how to frequently and accurately …documents.worldbank.org/curated/en/808821573630520498/... ·...
TRANSCRIPT
How to frequently and accurately measure poverty and forest dependence?
Emilie Perge,1 and Raisa Behal
September 2019
Abstract
The relationship between forest dependence and welfare remains partially explored, partially due to a lack of data. Data collection of household consumption and poverty correlates has been constrained by time consuming and costly tools, such as multipage household and community surveys. Forest-SWIFT is a complementary tool to a traditional household survey, developed to simultaneously measure poverty as well as forest dependence, using a 15-question country-specific mini-survey. Forest-SWIFT was piloted in Turkey, where the forest-dwelling population is also the poorest. The tool used recent data from the Household Budget Survey 2013 as well as the Socio-Economic Household Survey 2016 tracking poverty and forest-dependence across 100 forest villages in Turkey in 2017. Forest-SWIFT estimated poverty at 23.2 percent in rural forest villages, and forest dependence as 15 percent, the latter echoing findings from previous literature. Forest-SWIFT is efficient to bring more data on the relationship between poverty and forest activities and to monitor how this relationship evolves with the goal to have a tangible effect on policymaking.
1. Introduction
Poverty reduction features amongst the goals of the largest multilateral and bilateral development
institutions such as the World Bank, the UN, the IMF, even dominating the list of the Sustainable
Development Goals (SDGs). Besides identifying clear goals to reduce poverty, it is equally vital to
measure it. Tracking changes in poverty incidence can help us measure impact of programs,
improve activities and inform policy. A fifth of countries receiving assistance from the World Bank
do not have a single poverty estimate and another fifth only have one, decreasing the likelihood to
measure improvements (OPHI, 2014; Serajuddin et al., 2015). i Forest villages, which are most often
remote and difficult to reach, are easily marginalized from national development programs,
excluded from surveys, and underrepresented in samples despite the high incidence of poverty
(Chomitz 2007). Large-scale, nationally representative, household surveys such as the Living
Standards Measurement Surveys (LSMS) conducted by national statistics offices to assess poverty
and livelihoods, usually under-sample households in remote, low-density areas. As a result, there is
1 World Bank. Corresponding author [email protected]. The team is very grateful to the participants of the FLARE conference in Stockholm in October 2017 and to the participants of a reference group for their comments and suggestions. The team gives its thanks to Craig Meisner for giving us access to the data and to Facundo Cuevas for his help with HBS data. The team acknowledges the support from Shiva Shivakumaranand and Shinya Takamatsu with the models. The team is grateful for comments from Ximena Del Carpio on an earlier version of this work. This work was made possible thanks to a PROFOR grant (TF0A2918). All errors remain ours solely.
1
little quantitative information to guide the design and implementation of national policies for
poverty reduction among rural forest-villages (Oksanen and Mersman, 2005). Although poverty
within forests has often been assessed using case studies, such indicators are often not comparable
to national measures and satellite images (Sunderlin et al., 2005; Sunderlin et al., 2007), or
representative of forest households. Recent studies (Vedeld et al., 2007; Angelsen et al., 2014), have
gained valuable insights about the pervasive role of forests in rural poverty, strengthening the case
for improving measurements of forest income and gathering more granular data on forest use and
dependence. More data can help monitor the implications of natural resource degradation on
welfare and help design effective interventions and strategies to improve development and
conservation (Angelsen et al., 2014). In addition, information on forest activities is often missing
from traditional household surveys used by national statistics office to measure poverty, while
household incomes can be hard to measure from the Forestry Modules created by the Food and
Agricultural Organization (FAO), the Center on International Forestry Research (CIFOR), the
International Forestry Resources and Institutions Research Network (IFRI), and the World Bank. As
explained below, these modules allow governments, practitioners, and researchers to robustly
assess the livelihoods of forest households (forest income, consumption of forest products), forest
tenure, and access to forest resources (Bakkegaard et al., 2017; FAO et al., 2016).
The Poverty Environment Network (PEN) at the Center of International Forestry Research (CIFOR)
collected the most comprehensive dataset to date, composed of household and community-level
observations collected from 58 sites across 24 developing countries from January 2005 to May
2010 (Wunder et al., 2014). The team found that despite variation across geographic regions, the
average share of forest income, coming from collection of forest products, forest wages, payments
from forest ecosystem services, was 22.2 percent for the entire sample, at times even greater than
agricultural income. Moreover, poorer households had higher relative levels of forest income (i.e.
share of total income), and non-poor households had higher absolute levels (Angelsen et al., 2014).
