53076 v2 THE GOVERNMENT OF YEMEN, THE WORLD B A N K , A N D TH E U N I T E D N AT I O N S DEVELOPMENT PROGRAM Y E M E N P O V E RT Y A S S E S S M E N T (IN FOUR VOLUMES) VOLUME II: ANNEXES NOVEMBER 2007 TABLE OF CONTENTS ANNEX 1 SAMPLING DESIGN ............................................................................................................................ 1 I YEMEN HOUSEHOLD BUDGET SURVEY 2005-2006 SAMPLE DESIGN ..................1 ANNEX 2: CONSTRUCTING THE HBS DATABASE.......................................................................................... 3 I DETECTION AND AUTOMATIC CORRECTION OF OUTLIERS ............................3 IIHOUSEHOLDS CONSERVED IN THE FINAL DATABASES ....................................7 SAMPLING WEIGHTS ..........................................................................................8 III IVSTRUCTURE OF THE DATABASES.........................................................................8 V CONCLUSIONS AND RECOMMENDATIONS .........................................................9 APPENDIX 1YEMEN HOUSEHOLD BUDGET SURVEY 2005-2006 STRUCTURE OF THE DATABASES12 APPENDIX 2 CONSTRUCTING THE HOUSEHOLD BUDGET SURVEY DATABASE ...........................22 ANNEX 3 QUESTIONNAIRE ............................................................................................................................24 QUESTIONNAIRE PAGE 1 .........................................................................................24 QUESTIONNAIRE PAGE 2 .........................................................................................25 QUESTIONNAIRE PAGE 3 .........................................................................................26 QUESTIONNAIRE PAGE 4 .........................................................................................27 QUESTIONNAIRE PAGE 5 .........................................................................................28 QUESTIONNAIRE PAGE 6 .........................................................................................29 QUESTIONNAIRE PAGE 7 .........................................................................................30 QUESTIONNAIRE PAGE 8 .........................................................................................31 QUESTIONNAIRE PAGE 9 .........................................................................................32 QUESTIONNAIRE PAGE 10 .......................................................................................33 QUESTIONNAIRE PAGE 11 .......................................................................................34 QUESTIONNAIRE PAGE 12 .......................................................................................35 QUESTIONNAIRE PAGE 13 .......................................................................................36 QUESTIONNAIRE PAGE 14 .......................................................................................37 QUESTIONNAIRE PAGE 15 .......................................................................................38 QUESTIONNAIRE PAGE 16 .......................................................................................39 QUESTIONNAIRE PAGE 17 .......................................................................................40 QUESTIONNAIRE PAGE 18 .......................................................................................41 QUESTIONNAIRE PAGE 19 .......................................................................................42 QUESTIONNAIRE PAGE 20 .......................................................................................43 QUESTIONNAIRE PAGE 21 .......................................................................................44 QUESTIONNAIRE PAGE 22 .......................................................................................45 QUESTIONNAIRE PAGE 23 .......................................................................................46 QUESTIONNAIRE PAGE 24 .......................................................................................47 QUESTIONNAIRE PAGE 25 .......................................................................................48 QUESTIONNAIRE PAGE 26 .......................................................................................49 QUESTIONNAIRE PAGE 27 .......................................................................................50 QUESTIONNAIRE PAGE 28 .......................................................................................51 Page 4 QUESTIONNAIRE PAGE 29 .......................................................................................52 QUESTIONNAIRE PAGE 30 .......................................................................................53 QUESTIONNAIRE PAGE 31 .......................................................................................54 QUESTIONNAIRE PAGE 32 .......................................................................................55 QUESTIONNAIRE PAGE 33 .......................................................................................56 QUESTIONNAIRE PAGE 34 .......................................................................................57 QUESTIONNAIRE PAGE 35 .......................................................................................58 QUESTIONNAIRE PAGE 36 .......................................................................................59 QUESTIONNAIRE PAGE 37 .......................................................................................60 QUESTIONNAIRE PAGE 38 .......................................................................................61 QUESTIONNAIRE PAGE 39 .......................................................................................62 QUESTIONNAIRE PAGE 40 .......................................................................................63 QUESTIONNAIRE PAGE 41 .......................................................................................64 QUESTIONNAIRE PAGE 42 .......................................................................................65 QUESTIONNAIRE PAGE 43 .......................................................................................66 QUESTIONNAIRE PAGE 44 .......................................................................................67 QUESTIONNAIRE PAGE 45 .......................................................................................68 QUESTIONNAIRE PAGE 46 .......................................................................................69 QUESTIONNAIRE PAGE 47 .......................................................................................70 QUESTIONNAIRE PAGE 48 .......................................................................................71 QUESTIONNAIRE PAGE 49 .......................................................................................72 QUESTIONNAIRE PAGE 50 .......................................................................................73 QUESTIONNAIRE PAGE 51 .......................................................................................74 QUESTIONNAIRE PAGE 52 .......................................................................................75 QUESTIONNAIRE PAGE 53 .......................................................................................76 ANNEX 4: POVERTY LINE METHODOLOGY ............................................................................................77 I MEASURING POVERTY ......................................................................................77 1.1 WELFARE INDICATOR ..............................................................................................................................77 1. 2 INCOME VERSUS EXPENDITURE ............................................................................................................78 1.3 UNITS OF MEASUREMENT ......................................................................................................................78 1.4 POVERTY LINES ..........................................................................................................................................79 1.5 POVERTY MEASUREMENTS ....................................................................................................................81 II HOUSEHOLD SPECIFIC POVERTY LINES ............................................................82 2.1 CALORIC REQUIREMENTS ........................................................................................................................82 2.2 FOOD POVERTY LINE:..............................................................................................................................84 2.3 NON FOOD POVERTY LINE:....................................................................................................................85 ANNEX 5: POVERTY MAP ................................................................................................................................87 POVERTY MAPPING IN YEMEN.....................................................................................................................87 I INTRODUCTION.................................................................................................87 II METHODOLOGY ...............................................................................................87 A CONSUMPTION MODEL ............................................................................................................................88 B POVERTY INDICATORS .............................................................................................................................90 III DATA...............................................................................................................90 Page 5 C Census data....................................................................................................... 90 D SURVEY DATA ...........................................................................................................................................91 IV IMPLEMENTATION ....................................................................................92 E Select a Set of Variables that are Common to the Census and the HBS........... 92 F Estimate models of household consumption per capita using HBS data.......... 93 G Predict household consumption per capita using the census ........................... 94 V CONCERNS ON CURRENT RESULTS .....................................................94 VI REFERENCES ....................................................................................................94 FIGURE 1.................................................................................................................135 FIGURE A1..............................................................................................................172 ANNEX 6: HEALTH ........................................................................................................................................173 ANNEX 7: EDUCATION .................................................................................................................................192 ANNEX 8: IS PUBLIC EXPENDITURE TARGETING IN YEMEN PRO-POOR? ...................................197 I IS PUBLIC EXPENDITURE TARGETING IN YEMEN PRO-POOR?........................197 II BACKGROUND .................................................................................................198 III METHODOLOGY .............................................................................................202 IV DATA ISSUES....................................................................................................203 V FINDINGS .........................................................................................................206 REFERENCES.....................................................................................................................................................213 ANNEX 1: DECOMPOSITION OF THE NATIONAL POOR-AREA TARGETING DIFFERENTIAL ..214 ANNEX 2: UBN INDEX BY GOVERNORATE ...........................................................................................215 ANNEX 9: CONSTRUCTION OF SOCIAL ACCOUNTING MATRIX ........................................................216 ANNEX 10. UPDATES OF THE INPUT/OUTPUT TABLE FOR 2005....................................................226 I IO AND SAM TABLES FOR THE YEMENI ECONOMY ..........................226 I TECHNICAL NOTE ...........................................................................................226 II THE IO TABLE ...............................................................................................227 III THE SOCIAL ACCOUNTING MATRIX ...............................................................229 ANNEX 11. DEMAND SYSTEM ESTIMATION ..........................................................................................231 ANNEX 12: NATIONAL ACCOUNTS DATA ..............................................................................................242 ANNEX 13: MARKET SHARE ANALYSIS ...................................................................................................244 ANNEX 14: CALCULATION OF WELFARE GAINS ..................................................................................245 TABLES Table A.10. 1: The Accounts in our SAM for Yemen and its Structure FIGURES FIGURE A.6 1: RESIDENT ACCESS TO HEALTH CARE, BY DISTRICT ................................. 138 FIGURE A.6 2: PERCENTAGE OF RESIDENTS WHO SOUGHT MEDICAL CARE .................... 176 FIGURE A.6 3: PERCENTAGE OF RESIDENTS WHOM DID NOT SEEK MEDICAL CARE ....... 177 FIGURE A.6 4: INCREASES IN HOUSEHOLD EXPENDITURES ON HEALTH........................... 184 FIGURE A.8. 1: UBN INDEX BY GOVERNORATE ............................................................................205 FIGURE A.8. 2: PUBLIC EXPENDITURE PER CAPITA (2004) AND UBN INDEX (1994)..................207 FIGURE A.8. 3: TARGETING DIFFERENTIALS BY GOVERNORATE (IN THOUSAND RIALS)............210 Page 6 BOXES BOX A.8. 1: DECENTRALIZATION AND THE PROMISE OF EQUITY ....................................................199 BOX A.8. 2: THE STATUS OF FISCAL DECENTRALIZATION AND SUB-NATIONAL EXPENDITURES IN YEMEN .........................................................................................................................................201 BOX A.8. 3: MEASURING POVERTY ...................................................................................................169 BOX A.8. 4: EXPLAINING THE TARGETING DIFFERENTIALS: NORTH-SOUTH AND URBAN-RURAL DIMENSIONS.................................................................................................................................211 Page 7 Page 8 ANNEX 1 SAMPLING DESIGN I YEMEN HOUSEHOLD BUDGET SURVEY 2005-2006 SAMPLE DESIGN 1. The 2005-06 Household Budget Survey (HBS) is an important resource to estimate poverty, its proximate causes and effects of public action on poverty. The HBS provides the database for monitoring poverty as Yemen has just started implementing its second PRSP (2006-2010). This is the third HBS since the unification is 1990. 2. The main objectives of the HBS 2005/2006 are: Producing aggregates of the statistical indicators at the level of the urban and rural communities of each governorate in order to serve the purposes of economic and social development-planning on the central and local levels. Updating the National Accounts estimates in order to enable specialists and development planners to determine each governorate's share in the GDP, through the household's consumption structures. Collecting information about the variation in living standards between the urban and rural communities of each governorate, and between those of different governorates. Sample Frame and Stratification 3. The sample frame for the HBS was the 2004 Population Census. Yemen consists of 21 governorates. The study population was sorted into 38 strata. 17 governorates were represented by two strata (urban and rural,) whereas Sana'a City and Aden are only urban and Raima and Sana'a Region are only rural. This resulted in 19 urban strata and 19 rural strata. 4. Within each stratum, the sample was selected in two stages. In the first stage, a certain number of Census Enumeration Areas (EAs) were selected with probability proportional to size (pps,) using as a measure of size the number of households according to the pre-census estimates available in January 2005. In the second stage, 12 households were picked from each EA by systematic equal probability sampling (seps).1 5. In order to produce estimates of consumption in all governorates of both rural and urban populations, the total sample of 1,200 EAs was distributed across strata by a combination of allocation proportional to size and equal allocation (see Box 1.) The final sample allocation is as show in Figure 1. 1 This design varies significantly from that used for the HBS 1998, where the study community was allocated in 12 strata, 7 of which were urban and 5 were rural, and each stratum consisted of several governorates, except the capital (Sana'a) and the city of Aden, which were considered two distinct urban strata. The sample size of the HBS 1998 was set at 15120 households drawn from 420 PSUs, cluster size was set at 18 households. 1 Box 1: Allocation of Sample Across Strata The results of the 1998 Household Budget Survey were used to assign the sample size that needed to obtain accurate data at governorate level. The procedure used in allocating the sample households for the HBS 2004/2005 had the following steps: 1. 50% of the total sample was distributed proportional to the household counts of the strata. 2. 50% of the total sample size was distributed uniformly amongst strata. 3. Since the larger variation of the living conditions in urban communities result in higher expected standard error for these communities (based on data from HBS 1998), the sample was redistributed between urban and rural strata to achieve uniform expected relative standard errors for overall urban and rural strata (RSE 1.1%). The total sample allocation had total of 9,228 urban and 5,172 rural households. 4. The results were adjusted to make the number of households in each governorate a multiple of 144 (12 EAs of 12 households each,) to facilitate the random allocation of the sample into the 12 months of fieldwork. Figure 1: Yemen Household Budget Survey 2004-2005 (Sampling strata, allocation of the sample and Relative Standard Errors for Per Capita Consumption) HH counts (2994) PSUs Nominal Sample (HHs) R S E (%) Governorate Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural Total 11 Ibb 50,404 249,674 300,078 43 41 84 516 492 1,008 6.98 11.81 9.42 12 Abyan 13,795 42,332 56,127 30 18 48 360 216 576 5.71 7.32 5.29 13 Sec. of the Capital 247,668 247,668 156 156 1,872 1,872 4.44 4.44 14 Al Baydha 13,424 53,004 66,428 29 19 48 348 228 576 7.70 10.98 7.95 15 Taiz 79,029 283,521 362,550 56 40 96 672 480 1,152 6.14 10.13 7.51 16 Al Jawf 7,682 47,940 55,622 22 14 36 264 168 432 7.51 13.52 11.33 17 Hajjah 17,416 174,819 192,235 30 30 60 360 360 720 10.93 8.79 7.73 18 Al Hodiedah 109,974 236,347 346,321 75 33 108 900 396 1,296 7.07 5.85 4.52 19 Hadramout 56,084 63,137 119,221 41 19 60 492 228 720 17.92 24.68 15.67 20 Dhamar 24,639 161,267 185,906 31 29 60 372 348 720 11.92 5.98 5.40 21 Shabwah 8,657 41,101 49,758 21 15 36 252 180 432 7.83 13.94 11.75 22 Saadah 13,620 70,513 84,133 28 20 48 336 240 576 5.41 5.65 4.77 23 Sanaa 116,086 116,086 24 24 288 288 6.76 6.76 24 Aden 89,605 89,605 72 72 864 864 4.61 4.61 25 Lahaj 9,057 93,661 102,718 25 23 48 300 276 576 8.21 10.18 8.82 26 Marib 3,728 23,653 27,381 22 14 36 264 168 432 9.21 13.34 10.79 27 Al Mahweet 4,647 63,785 68,432 27 21 48 324 252 576 4.61 5.48 5.00 28 Al Mahrah 5,459 5,705 11,164 12 12 24 144 144 288 14.90 12.38 9.63 29 Amran 19,073 85,919 104,992 27 21 48 324 252 576 6.86 4.91 4.23 30 Al Dhalea 8,094 51,010 59,104 22 14 36 264 168 432 11.21 6.54 5.84 31 Raimah 55,086 55,086 24 24 288 288 6.40 6.40 Total 782,055 1,918,560 2,700,615 769 431 1,200 9,228 5,172 14,400 2.49 2.98 2.07 2 ANNEX 2: CONSTRUCTING THE HBS DATABASE I DETECTION AND AUTOMATIC CORRECTION OF OUTLIERS 1. We tried to fix the most extreme inconsistencies still remaining in the databases generated from the anthropometric and food consumption sections of the HBS. In order to automatically detect outliers we often used the same tools that will be used by subject matter specialists with analytic purposes later. We specifically used the World Health Organization (WHO) standard anthropometric tables, in order to assess the consistency between height, weight and age measures; food composition tables, in order to detect suspiciously low or high levels of food consumption, by way of the households' per capita energy intake; and specially developed unit price tables, in order to detect errors in the recording of quantities or amounts purchased. 2. Although our tools may have been the same, our objectives at this stage were very different from those that will be pursued by the thematic specialists in the analytic phase. For instance, in anthropometrics, we scrutinized the measures of children who seemed to be too heavy or too light for their age or height, but when doing this we were not trying to assess or qualify the nutritional status of Yemeni children ­ this will be the job of nutritional experts later on. We only wanted to detect possible measurement or recording errors. Similarly, we considered as doubtful the households who appeared to be consuming too few or too many calories, and the transactions with too small or too large unit prices, but our intention was not to assess the households' poverty status or the inter-regional or seasonal variation of prices ­ this will be done by poverty analysts and economic statisticians in the future. 3. Unwilling to qualify as inconsistent observations that are merely unlikely, but not necessarily impossible, our quality control criteria were in general much more lenient than those that specialists will use later to sort their subjects into analytic categories. For instance, whereas nutritionists will qualify as "wasted," "stunted," or "underweight" the kids for whom some of the measures are more that two standard deviations below their average values, we qualified as "inconsistent" the anthropometric measures beyond five standard deviations from the mean. In other words, very few, if any, of the kids we considered as outliers are likely to be genuinely small or large children ­ they are almost certainly outliers indeed. Non-standard conventions for missing values 4. In spite of instructions, some interviewers filled questionnaire fields with numbers such as "999" when they were unable to record the precise answers (they fortunately didn't have many opportunities to use this outdated convention to indicate that a question was not applicable, because the HBS questionnaire was explicitly designed to avoid such cases.) Such numbers can be very annoying at the analytic stage because they distort most results, including averages and standard errors. 3 5. We thoroughly scanned the HBS databases to detect these 999s and replace them with blanks (or periods, in their Stata and SPSS versions.) This was not trivial because the non- standard convention was not used uniformly (sometimes "99", "9.99" or other variants were used instead of "999",) and also because some of the 99s could occasionally represent genuine amounts (such as supermarket promotions.) Anthropometrics 6. The objective of this phase was to replace by blanks the values with strong evidence of being wrongly recorded in the field. However, we kept the ones for which we cannot affirm that they are incorrect or whether they simply reflect the reality in Yemen. Certainly the nutritional analysts who will work with these data will make further analysis using advanced nutritional techniques and they may decide to remove other values as well. 7. For identifying anthropometric outliers we worked in close collaboration with Dr Abdul Baki Alzaemey, who defined the corresponding criteria. The criteria ­ based on the most recent World Health Organization Anthropometric tables, released in June 2006, were the following: For children up to 60 months: Replace the weight by a blank if the Z-score of weight for age is less than -6 or greater than +5. Replace the height by a blank if the Z-score of height for age is less than -6 or greater than +6. For children 61 to 216 months (18 years)2: Replace the weight by a blank if the Z-score of weight for age is less than -5 or greater than +5. Replace the height by a blank if the Z-score of height for age is less than -4 or greater than +5. For individuals older than 18 years: Replace both weigh and height by blanks if the Body Mass Index (BMI) is less than 14.4 or greater than 44. 8. It is important to emphasize that we did not modify the original data on "Section 6: Anthropometrics" nor deleted any individual records. Instead we created two new variables with the values dictated by the above criteria: the new weight and height variables will be either equal to the originals or blanks. Food consumption 9. Section 14 contains the bulkiest and the most important part of the data collected by the HBS ­ the consumption and acquisition of food and some frequently purchased non-food items, 2 The HBS measured all children younger than 5 years throughout the year. In the last month of fieldwork (March 2006,) all household members were measured. 4 reported on a weekly basis. The detection and automatic correction of outliers in this section was concurrent with various other actions of data analysis and scrutiny, performed with the help of a dedicated program developed over the Excel/VBA platform. The program did not need to hold the whole database in memory. Instead it read the file twice on a record-by-record basis. The actions performed in each of the two program passes are described below. 10. The supporting workbook (Fix_S14B.xls) contains a spreadsheet with reference and summary data for all items in Section 14. Figure 2 below shows the first and last rows: Figure 2: Reference and Summary Data Used for Scrutiny of Section 14 11. Columns C to G contain external technical coefficients used to estimate the energy supplied by each item. For most items, the number of Kilo-calories is reported in column G, in reference to the so-called field unit used to record the quantities, coded in Column C as 1 (kilos,) 2 (pieces) or 3 (liters.) For certain items (such as bread or spices) for which the HBS only recorded the amounts spent, not the quantities, columns D to F contain conversion coefficients used to estimate the quantities and energy intakes from the amounts. 12. The other columns contain internal coefficients, obtained by the program from the database itself during the first pass: · Columns G to R refer to weekly purchases. Column G contains the number of weekly transactions reported, and Column R the median unit price. The other columns contain the mean and the standard deviation of the decimal logarithms of, respectively, the amounts spent, the quantities purchased, and the unit prices. · Columns S and T contain the mean and the standard deviation of the decimal logarithm of the quantity consumed in the week. 5 13. During the second pass, the program uses these internal coefficients to detect unlikely combinations of item codes, amounts and quantities (when applicable,) and to eventually fix the problems so detected. (All operations are done with decimal logarithms, but this will not be said explicitly in the rest of this explanation, for the sake of simplicity.) If the unit price of a transaction is more than 4 standard deviations away from the mean, it is considered to be an outlier, and then either the amount or the quantity is fixed, depending on which of the two is farther from its respective mean, and as long as the other one is less than 3 standard deviations from its mean. The magnitude to be fixed is estimated from the correct one using the median price, unless there are reasons to assume that the problem is due to an accidental shift in the location of the decimal point (a common error of both interviewers and data entry operators,) in which case the correction is done by multiplying the incorrect magnitude by an adequate power of 10. 14. The process will be illustrated with an example. One of the HBS interviewers reported that Household Number 2301801 purchased 1.5 kilos of imported wheat (food item code 102) for 7,400 Rials in the fourth survey week ­ an implicit unit price of almost 5,000 Rials per kilo. Based on the 11,303 purchases of imported wheat reported by the HBS for all households in the whole survey year, the program found that the mean and standard deviation of Log10 (unit price) for imported rice were, respectively, 1.74657 and 0.08187 (see row 4 in Figure 2,) meaning that the lower and upper bounds for the acceptable unit prices are 10 1.74657 - 4 x 0.08187 and 10 1.74657 + 4 x 0.08187, or 26.25 and 118.59 Rials per kilo. The implicit unit price of 5,000 Rials per kilo is therefore too high to be credible. Either the amount paid must have been less that 7,400 Rials or the quantity purchased must have been more than 1.5 kilos. To decide which of the two is more likely to have been the case, the program considers that the average amount paid in all 11,303 purchases is 10 3.18072 = 1,516 Rials and the average quantity purchased is 10 1.43416 = 27.17 kilos. Since the 7,400 Rials amount spent in the suspicious transaction is only +1.4 standard deviations above the average, whereas the 1.5 kilos are -2.3 standard deviations below the average, the program decides that the quantity must be wrong and needs to be fixed. The median unit price of all 11,303 purchases is 54 Rials per kilo, which suggests that the real quantity purchased must have been around 7,400 / 54 = 137 kilos. Since this is close to 150, the program decides that the quantity must have been 150.0 kilos, but was wrongly recorded as 1.500. 15. 2,970 of the 1.5 million transactions reported on Section 14 were fixed with this algorithm. As in the case of anthropometric measures, we did not modify any of the original data in Section 14. We just added three additional fields to each record ­ for the (eventually fixed) values of the amount spent, the quantity purchased and the quantity consumed from all sources (market, self-production or gifts.) 16. As a by-product of the scrutiny of Section 14, the program computed the total per capita energy intake, the total per capita food expenditure and the share of food in total expenditure for each of the 13,227 present at that moment in the HBS databases. In agreement with Mr. Srinivasan and Ms El-Laithy, we subsequently dropped from the HBS databases the households for which all three indicators were low enough to make further analyses unreliable. We sorted into this category 91 households reported as consuming less than 800 Kcal/capita/day, spending less than 1,000 Rials/capita/month on food and less than 10 percent of their budget on food. 6 II HOUSEHOLDS CONSERVED IN THE FINAL DATABASES 17. The target sample size was 14,400 households. Four questionnaires never arrived to the data entry office, therefore only 13,396 were entered. Of those, 996 households were qualified by the field workers as either (1) interview not complete, (2) household empty or destroyed or (3) refusal; and were subsequently dropped from the databases by the CSO prior to our arrival. The same was done with 173 households without any food consumption recorded in the diary. 18. As said before, during the course of this mission we identified and removed form the databases an additional 91 households with extremely low food consumption. Figure 3 below gives the distribution of the remaining 13,136 households by Governorate and survey month. Figure 3: Distribution of the Households in the HBS Databases (by Governorate and Survey Month) 7 III SAMPLING WEIGHTS 19. (For a better understanding of this section we reproduce in Appendix 2 a summary description of the HBS sampling design prepared by Mr Srinivasan.) 20. During the course of this mission, we computed the sampling weights (or raising factors,) needed to produce unbiased estimates from the survey. We first computed the probability pijh of selecting household ijh in Enumeration Area (EA) jh of stratum h as k h n jh m jh pijk = nh njh where kh is the number of EAs selected in stratum h; njh is the number of households in EA jh, according to the pre-census estimates available at the time the EAs were selected; nh is the number of households in stratum h, according to the pre-census estimates; mjh is the number of households in the final database in EA jh; and n'jh is the number of households in EA jh, according to the final census figures. We then computed the nominal weight wijh of household ijh as the inverse of its selection probability: 1 wijk = p ijh We finally computed the adjusted weight ijh of household ijh as nh ijh = wijh wijh h Where n'h is the number of households in stratum h, according to the final census figures. The final adjustment intends to have the HBS sum of weights match the official CSO number of households figures in all strata. IV STRUCTURE OF THE DATABASES 21. All data files were organized into 14 themes and delivered to CSO in three formats: SPSS (.sav), Stata (.dta) and dbf. Each theme corresponds to a specific statistical unit: 1. Households: contains data on the cover, dwelling conditions and household- level information on agriculture and credits. 2. Individuals: demographics, education, health and unemployment. 8 3. Enterprises: general information about each family enterprise 4. Jobs: information on each job conducted by a household member during the past 12 months. 5. Wages: specific information on each job conducted by a household member for wages. 6. Crops: crops grown during the past 12 months. 7. Types of land: information on various types of agricultural land owned or operated by the household during the past 12 months. 8. Enterprise incomes/expenditures: income and expenditures on specific items for each family enterprise. 9. Other sources of income: information on non-work income received by household members during the past 12 months. 10. Durable goods: durable goods owned by the household. 11. Credits: Credits or loans obtained by household members. 12. Food consumption: Acquisition and consumption of food and other frequently purchased items. 13. Non-food consumption: Acquisition of non-food items 14. Anthropometrics: Weight and height of children up to 6 years old (and all household members during the last survey month.) 22. All records in the fourteen files contain the following key information: Household identification number Sampling weight Stratum (governorate and urban/rural) Cluster Governorate Area (urban/rural) Survey month Household size The complete content of each file is given in Appendix 1. V CONCLUSIONS AND RECOMMENDATIONS No more data cleaning 23. The activities and actions developed during this mission are generally considered to be a part of the "data cleaning" phase of a household survey project. Two questions that can naturally be asked at this point are [1] Is the HBS database now totally consistent? and, [2] does 9 it need more "cleaning" before being delivered to end users for tabulation and analysis? The answer to the first question is probably not. The answer to the second question is definitely not. 24. We have already taken care of the most serious inconsistencies ­ those that could have led to wrong conclusions in poverty and nutritional analyses. Survey analysts are very likely to find more inconsistencies as a part of their endeavors, but this is not a reason for spending additional time and efforts to further refine the HBS data. There are in fact three powerful reasons for not doing this and delivering the HBS database to users as soon as possible. The first reason is that the databases generated by a survey as large and complex as the Yemen HBS can never be considered as perfect ­ there will be always something else that could be done, but the opportunity cost of doing it as a prerequisite for further tabulation is just too high. The database already represents a reality that is more than a year old, and it looses its policy-making value with each month that passes. Delivery is urgent. Another reason is that serious data analysts do understand that datasets from complex surveys are imperfect. They have analytic tools and expertise to deal with this situation and they prefer to do it themselves rather than relying of somebody else's criteria, especially when the later is not properly documented. The third reason is that solving the remaining inconsistencies may occasionally imply making imputations, which at this point ­ many miles and months away from the place and time where the data were collected ­ can only be made by guesswork. We strongly recommend not to submit the HBS to further "data cleaning." Use statistical software 25. The CSO has traditionally used tailor-made computer programs for tabulating census and survey data. We recommend that the institution evolves towards the use of standard statistical software (such as Ariel, Stata or SPSS) for this purpose. This will reduce the time and human resources needed to prepare tables and make the tables much more reliable. It will also foster analytic thinking throughout the institution, allow analysts to directly interact with the data, without intermediaries, and open the way to advanced models and techniques that can hardly be programmed on a case-by-case basis. 