experienced income shocks due to agricultural risks. To elevate the upward transition of
rural livelihood, policymakers should consider measures that aim at promoting and
enhancing education in rural area, strengthening land ownerships, encouraging investments

on farmland and farm diversification, building an effective monitoring system to ensure
that funds are being used in line with the objectives, and enhancing farm resilience.

KEYWORDS: Livelihood strategies, livelihood dynamics, livelihood transitions, rural
households, rural poverty, stochastic dominance analysis, principal

component analysis, cluster analysis, latent class analysis, panel data,
multinomial logistic regression, ordered logistic regression