Thailand Country Study 91
Chapter 6
THAILAND COUNTRY STUDY
National Context
I
n comparison with other Asian countries, Thailand is
a medium-sized country of about 62 million people,
with a gross national per capita income in 2001 of
nearly $2,000 ($6,550 in 1993 purchasing power parity
terms). Thailand achieved one of the highest economic
growth rates in the world during the period between 1975
and 1995. Broadly, Thailands development policy has re-
volved around an open door for trade and heavy invest-
ment in infrastructure to promote industrial development,
especially in labor-intensive industries. Thailand has
largely succeeded in meeting basic human needs and has
good social indicators: an average life expectancy of 69
and an adult illiteracy rate of only 5%. The economy
experienced a setback during the Asian financial crisis of
199798, but recovered fairly rapidly due to continuing
strong growth in exports.
Thailands long experience of sustained growth, good
communications, and labor force mobility has led to ris-
ing expectations and perceptions of increasing inequality
between the poor and the nonpoor. According to 1998
data, less than 0.5% of the population is living below the
extreme poverty line of $1 a day per person. However,
about 28% of the population is still poor by world stan-
dards, with incomes of less than $2 a day per person. The
Gini index is 41.4, showing that income inequality in Thai-
land is relatively high.

and poverty reduction in Thailand was carried out to pro-
vide a comparable framework to the studies conducted in
India and the PRC (Fan, Somchai, and Nuntaporn 2003).
The study focuses on rural poverty because of the concen-
tration of poverty in rural areas (20% in rural areas com-
pared to 6% in urban areas in 2000). Using regional-level
data over 20 years, it examines the impact of rural roads
and electricity expenditures on poverty reduction, as well
as the effects of irrigation, agricultural research and ex-
tension, and education expenditures. The model traces
the effects of public expenditures on poverty through their
effects on agricultural employment, nonagricultural em-
ployment, and food prices. The study showed that all of
these government investments had contributed to growth
in agricultural production and to the reduction of rural
poverty in Thailand.
Government spending on rural electricity had the larg-
est poverty reduction effect, as well as having a substantial
92 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
impact on growth in agricultural productivity. Among the
channels linking rural electricity to poverty reduction,
increase in nonfarm employment accounted for 75% of
the effect, and growth in agricultural productivity for only
20%.
17
Expenditures on agricultural research and exten-
sion had the second highest poverty reduction impact, fol-
lowed by expenditures on rural roads. Roads had little
effect on agricultural productivity, however; their poverty
reduction impacts came mainly from effects on nonfarm

improved infrastructure management, bet-
ter transport services, and greater involve-
ment of the private sector. In addition, it
encourages local participation in both
infrastructure construction and service pro-
vision. Lastly, it takes into account poten-
tial linkages with the infrastructure systems
of neighboring countries.
Roads. Several government agencies
are responsible for developing the national
road network, which covered more than
200,000 km in 1996. The Department of
Highways (DOH) is responsible for
interurban roads and highways, accounting
for almost half of the total network.
Rural roads are the responsibility of the
Accelerated Rural Development Depart-
ment, the Public Works Department, or the Royal Irriga-
tion Department, while urban streets and expressways are
managed by the Bangkok Metropolitan Administration
or the Expressway and Rapid Transit Authority, respec-
tively. Most of the DOH network is paved and regularly
maintained. These roads link the national capital to the
main centers of each province, and these centers in turn to
the (district) centers. Traffic on these roads is heavy, vary-
ing from less than 1,000 vehicles per day (vpd) on the
tertiary roads to more than 25,000 vpd on the most heavily
trafficked roads in the Central Region.
Few barriers constrain entry into the transport services
sector, and a wide variety of vehicles can be seen on the

ucts represented only a small fraction (1.7% and 1.2%,
respectively) of rail freight traffic.
The SRT operates at a net loss, mainly because it sub-
sidizes rates for third-class passenger service, which
accounts for 92% of all passengers. These rates have not
been increased since 1985, and they are about 50% lower
than the rates for intercity bus service. Nevertheless, the
railroad has been steadily losing passenger traffic, while
freight traffic is increasing. For this reason, the merits of
continuing to subsidize third-class passenger traffic as a
poverty reduction measure have been under discussion for
some time.
Energy Sector Policy
Electricity generation was originally the responsibil-
ity of the Electricity Generating Authority of Thailand
(EGAT). In the early 1990s, however, the Government
decided to allow private companies to invest in power gen-
eration plants. These are classified as small power pro-
ducers (SPPs) and independent power producers (IPPs).
Companies in both groups sell electricity to EGAT and
can also sell directly to the public. SPPs may produce up
to 150 megawatts but can sell only up to 90 megawatts to
EGAT. The total contribution of private producers to the
electricity supply system is still small, but is expected to
increase under the Governments privatization policy. If
this happens, lower costs and increased availability of elec-
tricity throughout the country are likely. Some SPPs use
renewable fuels such as bagasse (agricultural residues),
paddy husks, wood chips, sawdust, municipal waste, and
biogas. Although the present contribution of these projects