This paper presents findings from Forest-SWIFT (Survey of Well-being via Instant and Frequent
Tracking), a new data collection method to measure both poverty and forest dependence in a timely
and more frequent manner, which was piloted in Turkey in 2017. Forest-SWIFT is an extension of
an existing poverty estimation tool SWIFTii (OPHI, 2014), which uses baseline household data on
welfare to develop a short country-specific household survey to estimate current poverty incidence
as per the national poverty definition. These two tools complement traditional multitopic
household surveys to track poverty and welfare between two rounds of more comprehensive
2
household surveys. Forest-SWIFT additionally allows researchers to estimate forest dependence,
which is the share of income from forest-related activities out of households’ total income
(Angelsen et al., 2014), requiring data on value gained from forests and on total welfare (income or
consumption).
Both data on consumption and forest income were available for Turkey’s forest areas, which made
it the ideal location for a pilot. Turkey has 22.34 million hectares of forest area, which is
approximately 30 percent of the country’s land cover. The majority of the state-owned forests are
productive resources, supporting large scale processing and export of non-timber forest products
(NTFPs). According to official statistics, poverty in Turkey has been declining over the past decade.
Extreme poverty, constructed using a $1.90 per day in 2011 PPP poverty line, fell from 1.7 percent
in 2002 to 0.3 percent in 2013 and to 0.2 percent in 2016. In 2013rural extreme poverty remained
higher than the national average at 0.8 percent. Moderate poverty, using a $7 per day in 2011 PPP
poverty line, was also higher among rural populations at 34.9 percent compared to the country
average of 19 percent in 2013.iii
Using the data from the Forestry modules in Turkey and nationally representative data on poverty,
we developed models to measure both consumption and forest income with a set of only 20
questions and administered the survey across a sample of 1000 households, representative of all
households living in forest villages. The poverty model was built using data from rural households
in the 2013 Household Budget Survey (HBS) and a consumption-based poverty line of $7 a day (in
2011 PPP). Poverty incidence among the rural HBS sample was estimated at 34.9 percent. The
forest-dependence model was constructed using the 2016 Socio Economic Household Survey
(SEHS) of Turkey’s Forest Villages in which forest income is defined as the sum of values of forest
extraction, wages from forest employment, and payments for forest services. Forest-dependence
was 28 percent among the lowest income quintile, and only 8 percent among the highest quintile in
2016.
The poverty and forest models selected 14 and 10 variables respectively to confidently predict
consumption and forest income. When applied to the pilot data, poverty incidence in forest villages
is 23.2 percent with a $7 a day (in PPP of 2011) and forest dependence 15 percent. The Forest-
SWIFT pilot provided what we believe to be strong estimates of poverty and forest-dependence in a
timely and cost-effective manner. Although data more recent than 2014 were not available for
comparison, the reduction in poverty observed through our analysis is plausible. Turkey reduced
poverty headcount ratio from 11.5 percent in 2015 to 9.9 percent in 2016 using a $5.50 a day
3
poverty line in 2011 PPP.iv This pilot sheds light on areas for further improvement of the tool and
for recommendations for the design of baseline questionnaire.
The following section describes the Forest-SWIFT methodology. Section three provides the final
consumption and forest-income models. Section four highlights main results and forest household
characteristics, while section five summarizes the project and proposes avenues for further
improvements.
2. Forest-SWIFT Methodology
Forest-SWIFT is a data collection method developed to provide timely, quick, and accurate data on
poverty and forest dependence through a small set of country-specific questions. As an extension of
the Survey of Wellbeing via Instant and Frequent Tracking (SWIFT), a methodology developed by
the World Bank to estimate poverty incidence between consecutive comprehensive household
surveys (Ahmed et al., 2014; Yoshida et al., 2015), Forest-SWIFT additionally tracks forest
dependence, defined as the forest income share of permanent household income. These two survey
tools complement traditional household surveys which are collected on average every five years by
providing more frequent poverty measurements to monitor poverty changes. Although the model
can use either consumption or income as the welfare measure, the country preference takes
precedence to maintain comparability. Forest-SWIFT develops country-specific models for each
indicator, which often requires different variables per model.
Each model assumes a linear relationship between household total consumption/income, or forest
income ( yh ) and their correlates (xh ) with a projection error (uh). The inclusion of this error term
differentiates this model from other predictive tools
ln yh=xh ' β+uh (1)
Forest-SWIFT estimates the log transformation of the dependent variable to smooth asymmetries
and normalize the distribution of the variable, making it easier to estimate. Forest-SWIFT controls
for issues linked to over-fitting – when a model performs well within the sample but poorly outside
the dataset – by cross-validating the model (Kuhn and Johnson, 2013). The purpose of cross-
validation in Forest-SWIFT is to identify the optimal level of significance- or p-value- in the model,
which would balance the number of determinants and the goodness of fit across the sample. Cross-
validation consists of two steps: (a) splitting the sample in n-folds and running the model in n-1
folds and testing it on the nth fold v and (b) running multiple models per fold, testing various
thresholds of significance for model variables. This process of ‘stepwise’ selection entails adding
4
variables to the Ordinary Least Square (OLS) model sequentially if they bring enough information,
and simultaneously removing them if they do not. Each fold has a chance to be the testing data and
this process is repeated n times by changing the nth fold each time. The optimal p-value performs
best in terms of mean-squared errorsvi between actual and projected welfare, and the absolute
value of the difference between the actual and projected poverty (or forest-dependence) rates. This
concludes the cross-validation process and the stepwise OLS regression is run a final time on the
full sample of data using the selected p-value. The resulting regression is the SWIFT poverty model.
