Poverty map for the Palestinian Territories

"PECS and Census 2017"

Technical report1

June 2019

1 This technical report is written by Aziz Atamanov and Nethra Palaniswamy from the Poverty and Equity Global Practice at the World Bank in close collaboration with Jawad Al-Saleh and Fida Twam (Palestinian Central Bureau of Statistics). The team is thankful for collaboration and constant support provided by Ola Awad (President) and Haleema Saeed (Director General) of the Palestinian Central Bureau of Statistics and Benu Bidani (Manager, World Bank). We are particularly thankful to Paul Corral (peer reviewer), Ken Simler, Minh Cong Nguyen and Xiayun Tan for very useful suggestions and help.

2

Table of Contents

1. Introduction ...... 5 2. Methodology ...... 6 Methodological approach ...... 7 Practical implementation ...... 8 2. Data ...... 9 PECS and Census ...... 9 Technical challenges ...... 11 Small localities ...... 11 Lack of matching variables ...... 12 3. Model selection and results ...... 13 Selecting the model ...... 13 Performance of the national model ...... 14 Model performance ...... 14 Standard errors ...... 16 Validation ...... 18 4. Mapping poverty and inequality results ...... 18 Annex ...... 22 References ...... 40

3

Tables

Table 1. Geographic boundaries and classifications in the Palestinian territories ...... 10 Table 2. Means and standard deviations of variables in PECS 2016 and Census 2017 used in the national model ...... 12 Table 3. Parameters from the national and regional models ...... 14 Table 4. Standard errors and coefficients of variation for locality poverty estimates obtained from the national and regional modes ...... 14 Table 5. Results from the Beta model, GLS ...... 15 Table 6. Results from Alpha model ...... 16 Table A1. Number of localities and enumeration areas in each governorate in PECS and Census ...... 22 Table A2 Results from the OLS and GLS models ...... 23 Table A3. Coefficients and RMSE from Beta model and 10-fold cross validation ...... 24 Table A4. Small area poverty and inequality estimates for merged localities ...... 25

Figures

Figure 1. Example of merging selected localities in Hebron ...... 11 Figure 2. Governorate standard errors for direct PECS 2016 estimates and for small area estimates ...... 16 Figure 3. Standard errors for original and merge localities by different distribution groups ...... 16 Figure 4. Coefficients of variation and small area poverty estimate at the locality and governorate levels, 2017 ...... 17 Figure 5. Comparison between the direct poverty estimates from PECS and small area estimates at the regional and governorate levels ...... 18 Figure 6. Small area estimates at governorate level, 2017...... 19 Figure 7. Small area estimates at governorate level across location types, 2017 ...... 19 Figure 8. Small area poverty estimates at the locality level, 2017 ...... 20 Figure 9. Small area Gini estimates at the locality level, 2017 ...... 21 Figure A1. Comparison of direct with small area estimates at the locality level ...... 22 Figure A2. Governorate coefficients of variation for direct PECS 2016 estimates and for small area estimates ...... 25

4

1. Introduction

For close to three decades, the people in the West Bank and Gaza have lived in a volatile and aid- dependent economic environment, defined by external restrictions on trade and the mobility of people, in addition to open-ended political uncertainty and internal challenges.

Over the period 2011-2017, which is most recent year for which data is available, poverty increased by 3.4 percentage points in the Palestinian territories as a whole, from 25.8 to 29.2 percent (Atamanov and Palaniswamy, 2018). However, trends in the West Bank and Gaza regions diverged sharply. The proportion of the population living below the poverty line in the West Bank declined by 3.9 percentage points, from 17.8 to 13.9 percent. Poverty in Gaza rose by 14.2 percentage points, from 38.8 to 53 percent. This divergence in poverty trends at the regional level widened the gap in living standards between the two Palestinian territories significantly. While the average household in the West Bank had slightly higher living standards in 2017 than in 2011, this improvement remained fragile to a worsening political outlook2; and the precipitous decline in living standards in Gaza was accompanied by an alarming fall in access to essential public services like water and electricity. Declining aid flows and a worsening outlook on conflict are therefore likely to place significant downward pressures on living standards in both the Palestinian territories.

The uniquely fragmented geography of the Palestinian territories is characterized by the isolation of Gaza from the West Bank, and man-made barriers to mobility within the West Bank (World Bank, 2014). This geography has consequences for poverty and economic development. The restrictions in mobility - that started in the early 1990’s and intensified in the period leading up to 2010 –defined important spatial disparities in economic welfare between the two regions of the West Bank and Gaza, and within the West Bank over this time. In the West Bank, restrictions took the form of internal barriers such as road closures and checkpoints within the territory; and Gaza faced external restrictions that increasingly “closed” it to the external world. These restrictions defined within-territory barriers to trade that resulted in spatial price differences (World Bank, 2011); created inequalities in access to limited labor market opportunities (Abrahams, 2017); and had an adverse effect on local economic performance as measured by data on night time light emissions (Van der Weide et al, 2018). At the micro-level, they negatively affected the welfare and living standards of the Palestinian people in multiple ways: they increased unemployment, reduced wages and the number of days worked, and increased poverty in localities with greater restrictions on mobility (World Bank (2011) and Caali and Miaari (2013)).

Little has changed in the political and economic constraints facing the Palestinian Territories since 2010. During 2011-17, a worsening political economy - defined by continued restrictions on mobility within the West Bank and the nearly complete isolation of Gaza - deepened economic distortions and continued to erode the productive base, particularly in Gaza.

Given this fragmented geography and spatial disparities, poverty could vary widely even within a space of a few kilometers. However, conventional household budget surveys do not provide accurate poverty estimates at levels lower than the region (West Bank and Gaza), masking substantial variation in

2 Even a 5 percent drop in expenditures in the West Bank - which would be much lower than the 20 percent decline observed during the 2007 Gaza internal divide, for example - could increase poverty by as much as 16 percent. A 15% percent expenditure drop will increase poverty in the West Bank by as much as 50 percent. wellbeing not only across governorates, but also within them (World Bank 2014). Poverty mapping, or Small Area Estimation (SAE) is a statistical inference technique that allows estimation for very small areas, by combining information from censuses and household surveys. For example, SAE may produce poverty estimates at the levels for which Palestinian Expenditure and Consumption Survey (PECS) is not representative, namely governorates and localities. A poverty map, which provides estimates of poverty at the level of lower administrative units of government, is a useful tool to inform the effective and efficient allocation of resources and programs based on greatest needs. In addition, superimposing auxiliary information on the poverty map can shed light on the correlates of poverty. For example, areas with high poverty may have high unemployment rates, low access to clean water or regular electricity supply. Well-off areas may have high shares of population covered by social assistance. Having access to such disaggregated information, at the level of localities and governorates, can help policymakers deliver better targeted and efficient interventions. The last poverty map for the Palestinian territories is quite outdated. It was constructed in 2013/2014 using the General Census of Population and Housing 2007 and PECS 2009. Given the availability of a new PECS 2016 and Census 2017, it is important to update the old poverty map using these most recent sources of information. This technical report describes the methodology and data used to produce small area poverty estimates for the Palestinian territories. The first section presents the methodology. The second section describes the data, and the technical challenges in estimating poverty at the locality level. The third section discusses selection of the best model, its performance and conducts validation exercises. The fourth section shows poverty and inequality estimates for different levels. 2. Methodology

The small area estimation (SAE) exercise is typically implemented using a household survey and a census. SAE combines the strengths of household budget survey and census to estimate poverty headcount and other indicators at low levels of geographic disaggregation. Budget surveys or household income and expenditure surveys, such as the Palestinian Expenditure and Consumption Survey (PECS), collect detailed information on household expenditure, which is often the basis for national estimates of welfare and inequality measures. The surveys usually allow subnational estimates at the first administrative level such as regions, provinces, or urban and rural areas. For instance, the PECS is representative at the regional level, i.e, West Bank and Gaza and across rural, urban areas and camps. However, given the tradeoff between detail and sample size, these surveys do not typically have large enough sample sizes to allow for reliable inferences at lower administrative units like districts, sub- districts, or census tracts. On the other hand, in a census, all households are interviewed, but asked only a limited set of information, and generally do not include information in income or expenditure. SAE leverages the advantages of both sources of information, the high level of detail in surveys like the PECS, and the large sample sizes in the census. It models the relationship between expenditure and individual, household, and location characteristics in the survey, and uses this relationship and household characteristics in the census to predict household expenditure into the households in the census (World Bank, 2014).

6

Methodological approach

The poverty and inequality estimates presented in this report follows the small area estimation (SAE) method developed by Elbers et al. (2003), referred to as ELL. The ELL model relies on detailed consumption expenditure data from a household survey such as the PECS to estimate a model for household consumption per capita for a given a set of observable household characteristics. The estimated model is then applied to the same set of characteristics in the population census to impute household consumption, and then estimate expected levels of poverty or inequality across localities in the census. Since poverty rates are estimated, they are subject to errors. Nevertheless, experience to date suggests that estimates can be sufficiently precise for purposes of informing policy choices (Bedi, Coudouel, and Simler 2007; World Bank 2012). The ELL approach also provides estimates of the standard errors, so the users can make informed decisions about accuracy of estimates. Formally, ELL assumes that log consumption per capita satisfies:

, where

is consumption per capita of household h residing in area c, are household characteristics, and representing the residual, which is composed of the area component and the household component . These two residual components have expected values of zero and are independent of each other, witch . These unconditional parameters have been estimated using Henderson’s method III, a commonly used estimator for the variance parameters of a nested error model (see Henderson 1953 and Searle et al. 1992).

ELL also allows for heteroscedasticity. The conditional variance of the remaining residual is modeled

via a logistics truncation as a function of household and area characteristics in

order to obtain an estimate of the variance . Once all variance parameters have been estimated (and hence, an estimate of the full variance-covariance matrix is available), is re-estimated using Feasible Generalized least Squares (GLS). The small area estimates and associated standard errors are obtained by means of simulation. Let R denote the number of simulations. The estimator takes the form:

Where h(y) is a function that converts the vector y with (log) consumption per capita poverty measure (such as the headcount rate), and where denotes with elements

With each simulation, both the model parameters and the errors and are drawn from their estimated distributions. The parameter is drawn by re-estimating the model parameters using the rth bootstrap version of the survey sample. Alternatively, may be drawn from asymptotic distribution (this is referred to as “parametric drawing”). Parametric drawing is computationally fast. A potential disadvantage is that the true distribution of the estimator for the model parameter vector does not necessarily coincide with the asymptotic distribution.

7

Using bootstrapping is more computationally intensive but is expected to provide more accurate results when the sample size is small. The point estimates and their corresponding standard errors are obtained by computing respectively the average and the standard deviation over these simulated values. The difference between the true poverty rate W in a given area and the estimator of its expectation, given the above model, has three components: − = ( − ) + ( − ) + ( − ). The first component ( − ) is idiosyncratic error, due to the presence of the error term in the first stage regression; this error is higher for smaller target populations. The second component ( − ) is the model error, determined by the variance of model parameters; this error depends on the precision of the welfare model and on the distance between the X variables across the survey and the census. The model error does not change systematically with the size of the target population. The fact that it depends on the distance between the X variables across the survey and the census highlights the importance of getting a set of variables from both the survey and the census that match well. Finally, the third component ( − ) is the computation error, based on the method of computation and is generated by the fact that is based on a finite number of simulations. This component of the error can be made as small as desired with sufficient computational resources. Practical implementation

In practice, the construction of a poverty map involves several steps, which are described below. Construction of variables for the model. The first important step for the construction of the poverty map is to identify the variables common to the census and household survey, and make sure they are coded and named identically. Some variables may be similar, but not identical. For example, there may be six categories for marital status in Census and only three in household survey. The goal of this step is to create new identical variables in both datasets. Once these variables are created, it important to check whether they match in a statistical sense across census and survey. An absolute minimum standard for this check is to compare the means across datasets taking into account the survey design in the household budget survey. Checking minimum and maximum values is also important to make sure distributions overlap. These statistical matching tests should be conducted at the level at which model is going to be estimated. To reduce the contribution from location effects, the poverty mapping literature often uses cluster means of household level variables, which are calculated from the census data so that they are available for all census and survey clusters. Given that the number of clusters is typically much smaller in the survey, it is also important to check the means of these variables across the datasets.3 Beta-model. Before running the first-stage (beta model) to explain the variation in the logarithm of consumption per capita, the decision should be made about the level at which the modelling exercise will be implemented. Ideally the consumption model estimated for the survey should be at the national level with regional dummies and their interactions with other variables to capture intra-regional variation. However, in many countries, location specific models are estimated instead, justified by the differences consumption patterns across areas. It is also important to make sure that the number of observations is sufficient to run region-specific models, to ensure that the results are stable.