Providing electricity to households with no legal identifica-
tion has been a problem; people have connected illegally
to the lines serving their legally resident neighbors.
94 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
sites included two sites in the Northeast Region and one
in the Southern Region. In addition to being centers of
rural production, both regions are major destinations for
interregional transportation and are well served by both
road and rail systems. The Northeast Region (Map 6.1),
being the poorest, is also the one from which long-distance
migration for employment most frequently occurs. Migra-
tion is less important as a survival strategy in the Southern
Region , but the region relies heavily on transport to send its
primary products (e.g., rubber) to markets. The two urban
sites are slum settlements located in Nakhon Ratchasima
(provincial capital and major city of the Northeast Region),
and in Bangkok. These sites were chosen for reasons of con-
venience, as the Thai Development Research Institute
had
already conducted some research there and had built up good
relations with the communities concerned.
Northeast Region
Sample rural districts were selected on the basis of an
analysis of secondary data from a rural village database
maintained by the Thai Ministry of Interior. Village data
for 1990 and 1999 were analyzed to classify villages that
had experienced significant improvements in road trans-
port and electrification over that period. Significant
improvements were operationally defined as (i) a reduc-
tion of at least 50% in traveling time from the village to

capita incomes in this province are about twice those of
Buri Ram Province, which is a more typical area for the
Northeast Region. The 1999 per capita income in Nakhon
Ratchasima was about $940. Nakhon Ratchasima is home
to many prominent national politicians, which means that
the province is relatively better provided with publicly
supplied infrastructure than the national average. Overall
population density in Nakhon Ratchasima is rather low
(124 persons per km
2
in 1999), due to the presence of a
large national park in the province. The sample districts
selected in Nakhon Ratchasima are located on the far side
of this park, which means they are relatively distant from
the regions major road network.
Wung Kata and Klong Muang districts are relatively
poorer areas in Pak Chong County and Nakhon
Ratchasima Province. Wung Kata, in particular, is iso-
lated by its hilly terrain and its location on the far side of
Khao Yai National Park. Both districts suffer from prob-
lems of water availability and water quality. Agricultural
yields are higher in Klong Muang than in Wung Kata;
Klong Muang is slightly better connected to the road net-
work and has better road conditions in general. From the
county seat at Pak Chong, it takes about 1 hour on a tertiary
road to reach Wung Kata District. Most of the road is still
laterite, although some portions are paved with asphalt.
Because of its beautiful scenery, Wung Kata was the site of
much speculative land purchase during Thailands economic
bubble of the late 1980s and early 1990s.

cars. It takes 2 hours for people in these villages to reach
the county seat, and often much longer in the rainy season.
The villages in the second group are located closer to
the main road system. Most village roads that are not paved
are laterite rather than earthen. These villages benefit from
being located along the public transport routes that serve
the more remote communities, like the first group. Thus,
they have several options for daily travel outside the vil-
lages. These communities also have several stores selling
consumer products. Having good links to the national
road network makes it easy to obtain goods from major
markets, even by traveling to Bangkok.
The third village cluster, in Klong Muang District,
has been served by paved access roads for more than 10
years. However, one village (Nong Sai) has mainly earth
roads inside the village, while the other (Nong Sai Nea)
has concrete roads, as it is the site of an important temple.
Agricultural production patterns in all three groups are
similar, based on maize and cattle (including dairy pro-
duction) and some tapioca production.
Buri Ram Province is located farther toward the north-
east. It is more densely populated (147 persons/km
2
), more
96 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
agricultural, and less urbanized. Covering an area
approximately half that of Nakhon Ratchasima (includ-
ing the park), the value of Buri Rams provincial produc-
tion in 1999 was less than a third of that of its sister prov-
ince. Per capita income in Buri Ram Province in 1999 was

look for work; only children and elderly people
are left in the village. Most villagers have little
land (averaging 2 rais [0.16 ha] per family),
and droughts occur frequently. The villages are
located on laterite roads about 2 km away from
the nearest paved road.
The second pair of villages offers a con-
trast in road conditions, showing that roads
alone cannot always explain differences in
welfare. The road to one village, Pung Gu,
was recently paved. The other village, Sri
Takrong, is still 3 km from a paved road, but
the villagers in Sri Takrong appear economi-
cally better off because they carry on com-
mercial transactions with businesses in the
Prakomchai county seat. The last group of two villages
has good road access. One of them appears more affluent,
as it is located on a major intersection well served by pub-
lic transportation. However, the other village has not ben-
efited much from having good roads, possibly due to the
fact that, as in Pung Gu, most villagers do not own land.
Nakhon Ratchasima Buri Ram
Population Density 124.00 147.00
Km of Roads/Area (km
2
) 2.31 2.85
Km of Roads/Population 0.02 0.02
% Electrified Villages 98.30 99.10
% Electrified Households 96.60 97.10
Schools per 1,000 Population 0.60 0.60