26. A simple illustration of the superiority of statistical software over tailor-made programs is the need to obtain weighed estimates from survey data. This is a non-trivial challenge for any programmer, but is easily solved by anyone using standard statistical software. Recommendations for future surveys 27. A key factor for the success of complex surveys is the effective integration of computer- based quality controls to fieldwork. This can be achieved by implementing a high-quality data entry program and deploying dedicated data entry operators and PCs to perform data entry and consistency controls on a household-by-household basis as a part of field operations, so that 10 errors and inconsistencies are solved by means of eventual revisits to the households. The direct benefits of this methodology are: it significantly improve the quality of the information collected by the survey, because the errors and inconsistencies will be detected while the interviewers are still in the field rather than by office "cleansing" later. it generates databases that are ready for tabulation and analysis in a timely fashion; in fact, as the survey is conducted, thus giving the survey managers the ability to effectively monitor field operations. it fosters the application of uniform criteria by all the interviewers and throughout the whole period of data collection, 28. The improvements in quality and timing of this alternative are such that we strongly recommend that the CSO considers using it for future rounds of the HBS and in any other complex surveys. 29. In future rounds of the HBS, the CSO may also consider to re-visit some of the same households already visited by the survey in 2004-2006. A panel survey of this kind would have many analytic advantages. If this is to an option, we strongly recommend entering the names of household members in the HBS database now. This is almost costless, very easy to do now that the paper forms are still in good conditions and the names are legible, and it would facilitate enormously the organization of a panel survey in the future. (The names should obviously be kept in the CSO's internal database only ­ not delivered to external data users.) 11 APPENDIX 1YEMEN HOUSEHOLD BUDGET SURVEY 2005-2006 STRUCTURE OF THE DATABASES Page 12 Page 13 Page 14 Page 15 Page 16 Page 17 Page 18 Page 19 Page 20 Page 21 APPENDIX 2 CONSTRUCTING THE HOUSEHOLD BUDGET SURVEY DATABASE 1. The 2005-06 Household Budget Survey (HBS) is an important resource to estimate poverty, its proximate causes and effects of public action on poverty. The HBS provides the database for monitoring poverty as Yemen has just started implementing its second PRSP (2006-2010). This is the third HBS since the unification is 1990. 2. The main objectives of the HBS 2005/2006 are: 1. Producing aggregates of the statistical indicators at the level of the urban and rural communities of each governorate in order to serve the purposes of economic and social development-planning on the central and local levels. 2. Updating the National Accounts estimates in order to enable specialists and development planners to determine each governorate's share in the GDP, through the household's consumption structures. 3. Collecting information about the variation in living standards between the urban and rural communities of each governorate, and between those of different governorates. Sample Frame and Stratification 3. The sample frame for the HBS was the 2004 Population Census. Yemen consists of 21 governorates. The study population was sorted into 38 strata. 17 governorates were represented by two strata (urban and rural,) whereas Sana'a City and Aden are only urban and Raima and Sana'a Region are only rural. This resulted in 19 urban strata and 19 rural strata. 4. Within each stratum, the sample was selected in two stages. In the first stage, a certain number of Census Enumeration Areas (EAs) were selected with probability proportional to size (pps,) using as a measure of size the number of households according to the pre-census estimates available in January 2005. In the second stage, 12 households were picked from each EA by systematic equal probability sampling (seps).3 5. In order to produce estimates of consumption in all governorates of both rural and urban populations, the total sample of 1,200 EAs was distributed across strata by a combination of allocation proportional to size and equal allocation (see Box 2.) The final sample allocation is as show in Figure 3. 3 This design varies significantly from that used for the HBS 1998, where the study community was allocated in 12 strata, 7 of which were urban and 5 were rural, and each stratum consisted of several governorates, except the capital (Sana'a) and the city of Aden, which were considered two distinct urban strata. The sample size of the HBS 1998 was set at 15120 households drawn from 420 PSUs, cluster size was set at 18 households. 22 Box 2: Allocation of Sample across Strata The results of the 1998 Household Budget Survey were used to assign the sample size that needed to obtain accurate data at governorate level. The procedure used in allocating the sample households for the HBS 2004/2005 had the following steps: 1. 50% of the total sample was distributed proportional to the household counts of the strata. 2. 50% of the total sample size was distributed uniformly amongst strata. 3. Since the larger variation of the living conditions in urban communities result in higher expected standard error for these communities (based on data from HBS 1998), the sample was redistributed between urban and rural strata to achieve uniform expected relative standard errors for overall urban and rural strata (RSE 1.1%). The total sample allocation had total of 9,228 urban and 5,172 rural households. 4. The results were adjusted to make the number of households in each governorate a multiple of 144 (12 EAs of 12 households each,) to facilitate the random allocation of the sample into the 12 months of fieldwork. Figure 4: Yemen Household Budget Survey 2004-2005 Sampling Strata (Allocation of the Sample and Relative Standard Errors for Per Capita Consumption) HH counts (2994) PSUs Nominal Sample (HHs) R S E (%) Governorate Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural Total 11 Ibb 50,404 249,674 300,078 43 41 84 516 492 1,008 6.98 11.81 9.42 12 Abyan 13,795 42,332 56,127 30 18 48 360 216 576 5.71 7.32 5.29 13 Sec. of the Capital 247,668 247,668 156 156 1,872 1,872 4.44 4.44 14 Al Baydha 13,424 53,004 66,428 29 19 48 348 228 576 7.70 10.98 7.95 15 Taiz 79,029 283,521 362,550 56 40 96 672 480 1,152 6.14 10.13 7.51 16 Al Jawf 7,682 47,940 55,622 22 14 36 264 168 432 7.51 13.52 11.33 17 Hajjah 17,416 174,819 192,235 30 30 60 360 360 720 10.93 8.79 7.73 18 Al Hodiedah 109,974 236,347 346,321 75 33 108 900 396 1,296 7.07 5.85 4.52 19 Hadramout 56,084 63,137 119,221 41 19 60 492 228 720 17.92 24.68 15.67 20 Dhamar 24,639 161,267 185,906 31 29 60 372 348 720 11.92 5.98 5.40 21 Shabwah 8,657 41,101 49,758 21 15 36 252 180 432 7.83 13.94 11.75 22 Saadah 13,620 70,513 84,133 28 20 48 336 240 576 5.41 5.65 4.77 23 Sanaa 116,086 116,086 24 24 288 288 6.76 6.76 24 Aden 89,605 89,605 72 72 864 864 4.61 4.61 25 Lahaj 9,057 93,661 102,718 25 23 48 300 276 576 8.21 10.18 8.82 26 Marib 3,728 23,653 27,381 22 14 36 264 168 432 9.21 13.34 10.79 27 Al Mahweet 4,647 63,785 68,432 27 21 48 324 252 576 4.61 5.48 5.00 28 Al Mahrah 5,459 5,705 11,164 12 12 24 144 144 288 14.90 12.38 9.63 29 Amran 19,073 85,919 104,992 27 21 48 324 252 576 6.86 4.91 4.23 30 Al Dhalea 8,094 51,010 59,104 22 14 36 264 168 432 11.21 6.54 5.84 31 Raimah 55,086 55,086 24 24 288 288 6.40 6.40 Total 782,055 1,918,560 2,700,615 769 431 1,200 9,228 5,172 14,400 2.49 2.98 2.07 23 ANNEX 3 QUESTIONNAIRE QUESTIONNAIRE PAGE 1 24 QUESTIONNAIRE PAGE 2 25 QUESTIONNAIRE PAGE 3 26 QUESTIONNAIRE PAGE 4 27 QUESTIONNAIRE PAGE 5 28 QUESTIONNAIRE PAGE 6 29 QUESTIONNAIRE PAGE 7 30 QUESTIONNAIRE PAGE 8 31 QUESTIONNAIRE PAGE 9 32 QUESTIONNAIRE PAGE 10 33 QUESTIONNAIRE PAGE 11 34 QUESTIONNAIRE PAGE 12 35 QUESTIONNAIRE PAGE 13 36 QUESTIONNAIRE PAGE 14 37 QUESTIONNAIRE PAGE 15 38 QUESTIONNAIRE PAGE 16 39 QUESTIONNAIRE PAGE 17 40 QUESTIONNAIRE PAGE 18 41 QUESTIONNAIRE PAGE 19 42 QUESTIONNAIRE PAGE 20 43 QUESTIONNAIRE PAGE 21 44 QUESTIONNAIRE PAGE 22 45 QUESTIONNAIRE PAGE 23 46 QUESTIONNAIRE PAGE 24 47 QUESTIONNAIRE PAGE 25 48 QUESTIONNAIRE PAGE 26 49 QUESTIONNAIRE PAGE 27 50 QUESTIONNAIRE PAGE 28 51 QUESTIONNAIRE PAGE 29 52 QUESTIONNAIRE PAGE 30 53 QUESTIONNAIRE PAGE 31 54 QUESTIONNAIRE PAGE 32 55 QUESTIONNAIRE PAGE 33 56 QUESTIONNAIRE PAGE 34 57 QUESTIONNAIRE PAGE 35 58 QUESTIONNAIRE PAGE 36 59 QUESTIONNAIRE PAGE 37 60 QUESTIONNAIRE PAGE 38 61 QUESTIONNAIRE PAGE 39 62 QUESTIONNAIRE PAGE 40 63 QUESTIONNAIRE PAGE 41 64 QUESTIONNAIRE PAGE 42 65 QUESTIONNAIRE PAGE 43 66 QUESTIONNAIRE PAGE 44 67 QUESTIONNAIRE PAGE 45 68 QUESTIONNAIRE PAGE 46 69 QUESTIONNAIRE PAGE 47 70 QUESTIONNAIRE PAGE 48 71 QUESTIONNAIRE PAGE 49 72 QUESTIONNAIRE PAGE 50 QUESTIONNAIRE PAGE 51 QUESTIONNAIRE PAGE 52 QUESTIONNAIRE PAGE 53 ANNEX 4: POVERTY LINE METHODOLOGY I MEASURING POVERTY 1. Poverty analysis and assessment in Yemen has been driven by the concern to design appropriate poverty reduction strategies. However, debates about methods of poverty measurement are common because poverty is an elusive concept and no single measure can properly or adequately reflect its magnitude and features. Views differ on how individual welfare should be measured, how poverty lines should be set, and what poverty measures should be used. 2. The household raw data, for 1997 and 2005/06, provide a unique opportunity to evaluate the evolution of living standards over the period under consideration. 3. In what follows is a brief discussion of some of the conceptual issues underlying the practice of poverty measurements and comparisons, which will form the basis for our subsequent analysis on the size, evolution and profile of poverty in Yemen. 4. Poverty has traditionally been defined as a discrete characteristic- either one is poor or one is not. Given a particular indicator of welfare, a certain line or standard is drawn, and an individual or household falls on one side or the other. Analysis of poverty takes place at two different levels. Defining poverty consists of classifying the population into poor and non- poor. Measuring poverty seeks to aggregate the "amount" of poverty into a single statistic. 5. Constructing a poverty profile to show how poverty varies across sub-groups of a population is typically the first step in designing an anti-poverty policy. So how should a poverty profile be constructed? One appealing guiding principle is that within a given standard of living, poverty should not depend on which sub-group in the poverty profile the person with that standard of living happens to belong. Following Ravallion 1991, a poverty profile would be "consistent" if it respects this principle. Consistency requires that the poverty line is fixed in terms of the indicator of living standards used. Consistent poverty comparisons imply that two persons at the same real consumption level are deemed to be either "poor" or "not poor" irrespective of the time or place under consideration, or the presence or absence of policy change within the relevant domain. 1 Measuring Welfare 1.1 Welfare Indicator 6. There are different approaches to measuring welfare or well-being (Ravallion, 1994). For a given society, poverty exists if an individual (or household) is unable to attain a certain standard of living, or well-being, at the minimum levels accepted by the standards of that society. The issue is which factors or indicators constitute well- being or welfare? The approach we adopt is to measure welfare in terms of a money metric indicator, defined as the amount of money required - given a set of prices and the assumption of utility maximization - to attain a particular level of utility. This allows us to compare household's welfare levels, which cannot be observed, by comparing their observable consumption levels. Thus, consumption bases approach becomes particularly suited for measuring poverty in developing countries, since it 77 bases poverty comparisons in terms of deprivation from certain commodities and resources (both food and non-food) considered essential for a minimum level of well- being within a given society. However, there are other factors determining the standard of living and affecting welfare that cannot be readily reduced to a single monetary measure. Examples of such factors are access to education, access to basic health services, and access to safe potable water and basic housing amenities. Strictly interpreted, poverty means the inability of individuals to attain adequate or minimum nutrition, clothing, or shelter. More broadly, it encompasses those factors that enable the Individuals' command over resources, such as being healthy and literate. Poverty in this latter sense would constitute deprivation in capabilities, as measured by the UNDP Human Poverty Index. To measure poverty in this sense, the money metric welfare indicator should therefore be supplemented by other social indicators of well- being, such as infant mortality, school enrolment, life expectancy at birth, etc. 1. 2 Income versus expenditure 7. There are several conceptual and empirical considerations favoring the use of expenditure/ consumption, as opposed to income, as the basis for the welfare indicator in developing countries (Hentshcel and Lanjouw, 1996). One consideration is that since all income is not consumed, nor is all consumption financed out of income, consumption is arguably a more appropriate indicator if we are concerned with realized welfare. Expenditures/ consumption better reflect what households can command in terms of current income. They also reflect their access to credit markets or past savings when incomes are low or negative. A second consideration relates to the consumption options and income sources of the poor. Whereas poor households are likely to be purchasing and consuming only a narrow range of goods and services, their incomes may well derive from a variety of sources, many of which can be seasonal in nature. Expenditures/ consumption, are therefore a better indicator of longer run living standards than current income, since consumption tends to smooth variability and fluctuations in income streams. Thirdly, the practical problem of using income to indicate welfare lies in the measurement of incomes of individuals who operate their own business, where records of family businesses are often not kept. Lastly, survey respondents may be more willing to reveal their consumption patterns rather than their income. 1.3 Units of Measurement 8. Household budget surveys provide the most important source of data for poverty comparisons. These surveys record information on household income and consumption expenditures on various goods and services, and they are considered, therefore a good source of information on the distribution of welfare within society. In measuring poverty, a few issues must be considered when deploying household budget surveys. 9. After a comprehensive measure of household consumption is constructed, the critical issue of adjustment of household welfare for differences in household composition must be discussed. 10. Household surveys typically record aggregate outlays made by the household on various commodities. Poverty comparisons have thus tended to use the household - as 78 opposed to the individual as a unit of measurement. Total household consumption is likely to overstate the welfare level of persons in large households, since the goods and services consumed must be divided among more people. The most common adjustment made is to use per capita consumption. This may under-estimate welfare levels because households have very different compositions, and small children have smaller needs for food and some other items relative to adults. Further, there may be economies to scale in consumption for certain commodities. To correct for this, one can estimate household equivalence scales. Adult equivalence scales are therefore used to adjust the welfare measure for individuals to take into account differences in the age and gender structure of the household. Applying an adult equivalence scale means that household members are assigned a weight between zero and unity, depending on their age and gender. Adult equivalence scales typically assign a value of one to adult males and less than one to adult females and children (Ravallion, 1992). 11. However, calculating such scales is controversial. In this report, this controversy is overcome by controlling for difference in household composition and estimating household specific poverty line, as will be discussed in the following section. 12. Through out our analysis we used actual household consumption4 as welfare measures where actual consumption is the sum of values of market purchased goods, own produced goods and freely received goods. 1.4 Poverty Lines 13. Poverty lines can be absolute, relative or subjective. Much of the literature on poverty has been concerned with the respective merits of absolute and relative measures of poverty. 14. The choice of poverty lines is very critical as different methods can produce different rates of poverty and can sometimes cause a reverse in ranking, either between sub-groups or between different dates. When the purpose is to monitor progress in reducing absolute consumption poverty- defined in terms of command over basic consumption needs- one should not consider a person who chooses to buy fewer and more expensive calories poorer than another person who lives, for example, in a village, if both can afford exactly the same standard of living. (Ravallion, 1996). 15. One of the most common approaches is the Basic Needs Approach. By this approach, the poverty line is set as the cost in each sector and at each date of a normative "basic needs" bundle of goods. The difficulty is in identifying what constitutes "basic needs". For developing countries, the most important component of a basic needs poverty line is generally the food expenditure necessary to attain some recommended food energy intake. Thus, the food bundle is typically chosen to be sufficient to reach the predetermined calorie requirement, with a composition that is consistent with the consumption behavior of the poor. This bundle is then evaluated using prices prevailing in each sub-group (region) and at each date. Poverty lines can be then interpreted as Laspeyres cost-of-living numbers. Ravallion (1996) explained that the most compelling argument in favor of CBN method for making poverty comparisons is that it explicitly aims to control for differences in purchasing power over basic consumption needs. The CBN method can at least claim to provide a first order approximation of what we are trying to measure. The cost of bundle is known as the food poverty line. 4 See Annex __ for definitions and calculations of household income, expenditure and consumption 79 16. One could argue that sufficient calories intake does not ensure that basic food needs are met. However, Lipton (1986) argued that shortfalls in nutrients other than calories are almost always due to inadequate caloric intake or are not related to income increases. Protein deficiency is almost always cured once caloric needs are met. Deficiencies of vitamin, iron, magnesium iodine and other micronutrients occur on a large scale even without caloric shortage. However, cost-effective cures are likely to be achieved not by measures to raise income, intake or unit requirements of some or all foods, but by public action. 17. Another alternative is to set an ideal cheap diet to attain basic nutrition requirements and find its cost. However, attaining adequate nutrition is not the sole motive for human behavior (not even for most of the poor), nor is it the sole motive of food consumption. 18. Food poverty line is augmented by an allowance for expenditure on essential non- food goods. Following Engel's law, the non-food allowance can be estimated in two ways; (i) regressing the food share against total expenditures and identifying the non-food share in the expenditure distribution of households in which expenditure on food is equivalent to the food poverty line; or (ii) identifying the share of non-food expenditure for households in which total expenditure is equivalent to the food poverty line. The former approach yields an "upper" bound of the poverty line, while the latter yields a "lower" bound or the "ultra" poverty line, since it defines the total poverty line in terms of those households which had to displace food consumption to allow for non-food expenditures, is considered to be the minimum indispensable level of non-food requirements. 19. An alternative to this method is to estimate non food poverty line using non parametric approach, see Kakwani (2007). As explained by Kakwani 2007; select the households whose food expenditure lies between 90 percent and 110 percent of food poverty line. And then calculate the average non-food poverty line for the individuals belonging to these households. Adjustment to take account of economies of scale are also taken into account. Let j is the economies of scale parameter for the jth component of the non-food poverty line, which takes value 1 if the jth component is a purely private good and takes value 0 if the jth component is a purely public good. Suppose x j is the per capita mean poverty line for the jth non-food component and ni is the size of the ith household, then the economies of scale adjusted consumption of the jth non-food component by the ith household will be given by ( j -1) 20. xij = kx j ni 21. Absolute poverty lines have been widely used in developing countries since poverty research is dominated by the concern for the attainment of basic needs and the achievement of well-being in absolute terms. 22. Relative poverty lines have been more widely used in developed countries. These define poverty in terms of a proportion of the national mean. For instance, the poverty line can be set at 50 percent of the national mean. The poverty line in this sense would be sensitive solely to changes in the relative distribution of welfare i.e. on the parameters of the Lorenz curve (Ravallion, 1994). 23. Subjective poverty lines on the other hand define poverty in terms of individual judgments about what constitutes a socially acceptable minimum standard of living in a 80 given society. This approach is usually based on survey responses to a typical question such as: "What income level do you personally consider to be absolutely minimal?" (paraphrased from Kapten et al 1988 in Ravallion, 1992). Poverty measures based on the subjective approach tend to be an increasing function of income. That is, the higher the income of the individual surveyed, the higher the standard of living he or she considers as minimum. 1.5 Poverty Measurements 24. It has become standard practice in poverty comparisons to use the Foster-Greer- Thorbecke class of decomposable poverty measurements. These include three indices: the head count, the poverty gap and the poverty severity indices. 25. The head count index (P0) is a measure of the prevalence of poverty. It denotes the percentage of households that are poor ­ as defined by the poverty line - as a proportion of total population. This measure however, is insensitive to the distribution of the poor below the poverty line. This is captured by the following two indices, P1 and P2. The poverty gap index (P1) is a measure of the depth of poverty and denotes the gap between the observed expenditure levels of poor households and the poverty line. Assuming perfect targeting, the poverty gap index indicates the amount of resources (transfers) needed to bring all poor households up to the poverty line. The poverty severity index (P2) measures the degree of inequality in distribution below the poverty line and gives greater weight to households at the bottom of the income (or expenditure) distribution. 26. To illustrate, we suppose that as a result of a policy change, 10 percent of income is redistributed from a poor household whose income level places it at 30 percent below the poverty line to another household placed at 50 percent below the poverty line. The head count index in this case would not change, since the size of the redistribution does not enable either household to move up to the poverty line. The poverty gap index would not change either, since the redistribution occurred at levels below the poverty line. The effect of this redistribution policy will be captured by the P2 index, as the position of the lower level household in the distribution would improve. 2. Estimation of poverty lines in Yemen 27. The choice of the welfare indicator used in the estimation of the poverty line is a critical factor in making poverty assessments. Adjustments to spatial and time differentials can significantly influence the conclusions derived. Given the importance of correctly targeting poverty alleviation interventions at the regional level, this study has adopted a strong regional focus. Yemen is divided 20 governorates each is subdivided into urban and rural areas except Sanaa region and Aden. The estimated poverty lines ensure that regional differences in factors such as relative prices, activity levels, as well as size and age composition of poor households. This results in a rank distribution that is consistent with the chosen indicator of household welfare. Several poverty lines have been estimated to obtain a wide range of poverty comparisons among regions between 1997 and 2005/06. We present below methodologies used to estimate these poverty lines. 81 II HOUSEHOLD SPECIFIC POVERTY LINES 28. The report follows the cost of basic needs methodology to construct household region-specific poverty lines. The food poverty line varies for each household and for each region. Differences in poverty lines reflect variations in the food and non-food prices across regions. They also incorporate household differences in the size and age composition, and their food and non-food consumption preferences. 2.1 Caloric Requirements 29. The FAO has been concerned with the issue of determining the nutritional norms of individuals in different age and sex groups. These norms vary from country to country (and even different groups within a country) depending on factors such as race, climatic conditions, etc. 30. The nutritional needs of individuals are the starting point to construct food poverty line. It must be emphasized that these needs of individuals depend on several factors such as age, sex, location conditions and activity levels. 31. We adopted norms appropriate for Yemen. First we obtained the average weight and height of the Yemeni population 18 years old and over. Weights and heights data were collected for all household members surveyed during the last month of HBS. Table 1: Weight and Height by age, sex and location Urban Rural Male Female Male Female AGE_CAT Height Weight Height Weight Height Weight Height Weight from 18 to less than 30 162.64 58.30 153.10 53.69 161.82 57.18 155.30 51.96 from 30 to less than 60 164.02 66.06 154.38 61.76 163.59 61.03 154.91 55.61 60 and older 160.63 62.04 147.55 52.71 159.11 57.00 147.30 48.65 1. BMR is calculated for each individual 18 years and above, using equations in table 3, provided in "Energy and Protein Requirements; Report of a Joint FAO/WHO/UNU Expert Consultation", see table 2. 82 Table 2: Equations to Calculate BMR by Sex and Age Age range (years) BMR 10­18 (16.6W + 77H + 572) 18­30 (15.4W - 27H + 717) Men 30­60 (11.3W + 16H + 901) > 60 (8.8W + 1 128H - 1 071) 10­18 (7.4W + 482H + 217) Women 18­30 (13.3W + 334H + 35) 30­60 (8.7W - 25H + 865) 2. Individual caloric requirements are calculated by multiplying BMR by a factor to reflect an individual's activity level. Following WHO, we assumed that the activity levels for both males and females are moderate in urban areas and heavy in rural areas. Thus caloric requirements for individuals of age 18 years and above are obtained. For younger individuals, caloric requirements were obtained directly from WHO report, see Table 3. Table 3: Average Daily Energy Requirement of Adults Whose Occupational Work is Classified as Light, Moderate, or Heavy, Expressed as a Multiple of BMR Light Moderate Heavy Men 1.55 1.78 2.10 Women 1.56 1.64 1.82 3. Thus, for each household its own caloric requirements can be calculated, depending on its location, age of its members and their gender composition. 83 Table 4: Calculations for Caloric Requirements <1 335 335 <1 335 1_ 950 850 1_ 950 2_ 1125 1050 2_ 1125 3_ 1250 1150 3_ 1250 4_ 1350 1250 4_ 1350 5_ 1474 1325 5_ 1474 6_ 1575 1425 6_ 1575 7_ 1700 1550 7_ 1700 8_ 1825 1700 8_ 1825 9_ 1975 1850 9_ 1975 10_ 2200 1 950 10_ 2200 11_ 2200 1 950 11_ 2200 12_ 2400 2 100 12_ 2400 13_ 2400 2 100 13_ 2400 14_ 2650 2150 14_ 2650 15_ 2650 2150 15_ 2650 16_ 2650 2150 16_ 2650 17_ 2750 2150 17_ 2750 18_<30 2796 2180 18_<30 2796 30_60 2979 2237 30_60 2979 >60 2291 1841 >60 2291 2.2 Food Poverty Line: 4. Once the minimum caloric needs have been estimated, the next step is to determine the cost of obtaining the minimum level of calories. Cost is determined by how the calories are obtained on average by the first two quintiles, rather than by pricing out the cheapest way of obtaining the calories or following a recommended diet. For the first two quintile of households ranked by nominal per capita consumption, average quantities of all food items is constructed. Total calories generated by this bundle are calculated using calories contents in every food item. These quantities represent the bundle used to estimate the food poverty lines, which reflect consumption preferences of the poor. The bundle was priced using median market prices prevailing in each region, When market price of certain item in specific region is not available, we used median unit prices 5 derived from household questionnaire. Dividing cost of the chosen bundle by calories generated by it, the costs per calorie in each region were obtained. Household specific food poverty line is derived by multiplying calorie requirements for all household members by relevant cost of calories. Food poverty line takes into account household gender and age composition as well as its residential region. Food poverty line is used define extreme poverty, where households whose total actual consumption are below their food poverty lines, are considered ultra poor. 5 Unit values are obtained by dividing the reported value by its corresponding quantity. 84 5. This stage can be explained mathematically as follows: let Z denote the actual food consumption vector of the reference group of households initially considered poor; first two quintiles. The corresponding caloric values are represented by the vector k, and the food energy intake of the reference group is then kz = k.Z'. Let cost of this bundle for region r is Pr and caloric requirements of household h is Ch. Food poverty line for household h is then given by (kz /Pr)* Ch , thus the relative quantities in the diet of the poor are preserved in setting the poverty line. Table 5: Cost of 1000 Calories by Region Urban Rurak Ibb 49.497 49.43 Abyan 52.586 53.57 Sana'a City 53.357 0.00 Al-Baida 54.704 51.47 Taiz 55.387 52.96 Al-Jawf 51.67 49.52 Hajja 55.72 57.95 Al-Hodeida 49.854 51.84 Hadramout 54.965 57.50 Dhamar 51.884 58.51 Shabwah 53.244 59.98 Sa'adah 54.319 50.90 Sana'a Region 52.522 0.00 Aden 53.324 0.00 Laheg 49.083 49.89 Mareb 52.199 54.61 Al-Mahweet 51.11 51.41 Al-Maharh 51.753 56.41 Amran 55.614 64.70 Al-Dhale 52.462 50.00 Remah 53.824 0.00 2.3 Non food Poverty Line: 6. While the cost of the minimum food bundle is derived from estimated physiological needs, there is no equivalent methodology for determining the minimum non-food bundle. Following Engel's law, food shares are regressed against logarithm of total household expenditure relative to food poverty line and its square, logarithm of household size and its square, share of small and older children, share of adult males and females, and share of elderly. 7. That is si = + log( x i / z f ) + (log( x i / z f )) 2 + hi , (1) 1. Where si denotes food share of household i, xi is its actual consumption, zf if the food poverty line and hi is vector of household demographic characteristics. 2. The non-food allowance for each household can be estimated in two ways: (i) regressing the food share against total expenditures and identifying the non-food share in the expenditure distribution of households in which expenditure on food is 85 equivalent to the food poverty line; or (ii) by identifying the share of non-food expenditure for households in which total expenditure is equivalent to the food poverty line. The former approach yields an "upper" bound of the poverty line, while the latter yields a "lower" bound, since it defines the total poverty line in terms of those households which had to displace food consumption to allow for non-food expenditures, considered to be a minimum indispensable level of non-food requirements. Thus lower poverty line =(2-si)*zf (2). Upper poverty line is obtained by solving equation (1) iteratively. 3. By this approach household regional specific poverty lines are estimated (households with the same gender and age composition in each region have the same poverty lines). Obviously this approach takes into account location, age and gender composition as well as economies of scale, as food shares and hence non food estimates vary according to household size, age and gender composition. Hence differences in food shares result from the addition of members of specific age and gender. The sharing behaviors among household members are also reflected. 86 ANNEX 5: POVERTY MAP Poverty Mapping in Yemen6 I INTRODUCTION 1. This report describes how the poverty mapping method developed in Elbers, Lanjouw and Lanjouw (2002a), abbreviated with ELL, is implemented using data from Yemen. The idea is to measure consumption-based poverty at the disaggregated regional level by combing the information from the General Population, Housing, and Establishment Census in 2004 and the Household Budget Survey (HBS) in 2005-06 from Yemen. 2. Yemen has 21 governorates and 313 districts. The aim of this mission is to produce a poverty map at district level using the ELL method. The following tasks have been done to achieve this aim: · Select a set of variables that are common to the Census and the HBS, · Estimate models of household consumption per capita using HBS data for all the urban and rural areas, · Predict household consumption per capita using the Census data for all the urban and rural areas and estimate poverty indicators at district level. 3. The third reason is that solving the remaining inconsistencies may occasionally imply making imputations, which at this point ­ many miles and months away from the place and time where the data were collected ­ can only be made by guesswork. 