3 The typical error structure in poverty maps includes two levels: household and cluster. In the case of the current poverty map, cluster is locality. Tarozzi and Deaton (2009), however, caution that the misspecification in the error structure can lead to overstating the precision of poverty estimates. Errors can be correlated at higher levels as well, for example at the level of governorates. We are not able to incorporate more than two layers in SAE Stata package, but, as a potential remedy, tried to include governorate dummies and interact them with household level variables.

8

From the pool of variables not excluded due to comparability concerns, a variety of model selection techniques (lasso, forward stepwise, backward stepwise, etc.) can be employed to select the variables to identify the best performing model in explaining variation in consumption. It is also important to complement these methods by manual selection of variables by dropping highly multicollinear variables and by conducting out of sample prediction tests. Alpha-model. The alpha model estimates the household effect selecting from statistically matched variables and their interactions with the predictions and squared predictions from the equation. Simulation. This is the final step to simulate consumption per capita in Census by applying the parameter and error estimates from the survey to the census data. This step involves the selection of distribution drawing methods for cluster and idiosyncratic effects; the choice of poverty line and poverty and inequality indices; potential trimming of consumption, coefficients and errors; and finally, simulation of the parameters and selection of the aggregation level of estimates indices (Nguyen et al. 2018). 2. Data

PECS and Census

The Palestinian poverty map uses unit record data from PECS 2016 and the general Population and Housing Census of 2017. The PECS is the national household expenditure survey implemented by the Central Bureau of Statistics. It was implemented on an annual basis, from 2000-2011; and then after a five-year gap in 2016. The latest 2016 PECS is a 12-month survey that covered 3739 households and was implemented from October 2016 to September 2017. The PECS is a is a multi-topic survey that collects information on a range of topics that include food and non-food expenditures (based on a 30-day diary for most items and 12 month and 3-year recall for some unregularly purchased items), core data on socio-economic and health characteristics, dwelling conditions, and labor activities and incomes. It is used to calculate official poverty rates for the Palestinian Territories. The survey produces representative poverty estimates at the regional level (West Bank and Gaza) and strata level (urban, rural, refugee camp). Governorate and locality level estimates are not designed to be representative. The sample frame for this survey was based on Census 2007, which was updated in 2013. The 2017 Palestinian Census was conducted in December 2017 and reached 4,705,601 people. As in the previous poverty map exercise, there were issues related to estimating poverty for population in Jerusalem governorate (World Bank 2014). Jerusalem covers East Jerusalem (J1, under Israeli control) and the rest of Jerusalem governorate (J2). There are many Israeli settlements in J2 and consequently, many parts of the governorate are inaccessible to Palestinians. As a result, both census and survey data have extremely limited coverage of Jerusalem governorate. Therefore, it was decided not to include Jerusalem governorates in the poverty mapping exercise. The same decision was made for the previous poverty map.

9

Box 1. Poverty measurement in the Palestinian Territories

The National Commission for Poverty Alleviation (1998) established an official definition of poverty in the West Bank and Gaza. The poverty line was set at the median expenditure level of certain key items of the poorest 25 to 30 percent of households and has been calculated every year. A household with two adults and four children and their spending patterns were used as the reference to develop the poverty line. Given the relative nature of the poverty line, the Palestinian Central Bureau of Statistics (PCBS) also constructed a consistent poverty trend for 2004–2009 using the 1997 line adjusted for inflation. In early 2011, the PCBS redefined the poverty line by changing the reference household to two adults and three children and started a new consistent poverty series beginning in 2010. The welfare aggregate includes expenditures on food and nonfood items; expenditures on health, education, and rent; and the purchases of durables during last 12 month and purchases of transport during last 3 years. The aggregate is spatially adjusted using a Laspeyres price index derived for the West Bank, Gaza, and East Jerusalem using a subset of food and nonfood prices from the CPI dataset. The most recent poverty trend for the Palestinian territories ends in 2016, and it is based on the Palestinian Expenditure and Consumption Survey implemented over the period October 2016- September 2017. (Al-Saleh et al. 2018)

There are four levels of geographic areas in the Palestinian territories: regions, governorates, localities and enumeration areas. Given that the PECS 2016 was based on the sample frame from Census 2007, enumeration areas in PECS 2016 and Census 2017 are not identical. After dropping Jerusalem, the geographic division is shown in the following table (Table 1). The country is divided into two regions, 15 governorates and 556 localities. PECS 2016 covered 167 localities as shown in the annex (table A1). For the poverty map exercise a seven-digit hierarchical location id was created. Location ID=RGGLLLL, where R is one-digit region code (2), GG is two-digit governorate code (15) and LLLL is four-digit location code (556). A separate location id was created to aggregate poverty estimates at governorate level by location type (rural/urban and camp). Table 1. Geographic boundaries and classifications in the Palestinian territories The Palestinian Territories 2 regions West Bank Gaza 15 governorates Jenin North Gaza Tubas Gaza Tulkarm Deir Al Balah Nablus Khan Yunis Qalqiliya Rafah Salfit Ramallah & Al Bireh Jericho & Al Aghwar Bethlehem Hebron Jerusalem (dropped from estimation) 556 Localities Enumeration areas

10

Technical challenges

Small localities

Many localities in the Palestinian territories are very small, with most of them having a population size below 1000 households. Simulating poverty for such small localities will result in very high standard errors and will not be reliable. As in the previous round of poverty map exercise, the decision was made to try to aggregate, if possible, geographically contiguous4 localities to cover at least 1000 households in each. Local knowledge and information from the census such as demographics and labor market indicators were used to merge the most similar contiguous localities until this minimum threshold was reached. Figure 1 illustrates how merged localities look like: 21 original localities (grey color) were merged into five localities. As a result of this exercise, 151 merged localities were created. The smallest locality after merging had more than 1,000 households. The list of original and merged localities is provided in the annex.

Figure 1. Example of merging selected localities in Hebron

4 There are merged localities which were not directly adjacent to each other, but shared similar socio-economic characteristics.

11

Lack of matching variables The 2016 PECS was based on the old sample frame drawn from Census 2007. Even though the frame was updated in 2013, using an old frame results in a statistical mismatch between many variables created. At the national level, only for 81 variables do the census means belong to the 99 percent confidence interval from PECS. To illustrate, household size, which is typically a strong correlate of poverty and features in most poverty map models, was not statistically similar due to the use of the older sample frame in the PECS. Another problem was related to many missing values in the sector of employment in the Census. Consequently, this information could not be used in the model, even though this variable was crucial in the previous poverty map (World Bank 2014). Besides household level variables, we have constructed a range of variables measured as averages at the original locality level in Census and used them in the model. The inclusion of locality-level variables into the beta model aims to explain the variation in consumption per capita across target areas. Definitions of each variable used in the final national level model along with their means and standard deviations are shown in table 2. Table 2. Means and standard deviations of variables in PECS 2016 and Census 2017 used in the national model PECS Census Variables mean sd mean sd Proportion of female members in the household 50% 17% 49% 17% Household head has difficulty (great difficulty or can't do completely) in any ability 4% 20% 4% 19% Household head education level is higher than secondary 23% 42% 26% 44% Number of rooms per capita (adult equivalent) in the household 1.0 0.4 1.1 0.5 Number of children [0-14] in the household 2.8 2.0 2.6 1.9 Number of bedrooms household has 2.4 0.9 2.3 0.8 Household head (aged 15 and above) employed in Israel/settlement 11% 32% 12% 32% Household head (aged 15 and above) employed in national government 18% 39% 17% 37% Main source of heating is diesel 0.3% 5% 0.2% 4% Freezer 11% 31% 11% 32% Vacuum 36% 48% 39% 49% Phone line 35% 48% 34% 47% Computer 38% 49% 39% 49% Private car 26% 44% 26% 44% iPad 21% 40% 19% 39% Washing machine 96% 20% 96% 19% Dwelling type is villa 1% 9% 1% 11% Interaction, Gaza region and female head 2% 15% 3% 16% Interaction, West Bank region and head employed in national government 7% 26% 6% 23% Bethlehem Governorate 5% 22% 5% 22% Interaction, North Gaza governorate and number of children [0-14] in the household 29% 110% 26% 103% Interaction, Khan Yunis governorate and head employment status is employer 1% 9% 0% 5% Interaction, Gaza governorate and household head education level is higher than secondary 3% 17% 4% 20% Interaction, Hebron governorate and waste is burnt 1% 9% 1% 9% Interaction, Hebron governorate and number of children [0-14] in the household 47% 129% 46% 129% Interaction, Hebron governorate and private car 7% 26% 6% 23% Average share of households at the locality using gas as main source of energy 20% 21% 20% 21% Average share of households at the locality living in flat/apartment 60% 25% 62% 24% Average share of households at the locality using electricity as main source of energy 30% 17% 31% 17% Average share of households at the locality having solar boiler 56% 17% 55% 17% Average share of households at the locality using bottled water as main source 1% 3% 1% 2% Average share of households at the locality with a head having a chronic disease 19% 5% 20% 5% Source: PECS 2016 and Census 2017, authors’ calculations Note: Population weighted means except locality level means from Census.

12

3. Model selection and results

Selecting the model

The first step in the modelling process is to make a decision about the number of consumption models to be estimated. The model can be run either at the national or subnational level. The lowest level for the model should be at the strata level. The previous poverty map was based on two regional models constructed for West Bank and Gaza regions separately (World Bank 2014). In this round, we tried to run the model at the national and at the regional levels. The decision on which number of models to use will be based on comparing standard errors of the obtained poverty estimates. Testing statistical matching of the potential variables for each model was done at the level that the model is going to be estimated at. Thus, the selection of the variables for the West Bank regional model was done for the West Bank region, while selection of the variables for the Gaza model was done for the Gaza region alone. Three models were constructed in an iterative way using an OLS regression, starting from the full model with all possible variables and dropping each insignificant variable one by one. This procedure was complemented by using a lasso technique and manual adjustment of the model to achieve the highest adjusted R-squared (the R-Squared measures the ratio of explained variance by the model to the total variance of actual household consumption). In order to avoid the potential problem with over-fitting the model, we also used the Bayesian Information Criterion (BIC), which incorporates a penalty for model complexity. Selected specifications were tested for multicollinearity by calculating variance inflation factors (centered variance inflation factors for the final specification are provided in the annex). The models were revisited again based on significance of the variables in a GLS regression. Finally, the resulting models were tested for stability and out of sample prediction discussed further in the text. The key model parameters are shown in table 3.5 All models have high explanatory power. The adjusted R-squared is higher than 60 percent in the national and Gaza models. In all models, the share of cluster level errors in the variance of the total error is less than 5 percent. This is much lower than the 10 percent benchmark used in many technical reports. Yet, even given this information, it is still difficult to decide at which level the consumption models should be estimated.

5 All small area estimates in this note are done in SAE Stata package (Nguyen et al. 2018).

13

Table 3. Parameters from the national and regional models National model West Bank model Gaza separate model Number of observations 3397 2555 842 Adjusted R-squared 0.63 0.45 0.62 R-squared 0.63 0.46 0.63 Root MSE 0.37 0.40 0.34 F-stat 181.61 60.55 98.65 Alpha model diagnostics Number of observations 3397 2555 842 Adjusted R-squared 0.00 0.01 0.01 R-squared 0.00 0.01 0.02 Root MSE 2.275 2.24 2.25 F-stat 5.17 8.56 6.89 Model parameters Sigma ETA sq. 0.0019 0.0045 0.0004 Ratio of sigma eta sq over 0.0137 0.0285 0.0036 Variance of epsilon 0.14 0.15 0.12 Sampling variance of Sigma 0.0000 0.0000 0.0000 Source: PECS 2016, authors’ calculations. Note: Original localities (556) are clusters, bootstrap is done at the level of enumeration areas. Empirical Bayes methods is used.