The capital city of Nakhon Si Thammarat is located on
the coast. It is large and historically important, but is not
directly served by a trunk highway. Rather, the main high-
way passes through Thung Song County, another major
business center in the province. The sample districts in
Bang Chan County, which is not located on the coast, have
better access to the road network via Thung Song.
Villages in these two districts are primarily engaged in
rubber production. Rubber trees are the symbol of South-
ern Region agriculture, and have long been the major
source of economic prosperity in the South. Rubber price
supports also contribute to the economic welfare of the
regions people. Educational levels are high; the region is
known for its active participation in the political life of the
country. On average, household landholdings are signifi-
cantly larger than those in the Northeast Region. Although
the sample districts in the Southern Region are less well
served than the sample districts in the Northeast in terms
of physical infrastructure, they are still considerably bet-
ter off than those in the Northeast in terms of economic
productivity.
The two sample districts are about 90 km from Nakhon
Si Thammarat city center, and about 20 km from Toong
Song county seat, the provinces second most important
business center. The districts are reached by a tertiary high-
way from Thung Song. Compared to other districts in
Bang Chan County, they are relatively isolated. Many
98 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
households in these districts have no direct access to pub-
lic passenger transport. Consequently, almost all of them

its unusually low electrification rate.
In the urban sites, the transport intervention studied
was not so much road improvements as the availability and
quality of transport services, measured by access (walking)
times to pickup points for different transportation modes.
Slum dwellers in Bangkok could generally access motor-
cycles, minibuses, and buses by walking for less than 10
minutes, while for the Bailey community in the Northeast
the average was 12 minutes. Bangkok slum residents also
had access to boat service (10 minutes) and minivans (15
minutes). In contrast, for all slum residents, train service
was half an hour or more distant by walking. In Nakhon
Ratchasima, 77% of the slum residents interviewed had
no electricity connection. The reason for this low level of
connectivity is that the community is located along a rail-
way, and it is difficult and dangerous to lay electricity lines
across the rail line. In Bangkok, all slum dwellers had
access to electricity, although 30% used community meters
and 10% were connected through their neighbors.
Methodology
Definition of Poverty
The Thai country case study used three different defi-
nitions of poverty. The first definition is income-based or
objective poverty. The poverty classification used in the
study was calculated separately for the rural and urban
samples, based on the household data obtained in field
interviews. The median annual per capita income for the
rural household sample was close to 12,000 baht (B, about
$285), which is the same as the national official poverty
line for rural households in 2002. Households with per

nonpoor. It is interesting to observe that although many
more urban sample households (77) were in the near-poor
category than in the poor category (34), the great majority
of the nonpoor households (83 out of 98) had per capita
incomes more than two standard deviations above the
median (i.e., more than B20,380 or $485). This distribu-
tion illustrates the skewedness of income distribution in
Thailand, especially in urban areas.
The Thai study team was also interested in how
peoples perceptions of poverty affect their perceptions
about infrastructure improvements. For this reason, they
introduced the notion of subjective poverty, or poverty status
as reported by key informants (village and community
leaders). Using this method, relatively few of the rural
sample households were identified as poor (20%, as com-
pared to the 50% objectively poor). In urban areas, the
proportion subjectively classified as poor corresponded
more closely to the proportion of poor and near-poor. Strik-
ingly, about 40% of the sample households living in slum
settlements could be classified on the basis of income as
well-to-do,
18
but less than 10% were perceived by com-
munity leaders as being so. The team also measured rela-
tive poverty through self-reports, finding that the results
closely corresponded to the results using subjective
poverty. It shows that people perceive their own status and
are seen by their neighbors in relation to local rather than
national norms. Hence, in rural areas, especially poor
areas, objectively poor people may not be seen as poor,