4. In this report, section 2 provides a brief summary of the ELL method. Section 3 describes the data used. Section 4 describes the three tasks which have been implemented on the data and presents the results of the poverty estimates. Section 5 lists the remaining issues with the results. Poverty indicators are also estimated based on food consumption. The details of food poverty estimates are listed in Appendix A. II METHODOLOGY 5. The idea of the ELL method7 is to first estimate the joint distribution of y h , a variable on which the indicators of poverty are based, and a vector of variables x h using a smaller and richer sample (e.g. data from a survey). By restricting x h to be the variables on which a larger sample (e.g. data from a census) also provides 6 The author thanks P. Lanjouw from the World Bank for his guidance during all stages of this work, T. G. Srinivasan for providing access to the census and HBS data, and the staff of the Development Research Group at the World Bank, Washington D.C. for their help at the early stage of this work. 7 This paragraph is drawn from Elbers et. al. (2002b). 87 information, the distribution of y h for any sub-sample of the large sample can be generated by using the estimated distribution and the observed x h in the larger sample. This generated distribution of y h can then be used to generate the poverty indicators. The following is a brief summary of the method. A Consumption Model 6. Consumption per capita is often used to measure poverty. An estimated joint distribution of consumption per capita y h and a vector of observed variables x h is obtained using the ELL method by developing a linear model of y h on x h : ln y ch = xch + u ch , ' where y ch is the household consumption per capita for household h in location c, xch is the vector of explanatory variables, and u ch is an error term. It should be noted here that this model is only used for predicting y ch but not to measure the direct effect of xch on y ch , so the endogeneity of the explanatory variables is not of concern here. As the results of this model are going to be used to predict y ch in the census, it is preferred that the model fits most closely to the observations that represent a large part of the census population. Therefore population expansion factors are used as weights in this regression. The residual term u ch is defined as: u ch = c + ch , where c is a location component, and ch is a household component of the residual. The location component c is used to capture the part of the error term which is due to the location characteristics common to all households in that location. The household component of the residual ch reflects unobserved household characteristics which are not correlated with the location effect. 7. The variances of these two components of the error term reflect how much the household's predicted consumption deviates from its actual consumption. This deviation is one of the sources of the prediction error of the poverty indicators. The 88 idiosyncratic component ch falls approximately proportionately in sample size (Elbers et. al. 2002a), so for a large enough sample the idiosyncratic component of the error term does not cause serious problems to the precision of the estimates of poverty indicators. The location component c does not fall in sample size, so it is important to capture as much of the location effect in the consumption model as possible. One way to do this is to calculate the means of the observed variables (e.g. average level of education) at certain location level (e.g. enumeration area) using the census data, insert these variables into the survey data and use them as regressors in the consumption model. These variables of census means can often do a good job in capturing the location effect. 8. This consumption model is estimated using Generalized least squares (GLS). An Ordinary least squares (OLS) estimation is first performed to obtain the variance- ^ u ch from the OLS estimation can be covariance matrix of the error term. The residuals decomposed into two parts: u ch = u c. + (u ch - u c. ) = c + ech , ^ ^ ^ ^ ^ where a subscript "." indicates an average over that index. The variance of the location component can be estimated non-parametrically ^2 using c . ^ The component ech can be used to estimate the variances of ch . A logistic form is used in this estimation: 2 ech ln[ 2 ] = z ch + rch , T ^ A - ech 2 where zch are the variables which best explain variation in ech . In this way the prediction is bounded between zero and a maximum A. If A is set equal to (1.05) * max{ech } and B = exp{ z ch } , using the delta method the variance of ch is 2 T ^ estimated as: 89 AB 1 AB (1 - B) 2,ch = [ ^ ] + Var (r )[ ]. 1+ B 2 (1 + B ) 3 9. Once these two variances are calculated, they can be plugged into the variance- covariance matrix of the error term and the model can be estimated by GLS. B Poverty Indicators 10. The second set of tasks in the ELL method is to apply the estimates from the regression of the consumption model to the census data, predict the consumption from the census data and calculate the poverty indicators. 11. This task is done by simulation. For each simulation a vector of the parameters ~ is drawn from the multivariate normal distribution described by the GLS estimates of the consumption model and the associated variance-covariance matrix. The ~ location component of the error term c is drawn randomly with replacement from the ~ ~* set of c . To draw the household component ch , ch is first drawn for each ^ household with replacement from the set of all standardized residuals8, or from the standard residuals that correspond to the cluster from which the household's location ~ * × ,ch ^ effect is derived. The household component is then set to ch . For each ~ ~ ~ simulation, with the drawn values of , c , and ch , the value of per capita ^ consumption y ch is estimated as: ^ (' ~ ~ ~ y ch = exp x ch + c + ch ). ^ 12. Finally, the full vector of simulated consumption per capita y ch is used to calculate the mean and standard deviation of each poverty indicator. III DATA C Census data 13. The General Population, Housing, and Establishment Census was conducted by the Central Statistical Organization, Ministry of Planning & International Cooperation, Republic of Yemen in December 2004. The total number of households 8 Standardized residuals are calculated using the formula: ech 1 ech e* = -[ ,ch H ^ ch ,ch ^ ]. 90 covered in the census is 2,831,9299. For urban households, the administration contains six levels: governorate, district, sub-district, city, zone and neighborhood. The administration for rural households contains five levels: governorate, district, sub- district, village and sub-village. Table 1 lists the number of administrative levels in each governorate. All the administrative areas are then divided into 21,582 enumeration areas (EAs). Table 1 also lists the number of EAs in each governorate. We can see from Table 1 that for urban areas, the number of EAs is in between of the number of zones and neighborhoods for some governorates and smaller than the number of neighborhoods for other governorates. For rural areas the number of EAs is between the number of subdistricts and the number of villages. 14. Two kind of questionnaires are used in the census: the short questionnaire and the long questionnaire. The short questionnaire has seven components: housing unit properties, transport vehicles and durable goods, general & social data, data of disabled household members, married status and educational data. The long questionnaire is used for 10% of the households and it contains all the sections in the short one plus three sections: economic data, fertility data and mortality data. The long questionnaire provides richer information, but since it is only used by 10% of the households using the household level data from the long questionnaire often increases the standard errors of the estimates of the poverty indicators. In the case of Yemen the three extra sections covered in the long questionnaire provide little common information compared to the survey data. Therefore they are not used in generating variables at the household level. However, the economic data provided by the long questionnaire can be useful in predicting consumption and they can be used in generating variables of census means about average economic status at a certain location. D Survey Data 10 15. The Household Budget Survey 2005-06 was also conducted by the Central Statistical Organization of Yemen. The sample frame for the HBS was the 2004 General Population, Housing, and Establishment Census. Yemen consists of 21 governorates. The study population was sorted into 38 strata. The 17 governorates were represented by two strata (urban and rural), whereas Sana'a City and Aden are only urban and Remah and Sana'a Region are only rural. This resulted in 19 urban strata and 19 rural strata. 16. Within each stratum, the sample was selected in two stages. In the first stage, a certain number of Census Enumeration Areas (EAs) were selected with probability proportional to size (using as a measure of size the number of households according to the pre-census estimates available in January 2005). In the second stage, 12 households were picked from each EA by systematic equal probability sampling. 17. In order to produce estimates of consumption in all governorates of both rural and urban populations, the total sample of 1,200 EAs was distributed across strata by 9 Among these households, 231,565 households only contain data on dwelling because the houses were not occupied or the household does not have a household head. These households are dropped from later analysis. 10 The description of the design of the HBS is drawn from Godoy and Muñoz (2006). 91 a combination of allocation proportional to size and equal allocation. The final sample allocation is as shown in Table 2. 18. The HBS data contain information on household roster, activities, dwelling conditions, health, education, anthropometrics, income, durable goods and consumption. Among these, information on household roster, dwelling conditions, education and durable goods is also available in the census. IV IMPLEMENTATION E Select a Set of Variables that are Common to the Census and the HBS 19. All the common information in the census and the survey are listed in Table 3. Table 4 lists all the variables generated using the information listed in Table 3. The variables are in four categories: dwelling, durables, demography and education. Variables from these four categories are all very likely to be correlated to household consumption and can be good predictors of it. The high degree of comparability of selected variables is crucial for getting accurate estimates of poverty indicators. Two things need to be checked before a variable can be used as a candidate of the regressors in the consumption model. First, the questions in the questionnaires, on which a certain variable is based, must be truly the same in the census and the survey. This requires investigating the wording of questions in the questionnaires carefully. For example, in the Yemen case about the main source used for cooking, the survey questionnaire lists all the choices separately (i.e. 1 wood, 2 coal, 3 gas etc.) while the census questionnaire combines choices (e.g. 3 wood/coal/both, 4 wood/gas). One may want to generate a variable such as "the main source of cooking is coal", which will be equal to 1 if the answer is 2 in the survey and 3 in the census. Variables like this are problematic and should not be used in the analysis. Sometimes it is not very clear if the questions in the census and the survey are indeed the same, especially for the names of durable goods. Thus after being generated, the variables of durables were checked again to make sure that the names of the durables are the same in the census and the survey. Three variables were excluded (durable7, durable13 and durable14). 20. Second, the variables must have similar distributions in the census and the survey, such that the survey is representative. It is sometimes hard to judge if the distributions are similar. Experience from cases of other countries shows that the means of variables are the most important. In the analysis of the case of Yemen, means are also used as the most important property to judge if the distributions are the same. Whether the variables can have similar distributions depends mainly on the original design of the survey and the survey data. However, sometimes the way of generating the variables is also important. For example, for the durable goods one can generate integer variables "the number of a certain durable good owned by the household" or dummy variables "whether the household owns a certain durable good". It turns out that the later is a better idea. First, for most of the durables few households own more than one, so using the number of durables doesn't bring much more information than the dummy variables. Second the number of durables in general has a much wider range in the census than that in the survey; these outliers can change the mean and other properties of the distribution dramatically. Dummy variables only have values 0 and 1, so they do not have this problem. The distributions 92 of all the variables generated are compared in each stratum to make sure that a variable has similar distributions (especially means) in the census and the survey. The variables which pass this check are set as candidates which can be used as regressors in the consumption model. 21. As mentioned in section 2, unlike the idiosyncratic component the variance of the location component in the error term does not fall with the sample size. Thus it is essential to capture the location effect as much as possible in the consumption model. Variables of census means for each location are often used to achieve this. 22. Since households in the survey are drawn from the households in the EA of the census, it is natural to calculate the means at the EA level. In order to do this, it is required to be able to map each household in the survey and the census to the EAs. It is possible that the data do not provide enough information to do so. In such cases, the variables of census means can also be generated at other levels. For example, in the case of Yemen it is also reasonable to calculate the census means at the subdistrict level for the rural areas and at the zone level for the urban areas. In this case, one should be able to know which subdistricts/zones the households in the survey and the census belong to. Both EA level and subdistrict/zone level have been experimented with. It turns out that the former is better. First, for the urban areas variables of means calculated at EA level are much better in capturing the location effect. Second, for the rural areas the number of districts is not much more than the number of subdistricts for some governotes. Thus the variables of census means do not have much variation in some of the districts and these census means are often found to be able to change the estimates of poverty indicators dramatically. 23. Table 5 lists all the census means generated. It should be noted that these variables of census means can be generated not only for the variables which are common to both census and survey but also for the variables which only appear in the census but not in the survey. These variables of census means are then inserted into the survey. These variables in general have similar distributions in the census and in the survey but it is possible that for a certain variable the survey only covers the high/low range of the value of the variable. Thus it is also checked if the census means have similar means in the census and in the survey. Variables that pass this check are set to be the candidates of the regressors of the consumption model. F Estimate models of household consumption per capita using HBS data Consumption models are estimated for each stratum. Two criteria are used to evaluate the consumption models: the R square of the model and the ratio of the variance of the location effect to the variance of the error term. Table 6.1 to Table 6.38 show all the results of the regressions of the consumption models and the means and standard deviations of all the variables used in the models. For the 38 strata, the R squares vary from 0.38 to 0.70. The variables of census means calculated at EA level seem to capture the location effect well. The ratios of the variance of the location effect to the variance of the error term are below 0.078 for all strata except for rural Abyan. For some strata, the variables of census means can fully capture the location effect so it does not appear in the error term anymore. One principle of building these models is keeping the models simple. If a model can be built with reasonable R square and location effect without including interaction terms, interaction terms are not included. For strata with R squares lower than 0.40, interaction terms are included to get a better 93 fit. It is also noted that variables related to the size of households (e.g. hh_size, namales) sometimes contain outliers and bring high leverage to the model. If the variable (x) is not highly significant, it is deleted from the model. If it is highly significant, a variable equal to 1/(1+x) is generated to replace it. G Predict household consumption per capita using the census 24. The results of the consumption models shown in Table 6 are applied to predict household consumption per capita using the census data. The location effect is drawn semi-parametrically and the idiosyncratic component is drawn hierarchical semi-parametrically. If the simulated consumption per capita is higher than the highest consumption per capita or lower than the lowest one the survey, this draw is counted as missing and is not included in calculating the poverty indicators. The poverty line used is household specific. 25. I first produce the poverty indicators at the stratum level. Remember from the design of the survey that the survey is representative at this level, so the estimates calculated using the ELL method should be comparable to the estimates calculated directly from the survey data. Table 7 lists the estimates of headcount for each stratum using different data sources. For most of the strata, the estimates are close using the two data sources. Exceptions are estimates of rural Al-Jawf, rural Al-Maharh and urban Amran. The estimates of poverty indicators FGT0, FGT1 and FGT2 based on 100 simulations are listed in Table 8 and Table 9. The mean and plus/minus 2 standard errors of the estimates of headcount index are shown in Figure 1 and Figure 2 for rural and urban areas respectively. V CONCERNS ON CURRENT RESULTS 26. There is one main point of concerns. 27. For a large portion of the urban districts the number of households in each district is small. This causes high standard errors in the estimates of the poverty indicators. It should be borne in mind when using these estimates at district level that these estimates are noisy. Thus one should not make pairwise comparisons of poverty across districts without taking into account the statistical imprecision. VI REFERENCES Elbers, C., J. O. Lanjouw and P. Lanjouw. 2002a. "Micro-level Estimation of Welfare;." Working Paper No. 2911, The World Bank, Washington, D.C. Elbers, C., J. O. Lanjouw and P. Lanjouw. 2002b. "Micro-Level Estimation of Poverty and Inequality", Econometrica, 71(1): 355-64. Godoy, B. and J. Muñoz. 2006. "Preparation of the Household Budget Survey 2005-2006 databases for tabulation and analysis." Notes on Mission to Sana'a, Yemen; October 30 to November 8, 2006, The World Bank, Washington, D.C. 94 Table 1: Number of households, administrative areas and EAs in each governorate in census data Governorate Urban Rural households districts subdistricts cities zones neighb. EAs households districts subdistricts villages subvillages EAs 11 Ibb 54,126 17 19 19 23 356 371 259,492 20 251 2,717 17,208 2,054 12 Abyan 15,524 7 7 11 27 89 106 43,446 11 11 2,978 2,979 356 13 Sana'a City 260,825 12 12 12 89 791 1,637 4,971 1 3 52 172 41 14 Al-Baida 14,023 9 9 10 10 105 91 55,774 19 109 1,478 3,171 441 15 Taiz 88,474 16 17 17 17 317 621 304,262 20 233 1,983 16,407 2,286 16 Al-Jawf 7,236 11 11 11 11 121 44 49,230 12 47 481 2,466 293 17 Hajja 17,275 19 19 20 20 216 135 169,586 31 161 3,780 13,830 1,503 18 Al-Hodeida 120,603 24 33 34 34 283 803 246,919 24 135 2,298 5,796 1,878 19 Hadramout 66,375 24 25 30 82 149 428 75,605 30 37 3,837 3,847 550 20 Dhamar 25,879 8 8 8 8 134 183 173,069 12 312 3,373 13,419 1,416 21 Shabwah 9,637 11 11 11 19 43 74 43,412 17 24 3,337 3,540 398 22 Sa'adah 12,924 11 13 13 13 157 91 68,529 15 121 1,194 6,438 606 23 Sana'a Reg. 3,653 9 10 10 10 67 29 112,119 16 145 2,156 7,218 913 24 Aden 97,289 8 8 8 44 242 633 0 0 0 0 0 0 25 Laheg 9,720 9 9 10 10 110 65 104,882 14 40 4,124 5,840 757 26 Mareb 3,962 5 5 5 5 29 29 24,029 14 59 467 2,162 208 27 Al-Mahweet 4,674 6 6 6 6 71 35 60,849 9 114 1,213 4,647 531 28 Al-Maharh 5,220 6 6 6 18 51 40 7,636 9 12 367 369 75 29 Amran 18,728 15 16 16 16 179 152 80,408 20 125 1,629 5,707 753 30 Al-Dhale 8,445 8 8 8 8 140 62 52,640 9 41 1,688 2,900 409 31 Remah 617 3 3 3 3 31 5 49,862 6 89 737 6,679 480 95 Table 2: HBS final sample allocation No. of clusters No. of households Governorate Urban Rural Total Urban Rural Total 11 Ibb 43 41 84 516 492 1,008 12 Abyan 30 18 48 360 216 576 13 Sana'a City 156 0 156 1,872 0 1,872 14 Al-Baida 29 19 48 348 228 576 15 Taiz 56 40 96 672 480 1152 16 Al-Jawf 22 14 36 264 168 432 17 Hajja 30 30 60 360 360 720 18 Al-Hodeida 75 33 108 900 396 1,296 19 Hadramout 41 19 60 492 228 720 20 Dhamar 31 29 60 372 348 720 21 Shabwah 21 15 36 252 180 432 22 Sa'adah 28 20 48 336 240 576 23 Sana'a Region 0 24 24 0 288 288 24 Aden 72 0 72 864 0 864 25 Laheg 25 23 48 300 276 576 26 Mareb 22 14 36 264 168 432 27 Al-Mahweet 27 21 48 324 252 576 28 Al-Maharh 12 12 24 144 144 288 29 Amran 27 21 48 324 252 576 30 Al-Dhale 22 14 36 264 168 432 31 Remah 0 24 24 0 288 288 Total 769 431 1,200 9,228 5,172 14,400 Source: Figure 3, Godoy and Muñoz (2006) 96 Table 3: Common information in the census and the survey Census (question No. & code) Survey (question No. & code) Dwelling Type of house Section 2, 201 Section 3, 301 house/villa 1: house 1: house; 3: villa apartment 2: apartment 2: apartment hut 6: hut; 7: tin hut 6: hut tent 8: tent 7: tent establishment 3: establishment for 4: habitable establishment accommodation Main method of water Section 2, 206 Section 3, 305 supply public network 1: public network 1: public network private network 2: private network 3: private network cooperative network 3: cooperative network 2: cooperative network Main source of water Section 2, 205 Section 3, 306 supply well 1: deep well; 2: well; 4: covered 1: erteslan well; 2: normal well well rain collection 7: rooftop water harvest 7: traditional way in collecting rain Type of sewage system Section 2, 207 Section 3, 312 public network 1: public network 1: public network close pot 2: covered pit 2: close pot open pot 3: open pit 3: open pot Main source of lighting Section 2, 208 Section 3, 315 public network 1: public network 1: public network cooperative network 3: cooperative network 2: cooperative network private network 2: private network 3: private network private generator 4: private generator 4: house generator kerosene 5: kerosene 5: kerosene lantern gas 6: gas 6: gas lamp Main source for cooking Section 2, 209 Section 3, 319 electricity 5: electricity 5: electricity kerosene 2: kerosene 4: kerosene Legal status of the Section 2, 203 Section 3, 324 dwelling own 1: own 2: own rent 2: rented 1: rented Durables Durable goods Section 3, 301, 302 Section 12, 1201 private cars 301, 1: private car 1: private car taxi's 301, 2: taxi 2: taxi buses 301, 3: bus 3: autobus small trucks 301, 4: small truck 5: small truck large trucks 301, 5: large truck 6: truck motor bikes 301, 6: motorbike 8: motor bike mixers 301, 8: concrete mixer 10: mixer phones 302, 1: house phone 21: telephone mobiles 302, 2: mobile 22: mobile telephone refrigerators 302, 3: refrigerator 11: refrigerator 97 Table 3: Common information in the census and the survey (Continued) Census (question No. & code) Survey (question No. & code) washing machines 302, 4: washing machine 12: washing machine TVs 302, 7: TV 17: color TV; 18: black TV radios 302, 10: radio 16: radio/cassette recorder water heaters 302, 11: water heater 14: electrical water worm; 15: sun water worm sewing machines 302, 12: sewing machine 23: sewing machine PCs 302, 6: PC 27: PC satellite dishes 302, 8: satellite dish 20: satellite dish air conditioners 302, 9: air conditioner 26: air conditioner Demography Age of household Section 4, 408 Section 1, 103 members Relation of household Section 4, 406 Section 1, 105 members to head head 1: HH head 0: head spouse 2: HH head spouse 1: spouse Sex of household Section 4, 407 Section 1, 102 members male 1: male 1: male female 2: female 2: female Marital status of Section 6, 601 Section 1, 109 household members married 1: single 1: single single 2: married 2: married divorced 3: divorced 3: divorced widowed 4: widowed 4: widowed Education Education level of Section 7, 704 Section 5, 506 household members illiterate 1: illiterate 1: never read and write read/write 2: read/write 2: read and write primary 3: primary 3: primary university 8: university 10: university degree 98 Table 4: Household level variables generated using information common in both the census and the survey Variable name Definition housetype1 The type of the house of the household is house/villa housetype2 The type of the house of the household is apartment housetype3 The type of the house of the household is hut housetype4 The type of the house of the household is tent housetype5 The type of the house of the household is habitable establishment water1 The main source water supply is public network water2 The main source water supply is private network water3 The main source water supply is cooperative network sewage1 The type of sewage disposal system is public network sewage2 The type of sewage disposal system is close pot sewage3 The type of sewage disposal system is open pot light1 The main source of lighting is public network light2 The main source of lighting is cooperative network light3 The main source of lighting is private network light4 The main source of lighting is private generator light5 The main source of lighting is kerosene light6 The main source of lighting is gas cook1 The main source used for cooking is electricity cook2 The main source used for cooking is kerosene ownhouse1 The household owns the house ownhouse2 The household rents the house dum_durable1 The household owns private car(s) dum_durable2 The household owns taxi('s) dum_durable3 The household owns bus(es) dum_durable4 The household owns small truck(s) dum_durable5 The household owns large truck(s) dum_durable6 The household owns motor bike(s) dum_durable7* The household owns mixer(s) dum_durable8 The household owns phone(s) dum_durable9 The household owns mobile(s) dum_durable10 The household owns refrigerator(s) dum_durable11 The household owns washing machine(s) dum_durable12 The household owns TV(s) dum_durable13* The household owns radio(s) dum_durable14* The household owns water heater(s) dum_durable15 The household owns sewing machine(s) dum_durable16 The household owns PC(s) dum_durable17 The household owns satellite dish(es) dum_durable18 The household owns air conditioner(s) headage Age of the head spouseage Mean age of spouses of the head spouseno No. of spouses of the head hh_size Size of the household namales No. of adult males in the household (15age<60) nafemales No. of adult females in the household (15age<60) nkids No. of kids in the household (age<15) nelderly No. of elderlys in the household (age60) malep Percentage of males in the household femalep Percentage of females in the household amalep Percentage of adult males in the household afemalep Percentage of adult females in the household 99 Table 4: Household level variables generated using information common in both the census and the survey (continued) Variable name Definition kidp Percentage of kids in the household elderlyp Percentage of elderlys in the household marriedp Percentage of married people in the household singlep Percentage of single people in the household divorcedp Percentage of divorced people in the household widowp Percentage of widows in the household illiterp Percentage of illiterate members in the household primaryp Percentage of members who finish primary school in the household universityp Percentage of members with university diploma in the household headilliter The head is illiterate headread The head can only read and write headprim The head's highest education level is primary school headsecond The head's highest education level is higher than primary school and lower than university headuniv The head's highest education level is university or higher highilliter The person with highest level of education in the household is illiterate highread The person with highest level of education in the household can only read and write highprim The highest education level of the most well-educated person in the household is primary school highsecond The highest education level of the most well-educated person in the household is higher than primary school and lower than university highuniv The highest education level of the most well-educated person in the household is university * The variable is dropped from later analysis. Table 5: Variables of Census means Variable name Definition housetype1_ea The percentage of households whose type of the houses are house/villa housetype2_ea The percentage of households whose type of the houses are apartment housetype3_ea The percentage of households whose type of the houses are hut housetype4_ea The percentage of households whose type of the houses are tent housetype5_ea The percentage of households whose type of the houses are habitable establishment water1_ea The percentage of households whose main source of water supply is public network water2_ea The percentage of households whose main source of water supply is private network water3_ea The percentage of households whose main source of water supply is cooperative network sewage1_ea The percentage of households whose type of sewage disposal system is public network sewage2_ea The percentage of households whose type of sewage disposal system is close pot sewage3_ea The percentage of households whose type of sewage disposal system is open pot light1_ea The percentage of households whose main source of lighting is public network 100 Table 5: Variables of Census means (Continued) Variable name Definition light2_ea The percentage of households whose main source of lighting is cooperative network light3_ea The percentage of households whose main source of lighting is private network light4_ea The percentage of households whose main source of lighting is private generator light5_ea The percentage of households whose main source of lighting is kerosene light6_ea The percentage of households whose main source of lighting is gas cook1_ea The percentage of households whose main source used for cooking is electricity cook2_ea The percentage of households whose main source used for cooking is kerosene ownhouse1_ea The percentage of households who own their houses ownhouse2_ea The percentage of households who rent their houses dum_eaurable1_ea Percentage of households which own private car(s) dum_eaurable2_ea Percentage of households which own taxi('s) dum_eaurable3_ea Percentage of households which own bus(es) dum_eaurable4_ea Percentage of households which own small truck(s) dum_eaurable5_ea Percentage of households which own large truck(s) dum_eaurable6_ea Percentage of households which own motor bike(s) dum_eaurable8_ea Percentage of households which own phone(s) dum_eaurable9_ea Percentage of households which own mobile(s) dum_eaurable10_ea Percentage of households which own refrigerator(s) dum_eaurable11_ea Percentage of households which own washing machine(s) dum_eaurable12_ea Percentage of households which own TV(s) dum_eaurable15_ea Percentage of households which own sewing machine(s) dum_eaurable16_ea Percentage of households which own PC(s) dum_eaurable17_ea Percentage of households which own satellite dish(es) dum_eaurable18_ea Percentage of households which own air conditioner(s) headilliter_ea Percentage of households whose heads are illiterate headread_ea Percentage of households whose heads can only read and write headprim_ea Percentage of households whose heads only finish primary school headsecond_ea Percentage of households whose heads' highest education level is higher than primary school and lower than university headuniv_ea Percentage of households whose heads' highest education level is university or higher Illiter_ea Percentage of people who are illietrate primary_ea Percentage of people whose education level is primary school university_ea Percentage of people who finish university employed1_ea* Percentage of people who worked in the month before the census employed2_ea* Percentage of people who have worked before employ_nonself_ea* Percentage of prople who were emplyed by somebody else employ_self_ea* Percentage of prople who were self-emplyed work1_ea* Percentage of people who worked for gov. est./co. work2_ea* Percentage of people who worked for mixed est./co. work3_ea* Percentage of people who worked for private est./co. work4_ea* Percentage of people who worked for store/workshop/office work5_ea* Percentage of people who worked at home work6_ea* Percentage of people who worked as sidewalk salesman work7_ea* Percentage of people who worked as roaming salesman work8_ea* Percentage of people who worked as private construction site work9_ea* Percentage of people who worked as private farm work10_ea* Percentage of people who worked as private transport vehicle * The variable is generated using section 8 (economic data) of the long questionnaire. 