To ensure the quality and robustness of the models, Forest-SWIFT carries out two tests (if data are
available): backward imputation and validity test. The former applies the final model to a previous
round of data to check the stability of the model over time. The latter tests whether the error term
follows a normal distribution using a simulation method developed by Elbers, Lanjouw and
Lanjouw (2002, 2003).
The result is a small set of questions that considerably simplifies data collection and encourages
teams to collect data quicker and more frequently than traditional lengthy household surveys.
Forest-SWIFT data are collected through CAPI (Computer Assisted Personal Interviews) to ensure
quick analysis and results. Having similar questionnaire designs for baseline and Forest-SWIFT
surveys is essential. Differences in questionnaire design, recall periods, and labeling of questions,
can bias estimates and require re-estimations of models with fewer variables that can be selected
since there are fewer variable options once the survey is rolled out.
After creation of the questionnaire, the last phase of Forest-SWIFT is to predict consumption and
forest income based on the coefficients from the models. In this final prediction phase, Forest-
SWIFT utilizes multiple imputation estimations to apply the coefficients from the respective models
to the variables in the new dataset. Random error is simultaneously introduced by adding 1000
imputations with error per household estimate.
A household is identified as poor if its consumption is below a poverty line. As we have estimated
1000 imputations per household, we had 1000 estimates of consumption, and therefore poverty
status per household. To compare poor and non-poor households, we used a multiple imputation
(mi) command that split the sample in two and provide sample means for the two different
samples.
5
3. Constructing the models for Turkey
Turkish forests cover 22.34 million hectares, close to 30 percent of Turkey’s total land cover, and
are spread across the coastal mountain ranges along the Black, Marmara and Mediterranean seas.
More than half of Turkish forests are primarily coniferous with pine trees, while 60 percent of the
remainder is oak. About 40 percent of Turkish forests are productive forests: on average,
productive forests provide 7 million m3 of timber logs and 8 million m3 of fuelwood (Sirtioglu,
2010). In 2016, about 7 million people in Turkey lived in or near forests, depending on them for
their livelihoods (Sirtioglu, 2010; PROFOR, 2017).
Baseline data
Turkey recently collected forest income data in the 2016 Forest Socio-Economic Household Survey
(SEHS), which made the country a likely candidate to pilot Forest SWIFT, especially since
consumption data were also available in the 2013 Household Budget Survey (HBS). The 2016 SEHS
collected data on forest participation, income, and products from a of 2, 037 households, which was
a representative sample of forest dwelling households. The 2013 HBS, on the other hand, is
nationally representative sample of 10,058 householdsvii and collected information about
household composition, consumption, assets, employment categories, and education. Although the
data includes urban-rural classification, regions remain unidentified.viii
In the 2016 SEHS, forest income is measured as the total of monetary values of forest product
extractions, wages from forest employment, and payments from forest services. Although the
survey had questions on a wide range of product, forest income analysis is limited to products
collected by at least 10 households, which left 16 forest products: firewood, mushroom, herbs,
thyme, sage, hazelnut, linden, sting nettle, walnut, rosehip, pinecone, chestnut, industrial wood,
blackberry, trefoil, opium. Forest payments only included 15 observations. On average, households
diversify their livelihoods with 60% of households having three or more sources of revenue
(PROFOR, 2017). At the same time, 15 percent of households specialize, in agriculture or livestock
raising, 10 percent solely in forest-related activities, while 11 percent of households combine
forest-related activities and agriculture or livestock raising (PROFOR, 2017). Researchers in this
report find that using a relative poverty line equal to 60 percent of the median income in the
villages (TL 480 per capita per month in 2016 about $12 per capita per day in 2011PPP), 40
percent of their sample has households’ total income per capita below this line (PROFOR, 2017).