by more than half in only one village (Pa Pai Dang). The
cause of the difference here seems to be not a change in the
length or type of road, but a striking increase in vehicle
ownership. In Buri Ram, three of six communities experi-
enced significant changes in travel times, and this seems
to be at least partly due to improvements in road quality,
including paving. Three of seven communities in Nakhon
Si Thammarat saw significant changes in travel times, and
this also appears to be attributable to partial paving of
access roads. Vehicle ownership increased dramatically in
all communities over the past 10 years.
With respect to rural electricity, the measure of change
was the percentage of households within each village con-
19
This analysis is based on information from the Nrd2c database for 1990
and 1999. The study team also evaluated this information for changes
between 1992 and 2001.
18
Households were classified as well-to-do if they had incomes more
than two standard deviations above the sample median.
100 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
nected to electricity in 1990 and 1999. According to the
village level data, two villages in Nakhon Ratchasima had
no electricity at either time, and one that had no electricity
in 1990 was 100% electrified by 1999. The other three
sample villages from this province were approximately
50% electrified in 1990 and somewhat more so (ranging
from 67% to 80%) in 1999. In Buri Ram, two of six sample
communities had no electricity in 1990, but were 100%
electrified in 1999. The other four communities had elec-

Research Methods
The study aimed to adopt a double-difference approach
(before-and-after, with-and-without) at both the village
and the household level. Thus, it sought to compare wel-
fare changes over time between villages and households
with and without transport interventions, with and with-
out electricity, and with both types of changes, with the
objective of determining if impacts were significantly dif-
ferent between the poor and the nonpoor. The Thai study
team was particularly interested in letting respondents them-
selves explain how they perceived such effects. Conse-
quently, they built the main part of the study around house-
hold interviews, complemented by village-level informa-
tion and key informant interviews, limited participatory
focus groups, and supplemental secondary data analysis.
The household survey covered 913 rural households
and 209 urban households. The rural sample was
designed to include approximately 300 households
each from the selected sites in Nakhon Ratchasima,
Buri Ram, and Nakhon Si Thammarat. The urban
sample was designed to include approximately 100
households each from two urban settlements. As
described above, villages in rural areas were strati-
fied into three groups based on the quality of their
road access. A list of households in each community
was established in consultation with local authori-
ties. This list was further stratified according to sub-
jective socioeconomic status as reported by the
authorities, and households were then randomly selected
from the lists until the desired sample size was reached.

and also solicited views on the distribution of those
impacts within the community. At the end, the question-
naire asked for the respondents opinion about develop-
ment in general and about the need for more investment
in transport and energy infrastructure. Questions about
positive and negative impacts were asked separately, and
respondents were then asked to evaluate net impacts.
The questionnaire was administered in an open-ended
fashion, by inviting respondents to identify impacts and
the mechanisms through which these impacts took place,
rather than by providing them with a checklist. In addi-
tion to the household surveys, the team conducted inter-
views with local officials to obtain village-level informa-
tion. It also conducted two focus group discussions to vali-
date information provided in the interviews. The focus
group in Nakhon Ratchasima involved six women, drawn
from the womens group and the first aid volunteer group
in two adjacent sample communities. In Nakhon Si
Thammarat, it involved six employees of one district
office, five men and one woman.
Sample Community and
Household Characteristics
The rural sample communities in Nakhon Ratchasima
ranged in size from 50 to 500 households, or 2001,650
people. Most were farm households, although many house-
holds have multiple sources of income. About three fourths
of all households owned their own land, and about 10%
were renters. Some both rent and own land. Almost all
grew maize and/or sweet corn, while about 15% on aver-
age also grew commercial crops like cassava and sugar

(53%) of the sample households in Nakhon Si Thammarat
were poor (33% ultra-poor), while 47% were classified as
nonpoor. Thus, among the three rural sites, Buri Ram was
the poorest, Nakhon Si Thammarat occupied a middle
position, and Nakhon Ratchasima had the lowest inci-
dence of poverty in the study sample.
The rural survey sample was selected in such a way
that approximately equal numbers of households lived in
villages with poor road conditions, moderate road condi-
tions, and good road conditions. This stratification was
applied in each province, so there was little variation in
this distribution across provinces in the study sample.
However, the household questionnaire also looked at the
quality of immediate road access enjoyed by each sample
household; 63% of the households were served by laterite
roads, 20% by paved roads, 8% by concrete roads, and
10% by earth roads or tracks. Thus, most of the rural
sample had immediate access to motorable roads.
Residents help the Thailand study team to map some
of the features of their village.
102 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
For electricity, the household survey examined the
method of connection and the length of time that a house-
hold had been connected. Only 33 of the rural sample
households (4%) had no electricity; 84% of the sample had
a direct connection, and 12% were connected through their
neighbors. These proportions did not vary significantly
across the three provinces. About 23% of the sample had
had electricity for more than 10 years, 33% were con-
nected 510 years ago, and 20% became connected within