101 Table 6: Regression results of consumption models Table 6.1: Rural Ibb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_durable1 0.583 0.105 0.043 0.202 0.044 0.042 dum_durable1_ea 2.674 0.519 0.043 0.051 0.046 0.002 employed1 -0.846 0.242 0.220 0.094 0.223 0.010 employ_nonself 0.384 0.114 0.295 0.236 0.332 0.054 headread_d 1.433 0.285 0.156 0.097 0.157 0.008 headsecond_ea -1.131 0.344 0.117 0.068 0.138 0.007 highsecond 0.129 0.050 0.352 0.478 0.363 0.232 housetype2 -0.308 0.172 0.019 0.137 0.014 0.014 housetype3_ea 1.681 0.517 0.013 0.047 0.014 0.002 light5 -0.294 0.049 0.447 0.497 0.420 0.244 nelderly -0.104 0.035 0.398 0.660 0.370 0.404 ownhouse1_ea 0.712 0.164 0.839 0.135 0.851 0.017 primaryp 0.485 0.166 0.110 0.160 0.119 0.022 singlep -0.886 0.109 0.588 0.231 0.595 0.044 _intercept_ 11.110 0.177 obs. 463 R square 0.42 location effect 0.083 Table 6.2: Rural Abyan Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.323 0.121 0.080 0.272 0.062 0.058 dum_eaurable10 0.217 0.101 0.222 0.415 0.252 0.189 dum_eaurable12 0.115 0.074 0.396 0.489 0.418 0.244 dum_eaurable17 0.181 0.116 0.088 0.283 0.090 0.082 housetype1 -0.160 0.084 0.857 0.350 0.862 0.120 kidp -0.585 0.158 0.383 0.230 0.376 0.055 ownhouse1_ea 1.384 0.716 0.932 0.107 0.931 0.003 primaryp 0.543 0.246 0.133 0.181 0.115 0.019 singlep -0.654 0.168 0.587 0.213 0.576 0.048 water1 -0.311 0.090 0.140 0.347 0.150 0.128 _intercept_ 10.497 0.713 obs. 201 R square 0.40 location effect 0.148 102 Table 6.3: Rural Al-Baida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 0.594 0.201 0.238 0.157 0.226 0.020 dum_eaurable11 0.328 0.101 0.110 0.313 0.101 0.091 dum_eaurable11_ea 3.854 0.670 0.110 0.166 0.097 0.024 dum_eaurable14_ea -4.819 0.734 0.055 0.097 0.052 0.008 dum_eaurable3_ea -20.899 4.069 0.008 0.016 0.008 0.000 employed1 -9.877 1.380 0.245 0.094 0.263 0.010 employed2 13.265 1.734 0.261 0.092 0.281 0.009 headread_ea 1.209 0.268 0.226 0.127 0.202 0.014 headsecond_ea 5.487 0.755 0.099 0.068 0.097 0.006 headsingle 0.265 0.143 0.046 0.209 0.046 0.044 housetype1_ea 1.290 0.355 0.938 0.101 0.902 0.024 light5 -0.417 0.085 0.178 0.383 0.175 0.145 nafemales -0.038 0.020 2.193 1.541 2.207 1.931 nelderly -0.099 0.043 0.439 0.696 0.454 0.485 ownhouse1_ea -1.589 0.438 0.906 0.080 0.895 0.020 singlep -1.096 0.154 0.616 0.193 0.615 0.037 water4_ea -0.387 0.145 0.913 0.180 0.901 0.047 work7 -1.466 0.325 0.069 0.132 0.075 0.012 _intercept_ 10.546 0.477 obs. 222 R square 0.49 location effect -* *No location effect. Table 6.4: Rural Taiz Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12 0.176 0.049 0.332 0.471 0.346 0.227 dum_eaurable1_ea -5.475 1.104 0.026 0.029 0.020 0.000 dum_eaurable4 0.243 0.131 0.018 0.133 0.021 0.021 femalep -0.278 0.122 0.541 0.213 0.546 0.040 headiliter -0.156 0.046 0.667 0.471 0.694 0.213 kidp -0.898 0.082 0.409 0.255 0.414 0.064 light1 -0.348 0.079 0.194 0.395 0.151 0.128 light5 -0.333 0.063 0.654 0.476 0.698 0.211 nafemales -0.063 0.019 1.835 1.325 2.022 1.984 namales -0.089 0.020 1.387 1.321 1.405 1.791 sewage2 0.146 0.046 0.315 0.464 0.340 0.225 water2 0.336 0.079 0.060 0.237 0.070 0.065 work1 1.024 0.156 0.185 0.205 0.187 0.039 work10 2.326 0.455 0.038 0.079 0.030 0.003 work4 1.367 0.174 0.138 0.187 0.138 0.031 work5 -0.805 0.348 0.019 0.059 0.021 0.004 work6 0.905 0.279 0.035 0.084 0.034 0.007 work8 0.496 0.142 0.142 0.173 0.179 0.044 work9 1.037 0.155 0.209 0.234 0.202 0.040 _intercept_ 11.646 0.150 obs. 450 R square 0.47 location effect 0.067 103 Table 6.5: Rural Al-Jawf Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable18 1.386 0.371 0.001 0.037 0.008 0.008 dum_eaurable8_ea -1.690 0.996 0.012 0.044 0.013 0.001 headsecond_ea 0.989 0.349 0.138 0.136 0.124 0.009 housetype1_ea 0.444 0.109 0.733 0.292 0.698 0.080 light5 -0.181 0.076 0.732 0.443 0.717 0.204 marriedp 1.019 0.158 0.299 0.125 0.320 0.039 nafemales -0.056 0.027 1.766 1.220 1.787 1.287 _intercept_ 10.610 0.125 obs. 148 R square 0.42 location effect -* *No location effect. Table 6.6: Rural Hajja Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) cook2_ea -0.772 0.190 0.054 0.181 0.038 0.018 dum_eaurable15_ea 2.463 0.511 0.030 0.079 0.035 0.008 dum_eaurable1 0.334 0.093 0.043 0.203 0.060 0.056 dum_eaurable5 0.560 0.261 0.011 0.105 0.006 0.006 dum_eaurable6 0.325 0.152 0.014 0.116 0.021 0.021 dum_eaurable9_ea -2.330 0.542 0.093 0.106 0.087 0.011 employed1 -0.707 0.331 0.258 0.112 0.277 0.007 employ_nonself 1.075 0.310 0.123 0.165 0.143 0.040 employ_self 1.698 0.350 0.829 0.193 0.794 0.049 headage -0.006 0.002 41.659 14.961 42.337 228.559 headiliter_ea 0.535 0.315 0.709 0.158 0.697 0.026 headprim_ea 6.559 1.019 0.049 0.044 0.050 0.002 headread_ea -2.035 0.577 0.121 0.092 0.128 0.006 headsecond_ea 1.570 0.534 0.138 0.112 0.153 0.008 highprim -0.138 0.075 0.068 0.252 0.095 0.086 kidp -0.873 0.143 0.458 0.242 0.452 0.053 light1 0.410 0.126 0.078 0.268 0.089 0.081 light1_ea -0.924 0.207 0.076 0.237 0.072 0.050 light4_ea 1.113 0.546 0.032 0.081 0.034 0.006 nafemales -0.099 0.021 1.844 1.554 1.709 1.352 namales -0.068 0.019 1.788 1.428 1.809 1.782 sewage2 0.544 0.075 0.080 0.271 0.110 0.098 singlep -0.572 0.154 0.598 0.225 0.604 0.049 work5 3.420 0.680 0.016 0.050 0.021 0.002 work7 -0.888 0.334 0.052 0.105 0.049 0.013 work9 -0.656 0.187 0.533 0.278 0.542 0.089 _intercept_ 10.949 0.476 obs. 346 R square 0.58 location effect 0.055 104 Table 6.7: Rural Al-Hodeida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 0.370 0.172 0.256 0.187 0.264 0.020 amalep 0.473 0.157 0.261 0.171 0.257 0.022 dum_eaurable4 0.246 0.144 0.021 0.144 0.022 0.022 dum_eaurable5 0.461 0.192 0.006 0.075 0.012 0.012 dum_eaurable6 0.240 0.070 0.063 0.243 0.095 0.086 employed1 -7.410 1.490 0.328 0.115 0.346 0.016 employed2 8.002 1.430 0.342 0.113 0.356 0.017 headiliter_ea -0.923 0.312 0.786 0.123 0.773 0.020 headread_ea -1.392 0.407 0.106 0.069 0.132 0.015 headsingle 0.507 0.130 0.034 0.181 0.027 0.026 light1 0.646 0.188 0.035 0.184 0.033 0.032 light3_ea 1.965 0.286 0.022 0.102 0.030 0.020 light5_ea 1.237 0.200 0.862 0.243 0.890 0.043 marriedp 0.874 0.118 0.404 0.269 0.402 0.049 ownhouse2 0.712 0.280 0.008 0.091 0.002 0.002 water3 0.219 0.065 0.187 0.390 0.142 0.122 water3_ea 0.171 0.081 0.185 0.339 0.152 0.086 water4_ea -0.323 0.146 0.911 0.206 0.929 0.037 work7 -0.595 0.224 0.076 0.118 0.072 0.012 _intercept_ 10.372 0.281 obs. 377 R square 0.47 location effect 0.035 Table 6.8: Rural Hadramout Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11 0.152 0.047 0.223 0.416 0.298 0.210 dum_eaurable14_ea -1.733 0.266 0.113 0.182 0.131 0.027 dum_eaurable1 0.186 0.049 0.236 0.424 0.259 0.193 dum_eaurable2_ea -5.818 1.852 0.013 0.023 0.015 0.000 dum_eaurable4 0.476 0.101 0.056 0.231 0.052 0.049 dum_eaurable9_ea 0.861 0.258 0.197 0.183 0.202 0.025 headdivorced -0.337 0.177 0.017 0.129 0.008 0.008 kidp -0.646 0.180 0.398 0.220 0.397 0.046 light1_ea 0.456 0.110 0.414 0.451 0.478 0.195 light4 0.243 0.092 0.057 0.233 0.048 0.046 namales 0.094 0.026 2.349 2.129 2.422 2.726 nkids 0.125 0.025 4.014 3.289 3.945 12.955 singlep -0.454 0.129 0.568 0.208 0.580 0.037 hh_size -0.118 0.017 9.199 5.868 9.574 34.513 water1 -0.507 0.084 0.307 0.461 0.292 0.208 water3 0.189 0.076 0.088 0.283 0.064 0.060 work1 -0.264 0.100 0.129 0.145 0.166 0.058 _intercept_ 12.218 0.099 obs. 203 R square 0.58 105 location effect -* *No location effect. Table 6.9: Rural Dhamar Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep -0.504 0.217 0.213 0.167 0.232 0.023 dum_eaurable11 0.406 0.142 0.025 0.155 0.040 0.038 dum_eaurable3 -0.506 0.278 0.002 0.048 0.005 0.005 dum_eaurable4 0.265 0.119 0.026 0.160 0.038 0.037 dum_eaurable8 0.191 0.079 0.083 0.275 0.134 0.116 employed1 6.683 2.156 0.282 0.124 0.272 0.010 employed2 -6.017 2.280 0.290 0.123 0.279 0.009 employ_nonself -0.926 0.223 0.263 0.242 0.297 0.068 employ_self -0.758 0.244 0.700 0.249 0.663 0.061 headprim_ea 2.978 0.958 0.038 0.043 0.037 0.001 headread_ea 0.474 0.219 0.150 0.108 0.171 0.017 kidp -0.902 0.139 0.439 0.239 0.440 0.053 light4 0.416 0.201 0.023 0.151 0.011 0.011 marriedp 0.343 0.135 0.371 0.229 0.366 0.041 nelderly -0.141 0.039 0.436 0.699 0.462 0.498 primaryp 0.465 0.221 0.092 0.147 0.104 0.016 work3 -2.454 0.758 0.010 0.045 0.010 0.001 _intercept_ 12.337 0.247 obs. 315 R square 0.40 location effect 0.003 Table 6.10: Rural Shabwah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.411 0.089 0.256 0.437 0.245 0.186 light5_ea -0.652 0.116 0.265 0.361 0.261 0.125 light6 -0.315 0.124 0.096 0.295 0.117 0.104 ownhouse2_ea 8.078 2.105 0.021 0.061 0.018 0.000 singlep -0.705 0.226 0.615 0.184 0.616 0.026 hh_size -0.024 0.007 10.481 6.339 10.435 32.939 water1_ea -0.795 0.283 0.047 0.173 0.040 0.017 _intercept_ 11.927 0.155 obs. 151 R square 0.69 location effect 0.015 106 Table 6.11: Rural Sa'adah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 1.113 0.269 0.250 0.172 0.238 0.025 dum_eaurable10_ea 1.477 0.783 0.020 0.045 0.023 0.003 dum_eaurable12 0.163 0.058 0.306 0.461 0.352 0.229 dum_eaurable12_ea -0.476 0.185 0.306 0.265 0.323 0.054 dum_eaurable1 0.228 0.076 0.197 0.398 0.166 0.139 dum_eaurable4 0.268 0.085 0.085 0.279 0.101 0.091 employed1 -3.431 1.029 0.309 0.118 0.330 0.018 employed2 3.454 1.046 0.320 0.117 0.342 0.017 femalehead 0.470 0.178 0.054 0.225 0.026 0.025 headiliter_ea -1.242 0.327 0.734 0.184 0.748 0.022 headprim_ea -3.218 1.210 0.048 0.043 0.049 0.002 kidp -0.658 0.214 0.452 0.222 0.450 0.048 marriedp 0.371 0.160 0.347 0.199 0.363 0.037 namales -0.094 0.032 2.153 1.784 2.050 2.286 nkids 0.043 0.019 4.522 3.544 4.309 8.128 work10 -3.715 0.767 0.023 0.049 0.026 0.002 work9 0.444 0.162 0.612 0.267 0.605 0.061 _intercept_ 12.181 0.321 obs. 218 R square 0.41 location effect 0.068 Table 6.12: Rural Sana'a Region Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.238 0.058 0.141 0.348 0.162 0.136 dum_eaurable1_ea 0.367 0.124 0.141 0.161 0.157 0.032 dum_eaurable5 0.382 0.117 0.026 0.160 0.033 0.032 dum_eaurable6 0.255 0.105 0.027 0.163 0.041 0.039 dum_eaurable9 0.201 0.050 0.176 0.381 0.207 0.165 headiliter_ea -0.739 0.139 0.610 0.180 0.646 0.032 kidp -0.831 0.100 0.423 0.223 0.420 0.044 light1 0.230 0.046 0.497 0.500 0.478 0.251 nafemales -0.082 0.016 2.078 1.607 2.151 2.140 namales -0.059 0.016 2.091 1.863 2.052 2.195 primaryp 0.675 0.163 0.127 0.163 0.106 0.017 _intercept_ 12.189 0.130 obs. 256 R square 0.65 location effect 0.049 107 Table 6.13: Rural Laheg Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12 0.159 0.092 0.364 0.481 0.397 0.240 dum_eaurable12_ea -0.901 0.187 0.364 0.337 0.388 0.124 dum_eaurable14 -0.937 0.279 0.026 0.159 0.010 0.010 dum_eaurable14_ea -2.644 0.939 0.026 0.095 0.018 0.004 dum_eaurable1 0.488 0.166 0.063 0.242 0.043 0.042 dum_eaurable1_ea 3.506 0.572 0.063 0.091 0.057 0.008 dum_eaurable8_ea 2.164 0.517 0.099 0.198 0.081 0.028 elderlyp 1.160 0.217 0.091 0.204 0.094 0.039 employed1 5.318 1.253 0.207 0.095 0.214 0.011 employed2 -6.212 1.337 0.217 0.096 0.227 0.011 headread_ea 1.114 0.282 0.232 0.121 0.205 0.016 headuniv_ea 3.406 0.957 0.030 0.036 0.031 0.002 kidp -0.309 0.148 0.380 0.246 0.371 0.057 light5 -0.379 0.128 0.464 0.499 0.455 0.249 light6 -0.541 0.175 0.089 0.285 0.070 0.066 marriedp 0.427 0.143 0.358 0.238 0.358 0.046 nelderly -0.193 0.059 0.431 0.702 0.424 0.412 water3_ea 0.807 0.259 0.032 0.138 0.033 0.021 work1 -0.792 0.174 0.387 0.296 0.384 0.071 work4 -2.711 0.414 0.120 0.186 0.103 0.028 _intercept_ 11.736 0.223 obs. 246.000 R square 0.44 location effect -* *No location effect Table 6.14: Rural Mareb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11_ea -1.455 0.777 0.091 0.138 0.077 0.017 dum_eaurable13_ea 0.678 0.132 0.509 0.295 0.463 0.081 dum_eaurable17 0.425 0.156 0.061 0.239 0.053 0.051 dum_eaurable1 0.289 0.093 0.189 0.392 0.208 0.166 employed2 -1.061 0.265 0.207 0.110 0.200 0.018 headage -0.009 0.003 43.282 14.265 41.665 148.410 headiliter_ea -3.714 0.794 0.635 0.174 0.646 0.044 headread_ea -6.078 1.147 0.112 0.089 0.109 0.008 headsecond_ea -3.330 1.026 0.204 0.127 0.193 0.011 headuniv 0.472 0.236 0.030 0.036 0.020 0.019 marriedp 0.544 0.216 0.310 0.177 0.337 0.040 nkids -0.046 0.017 4.436 3.386 4.334 6.300 ownhouse1_ea -2.788 0.706 0.881 0.144 0.895 0.016 primaryp 0.846 0.214 0.134 0.175 0.132 0.036 work7 2.985 0.786 0.024 0.071 0.022 0.003 _intercept_ 17.631 1.207 obs. 158 R square 0.70 location effect 0.024 108 Table 6.15: Rural Al-Mahweet Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10 0.355 0.108 0.038 0.190 0.056 0.053 dum_eaurable13_ea 0.187 0.096 0.658 0.241 0.620 0.060 dum_eaurable15 0.259 0.109 0.037 0.189 0.050 0.048 dum_eaurable2 -0.636 0.353 0.007 0.083 0.004 0.004 dum_eaurable6 0.635 0.340 0.003 0.056 0.004 0.004 headiliter_ea -0.558 0.236 0.715 0.136 0.725 0.013 headmarried -0.687 0.096 0.893 0.309 0.914 0.079 headuniv 0.336 0.154 0.032 0.176 0.026 0.025 light6_ea -0.178 0.080 0.113 0.230 0.113 0.077 marriedp 0.978 0.116 0.360 0.221 0.380 0.051 primaryp 0.692 0.200 0.091 0.147 0.080 0.014 water3_ea -0.243 0.121 0.060 0.207 0.041 0.034 work8 0.315 0.155 0.139 0.168 0.131 0.025 _intercept_ 11.755 0.203 obs. 249 R square 0.41 location effect 0.005 Table 6.16: Rural Al-Maharh Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1×nelderlyinv* 0.424 0.111 0.141 0.107 0.180 0.127 dum_eaurable13_ea -0.834 0.173 0.332 0.066 0.392 0.054 dum_eaurable4_0 ×nelderlyinv -0.544 0.164 0.770 0.103 0.832 0.086 employed1 16.003 5.451 0.309 0.026 0.282 0.018 employed2 -14.516 5.455 0.313 0.026 0.286 0.019 light4_ea×nafemalesinv -1.152 0.373 0.029 0.007 0.046 0.013 nafemalesinv×nelderlyinv 0.729 0.323 0.379 0.050 0.359 0.025 nkids -0.086 0.017 3.145 7.531 3.104 5.712 _intercept_ 12.356 0.178 obs. 132 R square 0.42 location effect - *nelderlyinv=1/(1+nelderly); dum_durable4_0=1-dum_durable4; nafemalesinv=1/(1+nafemales). 109 Table 6.17: Rural Amran Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10 0.478 0.151 0.014 0.117 0.036 0.035 dum_eaurable15_ea 0.520 0.151 0.134 0.181 0.143 0.031 dum_eaurable2 -0.367 0.221 0.014 0.014 0.009 0.009 dum_eaurable9_ea -0.361 0.146 0.207 0.151 0.232 0.036 headsecond 0.128 0.067 0.150 0.357 0.144 0.124 light5 -0.306 0.063 0.403 0.490 0.383 0.237 nelderly -0.085 0.035 0.473 0.747 0.460 0.504 ownhouse1_ea -0.805 0.376 0.946 0.226 0.924 0.005 singlep -1.013 0.111 0.607 0.187 0.585 0.047 water2_ea 0.777 0.171 0.034 0.117 0.044 0.029 water3 -0.178 0.118 0.045 0.206 0.043 0.042 work4 -2.000 0.596 0.038 0.070 0.033 0.003 work6 -1.908 0.559 0.016 0.048 0.021 0.003 _intercept_ 12.653 0.351 obs. 224 R square 0.46 location effect -* *No location effect Table 6.18: Rural Al-Dhale Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1×sizeinv* 3.271 1.156 0.012 0.001 0.007 0.001 dum_eaurable10×sizeinv 2.528 1.019 0.010 0.002 0.010 0.001 dum_eaurable10_ea -0.725 0.281 0.093 0.020 0.106 0.030 dum_eaurable17 0.217 0.095 0.151 0.358 0.179 0.148 headsingle 0.427 0.173 0.043 0.041 0.031 0.030 headuniv 0.413 0.144 0.031 0.173 0.032 0.031 highsecond×sizeinv 1.514 0.507 0.057 0.006 0.056 0.005 housetype1×sizeinv 1.370 0.912 0.134 0.028 0.130 0.005 light1_ea -0.196 0.118 0.331 0.189 0.311 0.193 light5 -0.231 0.084 0.418 0.493 0.412 0.244 marriedp×sizeinv 1.957 0.900 0.054 0.005 0.055 0.006 sewage3_ea 0.301 0.100 0.250 0.433 0.277 0.117 water2 -0.615 0.197 0.045 0.208 0.025 0.025 _intercept_ 10.974 0.114 obs. 156 R square 0.40 location effect - *sizeinv=1/(1+hh_size); no location effect. 110 Table 6.19: Rural Remah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12 0.177 0.087 0.096 0.294 0.121 0.106 dum_eaurable15_ea 3.433 1.463 0.015 0.051 0.014 0.000 dum_eaurable17 -0.612 0.254 0.010 0.099 0.011 0.011 dum_eaurable4 0.758 0.181 0.007 0.085 0.019 0.019 dum_eaurable8_ea -5.708 1.331 0.013 0.025 0.016 0.001 dum_eaurable9_ea -1.363 0.433 0.069 0.086 0.067 0.005 headdivorced 0.519 0.221 0.011 0.104 0.011 0.011 headiliter_ea -0.699 0.229 0.703 0.155 0.722 0.022 headread_ea -0.486 0.272 0.183 0.116 0.192 0.018 headsingle 0.418 0.150 0.029 0.169 0.028 0.027 light4 0.591 0.178 0.026 0.160 0.026 0.025 light5_ea -0.591 0.123 0.806 0.268 0.817 0.069 light6 0.245 0.107 0.078 0.269 0.052 0.049 ownhouse2 0.341 0.177 0.039 0.195 0.018 0.018 singlep -1.047 0.106 0.579 0.231 0.556 0.053 work4 0.645 0.281 0.068 0.117 0.068 0.012 _intercept_ 13.015 0.256 obs. 270 R square 0.43 location effect 0.061 Table 6.20: Urban Ibb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable17_ea 0.606 0.229 0.357 0.145 0.356 0.019 dum_eaurable1 0.473 0.073 0.116 0.320 0.111 0.099 dum_eaurable5 0.886 0.210 0.011 0.104 0.011 0.011 dum_eaurable8 0.201 0.050 0.400 0.490 0.398 0.240 dum_eaurable9_ea 1.192 0.183 0.320 0.148 0.311 0.023 employed1 0.901 0.283 0.272 0.081 0.256 0.007 employ_nonself -0.418 0.126 0.519 0.206 0.469 0.043 headread_ea 0.753 0.304 0.179 0.080 0.173 0.006 headsecond 0.188 0.054 0.240 0.427 0.209 0.165 headuniv 0.345 0.075 0.099 0.299 0.101 0.091 light1_ea -0.541 0.139 0.893 0.213 0.890 0.049 light5 -0.353 0.161 0.018 0.134 0.019 0.019 nafemales -0.043 0.016 1.931 1.494 2.094 2.200 nkids -0.072 0.013 3.130 2.391 3.174 4.140 singlep -0.740 0.142 0.593 0.227 0.611 0.034 work9 -1.173 0.406 0.038 0.065 0.038 0.004 _intercept_ 12.036 0.161 obs. 470 R square 0.50 location effect -* *No location effect 111 Table 6.21: Urban Abyan Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea 0.555 0.241 0.677 0.467 0.647 0.052 dum_eaurable11_ea 0.815 0.299 0.386 0.487 0.375 0.040 dum_eaurable12_ea -1.389 0.288 0.789 0.408 0.772 0.027 dum_eaurable15_ea -1.004 0.390 0.157 0.364 0.165 0.012 dum_eaurable16_ea -5.152 1.991 0.014 0.119 0.014 0.000 dum_eaurable17_ea 1.243 0.237 0.513 0.500 0.494 0.041 dum_eaurable1 0.638 0.092 0.073 0.260 0.084 0.077 dum_eaurable3 0.586 0.181 0.014 0.118 0.022 0.022 dum_eaurable4_ea 11.976 2.624 0.010 0.014 0.011 0.000 headiliter_ea -0.716 0.353 0.332 0.107 0.323 0.009 headuniv 0.314 0.106 0.071 0.257 0.062 0.058 nafemalesinv* 0.898 0.165 0.404 0.045 0.363 0.025 nkids -0.071 0.010 2.793 2.553 2.969 6.723 sewage2 0.190 0.061 0.256 0.436 0.266 0.196 _intercept_ 11.368 0.186 obs. 318 R square 0.44 location effect - *nafemalesinv=1/(1+nafemales) No location effect 112 Table 6.22: Urban Sana'a City Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable15_ea -0.703 0.196 0.200 0.088 0.210 0.009 dum_eaurable16 0.247 0.044 0.120 0.325 0.138 0.119 dum_eaurable17_ea 0.388 0.139 0.429 0.162 0.437 0.029 dum_eaurable18 1.460 0.250 0.012 0.111 0.003 0.003 dum_eaurable1_ea 0.887 0.172 0.199 0.110 0.199 0.014 dum_eaurable2_ea 2.642 0.531 0.046 0.027 0.046 0.001 dum_eaurable4 0.363 0.094 0.028 0.165 0.023 0.022 dum_eaurable5 0.265 0.134 0.008 0.092 0.011 0.011 dum_eaurable8 0.227 0.033 0.470 0.499 0.446 0.247 employ_self -0.179 0.100 0.305 0.034 0.296 0.031 headprim 0.102 0.045 0.095 0.293 0.108 0.096 headuniv 0.138 0.048 0.195 0.396 0.225 0.175 housetype1_ea -0.764 0.244 0.557 0.235 0.576 0.061 housetype2_ea -0.658 0.247 0.352 0.215 0.353 0.054 nelderly -0.113 0.027 0.238 0.609 0.254 0.273 nkids -0.100 0.007 2.601 2.373 2.809 4.779 ownhouse1 0.062 0.033 0.468 0.499 0.453 0.248 sewage2 -0.101 0.040 0.377 0.485 0.392 0.239 sewage3_ea 0.649 0.213 0.012 0.064 0.012 0.004 singlep -0.532 0.076 0.565 0.245 0.576 0.048 universityp 0.536 0.129 0.073 0.161 0.081 0.027 water1 0.118 0.041 0.604 0.489 0.584 0.243 water2 0.296 0.078 0.042 0.201 0.036 0.035 water4_ea -0.208 0.071 0.887 0.213 0.895 0.040 work4 0.300 0.107 0.185 0.157 0.174 0.022 _intercept_ 12.828 0.253 obs. 1639 R square 0.46 location effect 0.049 Table 6.23: Urban Al-Baida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea 0.828 0.173 0.449 0.151 0.427 0.028 dum_eaurable13_ea -1.177 0.191 0.641 0.143 0.613 0.016 dum_eaurable1 0.244 0.071 0.159 0.366 0.134 0.117 dum_eaurable8 0.092 0.051 0.455 0.498 0.444 0.248 headdivorced 0.648 0.187 0.013 0.112 0.019 0.019 headsecond 0.152 0.060 0.208 0.406 0.223 0.174 headuniv 0.284 0.095 0.062 0.240 0.078 0.072 kidp -0.800 0.126 0.388 0.224 0.392 0.047 light1_ea -0.834 0.174 0.929 0.123 0.907 0.033 marriedp 0.704 0.129 0.363 0.214 0.380 0.047 nelderly -0.135 0.043 0.322 0.598 0.320 0.331 water1 0.230 0.049 0.524 0.499 0.549 0.248 _intercept_ 12.650 0.209 obs. 327 R square 0.45 location effect 0.062 113 Table 6.24: Urban Taiz Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea 1.801 0.481 0.581 0.200 0.594 0.041 dum_eaurable11_ea -1.253 0.542 0.481 0.195 0.498 0.043 dum_eaurable12_ea -1.353 0.310 0.798 0.119 0.799 0.017 dum_eaurable14_ea -0.648 0.366 0.134 0.101 0.141 0.010 dum_eaurable16 0.409 0.083 0.067 0.250 0.079 0.073 dum_eaurable17_ea -1.026 0.337 0.435 0.159 0.451 0.029 dum_eaurable18 0.535 0.194 0.030 0.171 0.012 0.012 dum_eaurable18_ea -1.311 0.601 0.030 0.045 0.028 0.001 dum_eaurable1 0.446 0.074 0.101 0.301 0.095 0.086 dum_eaurable3 0.320 0.133 0.020 0.138 0.022 0.021 dum_eaurable5_ea 5.330 2.606 0.007 0.010 0.007 0.000 dum_eaurable8 0.211 0.045 0.396 0.489 0.399 0.240 dum_eaurable8_ea 2.023 0.416 0.396 0.145 0.390 0.019 headage -0.008 0.002 41.556 14.214 44.127 185.790 headprim_ea -1.391 0.511 0.099 0.041 0.105 0.002 light5 -0.258 0.128 0.038 0.191 0.032 0.031 nelderly -0.096 0.050 0.279 0.572 0.262 0.281 nkids -0.110 0.012 2.500 2.249 2.447 4.089 sewage2_ea -0.559 0.104 0.227 0.315 0.225 0.103 singlep -0.831 0.105 0.576 0.251 0.586 0.051 water2_ea 1.343 0.356 0.021 0.083 0.023 0.005 _intercept_ 13.328 0.187 obs. 582.000 R square 0.48 location effect 0.057 Table 6.25: Urban Al-Jawf Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 1.674 0.316 0.261 0.155 0.254 0.014 dum_eaurable10 0.273 0.096 0.072 0.259 0.103 0.093 dum_eaurable18 0.671 0.150 0.028 0.164 0.040 0.038 dum_eaurable1_ea 0.482 0.189 0.156 0.132 0.141 0.020 headuniv_ea 3.752 1.347 0.024 0.028 0.022 0.000 highiliter -0.248 0.072 0.197 0.398 0.183 0.150 highprim -0.362 0.102 0.053 0.224 0.072 0.067 housetype1_ea 0.349 0.201 0.921 0.104 0.913 0.017 marriedp 0.472 0.156 0.311 0.145 0.303 0.033 nafemalesinv 2.249 0.272 0.391 0.026 0.364 0.018 _intercept_ 9.516 0.265 obs. 226.000 R square 0.42 location effect 0.089 * nafemalesinv=1/(1+nafemalesinv) 114 Table 6.26: Urban Hajja Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.489 0.082 0.090 0.287 0.113 0.101 elderlyp 0.454 0.224 0.063 0.164 0.067 0.031 headiliter -0.245 0.069 0.523 0.499 0.497 0.251 headmarried -0.476 0.133 0.879 0.326 0.896 0.094 headprim -0.342 0.123 0.076 0.265 0.082 0.076 headsingle -0.343 0.167 0.060 0.237 0.048 0.046 highprim 0.235 0.121 0.075 0.263 0.078 0.072 housetype1_ea -0.483 0.130 0.689 0.247 0.711 0.056 housetype4 -0.930 0.344 0.004 0.060 0.007 0.007 housetype5_ea 3.085 0.838 0.020 0.042 0.019 0.001 kidp -0.762 0.141 0.377 0.248 0.380 0.063 marriedp 0.549 0.148 0.352 0.229 0.346 0.053 nelderly -0.211 0.063 0.346 0.630 0.322 0.339 primaryp 0.423 0.153 0.218 0.247 0.199 0.047 sewage2_ea 0.551 0.097 0.350 0.322 0.360 0.113 work1 0.596 0.155 0.306 0.257 0.324 0.058 work9 0.908 0.158 0.102 0.158 0.135 0.049 _intercept_ 12.050 0.170 obs. 339.000 R square 0.49 location effect 0.034 Table 6.27: Urban Al-Hodeida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) cook2_1 -0.204 0.041 0.253 0.435 0.269 0.197 dum_eaurable1 0.534 0.086 0.049 0.216 0.044 0.042 dum_eaurable2_ea 7.390 2.179 0.006 0.009 0.006 0.000 dum_eaurable3_ea -2.300 0.891 0.015 0.021 0.015 0.001 dum_eaurable8 0.225 0.041 0.255 0.436 0.281 0.202 dum_eaurable9_ea 0.360 0.215 0.216 0.158 0.237 0.033 employed2 0.417 0.218 0.350 0.085 0.353 0.006 employ_nonself 0.265 0.101 0.450 0.240 0.441 0.065 headiliter -0.171 0.035 0.485 0.500 0.473 0.250 headiliter_ea 0.590 0.202 0.485 0.151 0.488 0.032 headmarried -0.093 0.051 0.837 0.370 0.844 0.132 housetype2 0.171 0.054 0.094 0.291 0.130 0.113 kidp -0.692 0.086 0.326 0.250 0.325 0.063 light1 0.225 0.055 0.727 0.445 0.732 0.197 light1_ea -0.260 0.077 0.716 0.330 0.729 0.116 nafemales -0.133 0.014 1.871 1.494 1.914 2.045 singlep -0.528 0.092 0.567 0.253 0.584 0.055 universityp 0.916 0.178 0.030 0.106 0.029 0.010 university_ea 5.696 1.590 0.023 0.022 0.025 0.001 work9 -0.404 0.197 0.034 0.086 0.034 0.007 _intercept_ 11.741 0.187 obs. 841 R square 0.57 location effect 0.014 115 Table 6.28: Urban Hadramout Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11_ea 1.099 0.299 0.559 0.211 0.544 0.039 dum_eaurable14_ea -1.502 0.219 0.170 0.267 0.176 0.075 dum_eaurable15_ea 0.911 0.246 0.328 0.195 0.315 0.038 dum_eaurable16 0.366 0.093 0.060 0.237 0.046 0.044 dum_eaurable18_ea 2.307 0.300 0.297 0.223 0.297 0.049 dum_eaurable2 -0.236 0.108 0.019 0.138 0.024 0.023 dum_eaurable3_ea -4.124 1.860 0.016 0.029 0.013 0.000 dum_eaurable8_ea -2.581 0.338 0.602 0.203 0.611 0.044 dum_eaurable9 0.130 0.040 0.375 0.484 0.413 0.243 employed1 -5.526 1.763 0.313 0.079 0.320 0.005 employed2 7.625 1.727 0.325 0.080 0.328 0.005 employ_self 1.393 0.212 0.434 0.220 0.449 0.039 headiliter -0.167 0.044 0.279 0.448 0.314 0.216 headuniv_ea 4.546 0.874 0.067 0.056 0.064 0.004 housetype1 0.306 0.110 0.836 0.371 0.850 0.128 housetype1_ea 1.504 0.229 0.836 0.199 0.833 0.039 housetype2 0.436 0.129 0.124 0.329 0.116 0.103 kidp -0.516 0.099 0.345 0.225 0.329 0.054 light1_ea 0.659 0.183 0.905 0.201 0.898 0.048 light4 -0.484 0.245 0.004 0.063 0.012 0.012 ownhouse2_ea 0.696 0.220 0.211 0.163 0.205 0.023 singlep -0.490 0.100 0.557 0.214 0.539 0.049 universityp 0.659 0.240 0.028 0.092 0.027 0.007 university_ea -11.177 2.756 0.024 0.017 0.025 0.000 water1_ea -0.863 0.218 0.879 0.256 0.875 0.065 _intercept_ 9.348 0.337 obs. 463 R square 0.48 location effect 0.088 Table 6.29: Urban Dhamar Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable14_ea 0.736 0.211 0.277 0.169 0.267 0.026 dum_eaurable15 0.136 0.075 0.216 0.411 0.218 0.171 dum_eaurable16 0.736 0.193 0.035 0.184 0.023 0.023 dum_eaurable1 0.547 0.109 0.115 0.319 0.082 0.076 dum_eaurable4 0.611 0.258 0.015 0.123 0.015 0.015 dum_eaurable5 0.448 0.215 0.011 0.104 0.019 0.018 dum_eaurable8 0.243 0.070 0.338 0.473 0.338 0.224 employed1 1.451 0.542 0.292 0.068 0.278 0.005 headsingle 0.292 0.144 0.054 0.226 0.047 0.045 nkids -0.044 0.020 3.285 2.498 3.227 5.206 singlep -0.576 0.176 0.589 0.223 0.579 0.048 hh_size -0.045 0.012 7.454 4.323 7.557 12.963 work1 0.589 0.195 0.320 0.180 0.313 0.031 work9 0.961 0.303 0.071 0.109 0.077 0.018 _intercept_ 11.345 0.217 obs. 342 116 R square 0.42 location effect 0.045 Table 6.30: Urban Shabwah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable16 0.324 0.143 0.040 0.196 0.037 0.036 dum_eaurable17_ea -0.672 0.160 0.561 0.216 0.572 0.028 dum_eaurable18 0.246 0.091 0.094 0.292 0.105 0.094 dum_eaurable1 0.353 0.072 0.218 0.413 0.202 0.162 headiliter -0.242 0.063 0.293 0.455 0.297 0.210 highread 0.104 0.062 0.307 0.461 0.276 0.201 housetype1_ea -0.954 0.092 0.701 0.323 0.679 0.112 marriedp 0.527 0.153 0.378 0.222 0.372 0.035 nkids -0.046 0.011 4.011 3.554 4.106 7.311 sewage1_ea 0.448 0.091 0.231 0.334 0.246 0.117 _intercept_ 12.304 0.144 obs. 198 R square 0.59 location effect -* *No location effect Table 6.31: Urban Sa'adah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10 0.184 0.070 0.194 0.395 0.205 0.164 dum_eaurable11 0.182 0.065 0.252 0.434 0.297 0.209 dum_eaurable14_ea -0.710 0.243 0.165 0.116 0.177 0.015 dum_eaurable1 0.390 0.061 0.202 0.402 0.222 0.173 employ_nonself -0.913 0.145 0.440 0.249 0.450 0.065 headuniv_ea 2.335 0.704 0.076 0.265 0.073 0.003 housetype3 0.357 0.174 0.032 0.176 0.023 0.022 light1 0.386 0.073 0.726 0.446 0.720 0.202 light1_ea -0.688 0.119 0.712 0.322 0.701 0.115 primary_ea 1.627 0.599 0.189 0.072 0.187 0.004 sewage3_ea 0.427 0.189 0.168 0.206 0.149 0.029 singlep -1.111 0.118 0.571 0.222 0.573 0.038 water4_ea -0.322 0.148 0.888 0.208 0.893 0.039 work5 -4.300 0.634 0.032 0.054 0.034 0.003 _intercept_ 12.623 0.195 obs. 324 R square 0.43 location effect -* *No location effect 117 Table 6.32: Urban Aden Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12_ea -0.554 0.225 0.841 0.126 0.845 0.015 dum_eaurable13_ea -0.314 0.092 0.517 0.210 0.503 0.046 dum_eaurable16 0.169 0.066 0.080 0.272 0.083 0.076 dum_eaurable18_ea 0.425 0.138 0.428 0.243 0.452 0.058 dum_eaurable1 0.306 0.051 0.165 0.371 0.149 0.127 dum_eaurable3 0.281 0.110 0.014 0.116 0.024 0.024 dum_eaurable9 0.239 0.036 0.412 0.492 0.417 0.244 headmarried -0.213 0.042 0.803 0.398 0.785 0.169 housetype1_ea 0.186 0.071 0.726 0.281 0.729 0.076 light1_ea 0.403 0.241 0.938 0.115 0.952 0.008 nafemales -0.093 0.013 1.927 1.548 1.958 1.758 singlep -1.013 0.076 0.548 0.238 0.555 0.051 universityp 0.576 0.119 0.069 0.154 0.083 0.025 university_ea 2.123 0.695 0.057 0.037 0.061 0.002 water2 0.689 0.199 0.004 0.063 0.005 0.005 work10 -0.889 0.313 0.035 0.058 0.037 0.004 work3 0.492 0.187 0.087 0.091 0.098 0.009 _intercept_ 12.363 0.199 obs. 716 R square 0.50 location effect 0.006 Table 6.33: Urban Laheg Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable2_ea -21.974 3.829 0.009 0.011 0.010 0.0001 headage 0.006 0.003 44.828 13.877 45.183 160.152 headprim_ea 5.295 1.170 0.062 0.046 0.065 0.001 light1 0.433 0.136 0.824 0.381 0.853 0.126 light2_ea 1.032 0.181 0.064 0.225 0.052 0.043 nelderly -0.120 0.053 0.368 0.666 0.409 0.496 ownhouse1 0.178 0.078 0.833 0.373 0.774 0.175 primaryp 1.092 0.143 0.331 0.274 0.292 0.065 singlep -0.790 0.158 0.567 0.236 0.533 0.058 university_ea 5.359 1.420 0.039 0.024 0.036 0.001 water3_ea -0.418 0.204 0.035 0.176 0.031 0.028 _intercept_ 10.656 0.222 obs. 273.000 R square 0.46 location effect 0.038 118 Table 6.34: Urban Mareb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable13_ea -0.659 0.264 0.577 0.179 0.600 0.037 dum_eaurable16 0.512 0.214 0.037 0.188 0.032 0.031 dum_eaurable18 0.334 0.105 0.141 0.348 0.155 0.132 dum_eaurable18_ea -3.348 1.140 0.141 0.069 0.131 0.005 dum_eaurable18_ea×sizeinv* 16.452 6.687 0.022 0.000 0.018 0.000 dum_eaurable5 0.408 0.164 0.027 0.162 0.051 0.049 headiliter_ea -0.731 0.298 0.389 0.142 0.432 0.026 headsecond×sizeinv 1.194 0.504 0.050 0.010 0.044 0.007 light1_ea 2.557 0.776 0.895 0.062 0.896 0.004 light4 0.785 0.414 0.006 0.075 0.010 0.010 ownhouse1 0.383 0.081 0.482 0.500 0.520 0.251 sizeinv 1.732 0.947 0.158 0.011 0.136 0.006 water4_ea -0.771 0.238 0.858 0.192 0.872 0.032 _intercept_ 10.432 0.598 obs. 224 R square 0.40 location effect - *sizeinv=1/(1+hh_size); no location effect. Table 6.35: Urban Al-Mahweet Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable14_ea 1.782 0.396 0.175 0.108 0.190 0.013 dum_eaurable2_ea -4.391 1.453 0.015 0.021 0.015 0.001 dum_eaurable3 0.854 0.419 0.003 0.052 0.003 0.003 dum_eaurable3_ea -21.371 8.416 0.003 0.005 0.003 0.000 dum_eaurable5_ea 7.384 2.691 0.011 0.010 0.011 0.000 dum_eaurable8 0.189 0.050 0.482 0.500 0.464 0.250 dum_eaurable8_ea -0.817 0.360 0.482 0.104 0.489 0.012 employ_nonself 0.729 0.201 0.677 0.204 0.656 0.054 headage -0.005 0.002 42.882 15.373 45.828 245.501 headdivorced 0.629 0.313 0.013 0.113 0.007 0.007 headread_ea 0.949 0.300 0.168 0.086 0.168 0.009 headuniv 0.159 0.079 0.128 0.335 0.117 0.104 highread -0.145 0.068 0.177 0.382 0.165 0.138 marriedp 1.043 0.288 0.349 0.228 0.346 0.039 marriedp×nkidsinv* -0.824 0.362 0.152 0.056 0.142 0.044 nkidsinv 1.033 0.155 0.387 0.102 0.352 0.078 nkidsinv×water3_ea 3.302 1.834 0.006 0.001 0.004 0.000 water1_ea -0.228 0.111 0.697 0.353 0.679 0.131 work1 -1.017 0.204 0.518 0.241 0.508 0.062 _intercept_ 11.162 0.248 obs. 289 R square 0.40 location effect - * nkidsinv=1/(1+nkids); no location effect. 119 Table 6.36: Urban Al-Maharh Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea 0.483 0.204 0.525 0.254 0.561 0.073 dum_eaurable16_ea -14.151 6.538 0.011 0.017 0.011 0.0003 dum_eaurable18 0.639 0.192 0.049 0.216 0.031 0.030 employed1 1.042 0.439 0.344 0.142 0.347 0.021 headage -0.006 0.002 43.065 13.993 47.476 217.166 headuniv_ea 9.756 2.554 0.056 0.064 0.057 0.003 marriedp 0.719 0.191 0.387 0.242 0.386 0.039 nkids -0.091 0.017 3.292 3.025 3.372 4.982 university_ea -37.644 4.958 0.016 0.020 0.016 0.0002 water1_ea -0.527 0.194 0.301 0.386 0.288 0.158 water4_ea 1.007 0.177 0.793 0.240 0.794 0.072 _intercept_ 11.189 0.269 obs. 137 R square 0.69 location effect -* *No location effect Table 6.37: Urban Amran Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable13_ea -0.459 0.171 0.734 0.169 0.733 0.027 dum_eaurable14_ea 1.098 0.202 0.193 0.134 0.197 0.020 dum_eaurable16 0.989 0.310 0.018 0.132 0.007 0.007 dum_eaurable1 0.473 0.073 0.156 0.363 0.169 0.141 dum_eaurable8 0.254 0.055 0.377 0.485 0.403 0.241 headuniv_ea -1.609 0.615 0.098 0.048 0.096 0.003 nkidsinv×singlep 1.306 0.339 0.169 0.028 0.139 0.006 nkidsinv×work10 3.381 1.329 0.019 0.001 0.015 0.001 ownhouse1_ea -0.680 0.223 0.659 0.142 0.667 0.017 singlep -0.863 0.151 0.587 0.214 0.597 0.035 work1 0.479 0.166 0.286 0.161 0.273 0.033 _intercept_ 12.042 0.222 obs. 302 R square 0.40 location effect 0.071 120 Table 6.38: Urban Al-Dhale Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable13_ea 0.322 0.147 0.578 0.234 0.577 0.078 dum_eaurable1_ea -2.335 0.558 0.130 0.091 0.148 0.013 dum_eaurable4 0.485 0.193 0.016 0.126 0.022 0.022 dum_eaurable5 0.663 0.195 0.015 0.123 0.025 0.025 dum_eaurable9_ea 1.818 0.372 0.298 0.147 0.317 0.026 headdivorced -0.613 0.229 0.010 0.099 0.017 0.016 headiliter_ea 1.776 0.452 0.310 0.121 0.309 0.010 headprim_ea -1.112 0.687 0.045 0.065 0.046 0.003 headsecond_ea 0.781 0.410 0.324 0.118 0.325 0.010 headwidow -0.702 0.143 0.046 0.210 0.051 0.048 light4 0.454 0.247 0.020 0.140 0.014 0.014 nafemales -0.057 0.032 1.727 1.531 1.874 1.732 singlep -0.870 0.159 0.583 0.244 0.600 0.055 hh_size -0.035 0.013 7.598 4.709 7.765 14.183 water2 0.538 0.144 0.050 0.218 0.081 0.075 _intercept_ 11.298 0.276 obs. 233 R square 0.50 location effect -* *No location effect 121 Table 7: Compare estimates of headcount using different data sources Governorate #hhno* avg_FGT0 se_FGT0 FGT0 (survey) Rural 11 Ibb 233,491 0.304 0.023 0.328 12 Abyan 38,120 0.525 0.046 0.504 14 Al-Baida 49,667 0.583 0.015 0.598 15 Taiz 266,914 0.429 0.021 0.415 16 Al-Jawf 45,325 0.555 0.025 0.526 17 Hajja 154,183 0.488 0.017 0.500 18 Al-Hodeida 224,491 0.350 0.019 0.364 19 Hadramout 58,818 0.350 0.012 0.404 20 Dhamar 150,379 0.227 0.021 0.253 21 Shabwah 34,657 0.522 0.028 0.568 22 Sa'adah 61,965 0.147 0.029 0.162 23 Sana'a Region 99,572 0.277 0.021 0.281 25 Laheg 87,266 0.463 0.018 0.495 26 Mareb 21,181 0.493 0.018 0.501 27 Al-Mahweet 55,251 0.304 0.024 0.315 28 Al-Maharh 5,849 0.090 0.015 0.063 29 Amran 73,020 0.650 0.029 0.706 30 Al-Dhale 48,051 0.456 0.024 0.464 31 Remah 45,963 0.326 0.025 0.341 Urban 11 Ibb 49,143 0.145 0.013 0.164 12 Abyan 13,473 0.348 0.017 0.314 13 Sana'a City 236,666 0.131 0.008 0.149 14 Al-Baida 12,786 0.157 0.023 0.167 15 Taiz 78,245 0.227 0.014 0.237 16 Al-Jawf 6,269 0.318 0.029 0.326 17 Hajja 16,137 0.221 0.022 0.209 18 Al-Hodeida 104,909 0.230 0.011 0.216 19 Hadramout 53,868 0.247 0.014 0.294 20 Dhamar 24,312 0.273 0.021 0.297 21 Shabwah 7,551 0.367 0.025 0.394 22 Sa'adah 11,549 0.195 0.013 0.182 24 Aden 82,967 0.159 0.012 0.169 25 Laheg 8,495 0.191 0.026 0.229 26 Mareb 3,325 0.184 0.021 0.180 27 Al-Mahweet 4,169 0.196 0.027 0.219 28 Al-Maharh 4,074 0.100 0.013 0.114 29 Amran 17,135 0.371 0.031 0.339 30 Al-Dhale 7,717 0.293 0.021 0.282 * The number of households in this table is different from the number listed in Table 1 because of two reasons: 1. The houses which are not occupied and/or do not have a household head are dropped (see note 3); 2. The households which have missing values in the variables used in the consumption model are dropped. 