Comparing the average values for rural households in the 2013 HBS and the whole sample in the
2016 SEHS, these values differ for some characteristics. First, forest households tend to be larger
6
than rural households, due to a larger share of prime age adults (i.e. fewer dependents such as
children and elders). Second, forest household heads are mostly men and better educated than their
rural counterparts. Third, labor force participation of household heads or adults is lower in forest
villages (table 1 in appendix). However, these differences are not significant and the observations in
these two datasets are comparable. We can confidently develop the model with these two surveys.
Consumption model with HBS 2013
The consumption model was designed to estimate the log per-capita expenditures (including all
frequent food and non-food expenditures, including health expenditures, imputed rents and
estimated depreciation of durable goods) on household characteristics, such as household
composition, assets, employment categories, and education for the rural sample. The poverty rate
among rural households was estimated at 34.9 percent using a poverty line of $7 (in terms of 2011
PPP).
Following the SWIFT methodology, an evaluation of the mean-squared errors of the consumption
estimates, and the absolute differences in the poverty estimates yields an optimal p-value of 0. 005
for the consumption model. The final stepwise regression selects 14 out of 23 potential
explanatory variables. Table 2 lists the final regression for the poverty model. The kernel density
plots for both original and measured values confirm that the estimated consumption density across
the population is similar to the original (Figure 1).
Table 1 Forest-SWIFT modelsConsumption Forest income
Age of HH head 0.006 (0.001)HH head is married (=1; 0 if not) -0.133 (0.046)HH head education is primary (=1; 0 if not) 0.137 (0.048)HH head education is secondary (=1; 0 if not) 0.231 (0.056)HH head education is tertiary or higher (=1; 0 if not) 0.393 (0.075)Head is employed (=1; 0 if not) 0.312 (0.047)Number of retired 0.347 (0.04)HH head is an employer (=1; 0 if not) 0.264 (0.091)HH head employed in agriculture (=1; 0 if not) -0.139 (0.029)HH with members working as unpaid worker in family business -0.995 (0.296)HH receiving forest wages 2.617 (0.184)Household size -0.071 (0.01) -0.261 (0.034)Dependency ratio -0.423 (0.044)Household having central heating system (=1; 0 if not) 0.305 (0.051)Non-overcrowding (=1; 0 if not) 0.23 (0.041)HH owns computer (=1; 0 if not) 0.272 (0.028) HH owns tractor 0.359 (0.112)HH owns chainsaw 0.292 (0.114)HH owns car/truck -0.229 (0.110)Village electrical network 0.249 (3.47)
7
Village water network 0.143 (-4.14)Village with net migration and high poverty 0.279 (-3.96)Village with no net migration and high poverty 0.127 (-4.85)Constant 8.15 (0.11) 5.304 (0.285)Number of obs 3006 1155R-squared 0.439 0.352Root MSE 0.524 1.535Note: non-overcrowding is defined as less than 1.5 persons per room.Standard errors in parenthesis. All variables are significant at 0.005 in consumption model and at 0.01 in the forest income model.
Figure 1 Kernel density distribution for consumption0
.2.4
.6kd
ensi
ty lo
gwel
fare
4 6 8 10 12x
kdensity logwelfare Original kdensity logwelfare imputed
Forest income model with SEHS 2016
The forest income model was designed to estimate the log of per capita forest income on a similar
set of household characteristics. For cross-validation purposes, the median per capita forest income
serves as the threshold in place of a poverty line. The model only includes households who report
non-zero, positive, forest-related income from any source, which represents around 60 percent of
the original sample. The final model selects 10 variables out of 25 using a p-value 0.01 (table 2). The
following kernel density plots confirm the matching distribution of the income estimates.
Figure 2 Kernel density distribution for forest income
8
0.0
5.1
.15
.2kd
ensi
ty ln
pc_f
binc
-5 0 5 10 15x
kdensity lnpc_fbinc kdensity lnpc_fbinc _mi_m=7
4. Results and discussion
Using the two models from above, we piloted a questionnaire including a household roster and 20
questions on forest collection and wage, dwellings and assets. We conducted the survey in 100 of
the 202 villages surveyed in 2016 SEHS creating an unbalanced panel dataset with the re-survey of
1000 households from the original sample. The sample is representative of national forest villages.
The survey took place over a three-week period, enumerators spending less than 20 minutes with
each household. GPS coordinates, cell phone numbers from respondents were recorded to monitor
survey progresses and to check quality of answers on collection of forest products. Questions in this
questionnaire focused on the determinants identified in the models, on participation in forest and
non-forest related activities, and participation in associations and groups.
When doing the last phase of Forest-SWIFT, the modeling appears to be more complicated than
expected because of issues during the collection of SEHS 2016 and Forest-SWIFT 2017 data. Issues
in questionnaire design arouse: while in SEHS 2016, the team collected data on more than 90
products, Forest-SWIFT questions were only for 9 products which increased response rate for
those questions dramatically. Moreover in 2016, if a household admitted to collection of a product,
such as firewood or pinecones, the 13 further questions about each product were asked. Collecting
data on a smaller set of forest products possibly encouraged villagers to answer to more questions.