Another important criteria used to assess performance of the estimated models includes checking standard errors and coefficients of variation (standard error over poverty rate) for locality level poverty estimates. Results are shown in table 4. Median standard errors and the coefficient of variation are lower for poverty rates obtained from the national model. We also checked the accuracy of locality level poverty rates by comparing them with direct poverty rates from PECS. These results are shown in the annex. Estimates for both models are highly correlated with direct estimates (not representative) without a distinct advantage in one set of results over the other. Given this, the decision was made to use the national level model. Table 4. Standard errors and coefficients of variation for locality poverty estimates obtained from the national and regional modes Model median se 75% se median cv National model 3.0% 5.2% 18.6% West Bank and Gaza separate models 3.9% 5.2% 21.8% Source: PECS 2016 and Census 2017, authors’ calculations Note: Original localities (556) are clusters, bootstrap is done at the level of enumeration areas. Empirical Bayes method is used.

Performance of the national model

Model performance Table 5 shows the coefficients from the final specification of the feasible GLS model along with standard errors. All coefficients are significant at five percent significance level and have the expected signs. For instance, a higher share of female members, children and the fact of living in localities with high share of households with disabled heads is associated with lower consumption per capita. In contrast, access to assets, better education of the head of household, and employment in the national government is associated with higher consumption per capita. The obtained coefficients are robust and do not vary much across OLS and GLS estimates (table A2). Multicollinearity also does not seem to be an issue in the

14 model. The highest variance inflation factor is around 2.6 which is much lower than 4-10 (typically considered as an indication of a potential multicollinearity). In order to check for overfitting and out of sample prediction, several tests were conducted. We split the data into 10 groups and for each group we fit the final specification using the other k-1 groups. The resulting parameters were used afterwards to predict the dependent variable in the unused group. We compare average coefficients from 10 regressions with coefficients from beta model. We also compare root mean squared errors (RMSE). None of the coefficients from 10-fold validation are very different from the ones in Beta model (see table A3 in the annex). The average root mean standard error is also very close to the one obtained from Beta-model. As an additional test, we have run a Leave-One-Out Cross-Validation (Daniels 2012). This validation estimates the model with all but the ith observations. The prediction error is calculated for each predicted value of the ith observation. The average RMSE is about 0.38 and very close to actual one from Beta-model. The results from the alpha model are presented in table 6. All coefficients are significant at one percent level.

Table 5. Results from the Beta model, GLS explanatory variables Log of consumption per capita Dependent variable: logarithm of consumption per capita coefficient se Proportion of female members in the household -0.172*** 0.04 Household head has difficulty (great difficulty or can't do completely) in any ability -0.109*** 0.03 Household head education level is higher than secondary 0.0798*** 0.02 Number of rooms per adult equivalent in the household 0.429*** 0.02 Number of children [0-14] in the household -0.0942*** 0.00 Number of bedrooms household has -0.0927*** 0.01 Household head (aged 15 and above) employed in Israel/settlement 0.111*** 0.02 Household head (aged 15 and above) employed in national government 0.109*** 0.02 Main source of heating is diesel 0.575*** 0.12 Freezer 0.0816*** 0.02 Vacuum 0.115*** 0.02 Phone line 0.104*** 0.02 Computer 0.0855*** 0.02 Private car 0.334*** 0.02 iPad 0.0979*** 0.02 Washing machine 0.112*** 0.03 Dwelling type is villa 0.308*** 0.07 Interaction, Gaza region and female head 0.134*** 0.05 Interaction, West Bank region and head employed in national government -0.135*** 0.03 Bethlehem Governorate 0.191*** 0.03 Interaction, North Gaza governorate and number of children [0-14] in the household 0.0270*** 0.01 Interaction, Khan Yunis governorate and head employment status is employer 0.315*** 0.07 Interaction, Gaza governorate and household head education level is higher than secondary 0.203*** 0.04 Interaction, Hebron governorate and waste is burnt -0.271*** 0.08 Interaction, Hebron governorate and number of children [0-14] in the household 0.0238*** 0.01 Interaction, Hebron governorate and private car -0.172*** 0.03 Average share of households at the locality using gas as main source of energy 0.227*** 0.05 Average share of households at the locality living in flat/apartment -0.225*** 0.03 Average share of households at the locality using electricity as main source of energy 0.388*** 0.04 Average share of households at the locality having solar boiler 0.137** 0.05 Average share of households at the locality using bottled water as main source 1.703*** 0.29 Average share of households at the locality with a head having a chronic disease -0.557*** 0.16 Constant 8.723*** 0.06 Observations 3,397 R-squared 0.633 Source: PECS 2016, authors’ calculations.

15

Table 6. Results from Alpha model coefficients p-value Interaction, North Gaza governorate and rented dwelling 1.85 0.000 Diesel is main source of heating 1.63 0.000 Interaction, other type of fuel for cooking and predicted consumption per capita 0.19 0.000 Observations 3,397 adj R-squared 0.00 F (3,3392) 5.17 Root MSE 2.27 Source: PECS 2016, authors’ calculations. Note: robust standard errors

Standard errors

The key criterion for assessing the performance of the model is the standard errors of the poverty estimates. These standard errors measure uncertainty associated with the predictions. The higher the standard errors are, the lower precision of the estimate is and the less confident one can be about the poverty estimate. A failure to consider standard errors in comparing poverty estimates may result in the wrong conclusion about statistical difference between two estimates. Poverty estimates at governorate level produced with small area estimation method have much smaller standard errors than direct non-representative estimates from PECS (figure 2). The 95 percent confidence intervals for small area poverty estimates become much narrower because of smaller standard errors. For example, poverty in Ramallah in PECS ranged from 8-24 percent in 2016, while in SAE it ranges from 8-11 percent. Using merged localities instead of the original ones also reduces standard errors substantially. Thus, the median standard error obtained for more than 550 original localities dropped from 5 percent to 3 percent when calculated for 151 merged localities (figure 3). Figure 2. Governorate standard errors for direct Figure 3. Standard errors for original and merge PECS 2016 estimates and for small area estimates localities by different distribution groups 8% PECS 2016 ELL, Census 2017 30% original localities merged localities 7% 6% 25% 5% 20% 4% 3% 15%

standard errors standard 2% 10%

1% errors standard

0%

5%

Gaza

Salfit

Jenin Rafah

Tubas 0%

Nablus

Hebron

1 5

Tulkarm

Qalqiliya

10 25 75 90 95 99

% %

Khan Yunis Khan

Bethlehem

North Gaza North

% % % % % %

Deir Al Balah Al Deir

median

Ramallah & Al Al & Bireh Ramallah Jericho & Al Al & Aghwar Jericho

Source: PECS 2016, authors’ calculations.

16

Figure 4 plots the coefficients of variation and poverty rates for three different levels at which the PECS is not representative: governorate (blue dots), location type of governorate (green dots) and merged localities (orange dots). Clearly the most precise poverty rates (estimates with the lowest standard errors) are obtained for the largest geographic units: governorates. For almost all of them, the ratio of standard errors to poverty estimates (coefficients of variation) do not exceed 15 percent. Estimates for different location types within governorates are also quite accurate – most coefficients of variation are below 20 percent. The least accurate estimates are at the locality level, despite relatively low median standard error, significant number of localities have quite high coefficients of variation (benchmark used here is 20 percent). As expected, many of them have quite low poverty rates which contributes to high coefficients of variation. With this caveat in mind, 82 localities from total 151 localities and 126 localities with poverty higher than 10 percent poverty rate have coefficients of variation below 20 percent.6

Figure 4. Coefficients of variation and small area poverty estimate at the locality and governorate levels, 2017

Source: PECS 2016 and Census 2017, authors’ calculations.

6 Coefficients of variation for direct West Bank and Gaza regional estimates are about eight and five percent accordingly.

17

Validation

The obtained small area poverty estimates are consistent with poverty rates derived from PECS at the regional and governorate levels. All of them belong to 95 percent survey confidence intervals (figure 5). Figure 5. Comparison between the direct poverty estimates from PECS and small area estimates at the regional and governorate levels

80%

70%

60%

50%

40%

30% poverty

20%

10%

0%

-10%

Gaza Gaza

Salfit

Jenin

Rafah

Nablus

Hebron

Qalqilia

Tulkarm

Palestine

West Bank West

Bethlehem

Khan Yunis Khan

North Gaza North

Valley

Ddeir al al Balah Ddeir

Tubas & Northen Northen & Tubas

Aariha & Alaghwar & Aariha Ramallah & & Albireh Ramallah Poverty map estimates PECS, lower bound PECS, upper bound Source: PECS 2016 and Census 2017, authors’ calculations.

4. Mapping poverty and inequality results

This section maps obtained results for poverty and inequality in the Palestinian territories. It is a short section because the team has made all results accessible online via interactive dashboards, in order to make the poverty map more user-friendly and accessible. In addition to indicators of poverty and inequality, the dashboards also visualize many different indicators from Census 2017 (such as employment, access to services, etc). All indicators of poverty and inequality are visualized as standalone indicators, as well as with these other indicators of living standards and welfare. Figure 6 shows small area poverty estimates at governorate level and figure 7 shows poverty estimates for each location type within governorates. Poverty does not vary much across governorates in the Palestinian territories. The difference between the most affluent governorate (Bethlehem) and the poorest (Qalqiliya) slightly larger than 10 percentage points. This gap in living standards widens substantially to more than 35 percentage points once we zoom into location types within governorates. The highest poverty rates across regions are observed in camps where poverty rates even in the West Bank can be as high as in some areas in Gaza. For example, poverty in Nablus camp is very high and is even close to the poverty in urban areas of North Gaza governorate.

18

Figure 6. Small area estimates at governorate level, 2017

Figure 7. Small area estimates at governorate level across location types, 2017

Source: PECS 2016 and Census 2017, authors’ calculations.

19

Poverty estimates at the merged locality level demonstrate even higher variability. The largest pockets of poverty are observed in Hebron governorate and the lowest poverty rates are observed in localities belonging to Ramallah & Al Bireh and Bethlehem governorates. Inequality, measured by Gini, is highest in localities with lower poverty such as Ramallah. Figure 8. Small area poverty estimates at the locality level, 2017

20

Figure 9. Small area Gini estimates at the locality level, 2017

Source: PECS 2016 and Census 2017, authors’ calculations. Note: Small area estimates were not produced for Jerusalem governorate.

21

Annex

Table A1. Number of localities and enumeration areas in each governorate in PECS and Census localities enumeration areas

PECS Census PECS Census

Jenin 22 84 35 Tubas 28 105 15 Tulkarm 41 143 35 Nablus 67 209 39 Qalqiliya 78 243 21 Salfit 88 263 14 Ramallah & Al Bireh 102 343 28 Jericho & Al Aghwar 108 357 21 n/a Bethlehem 117 408 14 Hebron 141 523 40 North Gaza 146 528 14 Gaza 150 533 21 Deir Al Balah 159 544 22 Khan Yunis 164 552 16 Rafah 167 556 14 Total 167 556 349 Source: PECS 2016 and Census 2017, authors’ calculations Note: Due to

Figure A1. Comparison of direct with small area estimates at the locality level a) National model b) Separate models

Source: PECS 2016 and Census 2017, authors’ calculations. Note: Original localities (556) are clusters, bootstrap is done at the level of enumeration areas. Empirical Bayes methods is used. Spearman rank correlation is higher than 75% for both models.