122 Table 8: Estimates of poverty indicators (rural areas) Governorate District #hhno avg_ ^ y avg_FGT0 se_FGT0 avg_FGT1 se_FGT1 avg_FGT2 se_FGT2 11 1 14,541 72,875 0.371 0.040 0.092 0.013 0.0333 0.0054 11 2 15,868 88,021 0.253 0.026 0.056 0.008 0.0186 0.0034 11 3 8,163 81,666 0.298 0.038 0.069 0.013 0.0237 0.0055 11 4 7,522 104,515 0.183 0.024 0.036 0.007 0.0110 0.0027 11 5 4,061 157,866 0.095 0.022 0.018 0.005 0.0055 0.0020 11 6 9,971 81,745 0.272 0.028 0.058 0.008 0.0188 0.0034 11 7 14,423 79,836 0.333 0.029 0.090 0.012 0.0343 0.0055 11 8 13,985 72,499 0.338 0.033 0.080 0.011 0.0280 0.0045 11 9 12,050 86,754 0.261 0.030 0.060 0.008 0.0210 0.0032 11 10 14,598 71,842 0.365 0.040 0.083 0.013 0.0274 0.0054 11 11 19,768 69,875 0.389 0.034 0.097 0.012 0.0350 0.0051 11 12 12,551 85,870 0.264 0.026 0.061 0.008 0.0209 0.0035 11 13 13,153 109,372 0.192 0.022 0.043 0.007 0.0144 0.0027 11 14 8,929 89,138 0.313 0.027 0.082 0.010 0.0308 0.0048 11 15 13,775 84,457 0.299 0.027 0.069 0.009 0.0235 0.0036 11 16 15,231 75,501 0.372 0.030 0.094 0.011 0.0341 0.0051 11 17 10,972 78,069 0.320 0.028 0.084 0.011 0.0315 0.0051 11 18 1,878 82,728 0.266 0.044 0.057 0.014 0.0188 0.0055 11 19 3,820 84,555 0.287 0.035 0.071 0.012 0.0258 0.0055 11 20 18,232 75,108 0.349 0.032 0.088 0.012 0.0319 0.0055 12 1 2,649 71,801 0.542 0.069 0.133 0.028 0.0459 0.0127 12 2 2,986 78,890 0.468 0.048 0.133 0.020 0.0527 0.0100 12 3 1,351 51,284 0.793 0.053 0.282 0.040 0.1263 0.0255 12 4 8,473 70,567 0.538 0.051 0.150 0.020 0.0582 0.0100 12 5 1,827 70,233 0.526 0.079 0.127 0.030 0.0433 0.0132 12 6 5,178 78,668 0.456 0.049 0.127 0.019 0.0496 0.0088 12 7 2,049 71,841 0.507 0.072 0.124 0.026 0.0430 0.0115 12 8 2,861 72,270 0.522 0.057 0.142 0.025 0.0538 0.0121 12 9 2,583 66,003 0.507 0.086 0.121 0.031 0.0425 0.0142 12 10 838 63,623 0.626 0.072 0.215 0.042 0.0967 0.0238 12 11 7,325 73,526 0.528 0.040 0.161 0.019 0.0674 0.0105 14 1 791 81,282 0.636 0.013 0.365 0.013 0.2295 0.0134 14 2 1,500 35,906 0.838 0.027 0.472 0.020 0.3086 0.0153 14 3 657 43,997 0.696 0.027 0.347 0.019 0.2115 0.0166 14 4 4,523 113,595 0.415 0.025 0.193 0.010 0.1144 0.0070 14 5 2,351 56,238 0.736 0.022 0.308 0.021 0.1590 0.0161 14 6 2,262 106,801 0.480 0.026 0.175 0.015 0.0877 0.0101 14 7 2,728 37,440 0.863 0.018 0.441 0.021 0.2584 0.0188 14 8 4,656 91,850 0.509 0.020 0.236 0.010 0.1350 0.0087 14 10 4,728 105,567 0.376 0.025 0.146 0.012 0.0796 0.0088 14 11 2,341 45,088 0.847 0.014 0.478 0.018 0.3074 0.0193 14 12 1,924 69,623 0.609 0.029 0.257 0.017 0.1376 0.0130 14 13 538 73,400 0.531 0.031 0.202 0.021 0.0922 0.0141 123 14 14 3,209 97,023 0.398 0.025 0.151 0.009 0.0788 0.0063 14 15 2,218 129,514 0.454 0.028 0.171 0.012 0.0865 0.0070 14 16 3,763 155,288 0.354 0.016 0.124 0.011 0.0582 0.0082 14 17 3,089 74,986 0.569 0.031 0.246 0.014 0.1371 0.0100 14 18 2,203 81,814 0.607 0.029 0.255 0.022 0.1330 0.0169 14 19 3,354 52,617 0.739 0.022 0.341 0.017 0.1934 0.0141 14 20 2,832 34,785 0.872 0.015 0.483 0.018 0.3059 0.0172 15 1 19,220 75,623 0.417 0.034 0.113 0.012 0.0434 0.0057 15 2 15,875 83,465 0.373 0.024 0.100 0.009 0.0380 0.0045 15 3 20,313 78,412 0.430 0.024 0.123 0.010 0.0487 0.0055 15 4 26,858 76,190 0.436 0.024 0.125 0.011 0.0492 0.0055 15 5 8,747 73,656 0.495 0.033 0.147 0.015 0.0593 0.0077 15 6 1,920 55,325 0.654 0.048 0.238 0.025 0.1106 0.0144 15 7 4,844 65,293 0.545 0.044 0.162 0.017 0.0656 0.0085 15 8 16,356 76,597 0.430 0.027 0.118 0.011 0.0450 0.0053 15 9 3,401 93,501 0.295 0.050 0.072 0.016 0.0255 0.0066 15 10 14,674 77,023 0.417 0.028 0.117 0.011 0.0460 0.0054 15 11 12,472 77,383 0.415 0.041 0.113 0.016 0.0434 0.0073 15 12 13,286 71,181 0.471 0.028 0.139 0.014 0.0558 0.0073 15 13 6,950 73,155 0.448 0.024 0.136 0.012 0.0566 0.0070 15 14 21,870 80,420 0.396 0.029 0.109 0.012 0.0425 0.0058 15 15 4,163 68,827 0.492 0.037 0.137 0.015 0.0519 0.0071 15 16 10,885 80,877 0.396 0.022 0.115 0.009 0.0470 0.0048 15 20 28,595 69,256 0.505 0.031 0.149 0.015 0.0601 0.0082 15 21 16,083 83,952 0.360 0.025 0.098 0.010 0.0374 0.0050 15 22 15,438 85,348 0.369 0.022 0.105 0.009 0.0415 0.0051 15 23 4,964 67,966 0.504 0.031 0.150 0.014 0.0604 0.0069 16 1 9,328 64,267 0.592 0.026 0.180 0.015 0.0727 0.0086 16 2 2,440 66,831 0.619 0.029 0.172 0.016 0.0635 0.0088 16 3 3,293 61,796 0.652 0.031 0.196 0.015 0.0791 0.0086 16 4 2,805 79,070 0.509 0.032 0.149 0.012 0.0585 0.0063 16 5 1,980 63,470 0.691 0.025 0.210 0.018 0.0834 0.0110 16 6 2,759 85,114 0.445 0.042 0.098 0.013 0.0304 0.0055 16 7 1,064 64,347 0.634 0.031 0.179 0.016 0.0673 0.0089 16 8 469 60,078 0.776 0.041 0.257 0.036 0.1065 0.0231 16 9 873 65,805 0.676 0.035 0.191 0.018 0.0725 0.0093 16 10 6,542 75,366 0.486 0.040 0.130 0.014 0.0478 0.0069 16 11 7,097 70,545 0.409 0.033 0.102 0.011 0.0361 0.0053 16 12 6,675 60,142 0.648 0.033 0.199 0.018 0.0809 0.0104 17 1 2,926 79,818 0.462 0.041 0.148 0.019 0.0643 0.0101 17 2 9,557 64,992 0.559 0.033 0.213 0.021 0.1050 0.0136 17 3 1,544 53,770 0.638 0.048 0.240 0.025 0.1177 0.0145 17 4 14,088 78,559 0.463 0.027 0.161 0.013 0.0750 0.0077 17 5 1,949 85,755 0.403 0.041 0.132 0.020 0.0589 0.0116 17 6 5,883 76,955 0.427 0.030 0.134 0.014 0.0579 0.0076 17 7 8,327 80,922 0.500 0.047 0.184 0.026 0.0882 0.0155 17 8 3,641 58,984 0.624 0.038 0.236 0.022 0.1157 0.0133 124 17 9 2,889 79,628 0.515 0.037 0.209 0.020 0.1071 0.0128 17 10 4,789 78,412 0.461 0.044 0.175 0.021 0.0872 0.0116 17 11 8,315 82,476 0.432 0.024 0.149 0.010 0.0695 0.0060 17 12 7,542 83,276 0.416 0.027 0.135 0.013 0.0595 0.0074 17 13 5,454 62,849 0.622 0.030 0.248 0.018 0.1257 0.0115 17 14 3,641 77,404 0.453 0.031 0.157 0.015 0.0727 0.0087 17 15 3,354 118,246 0.394 0.032 0.143 0.015 0.0696 0.0090 17 16 2,761 148,306 0.349 0.030 0.136 0.014 0.0688 0.0082 17 17 5,671 66,079 0.496 0.030 0.173 0.013 0.0802 0.0072 17 18 3,662 82,258 0.480 0.028 0.176 0.015 0.0847 0.0092 17 19 1,767 66,716 0.557 0.029 0.200 0.020 0.0939 0.0127 17 20 6,093 79,911 0.491 0.052 0.171 0.027 0.0787 0.0152 17 21 3,215 70,650 0.590 0.051 0.236 0.030 0.1195 0.0191 17 22 9,301 77,831 0.477 0.026 0.164 0.015 0.0754 0.0091 17 23 1,174 103,619 0.328 0.035 0.108 0.013 0.0481 0.0071 17 24 8,112 74,825 0.462 0.036 0.148 0.016 0.0650 0.0087 17 25 6,049 77,113 0.448 0.031 0.135 0.014 0.0562 0.0078 17 26 3,319 100,385 0.347 0.035 0.109 0.014 0.0479 0.0069 17 27 4,704 71,388 0.568 0.047 0.212 0.027 0.1024 0.0166 17 28 1,931 60,292 0.598 0.059 0.241 0.034 0.1231 0.0208 17 29 3,381 67,265 0.534 0.027 0.200 0.014 0.0979 0.0084 17 30 6,356 87,064 0.445 0.030 0.155 0.013 0.0728 0.0066 17 31 2,788 47,232 0.716 0.036 0.298 0.024 0.1559 0.0161 18 1 20,361 82,573 0.354 0.024 0.091 0.009 0.0331 0.0042 18 2 14,985 79,037 0.366 0.025 0.096 0.009 0.0361 0.0042 18 4 681 40,082 0.794 0.043 0.334 0.034 0.1661 0.0228 18 5 4,468 88,012 0.391 0.026 0.109 0.012 0.0416 0.0058 18 6 10,499 84,045 0.397 0.038 0.114 0.015 0.0452 0.0071 18 7 11,608 76,024 0.396 0.021 0.112 0.009 0.0441 0.0046 18 8 5,316 121,577 0.201 0.019 0.044 0.007 0.0139 0.0028 18 9 5,073 77,316 0.366 0.029 0.089 0.011 0.0307 0.0048 18 10 19,422 82,389 0.333 0.027 0.080 0.010 0.0279 0.0042 18 11 1,328 135,460 0.234 0.036 0.059 0.012 0.0211 0.0057 18 12 6,577 85,103 0.370 0.038 0.110 0.014 0.0450 0.0064 18 13 15,704 91,413 0.271 0.022 0.065 0.007 0.0230 0.0030 18 14 8,510 104,729 0.262 0.021 0.063 0.008 0.0219 0.0034 18 15 9,965 93,729 0.274 0.020 0.066 0.007 0.0228 0.0035 18 16 5,594 80,398 0.341 0.026 0.085 0.010 0.0302 0.0047 18 17 29,568 83,729 0.330 0.022 0.083 0.008 0.0298 0.0037 18 18 7,869 87,948 0.330 0.033 0.083 0.011 0.0294 0.0048 18 19 4,683 73,273 0.384 0.031 0.098 0.012 0.0354 0.0053 18 20 3,014 63,591 0.485 0.030 0.160 0.012 0.0702 0.0066 18 21 444 91,259 0.260 0.089 0.069 0.027 0.0265 0.0115 18 23 461 69,421 0.456 0.056 0.116 0.019 0.0410 0.0089 18 24 17,998 78,941 0.396 0.022 0.113 0.010 0.0446 0.0050 18 25 11,604 73,430 0.421 0.029 0.114 0.011 0.0427 0.0051 18 26 8,759 73,380 0.394 0.035 0.103 0.013 0.0384 0.0060 125 19 1 538 118,005 0.098 0.022 0.015 0.004 0.0036 0.0013 19 2 225 98,910 0.208 0.034 0.042 0.008 0.0125 0.0030 19 3 271 91,163 0.238 0.046 0.042 0.011 0.0114 0.0036 19 4 198 89,358 0.188 0.045 0.031 0.010 0.0080 0.0033 19 5 300 98,535 0.221 0.039 0.043 0.011 0.0124 0.0044 19 6 302 139,296 0.047 0.033 0.009 0.012 0.0031 0.0065 19 7 4,672 104,762 0.290 0.020 0.069 0.006 0.0232 0.0021 19 8 3,827 69,593 0.531 0.017 0.153 0.009 0.0565 0.0046 19 9 1,616 102,404 0.179 0.033 0.032 0.008 0.0088 0.0025 19 10 4,793 60,356 0.656 0.019 0.200 0.012 0.0760 0.0060 19 11 4,060 50,784 0.689 0.022 0.211 0.011 0.0810 0.0055 19 12 1,116 79,937 0.303 0.029 0.071 0.008 0.0236 0.0033 19 13 3,769 101,413 0.251 0.020 0.050 0.007 0.0146 0.0026 19 14 1,276 83,432 0.345 0.032 0.078 0.011 0.0249 0.0043 19 15 1,734 77,736 0.433 0.025 0.108 0.009 0.0366 0.0039 19 16 2,724 82,091 0.357 0.019 0.086 0.008 0.0288 0.0036 19 17 1,603 70,643 0.540 0.025 0.150 0.012 0.0544 0.0060 19 18 5,189 95,129 0.249 0.023 0.048 0.006 0.0142 0.0022 19 19 2,703 105,037 0.225 0.018 0.044 0.005 0.0130 0.0017 19 20 900 123,242 0.141 0.024 0.031 0.006 0.0099 0.0023 19 21 2,246 107,631 0.155 0.021 0.024 0.005 0.0060 0.0015 19 22 2,094 87,712 0.184 0.022 0.028 0.005 0.0071 0.0014 19 23 994 78,912 0.357 0.035 0.078 0.010 0.0246 0.0038 19 24 2,468 72,054 0.431 0.032 0.091 0.010 0.0280 0.0039 19 25 1,636 73,856 0.425 0.030 0.092 0.010 0.0289 0.0038 19 26 3,391 84,754 0.243 0.026 0.041 0.006 0.0108 0.0020 19 27 525 72,895 0.381 0.045 0.070 0.010 0.0196 0.0034 19 28 1,916 157,317 0.026 0.009 0.004 0.001 0.0008 0.0004 19 29 265 91,409 0.234 0.042 0.041 0.010 0.0116 0.0038 19 30 1,467 80,934 0.359 0.033 0.077 0.011 0.0239 0.0043 20 1 15,481 97,470 0.230 0.029 0.045 0.007 0.0137 0.0027 20 2 7,695 110,482 0.174 0.031 0.036 0.009 0.0112 0.0033 20 3 8,483 113,260 0.191 0.023 0.038 0.006 0.0119 0.0024 20 4 6,968 102,810 0.196 0.025 0.038 0.006 0.0113 0.0023 20 5 20,563 92,728 0.264 0.027 0.052 0.008 0.0156 0.0030 20 6 22,908 92,479 0.271 0.024 0.057 0.008 0.0180 0.0030 20 7 20,894 89,862 0.313 0.031 0.072 0.011 0.0243 0.0046 20 8 3,521 102,383 0.234 0.024 0.046 0.007 0.0136 0.0028 20 9 7,430 124,510 0.124 0.018 0.023 0.004 0.0070 0.0016 20 10 14,268 121,806 0.139 0.018 0.027 0.005 0.0081 0.0016 20 11 15,304 102,743 0.236 0.024 0.048 0.006 0.0153 0.0024 20 12 6,864 102,765 0.184 0.028 0.032 0.007 0.0088 0.0023 21 1 849 53,941 0.728 0.038 0.266 0.022 0.1245 0.0145 21 2 1,049 56,956 0.783 0.041 0.267 0.027 0.1161 0.0171 21 3 879 61,504 0.674 0.046 0.233 0.027 0.1064 0.0167 21 4 980 65,579 0.656 0.038 0.232 0.022 0.1053 0.0140 21 5 2,460 137,873 0.338 0.035 0.108 0.015 0.0486 0.0079 126 21 6 2,204 79,137 0.506 0.032 0.177 0.015 0.0830 0.0090 21 7 3,184 103,718 0.369 0.034 0.117 0.014 0.0522 0.0078 21 8 2,842 52,062 0.750 0.030 0.293 0.021 0.1438 0.0146 21 9 2,807 84,892 0.451 0.033 0.144 0.014 0.0640 0.0079 21 10 2,708 80,482 0.513 0.027 0.177 0.015 0.0812 0.0094 21 11 1,137 51,629 0.794 0.027 0.328 0.023 0.1651 0.0169 21 12 2,810 82,461 0.452 0.040 0.159 0.017 0.0773 0.0103 21 13 1,413 95,357 0.350 0.045 0.103 0.018 0.0444 0.0091 21 14 2,347 81,105 0.492 0.042 0.160 0.019 0.0714 0.0100 21 15 1,760 80,846 0.452 0.049 0.139 0.020 0.0606 0.0107 21 16 3,125 139,876 0.489 0.038 0.167 0.020 0.0773 0.0118 21 17 2,103 58,140 0.765 0.027 0.310 0.023 0.1552 0.0168 22 1 1,934 87,662 0.216 0.042 0.044 0.014 0.0130 0.0055 22 2 1,767 100,967 0.136 0.052 0.027 0.010 0.0088 0.0033 22 3 5,257 91,786 0.136 0.044 0.018 0.008 0.0037 0.0022 22 4 2,047 86,626 0.177 0.058 0.032 0.012 0.0098 0.0036 22 5 5,136 87,911 0.232 0.067 0.045 0.019 0.0132 0.0067 22 6 1,122 95,685 0.156 0.043 0.027 0.012 0.0076 0.0041 22 7 2,696 89,468 0.136 0.055 0.019 0.011 0.0041 0.0032 22 8 6,267 95,683 0.108 0.037 0.014 0.007 0.0032 0.0018 22 9 5,770 102,804 0.081 0.027 0.010 0.004 0.0021 0.0010 22 10 5,411 92,406 0.153 0.030 0.026 0.007 0.0073 0.0026 22 11 12,144 104,571 0.136 0.032 0.026 0.007 0.0076 0.0025 22 12 5,867 109,299 0.128 0.037 0.021 0.009 0.0054 0.0027 22 13 1,430 99,339 0.176 0.038 0.041 0.010 0.0140 0.0039 22 14 4,414 105,168 0.175 0.029 0.040 0.009 0.0130 0.0037 22 15 703 88,063 0.192 0.060 0.050 0.014 0.0181 0.0062 23 1 9,172 109,266 0.134 0.021 0.022 0.005 0.0056 0.0015 23 2 8,575 93,837 0.283 0.022 0.060 0.007 0.0192 0.0027 23 3 4,070 80,993 0.331 0.034 0.066 0.010 0.0194 0.0038 23 4 7,862 122,172 0.070 0.014 0.010 0.002 0.0021 0.0006 23 5 9,234 123,803 0.073 0.017 0.011 0.003 0.0030 0.0008 23 6 3,722 100,347 0.150 0.026 0.026 0.005 0.0073 0.0016 23 7 9,954 93,050 0.214 0.034 0.038 0.007 0.0104 0.0024 23 8 8,360 67,606 0.486 0.037 0.111 0.013 0.0358 0.0052 23 9 7,905 71,576 0.411 0.038 0.082 0.011 0.0237 0.0041 23 10 9,272 73,010 0.418 0.030 0.092 0.009 0.0290 0.0037 23 11 4,708 67,774 0.508 0.032 0.115 0.012 0.0365 0.0052 23 12 2,696 67,025 0.511 0.031 0.131 0.013 0.0462 0.0058 23 13 3,327 82,036 0.380 0.028 0.089 0.009 0.0304 0.0040 23 14 1,943 71,431 0.433 0.036 0.092 0.012 0.0285 0.0047 23 15 3,467 87,049 0.259 0.030 0.046 0.008 0.0125 0.0026 23 16 5,304 97,713 0.189 0.026 0.033 0.006 0.0088 0.0019 25 1 4,369 154,699 0.296 0.020 0.107 0.013 0.0515 0.0089 25 2 7,343 161,506 0.415 0.028 0.195 0.018 0.1146 0.0126 25 3 4,208 266,645 0.383 0.017 0.183 0.011 0.1076 0.0079 25 4 3,672 59,458 0.645 0.021 0.312 0.012 0.1845 0.0085 127 25 5 4,938 69,387 0.514 0.046 0.181 0.023 0.0846 0.0132 25 6 2,741 85,405 0.466 0.024 0.183 0.012 0.0942 0.0083 25 7 4,442 67,618 0.575 0.042 0.194 0.023 0.0859 0.0128 25 8 3,899 77,823 0.442 0.036 0.135 0.017 0.0564 0.0094 25 9 3,563 63,127 0.544 0.031 0.186 0.016 0.0863 0.0093 25 10 14,745 65,523 0.550 0.023 0.202 0.011 0.0989 0.0072 25 11 6,776 90,119 0.417 0.033 0.128 0.014 0.0546 0.0070 25 12 8,308 65,066 0.526 0.025 0.209 0.013 0.1090 0.0081 25 13 6,139 74,507 0.472 0.036 0.141 0.017 0.0567 0.0092 25 15 12,123 148,671 0.350 0.032 0.114 0.014 0.0500 0.0079 26 1 919 48,490 0.777 0.031 0.377 0.036 0.2146 0.0290 26 2 266 95,663 0.390 0.077 0.108 0.032 0.0434 0.0163 26 3 850 93,048 0.392 0.053 0.122 0.024 0.0530 0.0132 26 4 667 41,962 0.820 0.032 0.400 0.023 0.2307 0.0173 26 5 2,292 62,899 0.670 0.031 0.273 0.022 0.1424 0.0146 26 6 2,186 84,289 0.524 0.043 0.177 0.024 0.0804 0.0139 26 7 1,790 114,905 0.363 0.032 0.124 0.015 0.0577 0.0087 26 8 783 70,882 0.584 0.042 0.213 0.025 0.1039 0.0162 26 9 2,687 90,355 0.498 0.029 0.171 0.017 0.0806 0.0103 26 10 943 81,838 0.621 0.036 0.243 0.023 0.1207 0.0163 26 11 1,282 69,810 0.576 0.034 0.266 0.015 0.1506 0.0112 26 12 1,814 238,929 0.072 0.025 0.013 0.006 0.0034 0.0018 26 13 3,702 87,620 0.524 0.031 0.222 0.021 0.1186 0.0145 26 14 1,000 111,863 0.348 0.047 0.117 0.017 0.0527 0.0101 27 1 3,181 89,721 0.285 0.030 0.064 0.009 0.0206 0.0036 27 2 4,788 76,587 0.352 0.031 0.073 0.010 0.0218 0.0039 27 3 7,748 83,124 0.290 0.026 0.059 0.008 0.0174 0.0030 27 4 8,220 98,596 0.193 0.026 0.035 0.006 0.0099 0.0021 27 5 10,264 70,199 0.399 0.032 0.080 0.011 0.0234 0.0045 27 6 4,708 93,283 0.215 0.032 0.039 0.007 0.0107 0.0024 27 7 9,305 76,159 0.326 0.029 0.062 0.009 0.0177 0.0033 27 8 694 101,088 0.177 0.032 0.033 0.008 0.0096 0.0027 27 9 6,343 78,189 0.328 0.044 0.067 0.014 0.0204 0.0055 28 1 478 260,038 0.139 0.033 0.026 0.009 0.0065 0.0027 28 2 351 234,228 0.027 0.014 0.004 0.003 0.0008 0.0008 28 3 296 209,043 0.014 0.012 0.002 0.002 0.0004 0.0007 28 4 1,025 189,409 0.047 0.020 0.007 0.003 0.0015 0.0008 28 5 689 173,372 0.082 0.017 0.015 0.004 0.0036 0.0012 28 6 994 170,901 0.070 0.014 0.012 0.003 0.0028 0.0008 28 7 683 147,921 0.134 0.031 0.022 0.007 0.0052 0.0018 28 8 575 173,139 0.029 0.014 0.004 0.002 0.0007 0.0005 28 9 812 132,010 0.177 0.034 0.031 0.007 0.0075 0.0021 29 1 4,325 51,660 0.819 0.023 0.244 0.021 0.0932 0.0129 29 2 1,501 59,548 0.726 0.030 0.204 0.018 0.0767 0.0103 29 3 4,512 53,098 0.828 0.026 0.228 0.026 0.0791 0.0141 29 4 3,577 54,638 0.813 0.039 0.209 0.029 0.0692 0.0143 29 5 4,395 65,368 0.617 0.043 0.137 0.020 0.0418 0.0090 128 29 6 2,339 67,842 0.572 0.058 0.119 0.021 0.0343 0.0082 29 7 2,461 50,827 0.901 0.033 0.258 0.034 0.0906 0.0188 29 8 3,616 53,875 0.830 0.033 0.224 0.025 0.0781 0.0131 29 9 2,741 76,119 0.441 0.045 0.099 0.012 0.0333 0.0044 29 10 3,949 67,447 0.567 0.066 0.130 0.019 0.0449 0.0066 29 11 3,291 67,198 0.580 0.058 0.117 0.019 0.0340 0.0078 29 12 8,470 67,047 0.615 0.041 0.142 0.015 0.0465 0.0071 29 13 2,785 57,231 0.746 0.033 0.192 0.022 0.0657 0.0115 29 14 2,637 58,830 0.757 0.031 0.191 0.019 0.0633 0.0105 29 15 2,013 78,484 0.391 0.057 0.077 0.011 0.0249 0.0044 29 16 3,474 66,866 0.565 0.056 0.133 0.015 0.0457 0.0059 29 17 2,575 52,780 0.848 0.044 0.258 0.044 0.1021 0.0271 29 18 5,433 79,902 0.457 0.038 0.100 0.012 0.0316 0.0056 29 19 5,765 73,073 0.526 0.041 0.119 0.014 0.0387 0.0066 29 20 3,161 62,903 0.680 0.049 0.187 0.020 0.0682 0.0099 30 1 3,879 75,991 0.454 0.034 0.113 0.011 0.0389 0.0042 30 2 5,326 86,244 0.354 0.034 0.077 0.009 0.0243 0.0033 30 3 8,344 73,367 0.484 0.029 0.106 0.010 0.0324 0.0037 30 4 3,700 87,756 0.351 0.026 0.073 0.008 0.0219 0.0031 30 5 3,492 70,979 0.545 0.034 0.129 0.014 0.0420 0.0059 30 6 7,249 76,545 0.490 0.040 0.113 0.014 0.0361 0.0054 30 7 2,483 76,034 0.470 0.047 0.097 0.015 0.0288 0.0053 30 8 5,528 75,611 0.459 0.039 0.094 0.011 0.0278 0.0038 30 9 8,050 74,946 0.473 0.037 0.101 0.011 0.0305 0.0038 31 1 4,837 87,046 0.274 0.037 0.056 0.011 0.0175 0.0042 31 2 8,699 82,691 0.329 0.032 0.080 0.011 0.0287 0.0048 31 3 9,127 104,889 0.308 0.026 0.072 0.009 0.0248 0.0043 31 4 8,087 84,519 0.337 0.028 0.080 0.010 0.0277 0.0044 31 5 6,676 83,196 0.379 0.025 0.099 0.011 0.0371 0.0057 31 6 8,537 89,629 0.304 0.032 0.070 0.010 0.0237 0.0045 129 Table 9: Estimates of poverty indicators (urban areas) Governorate District #hhno avg_ ^ y avg_FGT0 se_FGT0 avg_FGT1 se_FGT1 avg_FGT2 se_FGT2 11 1 700 62,143 0.557 0.049 0.151 0.020 0.0537 0.0085 11 2 6,765 114,030 0.166 0.022 0.030 0.005 0.0080 0.0016 11 3 478 136,221 0.145 0.022 0.031 0.007 0.0097 0.0032 11 4 844 93,316 0.301 0.035 0.068 0.011 0.0214 0.0043 11 5 212 138,494 0.152 0.040 0.027 0.010 0.0069 0.0035 11 6 724 96,856 0.269 0.031 0.053 0.009 0.0148 0.0032 11 7 303 97,795 0.218 0.040 0.038 0.010 0.0100 0.0036 11 8 840 100,875 0.186 0.035 0.034 0.009 0.0093 0.0033 11 11 1,093 156,565 0.029 0.013 0.003 0.002 0.0007 0.0004 11 12 1,785 93,761 0.294 0.028 0.065 0.008 0.0207 0.0032 11 13 490 116,143 0.105 0.031 0.016 0.006 0.0039 0.0018 11 14 349 93,977 0.285 0.035 0.062 0.011 0.0189 0.0044 11 15 324 92,177 0.248 0.042 0.042 0.010 0.0104 0.0031 11 16 6,537 96,747 0.279 0.025 0.061 0.008 0.0191 0.0032 11 17 177 135,216 0.077 0.049 0.009 0.008 0.0018 0.0024 11 18 11,283 136,158 0.096 0.014 0.015 0.003 0.0037 0.0009 11 19 16,239 154,468 0.076 0.013 0.012 0.003 0.0029 0.0008 12 1 271 64,235 0.572 0.063 0.209 0.029 0.0974 0.0156 12 2 991 112,861 0.217 0.034 0.053 0.010 0.0190 0.0045 12 4 1,574 108,139 0.328 0.033 0.091 0.012 0.0356 0.0055 12 6 130 46,978 0.733 0.067 0.287 0.041 0.1388 0.0255 12 9 665 56,149 0.700 0.053 0.257 0.032 0.1186 0.0186 12 10 2,384 122,546 0.237 0.020 0.068 0.007 0.0272 0.0036 12 11 7,458 100,949 0.351 0.023 0.100 0.009 0.0402 0.0045 13 1 8,935 161,486 0.116 0.015 0.023 0.004 0.0069 0.0014 13 2 28,942 163,779 0.122 0.009 0.025 0.003 0.0080 0.0010 13 3 15,137 144,936 0.145 0.016 0.030 0.004 0.0095 0.0017 13 4 14,640 176,514 0.093 0.012 0.019 0.003 0.0057 0.0012 13 5 42,133 196,592 0.099 0.008 0.021 0.002 0.0066 0.0008 13 6 14,815 247,476 0.052 0.007 0.010 0.002 0.0029 0.0006 13 7 10,260 218,267 0.063 0.008 0.012 0.002 0.0036 0.0008 13 8 39,035 176,564 0.142 0.010 0.032 0.003 0.0105 0.0013 13 9 23,959 204,694 0.067 0.007 0.013 0.002 0.0038 0.0008 13 10 18,730 118,017 0.272 0.020 0.066 0.007 0.0228 0.0028 13 19 3,834 125,575 0.210 0.027 0.046 0.008 0.0148 0.0031 13 24 16,246 132,385 0.231 0.019 0.053 0.006 0.0178 0.0025 14 4 319 85,671 0.325 0.085 0.074 0.026 0.0240 0.0106 14 5 162 65,539 0.552 0.116 0.138 0.044 0.0462 0.0189 14 6 517 69,779 0.496 0.083 0.120 0.029 0.0399 0.0120 14 8 341 109,096 0.209 0.049 0.045 0.014 0.0144 0.0055 14 9 3,592 113,462 0.151 0.034 0.029 0.008 0.0087 0.0029 14 11 94 231,193 0.002 0.006 0.000 0.001 0.0001 0.0003 14 13 6,035 125,924 0.106 0.018 0.020 0.005 0.0059 0.0016 130 14 14 70 101,838 0.130 0.121 0.021 0.024 0.0053 0.0074 14 16 1,656 109,854 0.161 0.039 0.031 0.010 0.0093 0.0037 15 1 429 76,284 0.486 0.096 0.146 0.041 0.0593 0.0201 15 2 59 55,884 0.690 0.133 0.228 0.072 0.0968 0.0388 15 3 353 76,001 0.490 0.098 0.164 0.042 0.0717 0.0214 15 4 302 55,554 0.659 0.068 0.246 0.038 0.1153 0.0217 15 5 1,362 70,693 0.556 0.061 0.199 0.030 0.0909 0.0165 15 6 479 141,320 0.180 0.054 0.046 0.019 0.0170 0.0088 15 7 715 100,783 0.334 0.051 0.101 0.020 0.0413 0.0101 15 8 289 51,839 0.723 0.082 0.264 0.053 0.1196 0.0311 15 11 497 94,445 0.353 0.081 0.096 0.032 0.0366 0.0153 15 12 3,265 86,967 0.434 0.037 0.137 0.017 0.0579 0.0086 15 14 1,570 86,643 0.470 0.054 0.150 0.026 0.0632 0.0134 15 16 298 151,933 0.073 0.047 0.012 0.011 0.0032 0.0040 15 17 24,787 131,738 0.222 0.017 0.059 0.006 0.0227 0.0029 15 18 21,958 152,105 0.164 0.015 0.040 0.005 0.0145 0.0019 15 19 21,670 142,864 0.198 0.014 0.052 0.005 0.0195 0.0023 15 21 212 107,627 0.278 0.085 0.072 0.027 0.0268 0.0116 16 1 293 85,622 0.238 0.072 0.039 0.014 0.0098 0.0042 16 3 141 80,235 0.265 0.087 0.046 0.019 0.0123 0.0059 16 4 297 72,974 0.457 0.135 0.107 0.044 0.0346 0.0177 16 5 1,475 110,008 0.284 0.033 0.074 0.011 0.0268 0.0050 16 6 1,244 94,057 0.225 0.064 0.041 0.014 0.0119 0.0046 16 7 468 88,680 0.284 0.086 0.057 0.022 0.0164 0.0077 16 8 787 81,870 0.359 0.058 0.095 0.016 0.0346 0.0064 16 9 878 63,586 0.647 0.054 0.200 0.025 0.0782 0.0126 16 10 818 104,531 0.124 0.036 0.020 0.009 0.0051 0.0028 16 11 256 101,434 0.183 0.066 0.037 0.017 0.0110 0.0060 16 12 407 86,205 0.221 0.080 0.038 0.019 0.0101 0.0060 17 2 3,167 108,383 0.261 0.032 0.060 0.011 0.0207 0.0049 17 3 951 132,450 0.139 0.033 0.027 0.008 0.0082 0.0030 17 4 3,454 109,568 0.244 0.036 0.055 0.012 0.0184 0.0051 17 5 426 115,924 0.223 0.048 0.052 0.017 0.0178 0.0081 17 6 93 66,820 0.520 0.116 0.144 0.049 0.0540 0.0236 17 7 254 112,002 0.173 0.052 0.032 0.014 0.0092 0.0056 17 8 82 99,944 0.147 0.061 0.028 0.015 0.0080 0.0060 17 11 373 76,617 0.408 0.063 0.098 0.023 0.0335 0.0105 17 15 1,208 139,877 0.158 0.031 0.038 0.009 0.0141 0.0041 17 16 137 147,382 0.089 0.037 0.016 0.010 0.0045 0.0039 17 17 52 90,239 0.278 0.102 0.057 0.027 0.0167 0.0108 17 18 325 119,967 0.182 0.048 0.041 0.015 0.0134 0.0060 17 20 230 175,675 0.026 0.019 0.003 0.003 0.0007 0.0011 17 21 424 137,437 0.112 0.029 0.021 0.007 0.0062 0.0028 17 22 175 91,509 0.363 0.077 0.090 0.028 0.0321 0.0128 17 23 116 56,954 0.649 0.096 0.199 0.054 0.0794 0.0291 17 24 104 81,521 0.408 0.102 0.098 0.037 0.0330 0.0159 17 25 290 96,522 0.246 0.072 0.049 0.020 0.0145 0.0078 131 17 28 4,276 121,873 0.207 0.034 0.046 0.011 0.0150 0.0044 18 1 1,682 73,754 0.418 0.025 0.112 0.010 0.0421 0.0050 18 2 2,164 94,369 0.231 0.031 0.048 0.009 0.0149 0.0034 18 3 457 101,775 0.284 0.053 0.089 0.030 0.0390 0.0178 18 4 195 104,739 0.209 0.055 0.042 0.015 0.0122 0.0060 18 5 1,008 83,468 0.320 0.032 0.075 0.011 0.0253 0.0051 18 6 637 85,314 0.268 0.029 0.055 0.008 0.0167 0.0033 18 7 3,430 84,559 0.317 0.029 0.073 0.010 0.0243 0.0040 18 8 738 65,817 0.466 0.093 0.114 0.034 0.0389 0.0149 18 9 2,941 82,989 0.312 0.028 0.069 0.009 0.0223 0.0038 18 10 7,128 101,705 0.217 0.014 0.048 0.005 0.0159 0.0020 18 11 263 85,092 0.311 0.046 0.082 0.018 0.0307 0.0094 18 13 4,759 78,146 0.364 0.019 0.090 0.008 0.0316 0.0038 18 14 1,325 92,711 0.248 0.028 0.052 0.008 0.0157 0.0032 18 15 284 79,636 0.286 0.042 0.069 0.014 0.0247 0.0067 18 16 1,830 89,443 0.271 0.021 0.061 0.007 0.0197 0.0029 18 17 6,731 89,192 0.286 0.020 0.065 0.007 0.0216 0.0029 18 19 2,018 83,044 0.323 0.037 0.079 0.013 0.0280 0.0058 18 20 1,741 83,090 0.305 0.030 0.068 0.009 0.0222 0.0039 18 21 20,706 115,721 0.181 0.011 0.038 0.004 0.0119 0.0015 18 22 12,561 159,491 0.105 0.011 0.020 0.003 0.0061 0.0010 18 23 23,081 112,743 0.213 0.012 0.047 0.004 0.0154 0.0017 18 24 4,787 98,897 0.232 0.023 0.051 0.007 0.0166 0.0030 18 25 2,781 86,199 0.297 0.020 0.069 0.007 0.0229 0.0031 18 26 1,662 83,710 0.309 0.028 0.070 0.009 0.0231 0.0041 19 1 195 63,099 0.486 0.127 0.169 0.049 0.0756 0.0229 19 2 230 224,009 0.000 0.001 0.000 0.000 0.0000 0.0000 19 6 132 49,189 0.800 0.174 0.255 0.096 0.1022 0.0490 19 7 1,171 115,474 0.291 0.062 0.080 0.022 0.0304 0.0097 19 8 1,430 154,519 0.038 0.020 0.005 0.004 0.0011 0.0011 19 9 474 100,480 0.250 0.096 0.054 0.025 0.0175 0.0094 19 10 5,609 115,484 0.209 0.034 0.041 0.009 0.0117 0.0031 19 11 5,011 114,549 0.377 0.032 0.099 0.013 0.0356 0.0063 19 13 1,507 148,498 0.105 0.031 0.024 0.010 0.0083 0.0040 19 14 1,741 89,843 0.390 0.049 0.107 0.018 0.0406 0.0081 19 15 6,647 93,577 0.346 0.025 0.100 0.009 0.0392 0.0038 19 16 148 204,650 0.005 0.017 0.001 0.003 0.0001 0.0006 19 17 4,206 86,107 0.401 0.053 0.109 0.020 0.0404 0.0087 19 18 217 116,042 0.262 0.107 0.105 0.039 0.0504 0.0180 19 19 308 175,177 0.060 0.064 0.009 0.012 0.0022 0.0033 19 20 68 394,768 0.000 0.000 0.000 0.000 0.0000 0.0000 19 21 238 111,628 0.353 0.093 0.098 0.043 0.0356 0.0201 19 22 160 53,219 0.751 0.166 0.239 0.090 0.0946 0.0460 19 24 140 38,413 0.912 0.079 0.373 0.071 0.1751 0.0451 19 25 287 137,737 0.157 0.070 0.034 0.021 0.0109 0.0086 19 26 1,116 154,823 0.150 0.061 0.039 0.020 0.0138 0.0082 19 27 331 94,821 0.246 0.201 0.049 0.052 0.0142 0.0184 132 19 28 389 174,627 0.077 0.062 0.015 0.016 0.0046 0.0054 19 29 22,113 148,128 0.177 0.019 0.043 0.007 0.0150 0.0029 20 1 533 87,671 0.405 0.053 0.129 0.024 0.0550 0.0126 20 2 3,305 100,704 0.304 0.029 0.074 0.011 0.0259 0.0047 20 3 92 55,408 0.651 0.102 0.227 0.060 0.1001 0.0331 20 4 132 105,615 0.311 0.097 0.083 0.037 0.0306 0.0173 20 7 328 82,696 0.490 0.083 0.147 0.039 0.0597 0.0203 20 8 19,145 123,736 0.262 0.023 0.065 0.008 0.0234 0.0033 20 9 113 127,962 0.178 0.075 0.039 0.021 0.0123 0.0078 20 11 664 127,528 0.178 0.036 0.040 0.010 0.0135 0.0041 21 4 129 62,602 0.585 0.065 0.158 0.024 0.0590 0.0113 21 5 246 113,923 0.190 0.043 0.045 0.012 0.0153 0.0050 21 7 1,467 99,099 0.284 0.027 0.072 0.008 0.0266 0.0030 21 10 869 69,937 0.460 0.048 0.118 0.014 0.0442 0.0053 21 11 149 46,342 0.916 0.033 0.279 0.034 0.1043 0.0184 21 12 348 62,427 0.593 0.055 0.169 0.023 0.0650 0.0108 21 13 2,626 135,175 0.190 0.016 0.052 0.006 0.0198 0.0026 21 14 270 63,613 0.543 0.063 0.150 0.022 0.0573 0.0093 21 15 549 59,890 0.593 0.066 0.164 0.024 0.0622 0.0098 21 16 660 75,949 0.403 0.047 0.104 0.013 0.0389 0.0052 21 17 238 69,903 0.470 0.100 0.124 0.029 0.0469 0.0124 22 1 370 130,131 0.071 0.024 0.012 0.006 0.0029 0.0019 22 2 472 164,391 0.135 0.019 0.029 0.006 0.0092 0.0023 22 3 56 141,060 0.118 0.049 0.029 0.015 0.0108 0.0074 22 5 643 134,388 0.094 0.024 0.016 0.005 0.0044 0.0017 22 7 492 165,907 0.124 0.045 0.027 0.012 0.0088 0.0045 22 8 379 115,549 0.182 0.043 0.037 0.011 0.0114 0.0039 22 9 306 108,368 0.149 0.038 0.025 0.009 0.0065 0.0030 22 10 1,692 100,091 0.230 0.019 0.051 0.005 0.0166 0.0022 22 11 1,389 95,346 0.260 0.026 0.062 0.009 0.0205 0.0035 22 14 85 139,216 0.029 0.025 0.003 0.003 0.0006 0.0007 22 15 5,665 110,356 0.215 0.015 0.050 0.004 0.0168 0.0016 24 1 10,962 108,371 0.266 0.020 0.061 0.007 0.0204 0.0029 24 2 13,925 125,410 0.203 0.016 0.045 0.005 0.0146 0.0022 24 3 15,338 155,247 0.108 0.010 0.020 0.003 0.0060 0.0009 24 4 9,019 122,561 0.209 0.018 0.045 0.006 0.0143 0.0024 24 5 8,025 136,835 0.128 0.013 0.024 0.004 0.0070 0.0013 24 6 7,533 157,479 0.109 0.011 0.021 0.003 0.0062 0.0011 24 7 11,811 152,888 0.107 0.010 0.021 0.003 0.0061 0.0010 24 8 6,354 161,529 0.116 0.011 0.023 0.003 0.0071 0.0012 25 1 541 253,029 0.022 0.013 0.004 0.002 0.0012 0.0008 25 4 200 132,115 0.107 0.071 0.021 0.016 0.0065 0.0056 25 6 271 79,185 0.533 0.089 0.163 0.051 0.0605 0.0256 25 7 1,149 89,993 0.237 0.052 0.053 0.013 0.0179 0.0046 25 8 161 63,977 0.582 0.141 0.139 0.046 0.0459 0.0175 25 9 440 73,372 0.560 0.049 0.180 0.032 0.0714 0.0176 25 10 204 67,470 0.537 0.127 0.131 0.046 0.0439 0.0187 133 25 14 4,005 142,953 0.115 0.023 0.025 0.006 0.0082 0.0021 25 15 1,524 121,953 0.184 0.041 0.036 0.010 0.0108 0.0033 26 7 224 132,138 0.049 0.027 0.006 0.005 0.0013 0.0011 26 8 106 147,039 0.040 0.028 0.007 0.006 0.0019 0.0021 26 9 789 99,974 0.314 0.036 0.072 0.012 0.0239 0.0054 26 12 1,671 158,936 0.099 0.023 0.018 0.006 0.0053 0.0022 26 13 535 114,183 0.253 0.039 0.052 0.011 0.0159 0.0044 27 1 888 98,265 0.199 0.047 0.031 0.010 0.0074 0.0028 27 2 674 86,880 0.278 0.053 0.034 0.011 0.0079 0.0032 27 3 561 112,480 0.050 0.022 0.005 0.003 0.0008 0.0005 27 4 243 91,316 0.156 0.065 0.020 0.011 0.0039 0.0027 27 6 210 95,259 0.123 0.061 0.013 0.009 0.0023 0.0019 27 8 1,593 94,472 0.209 0.043 0.035 0.010 0.0091 0.0036 28 3 298 83,815 0.286 0.060 0.076 0.018 0.0300 0.0076 28 4 1,516 194,503 0.080 0.020 0.023 0.006 0.0092 0.0029 28 6 494 105,123 0.148 0.034 0.029 0.008 0.0088 0.0031 28 7 821 151,012 0.045 0.015 0.008 0.003 0.0022 0.0011 28 8 718 170,434 0.078 0.017 0.021 0.006 0.0082 0.0030 28 9 227 109,834 0.128 0.045 0.020 0.009 0.0050 0.0031 29 1 320 62,959 0.682 0.078 0.194 0.047 0.0704 0.0243 29 2 742 97,020 0.406 0.089 0.098 0.035 0.0335 0.0161 29 3 117 55,548 0.817 0.135 0.235 0.084 0.0850 0.0435 29 4 282 57,255 0.742 0.097 0.205 0.051 0.0720 0.0246 29 5 273 84,955 0.411 0.143 0.079 0.042 0.0223 0.0150 29 6 274 58,691 0.773 0.083 0.259 0.054 0.1082 0.0314 29 8 240 56,829 0.759 0.083 0.240 0.054 0.0950 0.0302 29 9 161 66,391 0.670 0.149 0.175 0.063 0.0610 0.0278 29 10 220 86,119 0.407 0.079 0.087 0.027 0.0257 0.0106 29 11 1,752 85,090 0.426 0.047 0.093 0.017 0.0289 0.0073 29 13 122 61,041 0.741 0.134 0.197 0.065 0.0677 0.0297 29 15 9,391 108,163 0.259 0.033 0.049 0.008 0.0141 0.0030 29 16 176 60,328 0.768 0.087 0.221 0.048 0.0801 0.0250 29 17 1,303 87,609 0.473 0.062 0.116 0.023 0.0390 0.0101 29 19 1,762 94,644 0.348 0.050 0.072 0.014 0.0212 0.0053 30 1 945 140,251 0.212 0.043 0.053 0.015 0.0191 0.0065 30 2 1,896 109,046 0.288 0.040 0.067 0.015 0.0227 0.0065 30 3 1,237 93,633 0.325 0.035 0.074 0.012 0.0249 0.0053 30 4 560 110,939 0.248 0.044 0.058 0.014 0.0199 0.0061 30 5 269 101,415 0.376 0.049 0.109 0.024 0.0424 0.0126 30 6 2,365 98,704 0.304 0.034 0.074 0.012 0.0267 0.0052 30 7 59 112,121 0.120 0.060 0.032 0.013 0.0127 0.0062 30 9 386 94,921 0.419 0.040 0.152 0.022 0.0722 0.0140 134 Figure 1 1 .8 Rural headcount index plus/minus 2 standard errors Headcount index .4 .2 0 .6 0 100 200 300 Districts sorted by headcount index Figure 2 Urban headcount index plus/minus 2 standard errors 1.5 1 Headcount index .5 0 -.5 0 50 100 150 200 250 Districts sorted by headcount index 135 Appendix A A.1. In this section, the ELL method described in the main text is applied to measure food poverty in Yemen. Food consumption per capita is used to replace consumption per capita in the main text to measure poverty. All the steps of implementing the method are as described in the main text. Table A1 shows the results of all the food consumption models. Table A2 compares the estimates of headcount at governorate level using two methods (ELL method and directly calculated using the survey data). Table A3 and A4 list the estimates of food poverty indicators for each district. Figure A1 and A2 show the mean and plus/minus 2 standard errors of the poverty indicators. A.2. In general, the estimates of the food poverty indicators are less good than the total consumption indicators shown in the main text. This mainly comes from the difficulty to get good models to explain the food consumption per capita for some strata. It can be seen that the R squares of the food consumption models are in general lower than the ones of the total consumption models. A.3. Sharing the same concern with the estimates of the poverty indicators using total consumption, the estimates of food poverty indicators have very big standard errors for the urban districts with a small number of households. 136 Table A1: Regression results of food consumption models Table A1.1: Rural Ibb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 0.630 0.204 0.211 0.173 0.266 0.022 dum_eaurable1 0.212 0.089 0.043 0.202 0.043 0.041 dum_eaurable10_ea -0.863 0.230 0.080 0.141 0.106 0.027 dum_eaurable15_ea 1.378 0.490 0.040 0.055 0.040 0.002 dum_eaurable1_ea 4.130 0.634 0.043 0.051 0.046 0.002 employed2 -1.733 0.248 0.233 0.095 0.237 0.012 headprim_ea -2.162 0.561 0.049 0.036 0.055 0.002 headread_ea 1.597 0.244 0.156 0.097 0.156 0.008 headsecond_ea -0.961 0.309 0.117 0.068 0.136 0.007 housetype3_ea 2.151 0.552 0.013 0.047 0.014 0.002 light1_ea 0.324 0.067 0.362 0.435 0.442 0.210 light2 0.468 0.135 0.027 0.163 0.021 0.021 nafemales -0.097 0.023 1.828 1.283 1.943 1.553 nkids -0.029 0.012 3.534 2.579 3.324 4.994 ownhouse1 0.129 0.059 0.854 0.353 0.885 0.102 ownhouse1_ea 0.492 0.147 0.839 0.135 0.852 0.017 primaryp 0.577 0.127 0.110 0.160 0.123 0.024 primary_ea 1.590 0.495 0.109 0.058 0.132 0.005 singlep -0.410 0.114 0.588 0.231 0.593 0.046 water2_ea 0.294 0.142 0.075 0.196 0.059 0.026 work9 -0.251 0.126 0.298 0.248 0.254 0.042 _intercept_ 10.258 0.159 obs. 463 R square 0.44 location effect 0.046 137 Table A1.2: Rural Abyan Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12 0.149 0.075 0.396 0.489 0.424 0.245 dum_eaurable12_ea -1.298 0.198 0.396 0.311 0.418 0.098 dum_eaurable16 3.042 0.528 0.002 0.044 0.005 0.005 dum_eaurable2_1 0.446 0.238 0.009 0.094 0.014 0.014 dum_eaurable9_ea 2.355 0.298 0.132 0.123 0.149 0.037 headiliter_ea 3.066 0.291 0.491 0.168 0.524 0.036 headsecond_ea 4.545 0.438 0.202 0.124 0.219 0.020 headsingle 0.308 0.114 0.057 0.231 0.069 0.065 housetype1_ea 0.358 0.148 0.857 0.233 0.896 0.047 kidp -0.491 0.150 0.383 0.230 0.375 0.055 ownhouse1_ea -3.108 0.701 0.932 0.107 0.929 0.003 singlep -0.906 0.165 0.587 0.213 0.571 0.050 work8 -5.375 0.664 0.035 0.070 0.037 0.006 _intercept_ 11.445 0.632 obs. 201 R square 0.59 location effect 0.077 Table A1.3: Rural Al-Baida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11 0.354 0.108 0.110 0.313 0.103 0.093 dum_eaurable11_ea 1.930 0.352 0.110 0.166 0.097 0.024 dum_eaurable14_ea -2.771 0.594 0.055 0.097 0.053 0.008 headiliter_ea 6.977 1.097 0.646 0.171 0.684 0.025 headread_ea 8.451 1.214 0.226 0.127 0.204 0.014 headsecond_ea 7.764 1.005 0.099 0.068 0.097 0.006 headuniv -0.389 0.181 0.015 0.120 0.027 0.027 malep 0.553 0.190 0.503 0.169 0.497 0.027 singlep -0.913 0.161 0.616 0.193 0.612 0.038 water4_ea 0.584 0.167 0.913 0.180 0.899 0.048 work10 -6.599 0.790 0.030 0.057 0.026 0.003 work4 1.197 0.236 0.109 0.146 0.098 0.024 _intercept_ 2.788 1.187 obs. 222 R square 0.46 location effect -* *No location effect 138 Table A1.4: Rural Taiz Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12 0.131 0.049 0.332 0.471 0.323 0.219 dum_eaurable1_ea -6.751 1.304 0.026 0.029 0.020 0.0003 employed1 10.611 1.546 0.186 0.097 0.200 0.014 employed2 -9.385 1.475 0.200 0.099 0.214 0.014 employ_nonself 1.145 0.165 0.372 0.258 0.369 0.050 kidp -0.621 0.174 0.409 0.255 0.416 0.064 light1 0.149 0.064 0.194 0.395 0.158 0.133 light4 0.439 0.142 0.024 0.154 0.020 0.020 light6_ea 2.033 0.404 0.038 0.104 0.033 0.005 nafemales -0.096 0.018 1.387 1.321 1.968 1.778 namales -0.082 0.019 1.387 1.321 1.402 1.737 nkids -0.037 0.019 3.135 2.440 3.076 5.139 ownhouse1_ea 0.672 0.296 0.929 0.089 0.932 0.018 sewage2 0.187 0.045 0.315 0.464 0.352 0.229 water2 0.479 0.092 0.060 0.237 0.078 0.072 water2_ea -0.748 0.183 0.059 0.164 0.048 0.019 work10 3.437 0.444 0.038 0.079 0.028 0.003 work8 0.426 0.144 0.142 0.173 0.176 0.044 work9 1.337 0.176 0.209 0.234 0.199 0.039 _intercept_ 9.560 0.354 obs. 450 R square 0.48 location effect 0.083 Table A1.5: Rural Al-Jawf Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 1.376 0.248 0.228 0.137 0.247 0.019 dum_eaurable18 1.587 0.295 0.001 0.037 0.008 0.008 dum_eaurable1_ea 0.493 0.200 0.127 0.161 0.133 0.022 dum_eaurable8_ea -3.987 0.815 0.012 0.044 0.012 0.001 headread_ea -1.024 0.417 0.066 0.088 0.066 0.006 headsecond_ea 3.340 0.419 0.138 0.136 0.125 0.009 headuniv_ea -8.282 1.515 0.017 0.038 0.019 0.001 housetype1_ea 0.293 0.093 0.733 0.292 0.691 0.082 light5 0.216 0.066 0.732 0.443 0.715 0.205 marriedp 0.432 0.144 0.299 0.125 0.324 0.038 nafemales -0.226 0.031 1.766 1.220 1.783 1.249 ownhouse2 0.593 0.261 0.007 0.081 0.006 0.006 _intercept_ 9.676 0.107 obs. 148 R square 0.63 location effect 0.11 139 Table A1.6: Rural Hajja Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) cook2_ea -0.424 0.198 0.054 0.181 0.036 0.017 dum_eaurable1 0.174 0.102 0.043 0.203 0.058 0.055 dum_eaurable12_ea -0.409 0.238 0.145 0.217 0.114 0.016 dum_eaurable15_ea 1.665 0.387 0.030 0.079 0.038 0.009 dum_eaurable5 0.630 0.280 0.011 0.105 0.006 0.006 dum_eaurable6 0.338 0.163 0.014 0.116 0.019 0.018 dum_eaurable9_ea -1.437 0.356 0.093 0.106 0.086 0.011 employed1 -0.634 0.344 0.258 0.112 0.276 0.007 employ_self 1.046 0.189 0.829 0.193 0.795 0.050 headage -0.005 0.002 41.659 14.961 42.754 221.164 headprim_ea 4.001 0.662 0.049 0.044 0.050 0.002 headuniv_ea 3.519 1.344 0.016 0.026 0.015 0.001 highprim -0.209 0.083 0.068 0.252 0.093 0.085 kidp -0.814 0.155 0.458 0.242 0.448 0.054 light4 -0.299 0.127 0.033 0.179 0.036 0.034 nafemales -0.126 0.022 1.788 1.428 1.700 1.314 namales -0.072 0.020 1.844 1.554 1.803 1.765 ownhouse1_ea -0.959 0.212 0.925 0.103 0.904 0.019 sewage2 0.515 0.078 0.080 0.271 0.111 0.099 singlep -0.595 0.166 0.598 0.225 0.603 0.048 work5 2.259 0.666 0.016 0.050 0.021 0.002 work9 -0.682 0.159 0.533 0.278 0.544 0.087 _intercept_ 12.320 0.263 obs. 346 R square 0.56 location effect 0.170 140 Table A1.7: Rural Al-Hodeida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 0.489 0.186 0.261 0.171 0.261 0.021 amalep 1.040 0.228 0.256 0.187 0.254 0.022 dum_eaurable12_ea 1.502 0.202 0.155 0.168 0.168 0.030 dum_eaurable5 0.386 0.205 0.006 0.075 0.008 0.008 dum_eaurable6_ea -4.515 0.563 0.063 0.071 0.060 0.004 employed1 -6.331 1.510 0.328 0.115 0.345 0.016 employed2 6.479 1.453 0.342 0.113 0.360 0.017 employ_nonself -3.246 0.663 0.108 0.150 0.106 0.017 employ_self -1.267 0.551 0.870 0.162 0.866 0.019 headage 0.006 0.002 43.742 16.789 42.535 203.554 headprim_ea 2.915 0.815 0.039 0.034 0.040 0.002 headsingle 0.369 0.144 0.034 0.181 0.037 0.035 housetype1 0.134 0.049 0.537 0.499 0.504 0.251 light1_ea 2.553 0.388 0.034 0.148 0.026 0.017 light3_ea 5.920 0.500 0.022 0.102 0.025 0.016 light5_ea 2.693 0.274 0.862 0.243 0.887 0.045 marriedp 0.316 0.139 0.404 0.269 0.399 0.051 namales -0.168 0.029 1.446 1.180 1.525 1.087 ownhouse2 0.900 0.298 0.008 0.091 0.006 0.006 primaryp 0.486 0.215 0.053 0.129 0.053 0.013 water2_ea 0.427 0.108 0.104 0.236 0.136 0.083 water3_ea -0.208 0.099 0.185 0.339 0.159 0.091 work7 -0.935 0.257 0.076 0.118 0.074 0.012 _intercept_ 8.774 0.594 obs. 377 R square 0.52 location effect 0.11 Table A1.8: Rural Hadramout Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11 0.186 0.053 0.223 0.416 0.278 0.202 dum_eaurable4_ea -3.033 0.406 0.056 0.084 0.064 0.009 employed1 9.084 2.124 0.262 0.106 0.231 0.008 employed2 -9.895 2.005 0.274 0.106 0.242 0.008 employ_nonself -1.002 0.144 0.328 0.230 0.366 0.063 kidp -0.351 0.115 0.398 0.220 0.389 0.046 size -0.026 0.004 9.199 5.868 9.353 34.503 water1_ea -0.472 0.065 0.301 0.417 0.330 0.166 work5 3.523 0.533 0.021 0.057 0.035 0.006 _intercept_ 11.930 0.145 obs. 203 R square 0.53 location effect 0.047 141 Table A1.9: Rural Dhamar Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10 0.494 0.142 0.030 0.170 0.031 0.030 dum_eaurable10_ea -1.901 0.530 0.030 0.066 0.035 0.004 elderlyp 1.313 0.211 0.091 0.203 0.076 0.027 employed2 1.680 0.260 0.290 0.123 0.277 0.008 employ_nonself -0.342 0.101 0.263 0.242 0.305 0.068 headdivorced 0.628 0.199 0.010 0.098 0.013 0.013 headuniv_ea 6.894 1.175 0.023 0.029 0.028 0.001 kidp -0.301 0.127 0.439 0.239 0.442 0.053 light1 -0.101 0.063 0.272 0.445 0.322 0.219 marriedp 0.489 0.134 0.371 0.229 0.375 0.042 nelderly -0.241 0.044 0.436 0.699 0.438 0.484 water1 0.281 0.073 0.126 0.332 0.161 0.136 _intercept_ 10.325 0.126 obs. 315 R square 0.41 location effect 0.074 Table A1.10: Rural