Participation in forest extraction in 2017 increased on average by 12-percentage points. For these
reasons, indicators on forest collection were excluded from the post-survey revisions of the forest
income model.
9
In addition, when collecting Forest-SWIFT data, Turkish households received selected assets
through social assistance program. As such the presence of assets such as freezers and solar panels
no longer correlated with wealth and could not be included in the consumption and forest income
models.
The Forest-SWIFT data bring the following results. Participation in forest related activities is high
(95 percent) with 83 percent of households generating income values from these activities and 77
percent extracting forest products for their home-consumption (table 2). The most important
source of forest income remains forest-wage activities. The forest products collected the most were
walnut, hazelnut, linden, and herbs, while hazelnut and walnut were the most likely to be
commercialized (table 2 in appendix). On average, 37 percent of households collect between 4 and
9 different forest products but only 14 percent of households sell 1 forest product; households
mainly collect forest products for their home-consumption (table 3 in appendix).
Table 2 Participation in forest and non-forest activities (percent)Participation (percent of households)
FOREST-RELATED 94.78%Income from Forestry and/or NTFP Production 83.20%
Wage 82.35%Market Sales from Collections 14.06%
Other, Unidentified 0.34%Subsistence Value from Collections 76.85%
NON-FOREST RELATED INCOME 96.57%Wage
Agricultural Income 59.12%Pensions 70.68%
Other, Unidentified 47.99%Source: authors’ estimations using 2017 Forest-SWIFT.Note: Results are weighted at the household level
Forest households in 2017 are better-off than rural households in 2013 but one out of four
households are poor. Predicted consumption values with SWIFT 2017 data are higher than
consumption values in 2013 (table 3). Over time poverty has reduced which has been the case
when looking at national poverty rates. It is true that this level of poverty is different from the one
reported in PROFOR (2017) but one cannot compare these two as we are using a different poverty
line than the PROFOR report, and a consumption aggregate rather than income.
Table 3 Results for imputed poverty rate and consumption per capita (Rural)HBS 2013 (original) SWIFT 2017
Poverty rate (percent) 34.9 23.2Mean per capita Consumption 5,906 6513.206
10
Source: authors’ estimations using HBS and Forest-SWIFT data. Note: Consumption values in HBS 2013 and in Forest-SWIFT 2017 are all in Turkish Lira 2013. The poverty line equal of US$ 7 per capita per day was converted to TL 2013 using 2011 PPP.
Forest income is greater in 2017 than in 2016. In 2017, forest income is on average equal to TL
1223.82 in value of 2016 (table 4). Households are more likely to participate in forest-related
activities in 2017. This difference in participation is accounted for by the fact that response rates
were higher in 2017 than in 2016. This higher response rate can be attributed to the change in
questionnaire design and length. While the original questionnaire (2016 SEHS) had forest-related
questions about 90 items, our follow up survey focused on a subsample of forest items, potentially
reducing respondents’ fatigue. Higher response rate results in increased accuracy and this increase
in forest income over time.
Table 4 Results for Imputed Forest Income and ratio below median forest income
SEHS 2016 Forest-SWIFT 2017
Ratio below median forest income (percent) 50.02 46.70Mean log per capita forest income 4.631 5.003Mean per capita income 893.49 1223.82Forest dependence (percent) n.a. 19.2Note: Forest incomes with SEHS 2016 and Forest-SWIFT 2017 are in Turkish Lira 2016Suggested formula for inverse natural log = exp(m+ sigma (m)^2/2)Forest dependence is equal to the average household forest income over household average consumption that is a proxy of permanent income.
Using the predicted values of consumption and of forest income, the average relative contribution
of forests to total income is approximately 19 percent. Consumption represents households’
permanent income. This result is comparable to the levels of forest dependence reported by
Angelsen and colleagues (2014) in their global study. Although forest may not be perceived as their
main source of income, households depend on this source for their livelihoods.
Households who are highly dependent on forest resources are more likely to be poor. These
households have a lower share of members working in wage activities but more households
working for their own account. In addition, forest dependent households hold fewer assets and
except for chainsaw that is used for forest-related activities. These households are less likely to own
agricultural land.