22

Table A2 Results from the OLS and GLS models OLS GLS Number of bedrooms household has -0.09272 -0.09263 Computer 0.085 0.082 Average share of households at the locality living in flat/apartment -0.225 -0.228 Average share of households at the locality with a head having a chronic disease -0.557 -0.598 Average share of households at the locality using gas as main source of energy 0.227 0.225 Average share of households at the locality using electricity as main source of energy 0.388 0.383 Average share of households at the locality having solar boiler 0.137 0.121 Average share of households at the locality using bottled water as main source 1.703 1.733 Dwelling type is villa 0.308 0.307 Proportion of female members in the household -0.172 -0.180 Freezer 0.082 0.098 Interaction, Hebron governorate and number of children [0-14] in the household 0.024 0.024 Interaction, Hebron governorate and private car -0.172 -0.169 Interaction, Hebron governorate and waste is burnt -0.271 -0.275 Interaction, North Gaza governorate and number of children [0-14] in the household 0.027 0.028 Interaction, Gaza governorate and household head education level is higher than secondary 0.203 0.209 Interaction, Khan Yunis governorate and head employment status is employer 0.315 0.311 Bethlehem Governorate 0.191 0.209 Household head has difficulty (great difficulty or can't do completely) in any ability -0.109 -0.109 Household head (aged 15 and above) employed in Israel/settlement 0.111 0.109 Household head (aged 15 and above) employed in national government 0.109 0.119 Household head education level is higher than secondary 0.080 0.077 Main source of heating is diesel 0.575 0.571 iPad 0.098 0.104 Number of children [0-14] in the household -0.094 -0.093 Phone line 0.104 0.109 Private car 0.334 0.333 Interaction, West Bank region and head employed in national government -0.135 -0.143 Interaction, Gaza region and female head 0.134 0.138 Number of rooms per capita (adult equivalent) in the household 0.429 0.428 Vacuum 0.115 0.111 Washing machine 0.112 0.112 Constant 8.72 8.74 Source: PECS 2016 and Census 2017, authors’ calculations.

23

Table A3. Coefficients and RMSE from Beta model and 10-fold cross validation average beta regressors (10 folds) model Proportion of female members in the household -0.17 -0.17 Household head has difficulty (great difficulty or can't do completely) in any ability -0.11 -0.11 Household head education level is higher than secondary 0.08 0.08 Number of rooms per capita (adult equivalent) in the household 0.43 0.43 Number of children [0-14] in the household -0.09 -0.09 Number of bedrooms household has -0.09 -0.09 Household head (aged 15 and above) employed in Israel/settlement 0.11 0.11 Household head (aged 15 and above) employed in national government 0.11 0.11 Main source of heating is diesel 0.58 0.57 Freezer 0.08 0.08 Vacuum 0.11 0.11 Phone line 0.10 0.10 Computer 0.09 0.09 Private car 0.33 0.33 iPad 0.10 0.10 Washing machine 0.11 0.11 Dwelling type is villa 0.31 0.31 Interaction, Gaza region and female head 0.13 0.13 Interaction, West Bank region and head employed in national government -0.13 -0.13 Bethlehem Governorate 0.19 0.19 Interaction, North Gaza governorate and number of children [0-14] in the household 0.03 0.03 Interaction, Khan Yunis governorate and head employment status is employer 0.31 0.31 Interaction, Gaza governorate and household head education level is higher than secondary 0.20 0.20 Interaction, Hebron governorate and waste is burnt -0.27 -0.27 Interaction, Hebron governorate and number of children [0-14] in the household 0.02 0.02 Interaction, Hebron governorate and private car -0.17 -0.17 Average share of households at the locality using gas as main source of energy 0.23 0.23 Average share of households at the locality living in flat/apartment -0.22 -0.22 Average share of households at the locality using electricity as main source of energy 0.39 0.39 Average share of households at the locality having solar boiler 0.14 0.14 Average share of households at the locality using bottled water as main source 1.70 1.70 Average share of households at the locality with a head having a chronic diSAEse -0.56 -0.56 Constant 8.72 8.72 RSME 0.38 0.37 Source: PECS 2016, authors’ calculations.

24

Figure A2. Governorate coefficients of variation for direct PECS 2016 estimates and for small area estimates 50%

45% PECS 2016 ELL, Census 2017

40%

35%

30%

25%

20%

coefficient ofvariation coefficient 15%

10%

5%

0%

Source: PECS 2016, authors’ calculations.

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1502810 1502810 Deir Samit 8071 Hebron 1448 17.94 4.4 27.49 2703430 2703430 'Abasan al Jadida 9220 Khan Yunis 4852 52.62 8.2 29.72 2703470 2703470 Khuza'a 11302 Khan Yunis 5299 46.88 6.9 28.80 1301535 1301471 'Abud 2140 Ramallah & Al Bi 285 13.3 2.3 29.13 1301545 1301500 Khirbet Abu Falah 4367 Ramallah & Al Bi 552 12.65 2.1 32.04 1301555 1301610 Al Mazra'a ash Sharqiya 4038 Ramallah & Al Bi 559 13.85 2.9 34.14 1301560 1301620 Deir Nidham 871 Ramallah & Al Bi 86 9.9 1.5 30.97 1301590 1301650 Kafr Malik 2928 Ramallah & Al Bi 576 19.66 3 33.11 1301605 1301715 Qibya 6053 Ramallah & Al Bi 955 15.78 3 30.03 1301610 1301610 Silwad 6303 Ramallah & Al Bi 873 13.85 2.9 34.14 1502725 1502635 Wadi ar Rim 223 Hebron 73 32.71 6.8 28.84 1502730 1502810 Suba 138 Hebron 25 17.94 4.4 27.49 1502765 1502810 Beit Maqdum 1103 Hebron 198 17.94 4.4 27.49 1502770 1502810 El Kaum 1455 Hebron 261 17.94 4.4 27.49 1502781 1502680 Al Bouaierah (Aqabat Injeleh) 1524 Hebron 404 26.51 3 28.55 1502782 1502680 Khallet Edar 2927 Hebron 776 26.51 3 28.55 1502785 1502810 Humsa 64 Hebron 11 17.94 4.4 27.49

25

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1502795 1502810 Al Muwarraq 835 Hebron 150 17.94 4.4 27.49 1502800 1502810 Tarusa 138 Hebron 25 17.94 4.4 27.49 1502880 1502680 Birin 149 Hebron 40 26.51 3 28.55 1502830 1502635 Masafer Bani Na'im Khallet Al Masafer 564 Hebron 184 32.71 6.8 28.84 1502881 1502680 Al'en 156 Hebron 41 26.51 3 28.55 1502835 1502835 Beit 'Awwa 10381 Hebron 1628 15.68 3.6 27.84 1502840 1502840 Dura 39128 Hebron 5058 12.93 3.5 27.66 1452185 1452275 'Ayda Camp 2794 Bethlehem 289 10.34 2.7 28.16 1452190 1452360 Khallet an Nu'man 111 Bethlehem 9 8.39 2.3 29.33 1452195 1452275 Al 'Aza Camp 1507 Bethlehem 156 10.34 2.7 28.16 1452200 1452360 Al Khas 394 Bethlehem 33 8.39 2.3 29.33 1452205 1452360 Al Haddadiya 81 Bethlehem 7 8.39 2.3 29.33 1452225 1452360 Dar Salah 4534 Bethlehem 380 8.39 2.3 29.33 1452230 1452230 Husan 6973 Bethlehem 750 10.75 3.1 28.51 1452255 1452255 Beit Sahur 12998 Bethlehem 177 1.36 0.7 31.57 1452270 1452300 Al Khadr 11833 Bethlehem 1194 10.09 3.3 29.63 1452265 1452265 Ad Doha 12611 Bethlehem 841 6.67 1.9 30.14 1452275 1452275 Ad Duheisha Camp 8705 Bethlehem 900 10.34 2.7 28.16 1452280 1452385 Hindaza and Bureid'a 7437 Bethlehem 807 10.85 2.5 29.51 1251300 1251310 Marda 2313 Salfit 493 21.29 2.9 29.91 1251320 1251360 Mas-ha 2308 Salfit 316 13.68 2.3 30.02 1251440 1251370 Khirbet Qeis 266 Salfit 32 11.9 2.2 29.21 1200945 1200970 Jit 2405 Qalqilia 344 14.31 1.7 29.28 1452285 1452360 Ash Shawawra 4117 Bethlehem 345 8.39 2.3 29.33 1452300 1452300 Artas 5679 Bethlehem 573 10.09 3.3 29.63 1452325 1452325 Nahhalin 8648 Bethlehem 585 6.77 2.2 27.85 1452335 1452385 Beit Ta'mir 1579 Bethlehem 171 10.85 2.5 29.51 1452345 1452385 Khallet al Louza 638 Bethlehem 69 10.85 2.5 29.51 1452355 1452325 Al Jab'a 1109 Bethlehem 75 6.77 2.2 27.85 1452360 1452360 Za'tara 7761 Bethlehem 651 8.39 2.3 29.33 1452375 1452385 Al Fureidis 1083 Bethlehem 117 10.85 2.5 29.51 1452385 1452385 Jannatah (Badd Falouh) 7258 Bethlehem 787 10.85 2.5 29.51 1452400 1452385 Wadi Rahhal 1800 Bethlehem 195 10.85 2.5 29.51 1452405 1452385 Jub adh Dhib 142 Bethlehem 15 10.85 2.5 29.51 1452415 1452325 Khallet Sakariya 140 Bethlehem 9 6.77 2.2 27.85 1452430 1452495 Khallet al Haddad 502 Bethlehem 70 14.02 2.8 29.69 1452440 1452495 Al Ma'sara 1073 Bethlehem 150 14.02 2.8 29.69 1452445 1452495 Wadi an Nis 990 Bethlehem 139 14.02 2.8 29.69 1200965 1200970 Baqat al Hatab 1943 Qalqilia 278 14.31 1.7 29.28 1200970 1200970 Hajja 2659 Qalqilia 380 14.31 1.7 29.28

26

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1201015 1200970 Far'ata 872 Qalqilia 125 14.31 1.7 29.28 2653200 2653200 Deir al Balah Camp 6914 Deir Al-Balah 4321 62.5 8.1 27.09 2653275 2653275 Wadi as Salqa 6647 Deir Al-Balah 3897 58.63 7.1 28.92 1150935 1150955 Deir al Hatab 2834 Nablus 550 19.42 2.9 28.55 1150955 1150955 Salim 6257 Nablus 1215 19.42 2.9 28.55 1502855 1502680 Qalqas 1700 Hebron 451 26.51 3 28.55 1502860 1502835 Sikka 909 Hebron 143 15.68 3.6 27.84 1502870 1502950 Wadi 'Ubeid 971 Hebron 215 22.14 3.5 28.99 1502890 1502835 Tawas 203 Hebron 32 15.68 3.6 27.84 1502895 1502950 Khursa 3463 Hebron 767 22.14 3.5 28.99 1502900 1502950 Tarrama 639 Hebron 141 22.14 3.5 28.99 1502905 1502530 Al Fawwar Camp 7601 Hebron 1252 16.47 3.2 26.25 1251275 1251305 Qarawat Bani Hassan 5369 Salfit 460 8.57 2.1 28.85 1251290 1251310 Qira 1245 Salfit 265 21.29 2.9 29.91 1251295 1251310 Kifl Haris 3977 Salfit 847 21.29 2.9 29.91 1502910 1502835 Al Majd 2265 Hebron 355 15.68 3.6 27.84 2552681 2552695 Um Al-Nnaser (Al Qaraya al Badawiya) 4679 North Gaza 2593 55.41 5.5 29.37 1502915 1502950 Marah al Baqqar 297 Hebron 66 22.14 3.5 28.99 1502920 1502950 Hadab al Fawwar 2342 Hebron 518 22.14 3.5 28.99 2653065 2653065 An Nuseirat Camp 31425 Deir Al-Balah 17891 56.93 4.7 27.81 2703420 2703420 Khan Yunis 202394 Khan Yunis 114618 56.63 4.8 30.27 2703425 2703425 Bani Suheila 41126 Khan Yunis 22814 55.47 8.7 29.15 2703445 2703445 'Abasan al Kabira 26565 Khan Yunis 12659 47.65 6.9 31.16 2703485 2703485 Al Fukhkhari 6394 Khan Yunis 3164 49.48 9 29.35 2753505 2753505 Al Shokat 16290 Rafah 9220 56.6 7.9 26.72 2753500 2753500 Al-Nnaser 8899 Rafah 5426 60.97 6.8 28.74 2753495 2753495 Rafah Camp 36206 Rafah 21625 59.73 7.4 27.11 2753490 2753490 Rafah 171571 Rafah 93468 54.48 3.9 29.23 2653140 2653140 Al Bureij Camp 27723 Deir Al-Balah 17255 62.24 6.2 27.93 2653145 2653145 Al Bureij 15329 Deir Al-Balah 8214 53.58 6.9 28.92 2653180 2653180 Az Zawayda 23518 Deir Al-Balah 12836 54.58 7.3 30.69 2653210 2653210 Al Maghazi Camp 17965 Deir Al-Balah 10525 58.59 7.5 28.14 1452390 1452325 Khallet al Balluta 73 Bethlehem 5 6.77 2.2 27.85 2653215 2653215 Al Maghazi 9564 Deir Al-Balah 4985 52.13 5.5 29.02 2653240 2653240 Deir al Balah 73483 Deir Al-Balah 42480 57.81 5 30.43 2653250 2653215 Al Musaddar 2561 Deir Al-Balah 1335 52.13 5.5 29.02 2703370 2703370 Al Qarara 28785 Khan Yunis 13150 45.68 8.1 29.93 2703410 2703410 Khan Yunis Camp 40871 Khan Yunis 22051 53.95 6.1 28.47 1050420 1050550 Bardala 1584 Tubas & Northen 376 23.76 3.5 29.40 1502925 1502835 Deir al 'Asal at Tahta 613 Hebron 96 15.68 3.6 27.84