the following:
 Number of roads to district offices in 1992 and 2001,
and change in this number between 1992 and 2001;
 Length of paved roads to district offices in 1992 and
2001, and change;
 Length of laterite roads to district offices in 1992 and
2001, and change;
 Average travel time to district offices in 1984, 1992,
and 2001, and changes in 19841992, 19922001, and
19842001;
 Percentage of households in the village with electricity
in 1992 and 2001, and change;
 Years since a household gained immediate road access;
 Years that a household has had electricity; and
 Annual amount paid by a household for electricity.
The first five variables were taken from the Nrd2c
database for villages and attributed to the sample house-
holds, while the last three were taken directly from the house-
hold surveys. Village dummy variables were also introduced
into the analysis to account for other situational factors that
might have influenced changes in income, expenditure, or
education. Ordinary least square regressions with stepwise
selection were run for the entire rural sample and for poor
households separately. The regressions do not have a very
good fit (values of R
2
on the order of 0.1-0.3), as is com-
mon in cross-sectional regressions using household data.
Only one of the regressions yielded significant results
(p<0.05) with respect to household income, both for the

earth to laterite roads, since none of the intervention
variables concerning laterite roads is significantly
related to incomes, either for all households or for poor
households.
For all households, the positive relationship between
electricity bills and household income could mean either
that higher electricity use enhanced incomes, or that higher
income permitted more electricity use. However, the
degree of electricity penetration in 2001 was negatively
correlated with the income of poor households. This was
not the expected outcome, since it was hypothesized that
the availability of electricity should open up more income-
earning opportunities for the poor. This result may reflect
an incipient inequality problem within the more
electrified rural communities. In fact, poor households in
these more modern villages were even poorer than the
Coefficients Standard Probability
Errors
Constant 8.422 0.073 0.000
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) NS
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) 0.035 0.003 0.000
Increase in Length of Paved Roads NS
Length of Laterite Roads to District (1992) NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS

work needs to be done to determine whether this phenom-
enon was unique to the study sample.
When household expenditures were used as the depen-
dent variable, more intervention variables became sig-
nificant (Tables 6.6 and 6.7). The length of paved roads to
the district remained the most significant determinant of
household expenditures for all households. The change in
length of paved roads was significant for all households
and also for poor households. Interestingly, the length of
laterite roads to the district office in 1992 also had a sig-
nificant effect on household expenditures for all house-
holds (but not for poor households) in 2001. This may
reflect the effects of prior improvements from earth to
laterite roads, which stimulated growth in commerce and
farmer involvement in the cash economy. Recent
reductions in average travel time to the district center were
Coefficients Standard Probability
Errors
Constant 8.422 0.073 0.000
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) NS
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) 0.035 0.003 0.000
Increase in Length of Paved Roads NS
Length of Laterite Roads to District (1992) NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS

Increasing the percentage of households with access to
electricity had the effect of inducing higher spending by
both poor and nonpoor households. Since it did not have a
similar effect on incomes for either group, this finding sug-
gests that such spending was related to consumption rather
than productive investment. In fact, the household inter-
views and focus group discussions showed that households
tended to imitate others consumption patterns when it
came to electric goods. For example, it was common for
families to want to own a television set when their neigh-
bors owned one. Higher expenditures for all households
were also correlated with the length of time that a house-
hold had been electrified. Again, village dummies pro-
duced significant results.
With respect to education, both the number and
increasing length of paved roads linking the village to the
Coefficients Standard Probability
Errors
Constant 8.732 0.128 0.000
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) NS
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) 0.035 0.003 0.000
Increase in Length of Paved Roads NS
Length of Laterite Roads to District (1992) 0.009 0.005 NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS

households, the number of roads was significant, even
though the length of paved roads was not. This result may
be explained by the fact that poor households were not
usually located near village centers, and thus may have
benefited from having more alternative routes to places
outside the village. A lower average travel time to the dis-
trict center in 1992 also predicted higher average years of
education per household in 2001, for all households but
not for poor households. This parameter may reflect the
opportunity to access higher education, which may only
be available in the district centers.
Statistically significant relationships with educational
levels existed for the increase in the share of households
electrified, the number of years that a household had been
Coefficients Standard Probability
Errors
Constant 8.814 0.087 0.000
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) NS
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) NS
Increase in Length of Paved Roads 0.044 0.006 0.000
Length of Laterite Roads to District (1992) NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS
Average Travel Time to District (1992) 0.010 0.001 0.000
Average Travel Time to District (2001) NS