Table 5 Differences across engagement in forest activities and by poverty
VariableNon-Poor Poor
Not engage
d in forest
Engaged in
forest
High Forest
Dependence
Low Forest
dependence
11
Household Size 3.46 3.90 2.90 3.60 3.7 2.9Share of prime aged members (%) 74.0 63.3 62.0 72.2 72.7 67.0Sex ratio (%) 45.4 47.2 47.3 45.7 45.8 46.1Share of employed members (%) 31.0 27.7 9.1 31.4 31.2 26.1Share of retired members (%) 22.2 18.0 37.6 20.3 19.4 29.5Share of members working on own account (%)
31.4 29.8 5.8 32.4 32.0 26.5
Share of housewives (%) 30.2 29.2 33.7 29.8 29.2 33.3Share of members that attended school (%)
90.2 87.8 87.1 89.8 89.9 88.6
Household head age (years) 54.1 53.5 56.1 53.9 53.5 56.0Household head is male (%) 97.4 95.9 95.2 97.1 97.0 97.3Household head is married (%) 90.5 91.9 85.0 91.1 91.1 89.7Household head attended school (%) 95.9 93.9 90.2 95.7 95.6 94.8HH Owns Car/Truck (%) 61.3 55.4 27.6 61.8 63.4 44.8HH Owns Chainsaw (%) 57.6 52.8 34.0 57.8 57.1 54.2HH Owns tractor (%) 44.8 42.0 22.1 45.4 45.2 39.6HH Owns water pump (%) 17.9 15.9 3.1 18.2 19.0 10.5HH Owns color tv (%) 99.2 99.2 98.4 99.2 99.3 99.0HH Owns computer (%) 18.0 7.8 14.7 15.9 16.8 11.5HH owns Agricultural Land (%) 77.3 78.0 29.6 80.1 72.5 62.5
Forest dependence (%) 16.0 30.5 n.a. 19.2 3.26 117.3
7Forest Income per capita (2013 TL) 1,047.4 747.2 n.a. 981.2 194.5 5819.8Poverty rate (%) n.a. n.a. 18.56 22.21 20.69 27.90Consumption (2013 TL) 7,621.6 2,585.1 7,237.8 6,473.3 6,640.6 5,951.7*Weighted at the household level**Above/below forest dependence average of 19 percent.
Poor forest households are more dependent on forest income for their livelihoods than non-poor
forest households. On average, poor forest household generate 30.5 percent of their income
through forest-related activities. In absolute terms, poor households benefit less from forest related
activities but are more dependence (Angelsen et al., 2014). Forest-related activities appear to be a
complement to agricultural activities; forest-dependent households are engaged in agriculture but
to a lower extent than other households. Forest-dependent households have less diversified
livelihood activities.
Differences between poor and non-poor are striking with respect to asset ownership, internet
access, and employment rates. On average, poor households have fewer assets although some
differences are not large for televisions. On average, poor households are bigger with more
dependents and more female members. Poor households have lower rates of employment, and a
smaller number of retired members who receive pensions. Household heads are slightly less likely
to have attended school. On average poor households have less access to internet than others.
12
Poverty in Turkey is then higher in forest areas than in the rest of the country. Forest-SWIFT
methodology provides a measure comparable to the national poverty measure. The results are well
in line with measures on forest dependence at the global level. As found in earlier studies in Turkey,
households actively engaged in forest activities are less educated (Atmiş et al., 2007). Although not
reported here, women and youth play an important role in forest management (Atmiş et al., 2007).
Often these groups of population are the most exposed to poverty emphasizing the importance of
forest products for their livelihoods.
Households who are poor and forest dependent present similar characteristics; forest investments
to improve the livelihoods of forest dependent people are likely to improve the living conditions of
the poor if targeting is well done. Forest-SWIFT can help teams to target forest-dependent poor
households at baseline with a simple questionnaire asking only the questions used here. A follow-
up questionnaire can easily be implemented to gather more evidence throughout the projects.
In addition, Turkey National Statistics Office could design a questionnaire to have all questions to
predict forest income as part of their national household budget survey to accurately measure
forest dependence. Although sampling might remain an issue, an oversampling of forest areas could
be a good solution for Turkey to gather more evidence on forest households and bring development
policies tailored to their needs. Considering the rapid exodus of forest households towards urban
areas (Atmiş et al., 2007), it is important to find solutions to Government of Turkey is very
concerned with this issue as forest households provide ecosystem services that are very costly to
replace. Forest-SWIFT can be used to see if incentives given to forest households to provide
ecosystem services are enough to ensure such provision and to take households out of poverty.
Such an analysis could be done using distributional impact analysis in which one can establish a
monetary amount to be given and see the impact on consumption status, assuming everything else
remains constant.
5. Conclusion
Forestry Modules and Forest-SWIFT bring more evidence on poverty and forest income in forest
environments. In Turkey, the Forestry Module was designed to be representative of forest villages
and to measure forest income. Forest activities are an important source of income for forest
villagers, especially since poor households depend more on these activities than non-poor ones.