27

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1502940 1502950 Wadi ash Shajina 758 Hebron 168 22.14 3.5 28.99 1502950 1502950 As Sura 3920 Hebron 868 22.14 3.5 28.99 1050455 1050550 Kardala 200 Tubas & Northen 48 23.76 3.5 29.40 1301595 1301595 Shuqba 5425 Ramallah & Al Bi 930 17.14 2.8 28.83 1502435 1502450 Khirbet ad Deir 356 Hebron 65 18.27 5.6 26.31 1502450 1502450 Surif 17196 Hebron 3141 18.27 5.6 26.31 1502530 1502530 Al 'Arrub Camp 8890 Hebron 1464 16.47 3.2 26.25 1502545 1502540 Jala 370 Hebron 151 40.92 8.8 27.30 1502560 1502560 Kharas 9085 Hebron 1472 16.2 3.8 27.35 1502750 1502750 Taffuh 15695 Hebron 2212 14.09 5.2 25.78 1502655 1502640 Beit Kahil 8827 Hebron 1806 20.46 4.2 27.68 1010005 1010045 Zububa 2277 Jenin 178 7.83 1.5 29.09 1201020 1200970 Immatin 2755 Qalqilia 394 14.31 1.7 29.28 1201035 1200970 Al Funduq 1125 Qalqilia 161 14.31 1.7 29.28 1150825 1151000 An Nassariya 1886 Nablus 405 21.5 2.5 31.80 1150840 1151000 Al 'Aqrabaniya 938 Nablus 202 21.5 2.5 31.80 1150855 1150990 Qusin 2240 Nablus 430 19.19 2.9 29.36 1150860 1150860 Beit Iba 4070 Nablus 489 12 1.9 29.85 1150810 1150860 Deir Sharaf 2945 Nablus 354 12 1.9 29.85 1150865 1151000 Beit Hasan 1597 Nablus 343 21.5 2.5 31.80 1150885 1151000 'Ein Shibli 313 Nablus 67 21.5 2.5 31.80 1150910 1150955 'Azmut 3435 Nablus 667 19.42 2.9 28.55 1150920 1150920 Nablus 155632 Nablus 17352 11.15 1.8 30.29 2552695 2552695 Beit Lahiya 88671 North Gaza 49131 55.41 5.5 29.37 2552740 2552740 Beit Hanun 51591 North Gaza 24821 48.11 7 28.77 2552755 2552755 Jabalya Camp 48855 North Gaza 29464 60.31 6.7 28.72 2552790 2552790 Jabalya 170208 North Gaza 80423 47.25 4.8 30.22 1201065 1201100 Kafr Laqif 1039 Qalqilia 195 18.73 2.5 29.92 1201085 1200970 Jinsafut 2571 Qalqilia 368 14.31 1.7 29.28 2602775 2602775 Ash Shati' Camp 40546 Gaza 25316 62.44 6.1 28.85 2602825 2602825 Gaza 578882 Gaza 295002 50.96 3.1 33.85 2602900 2602945 Madinat Ezahra 5274 Gaza 2509 47.56 4.5 33.52 2602945 2602945 Al Mughraqa 11405 Gaza 5425 47.56 4.5 33.52 2603045 2602945 Juhor ad Dik 4565 Gaza 2171 47.56 4.5 33.52 1010465 1010465 Kafr Ra'i 8306 Jenin 698 8.41 1.9 28.54 1502780 1502780 Hebron (Al Khalil) 199402 Hebron 28144 14.11 3.2 27.58 2653070 2653070 An Nuseirat 54287 Deir Al-Balah 26085 48.05 4.9 29.44 1452180 1452180 Al 'Ubeidiya 14303 Bethlehem 1643 11.49 3.5 28.35 1150660 1150680 Bazzariya 2790 Nablus 390 13.97 1.9 29.69 1150680 1150680 Burqa 4146 Nablus 579 13.97 1.9 29.69

28

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1201190 1201175 'Izbat Jal'ud 146 Qalqilia 18 12.6 2.3 30.16 1201205 1201175 Al Mudawwar 350 Qalqilia 44 12.6 2.3 30.16 1201210 1201175 'Izbat Salman 896 Qalqilia 113 12.6 2.3 30.16 1201225 1201175 'Izbat al Ashqar 451 Qalqilia 57 12.6 2.3 30.16 1201255 1201175 Beit Amin 1279 Qalqilia 161 12.6 2.3 30.16 1201260 1201175 Sanniriya 3609 Qalqilia 455 12.6 2.3 30.16 1201280 1201175 'Azzun 'Atma 2068 Qalqilia 261 12.6 2.3 30.16 1301460 1301485 Mazari' an Nubani 2421 Ramallah & Al Bi 367 15.16 2.1 29.72 1351690 1351690 Al 'Auja 5138 Ariha & Al Aghwa 1141 22.2 3 31.22 1351510 1351690 Fasayil 1637 Ariha & Al Aghwa 363 22.2 3 31.22 1351116 1351690 Marj al Ghazal 238 Ariha & Al Aghwa 53 22.2 3 31.22 1150765 1150680 Sabastiya 3187 Nablus 445 13.97 1.9 29.69 1150770 1150680 Ijnisinya 584 Nablus 82 13.97 1.9 29.69 1150775 1150820 Talluza 2791 Nablus 324 11.62 2.4 29.50 1150785 1150860 An Naqura 1783 Nablus 214 12 1.9 29.85 1150805 1151000 Al Badhan 3166 Nablus 681 21.5 2.5 31.80 1251250 1251310 Deir Istiya 3599 Salfit 766 21.29 2.9 29.91 1452170 1452230 Al Walaja 2643 Bethlehem 284 10.75 3.1 28.51 1301475 1301485 'Arura 3088 Ramallah & Al Bi 468 15.16 2.1 29.72 1301505 1301485 Deir as Sudan 2146 Ramallah & Al Bi 325 15.16 2.1 29.72 1301520 1301485 Jilijliya 628 Ramallah & Al Bi 95 15.16 2.1 29.72 1301525 1301485 'Ajjul 1393 Ramallah & Al Bi 211 15.16 2.1 29.72 1301530 1301500 Al Mughayyir 2854 Ramallah & Al Bi 361 12.65 2.1 32.04 1301540 1301471 An Nabi Salih 519 Ramallah & Al Bi 69 13.3 2.3 29.13 1301546 1301635 Rawabi 706 Ramallah & Al Bi 41 5.75 1.3 34.79 1301550 1301695 Umm Safa 677 Ramallah & Al Bi 76 11.23 2.2 30.59 1452175 1452230 Battir 4646 Bethlehem 499 10.75 3.1 28.51 1502540 1502540 Beit Ummar 16887 Hebron 6910 40.92 8.8 27.30 1502550 1502560 Hitta 1149 Hebron 186 16.2 3.8 27.35 1502555 1502680 Shuyukh al 'Arrub 1948 Hebron 517 26.51 3 28.55 1502735 1502635 Qinan an Namir 158 Hebron 52 32.71 6.8 28.84 1502575 1502680 Umm al Butm 82 Hebron 22 26.51 3 28.55 1502580 1502680 Hamrush 53 Hebron 14 26.51 3 28.55 1301575 1301695 Jibiya 155 Ramallah & Al Bi 17 11.23 2.2 30.59 1301585 1301695 Burham 579 Ramallah & Al Bi 65 11.23 2.2 30.59 1301600 1301695 Kobar 4293 Ramallah & Al Bi 482 11.23 2.2 30.59 1010025 1010060 'Arabbuna 1005 Jenin 79 7.88 1.4 28.70 1010030 1010060 Al Jalama 2224 Jenin 175 7.88 1.4 28.70 1010035 1010035 Silat al Harithiya 11226 Jenin 787 7.01 2.4 28.80 1010040 1010045 As Sa'aida 85 Jenin 7 7.83 1.5 29.09

29

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1010045 1010045 'Anin 4134 Jenin 324 7.83 1.5 29.09 1502585 1502560 Nuba 5601 Hebron 908 16.2 3.8 27.35 1502595 1502680 Kuziba 1376 Hebron 365 26.51 3 28.55 1502615 1502615 Beit Ula 14454 Hebron 2754 19.05 5.7 27.43 1502620 1502620 Sa'ir 20607 Hebron 8164 39.62 7.9 26.69 1251400 1251425 Bruqin 3941 Salfit 619 15.71 2.3 29.36 1251415 1251370 Farkha 1607 Salfit 191 11.9 2.2 29.21 1251425 1251425 Kafr ad Dik 5406 Salfit 849 15.71 2.3 29.36 1502630 1502630 Halhul 26888 Hebron 5222 19.42 6.1 27.09 1502635 1502635 Ash Shuyukh 11988 Hebron 3921 32.71 6.8 28.84 1502640 1502640 Tarqumiya 19205 Hebron 3930 20.46 4.2 27.68 1502680 1502680 Beit 'Einun 1918 Hebron 509 26.51 3 28.55 1502681 1502680 Qla’a Zeta 1064 Hebron 282 26.51 3 28.55 1502778 1502680 Al Bouaierah (Al Baq'a) 1382 Hebron 366 26.51 3 28.55 1502685 1502685 Idhna 25854 Hebron 3676 14.22 3.8 26.05 1251305 1251305 Biddya 10177 Salfit 872 8.57 2.1 28.85 1251310 1251310 Haris 4029 Salfit 858 21.29 2.9 29.91 1251330 1251370 Iskaka 1167 Salfit 139 11.9 2.2 29.21 1251340 1251360 Sarta 3293 Salfit 450 13.68 2.3 30.02 1251355 1251360 Izbat Abu 'Adam 7 Salfit 1 13.68 2.3 30.02 1251360 1251360 Az Zawiya 5875 Salfit 803 13.68 2.3 30.02 1251370 1251370 Salfit 10890 Salfit 1296 11.9 2.2 29.21 1251395 1251360 Rafat 2456 Salfit 336 13.68 2.3 30.02 1301490 1301500 Turmus'ayya 2449 Ramallah & Al Bi 310 12.65 2.1 32.04 1452235 1452325 Wadi Fukin 1328 Bethlehem 90 6.77 2.2 27.85 1301495 1301595 Al Lubban al Gharbi 1556 Ramallah & Al Bi 267 17.14 2.8 28.83 1301500 1301500 Sinjil 5707 Ramallah & Al Bi 722 12.65 2.1 32.04 1251430 1251425 Deir Ballut 3772 Salfit 592 15.71 2.3 29.36 1200905 1201100 Falamya 757 Qalqilia 142 18.73 2.5 29.92 1200925 1200970 Kafr Qaddum 3280 Qalqilia 469 14.31 1.7 29.28 1200985 1201100 Jayyus 3478 Qalqilia 652 18.73 2.5 29.92 1200995 1201100 Khirbet Sir 645 Qalqilia 121 18.73 2.5 29.92 1301472 1301471 Deir Ghassana 1682 Ramallah & Al Bi 224 13.3 2.3 29.13 1301485 1301485 'Abwein 3474 Ramallah & Al Bi 527 15.16 2.1 29.72 1301515 1301595 Rantis 3159 Ramallah & Al Bi 541 17.14 2.8 28.83 1050580 1050550 Al Malih 349 Tubas & Northen 83 23.76 3.5 29.40 1050656 1050755 Khirbet Yarza 31 Tubas & Northen 4 13.12 2.1 28.69 1503075 1503335 Beit Mirsim 415 Hebron 123 29.68 5 28.57 1351045 1351690 Marj Na'ja 828 Ariha & Al Aghwa 184 22.2 3 31.22 1351110 1351690 Az Zubeidat 1679 Ariha & Al Aghwa 373 22.2 3 31.22