electricity. Given the respective time frames, it seems likely
that more education encouraged greater use of electricity
by the poor, rather than the other way around.
The study team also ran transport and energy inter-
vention variables, along with other household-level vari-
ables, against satisfaction scores given by respondents on
changes that had occurred over the past 10 years in family
income, family well-being, family convenience, and fam-
ily happiness, as well as in the village economy and society
(Table 6.10). The main finding was that households with
more assets were more likely to report positive changes
Coefficients Standard Probability
Errors
Constant 3.504 0.354 0.0
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) 0.438 0.150 0.000
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) NS
Increase in Length of Paved Roads 0.065 0.016 0.000
Length of Laterite Roads to District (1992) NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS
Average Travel Time to District (1992) 0.005 0.002 0.000
Average Travel Time to District (2001) NS
Change in Travel Time (19841992) NS
Change in Travel Time (19922001) NS
Change in Travel Time (19842001) NS

district office were associated with a perception of greater
family happiness. A greater length of laterite road in 1992
was associated with positive changes in family well-being,
and a greater length of paved road in 1992 with greater
family convenience. The current length of paved roads is
correlated with perceptions of positive changes in the vil-
lage economy and society.
Other factors possibly influencing peoples perceptions
of change were their occupation, their status as natives of
Coefficients Standard Probability
Errors
Constant 4.097 0.267 0.000
Transport Variables
Number of Roads to District (1992) NS
Number of Roads to District (2001) 0.438 0.150 NS
Increase in Number of Roads NS
Length of Paved Roads to District (1992) NS
Length of Paved Roads to District (2001) NS
Increase in Length of Paved Roads 0.065 0.016 0.000
Length of Laterite Roads to District (1992) NS
Length of Laterite Roads to District (2001) NS
Difference in Length of Laterite Roads NS
Average Travel Time to District (1984) NS
Average Travel Time to District (1992) 0.005 0.002 0.040
Average Travel Time to District (2001) NS
Change in Travel Time (19841992) NS
Change in Travel Time (19922001)
Change in Travel Time (19842001) NS
Years of Household Immediate Road Access NS
Energy Variables

Given the policy-oriented focus of the study, the Thai-
land study team set out to determine if the poor had differ-
ent views about the impacts of transport and energy changes
than the public at large. There were three possible out-
comes: (i) the poor benefit more from transport and en-
ergy changes than the public at large, (ii) the poor benefit
equally with the public at large, and (iii) the poor do not
benefit as much as the public at large, and may even be
negatively affected by such investments. These outcomes
corresponded to positions taken by different stakeholders
in national debates over the merits of additional infra-
structure investment. The aim of the study was to inform
this debate by providing data from the point of view of the
poor themselves.
The study examined perceived impacts on occupations,
household income and expenditure, the availability of
Family Family Family Family Village Village
Income Well-Being Conven- Happiness Economy Society
-ience
Asset Ownership
Stereos + +
Bicycles + +
Refrigerators + +
Gas Stoves + + +
Televisions + + + +
Telephones + + + +
Mechanical Plows + +
Radio/Cassette Player +
Transport
Travel Time to District +

mobility. Table 6.11 shows the percentage of poor and
nonpoor households reporting net impacts. Statistical tests
using logistic log-linear models shows that, for most
impacts, the differences between poor and nonpoor
respondents were not statistically significant. Where their
views differed, it was sometimes not in the ways that would
be expected. A similar result was found for electricity. This
would tend to confirm the view that infrastructure, as a
public good, benefits all people more or less equally.
Most households reported that rural road improve-
ments had no significant impact on occupational choice
(but see Box 6.1). This finding was significantly stronger
Impact Result Percent of Respondents
All Households Nonpoor Poor
(n=913) (n=441) (n=454)
Occupational Change No Impact** 87.6 82.1 93.0
Household Income Increase Income** 50.7 55.7 45.4
No Impact** 44.2 41.4 47.3
Decrease Income 5.0 2.9 7.3
Household Increase Expenditure 81.9 80.5 83.1
Expenditure
Goods Availability More Goods Available 96.5 96.8 96.5
Satisfaction With Goods Satisfied 92.7 94.1 91.8
Household Debts No Impact** 90.0 87.8 92.7
Household Education Improves Education 91.9 90.5 93.1
Household Health Improves Health** 90.3 93.4 87.0
Free Time Availability More Free Time* 67.3 69.4 65.3
No Impact** 25.1 20.9 29.1
Safety Increase Safety 58.6 56.9 59.9
Decrease Safety 25.0 28.6 22.0