Using Forest-SWIFT data, we have a robust measure of poverty based on consumption projections
13
without having to collect data on consumption. Collecting consumption data is costly and time-
intensive; Forest-SWIFT allows us to track poverty using a small cost-effective survey.
With our data from 2017, we the poverty rate among forest villagers in Turkey is 22 percent. This
poverty rate is in line with poverty reduction observed in Turkey in the last decade. Forest
participating households generate about 15 percent of their livelihoods from forest-related
activities, which is not a negligible part. Households are highly engaged in forest-wage activities and
use NTFPs for their home consumption. One out of ten households sell forest products, mostly
hazelnuts.
Upon the first application of both surveys in Turkey, the Forestry Module and Forest-SWIFT
emphasize the importance and power of having such data. The issues identified during the
modeling do not undermine the fact that these datasets provide detailed evidence on households’
livelihoods. The resulting poverty rate is representative at the forest level and additional data
collection should be conducted on a regular basis to map changes in poverty. Future research could
exploit the panel structure of the datasets to analyze trends and roots of these changes. Up to date
data could help government and practitioners improve the design of programs and policies to pull
households out of poverty, besides improving targeting of households and monitoring impact of
interventions.
14
References
Ahmed F., C. Dorji, S. Takamatsu, and N. Yoshida. 2014. “Hybrid Survey to Improve the Reliability of
Poverty Statistics in a Cost-Effective Manner” World Bank Policy and Research Paper 6909. World
Bank, Washington D.C. Angelsen A., H.O. Larsen, J.F. Lund, C. Smith-Hall, S. Wunder (Eds.). 2011.
Measuring livelihoods and environmental dependence: Methods for research and fieldwork,
Earthscan, London http://www.cifor.org/publications/pdf_files/Books/BAngelsen1102.pdf
Angelsen A., P. Jagger, R. Babigumira, B. Belcher, N. J. Hogarth, S. Bauch, J. Borner, C. Smith-Hall, and
S. Wunder. 2014. “Environmental Income and Rural Livelihoods: A Global-Comparative Analysis.”
World Development. Vol 64
Atmiş, E., S. Özdenb, and W. Lise. 2007. “Public participation in forestry in Turkey.” Ecological
Economics. Vol. 62(2007), pp 352-359.
Atmiş, E., H. Batuhan Günşen, B. Bayramoğlu Lise, and W. Lise. 2009. “Factors Affecting Forest
Cooperative’s Participation in Forestry in Turkey.” Forest Policy and Economics 11 (2): 102–8.
https://doi.org/10.1016/j.forpol.2008.10.002.
Byron, N., Arnold, M., 1999. “What futures for the people of the tropical forests?” World
Development 27, 789–805.
Deaton, A. 1997. The Analysis of Household Surveys: A Microeconometric Approach to
Development Policy. Washington, D.C.: The Johns Hopkins University Press for the World Bank.
Elbers C., J.O. Lanjouw and P. Lanjouw, 2002. “Micro-Level Estimation of Welfare,” Policy Research
Department Working Paper No. WPS2911, The World Bank.
Elbers C., J.O. Lanjouw and P. Lanjouw, 2003. “Micro-level estimation of poverty and inequality.”
Econometrica, Vol. 71, No. 1 (January, 2003), 355–364
FAO, CIFOR, IFRI and World Bank. 2016. National socioeconomic surveys in forestry: guidance and
survey modules for measuring the multiple roles of forests in household welfare and livelihoods, by
R.K. Bakkegaard, A. Agrawal, I. Animon, N. Hogarth, D. Miller, L. Persha, E. Rametsteiner, S. Wunder
and A. Zezza. FAO Forestry Paper 179.
Kuhn, Max and Kjell, Johnson. 2013. “Applied predictive modeling” Springer
15
Oxford Poverty and Human Development Initiative (OPHI), 2014. “Towards frequent and accurate
poverty data”, Briefing Paper. United Nations Sustainable Development Solutions Network
(UNSDSN), http://unsdsn.org/wp-content/uploads/2014/09/Towards-Frequent-and-Accurate-
Poverty-Data-WEB.pdf
PROFOR/World Bank. 2017. “Poverty, forest dependence, and migration in the forest communities
of Turkey. Evidence and policy impacts” PROFOR Report. Washington D.C.
https://www.profor.info/sites/profor.info/files/PROFOR_TurkeyForestCommunitiesSocioeconomi
cs_web.pdf
Serajuddin U., H. Uematsu, C. Wieser, N. Yoshida, and A. Dabalen. 2014. “Data Deprivation. Another
Deprivation to End” World Bank Policy Research Working Paper7252/ 1-22.