30

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1351140 1351690 Al Jiftlik 3046 Ariha & Al Aghwa 676 22.2 3 31.22 1010050 1010060 'Arrana 2371 Jenin 187 7.88 1.4 28.70 1010055 1010060 Deir Ghazala 1107 Jenin 87 7.88 1.4 28.70 1010070 1010045 Khirbet Suruj 34 Jenin 3 7.83 1.5 29.09 1150695 1150820 Yasid 2501 Nablus 291 11.62 2.4 29.50 1150705 1150680 Beit Imrin 3318 Nablus 464 13.97 1.9 29.69 1150745 1150680 Nisf Jubeil 470 Nablus 66 13.97 1.9 29.69 1150820 1150820 'Asira ash Shamaliya 8795 Nablus 1022 11.62 2.4 29.50 1010060 1010060 Faqqu'a 4324 Jenin 341 7.88 1.4 28.70 1150835 1150860 Zawata 2533 Nablus 304 12 1.9 29.85 1150875 1150860 Beit Wazan 1310 Nablus 157 12 1.9 29.85 1150880 1150990 'Ein Beit el Ma Camp 3583 Nablus 688 19.19 2.9 29.36 1150930 1150930 'Askar Camp (al Qadeem) 6528 Nablus 2790 42.73 6.7 27.02 1150931 1150930 'Askar Camp (al Jadeed) 4760 Nablus 2034 42.73 6.7 27.02 1150960 1150960 Balata Camp 14598 Nablus 6481 44.39 7.1 27.19 1150950 1150990 Sarra 3368 Nablus 646 19.19 2.9 29.36 1010010 1010045 Rummana 3566 Jenin 279 7.83 1.5 29.09 1010015 1010045 Ti'innik 1273 Jenin 100 7.83 1.5 29.09 1010020 1010045 At Tayba 2160 Jenin 169 7.83 1.5 29.09 1100250 1100290 'Akkaba 328 Tulkarm 48 14.55 2.5 28.05 1150975 1150990 'Iraq Burin 1007 Nablus 193 19.19 2.9 29.36 1150990 1150990 Tell 5149 Nablus 988 19.19 2.9 29.36 1151050 1151010 Madama 2089 Nablus 347 16.63 2.4 29.69 1151095 1151010 'Asira al Qibliya 2931 Nablus 487 16.63 2.4 29.69 1452209 1452210 Bir Onah 1386 Bethlehem 31 2.25 0.9 32.55 1151160 1151245 'Urif 3619 Nablus 596 16.46 2.8 28.90 1151176 1151090 Khirbet Tana 16 Nablus 2 15.13 4.2 27.50 1151230 1151245 Zeita Jamma'in 2736 Nablus 450 16.46 2.8 28.90 1151200 1151270 Yanun 92 Nablus 23 24.56 4.2 29.03 1452210 1452210 Beit Jala 12673 Bethlehem 285 2.25 0.9 32.55 1502815 1502815 Bani Na'im 24498 Hebron 7835 31.98 8.2 27.75 1503400 1503335 Khirbet ar Rahwa 51 Hebron 15 29.68 5 28.57 1151311 1151270 Alttawel and Tall al Khashaba 107 Nablus 26 24.56 4.2 29.03 1151385 1151270 Majdal Bani Fadil 2903 Nablus 713 24.56 4.2 29.03 1100870 1100900 Kur 285 Tulkarm 26 8.97 1.8 28.70 1100915 1100900 Kafr 'Abbush 1696 Tulkarm 152 8.97 1.8 28.70 1100895 1100900 Kafr Zibad 1189 Tulkarm 107 8.97 1.8 28.70 1100900 1100900 Kafr Jammal 2784 Tulkarm 250 8.97 1.8 28.70 1100795 1100800 Saffarin 735 Tulkarm 95 12.99 2.4 28.17 1100730 1100665 Ramin 1949 Tulkarm 263 13.49 2.7 29.18

31

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1100800 1100800 Beit Lid 5467 Tulkarm 710 12.99 2.4 28.17 1151445 1151365 Duma 2670 Nablus 497 18.61 2.7 29.21 1151450 1151365 Khirbet Sarra 27 Nablus 5 18.61 2.7 29.21 1100690 1100665 Kafr al Labad 4629 Tulkarm 624 13.49 2.7 29.18 1100665 1100665 Anabta 7877 Tulkarm 1063 13.49 2.7 29.18 1100620 1100635 Nur Shams Camp 6264 Tulkarm 2094 33.43 5.4 26.44 1100645 1100645 Tulkarm 63926 Tulkarm 9323 14.58 2.9 30.09 1503281 1503115 Almefqara 73 Hebron 35 48.1 3.9 30.65 1010215 1010215 Deir Abu Da'if 6908 Jenin 932 13.49 3.6 29.33 1010265 1010265 Ya'bad 15700 Jenin 1941 12.36 2.6 28.45 1010285 1010265 Imreiha 323 Jenin 40 12.36 2.6 28.45 1010286 1010265 Firasin 26 Jenin 3 12.36 2.6 28.45 1010287 1010265 Kherbet Al Hamam 20 Jenin 2 12.36 2.6 28.45 1010300 1010220 Ash Shuhada (Mothalth Ash Shuhada) 2244 Jenin 303 13.51 2.8 28.71 1010310 1010405 Al Mughayyir 3186 Jenin 583 18.29 2.4 29.24 1010315 1010405 Al Mutilla 314 Jenin 57 18.29 2.4 29.24 1010320 1010405 Bir al Basha 1691 Jenin 309 18.29 2.4 29.24 1010325 1010405 Tannin 41 Jenin 7 18.29 2.4 29.24 1010335 1010405 Al Hafira (Hafirat Arraba) 86 Jenin 16 18.29 2.4 29.24 1010370 1010370 Arraba 11255 Jenin 1249 11.1 2.5 28.47 1010385 1010405 Telfit 179 Jenin 33 18.29 2.4 29.24 1010395 1010405 Mirka 2160 Jenin 395 18.29 2.4 29.24 1010435 1010625 Az Zababida 4173 Jenin 514 12.31 2.3 31.09 1010445 1010370 Fahma 3131 Jenin 347 11.1 2.5 28.47 1010460 1010370 Az Zawiya 986 Jenin 109 11.1 2.5 28.47 1010485 1010600 Al Kufeir 56 Jenin 10 17.63 3.3 28.77 1010495 1010600 Sir 840 Jenin 148 17.63 3.3 28.77 1010500 1010465 'Ajja 6042 Jenin 508 8.41 1.9 28.54 1010510 1010600 Sanur 4938 Jenin 870 17.63 3.3 28.77 1010515 1010465 Ar Rama 1198 Jenin 101 8.41 1.9 28.54 1010565 1010520 Al Judeida 5834 Jenin 746 12.79 2.6 29.01 1100345 1100290 An Nazla ash Sharqiya 1583 Tulkarm 230 14.55 2.5 28.05 1301615 1301610 Yabrud 571 Ramallah & Al Bi 79 13.85 2.9 34.14 1301625 1301805 Shabtin 1129 Ramallah & Al Bi 132 11.66 2 29.15 1251315 1251370 Yasuf 2038 Salfit 243 11.9 2.2 29.21 1151180 1151010 Odala 1564 Nablus 260 16.63 2.4 29.69 1151185 1151185 Huwwara 6650 Nablus 1371 20.61 3.6 29.64 1151245 1151245 Jamma'in 7425 Nablus 1222 16.46 2.8 28.90 1151215 1151215 Beita 11665 Nablus 2413 20.68 4.7 28.42 1151285 1151185 Za'tara 63 Nablus 13 20.61 3.6 29.64

32

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1010210 1010095 Kafr Qud 1523 Jenin 125 8.24 2 28.85 1010220 1010220 Birqin 6987 Jenin 944 13.51 2.8 28.71 1301636 1301700 Ad Doha 208 Ramallah & Al Bi 22 10.58 2.9 34.59 1301640 1301700 'Ein Siniya 919 Ramallah & Al Bi 97 10.58 2.9 34.59 1301670 1301695 Abu Shukheidim 2423 Ramallah & Al Bi 272 11.23 2.2 30.59 1301745 1301715 Al Midya 1524 Ramallah & Al Bi 240 15.78 3 30.03 1010295 1010215 Umm at Tut 1171 Jenin 158 13.49 3.6 29.33 1010305 1010405 Jalqamus 2573 Jenin 471 18.29 2.4 29.24 1010365 1010405 Ad Damayra 242 Jenin 44 18.29 2.4 29.24 1502960 1503215 Ar Rihiya 5724 Hebron 1794 31.35 5.5 26.53 1502965 1502680 Zif 1055 Hebron 280 26.51 3 28.55 1502980 1502950 Imreish 2196 Hebron 486 22.14 3.5 28.99 1502970 1502835 Deir al 'Asal al Fauqa 1849 Hebron 290 15.68 3.6 27.84 1100350 1100350 Baqa ash Sharqiya 4771 Tulkarm 599 12.56 2.7 29.41 1100355 1100290 An Nazla al Wusta 419 Tulkarm 61 14.55 2.5 28.05 1100380 1100290 An Nazla al Gharbiya 1083 Tulkarm 158 14.55 2.5 28.05 1100425 1100350 Zeita 3002 Tulkarm 377 12.56 2.7 29.41 1100440 1100475 Seida 3683 Tulkarm 544 14.78 2.9 28.43 1010400 1010405 Wadi Du'oq 167 Jenin 31 18.29 2.4 29.24 1010401 1010405 Fahma al Jadida 380 Jenin 69 18.29 2.4 29.24 1010405 1010405 Raba 3839 Jenin 702 18.29 2.4 29.24 1010410 1010405 Al Mansura 218 Jenin 40 18.29 2.4 29.24 1010415 1010625 Misliya 2828 Jenin 348 12.31 2.3 31.09 1010430 1010405 Al Jarba 62 Jenin 11 18.29 2.4 29.24 1100570 1100570 Bal'a 7628 Tulkarm 1021 13.39 3.3 29.76 1100595 1100570 Iktaba 2923 Tulkarm 391 13.39 3.3 29.76 1503170 1503335 Al Burj and Al Bira 3188 Hebron 946 29.68 5 28.57 1503210 1503115 Umm al Khair 682 Hebron 328 48.1 3.9 30.65 1010505 1010465 'Anza 1900 Jenin 160 8.41 1.9 28.54 1010520 1010520 Meithalun 8159 Jenin 1044 12.79 2.6 29.01 1151195 1151215 'Einabus 2887 Nablus 597 20.68 4.7 28.42 1010125 1010095 Al 'Araqa 2615 Jenin 215 8.24 2 28.85 1010155 1010095 Al Hashimiya 1280 Jenin 105 8.24 2 28.85 1010275 1010220 Kufeirit 3008 Jenin 406 13.51 2.8 28.71 1503211 1503115 Sadit athaleh 45 Hebron 22 48.1 3.9 30.65 1503215 1503215 Al Karmil 9688 Hebron 3037 31.35 5.5 26.53 1503217 1503115 Majd AlBa' 114 Hebron 55 48.1 3.9 30.65 1503220 1503115 Qinan Jaber 347 Hebron 167 48.1 3.9 30.65 1503235 1503335 Somara 243 Hebron 72 29.68 5 28.57 1503245 1503245 Adh Dhahiriya 35667 Hebron 3564 9.99 2.9 26.82