Shabwah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.187 0.104 0.256 0.437 0.250 0.189 dum_eaurable2_ea 17.972 3.335 0.008 0.014 0.009 2E-4 employ_nonself -0.348 0.219 0.475 0.267 0.515 0.069 headiliter_ea -1.754 0.320 0.502 0.183 0.478 0.037 headsingle 0.385 0.194 0.038 0.190 0.042 0.041 light2 0.292 0.155 0.098 0.297 0.081 0.075 light6 -0.492 0.135 0.096 0.295 0.090 0.082 nkids -0.026 0.013 4.958 3.782 4.507 11.344 singlep -0.865 0.269 0.615 0.184 0.617 0.025 water3_ea -0.646 0.222 0.080 0.223 0.109 0.065 work1 -1.463 0.352 0.237 0.201 0.231 0.032 _intercept_ 12.333 0.297 obs. 151 R square 0.50 location effect -* * No location effect 142 Table A1.11: Rural Sa'adah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 0.656 0.282 0.231 0.134 0.230 0.017 amalep×nafemalesinv* 1.120 0.382 0.109 0.147 0.095 0.014 dum_eaurable17×nafemalesinv 5.078 1.922 0.007 0.053 0.008 0.003 dum_eaurable1 0.276 0.073 0.197 0.398 0.168 0.141 dum_eaurable12 0.147 0.058 0.306 0.461 0.336 0.224 dum_eaurable15 0.279 0.101 0.057 0.231 0.074 0.069 dum_eaurable17 -1.596 0.704 0.019 0.137 0.021 0.021 dum_eaurable4 0.255 0.084 0.085 0.279 0.113 0.101 dum_eaurable5_ea 4.054 0.938 0.013 0.025 0.016 0.001 headiliter_ea -0.516 0.192 0.734 0.184 0.751 0.023 kidp -0.399 0.165 0.452 0.222 0.451 0.047 light5_ea 0.673 0.194 0.386 0.354 0.383 0.117 light5_ea×nafemalesinv -1.517 0.474 0.161 0.181 0.153 0.024 malep -0.481 0.185 0.516 0.172 0.516 0.025 nafemalesinv 3.922 1.140 0.404 0.188 0.403 0.030 nafemalesinv×ownhouse1_ea -2.981 1.243 0.362 0.170 0.358 0.022 ownhouse1_ea 2.702 0.748 0.900 0.113 0.895 0.014 work10 -3.851 0.787 0.023 0.049 0.024 0.002 _intercept_ 8.306 0.705 obs. 218 R square 0.45 location effect 0.004 *nafemalesinv=1/(1+nafemales) Table A1.12: Rural Sana'a Region Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1_ea -0.306 0.120 0.141 0.161 0.155 0.032 dum_eaurable5 0.248 0.102 0.026 0.160 0.032 0.031 dum_eaurable9 0.118 0.043 0.176 0.381 0.208 0.166 headiliter_ea -0.864 0.123 0.610 0.180 0.646 0.031 kidp -0.720 0.095 0.423 0.223 0.419 0.043 light4 -0.492 0.166 0.023 0.151 0.011 0.011 marriedp 0.295 0.098 0.362 0.210 0.365 0.038 nafemales -0.101 0.012 2.078 1.607 2.156 2.136 primaryp 0.493 0.145 0.127 0.163 0.107 0.018 work1 -0.445 0.096 0.245 0.233 0.223 0.043 work4 1.306 0.365 0.037 0.082 0.033 0.002 _intercept_ 11.683 0.134 obs. 256 R square 0.52 location effect 0.065 143 Table A1.13: Rural Laheg Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 1.005 0.256 0.256 0.191 0.254 0.026 dum_eaurable1 0.555 0.175 0.063 0.242 0.047 0.045 dum_eaurable10_ea 0.700 0.202 0.193 0.292 0.169 0.079 dum_eaurable12 0.148 0.090 0.364 0.481 0.383 0.237 dum_eaurable12_ea -0.862 0.184 0.364 0.337 0.393 0.123 dum_eaurable14 -0.900 0.299 0.026 0.159 0.012 0.012 dum_eaurable14_ea -1.923 0.679 0.026 0.095 0.019 0.004 dum_eaurable15 0.196 0.118 0.095 0.293 0.104 0.094 dum_eaurable1_ea 3.125 0.593 0.063 0.091 0.057 0.008 elderlyp 1.039 0.184 0.091 0.204 0.088 0.034 employed1 -0.880 0.328 0.207 0.095 0.212 0.010 headmarried 0.161 0.098 0.863 0.344 0.873 0.111 headread_ea 1.255 0.293 0.232 0.121 0.204 0.015 headsecond_ea -1.409 0.397 0.213 0.121 0.199 0.018 namales -0.089 0.030 1.808 1.704 1.786 2.392 ownhouse1_ea -4.135 0.921 0.939 0.109 0.961 0.002 work2 7.567 2.250 0.004 0.020 0.004 4E-4 work4 -1.567 0.285 0.120 0.186 0.109 0.030 _intercept_ 14.476 0.920 obs. 246 R square 0.39 location effect 0.03 Table A1.14: Rural Mareb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable11_ea -3.259 0.984 0.091 0.138 0.076 0.017 dum_eaurable13_ea 1.267 0.155 0.509 0.295 0.465 0.081 dum_eaurable17 0.285 0.170 0.061 0.239 0.049 0.047 dum_eaurable18 0.540 0.215 0.033 0.178 0.034 0.033 employed1 -15.210 7.728 0.197 0.107 0.197 0.017 employed2 14.137 7.544 0.207 0.110 0.201 0.018 headage -0.010 0.003 43.282 14.265 41.699 146.252 headiliter_ea -4.312 0.942 0.635 0.174 0.642 0.044 headread_ea -7.256 1.314 0.112 0.089 0.110 0.009 headsecond_ea -3.108 1.176 0.204 0.127 0.193 0.011 headuniv 0.659 0.251 0.030 0.172 0.021 0.020 nkids -0.087 0.015 4.436 3.386 4.341 6.313 ownhouse1_ea -3.929 0.955 0.881 0.144 0.894 0.016 primaryp 0.877 0.218 0.134 0.175 0.137 0.036 sewage2 0.340 0.113 0.338 0.473 0.333 0.223 work7 4.287 0.844 0.024 0.071 0.023 0.003 _intercept_ 18.473 1.527 obs. 158 R square 0.74 144 location effect -* *No location effect Table A1.15: Rural Al-Mahweet Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) Afemalep^2 0.540 0.221 0.090 0.126 0.095 0.014 dum_eaurable10 0.276 0.104 0.038 0.190 0.058 0.055 dum_eaurable13_ea 0.411 0.105 0.658 0.241 0.626 0.060 headiliter_ea -0.795 0.276 0.715 0.136 0.723 0.013 headmarried -0.379 0.117 0.893 0.309 0.905 0.086 headsecond_ea -0.847 0.476 0.107 0.074 0.108 0.006 light6_ea×light6_ea -0.208 0.085 0.066 0.192 0.088 0.069 malep -1.287 0.350 0.495 0.185 0.509 0.029 Malep^3 1.982 0.407 0.171 0.176 0.172 0.022 marriedp 0.498 0.274 0.360 0.221 0.382 0.049 marriedp^3 0.468 0.245 0.111 0.223 0.121 0.051 primaryp 1.775 0.505 0.091 0.147 0.080 0.014 Primaryp^2 -2.894 1.415 0.030 0.095 0.022 0.002 water3_1 -0.364 0.104 0.061 0.239 0.050 0.048 work4^3 12.754 3.983 0.002 0.008 0.002 1E-4 _intercept_ 11.390 0.245 obs. 249 R square 0.43 location effect -* *No location effect Table A1.16: Rural Al-Maharh Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1_1 0.175 0.080 0.171 0.377 0.214 0.169 dum_eaurable13_ea -1.489 0.157 0.332 0.256 0.389 0.051 dum_eaurable4_0*×nelderlyinv -0.298 0.138 0.770 0.321 0.821 0.088 employed1 3.264 0.368 0.309 0.160 0.295 0.018 kidp -0.774 0.153 0.373 0.244 0.404 0.052 light2_ea -1.590 0.550 0.034 0.081 0.035 0.006 nafemalesinv ×nelderlyinv 1.552 0.294 0.379 0.223 0.363 0.024 ownhouse1_ea 3.283 0.550 0.865 0.163 0.897 0.010 _intercept_ 8.053 0.592 obs. 132 R square 0.67 location effect 0.14 * dum_eaurable4_0=1- dum_eaurable4; nelderlyinv=1/(1+nelderly); nafemalesinv=1/(1+nafemales) 145 Table A1.17: Rural Amran Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 0.430 0.200 0.234 0.148 0.231 0.016 dum_eaurable10 0.530 0.148 0.014 0.117 0.038 0.037 headdivorced 0.788 0.357 0.006 0.076 0.005 0.005 headmarried -0.202 0.121 0.936 0.245 0.939 0.058 headsingle -0.499 0.232 0.025 0.157 0.014 0.014 light1_ea -0.470 0.093 0.183 0.355 0.194 0.127 light4_ea -1.303 0.340 0.026 0.077 0.026 0.005 light5 -0.441 0.066 0.403 0.490 0.372 0.235 singlep -0.864 0.118 0.607 0.187 0.590 0.046 work8 -0.638 0.271 0.059 0.094 0.053 0.009 work9 0.423 0.104 0.534 0.277 0.516 0.083 _intercept_ 11.080 0.126 obs. 224 R square 0.40 location effect 0.084 Table A1.18: Rural Al-Dhale Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10 0.178 0.096 0.093 0.290 0.096 0.087 dum_eaurable6_ea -4.379 1.066 0.017 0.053 0.013 0.001 headage 0.004 0.002 44.002 14.971 43.662 195.229 headuniv_ea 6.282 1.515 0.031 0.034 0.026 0.001 housetype1×sizeinv* 4.785 0.419 0.134 0.091 0.123 0.004 light5 -0.281 0.062 0.418 0.493 0.405 0.243 namales 0.068 0.020 2.023 1.612 1.900 1.720 sewage3_ea 0.436 0.152 0.246 0.297 0.292 0.119 universityp 3.725 0.875 0.008 0.046 0.008 0.001 water2 -0.994 0.182 0.045 0.208 0.031 0.030 water2_ea 1.257 0.259 0.044 0.134 0.044 0.018 work9 0.264 0.107 0.437 0.306 0.432 0.155 _intercept_ 9.209 0.156 obs. 156 R square 0.58 location effect - *sizeinv=1/(1+hhsize); No location effect 146 Table A1.19: Rural Remah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable15_ea -3.388 1.473 0.015 0.051 0.014 0.0004 dum_eaurable17_ea 4.059 1.290 0.010 0.029 0.011 0.001 dum_eaurable4 0.594 0.191 0.007 0.085 0.017 0.017 dum_eaurable7_ea -78.122 23.662 3E-5 3E-6 3E-4 3E-6 dum_eaurable8_ea -2.863 1.386 0.013 0.114 0.016 0.001 headdivorced 0.832 0.237 0.011 0.104 0.011 0.011 headiliter_ea -1.403 0.302 0.703 0.155 0.723 0.022 headread_ea -0.959 0.350 0.183 0.116 0.192 0.018 headsecond 0.134 0.085 0.130 0.337 0.108 0.097 headsingle 0.373 0.163 0.029 0.169 0.025 0.025 headuniv 0.431 0.241 0.021 0.144 0.011 0.010 highiliter 0.135 0.056 0.308 0.462 0.316 0.217 light4 0.409 0.167 0.026 0.160 0.025 0.024 light6 0.398 0.114 0.078 0.269 0.055 0.052 nkids -0.040 0.012 3.996 3.285 3.527 8.277 ownhouse1_ea -1.919 0.461 0.907 0.105 0.912 0.004 singlep -0.652 0.149 0.579 0.231 0.559 0.053 _intercept_ 14.116 0.523 obs. 270 R square 0.39 location effect 0.170 Table A1.20: Urban Ibb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.303 0.073 0.116 0.320 0.112 0.100 dum_eaurable10_ea 1.691 0.374 0.510 0.216 0.520 0.054 dum_eaurable11_ea -0.778 0.359 0.473 0.189 0.495 0.042 dum_eaurable18_ea -9.050 2.387 0.009 0.014 0.010 1E-4 dum_eaurable3_ea 2.369 1.373 0.027 0.023 0.027 4E-4 dum_eaurable5 0.394 0.213 0.011 0.104 0.012 0.011 dum_eaurable8 0.176 0.050 0.400 0.490 0.395 0.240 dum_eaurable8_ea -0.483 0.247 0.400 0.136 0.416 0.024 employ_nonself -0.459 0.121 0.519 0.206 0.471 0.043 headprim_ea -2.524 0.514 0.088 0.045 0.087 0.002 headsecond 0.153 0.055 0.240 0.427 0.213 0.168 headuniv 0.212 0.076 0.099 0.299 0.098 0.089 headuniv_ea -2.011 0.672 0.099 0.052 0.102 0.002 housetype2_ea -0.520 0.169 0.306 0.231 0.310 0.053 nafemales -0.044 0.017 1.931 1.494 2.096 2.165 namales -0.040 0.016 1.975 1.650 2.109 2.578 nelderly -0.102 0.041 0.271 0.579 0.268 0.314 nkids -0.062 0.013 3.130 2.391 3.186 4.185 ownhouse2_ea 0.581 0.236 0.370 0.141 0.363 0.016 singlep -0.810 0.146 0.593 0.227 0.610 0.034 work9 -2.495 0.430 0.038 0.065 0.038 0.004 147 _intercept_ 11.917 0.147 obs. 470 R square 0.42 location effect -* *No location effect Table A1.21: Urban Abyan Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 1.222 0.210 0.315 0.217 0.308 0.026 dum_eaurable1 0.566 0.098 0.073 0.260 0.084 0.077 dum_eaurable10_ea 0.791 0.300 0.677 0.175 0.641 0.053 dum_eaurable11_ea 2.629 0.561 0.386 0.181 0.377 0.040 dum_eaurable12_ea -2.654 0.471 0.789 0.126 0.773 0.026 dum_eaurable13_ea 0.850 0.347 0.504 0.167 0.502 0.038 dum_eaurable15_ea -2.893 0.683 0.157 0.097 0.162 0.012 dum_eaurable17_ea 2.390 0.394 0.513 0.173 0.491 0.042 dum_eaurable18_ea -1.621 0.360 0.142 0.141 0.141 0.020 dum_eaurable4_ea 12.151 3.001 0.010 0.014 0.011 0.000 dum_eaurable8_ea -1.857 0.431 0.382 0.151 0.374 0.029 headage 0.005 0.002 45.701 13.669 46.404 148.071 headiliter_ea -2.269 0.500 0.332 0.107 0.321 0.009 headuniv 0.347 0.114 0.071 0.257 0.059 0.056 nafemalesinv* 1.772 0.222 0.404 0.213 0.363 0.025 ownhouse1 -0.136 0.075 0.834 0.372 0.830 0.142 sewage3_ea 0.782 0.196 0.139 0.249 0.147 0.077 work4 1.112 0.352 0.108 0.134 0.107 0.014 _intercept_ 10.053 0.286 obs. 318 R square 0.42 location effect - *nafemalesinv=1/(1+nafemales); No location effect 148 Table A1.22: Urban Sana'a City Mean sd sd Variable Coefficient Std. Err. (census) (census) Mean (survey) (survey) dum_eaurable15_ea -0.466 0.185 0.200 0.088 0.211 0.009 dum_eaurable16_1 0.164 0.041 0.120 0.325 0.141 0.121 dum_eaurable16_ea 1.969 0.850 0.120 0.097 0.123 0.013 dum_eaurable16_ea^2 -9.740 3.210 0.024 0.040 0.028 0.004 dum_eaurable16_ea^3 11.224 3.201 0.006 0.020 0.010 0.002 dum_eaurable18_1 0.901 0.232 0.012 0.111 0.003 0.003 dum_eaurable1_ea^3 2.537 0.585 0.016 0.033 0.017 0.002 dum_eaurable4_1 0.194 0.087 0.028 0.165 0.021 0.021 dum_eaurable8_1 0.160 0.028 0.470 0.499 0.444 0.247 dum_eaurable9_ea 0.480 0.197 0.507 0.141 0.508 0.021 employ_nonself^2 0.647 0.159 0.474 0.243 0.489 0.062 employ_self 0.466 0.228 0.305 0.185 0.293 0.030 headage -0.003 0.001 40.369 13.570 43.251 165.235 headiliter_ea 0.727 0.269 0.267 0.107 0.261 0.009 headprim_1 0.106 0.042 0.095 0.293 0.110 0.098 headprim_ea^3 23.624 6.764 0.001 0.002 0.002 1E-4 headread_ea^2 3.043 0.734 0.030 0.032 0.031 0.001 headsecond_ea 4.552 0.991 0.292 0.062 0.291 0.004 headsecond_ea^3 -10.968 3.169 0.028 0.017 0.028 3E-4 headuniv_1 0.207 0.035 0.195 0.396 0.223 0.174 housetype1_1 -0.088 0.029 0.557 0.497 0.547 0.248 Marriedp^3 0.351 0.067 0.146 0.263 0.124 0.056 nelderly -0.102 0.028 0.238 0.609 0.256 0.275 nkids -0.223 0.025 2.601 2.373 2.825 4.804 nkids^2 0.022 0.005 12.393 42.780 12.653 402.126 nkids^3 -0.001 0.000 88.139 4189.858 75.438 74026.892 sewage1_ea^2 0.154 0.059 0.564 0.433 0.536 0.191 sewage3_ea 0.817 0.223 0.012 0.064 0.012 0.005 universityp^3 0.424 0.145 0.020 0.112 0.020 0.011 water1_1 0.136 0.048 0.604 0.489 0.589 0.242 water1_ea -0.481 0.133 0.590 0.408 0.610 0.162 water1_ea^3 0.251 0.128 0.467 0.394 0.489 0.154 water2_1 0.154 0.078 0.042 0.201 0.034 0.033 water4_ea -0.208 0.072 0.887 0.213 0.895 0.040 work1^2 -0.338 0.136 0.147 0.136 0.162 0.023 work7^2 6.502 3.262 0.006 0.018 0.006 4E-4 work7^3 -20.136 8.467 0.001 0.006 0.002 1E-4 _intercept_ 9.449 0.357 obs. 1639 R square 0.39 location effect 0.025 149 Table A1.23: Urban Al-Baida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.263 0.079 0.159 0.366 0.133 0.116 dum_eaurable12_ea 0.953 0.297 0.815 0.132 0.807 0.010 dum_eaurable13_ea -0.652 0.222 0.641 0.143 0.612 0.017 dum_eaurable18_ea -5.562 1.835 0.014 0.019 0.011 3E-4 dum_eaurable5_ea 16.432 3.039 0.016 0.015 0.015 1E-4 dum_eaurable8 0.118 0.057 0.455 0.498 0.439 0.247 employed2 1.623 0.405 0.322 0.092 0.319 0.005 headdivorced 0.471 0.206 0.013 0.112 0.017 0.017 kidp -0.780 0.136 0.388 0.224 0.391 0.047 marriedp 0.776 0.139 0.363 0.214 0.378 0.046 nafemales -0.115 0.020 2.017 1.540 2.130 2.027 university_ea -6.277 2.963 0.015 0.009 0.016 1E-4 water1_ea 0.197 0.067 0.509 0.410 0.499 0.179 work10 2.141 0.669 0.039 0.047 0.030 0.002 work9 0.348 0.139 0.159 0.202 0.162 0.054 _intercept_ 9.767 0.339 obs. 327 R square 0.42 location effect -* *No location effect 150 Table A1.24: Urban Taiz Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12_ea^3 -1.5708 0.2795 0.539 0.187 0.546 0.043 dum_eaurable18_ea^2 403.0795 62.643 0.003 0.013 0.002 3E-5 dum_eaurable18_ea^3 -1636.106 262.7575 0.001 0.005 2E-4 1E-6 dum_eaurable1 0.2357 0.0808 0.101 0.301 0.092 0.084 dum_eaurable11_ea 1.2389 0.4416 0.481 0.195 0.495 0.043 dum_eaurable14_ea -0.9563 0.4939 0.134 0.101 0.140 0.010 dum_eaurable16 0.4862 0.0897 0.067 0.250 0.081 0.075 dum_eaurable18 0.5874 0.2132 0.030 0.171 0.012 0.012 dum_eaurable18_ea -23.8544 3.9038 0.030 0.045 0.028 0.001 dum_eaurable8 0.1328 0.05 0.396 0.489 0.400 0.240 dum_eaurable8_ea 2.4542 0.5339 0.396 0.145 0.390 0.019 employ_self^3 -1.0469 0.2871 0.074 0.128 0.079 0.027 headread_ea 1.9674 0.4284 0.143 0.067 0.151 0.006 headsecond_ea^3 -3.7498 2.1384 0.024 0.018 0.025 2E-4 headuniv_ea^3 25.8864 6.0094 0.005 0.007 0.006 1E-4 housetype3_ea 2.006 0.4704 0.017 0.059 0.019 0.006 kidp^2 0.6678 0.9237 0.168 0.170 0.181 0.030 kidp^2 0.4222 1.1325 0.095 0.120 0.102 0.015 nelderly -0.1566 0.0474 0.279 0.572 0.256 0.281 nkids -0.1607 0.0212 2.500 2.249 2.457 4.240 ownhouse1_ea 2.6185 0.7354 0.517 0.500 0.503 0.024 ownhouse2_ea 2.7389 0.7017 0.450 0.498 0.454 0.028 sewage2_ea 1.1529 0.448 0.227 0.315 0.229 0.104 sewage2_ea^2 -2.1528 0.5258 0.227 0.315 0.157 0.067 singlep^2 -3.3888 0.6401 0.394 0.244 0.396 0.044 singlep^2 3.0947 0.6946 0.285 0.236 0.278 0.037 university_ea -13.292 2.6255 0.050 0.026 0.052 0.001 work4 -1.4352 0.3154 0.232 0.172 0.233 0.029 work4^3 2.4167 0.6503 0.037 0.067 0.036 0.005 work8^3 3.0032 1.2136 0.001 0.004 0.006 0.001 _intercept_ 9.2042 0.6649 obs. 582 R square 0.43 location effect -* *No location effect 151 Table A1.25: Urban Al-Jawf Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 1.902 0.300 0.261 0.155 0.255 0.014 dum_eaurable10 0.340 0.092 0.072 0.259 0.104 0.094 dum_eaurable16 -0.491 0.247 0.005 0.067 0.009 0.009 dum_eaurable18 0.410 0.144 0.028 0.164 0.041 0.039 dum_eaurable1_ea 2.009 0.290 0.156 0.132 0.143 0.020 dum_eaurable6 0.223 0.095 0.062 0.241 0.078 0.072 dum_eaurable9_ea -1.103 0.213 0.270 0.176 0.256 0.043 headuniv_ea 6.030 1.357 0.024 0.028 0.023 4E-4 highiliter -0.137 0.068 0.197 0.398 0.180 0.148 housetype1_ea 0.584 0.214 0.921 0.104 0.912 0.017 light5_ea 0.178 0.096 0.249 0.284 0.249 0.100 marriedp 0.269 0.146 0.311 0.145 0.301 0.033 nafemalesinv 2.074 0.258 0.391 0.161 0.365 0.018 sewage3_ea -0.318 0.092 0.287 0.289 0.315 0.089 _intercept_ 8.362 0.280 obs. 226 R square 0.47 location effect 0.034 Table A1.26: Urban Hajja Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep -0.700 0.186 0.258 0.165 0.275 0.030 dum_eaurable1 0.263 0.085 0.090 0.287 0.116 0.103 dum_eaurable11_ea -0.734 0.319 0.305 0.231 0.303 0.053 dum_eaurable17_ea 0.904 0.304 0.300 0.216 0.296 0.050 dum_eaurable1_ea 1.690 0.504 0.090 0.072 0.099 0.006 headiliter -0.259 0.063 0.523 0.499 0.494 0.251 headmarried -0.416 0.111 0.879 0.326 0.895 0.094 headprim -0.362 0.126 0.076 0.265 0.081 0.075 highprim 0.293 0.125 0.075 0.263 0.078 0.072 housetype1_ea -1.515 0.255 0.689 0.247 0.714 0.056 housetype3_ea -1.131 0.256 0.175 0.249 0.162 0.065 kidp -0.980 0.152 0.377 0.248 0.379 0.064 marriedp 0.772 0.149 0.352 0.229 0.344 0.053 nelderly -0.135 0.051 0.346 0.630 0.321 0.336 ownhouse1_ea 0.323 0.168 0.728 0.182 0.760 0.032 sewage2_ea 0.340 0.112 0.350 0.322 0.359 0.114 water1_ea -0.569 0.130 0.389 0.441 0.361 0.185 water3 0.829 0.190 0.046 0.209 0.035 0.034 _intercept_ 12.577 0.266 obs. 339 R square 0.42 location effect 0.037 152 Table A1.27: Urban Al-Hodeida Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 0.490 0.099 0.275 0.179 0.319 0.049 cook2 -0.185 0.039 0.253 0.435 0.263 0.194 cook2_ea 0.399 0.100 0.246 0.213 0.252 0.051 dum_eaurable1 0.332 0.082 0.049 0.216 0.045 0.043 dum_eaurable2_ea 6.543 2.113 0.006 0.009 0.006 1E-4 dum_eaurable3_ea -1.668 0.771 0.015 0.021 0.015 0.001 dum_eaurable8 0.160 0.039 0.255 0.436 0.282 0.203 headiliter -0.131 0.034 0.485 0.500 0.476 0.250 headmarried -0.119 0.049 0.837 0.370 0.844 0.132 headprim_ea -1.149 0.321 0.098 0.050 0.086 0.002 housetype2 0.129 0.053 0.094 0.291 0.126 0.110 housetype2_ea 0.522 0.148 0.094 0.154 0.104 0.033 kidp -0.325 0.107 0.326 0.250 0.325 0.063 light2 -0.724 0.162 0.027 0.161 0.011 0.011 nafemales -0.109 0.014 2.032 4.080 1.907 2.008 ownhouse2_ea -0.710 0.189 0.186 0.149 0.180 0.025 singlep -0.614 0.094 0.567 0.253 0.585 0.055 universityp 0.723 0.167 0.030 0.106 0.030 0.010 university_ea 5.493 0.997 0.023 0.022 0.025 0.001 work9 -0.831 0.190 0.034 0.086 0.035 0.007 _intercept_ 11.453 0.095 obs. 841 R square 0.51 location effect 0.035 Table A1.28: Urban Hadramout Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) amalep 0.388 0.137 0.313 0.207 0.295 0.029 dum_eaurable16 0.170 0.093 0.060 0.237 0.049 0.046 dum_eaurable2 -0.177 0.110 0.019 0.138 0.025 0.025 dum_eaurable3_ea -3.081 1.317 0.016 0.029 0.013 2E-4 dum_eaurable6_ea -0.914 0.096 0.164 0.209 0.182 0.048 employed2 0.896 0.331 0.325 0.080 0.329 0.005 employ_self 0.358 0.125 0.434 0.220 0.446 0.040 headiliter -0.141 0.044 0.279 0.448 0.306 0.213 headiliter_ea -0.665 0.205 0.279 0.123 0.274 0.018 headuniv_ea 3.193 0.643 0.067 0.056 0.066 0.004 highprim 0.240 0.105 0.035 0.185 0.032 0.031 kidp -0.380 0.120 0.345 0.225 0.334 0.054 light1 0.377 0.210 0.930 0.255 0.936 0.060 light5 0.877 0.247 0.019 0.136 0.013 0.013 singlep -0.345 0.105 0.557 0.214 0.529 0.054 universityp 0.529 0.245 0.028 0.092 0.027 0.007 university_ea -8.246 2.158 0.024 0.017 0.025 4E-4 153 water2 0.581 0.219 0.053 0.224 0.056 0.053 _intercept_ 10.456 0.253 obs. 463 R square 0.41 location effect 0.14 Table A1.29: Urban Dhamar Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable12_ea -0.965 0.222 0.785 0.169 0.795 0.031 dum_eaurable17_ea 1.426 0.340 0.286 0.142 0.291 0.017 dum_eaurable1_ea 3.576 0.511 0.115 0.067 0.108 0.003 dum_eaurable3_ea -5.906 1.338 0.023 0.022 0.023 0.001 dum_eaurable6_ea 6.048 1.655 0.020 0.023 0.018 3E-4 dum_eaurable8 0.196 0.058 0.338 0.473 0.337 0.224 employed2 1.812 0.442 0.300 0.068 0.285 0.005 headprim_ea -3.383 0.703 0.080 0.052 0.079 0.002 headsingle 0.385 0.122 0.054 0.226 0.047 0.045 light6 -0.442 0.117 0.055 0.227 0.049 0.047 nkids -0.047 0.017 3.285 2.498 3.281 5.204 singlep -0.524 0.147 0.589 0.223 0.579 0.049 hhsize -0.028 0.010 7.454 4.323 7.534 12.646 work9 1.047 0.271 0.071 0.109 0.077 0.018 _intercept_ 10.914 0.184 obs. 342 R square 0.43 location effect 0.047 Table A1.30: Urban Shabwah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.228 0.081 0.218 0.413 0.216 0.170 dum_eaurable16 0.304 0.164 0.040 0.196 0.038 0.037 dum_eaurable17_ea -1.047 0.246 0.561 0.216 0.576 0.028 employed2 -1.422 0.423 0.297 0.104 0.325 0.014 headiliter -0.212 0.075 0.293 0.455 0.297 0.210 headprim_ea 1.544 0.793 0.067 0.039 0.063 0.002 headsecond 0.184 0.074 0.300 0.458 0.342 0.226 housetype1_ea -1.497 0.184 0.701 0.323 0.666 0.119 nkids -0.031 0.014 4.011 3.554 4.126 6.894 sewage1_ea 0.621 0.137 0.231 0.334 0.250 0.120 singlep -0.547 0.192 0.584 0.223 0.606 0.031 water3 -0.492 0.222 0.062 0.241 0.056 0.053 work5 1.520 0.582 0.034 0.073 0.034 0.006 _intercept_ 12.842 0.324 obs. 198 R square 0.54 location effect -* *No location effect 154 Table A1.31: Urban Sa'adah Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable17_ea^3 4.757 1.339 0.020 0.032 0.021 0.001 dum_eaurable1 0.339 0.065 0.202 0.402 0.218 0.171 dum_eaurable10 0.227 0.066 0.194 0.395 0.209 0.166 dum_eaurable3 0.293 0.164 0.017 0.130 0.019 0.019 dum_eaurable3_ea^3 -2716.399 646.641 3E-5 6E-5 0.003 0.001 dum_eaurable8 0.084 0.055 0.338 0.473 0.331 0.222 employ_nonself^3 -1.855 0.263 0.169 0.211 0.180 0.052 headprim_ea -0.941 0.501 0.095 0.049 0.093 0.003 light1 0.343 0.072 0.726 0.446 0.725 0.200 light1_ea^3 -0.363 0.149 0.538 0.331 0.538 0.113 light5_1 0.347 0.143 0.056 0.229 0.033 0.032 nafemales -0.056 0.017 2.071 1.801 2.275 2.270 nkidsinv 0.328 0.111 0.365 0.311 0.305 0.064 primary_ea 2.257 0.718 0.189 0.072 0.189 0.004 sewage3 0.086 0.062 0.172 0.377 0.189 0.154 sewage3_ea^3 0.877 0.613 0.040 0.118 0.024 0.004 singlep^3 -0.640 0.165 0.258 0.195 0.244 0.025 work4^3 1.225 0.707 0.025 0.050 0.034 0.003 work5 -18.395 3.224 0.032 0.054 0.034 0.003 work5^2 222.834 53.169 0.004 0.011 0.004 1E-4 work5^3 -752.706 206.089 0.001 0.003 0.001 7E-6 _intercept_ 10.792 0.156 obs. 324 R square 0.36 location effect -* *No location effect Table A1.32: Urban Aden Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.142 0.053 0.165 0.371 0.144 0.123 dum_eaurable2_ea -1.535 0.936 0.014 0.022 0.015 4E-4 dum_eaurable3 0.337 0.115 0.014 0.116 0.024 0.023 dum_eaurable6 0.966 0.453 0.001 0.033 0.001 0.001 dum_eaurable9 0.231 0.037 0.412 0.492 0.411 0.243 femalehead 0.116 0.047 0.164 0.371 0.181 0.148 kidp -0.310 0.146 0.298 0.236 0.288 0.057 nafemales -0.104 0.016 2.113 2.790 1.984 1.844 nkids -0.027 0.016 2.290 2.263 2.148 4.713 ownhouse1_ea -2.143 0.791 0.789 0.114 0.797 0.011 ownhouse2 0.130 0.050 0.183 0.387 0.157 0.133 ownhouse2_ea -2.463 0.819 0.180 0.102 0.182 0.010 singlep -0.525 0.087 0.548 0.238 0.555 0.051 universityp 0.663 0.118 0.069 0.154 0.082 0.025 water2 0.476 0.193 0.004 0.063 0.005 0.005 work3 0.804 0.182 0.087 0.091 0.098 0.010 155 work6 -1.291 0.318 0.023 0.047 0.023 0.003 _intercept_ 13.450 0.777 obs. 716 R square 0.41 location effect 0.12 Table A1.33: Urban Laheg Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) afemalep 0.988 0.239 0.283 0.178 0.312 0.030 dum_eaurable2_ea -30.626 3.611 0.009 0.011 0.007 1E-4 headage 0.005 0.002 44.828 13.877 44.852 174.849 headiliter_ea 0.879 0.435 0.311 0.116 0.303 0.017 headprim_ea 5.923 1.118 0.062 0.046 0.065 0.001 highprim 0.308 0.135 0.030 0.172 0.043 0.041 light1 0.511 0.117 0.824 0.381 0.791 0.166 light2_ea 0.499 0.213 0.064 0.225 0.093 0.078 nafemales -0.156 0.034 1.872 1.518 1.915 1.684 ownhouse1 0.224 0.076 0.833 0.373 0.842 0.134 primaryp 0.753 0.142 0.331 0.274 0.344 0.071 singlep -0.435 0.168 0.567 0.236 0.589 0.055 university_ea 6.651 1.386 0.039 0.024 0.047 0.001 _intercept_ 9.327 0.257 obs. 273 R square 0.46 location effect 0.079 Table A1.34: Urban Mareb Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea -0.822 0.275 0.442 0.144 0.450 0.028 dum_eaurable16 0.778 0.210 0.037 0.188 0.033 0.032 dum_eaurable18 0.220 0.105 0.141 0.348 0.140 0.121 headsecond×sizeinv* 1.305 0.495 0.050 0.099 0.046 0.007 headsecond_ea 1.535 0.433 0.289 0.088 0.276 0.010 light1_ea 3.653 0.742 0.895 0.062 0.898 0.004 light 0.938 0.399 0.006 0.075 0.010 0.010 ownhouse1 0.242 0.080 0.465 0.292 0.521 0.251 sizeinv 4.151 0.516 0.157 0.106 0.135 0.006 water4_ea -0.642 0.248 0.858 0.192 0.871 0.032 work1 -0.521 0.213 0.350 0.191 0.390 0.041 _intercept_ 7.510 0.581 obs. 224 R square 0.43 location effect - *sizeinv=1/(1+sizeinv); No location effect 156 Table A1.35: Urban Al-Mahweet Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 -0.213 0.091 0.100 0.300 0.076 0.071 dum_eaurable10_ea 0.749 0.316 0.324 0.119 0.327 0.018 dum_eaurable14_ea 1.430 0.414 0.175 0.108 0.189 0.013 dum_eaurable1_ea -2.258 0.986 0.100 0.039 0.104 0.001 dum_eaurable3_ea -31.660 8.094 0.003 0.005 0.003 2E-5 dum_eaurable8 0.114 0.050 0.482 0.500 0.453 0.249 headage -0.004 0.002 42.882 15.373 45.183 226.657 headprim_ea -1.838 0.634 0.048 0.044 0.049 0.002 highread -0.119 0.069 0.168 0.374 0.163 0.137 nkidsinv* 0.553 0.120 0.387 0.319 0.343 0.075 nkidsinv×water3_ea 7.961 1.930 0.006 0.031 0.004 2E-4 ownhouse2_ea 0.620 0.275 0.308 0.097 0.290 0.010 singlep -0.630 0.153 0.589 0.234 0.599 0.045 water1_ea -0.484 0.095 0.697 0.353 0.685 0.131 work1 -0.763 0.161 0.518 0.241 0.514 0.064 _intercept_ 11.361 0.224 obs. 289 R square 0.38 location effect - *nkidsinv=1/(1+nkids); No location effect Table A1.36: Urban Al-Maharh Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable10_ea 0.779 0.147 0.525 0.254 0.563 0.072 dum_eaurable15_ea -1.176 0.306 0.238 0.173 0.265 0.023 dum_eaurable16 1.287 0.349 0.011 0.105 0.007 0.007 employ_nonself -1.990 0.435 0.459 0.251 0.443 0.056 headsecond_ea 1.810 0.533 0.239 0.133 0.250 0.018 headuniv_ea 12.504 2.026 0.056 0.064 0.058 0.003 marriedp 0.560 0.183 0.387 0.242 0.382 0.039 nafemales -0.068 0.025 1.954 1.618 1.997 1.665 nkids -0.045 0.016 3.292 3.025 3.388 5.077 university_ea -43.844 5.232 0.016 0.020 0.016 2E-4 _intercept_ 11.548 0.189 obs. 137 R square 0.67 location effect -* *No location effect 157 Table A1.37: Urban Amran Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable1 0.307 0.070 0.156 0.363 0.167 0.140 dum_eaurable13_ea -0.514 0.165 0.734 0.169 0.733 0.027 dum_eaurable14_ea 2.028 0.338 0.193 0.134 0.198 0.021 dum_eaurable17_ea -0.741 0.385 0.179 0.113 0.176 0.008 dum_eaurable3_ea -5.083 1.968 0.021 0.020 0.020 4E-4 dum_eaurable8 0.216 0.055 0.377 0.485 0.404 0.242 headprim_ea 1.024 0.586 0.083 0.040 0.088 0.002 highread -0.136 0.070 0.212 0.409 0.187 0.153 housetype1_ea 0.607 0.267 0.788 0.184 0.803 0.021 nelderlyinv 0.294 0.106 0.847 0.249 0.848 0.060 nkidsinv*×singlep 1.230 0.331 0.169 0.166 0.139 0.006 nkidsinv×work10 3.590 1.402 0.019 0.035 0.016 0.001 ownhouse1_ea -1.034 0.274 0.659 0.142 0.667 0.018 singlep -0.692 0.158 0.587 0.214 0.598 0.034 water4_ea 0.574 0.136 0.926 0.164 0.905 0.039 work5 -1.166 0.656 0.021 0.050 0.023 0.002 _intercept_ 10.124 0.262 obs. 302 R square 0.38 location effect 0.080 *nkidsinv=1/(1+nkids) Table A1.38: Urban Al-Dhale Variable Coefficient Std. Err. Mean (census) sd (census) Mean (survey) sd (survey) dum_eaurable17_ea 0.598 0.158 0.374 0.215 0.383 0.057 dum_eaurable4 0.521 0.195 0.016 0.126 0.025 0.024 dum_eaurable5 0.670 0.192 0.015 0.123 0.021 0.021 headdivorced -0.516 0.231 0.010 0.099 0.016 0.016 headread_ea -0.717 0.314 0.273 0.103 0.268 0.011 headwidow -0.395 0.142 0.046 0.210 0.051 0.048 light4 0.551 0.251 0.020 0.140 0.014 0.014 nafemales -0.152 0.025 1.727 1.531 1.874 1.697 nkids -0.065 0.016 3.238 2.861 3.459 5.673 primary_ea 1.266 0.666 0.247 0.077 0.240 0.003 singlep -0.407 0.170 0.583 0.244 0.600 0.054 water2 0.367 0.142 0.050 0.218 0.072 0.067 _intercept_ 11.094 0.249 obs. 233 R square 0.42 location effect 0.12 158 Table A2: Compare estimates of headcount (food poverty) using different data sources Governorate #hhno* avg_FGT0 se_FGT0 FGT0 (survey) Rural 11 233,491 0.603 0.016 0.625 12 38,120 0.761 0.020 0.725 14 49,667 0.774 0.025 0.805 15 266,914 0.642 0.019 0.660 16 45,325 0.801 0.033 0.852 17 154,183 0.662 0.027 0.672 18 224,491 0.547 0.022 0.529 19 58,818 0.606 0.025 0.614 20 150,379 0.505 0.031 0.504 21 34,657 0.781 0.030 0.792 22 61,965 0.449 0.037 0.407 23 99,571 0.667 0.025 0.657 25 87,265 0.651 0.031 0.662 26 21,181 0.586 0.032 0.639 27 54,322 0.573 0.031 0.608 28 5,903 0.221 0.029 0.233 29 73,020 0.917 0.026 0.899 30 48,051 0.795 0.023 0.810 31 45,963 0.553 0.029 0.564 Urban 11 49,143 0.468 0.014 0.500 12 13,473 0.610 0.018 0.644 13 236,515 0.465 0.012 0.477 14 12,786 0.495 0.027 0.447 15 77,660 0.513 0.015 0.511 16 7,064 0.770 0.018 0.741 17 16,137 0.471 0.022 0.504 18 104,909 0.502 0.015 0.490 19 53,868 0.583 0.036 0.641 20 24,312 0.602 0.021 0.607 21 7,551 0.640 0.026 0.631 22 11,549 0.469 0.018 0.447 24 82,967 0.500 0.021 0.509 25 8,495 0.510 0.032 0.513 26 3,325 0.474 0.039 0.465 27 4,169 0.684 0.029 0.682 28 4,074 0.274 0.020 0.313 29 17,135 0.722 0.022 0.705 30 7,717 0.606 0.040 0.602 * The number of households in this table is different from the number listed in Table 1 because of two reasons: 1. The houses which are not occupied and/or do not have a household head are dropped (see note 3); 2. The households which have missing values in the variables used in the consumption model are dropped. 159 Table A3: Estimates of poverty indicators (Urban areas) Governorate District #hhno avg_ ^ y avg_FGT0 se_FGT0 avg_FGT1 se_FGT1 avg_FGT2 se_FGT2 11 1 32,482 14,541 0.741 0.015 0.284 0.010 0.135 0.007 11 2 40,746 15,868 0.594 0.022 0.195 0.012 0.084 0.007 11 3 40,828 8,163 0.585 0.033 0.196 0.016 0.086 0.009 11 4 54,042 7,522 0.417 0.031 0.124 0.012 0.051 0.006 11 5 61,780 4,061 0.355 0.041 0.104 0.016 0.042 0.008 11 6 44,874 9,971 0.501 0.034 0.153 0.013 0.063 0.006 11 7 37,723 14,423 0.613 0.026 0.215 0.012 0.098 0.007 11 8 32,894 13,985 0.704 0.021 0.245 0.012 0.108 0.007 11 9 44,539 12,050 0.554 0.019 0.182 0.011 0.078 0.006 11 10 36,930 14,598 0.655 0.024 0.221 0.013 0.096 0.008 11 11 34,975 19,768 0.678 0.023 0.234 0.011 0.103 0.006 11 12 42,468 12,551 0.547 0.028 0.172 0.013 0.072 0.007 11 13 46,288 13,153 0.506 0.025 0.160 0.013 0.068 0.007 11 14 39,295 8,929 0.622 0.019 0.235 0.012 0.111 0.008 11 15 43,351 13,775 0.556 0.026 0.179 0.011 0.076 0.006 11 16 37,134 15,231 0.650 0.027 0.231 0.014 0.105 0.008 11 17 37,931 10,972 0.625 0.025 0.225 0.013 0.103 0.007 11 18 39,126 1,878 0.613 0.044 0.171 0.019 0.064 0.009 11 19 39,911 3,820 0.583 0.038 0.184 0.015 0.078 0.008 11 20 35,905 18,232 0.682 0.020 0.242 0.011 0.108 0.007 12 1 28,524 2,649 0.876 0.024 0.428 0.023 0.251 0.020 12 2 30,594 2,986 0.783 0.030 0.389 0.018 0.244 0.014 12 3 73,343 1,351 0.496 0.064 0.191 0.026 0.096 0.017 12 4 28,975 8,473 0.847 0.014 0.447 0.021 0.277 0.020 12 5 30,414 1,827 0.803 0.023 0.432 0.027 0.271 0.027 12 6 58,929 5,178 0.544 0.046 0.223 0.020 0.122 0.013 12 7 42,534 2,049 0.704 0.035 0.307 0.025 0.167 0.017 12 8 30,285 2,861 0.838 0.037 0.372 0.028 0.205 0.020 12 9 58,224 2,583 0.800 0.035 0.398 0.030 0.237 0.024 12 10 40,151 838 0.737 0.021 0.410 0.023 0.263 0.021 12 11 35,031 7,325 0.839 0.020 0.421 0.023 0.252 0.021 14 1 61,145 791 0.566 0.029 0.278 0.021 0.156 0.019 14 2 29,633 1,500 0.876 0.031 0.350 0.029 0.170 0.021 14 3 36,214 657 0.700 0.050 0.279 0.026 0.133 0.020 14 4 51,956 4,523 0.573 0.040 0.219 0.019 0.107 0.012 14 5 33,806 2,351 0.809 0.035 0.276 0.022 0.120 0.014 14 6 37,888 2,262 0.804 0.026 0.337 0.024 0.176 0.020 14 7 23,352 2,728 0.937 0.018 0.459 0.022 0.273 0.021 14 8 36,304 4,656 0.717 0.037 0.331 0.018 0.193 0.013 14 10 36,917 4,728 0.690 0.033 0.279 0.015 0.153 0.010 14 11 27,755 2,341 0.874 0.014 0.437 0.021 0.249 0.020 14 12 28,446 1,924 0.866 0.032 0.391 0.026 0.214 0.023 14 13 34,208 538 0.796 0.064 0.217 0.034 0.077 0.018 14 14 31,879 3,209 0.766 0.033 0.373 0.018 0.220 0.019 160 14 15 23,178 2,218 0.922 0.024 0.470 0.030 0.283 0.031 14 16 30,614 3,763 0.790 0.033 0.371 0.021 0.223 0.015 14 17 28,492 3,089 0.840 0.024 0.391 0.023 0.223 0.019 14 18 43,400 2,203 0.598 0.055 0.225 0.022 0.121 0.011 14 19 27,469 3,354 0.843 0.033 0.410 0.018 0.246 0.018 14 20 31,422 2,832 0.811 0.046 0.317 0.023 0.158 0.014 15 1 41,991 19,220 0.596 0.033 0.231 0.017 0.117 0.010 15 2 44,727 15,875 0.590 0.026 0.232 0.011 0.119 0.007 15 3 37,678 20,313 0.672 0.024 0.283 0.014 0.151 0.009 15 4 38,549 26,858 0.669 0.017 0.290 0.012 0.158 0.009 15 5 47,553 8,747 0.575 0.037 0.234 0.020 0.122 0.013 15 6 30,743 1,920 0.745 0.044 0.333 0.029 0.185 0.020 15 7 39,252 4,844 0.628 0.041 0.255 0.020 0.134 0.012 15 8 39,546 16,356 0.646 0.026 0.259 0.014 0.134 0.009 15 9 61,231 3,401 0.352 0.052 0.098 0.020 0.038 0.010 15 10 40,830 14,674 0.639 0.024 0.264 0.015 0.139 0.011 15 11 39,207 12,472 0.652 0.034 0.268 0.022 0.140 0.014 15 12 38,745 13,286 0.643 0.022 0.269 0.014 0.143 0.010 15 13 39,304 6,950 0.632 0.031 0.258 0.017 0.135 0.011 15 14 36,645 21,870 0.670 0.026 0.277 0.017 0.146 0.012 15 15 33,263 4,163 0.705 0.032 0.295 0.022 0.156 0.015 15 16 44,516 10,885 0.597 0.026 0.252 0.014 0.136 0.009 15 20 38,227 28,595 0.656 0.024 0.271 0.017 0.143 0.013 15 21 35,903 16,083 0.690 0.018 0.294 0.013 0.157 0.010 15 22 40,000 15,438 0.631 0.026 0.258 0.016 0.135 0.011 15 23 33,519 4,964 0.696 0.030 0.303 0.019 0.164 0.013 16 1 34,905 9,328 0.799 0.040 0.296 0.024 0.134 0.016 16 2 28,803 2,440 0.935 0.031 0.389 0.036 0.191 0.028 16 3 28,181 3,293 0.922 0.036 0.400 0.034 0.202 0.028 16 4 43,633 2,805 0.718 0.039 0.323 0.023 0.171 0.021 16 5 32,114 1,980 0.818 0.046 0.360 0.023 0.191 0.016 16 6 42,510 2,759 0.738 0.073 0.268 0.033 0.120 0.020 16 7 30,909 1,064 0.943 0.032 0.350 0.042 0.152 0.028 16 8 25,715 469 0.973 0.022 0.453 0.038 0.248 0.033 16 9 34,990 873 0.963 0.038 0.318 0.046 0.121 0.028 16 10 38,836 6,542 0.700 0.044 0.286 0.018 0.140 0.011 16 11 35,216 7,097 0.735 0.035 0.291 0.019 0.143 0.012 16 12 31,023 6,675 0.863 0.038 0.330 0.025 0.153 0.017 17 1 39,326 2,926 0.682 0.061 0.262 0.036 0.128 0.023 17 2 41,227 9,557 0.648 0.046 0.259 0.033 0.133 0.024 17 3 38,270 1,544 0.686 0.058 0.278 0.036 0.145 0.025 17 4 45,576 14,088 0.600 0.037 0.235 0.021 0.120 0.014 17 5 52,984 1,949 0.513 0.060 0.187 0.030 0.091 0.019 17 6 41,679 5,883 0.629 0.050 0.230 0.029 0.109 0.018 17 7 34,695 8,327 0.750 0.030 0.336 0.030 0.186 0.025 17 8 30,693 3,641 0.808 0.048 0.355 0.045 0.192 0.033 17 9 43,172 2,889 0.636 0.056 0.263 0.038 0.140 0.026 161 17 10 31,005 4,789 0.743 0.043 0.350 0.035 0.204 0.028 17 11 40,951 8,315 0.660 0.039 0.267 0.022 0.139 0.015 17 12 47,042 7,542 0.577 0.052 0.202 0.027 0.094 0.016 17 13 35,687 5,454 0.738 0.041 0.321 0.037 0.175 0.028 17 14 35,887 3,641 0.704 0.036 0.294 0.027 0.155 0.020 17 15 49,375 3,354 0.562 0.097 0.235 0.062 0.127 0.042 17 16 51,093 2,761 0.607 0.089 0.272 0.059 0.155 0.043 17 17 35,106 5,671 0.678 0.046 0.270 0.026 0.139 0.016 17 18 43,562 3,662 0.624 0.065 0.247 0.041 0.127 0.028 17 19 37,049 1,767 0.706 0.042 0.275 0.026 0.139 0.018 17 20 40,206 6,093 0.680 0.095 0.276 0.070 0.142 0.049 17 21 45,098 3,215 0.620 0.126 0.251 0.080 0.131 0.054 17 22 44,144 9,301 0.626 0.040 0.231 0.022 0.111 0.014 17 23 43,442 1,174 0.596 0.075 0.217 0.037 0.104 0.021 17 24 45,393 8,112 0.598 0.055 0.216 0.029 0.103 0.017 17 25 47,483 6,049 0.557 0.049 0.187 0.026 0.084 0.015 17 26 49,371 3,319 0.533 0.055 0.198 0.027 0.100 0.017 17 27 33,715 4,704 0.742 0.041 0.319 0.034 0.172 0.025 17 28 35,358 1,931 0.735 0.131 0.328 0.106 0.183 0.079 17 29 36,130 3,381 0.724 0.037 0.304 0.031 0.162 0.024 17 30 36,963 6,356 0.696 0.038 0.294 0.023 0.157 0.016 17 31 24,550 2,788 0.887 0.033 0.448 0.036 0.269 0.030 18 1 51,990 20,361 0.462 0.029 0.167 0.013 0.083 0.008 18 2 44,641 14,985 0.545 0.027 0.217 0.014 0.117 0.009 18 4 7,959 681 0.995 0.007 0.738 0.043 0.564 0.055 18 5 153,330 4,468 0.592 0.037 0.280 0.022 0.168 0.017 18 6 121,284 10,499 0.559 0.058 0.229 0.035 0.121 0.023 18 7 37,040 11,608 0.619 0.028 0.287 0.014 0.172 0.010 18 8 150,193 5,316 0.370 0.035 0.117 0.015 0.052 0.009 18 9 40,051 5,073 0.605 0.030 0.236 0.019 0.122 0.013 18 10 42,159 19,422 0.573 0.023 0.238 0.016 0.129 0.012 18 11 75,866 1,328 0.270 0.050 0.085 0.022 0.038 0.013 18 12 48,406 6,577 0.526 0.045 0.218 0.031 0.120 0.022 18 13 48,054 15,704 0.488 0.032 0.176 0.015 0.088 0.009 18 14 73,799 8,510 0.468 0.029 0.194 0.018 0.106 0.013 18 15 52,827 9,965 0.444 0.037 0.143 0.018 0.064 0.011 18 16 42,962 5,594 0.549 0.028 0.214 0.016 0.111 0.011 18 17 44,807 29,568 0.588 0.027 0.239 0.016 0.128 0.011 18 18 44,004 7,869 0.556 0.054 0.206 0.034 0.103 0.023 18 19 39,711 4,683 0.597 0.031 0.237 0.020 0.124 0.014 18 20 46,247 3,014 0.631 0.040 0.302 0.028 0.182 0.022 18 21 62,487 444 0.339 0.179 0.094 0.067 0.037 0.031 18 23 31,945 461 0.751 0.069 0.307 0.052 0.157 0.037 18 24 66,611 17,998 0.628 0.024 0.291 0.015 0.171 0.011 18 