Vedeld, Paul, Arild Angelsen, Jan Bojö, Espen Sjaastad, and Gertrude Kobugabe Berg. 2007. “Forest
Environmental Incomes and the Rural Poor.” Economic Perspectives and Analyses of Multiple
Forest Values and Sustainable Forest Management 9 (7): 869–79.
https://doi.org/10.1016/j.forpol.2006.05.008.
Vira, B., C. Wildburger and S. Mansourian. 2015. “Forests, Trees and Landscapes for Food Security
and Nutrition. A Global Assessment Report.” IUFRO World Series. Volume 33. Vienna
Wunder, S., Angelsen, A., & Belcher, B. (2014). Forests, Livelihoods, and Conservation: Broadening
the Empirical Base. World Development, 64, Supplement 1, S1–S11.
https://doi.org/10.1016/j.worlddev.2014.03.007
Yildirim H. T. and Kose M. 2013. “Non-Wood Forest Products Production Policy and Marketing in
Turkey.” Conference Paper. Forest Products Society's 67th International Convention Forest
Products in a Global Sustainable Economy.
Yoshida, N., R. Munoz, A. Skinner, C. Kyung-eun Lee, M. Brataj, W. Durbin and D. Sharma. 2015.
“SWIFT Data Collection Guidelines version 2”, World Bank. Washington D.C.
16
Appendix
Table 1 Summary statistics for SEHS 2016 and HBS 2013
HBS 2013 (rural) SEHS 2016
Household size 3.88 4.52Dependence ratio 0.607 0.532Head CharacteristicsAge 52.08 53.35Male 87.33% 96.48%No school 19.70% 9.06%Primary school 80.30% 90.06%Employed 71.81% 65.48%Prime aged adults’ characteristicsNo school 18.49% 9.85%Primary school 81.51% 90.04%Employed 61.01% 49.60%Neither student, nor employed (15-29 yrs) 29.45% 21.16%Unemployed 2.74% 48.89%Labor force participation 61.01% 56.71%Female labor force participation 38.67% 26.30%HH Characteristicsa
Household with access to gas 1.82% 0.00%Household with access to piped water inside or outside house 98.43% 89.42%HH AssetsHH Assets: Total out of 5 3.27 3.94Household with access to refrigerator 97.84% 97.50%Household with fixed or at least one cell phone 96.66% 98.53%Household with access to internet at home 16.44% 7.64%Household with access to washing machine 92.29% 94.27%Observations 3,006 1,256Source: authors’ computation using SEHS 2016 and HBS 2013. Weights applied.Note: All statistics are at the HH level. a SEHS 2016 collected this data at the community level
Table 2 Most important forest products and sales
Products Extraction (percent) Sales (percent)Hazelnut 50.19% 12.93%Industrial wood 41.95% 0.32%Linden 58.13% 0.03%Opium 0.95% 0.33%Pinecone 15.91% 0.31%Sting nettle 37.14% 0.00%Trefoil 21.68% 0.11%Walnut 76.84% 4.12%Herbs 56.87% 0.07%
17
Note: Weighted at the household levelPercent of households extracting and selling forest products measured on sample of households extracting at least one forest product (970 of 1256 HH).
Table 3 Total forest products collected
Total (of 9) Collected For Sale Only0 22.58 85.941 16.54 14.012 10.76 0.053 12.544 9.165 12.126 9.317 6.248 0.729 0.03
18
End notes
19
i The World Bank works in more than 155 countries worldwide with the twin goals of reducing poverty and boosting shared prosperity. However, between 2002 and 2011, 29 countries did not have single poverty estimate and 28 only had one (Serajuddin et al., 2015). ii https://www.ifc.org/wps/wcm/connect/a8a598a6-572d-4146-9f4d-0ea35c395a06/SWIFT-brochure-13.pdf?MOD=AJPERES iii Turkey stopped calculating poverty at the urban and rural levels after 2013. Poverty rate in 2017 at the national level using a poverty line equal to 50 percent of the median household income was 13.5 percent (https://data.worldbank.org/topic/poverty?locations=TR). iv http://povertydata.worldbank.org/poverty/country/TUR v We pick ten folds but it could be any number of folds.vi The average of the sum of squared differences between yh and yh=xh
'∗βvii Here we are using World Bank’s 2013 ECAPOV Turkey dataset. ECAPOV- the World Bank’s harmonized poverty database for Eastern and Central Asia- harmonized Turkey’s 2013 Household Budget Survey (HBS) to create a number of standard indicators that can be compared across countries in the region. These indicators have been then published in the World Bank open data portal and World Bank World Development Indicators (WDI) http://databank.worldbank.org/data/reports.aspx?source=2&Topic=11viii In 2014, 2015, and 2016, TUIK has not released any statistics at urban/rural disaggregation due to changes introduced in administrative divisions throughout the country. Therefore, the numbers belong to 2013. In 2013, administrative regions could not be identified.