33

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1503246 1503335 Iqtet 8 Hebron 2 29.68 5 28.57 1503248 1503115 Kafr Jul 19 Hebron 9 48.1 3.9 30.65 1503255 1503115 At Tuwani 193 Hebron 93 48.1 3.9 30.65 1503256 1503115 Sosya 198 Hebron 95 48.1 3.9 30.65 1503265 1503115 An Najada 447 Hebron 215 48.1 3.9 30.65 1503270 1503335 Khirbet Deir Shams 80 Hebron 24 29.68 5 28.57 1503280 1503115 Ar Rakeez 15 Hebron 7 48.1 3.9 30.65 1503285 1503335 Khirbet Shuweika 281 Hebron 83 29.68 5 28.57 1503295 1503335 'Anab al Kabir 461 Hebron 137 29.68 5 28.57 1100710 1100800 Kafa 808 Tulkarm 105 12.99 2.4 28.17 1100655 1100665 'Izbat Abu Khameish 54 Tulkarm 7 13.49 2.7 29.18 1100685 1100665 'Izbat al Khilal 137 Tulkarm 18 13.49 2.7 29.18 1100715 1100665 Al Haffasi 206 Tulkarm 28 13.49 2.7 29.18 1100815 1100900 Ar Ras 634 Tulkarm 57 8.97 1.8 28.70 1100780 1100800 Khirbet Jubara 305 Tulkarm 40 12.99 2.4 28.17 1100845 1100900 Kafr Sur 1256 Tulkarm 113 8.97 1.8 28.70 1100760 1100800 Shufa 1317 Tulkarm 171 12.99 2.4 28.17 1503300 1503115 Khirbet Asafi (Al Fauqa and Al Tahta) 73 Hebron 35 48.1 3.9 30.65 1503310 1503115 Shi'b al Batim 176 Hebron 85 48.1 3.9 30.65 1503315 1503115 Qawawis 31 Hebron 15 48.1 3.9 30.65 1301620 1301620 Beitillu 3444 Ramallah & Al Bi 341 9.9 1.5 30.97 1301630 1301620 Jammala 1654 Ramallah & Al Bi 164 9.9 1.5 30.97 1301635 1301635 Birzeit 5825 Ramallah & Al Bi 335 5.75 1.3 34.79 1301645 1301610 Silwad Camp 453 Ramallah & Al Bi 63 13.85 2.9 34.14 1301650 1301650 Deir Jarir 4441 Ramallah & Al Bi 873 19.66 3 33.11 1100725 1100800 'Izbat Shufa 1420 Tulkarm 184 12.99 2.4 28.17 1100735 1100800 Far'un 4028 Tulkarm 523 12.99 2.4 28.17 1100635 1100635 Tulkarm Camp 9685 Tulkarm 3238 33.43 5.4 26.44 1301655 1301620 Deir 'Ammar 3332 Ramallah & Al Bi 330 9.9 1.5 30.97 1301660 1301620 Deir 'Ammar Camp 1872 Ramallah & Al Bi 185 9.9 1.5 30.97 1301665 1301715 Budrus 1586 Ramallah & Al Bi 250 15.78 3 30.03 1301675 1301635 Jifna 2828 Ramallah & Al Bi 162 5.75 1.3 34.79 1301680 1301700 Dura al Qar' 3013 Ramallah & Al Bi 319 10.58 2.9 34.59 1301685 1301650 At Tayba 1304 Ramallah & Al Bi 256 19.66 3 33.11 1301695 1301695 Al Mazra'a al Qibliya 5144 Ramallah & Al Bi 577 11.23 2.2 30.59 1301700 1301700 Al Jalazun Camp 8145 Ramallah & Al Bi 862 10.58 2.9 34.59 1301705 1301635 Abu Qash 2222 Ramallah & Al Bi 128 5.75 1.3 34.79 1301710 1301805 Deir Qaddis 2437 Ramallah & Al Bi 284 11.66 2 29.15 1301715 1301715 Ni'lin 5079 Ramallah & Al Bi 801 15.78 3 30.03 1301720 1301610 'Ein Yabrud 2500 Ramallah & Al Bi 346 13.85 2.9 34.14

34

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1301725 1301805 Kharbatha Bani Harith 3450 Ramallah & Al Bi 402 11.66 2 29.15 1301730 1301620 Ras Karkar 1944 Ramallah & Al Bi 192 9.9 1.5 30.97 1301735 1301700 Surda 1282 Ramallah & Al Bi 136 10.58 2.9 34.59 1301740 1301620 Al Janiya 1288 Ramallah & Al Bi 128 9.9 1.5 30.97 1010085 1010120 Umm ar Rihan 438 Jenin 20 4.58 1.4 28.84 1010105 1010120 Khirbet 'Abdallah al Yunis 152 Jenin 7 4.58 1.4 28.84 1010110 1010060 Mashru' Beit Qad )Ash Shamali) 419 Jenin 33 7.88 1.4 28.70 1010115 1010120 Dhaher al Malih 191 Jenin 9 4.58 1.4 28.84 1010120 1010120 Barta'a ash Sharqiya 4556 Jenin 209 4.58 1.4 28.84 1010135 1010215 Al Jameelat 50 Jenin 7 13.49 3.6 29.33 1010140 1010060 Beit Qad)Al Janubi) 1508 Jenin 119 7.88 1.4 28.70 1010145 1010035 Tura al Gharbiya 1029 Jenin 72 7.01 2.4 28.80 1010150 1010035 Tura ash Sharqiya 228 Jenin 16 7.01 2.4 28.80 1010165 1010035 Nazlat ash Sheikh Zeid 827 Jenin 58 7.01 2.4 28.80 1010170 1010035 At Tarem 486 Jenin 34 7.01 2.4 28.80 1010175 1010120 Khirbet al Muntar al Gharbiya 37 Jenin 2 4.58 1.4 28.84 1010185 1010185 Jenin Camp 10210 Jenin 3248 31.82 7.1 27.16 1010190 1010215 Jalbun 2758 Jenin 372 13.49 3.6 29.33 1010195 1010215 'Aba (Al Gharbiya) 319 Jenin 43 13.49 3.6 29.33 1010205 1010120 Khirbet al Muntar ash Sharqiya 32 Jenin 1 4.58 1.4 28.84 1010225 1010120 Umm Dar 641 Jenin 29 4.58 1.4 28.84 1010230 1010120 Al Khuljan 604 Jenin 28 4.58 1.4 28.84 1010235 1010215 Wad ad Dabi') 'Aba ash Sharqiya) 560 Jenin 76 13.49 3.6 29.33 1010240 1010265 Dhaher al 'Abed 458 Jenin 57 12.36 2.6 28.45 1010245 1010265 Zabda 1227 Jenin 152 12.36 2.6 28.45 1100475 1100475 'Illar 7271 Tulkarm 1075 14.78 2.9 28.43 1100480 1100480 'Attil 10110 Tulkarm 1314 13 2.7 29.26 1100530 1100530 Deir al Ghusun 9667 Tulkarm 1405 14.54 2.8 28.91 1100545 1100530 Al Jarushiya 1154 Tulkarm 168 14.54 2.8 28.91 1100555 1100530 Al Masqufa 154 Tulkarm 22 14.54 2.8 28.91 1452475 1452495 Al Halqum 263 Bethlehem 37 14.02 2.8 29.69 1452480 1452525 Umm Salamuna 1175 Bethlehem 244 20.73 5.4 28.87 1452490 1452495 Al Manshiya 527 Bethlehem 74 14.02 2.8 29.69 1452510 1452495 Wadi Immhamid 148 Bethlehem 21 14.02 2.8 29.69 1452495 1452495 Tuqu' 8674 Bethlehem 1216 14.02 2.8 29.69 1201040 1201040 Qalqiliya 48177 Qalqilia 13180 27.36 5 29.44 1452455 1452495 Khirbet ad Deir 1988 Bethlehem 279 14.02 2.8 29.69 1452460 1452495 Jurat ash Sham'a 1759 Bethlehem 247 14.02 2.8 29.69 1452465 1452325 Khallet 'Afana 58 Bethlehem 4 6.77 2.2 27.85 1452470 1452495 Marah Ma'alla 1061 Bethlehem 149 14.02 2.8 29.69

35

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1201055 1201100 An Nabi Elyas 1399 Qalqilia 262 18.73 2.5 29.92 1201070 1201125 'Arab Abu Farda 131 Qalqilia 21 15.77 2.9 28.57 1201075 1201100 'Izbat at Tabib 258 Qalqilia 48 18.73 2.5 29.92 1201100 1201100 'Azzun 9269 Qalqilia 1736 18.73 2.5 29.92 1201105 1201125 'Arab ar Ramadin al Janubi 286 Qalqilia 45 15.77 2.9 28.57 1201115 1201175 'Isla 1111 Qalqilia 140 12.6 2.3 30.16 1201116 1201175 Arab Al-Khouleh 11 Qalqilia 1 12.6 2.3 30.16 1201120 1201125 Wadi ar Rasha 152 Qalqilia 24 15.77 2.9 28.57 1301565 1301635 'Atara 2477 Ramallah & Al Bi 142 5.75 1.3 34.79 1201125 1201125 Habla 7053 Qalqilia 1112 15.77 2.9 28.57 1301820 1301855 Beit 'Ur at Tahta 5009 Ramallah & Al Bi 572 11.43 2.8 28.60 1201130 1201125 Ras at Tira 484 Qalqilia 76 15.77 2.9 28.57 1201155 1201125 Ras 'Atiya 2129 Qalqilia 336 15.77 2.9 28.57 1201170 1201125 Ad Dab'a 412 Qalqilia 65 15.77 2.9 28.57 1201175 1201175 Kafr Thulth 4851 Qalqilia 611 12.6 2.3 30.16 1201005 1201125 'Arab ar Ramadin ash Shamali 84 Qalqilia 13 15.77 2.9 28.57 1301455 1301471 Qarawat Bani Zeid 3394 Ramallah & Al Bi 452 13.3 2.3 29.13 1301470 1301471 Kafr 'Ein 1946 Ramallah & Al Bi 259 13.3 2.3 29.13 1301471 1301471 Beit Reema 4308 Ramallah & Al Bi 573 13.3 2.3 29.13 1301895 1301895 Beit Liqya 8981 Ramallah & Al Bi 861 9.59 2.3 28.42 1351970 1351920 Deir al Qilt 1 Ariha & Al Aghwa 0 11.06 1.7 32.18 1351840 1351690 An Nuwei'ma 1651 Ariha & Al Aghwa 367 22.2 3 31.22 1351865 1351975 'Ein as Sultan Camp 4168 Ariha & Al Aghwa 632 15.17 3.2 29.02 1452500 1452525 Marah Rabah 1711 Bethlehem 355 20.73 5.4 28.87 1452520 1452495 Khirbet Tuqu' 130 Bethlehem 18 14.02 2.8 29.69 1452535 1452495 Al Maniya 1332 Bethlehem 187 14.02 2.8 29.69 1452565 1452495 Kisan 554 Bethlehem 78 14.02 2.8 29.69 1452660 1452495 'Arab ar Rashayida 2038 Bethlehem 286 14.02 2.8 29.69 1452525 1452525 Beit Fajjar 13376 Bethlehem 2773 20.73 5.4 28.87 1010080 1010080 Al Yamun 20369 Jenin 2566 12.6 4 28.59 1010095 1010095 Kafr Dan 6463 Jenin 532 8.24 2 28.85 1010585 1010520 Al 'Asa'asa 560 Jenin 72 12.79 2.6 29.01 1010605 1010605 Jaba' 10210 Jenin 1609 15.75 4.8 26.93 1010615 1010600 Al Fandaqumiya 4182 Jenin 737 17.63 3.3 28.77 1301850 1301895 Beit Sira 3307 Ramallah & Al Bi 317 9.59 2.3 28.42 1301855 1301855 Kharbatha al Misbah 6327 Ramallah & Al Bi 723 11.43 2.8 28.60 1503010 1502835 Beit ar Rush at Tahta 426 Hebron 67 15.68 3.6 27.84 1503090 1502835 Beit ar Rush al Fauqa 1378 Hebron 216 15.68 3.6 27.84 1503095 1502950 Karma 1772 Hebron 392 22.14 3.5 28.99 1503100 1503215 Beit 'Amra 3533 Hebron 1108 31.35 5.5 26.53