25 42,584 11,604 0.538 0.039 0.205 0.019 0.106 0.011 18 26 46,244 8,759 0.594 0.036 0.248 0.021 0.135 0.014 19 1 40,477 538 0.550 0.066 0.156 0.028 0.066 0.019 162 19 2 22,365 225 0.911 0.040 0.430 0.034 0.237 0.029 19 3 49,960 271 0.445 0.103 0.076 0.027 0.019 0.009 19 4 65,888 198 0.150 0.046 0.034 0.010 0.012 0.005 19 5 58,554 300 0.258 0.134 0.035 0.024 0.008 0.006 19 6 61,149 302 0.121 0.081 0.018 0.014 0.004 0.004 19 7 33,001 4,672 0.793 0.019 0.320 0.019 0.156 0.013 19 8 38,142 3,827 0.731 0.028 0.253 0.020 0.111 0.012 19 9 40,125 1,616 0.621 0.050 0.188 0.016 0.082 0.008 19 10 36,842 4,793 0.739 0.031 0.257 0.018 0.115 0.011 19 11 36,091 4,060 0.741 0.031 0.253 0.017 0.114 0.010 19 12 36,967 1,116 0.637 0.056 0.208 0.019 0.094 0.009 19 13 44,068 3,769 0.588 0.034 0.185 0.017 0.078 0.011 19 14 38,074 1,276 0.759 0.038 0.223 0.028 0.086 0.017 19 15 44,229 1,734 0.647 0.052 0.204 0.030 0.086 0.016 19 16 41,598 2,724 0.667 0.024 0.246 0.016 0.116 0.012 19 17 33,669 1,603 0.790 0.031 0.294 0.020 0.139 0.013 19 18 56,918 5,189 0.417 0.031 0.116 0.012 0.046 0.006 19 19 39,565 2,703 0.742 0.035 0.241 0.023 0.102 0.014 19 20 38,581 900 0.715 0.036 0.245 0.025 0.109 0.016 19 21 72,867 2,246 0.458 0.032 0.148 0.011 0.066 0.007 19 22 49,194 2,094 0.462 0.041 0.113 0.019 0.040 0.010 19 23 42,533 994 0.691 0.047 0.232 0.024 0.102 0.015 19 24 47,512 2,468 0.489 0.045 0.126 0.016 0.047 0.008 19 25 48,903 1,636 0.547 0.041 0.156 0.028 0.059 0.016 19 26 65,119 3,391 0.206 0.044 0.040 0.010 0.012 0.004 19 27 81,529 525 0.103 0.037 0.016 0.007 0.004 0.002 19 28 56,682 1,916 0.363 0.047 0.073 0.013 0.022 0.005 19 29 36,605 265 0.662 0.070 0.283 0.023 0.149 0.022 19 30 43,055 1,467 0.593 0.060 0.172 0.024 0.070 0.012 20 1 67,602 15,481 0.313 0.033 0.069 0.010 0.023 0.005 20 2 66,746 7,695 0.307 0.059 0.062 0.016 0.019 0.006 20 3 51,952 8,483 0.561 0.051 0.126 0.021 0.039 0.009 20 4 51,347 6,968 0.540 0.070 0.119 0.024 0.038 0.010 20 5 49,399 20,563 0.606 0.038 0.141 0.020 0.045 0.009 20 6 51,223 22,908 0.559 0.038 0.126 0.018 0.040 0.008 20 7 48,226 20,894 0.625 0.038 0.165 0.023 0.059 0.012 20 8 58,061 3,521 0.491 0.052 0.134 0.026 0.050 0.014 20 9 62,396 7,430 0.381 0.046 0.084 0.017 0.028 0.008 20 10 54,581 14,268 0.480 0.060 0.123 0.028 0.046 0.015 20 11 55,130 15,304 0.507 0.048 0.114 0.019 0.036 0.008 20 12 49,614 6,864 0.574 0.058 0.133 0.023 0.042 0.010 21 1 22,461 849 0.944 0.028 0.529 0.028 0.332 0.029 21 2 36,115 1,049 0.812 0.046 0.317 0.030 0.157 0.020 21 3 22,845 879 0.925 0.029 0.540 0.025 0.359 0.027 21 4 40,844 980 0.778 0.037 0.339 0.027 0.184 0.019 21 5 31,868 2,460 0.847 0.032 0.393 0.036 0.221 0.031 21 6 35,168 2,204 0.817 0.038 0.345 0.029 0.182 0.021 163 21 7 50,595 3,184 0.600 0.055 0.225 0.026 0.115 0.016 21 8 29,612 2,842 0.893 0.033 0.401 0.038 0.216 0.029 21 9 29,909 2,807 0.889 0.025 0.424 0.028 0.244 0.026 21 10 25,951 2,708 0.899 0.026 0.489 0.025 0.312 0.024 21 11 24,206 1,137 0.893 0.025 0.554 0.031 0.374 0.037 21 12 57,602 2,810 0.576 0.036 0.242 0.020 0.133 0.016 21 13 31,705 1,413 0.878 0.022 0.422 0.034 0.239 0.032 21 14 40,488 2,347 0.762 0.046 0.310 0.028 0.162 0.018 21 15 50,392 1,760 0.651 0.055 0.261 0.027 0.140 0.017 21 16 48,060 3,125 0.691 0.040 0.303 0.039 0.165 0.032 21 17 33,998 2,103 0.823 0.040 0.389 0.037 0.220 0.029 22 1 46,148 1,934 0.475 0.047 0.103 0.012 0.033 0.005 22 2 50,901 1,767 0.439 0.055 0.102 0.017 0.035 0.007 22 3 47,306 5,257 0.560 0.054 0.110 0.018 0.031 0.006 22 4 54,399 2,047 0.601 0.050 0.137 0.017 0.046 0.007 22 5 48,143 5,136 0.507 0.056 0.103 0.016 0.031 0.006 22 6 48,440 1,122 0.632 0.045 0.163 0.015 0.059 0.006 22 7 44,833 2,696 0.512 0.069 0.111 0.018 0.037 0.006 22 8 46,241 6,267 0.518 0.050 0.098 0.016 0.027 0.006 22 9 50,715 5,770 0.449 0.049 0.078 0.013 0.021 0.004 22 10 47,083 5,411 0.526 0.048 0.102 0.016 0.029 0.006 22 11 57,943 12,144 0.344 0.035 0.074 0.010 0.024 0.004 22 12 60,038 5,867 0.293 0.028 0.066 0.008 0.022 0.003 22 13 48,528 1,430 0.433 0.060 0.100 0.022 0.034 0.010 22 14 69,291 4,414 0.422 0.038 0.101 0.012 0.035 0.005 22 15 59,351 703 0.312 0.057 0.064 0.016 0.020 0.007 23 1 44,930 9,172 0.632 0.031 0.165 0.013 0.056 0.006 23 2 44,449 8,575 0.627 0.028 0.173 0.011 0.062 0.005 23 3 39,896 4,070 0.709 0.028 0.204 0.013 0.075 0.007 23 4 48,167 7,862 0.559 0.035 0.137 0.013 0.045 0.006 23 5 48,216 9,234 0.574 0.043 0.144 0.018 0.048 0.008 23 6 41,710 3,722 0.689 0.042 0.180 0.022 0.061 0.010 23 7 41,031 9,954 0.721 0.025 0.198 0.012 0.070 0.006 23 8 38,231 8,360 0.743 0.026 0.211 0.011 0.077 0.005 23 9 40,566 7,905 0.732 0.032 0.188 0.017 0.062 0.008 23 10 39,949 9,272 0.752 0.031 0.200 0.014 0.068 0.006 23 11 45,914 4,708 0.656 0.048 0.173 0.020 0.059 0.009 23 12 37,762 2,696 0.746 0.026 0.233 0.013 0.090 0.007 23 13 40,349 3,327 0.719 0.025 0.207 0.012 0.076 0.006 23 14 42,111 1,943 0.705 0.036 0.181 0.017 0.060 0.008 23 15 46,486 3,467 0.586 0.039 0.149 0.014 0.051 0.006 23 16 46,720 5,304 0.598 0.034 0.158 0.012 0.056 0.006 25 1 65,386 4,369 0.423 0.073 0.141 0.034 0.064 0.019 25 2 52,180 7,342 0.638 0.058 0.302 0.045 0.177 0.035 25 3 37,284 4,208 0.719 0.043 0.321 0.036 0.179 0.028 25 4 29,158 3,672 0.831 0.036 0.399 0.029 0.232 0.023 25 5 49,526 4,938 0.582 0.056 0.193 0.027 0.085 0.016 164 25 6 44,586 2,741 0.652 0.040 0.255 0.032 0.128 0.023 25 7 42,325 4,442 0.650 0.052 0.211 0.029 0.090 0.017 25 8 50,238 3,899 0.474 0.058 0.120 0.019 0.043 0.009 25 9 43,790 3,563 0.562 0.061 0.167 0.024 0.069 0.012 25 10 33,159 14,745 0.803 0.026 0.347 0.025 0.186 0.020 25 11 36,647 6,776 0.723 0.051 0.269 0.032 0.131 0.021 25 12 35,759 8,308 0.726 0.055 0.272 0.033 0.133 0.021 25 13 55,331 6,139 0.446 0.060 0.109 0.021 0.038 0.009 25 15 62,528 12,123 0.624 0.050 0.222 0.032 0.101 0.020 26 1 23,873 919 0.846 0.045 0.520 0.036 0.351 0.032 26 2 54,429 266 0.548 0.104 0.189 0.052 0.089 0.031 26 3 47,287 850 0.561 0.048 0.209 0.027 0.102 0.018 26 4 15,542 667 0.953 0.020 0.601 0.024 0.417 0.023 26 5 36,717 2,292 0.726 0.050 0.395 0.025 0.256 0.018 26 6 49,639 2,186 0.658 0.037 0.274 0.030 0.145 0.023 26 7 73,738 1,790 0.417 0.037 0.181 0.020 0.101 0.013 26 8 35,044 783 0.741 0.033 0.363 0.025 0.219 0.020 26 9 61,525 2,687 0.544 0.051 0.222 0.027 0.117 0.017 26 10 48,043 943 0.689 0.025 0.342 0.029 0.199 0.025 26 11 34,448 1,282 0.750 0.043 0.398 0.030 0.250 0.025 26 12 150,256 1,814 0.223 0.053 0.059 0.021 0.022 0.010 26 13 52,215 3,702 0.613 0.059 0.314 0.031 0.197 0.021 26 14 76,852 1,000 0.374 0.050 0.159 0.021 0.087 0.013 27 1 53,356 2,252 0.304 0.047 0.048 0.010 0.012 0.003 27 2 40,988 4,788 0.629 0.034 0.145 0.016 0.046 0.007 27 3 46,994 7,748 0.545 0.038 0.119 0.013 0.037 0.006 27 4 48,817 8,220 0.462 0.044 0.087 0.013 0.023 0.005 27 5 37,643 10,264 0.758 0.032 0.187 0.023 0.060 0.012 27 6 50,037 4,708 0.416 0.066 0.068 0.016 0.017 0.005 27 7 43,442 9,305 0.606 0.033 0.126 0.015 0.036 0.006 27 8 51,370 694 0.339 0.071 0.049 0.014 0.011 0.004 27 9 41,675 6,343 0.616 0.036 0.152 0.024 0.053 0.014 28 1 133,075 478 0.308 0.042 0.104 0.021 0.043 0.011 28 2 208,641 351 0.196 0.052 0.050 0.022 0.017 0.010 28 3 114,601 296 0.055 0.043 0.008 0.007 0.002 0.002 28 4 100,139 1,025 0.161 0.065 0.037 0.020 0.013 0.008 28 5 86,078 689 0.303 0.046 0.101 0.019 0.042 0.009 28 6 135,159 994 0.122 0.034 0.028 0.010 0.009 0.004 28 7 116,445 683 0.321 0.034 0.101 0.021 0.039 0.011 28 8 97,208 575 0.099 0.070 0.019 0.017 0.006 0.006 28 9 83,013 812 0.325 0.077 0.075 0.026 0.024 0.011 29 1 31,598 4,325 0.908 0.031 0.322 0.032 0.135 0.021 29 2 28,875 1,501 0.951 0.018 0.395 0.028 0.188 0.023 29 3 30,231 4,512 0.942 0.021 0.347 0.033 0.148 0.023 29 4 32,595 3,577 0.901 0.032 0.307 0.035 0.123 0.022 29 5 32,469 4,395 0.911 0.033 0.319 0.028 0.134 0.019 29 6 30,984 2,339 0.930 0.027 0.349 0.025 0.154 0.018 165 29 7 27,472 2,461 0.985 0.009 0.406 0.033 0.185 0.026 29 8 29,269 3,616 0.950 0.019 0.373 0.030 0.167 0.022 29 9 31,705 2,741 0.934 0.024 0.342 0.022 0.149 0.015 29 10 33,162 3,949 0.913 0.038 0.314 0.029 0.132 0.017 29 11 33,744 3,291 0.877 0.045 0.300 0.025 0.131 0.015 29 12 33,824 8,470 0.896 0.038 0.299 0.024 0.125 0.014 29 13 31,144 2,785 0.931 0.027 0.336 0.027 0.144 0.018 29 14 29,544 2,637 0.952 0.020 0.377 0.026 0.175 0.020 29 15 26,817 2,013 0.962 0.019 0.433 0.033 0.218 0.030 29 16 36,820 3,474 0.810 0.058 0.256 0.023 0.105 0.013 29 17 34,004 2,575 0.877 0.044 0.307 0.031 0.136 0.022 29 18 29,103 5,433 0.941 0.019 0.390 0.030 0.187 0.025 29 19 32,291 5,765 0.918 0.029 0.329 0.023 0.143 0.015 29 20 31,242 3,161 0.948 0.022 0.358 0.029 0.161 0.021 30 1 34,246 3,879 0.770 0.048 0.311 0.034 0.157 0.023 30 2 35,970 5,326 0.769 0.030 0.280 0.019 0.131 0.013 30 3 29,954 8,344 0.856 0.024 0.354 0.018 0.177 0.014 30 4 36,888 3,700 0.749 0.029 0.270 0.026 0.125 0.018 30 5 34,504 3,492 0.762 0.040 0.286 0.023 0.138 0.015 30 6 37,394 7,249 0.769 0.037 0.307 0.022 0.154 0.015 30 7 43,303 2,483 0.669 0.045 0.231 0.022 0.105 0.014 30 8 31,834 5,528 0.838 0.022 0.341 0.026 0.169 0.021 30 9 31,708 8,050 0.842 0.026 0.340 0.022 0.167 0.016 31 1 49,569 4,837 0.481 0.055 0.138 0.021 0.057 0.010 31 2 46,757 8,699 0.543 0.043 0.177 0.019 0.080 0.010 31 3 44,900 9,127 0.556 0.031 0.193 0.016 0.091 0.010 31 4 50,997 8,087 0.537 0.029 0.192 0.017 0.090 0.011 31 5 40,953 6,676 0.631 0.033 0.231 0.019 0.111 0.012 31 6 49,504 8,537 0.534 0.042 0.176 0.020 0.079 0.011 166 Table A4: Estimates of poverty indicators (urban areas) Governorate District #hhno avg_ ^ y avg_FGT0 se_FGT0 avg_FGT1 se_FGT1 avg_FGT2 se_FGT2 11 1 31,582 700 0.730 0.042 0.280 0.029 0.138 0.019 11 2 40,130 6,765 0.553 0.021 0.177 0.012 0.077 0.007 11 3 23,294 478 0.848 0.037 0.387 0.032 0.213 0.023 11 4 30,726 844 0.762 0.034 0.311 0.031 0.158 0.022 11 5 48,019 212 0.476 0.058 0.149 0.028 0.064 0.016 11 6 28,923 724 0.754 0.038 0.316 0.027 0.166 0.018 11 7 40,754 303 0.556 0.043 0.203 0.024 0.099 0.015 11 8 32,350 840 0.674 0.042 0.243 0.023 0.114 0.015 11 11 49,435 1,093 0.427 0.028 0.138 0.013 0.061 0.008 11 12 38,766 1,785 0.604 0.041 0.213 0.023 0.100 0.014 11 13 54,512 490 0.335 0.051 0.085 0.017 0.031 0.008 11 14 34,006 349 0.680 0.047 0.263 0.027 0.130 0.018 11 15 36,372 324 0.651 0.055 0.224 0.032 0.101 0.019 11 16 41,661 6,537 0.548 0.020 0.203 0.011 0.099 0.007 11 17 27,479 177 0.792 0.040 0.352 0.030 0.189 0.022 11 18 54,277 11,283 0.370 0.017 0.117 0.008 0.051 0.005 11 19 53,017 16,239 0.390 0.016 0.136 0.009 0.064 0.005 12 1 23,229 271 0.909 0.034 0.449 0.041 0.258 0.031 12 2 49,985 991 0.502 0.041 0.175 0.023 0.081 0.014 12 4 53,188 1,574 0.544 0.032 0.211 0.018 0.108 0.013 12 6 17,034 130 0.982 0.019 0.571 0.052 0.357 0.051 12 9 25,625 665 0.775 0.044 0.379 0.035 0.221 0.027 12 10 58,058 2,384 0.457 0.031 0.155 0.013 0.075 0.008 12 11 39,795 7,458 0.651 0.022 0.267 0.016 0.139 0.011 13 1 53,073 8,935 0.501 0.026 0.148 0.011 0.060 0.006 13 2 52,726 28,942 0.504 0.017 0.155 0.008 0.064 0.004 13 3 48,386 15,137 0.549 0.028 0.173 0.014 0.073 0.008 13 4 57,472 14,640 0.444 0.024 0.136 0.010 0.058 0.005 13 5 63,688 42,131 0.390 0.015 0.111 0.006 0.044 0.003 13 6 67,016 14,814 0.367 0.017 0.101 0.007 0.039 0.003 13 7 63,888 10,260 0.387 0.019 0.109 0.008 0.043 0.004 13 8 59,377 39,035 0.439 0.015 0.134 0.007 0.056 0.004 13 9 58,917 23,958 0.434 0.018 0.127 0.008 0.051 0.004 13 10 44,032 18,583 0.624 0.018 0.209 0.010 0.092 0.006 13 19 47,073 3,834 0.559 0.036 0.169 0.016 0.069 0.008 13 24 51,698 16,246 0.515 0.020 0.158 0.009 0.066 0.005 14 4 48,510 319 0.454 0.101 0.109 0.038 0.036 0.017 14 5 45,180 162 0.498 0.123 0.105 0.041 0.032 0.016 14 6 30,532 517 0.845 0.038 0.300 0.031 0.131 0.020 14 8 56,731 341 0.534 0.034 0.181 0.025 0.076 0.015 14 9 46,351 3,592 0.566 0.043 0.151 0.021 0.055 0.011 14 11 44,481 94 0.469 0.195 0.104 0.065 0.032 0.026 14 13 55,000 6,035 0.411 0.035 0.102 0.015 0.036 0.007 167 14 14 36,139 70 0.738 0.111 0.207 0.062 0.073 0.032 14 16 47,176 1,656 0.524 0.057 0.148 0.027 0.058 0.014 15 1 23,796 277 0.870 0.041 0.450 0.048 0.274 0.040 15 2 21,260 59 0.915 0.043 0.484 0.055 0.294 0.050 15 3 17,561 353 0.934 0.033 0.541 0.049 0.353 0.044 15 4 26,037 302 0.805 0.041 0.430 0.031 0.273 0.024 15 5 48,443 1,338 0.531 0.044 0.258 0.027 0.155 0.019 15 6 39,799 479 0.624 0.056 0.300 0.034 0.178 0.025 15 7 29,656 715 0.753 0.039 0.391 0.031 0.243 0.023 15 8 16,181 289 0.955 0.032 0.569 0.059 0.375 0.056 15 11 33,948 497 0.716 0.053 0.302 0.036 0.161 0.025 15 12 33,631 3,265 0.699 0.029 0.321 0.020 0.185 0.014 15 14 42,251 1,570 0.589 0.045 0.229 0.023 0.118 0.014 15 16 70,472 298 0.273 0.048 0.083 0.020 0.036 0.011 15 17 49,178 24,378 0.528 0.015 0.217 0.009 0.116 0.006 15 18 58,035 21,958 0.445 0.019 0.168 0.010 0.085 0.007 15 19 52,636 21,670 0.497 0.019 0.195 0.011 0.101 0.007 15 21 58,266 212 0.360 0.071 0.099 0.027 0.040 0.014 16 1 32,163 293 0.800 0.051 0.255 0.041 0.102 0.024 16 3 28,986 141 0.829 0.054 0.281 0.037 0.119 0.022 16 4 24,401 297 0.977 0.017 0.433 0.047 0.213 0.038 16 5 37,353 1,475 0.701 0.032 0.289 0.020 0.152 0.013 16 6 33,126 1,244 0.806 0.037 0.282 0.026 0.122 0.016 16 7 36,209 468 0.768 0.069 0.210 0.041 0.074 0.021 16 8 33,013 787 0.781 0.028 0.341 0.023 0.175 0.019 16 9 22,451 878 0.963 0.012 0.504 0.019 0.296 0.018 16 10 52,575 818 0.425 0.051 0.101 0.018 0.034 0.008 16 11 32,358 256 0.730 0.058 0.230 0.033 0.093 0.018 16 12 25,106 407 0.857 0.041 0.376 0.039 0.198 0.032 17 2 67,836 3,167 0.388 0.031 0.114 0.014 0.046 0.008 17 3 65,918 951 0.442 0.055 0.134 0.024 0.055 0.013 17 4 63,201 3,454 0.341 0.036 0.091 0.014 0.034 0.007 17 5 49,545 426 0.500 0.062 0.152 0.030 0.063 0.016 17 6 45,854 93 0.500 0.113 0.145 0.046 0.057 0.024 17 7 52,579 254 0.433 0.100 0.110 0.035 0.041 0.016 17 8 34,078 82 0.698 0.100 0.231 0.047 0.101 0.026 17 11 40,930 373 0.585 0.071 0.180 0.036 0.074 0.019 17 15 55,504 1,208 0.471 0.058 0.140 0.027 0.058 0.014 17 16 59,925 137 0.305 0.085 0.072 0.027 0.025 0.012 17 17 73,839 52 0.132 0.100 0.024 0.022 0.007 0.007 17 18 36,257 325 0.674 0.068 0.265 0.038 0.134 0.025 17 20 45,606 230 0.565 0.062 0.205 0.041 0.097 0.027 17 21 40,777 424 0.650 0.067 0.254 0.037 0.127 0.024 17 22 48,225 175 0.619 0.053 0.267 0.048 0.140 0.037 17 23 31,049 116 0.813 0.091 0.308 0.065 0.146 0.041 17 24 45,033 104 0.583 0.105 0.178 0.049 0.073 0.027 17 25 48,898 290 0.497 0.091 0.159 0.045 0.069 0.025 168 17 28 48,866 4,276 0.578 0.034 0.228 0.020 0.115 0.013 18 1 40,436 1,682 0.596 0.031 0.188 0.015 0.080 0.009 18 2 49,836 2,164 0.413 0.037 0.101 0.012 0.035 0.005 18 3 37,368 457 0.590 0.055 0.277 0.030 0.159 0.021 18 4 19,738 195 0.962 0.031 0.543 0.067 0.336 0.063 18 5 45,105 1,008 0.504 0.040 0.133 0.016 0.049 0.008 18 6 46,900 637 0.466 0.046 0.132 0.018 0.052 0.009 18 7 45,021 3,430 0.505 0.027 0.137 0.012 0.052 0.006 18 8 28,737 738 0.854 0.063 0.331 0.066 0.157 0.046 18 9 32,203 2,941 0.743 0.021 0.314 0.022 0.165 0.019 18 10 43,516 7,128 0.533 0.024 0.164 0.011 0.068 0.006 18 11 38,330 263 0.621 0.067 0.203 0.032 0.087 0.018 18 13 41,497 4,759 0.569 0.023 0.173 0.011 0.072 0.006 18 14 55,622 1,325 0.348 0.033 0.085 0.011 0.030 0.005 18 15 37,986 284 0.592 0.063 0.180 0.030 0.075 0.016 18 16 45,726 1,830 0.486 0.027 0.134 0.011 0.051 0.006 18 17 45,602 6,731 0.495 0.023 0.143 0.010 0.056 0.005 18 19 44,852 2,018 0.506 0.027 0.146 0.013 0.059 0.007 18 20 44,377 1,741 0.506 0.041 0.132 0.017 0.048 0.008 18 21 49,195 20,706 0.478 0.017 0.141 0.009 0.057 0.005 18 22 59,610 12,561 0.395 0.022 0.112 0.009 0.044 0.005 18 23 47,195 23,081 0.514 0.019 0.152 0.010 0.061 0.005 18 24 48,562 4,787 0.462 0.029 0.140 0.012 0.059 0.006 18 25 41,118 2,781 0.562 0.026 0.173 0.013 0.072 0.007 18 26 44,220 1,662 0.525 0.036 0.144 0.015 0.054 0.008 19 1 44,201 195 0.590 0.153 0.172 0.068 0.069 0.038 19 2 56,399 230 0.387 0.177 0.068 0.048 0.017 0.016 19 6 42,848 132 0.626 0.199 0.200 0.121 0.082 0.068 19 7 42,958 1,171 0.631 0.082 0.169 0.033 0.062 0.018 19 8 41,439 1,430 0.673 0.069 0.163 0.033 0.054 0.017 19 9 39,063 474 0.735 0.199 0.191 0.091 0.066 0.042 19 10 31,456 5,609 0.894 0.033 0.334 0.038 0.149 0.026 19 11 35,611 5,011 0.827 0.036 0.299 0.030 0.131 0.019 19 13 56,775 1,507 0.275 0.104 0.045 0.022 0.011 0.007 19 14 46,875 1,741 0.537 0.087 0.111 0.030 0.034 0.013 19 15 48,062 6,647 0.497 0.069 0.099 0.021 0.029 0.008 19 16 53,440 148 0.333 0.290 0.055 0.068 0.014 0.023 19 17 39,943 4,206 0.739 0.055 0.196 0.032 0.069 0.016 19 18 36,653 217 0.762 0.206 0.297 0.107 0.145 0.062 19 19 61,776 308 0.181 0.194 0.030 0.040 0.008 0.012 19 20 43,889 68 0.581 0.317 0.119 0.110 0.034 0.043 19 21 49,623 238 0.421 0.230 0.084 0.067 0.025 0.026 19 22 24,553 160 0.963 0.112 0.457 0.135 0.238 0.105 19 24 54,942 140 0.332 0.305 0.059 0.087 0.016 0.032 19 25 58,159 287 0.266 0.216 0.048 0.055 0.014 0.020 19 26 45,642 1,116 0.556 0.114 0.120 0.043 0.037 0.019 19 27 48,613 331 0.440 0.275 0.104 0.080 0.036 0.032 169 19 28 67,137 389 0.192 0.117 0.042 0.028 0.014 0.011 19 29 51,292 22,113 0.449 0.055 0.092 0.017 0.028 0.007 20 1 51,595 533 0.462 0.073 0.147 0.034 0.064 0.019 20 2 42,505 3,305 0.582 0.035 0.212 0.018 0.102 0.011 20 3 25,875 92 0.853 0.064 0.377 0.060 0.198 0.044 20 4 30,393 132 0.828 0.064 0.352 0.060 0.180 0.044 20 7 56,407 328 0.460 0.068 0.156 0.035 0.070 0.020 20 8 43,117 19,145 0.619 0.023 0.230 0.012 0.110 0.008 20 9 58,612 113 0.371 0.121 0.108 0.049 0.045 0.025 20 11 77,270 664 0.341 0.049 0.098 0.019 0.039 0.009 21 4 26,415 129 0.951 0.029 0.412 0.040 0.208 0.032 21 5 55,312 246 0.520 0.139 0.183 0.088 0.083 0.055 21 7 40,817 1,467 0.657 0.046 0.231 0.028 0.106 0.018 21 10 34,346 869 0.802 0.039 0.285 0.028 0.127 0.019 21 11 24,137 149 0.952 0.038 0.445 0.051 0.246 0.041 21 12 27,970 348 0.917 0.031 0.400 0.037 0.206 0.031 21 13 62,452 2,626 0.339 0.030 0.127 0.010 0.066 0.009 21 14 25,393 270 0.955 0.029 0.439 0.047 0.232 0.038 21 15 35,680 549 0.732 0.110 0.294 0.066 0.148 0.041 21 16 34,470 660 0.805 0.056 0.271 0.038 0.117 0.023 21 17 40,138 238 0.687 0.141 0.190 0.071 0.071 0.036 22 1 60,949 370 0.264 0.083 0.055 0.025 0.017 0.010 22 2 52,277 472 0.486 0.072 0.141 0.031 0.055 0.016 22 3 63,224 56 0.292 0.073 0.066 0.020 0.022 0.009 22 5 46,407 643 0.509 0.058 0.142 0.026 0.054 0.014 22 7 66,661 492 0.296 0.059 0.069 0.021 0.023 0.009 22 8 56,844 379 0.340 0.072 0.077 0.022 0.025 0.009 22 9 41,626 306 0.600 0.065 0.193 0.034 0.082 0.020 22 10 44,256 1,692 0.555 0.026 0.193 0.015 0.089 0.009 22 11 45,020 1,389 0.523 0.038 0.161 0.019 0.067 0.011 22 14 56,537 85 0.330 0.208 0.087 0.072 0.033 0.033 22 15 51,481 5,665 0.460 0.020 0.156 0.010 0.071 0.006 24 1 50,550 10,962 0.562 0.027 0.180 0.014 0.078 0.008 24 2 49,482 13,925 0.541 0.027 0.177 0.014 0.077 0.008 24 3 51,878 15,338 0.483 0.028 0.154 0.014 0.067 0.008 24 4 45,051 9,019 0.578 0.032 0.187 0.018 0.081 0.010 24 5 51,084 8,025 0.476 0.030 0.140 0.014 0.057 0.008 24 6 58,486 7,533 0.403 0.029 0.112 0.012 0.044 0.006 24 7 54,492 11,811 0.451 0.026 0.132 0.012 0.054 0.006 24 8 55,080 6,354 0.448 0.032 0.138 0.015 0.059 0.009 25 1 70,393 541 0.445 0.062 0.168 0.033 0.081 0.020 25 4 64,130 200 0.241 0.114 0.065 0.035 0.026 0.016 25 6 26,247 271 0.670 0.102 0.314 0.067 0.181 0.044 25 7 31,443 1,149 0.720 0.064 0.253 0.038 0.114 0.021 25 8 22,671 161 0.940 0.047 0.425 0.076 0.219 0.059 25 9 35,564 440 0.657 0.074 0.242 0.044 0.113 0.027 25 10 32,031 204 0.608 0.099 0.227 0.055 0.107 0.032 170 25 14 60,367 4,005 0.353 0.039 0.104 0.017 0.043 0.008 25 15 48,067 1,524 0.555 0.050 0.180 0.030 0.076 0.017 26 7 53,236 224 0.438 0.097 0.098 0.032 0.032 0.013 26 8 66,062 106 0.147 0.087 0.024 0.018 0.006 0.005 26 9 42,128 789 0.624 0.056 0.200 0.032 0.085 0.018 26 12 64,905 1,671 0.346 0.036 0.087 0.015 0.031 0.007 26 13 43,435 535 0.593 0.053 0.190 0.030 0.080 0.018 27 1 39,374 888 0.642 0.047 0.183 0.024 0.071 0.013 27 2 39,417 674 0.662 0.072 0.163 0.031 0.056 0.015 27 3 56,956 561 0.465 0.068 0.094 0.020 0.027 0.008 27 4 33,596 243 0.815 0.048 0.250 0.036 0.096 0.021 27 6 42,483 210 0.550 0.105 0.114 0.034 0.033 0.013 27 8 34,324 1,593 0.799 0.035 0.250 0.021 0.102 0.011 28 3 53,376 298 0.392 0.047 0.112 0.022 0.044 0.011 28 4 89,034 1,516 0.163 0.025 0.044 0.008 0.017 0.004 28 6 44,076 494 0.518 0.046 0.182 0.019 0.084 0.010 28 7 71,831 821 0.189 0.031 0.042 0.010 0.014 0.005 28 8 54,831 718 0.416 0.038 0.142 0.014 0.064 0.007 28 9 67,422 227 0.264 0.059 0.062 0.019 0.022 0.008 29 1 26,564 320 0.940 0.039 0.425 0.053 0.221 0.043 29 2 42,516 742 0.699 0.053 0.292 0.035 0.151 0.024 29 3 24,084 117 0.978 0.023 0.479 0.071 0.256 0.064 29 4 23,508 282 0.976 0.022 0.462 0.052 0.244 0.046 29 5 35,588 273 0.754 0.107 0.267 0.065 0.119 0.040 29 6 27,468 274 0.941 0.028 0.425 0.050 0.216 0.042 29 8 17,309 240 0.986 0.020 0.605 0.046 0.400 0.047 29 9 27,275 161 0.917 0.069 0.415 0.067 0.228 0.050 29 10 23,118 220 0.953 0.049 0.483 0.081 0.279 0.070 29 11 33,613 1,752 0.786 0.039 0.299 0.026 0.142 0.016 29 13 16,031 122 0.999 0.002 0.664 0.074 0.455 0.093 29 15 40,368 9,391 0.667 0.028 0.251 0.018 0.121 0.012 29 16 13,648 176 0.996 0.007 0.689 0.056 0.496 0.070 29 17 40,563 1,303 0.675 0.061 0.256 0.034 0.124 0.021 29 19 38,339 1,762 0.693 0.046 0.257 0.029 0.122 0.018 30 1 46,413 945 0.539 0.092 0.172 0.043 0.075 0.023 30 2 40,907 1,896 0.631 0.059 0.208 0.036 0.092 0.022 30 3 38,961 1,237 0.682 0.070 0.234 0.036 0.107 0.020 30 4 48,733 560 0.451 0.109 0.123 0.040 0.049 0.018 30 5 32,080 269 0.801 0.085 0.329 0.066 0.168 0.045 30 6 46,284 2,365 0.560 0.074 0.175 0.029 0.076 0.015 30 7 39,269 59 0.618 0.160 0.204 0.078 0.093 0.043 30 9 34,697 386 0.769 0.076 0.315 0.061 0.160 0.044 171 Figure A1 Rural headcount index (food) plus/minus 2 standard errors 1 .8 Headcount index .4 .2 0 .6 0 100 200 300 Districts sorted by headcount index Figure A2 Urban headcount index (food) plus/minus 2 standard errors 1.5 1 Headcount index .5 0 -.5 0 50 100 150 200 250 Districts sorted by headcount index 172 ANNEX 6: HEALTH 1. The Ministry of Population and Public Health (MOPHP) operates a four-tired health system which is comprised by health centers and health units at the village level and districts for primary health care, district and governorate hospitals for the secondary care and referral hospitals in Sana'a, Aden and other big cities. According to the latest data available from the MOPHP in 2002, there were 15 major hospitals, 22 governorate hospitals, 111 rural and district hospitals, 614 health centers and 2,025 health units under the MOPHP's system. Since 2002, the responsibilities of operating these health facilities have been shifted to the local level due to the implementation of decentralization policy of the Government, and additional numbers of health facilities were constructed. Available data show that the number of installed hospital bed was 0.6 per 1,000 population11. This is much lower than other countries in the MENA region. It must be noted that the number of operating health facilities with adequate equipment or health staff is much less than what physically exists. The National Health Account Study12 found that Government expenditure on health was nearly 1.8% of Gross Domestic Product in 2003 and accounted 32% of the total health expenditure of the country. Per capita public health expenditure was only US $11. Government spending on health was 4.9%. Currently, Yemen does not have a compulsory health insurance system that provides financial protection to the poor in the case of catastrophic illnesses. 2. The MOPHP has initiated the Health Sector Reform Strategy in 1998 to improve the performance of their health system. The Strategy attempted to address improvement of management systems, decentralization of management functions, and cost sharing and strengthening the stewardship role of the MOPHP. The Strategy tried to improve critical issues in health sector such as improvement of efficiency, equity, and access to health care. In the last ten years, the Health Sector Reform Strategy has faced political and administrative challenges in its implementation process. In spite of high expectations from those who involved, there has not been so much progress witnessed over the last ten year. The Government of Yemen planed to reformulate the Health Sector Strategy of 1998 into the third five-year plan (2006-2010). 3. With respect to health outcomes, there have been gradual improvements in the recent years. However, Yemen continues to be one of the countries with the worst health outcomes. As of 2005, infant and under-five mortalities are 76 per 1,000 live births102 per 1,000 live births respectively13. They are the second highest in the MENA regions. Moreover, according to the most recent health survey14, nearly 50% of under-five year old children are underweight. At this moment, most of the MDG targets related to health will not be achieved by 2015, provided that the current level of interventions continues. 11 The data source is World Health Organization's World Health Report 2006. This figure is believed to be the bed capacity in the public sector. There was not an accurate estimate available at the time of writing this report. 12 Yemen National Health Accounts: Estimate for 2003, National Health Accounts Team, Republic of Yemen, Partners for Health Reformplus, June 2006 13 The data source is United Nations Children's Fund, State of the World Children 2007. 14 Pan Arab Family Health Survey 2003 173 4. The previous Poverty Assessment in 2002 highlighted the need for increase both the access and the quality of services. The previous Poverty Assessment suggested that the priority interventions for Yemen would be to (i) strengthen maternal and child health programs in rural areas, (ii) redirect health care resources towards primary health care, and (iii) increase the public finance resource to the health sector. The World Bank's analysis, Inequality in health, nutrition and population in Yemen15 in 2003 also recommended the Government of Yemen undertake more aggressive pro-poor targeting in their health service provisions and ensure the effectiveness of the targeting, particularly reduction of maternal and child mortality. The analysis supported introduction of health insurance program targeted the poor and emphasized needs for inter-sectoral approaches to improve transport, communication and educational attainment of women, particularly in relation to reduction of maternal and child mortality, and child malnutrition. 5. This chapter aims to examine if the health inequality between the poor and the rich has been alleviated over the course of years, due to the collective efforts of the government, donor agencies and other non-governmental agencies. This chapter does not necessarily intend to evaluate the past performance of the Health Sector Reform Strategy in place in the last ten years, as observed changes may not be only contribution of the Health Sector Reform Strategy. At the end of the section, a set of policy recommendations were made for future consideration. A Access to health care 6. According to the 2005 HBS, on national average, nearly 70% of the surveyed individuals sought medical treatment at a health facility, either public or private, in the same district or area when ill. This indicates that the surveyed individuals have relatively good physical access to health care. Regarding the geographical disparity, the urban and rural gap was relatively small, only 10% point. Equally the different between the richest quintile and poorest expenditure quintile was not significantly large. In the previous Poverty Assessment used a slightly different measure to assess access to health care, percentage of population with at least one health functional health facility only in the same area16. In the previous Poverty Assessment, nationally, only 38% of the population lived in an area where at least one health facility, whether public or private. On average, some 94% of urban population lives in the area with at least one health facility, compared with 21% in the rural area. 15 Preker A.S. Nandini Omman, Elizabeth Lule, Deeborah Vazirani, and Ritu Chhabra, Yemen Inequalities in Health, Nutrition and Population, World Bank June 2003 16 The comparability in the definition of "area" between these two HBSs was not clear at the same of writing this section. Additionally, the 1998 HBS looked into only the distribution of health facilities by available data. 174 Figure A.6 1: Resident Access to Health Care, by District Acce s s to he alth care , % of population who s e e k care in the s ame are a or dis trict 100 94 90 75.6 74.6 80 68.4 64.6 65.4 70 60 50 38 40 31.1 30 21 22.4 20 10 0 National Urban Rural Quintile 1 Quntile 5 average 1998 2005 7. In addition to overall trend of the improved physical access to health care, more percentage of individuals seems to now seek medical care at the time of sickness. This trend is observed in both the poor and non-poor groups. In the previous Poverty Assessment, only 38.8% of the poor who were ill during the preceding month of the survey sought treatment. In the 2005 HBS, 61.5% of the poor sought treatment in case of illness. The access to health care seems to have improved further for the non-poor. The percentage of the non-poor individuals who were ill during the preceding the interview sought treatment has increased from 41.1% in 1998 to 76.8% in 2005. 8. Despite the overall upward trend, the percentage to seek medical care still varies by household's expenditure levels. In the 2005 HBS, the percentage of individuals who were ill during the preceding the survey and sought treatment ranges from 57.1% in the poorest quintile17 to 80.4% in the richest quintile. Compared with the 1998 HBS figures, the difference of the percentages between the poorest and richest has widened. The percentage increase in seeking the care at the time of illness did not catch up the improved physical access to health facility for the individuals in the poorest expenditure group. What are reasons for the individuals in the poorest expenditure group not to seek the medical care, other than physical access to health facilities? 17 Quintiles are defined by household expenditure divided by the total number of household members. 175 Figure A.6 2: Percentage of Residents Who Sought Medical Care % of thos e who s ought me dical care among those whowe re ill during the pre ce ding month of the inte rvie w 90 80.4 76.1 76.3 80 65.9 70 57.1 60 50 % 40 31.1 30 22.4 20 10 0 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Expenditure Quintile 1998 2005 9. As show in Figure XXX, the reasons for not seeking medical care in case of illness demonstrate variations among the household expenditure groups. For the poor, inability to pay for care seems to be the most significant reason for not seeking the medical care. Unavailability of needed medical service and difficulty in physical access were the second and third reasons for not seeking the medical care18. This trend changes as the household expenditure increases. Inability to pay for medical care and unavailability of care become less significant barriers the expenditure level goes up. In the richest quintile, the major reason for not seeking care is that the illness was considered too minor to receive the medical attention. 18 The 2005 HBS did not assess how many minutes it takes to go to a nearby health facility. 176 Figure A.6 3: Percentage of Residents Whom Did Not Seek Medical Care Reasons for not seeking medical care 70% 60% 50% 40% 30% 20% 10% 0% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Cannot affort it T ravel is t oo difficult No female doct or/nurse available No male doct or/nurse available Services not available Illness was minor Bad services Social reasons Ot her 10. This finding coincides with the result of the recent Bank financed qualitative research study, Qualitative Assessment of Community Based Health Related Programs: Five Programs and Six Locations in Yemen. The study conducted twenty four focus groups with community members who were beneficiaries of existing community-based development projects financed by different types of donors. The qualitative study indicates that the focus group participants raised concerned over lack or access or difficulties to health centers, lack of quality health services and financial barrier. The poor seems to have more difficulties in accessing health care, due to the cost-sharing schemes which were introduced in the early 1990s. B Utilization of services/ Health Behavior 11. In regard to utilization of health services, the 2005 HBS indicated that throughout the expenditure groups, nearly 80% of individuals seek medical care only when they are ill. Some 10% of the surveyed individuals claimed that they do not visit a health professional at all. This trend indicates that Yemenis, whether the poor or the rich, do not have a custom of paying a preventive medical care visit. 177 Table A.6. 1: Frequency of Visiting Health Care Professionals Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Rural Monthly 0.2% 0.2% 0.4% 0.5% 0.9% 0.7% 0.3% Every 2 month 0.1% 0.1% 0.2% 0.2% 0.4% 0.3% 0.1% Every 3 month 0.0% 0.1% 0.0% 0.1% 0.3% 0.2% 0.1% Every 4 month 0.0% 0.0% 0.0% 0.0% 0.1% 0.1% 0.0% Every 5 month 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Twice a year 0.1% 0.0% 0.0% 0.1% 0.2% 0.2% 0.0% No regularity 0.2% 0.3% 0.4% 0.3% 0.9% 0.8% 0.3% When ill 79.3% 82.9% 83.8% 84.6% 85.6% 87.9% 81.4% Don't see health prpfessional 20.2% 16.4% 15.2% 14.2% 11.6% 9.8% 17.7% 100.0% 100.0% 100.0% 100.0% 100% 100.0% 100.0% Source: World Bank staff estimates from the 2005 HBS 12. In spite of the fact that more individuals seem to seek medical care when ill, the percentage of women who received assisted delivery remain very low in Yemen. The national average is less than one-third of women who delivered a child. The percentage increases from 19% in the poorest to 40% in the richest. A geographical gap between the urban and rural is also noticeably huge, 13% point difference. A comparison with the 1997 PAPCHILD data implies that there has been some improvement, in particular, in the poorest segments of women. Illiteracy and cultural norms among Yemeni women may limit their demand for utilizing the assisted delivery. Perceptions of risk during the periods of pregnancy, birth and postpartum affect strongly care seeking behavior of women. This impediment could be aggravated by lack of physical access to a facility or financial barrier. There was notable disparity among the governorates: the percentage of assisted delivery was extremely low in the governorates of Abyan (14.7%), Al-Jawf (11.4%), Hajja (9.8%), Dhamar (13.6%), Shabwah (18.0%), Sana'a (18.0%) and Al- Mahwe (17.6%). Table: Assisted delivery by expenditure quintile (% of women) National Quintile Quintile Quintile Quintile Quintile Urban Rural average 1 2 3 4 5 1997 21.7 6.8 13.2 15.6 28.7 49.7 n/a n/a 2005 29.4 19.8 23.5 27.1 32.8 40.3 49.9 21.2 Source: PAPCHILD 1997 and HBS 2005 Note: The PAPCHILD figures are percentage of delivery by a medically trained person. The questionnaire of HBS 2005 did not clearly distinguish if the question meant to ask a delivery by a doctor, midwife or a medically trained professional. 13. Exclusive breast-feeding practice for the first six month after the birth seemed to be universally accepted among those who were interviewed. There was no noticeable disparity among the expenditure groups and the non-poor and the poor. However, several governorates have shown lower prevalence of this practice: Hajja (63.0%), Shabwah (64.9%), and Remah (53.9%). 14. Child malnutrition remained persistently a concerning issue in Yemen. Nearly one third of children two to five year old are severely stunting. There is a larger disparity between in the urban and rural area on the prevalence of severe stunting than other types 178 of malnutrition. Poverty is clearly associated to the prevalence of severe stunting and underweight. Since the WHO guideline for international reference population was revised in 2005, it was not possible to compare the prevalence of child malnutrition with the results of earlier health surveys. On the other hand, available data infers that the prevalence of child malnutrition has not been reduced over the course of years. Severe stunting was more prevalent in the governorates of Al-Jawf (52.0%), Al-Mahweet (43.2%), Amran (39.4%) and Sana'a (47.6%). Severe underweight was more common in the governorates of Al-Hodeida (19.9%), Mareb (14.8%), Amran (18.3%) and Al-Dhale (31.0%). More prevalent severe wasting was observed in the governorates of Al-Mahrh (27.7%), Al-Dhale (23.4%), Mareb (22.7%) and Laheg (22.9%). Table: Prevalence of severe malnutrition (%) National Non- average Poor poor Urban Rural Boy Girl Severe stunting 27.5% 33.3% 27.8% 23.5% 33.2% 29.2% 26.7% Severe underweight 12.4% 14.4% 11.7% 10.7% 14.8% 12.4% 10.9% Severe wasting 10.2% 10.4% 11.2% 10.6% 11.6% 11.0% 9.4% Source: World Bank staff estimates from the 2005 HBS Note: Following the new WHO guidelines, prevalence of stunting and wasting was calculated children between 2 to 5 years old. For underweight, the relevant age group remained children under five years old. 15. Immunization coverage of one year old children varies by geographical location as well as by the type of vaccine. For example, thanks to the aggressive eradication campaign by the Government, nearly 100% of coverage has achieved throughout different expenditure groups. There was no substantial disparity between the urban area and rural area or gender disparity was observed either. 