for sales or employment that would supplement the activ-
ity that they regard as their primary occupation.
Among those who felt that road improvements had
reduced their household incomes, the main reasons were
the general economic slowdown due to the Asian finan-
cial crisis, fewer jobs available, lower product prices, and
lower sales. This suggests that, especially among the
ultra-poor, a small minoritys livelihood strategies cannot
stand up to the competition introduced by road improve-
ments. Interestingly, one nonpoor respondent cited higher
wages paid as a negative consequence of road improve-
ments, while four respondents (two ultra-poor and two
Only about half of all households thought that rural
road improvements had increased their household income.
Poor households were significantly less likely to think so
than nonpoor households. Most of the rest of the respon-
dents felt that road improvements had had no impact on
their incomes. However, about 5% of all households,
including 7% of the poor and close to 10% of the ultra-
poor, felt that road improvements had actually decreased
their incomes. Respondents gave many reasons why roads
might increase incomes. The most frequently cited were
an increase in job opportunities both inside and outside
the village, higher sales of local products, and overall eco-
nomic improvement. Lower transport costs, higher prod-
uct prices, and more farm-gate sales were also mentioned.
When this response was combined with the response about
nonpoor) cited an oversupply of labor (migrants from even
poorer regions), suggesting that road improvements also
introduced greater competition in the local labor market.

for the next generation is even brighter. He foresees that his children will not work in the village anymore, but will seek work further afield.
Source: Thailand study team.
112 Assessing the Impact of Transport and Energy Infrastructure on Poverty Reduction
resources, increased educational expenses, and the need
to pay for road maintenance.
A small minority (about 3% of the sample), predomi-
nantly nonpoor, felt that road improvements decreased
household expenditure. In these cases, the reasons cited
included lower passenger and goods transport costs,
decreased need to travel to buy goods, lower product prices,
lower expenditures on gasoline, and fewer people at home
because of migration to find jobs elsewhere.
Respondents overwhelmingly confirmed that more
goods were available in local markets as a result of trans-
port improvements, and that this was a positive impact for
both the poor and nonpoor. The primary reason for satis-
faction with this result was the reduced risk of shortages, a
serious problem for all (but especially for the poor) in
remote rural areas. Respondents were also happy to be
able to choose from a wider selection of goods. Other
reasons mentioned included more shops, cheaper goods,
more good-quality food available, and greater convenience
(including savings in transport time). The very small
minority (less than 2% of the sample) that thought having
more goods locally available was not a good thing cited
the greater availability of expensive goods and the conse-
quent temptation to overspend.
The study team particularly wanted to examine the
relationship between rural infrastructure improvements
and household debts. According to one popular view in

provements as having a positive impact on family health.
In this case, however, the poor were significantly less likely
than the nonpoor to report such positive impacts. The main
reason given was more convenient traveling to health care
centers, followed by prompt access to health care, which
may reflect the greater ability of health care providers to
reach their clients in their villages or homes. Interest-
ingly, quite a number of respondents (67, or 8% of the
total) mentioned reduction in dusta result of road pav-
ingas a significant source of positive health impacts. A
few respondents also mentioned the effects of increased
income on health. About 3% of the sample identified nega-
tive impacts, mainly in connection with the dust generated
on laterite roads. A few respondents also mentioned
vehicular air and noise pollution.
Views about road impacts on the availability of free
time were rather mixed, although little variation between
the views of the poor and the nonpoor emerged. About
two thirds of the sample felt that road improvements
resulted in more free time, while about one fourth felt that
there was no net impact, and the remainder saw a net nega-
tive impact. The main reasons for more free time were
20
This type of consumer debt may be related to the purchase of
television sets, which are widely available in rural Asia. From
another perspective, such expenditure could be regarded as an
investment in information (see results for electricity on p. 118).
For survey respondents, road improvements mean more
convenient
i.e., fastertravel to health care centers.

Respondents in all income classes strongly agreed
that rural road improvements increased access to infor-
mation. Those who did not share this view mainly felt
that roads had no impact in this area. The main mecha-
nism of impact transmission in this case was overwhelm-
ingly believed to be the greater ease of personal travel
outside the community. Only a very small number of
respondents mentioned the greater availability of news-
papers, postal services, or telecommunications, or the ar-
rival of newcomers bringing new information into the
community.
With respect to access to common resources, about
two thirds of the respondents perceived a positive impact,
with a significantly more positive response among the poor
than among the nonpoor. Most of the remaining respon-
dents saw no net impact on access to common resources.
The reason given by almost all the respondents was greater
road access to common resource areas. Only 3% of the
sample saw a net negative impact, and this was mainly
attributed to outsiders getting greater access to common
resources that were felt to belong to the village. It may be
noteworthy that negative road impacts on common
resources were mostly perceived by nonpoor respondents.
Respondents agreed overwhelmingly (over 90% in all
cases) that rural road improvements led to better social
relations, both within rural villages and between those vil-
lages and the outside world (bonding and bridging
social capital). About 7% perceived no impact, and only
about 2% thought roads had a negative effect on social capi-
tal. The main reason given for bonding effects was the