36

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1503105 1503115 Al Ka’abneh -Om Adaraj (Alzoyedeen) 1450 Hebron 697 48.1 3.9 30.65 1503005 1503115 Al Buweib 730 Hebron 351 48.1 3.9 30.65 1503006 1503115 Toba 76 Hebron 37 48.1 3.9 30.65 1503305 1503115 Al Maq'ora 45 Hebron 22 48.1 3.9 30.65 1503126 1503115 Khashem Adaraj (Al-Hathaleen) 984 Hebron 473 48.1 3.9 30.65 1503130 1503115 Khashem al Karem 150 Hebron 72 48.1 3.9 30.65 1503135 1503335 Kurza 938 Hebron 278 29.68 5 28.57 1503145 1503335 Rabud 2801 Hebron 831 29.68 5 28.57 1503150 1503115 Umm Lasafa and Abu Shabban 1633 Hebron 785 48.1 3.9 30.65 1503110 1503335 Wadi al Kilab 49 Hebron 15 29.68 5 28.57 1503111 1503115 Umm Ashoqhan 602 Hebron 290 48.1 3.9 30.65 1503115 1503115 Khallet al Maiyya 2138 Hebron 1028 48.1 3.9 30.65 1503117 1503115 Umm Al Amad (Sahel Wadi Elma) 398 Hebron 191 48.1 3.9 30.65 1503120 1503120 Yatta 63116 Hebron 14988 23.75 4.9 26.72 1503125 1503115 Ar Rifa'iyya and Ad Deirat 1305 Hebron 628 48.1 3.9 30.65 1352075 1351920 An Nabi Musa 343 Ariha & Al Aghwa 38 11.06 1.7 32.18 1351975 1351975 Aqbat Jaber Camp 8593 Ariha & Al Aghwa 1304 15.17 3.2 29.02 1502985 1502835 Iskeik 211 Hebron 33 15.68 3.6 27.84 1301810 1301810 Ramallah 34483 Ramallah & Al Bi 606 1.76 0.8 34.67 1301825 1301825 Beituniya 24905 Ramallah & Al Bi 1128 4.53 1.4 29.79 1351920 1351920 Jericho (Ariha) 18992 Ariha & Al Aghwa 2100 11.06 1.7 32.18 1351845 1351690 'Ein ad Duyuk al Fauqa 833 Ariha & Al Aghwa 185 22.2 3 31.22 1010180 1010180 Jenin 48688 Jenin 3274 6.72 2.2 30.26 1010590 1010625 Al 'Attara 1220 Jenin 150 12.31 2.3 31.09 1010600 1010600 Siris 5903 Jenin 1041 17.63 3.3 28.77 1010625 1010625 Silat adh Dhahr 7262 Jenin 894 12.31 2.3 31.09 1301830 1301830 Al Am'ari Camp 4690 Ramallah & Al Bi 1165 24.85 6.5 28.13 1301835 1301830 Qaddura Camp 856 Ramallah & Al Bi 213 24.85 6.5 28.13 1301860 1301855 Beit 'Ur al Fauqa 1033 Ramallah & Al Bi 118 11.43 2.8 28.60 1301890 1301855 At Tira 1488 Ramallah & Al Bi 170 11.43 2.8 28.60 1503320 1503320 As Samu' 25873 Hebron 7481 28.91 7.3 26.41 1503324 1503115 Khirbet Alrthem 16 Hebron 8 48.1 3.9 30.65 1503325 1503115 Edqeqa (Khirbet Tawil ash Shih) 299 Hebron 144 48.1 3.9 30.65 1503335 1503335 Ar Ramadin 4124 Hebron 1224 29.68 5 28.57 1503345 1503115 Maghayir al 'Abeed 16 Hebron 8 48.1 3.9 30.65 1503350 1503115 Khirbet al Fakheit 310 Hebron 149 48.1 3.9 30.65 1503360 1503115 Khirbet Bir al 'Idd 134 Hebron 64 48.1 3.9 30.65 1503365 1503115 Haribat an Nabi 41 Hebron 20 48.1 3.9 30.65 1503375 1503115 Khirbet Zanuta 130 Hebron 63 48.1 3.9 30.65 1503380 1503115 Imneizil 277 Hebron 133 48.1 3.9 30.65

37

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1503395 1503115 Khirbet Ghuwein al Fauqa 55 Hebron 26 48.1 3.9 30.65 1503405 1503335 'Arab al Fureijat 429 Hebron 127 29.68 5 28.57 1503385 1503115 Khirbet al Kharaba 17 Hebron 8 48.1 3.9 30.65 1301570 1301595 Deir Abu Mash'al 4207 Ramallah & Al Bi 721 17.14 2.8 28.83 1010340 1010340 Qabatiya 23941 Jenin 2152 8.99 2.4 28.08 1452240 1452240 Bethlehem (Beit Lahm) 26384 Bethlehem 1651 6.26 2.3 32.09 1452208 1452210 Khallet Hamameh 1577 Bethlehem 35 2.25 0.9 32.55 1301815 1301785 Burqa 2034 Ramallah & Al Bi 228 11.19 1.9 39.69 1301750 1301785 Rammun 2390 Ramallah & Al Bi 268 11.19 1.9 39.69 1301755 1301755 Kafr Ni'ma 4628 Ramallah & Al Bi 402 8.7 1.7 28.49 1301760 1301755 Bil'in 2124 Ramallah & Al Bi 185 8.7 1.7 28.49 1301765 1301785 Beitin 2228 Ramallah & Al Bi 249 11.19 1.9 39.69 1301770 1301620 'Ein Qiniya 717 Ramallah & Al Bi 71 9.9 1.5 30.97 1301775 1301650 Badiw al Mu'arrajat 825 Ramallah & Al Bi 162 19.66 3 33.11 1301780 1301755 Deir Ibzi' 2574 Ramallah & Al Bi 224 8.7 1.7 28.49 1301785 1301785 Deir Dibwan 4143 Ramallah & Al Bi 464 11.19 1.9 39.69 1301790 1301790 Al Bireh 41511 Ramallah & Al Bi 1686 4.06 1.4 32.56 1301800 1301755 'Ein 'Arik 1763 Ramallah & Al Bi 153 8.7 1.7 28.49 1301805 1301805 Saffa 4347 Ramallah & Al Bi 507 11.66 2 29.15 1100290 1100290 Qaffin 10417 Tulkarm 1515 14.55 2.5 28.05 1100330 1100350 Nazlat 'Isa 2245 Tulkarm 282 12.56 2.7 29.41 1151000 1151000 Beit Dajan 4454 Nablus 957 21.5 2.5 31.80 1151010 1151010 Rujeib 5956 Nablus 990 16.63 2.4 29.69 1151025 1150955 Kafr Qallil 3025 Nablus 588 19.42 2.9 28.55 1151030 1151000 Furush Beit Dajan 722 Nablus 155 21.5 2.5 31.80 1151080 1151010 Burin 2840 Nablus 472 16.63 2.4 29.69 1151090 1151090 Beit Furik 13442 Nablus 2034 15.13 4.2 27.50 1151135 1151185 'Awarta 7044 Nablus 1452 20.61 3.6 29.64 1151265 1151270 Osarin 2050 Nablus 504 24.56 4.2 29.03 1151270 1151270 Aqraba 10010 Nablus 2459 24.56 4.2 29.03 1151325 1151335 Yatma 3358 Nablus 405 12.06 2.3 28.03 1151335 1151335 Qabalan 8183 Nablus 987 12.06 2.3 28.03 1151345 1151365 Jurish 1539 Nablus 286 18.61 2.7 29.21 1151365 1151365 Qusra 5410 Nablus 1007 18.61 2.7 29.21 1151375 1151365 Talfit 3586 Nablus 667 18.61 2.7 29.21 1151380 1151245 As Sawiya 2757 Nablus 454 16.46 2.8 28.90 1151405 1151335 Al Lubban ash Sharqiya 2636 Nablus 318 12.06 2.3 28.03 1151410 1151365 Qaryut 2556 Nablus 476 18.61 2.7 29.21 1151420 1151365 Jalud 742 Nablus 138 18.61 2.7 29.21 1151435 1151335 'Ammuriya 370 Nablus 45 12.06 2.3 28.03 1502955 1502950 Deir Razih 287 Hebron 64 22.14 3.5 28.99 1050470 1050550 Khirbet Tell el Himma 76 Tubas & Northen 18 23.76 3.5 29.40 1050551 1050550 Al Farisiya 116 Tubas & Northen 28 23.76 3.5 29.40 1050560 1050550 Al 'Aqaba 166 Tubas & Northen 39 23.76 3.5 29.40 1050450 1050550 'Ein el Beida 1122 Tubas & Northen 267 23.76 3.5 29.40 1050550 1050550 Tayasir 2837 Tubas & Northen 674 23.76 3.5 29.40

38

Table A4. Small area poverty and inequality estimates for merged localities Original Merged Locality name Population, Governorate Number of Poverty Standard Gini localities localities Census 2017 poor rate error of people poverty estimate 1050525 1050535 Salhab 25 Tubas & Northen 5 21.18 5.6 28.43 1050650 1050700 Kashda 65 Tubas & Northen 13 19.55 3.4 33.44 1050670 1050700 Ras al Far'a 1232 Tubas & Northen 241 19.55 3.4 33.44 1050700 1050700 El Far'a Camp 5545 Tubas & Northen 1084 19.55 3.4 33.44 1050740 1050700 Wadi al Far'a 3941 Tubas & Northen 770 19.55 3.4 33.44 1050755 1050755 Tammun 12930 Tubas & Northen 1697 13.12 2.1 28.69 1050790 1050755 Khirbet 'Atuf 213 Tubas & Northen 28 13.12 2.1 28.69 1050720 1050755 Khirbet ar Ras al Ahmar 73 Tubas & Northen 10 13.12 2.1 28.69 1050871 1050755 Al Hadidiya 180 Tubas & Northen 24 13.12 2.1 28.69 1050535 1050535 'Aqqaba 8121 Tubas & Northen 1720 21.18 5.6 28.43 1050610 1050610 Tubas 21253 Tubas & Northen 2109 9.92 2.4 29.04 1050490 1050535 Ibziq 127 Tubas & Northen 27 21.18 5.6 28.43

39

References

Abrahams, Alexei, 2018. Hard traveling: Commuting costs and urban unemployment with deficient labor demand. ESOC Working Paper No. 8. June 1, 2018.

Jawad Al-Salehi & Feda Twam & Aziz Atamanov & Nethra Palaniswamy, 2018. "Measuring Poverty in West Bank and Gaza," World Bank Other Operational Studies 30526, The World Bank.

Atamanov, Aziz and Palaniswamy, Nethra. 2018. West Bank and Gaza Poverty and Shared Prosperity Diagnostic 2011-2017. World Bank, Washington, DC.

Bedi, T., A. Coudouel and K. Simler, 2007. More than a pretty picture: using poverty maps to design better policies and interventions. Washington DC: The World Bank Group.

Calì, Massimiliano & Miaari, Sami H., 2018. "The labor market impact of mobility restrictions: Evidence from the West Bank," Labour Economics, Elsevier, vol. 51(C), pages 136-151.

Daniels, Benjamin, 2012. "CROSSFOLD: Stata module to perform k-fold cross-validation" (https://ideas.repec.org/c/boc/bocode/s457426.html).

Elbers C., J. O. Lanjouw and P. Lanjouw 2003. “Micro-Level Estimation of Poverty and Inequality”. Econometrica 71(1): 355–364.

Nguyen, Minh Cong; Corral Rodas, Paul Andres; Azevedo, Joao Pedro Wagner De; Zhao, Qinghua. 2018. SAE: A Stata Package for Unit Level Small Area Estimation (English). Policy Research working paper; no. WPS 8630. World Bank, Washington, DC.

Tarozzi, A & Deaton, A 2009, ‘Using census and survey data to estimate poverty and inequality for small areas’, The Review of Economics and Statistics, vol. 91, no. 4, pp. 773-792.

Van Der Weide, Roy; Rijkers, Bob; Blankespoor, Brian; Abrahams, Alexei Sisulu, 2018. Obstacles on the road to Palestinian economic growth (English). Policy Research working paper; no. WPS 8385. World Bank, Washington, DC.

World Bank 2011. West Bank and Gaza Coping with Conflict? Poverty and Inclusion in the West Bank and Gaza. Report No. 61293-GZ. World Bank, Washington, DC.

World Bank 2012. “Pilot Study of Small Area Poverty Estimation Methods for the New Member States of the European Union,” Report prepared for the Scientific Steering Committee of the World Bank and European Commission Project on Small Area Poverty Estimation, World Bank, Washington, DC.

World Bank 2014. Seeing is believing. Poverty in the Palestinian Territories. Washington, D.C.: World Bank Group. Washington, D.C.: World Bank Group.

40