hold income.
21
Only a few respondents cited the possibil-
ity of starting a home-based business and receiving higher
prices for local products or higher wages. Others men-
tioned the possibility of working for longer hours. Elec-
tric water pumps were also cited as time-saving devices that
released household energies for additional production.
The poor and the nonpoor shared these views. How-
ever, the nonpoor were distinctly more likely to mention
increasing jobs within the village as mechanisms for
income improvement, and distinctly less likely to men-
tion increasing jobs outside the village. This finding sug-
gests that the nonpoor were more likely to be able to invest
and capture the benefits of electricity through local busi-
nesses, while the poor depended on other investors to cre-
ate electricity-related job opportunities. Those who said
that electricity had a negative impact on their income were
predominantly from the ultra-poor group. They attrib-
uted this effect to a general economic slowdown and fewer
jobs being available (Box 6.2).
In contrast to the limited impacts on income, a higher
share of households perceived that electricity increased
their household expenditure. This was especially true for
the upper range of the poor and for the nonpoor, but less
so for the ultra-poor. The main reason, not surprisingly,
was higher electricity bills. Ultra-poor households cited
shis reason slightly less frequently, presumably because
they were less likely to have an electricity connection.
22

Bridging Social Capital Better Relations 78.9 78.0 79.3
Outside Village
* Significant difference between poor and nonpoor households at p<0.05; **Significant difference at p<0.01.
Note: All Households includes results from 18 unclassified households.
Source: Thailand study team field survey.
Table 6.12. Perceived Impacts of Rural Electricity Improvements
Thailand Country Study 115
appliances. A smaller group felt that electricity caused
goods to become more expensive. Only a few individuals
felt that electricity lowered costs. Half of them mentioned
the cost of lighting, and the others did not specify the costs
concerned.
As with road improvements, survey respondents gen-
erally rejected the suggestion that electricity improvements
had anything to do with their debts. However, a minority
(about 14%) did feel that electricity had increased their
debts. This effect was more often felt among the nonpoor
than among the poor. This finding suggests that the poor
(and especially the ultra-poor) were more careful about
incurring consumer debt than are the nonpoor, who can
afford to take greater risks and spend a larger share of
their income on nonessentials.
Respondents clearly felt that electricity improvements
had a positive effect on household education. The main
reason, cited by 72% of respondents, was the availability
of light for doing homework in the evening. While no
significant difference emerged on this point between the
poor and the nonpoor, the impact on the ultra-poor was
slightly less marked than the impact on the nonpoor and
the less poor. Most other respondents felt that electricity

cation and health generally seemed to confirm the
hypothesis that positive effects were associated with
access to electricity for both the poor and the nonpoor. The
fact that the ultra-poor seemed to benefit somewhat less
may result from the fact that fewer of the ultra-poor house-
holds in the rural survey sample had access to electricity.
About two thirds of all respondents said that electric-
ity provided them with more free time. Poor households
were somewhat, but not significantly, less likely to per-
Box 6.2. It is Easier to Earn Income Now
Uncle Tong Luan, 50, a villager in Klong Muang tambon, Pak Chong District, Nakhon Ratchasima Province, is now herding milk
cows, a career he started about 10 years ago when electricity first came to the village and brought benefits to the villagers. The privately
owned milk center, which uses electricity to extract milk from the cows, was established at the village entrance. Villagers like Tong Luan
started raising milk cows and now earn a higher, more stable income. Moreover, raising cows is not hard work, so even older people like
Tong Luan can earn a decent income.
Before, he was a farmer, just like everyone else. But Klong Muang also lacked roads, so farming was not very lucrative: much
time was consumed in transporting products, and if transport was lacking, the villagers had to sell their products at the same time,
the time when outsiders came to buy the products. That led to depressed prices and less income. So roads helped to raise selling
prices.
When the roads came, the villagers had more options for selling their crops and selling prices rose. Realizing the benefits they
had gained, the villagers always worked together to ensure that road maintenance is carried out. However, Tong Luan feels that the
government should improve the condition of the roads, since earthen and laterite roads cannot withstand heavy rains.
Roads also interact with electricity. Tong Luan notes that better roads make transporting milk easier, especially for him, because
he lives far from the village center. He was one of the last villagers to get electricity because he lives so far away; Tong Luan believes
that the better-off families in the village have benefited more from electricity, since they were electrified earlier.
Source: Thailand study team.