179 Table: Immunization coverage by expenditure quintile (%) National Quintile Quintile Quintile Quintile Quintile Urban Rural average 1 2 3 4 5 Polio 99.1 99.0 99.1 99.0 99.1 99.2 99.0 99.1 DPT3 86.4 84.3 84.1 85.6 88.8 89.0 91.2 84.3 Measles 74.9 70.1 71.9 77.1 77.0 79.2 83.4 72.0 Hepatitis 63.1 56.6 59.3 63.5 65.5 71.9 77.5 58.1 Source: World Bank staff estimates from the 2005 HBS 16. DPT3 had nearly 86% coverage at the national level and the disparity of the coverage between the richest and the poorest was only 5% point. The disparity in the urban and rural was equally small. The governorates of Abyan (60.5%), Sana'a (69.4%), and Remah (66.0%) have recorded lower coverage of DPT3. 17. The percentage of Measles immunization is one of the Millennium Development Goal Indicators. The national average was less than 80% and it appears that Yemen has a long way to go to achieve the target. There was a 9% point disparity between the richest and poorest. The gap between the urban and rural was 12% point, further noticeable. Measles vaccination coverage was much below the average in the governorate of Ibb (68.2%), Hajja (47.0%), Sa'adah (56.2%) and Sana'a (59.5%). 18. Hepatitis vaccine coverage is the most troublesome among the four vaccination types discussed in this section. The national average was only 63%. There were significant gaps in between the richest and the poorest (8% point) and between urban and rural (19% point). The governorates of Hajja (29.6%), Al-Jawf (33.9%), and Sana'a (42.9%) observed extremely low coverages. It appears that the coverage of Measles and Hepatitis vaccinations may be more closely correlated to the geographical location of governorate than the expenditure level of the household where children belong to. C Household Health Expenditure 19. According to the National Health Account Study 2003, during the period of 1999 and 2003, proportion of public expenditure on health remained steady, around 1.5 to 2% of GDP and 4 to 5% of the government expenditure. On the other hand, the proportion of the private source has slightly increased from 57% to 60% between 1998 and 2003. The 2005 HBS data indicated the proportion of out-of-pocket payments for health services against total expenditure have also increased between 1998 and 2005. At the same token, compared with the results of the 1998 HBS, on average, the proportion spent for health as of total household expenditure has slightly increased from 2.3% in 1998 to 2.9%. Based on the 2005 HBS, per capita household expenditure on health was 4,866 Yemeni Rials19 which was equivalent to US$ 25.320. The per capita household expenditure on health presents a very wide range of expenditure level from 752 Yemeni Rials in the poorest quintile to 17,273 Yemeni Rials in the richest quintile (Table XXX). On the other hand, 19 The 2005 HBS questionnaire did not capture health care costs for chronic disease and disability. Thus, these estimates may be underreported. 20 Official change rate in 2005, US$1 = 192 Yemeni Rials 180 the household expenditures on health in the urban and rural area were 7,379 Yemeni Rials and 3,919 Yemeni Rials respectively. This makes the urban and rural ratio, approximately 2:1. 20. The share of the household expenditure on health is progressive among expenditure groups. Households in the poorest quintile spent a lower percentage of their total household budget to health (2.0%) than the richest quintile (4.5%). This progressive spending pattern is observed both in urban and rural households. In urban areas, the budget share on health for the poor and the non-poor are 1.8% and 3.2% respectively, compared to 1.9% and 3.4% in rural areas. 21. In the 1998 HBS, the proportion of household expenditure on health was higher in the rural areas than the urban areas. Between 1998 and 2005, this trend was reversed by a sharp increase of the health share from 2.04% to 3.0% in the urban area than the rural area 2.4% to 2.8%. The analysis in this chapter does not allow us to examine reasons for the sharp increase of the share of household expenditure on health in the urban area. Nonetheless, one possible explanation might be a sharper increase of medical care cost in the urban areas over this time period. 22. Among different categories of medical expenses, medicine and prescription medicine absorb the highest share of total expenditure on health care for the poorest (46.8%). This large share of medicine and prescription drugs is observed among most of expenditure groups, except the richest quintile. In the richest quintile, the proportion is only 19.2%. In the 1998 HBS, a similar category, medicine and medical preparation took up 60 to 70% of the household expenditure on health. Since the comparable proportion was only 26.8% in 2005, it can be concluded that the burden of medicine and prescription drugs cost was somewhat reduced, on average, between 1998 and 2005. The situation of drug supply has long been regarded as unsatisfactory in Yemen. 21 As a large private sector dominates the supply of drugs with expensive prices, anecdotal evidence indicates that the very poor may be able to make drug purchase from the private sector by taking loans or selling personal possessions since the needed drugs are scarcely available from the public source 22 . The category of "other health care expenses 23 " was the second highest share (42.3%) in the poorest quintile. 23. Most noticeable spending pattern of the better off is a significantly high proportion of medical costs sought outside of Yemen. Even in the 1998 HBS, this tendency was 21 Between 1996 and 2002, the Government of Yemen initiated Yemen Drug Action Programme (YEMDAP) and created "Drug Funds" in order to improve availability of medicine of good quality through the public sector. Until the Programme ended in 2002, a marked degree of improvement was indeed achieved. Following the end of "Drug Fund", National Drug Programme was created in the MOPHP to improve availability of drugs in the country. Although both of the programmes were not financially viable or successful, they might have reduced the burden of medicine costs among the household, compared with the time there was no such a programme. 22 The Synthesis Report, The Pharmaceutical Sector in Yemen, Short-term and Long-term Plans for Action (Draft) by Dr. Graham Dukes 23 According to the instruction book of the HBS 2005, it was not quite clear what to be categorized to "other health care expenses." It was possible that some interviewees, in particular, the poorer groups did not bother to disaggregate 181 observed, however, the proportion of household expenditure on health for treatment abroad was much less. The 2005 HBS indicated that the richest quintile spent nearly a half of their household spending on health. This is a significant increased from what was observed in the previous Poverty Assessment (around 10% among the non-poor.) Individuals in other quintiles did not seek medical treatment outside of Yemen. This clearer distinction on the treatment abroad implies that higher proportion of individuals in the richest quintile are willing to pay high cost for quality medical care available outside of Yemen than in 1998. 182 Table A.6. 2: Household Expenditures on Health Care Services Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Rural Non-poor Poor Total Per capita household expenditure on health (YR) 752 1,141 2,021 3,148 17,273 7,379 3,919 7,013 909 4,866 % of household expenditure on health 2.0% 2.0% 2.7% 3.1% 4.5% 3.0% 2.8% 3.4% 2.0% 2.9% Composition of household expenditure Medicine and Prescription drugs 47.1% 48.3% 46.4% 47.4% 18.5% 25.1% 28.1% 25.5% 45.5% 26.8% Doctor's fee 0.9% 3.0% 2.4% 2.0% 1.4% 2.7% 0.9% 1.6% 2.3% 1.7% Hospital stay 2.8% 1.3% 8.4% 9.1% 4.8% 5.8% 5.2% 5.7% 2.3% 5.4% Surgical fees 1.5% 10.9% 9.1% 6.7% 9.3% 9.9% 7.9% 8.9% 7.0% 8.7% Medical services (injections, nurse aid) 0.0% 0.2% 0.8% 0.2% 0.5% 0.8% 0.2% 0.5% 0.2% 0.4% Medical examinations 1.0% 1.4% 1.2% 1.4% 0.8% 1.5% 0.6% 1.0% 1.0% 1.0% Medical supplies (glasses, hearing aids) 0.3% 0.2% 0.3% 2.2% 0.5% 0.8% 0.6% 0.7% 0.3% 0.7% Medical appliances 1.0% 1.0% 1.6% 2.2% 1.5% 2.0% 1.3% 1.6% 0.9% 1.6% Medical paraphernalia (cotton, syringes) 0.0% 0.1% 0.1% 0.1% 0.1% 0.2% 0.0% 0.1% 0.1% 0.1% Midwife and delivery expense 3.1% 1.5% 0.8% 0.9% 0.4% 0.9% 0.5% 0.6% 1.9% 0.7% Medical treatment outside Yemen 0.0% 0.0% 0.0% 0.0% 44.8% 36.9% 28.2% 34.1% 0.0% 31.8% Other healthcare expenses 42.3% 31.9% 28.9% 27.7% 17.2% 13.6% 26.3% 19.8% 38.5% 21.0% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: World Bank staff estimates from the 2005 HBS 183 24. When a household has any disabled or chronically ill member, the household expenditure on health increases nearly by 20 to 40% (Table XXX). This trend is universal throughout the expenditure groups, however, third and forth quintiles observed much higher ratios than the national average. The ratio in the urban area is higher than the one in rural area. This may attribute to higher care cost for the disabled or chronically ill household members. Or the services needed for disable or chronically ill are not readily available in the rural areas. Figure A.6 4: Increases in Household Expenditures on Health Ratios of household expenditure on health (HH with disability /chronically ill vs. average HH) 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Quint ile Quint ile Quint ile Quint ile Quint ile Urban Rural Non- P oor T ot al 1 2 3 4 5 poor 25. On the contrary, households with children under 15 year old spend equivalent or slightly less amount on health, if compared with the spending of the average households. Interestingly, if such household is non-poor or located in the urban area, the household expenditure on health would be reduced by 10 to 25%. 184 Table A.6. 3: Household Expenditures on Health Care Services (Household with members with disability or chronic illness) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Rural Non-poor Poor Total Per capita household expenditure on health (YR) 904 1,849 3,192 4,915 20,544 10,122 4,546 9,372 1,306 6,243 Composition of household expenditure Medicine and Prescription drugs 39.0% 45.6% 42.3% 40.6% 23.5% 24.9% 34.3% 28.8% 39.5% 29.7% Doctor's fee 0.7% 2.9% 2.1% 1.9% 1.3% 2.1% 1.0% 1.5% 2.4% 1.5% Hospital stay 1.8% 1.5% 7.6% 13.2% 7.7% 6.9% 8.9% 8.4% 2.1% 7.9% Surgical fees 0.6% 12.0% 9.7% 8.9% 12.5% 9.6% 12.9% 11.7% 6.4% 11.3% Medical services (injections, nurse aid) 0.0% 0.3% 1.1% 0.2% 0.5% 0.8% 0.3% 0.5% 0.3% 0.5% Medical examinations 0.9% 1.4% 0.9% 1.4% 1.1% 1.6% 0.6% 1.2% 0.9% 1.1% Medical supplies (glasses, hearing aids) 0.5% 0.3% 0.3% 3.6% 0.7% 0.7% 1.3% 1.1% 0.4% 1.0% Medical appliances 1.2% 1.1% 1.2% 2.1% 1.9% 2.0% 1.6% 1.9% 0.7% 1.8% Medical paraphernalia (cotton, syringes) 0.0% 0.1% 0.1% 0.1% 0.1% 0.2% 0.0% 0.1% 0.1% 0.1% Midwife and delivery expense 5.3% 0.7% 0.5% 0.5% 0.2% 0.3% 0.6% 0.3% 2.3% 0.5% Medical treatment outside Yemen 0.0% 0.0% 0.0% 0.0% 29.6% 34.4% 5.2% 21.4% 0.0% 19.6% Other healthcare expenses 49.9% 34.1% 34.4% 27.5% 20.9% 16.5% 33.1% 23.1% 44.9% 24.9% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: World Bank staff estimates from the 2005 HBS 185 Table A.6. 4: Household Expenditures on Health Care Services (Household with children under 15 year old ) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Urban Rural Non-poor Poor Total Per capita household expenditure on health (YR) 758 1,137 2,048 3,212 16,107 5,626 3,895 6,305 907 4,352 Composition of household expenditure Medicine and Prescription drugs 46.8% 47.7% 45.9% 47.3% 19.1% 31.0% 27.0% 27.1% 44.6% 28.4% Doctor's fee 0.8% 3.1% 2.5% 2.0% 1.5% 3.3% 1.0% 1.7% 2.4% 1.8% Hospital stay 2.8% 1.3% 8.9% 9.4% 5.1% 6.8% 5.3% 6.1% 2.3% 5.8% Surgical fees 1.5% 11.3% 9.5% 6.9% 9.7% 11.7% 7.7% 9.2% 7.2% 9.1% Medical services (injections, nurse aid) 0.0% 0.1% 0.8% 0.2% 0.6% 1.0% 0.2% 0.5% 0.1% 0.5% Medical examinations 0.9% 1.5% 1.0% 1.4% 0.8% 1.7% 0.6% 1.0% 0.9% 1.0% Medical supplies (glasses, hearing aids) 0.3% 0.2% 0.3% 2.4% 0.6% 1.1% 0.7% 0.8% 0.3% 0.8% Medical appliances 1.0% 1.0% 1.7% 2.0% 1.6% 2.2% 1.3% 1.6% 0.9% 1.6% Medical paraphernalia (cotton, syringes) 0.0% 0.1% 0.1% 0.1% 0.1% 0.3% 0.0% 0.1% 0.1% 0.1% Midwife and delivery expense 3.1% 1.6% 0.8% 1.0% 0.5% 1.3% 0.5% 0.7% 2.0% 0.8% Medical treatment outside Yemen 0.0% 0.0% 0.0% 0.0% 41.9% 24.2% 30.2% 30.4% 0.0% 28.1% Other healthcare expenses 42.6% 32.1% 28.6% 27.5% 18.5% 15.5% 25.5% 20.7% 39.2% 22.1% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Source: World Bank staff estimates from the 2005 HBS 186 Benefit Incidence of the Government Health Service Provisions 26. In general, a benefit incidence analysis looks into (i) utilization patters of government's health services by expenditure groups and (ii) distribution of subsidy given to an individual in different expenditure groups who utilizes public health services. However, the 2005 HBS questionnaire did not capture the number of per capita visit and household health expenditure by different type of health facilities. Without these basic data, it was not possible to compute the patterns in utilization of public health facilities or distribution of health subsidies. In this chapter, the benefit incidence analysis is limited only to a question, if individuals in different expenditure groups seek medical care in public health facilities in case of illness and analyze the reasons for their preference. 27. According to the 2005 HBS, individuals in the poorest quintile do not necessarily receive their medical care from the public health facilities. In the poorest quintile, more individuals sought the care at the private health facilities, which include private clinic, private hospital, private doctor consultation, and private consultation with a certified health professional. Among the private facilities, private clinic visit represent 33% of the surveyed individuals in the poorest quintile. This percentage was outstandingly higher than the rest of expenditure groups. The percentages between the public facilities and private ones are slightly changed in the second poorest income quintile, however, in sum, this graph indicates that the government's health system does not effectively outreach to the poor. There is no clear evidence that the poor is a particular beneficiary of the government's health service provision. Care seeking behavior of individuals by expenditure quintile (% of individuals) 60% 50.1% 50% 4 6.3 % 4 5.6 % 43 .7% 3 9 .63 9.3 % % 40% 3 2 .6 % 3 0 .6 % 3 0 .0 % 2 8 .1% 30% 2 6 .2 % 2 3 .0 % 2 1.8 % 2 1.1% 2 1.6 % 20% 10% 0.0% 0 .0% 0.0 % 0.0% 0 .2% 0% Quint ile 1 Quint ile 2 Quint ile 3 Quint ile 4 Quint ile 5 Public healt h facilit ies Privat e healt h facilit ies Ot hers Out of Yemen Note: Public health facilities include public health center and public hospital. Private health facilities include private clinic, private hospital, private doctor consultation, private consultation with a certified health professional. Other includes pharmacy and traditional medicine. If we can desegregate pharmacy to public or private one, the percentage of the private facilities will go up further. 187 28. This finding is also endorsed by the qualitative research study mentioned earlier. In case of serious illness, the focus group participants showed their strong preference for seeking better quality health services at private health facilities rather than public ones. At the same time, the participants noted that the private treatment fees were so expensive that they could not afford the treatment easily. Table XX summaries the perceived benefits of visiting public health facilities versus private health facilities based on the focus group discussions. In public hospitals, patients have to pay only 40 Yemeni Rials24, however they have to pay additional fees for medicines and examinations. This could possibly add up to a considerable financial burden even though they seek care in public hospitals. Focus group discussion also indicated that cases of medical doctors from public health facilities have their own private practices for making additional incomes. Thus, this leads to short operating hours of public health facilities and absenteeism of doctors or health workers. Table: Perceived Benefits of Visiting Public or Private Health Facilities Public health facilities Private health facilities Reasons for visiting · Shorter travel distance · More available · Cheaper · More professionals/specialists · Drugs are cheaper · Shorter waiting times · Longer opening hours · Tests/exams are available · Better equipment Reasons for not visiting · Too many exams · Expensive · Long waiting time · Too far away · Unprofessional staff attitudes · No specialist available · Health unit close early · No female doctors in health units · No drugs and materials · Too many medical exams Source: Yoshimi Nishino, Qualitative Assessment of Community Based Health Related Programs: Five Programs and Six Locations in Yemen 29. The following figures present the relation between per capita public health expenditure and some of the measure health outcomes measured in the 2005 HBS. The per capita public health expenditure by governorate was a yearly average of the MOPHP and governorate expenditure on health between 1999 and 2003, taken from the NHA 2003 Study25. First of all, there is a large range in the per capita public health expenditure by governorate. This indicates that there might be lack of equity consideration in allocating the public health budget. Secondly, according to these figures, there seems to be lose correlations between the level of per capita public health expenditure and the level of measles immunization coverage or prevalence of severe underweight. However, at the same time, there are huge differences in the level of the immunization coverage or 24 The different report shows a much higher co-payment at the hospital. This could be due to the fact that in both public and private health facilities, the pricing for health services were at ad-hoc basis and not transparent. 25 This was the most recent per capita public health expenditure available for this type of analysis. 188 prevalence of severe underweight among the governorates with comparable per capita public health expenditure. This implies that there exist huge gaps in the planning and implementation capacity of basic health services among governorates. Relation between per capita public health expenditure and immunization coverage by governorate 100 90 easles im unization coverage, 2006 (%) 80 70 60 m 50 40 30 y = 8.8627Ln(x) + 16.364 20 2 R = 0.1213 M 10 0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 per capita public health expenditure (yearly average 1999-2003, YR) Source: The World Bank estimates from the 2005 HBS, National Health Accounts Study 2003 Relation between per capita public health expenditure and prevalence of severe underweight 35.0 -0.0003x 30.0 y = 14.714e eight, 2006 (%) Prevalence of severe 2 25.0 R = 0.1548 20.0 15.0 underw 10.0 5.0 0.0 - 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Per capita public health expenditure, 1999-2003 (yearly average, YR) Source: The World Bank estimates from the 2005 HBS, National Health Accounts Study 2003 D Conclusions and Ways Forward 30. Based on the analysis from the 2005 HBS, it appears that the access to health facilities has improved over the course of the years. Nearly 70% of household has at least one health facility in the same area or district in 2005. The percentage of people who seek medical care in the case of illness both have increased. However, it must be noted that the gap between the richest group and the poorest group has widened. The access to health facility has improved more favor for the rich. As identified in the analysis, the poor find it difficult to access to health care in case of illness, due to their inability to pay for medical 189 services. In particular, increased share of household expenditure on health presents an alarming concern. The poor might be exposed to greater risks of deeper impoverishment in case of serious illness than before. Or they are less likely to seek care if they remain unable to pay for the cares. Given the analysis in this section, it appears that there has been slight improvement in access to health care, in particular physical access to a health facility, since the Health Sector Reform was initiated in 1998. Nonetheless, there still exist considerable challenges in improving in equity and efficiency in providing health care services. Additionally, the stewardship or leadership role of the MOPHP remains weak. The health section recommends three options for further consideration as below. 31. First, as the previous Poverty Assessment and the Bank's analysis of health inequality suggested, the Government of Yemen need to divert more of their resources to health sector and then apply more rigorous targeting to the poor segments of the society. This would be a policy option in a short-term. Currently, the government health program is still under funded with only 4.9%26 of the government expenditure allocated to health. It would be advisable to increase the level of government expenditure to health close to some 10%. Provided that the level of the government funding is increased, efficiency of the government's health service provision must be strengthened and monitored at the same token. In the light of targeting the poor more aggressively, the current trend of decentralizing health care service provision would be beneficial and can encourage community level interventions. On the other hand, additional efforts need to be made to further streamline the planning and budgeting process and provision of basic health services, which often is one of the major bottle necks in the decentralization process. Continuous capacity building trainings would be pivotal for managers and health workers in the lower levels of the system. The disparity of geographical resource allocation and lack of rational methodology still persists. It is hoped that the analysis in this section provides a snapshot view of which segments of the society or which governorate needs priority interventions, especially in relation to improvement of assisted delivery, immunization and child malnutrition of which the 2005 HBS's questionnaire captured data. 32. Secondarily, in addition to strengthening the capacity of the MOHPH, it would be worthwhile examining efficacy of partnering with community-based organizations (CBOs) or non-governmental organizations (NGOs) for provisions of basic health care services. CBOs or NGOs are generally believed to effectively reach the poor or disadvantages groups of the society. 33. In regard to alleviating the financial barrier for the poor, one may argue that one option would be to introduce a national health insurance scheme. The Government of Yemen has been exploring possibilities of introducing a national health insurance program in the country with assistance from bilateral or multilateral donor agencies. The current version of Poverty Reduction Strategy Report and Five Year Plan support the idea of introducing the national health insurance plan. However, in reality, the country has experienced political and administrative setbacks in passing Health and Work Insurance 26 Yemen National Health Accounts: Estimate for 2003, National Health Accounts Team, Republic of Yemen, Partners for Health Reformplus, June 2006 190 Law and Police Health Insurance Scheme and has not received sufficient support from the Cabinet and other parts of the society27. As a pilot basis, the Government of Yemen is trying to introduce a community-based health insurance in Al-Shamayatayn in the governorate of Taiz28. The initial results of the pilot scheme present that the community- based health insurance scheme may face a series of institutional constraints and challenges before it is becomes operational and is scaled-up to a larger population in context of Yemen. At this point of time, it appears that a cascaded approach, such as the short cycle of a formulation of framework, pilot scheme implementation, evaluation and adjustment of the framework will be more beneficial for Yemen at this point as the country still need to demonstrate a showcase to buy-in the interests and endorsement from relevant stakeholders. Provision of health insurance would be very important to remove the financial barriers to access health care among the poor, however, that will require a careful designing of the scheme and consensus building among the policy makers. 27 Military Health Insurance Law was approved that requires contribution rates of 3% for soldiers and 5% for civilian officers. 28 GTZ, WHO, World Bank and ILO, Towards a national health insurance in Yemen 191 ANNEX 7: EDUCATION 1. Poverty rates are the highest for households headed by an illiterate person; the poverty rate decreased since 1998, but still remains large. According to the 1998 HBS, the poverty rate for households headed by an illiterate person was 47.3% nationally, 48.8% in rural areas, 39.9% in urban areas. In 2005 HBS, these declined to 44%, 47%, and 34% respectively. The lowest poverty rate was found among the household headed by a person with university and above education although the poverty rate for urban and rural areas are very large: 5% to 29%. 60% Urban 50% Rural 40% Total 30% 20% 10% 0% Primary Total Illiterate Pre-Sec. Preparatory read and and above Unknown read only Secondary Secondary University Vocational Institute Pre-Sec. Diploma write Basic/ Post No formal Formal schooling schooling 2. 68.3% of the poor had no formal education. Among the all the poor households, 49% of them are headed by the illiterate households, and cumulative share of households with household heads who has no formal education is 68.3%. The share decreased since 1998 when the cumulative share of this category was 86.7%. Table 7.1: Educational attainment for the poor and non-poor by urban-rural status Non-poor Poor Urban Rural Total Urban Rural Total No formal education Total 35.9 58.2 50.7 59.9 69.9 68.3 No formal schooling Illiterate 21.6 39.3 33.4 43.5 50.1 49.1 read only 2.3 4.7 3.9 3.2 5.4 5.1 read and write 12.0 14.2 13.4 13.3 14.4 14.2 Formal education Total 61.8 40.3 47.6 37.2 27.4 29.0 Formal schooling Primary 14.2 13.1 13.5 14.9 9.1 10.0 Basic/ Preparatory 11.9 9.7 10.5 8.7 8.3 8.4 Pre-Sec. Vocational 0.9 0.6 0.7 0.5 0.3 0.3 Pre-Sec. Institute 0.7 1.7 1.4 0.5 0.8 0.7 Secondary 14.0 9.6 11.1 7.6 6.1 6.3 Post Secondary Diploma 3.6 2.2 2.7 1.8 0.9 1.0 University and above 16.5 3.5 7.8 3.3 2.0 2.2 Unknown 2.3 1.5 1.8 2.9 2.6 2.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 192 3. The situation of enrollment among age 6-14 children improved overall with drastic pick up on rural girls' enrollment. The overall enrollment rate of age 6-14 children increased from 60% in 1998 to 66% in 2005. This is achieved especially by an increase in the enrollment of rural girls. While the enrollment rate of boy increased from 75% to 76% between these two surveys, girls' enrollment increased from 43% to 56%. In the urban-rural perspective, while the urban enrollment rate remained at 81%, rural enrollment increased from 54% to 62%. This enrollment rate is not a gross enrollment rate (GER) nor a net enrollment rate (NER), but it is similar to NER29. It is important to note that the NER as calculated in MOE's statistics is slightly higher than what is discovered in the HBS. This difference implies that there are students who are officially registered in the schools, but they don't feel they are enrolled because they are not regularly attending school or they decided not to go to school anymore during the school year30. 4. Although the overall enrollment rate increased, the gap in enrollment rate between the poorest and the richest deciles are growing. When the enrollment rate is calculated for each income decile, it is discovered that the enrollment rate for the poorest decile decreased and the richest declie increased. In 2005, 50 percent of the children aged 6-14 have access to school, which is down by 6 percentage point lower than HBS 1998 (56%). On the other hand the 82 percent of the richest children have access to school, which is up by 15 percentage point from HBS 1998 (67%). The trends in the urban-rural dimension gives a detailed description for this trend. While the enrollment rates in the poorest decile declined for both urban and rural areas by 5 percentage points each, the richest deciles increased enrollment by 1 percentage point in urban areas and by 21 percentage points in rural areas. From gender dimension, again the enrollment rate for the poorest decile declined for boys (by 6 percentage points) and girls (by 4 percentage points) whereas the enrollment rate of the richest decile increased for boys (by 3 percentage points) and girls (by 26 percentage points). From these patterns, the overall enrollment gain mainly occurred by the increase in enrollment of girls in rural rich households. As a number of schools are constructed in rural areas, rural population have gained more access to school. However, those who can go to school in the rural households are mainly the richest households. 5. The reasons for not enrolling in school is widely unknown. The statistics shows reasons for never enrolled and reasons for dropping out are 99% unknown for age 6-11. Although reasons for dropping out from school for age 12-14 children are reported, the responding behavior is too different from other questions. (It does not make sense that for age 12-14 children, reasons for never attending school shows 99% not-stated, and reasons for dropping out shows 9% not-stated.) The data quality must be re-checked. 29 It is not GER because it doesn't include enrollment of other ages (below 6 and above 14); it is not NER for basic education because some of the students registered as enrolled are in secondary schools. 30 Another possible explanation is data error in either or both datasets. It is common to have different enrollment rates depending on the data source. 193 Figure 7.2: Age 6-14 Enrollment Rate by Income Decile, Gender, and Urban-Rural Status 100 Age 6-14 Enrollment Rate (%) 90 80 70 60 50 40 Male urban 30 Male rural 20 Female urban 10 Female rural 0 1 2 3 4 5 6 7 8 9 10 Income Decile 6. The gap in the enrollment rate by governorate still exists largely especially for girls. Figure XX shows the enrollment rate of age 6-14 children by gender (Note this is not a basic education NER because it includes non-basic education students). Compared to NPS 1999, the gap between the highest and lowest girls' enrollment rate declined from 67 percentage points to 52 percentage points (Sana'a City 84%, Saadah 32%). Yet, disadvantage of certain governorates in terms of enrollment rate still persists. Boys' enrollment gap by governorate on the other hand is not as large, but it is still 29 percentage points (Sana'a City 87%, Al-Hodeidah 58%) 194 Figure 7.3: Age 6-14 Enrollment Rate by Governorate and Gender 100 Male 90 Age 6-14 Enrollment Rate (%) Female 80 70 60 50 40 30 20 10 0 Al-Jowf Ibb Sana'a City Taiz Dhamar Aden Al-Mahra Al-Daleh Abyan Shabwah Mareb Amran Ryma Al-Baida Al-Hodeidah Saadah Lahej Hadramout Al-Mahweet Sana'a Hajja Source: World Bank staff estimates based on 2005 HBS Note: It is not an NER for Basic Education as it includes non-basic education students. 7. Illiteracy rates for age 15-24 population has decreased overall due to significant decrease in illiteracy among rural females; however, urban illiteracy rate and male illiteracy rate have gone up. Illiteracy rate among age 15-24 population decreased from 34.8% to 28.5% between 1998 and 2005 due largely to the decrease in the rural female illiteracy from 73.0% to 49.9%. However, male illiteracy rate has gone up both in urban (from 3.8% to 9.4%) and rural areas (from 14.9% to 19.1%). 195 Figure 7.4: Age 15-24 Illiteracy Rate by Income Decile, Gender, and Urban-Rural Status 80.0 70.0 60.0 Illiteracy Rate (%) 50.0 Male urban Male rural 40.0 Female urban 30.0 Female rural 20.0 10.0 0.0 1 2 3 4 5 6 7 8 9 10 Source: World Bank staff estimates based on 2005 HBS 196 ANNEX 8: IS PUBLIC EXPENDITURE TARGETING IN YEMEN PRO-POOR? I IS PUBLIC EXPENDITURE TARGETING IN YEMEN PRO-POOR? ABSTRACT 1. Since 2000, Yemen has initiated an ambitious program of decentralization to strengthen roots of democracy. In this context, fears have been expressed that devolution of spending power could aggravate problems of fiscal management. However, decentralized authority in spending could help better target the poor in delivering public services. This paper is an attempt at examining targeting efficiency of aggregate public expenditure at decentralized levels. At the current stage of fiscal decentralization in Yemen, central government still makes budgetary decisions but only the execution is carried out locally. Examining targeting efficiency at decentralized levels requires data on expenditures and poverty at the decentralized level. Though expenditures at decentralized levels are readily available, reliable estimates of poverty are not. In the absence of a household survey designed to provide poverty estimates at sufficiently decentralized levels, the paper uses a method suggested by Ravallion (2000) to identify the latent differences in mean program allocations to the poor and non-poor using unmet basic needs measure of poverty constructed from census data (thus representative at any geographic level) . 2. The analysis finds evidence of anti-poor bias in public expenditure allocations. The poor receive YR 6290 (US $33) less than the non-poor in per-capita terms which is almost the same as the per-capita public expenditure on all Yemenis. Most of the anti-poor bias occurs at the level of governorates. At the governorate level, the poor get YR10,060 (US $52) per-capita less than the non-poor. Intra-governorate allocation of expenditures across districts offsets some of the anti-poor bias observed at the governorate level. Four of the 20 governorates (Taiz, Al- Hodeidah, Ibb and Hajjaah), accounting for about 40 percent of Yemen's population, allocate more to the poor than the non-poor, or, allocate equal amount. The worst five governorates in this respect are Shabwah, Amran, Sanaa, Lahj and Hadramout. To the extent that the observed inequity in governorate level expenditure allocation captures more than the higher cost of provision of services in inaccessible regions, Yemen could improve targeting efficiency of public expenditures. 197 II BACKGROUND 3. Yemen, with its 20 million people and gross national income per-capita of US$ 510, is among the poorest countries in MENA. The incidence of poverty in Yemen was 41.8 percent at the national level, 45.0 percent in rural areas, and 30.8 percent in urban areas31. 4. The country faces major constraints to sustainable and employment-generating growth as well as good public service delivery that stand in the way of rapid improvements in the quality of life of the poor32. The Government of Yemen is clearly committed to poverty reduction and has prepared a national poverty strategy that aims to reach the Millennium Development Goals (MDGs) by 2015. As a result of recent policy efforts, social indicators have gradually improved, but they still rank with some sub-Saharan African and South Asian countries. 5. Yemen faces the difficult challenge of ensuring the benefits of growth are distributed across poorer sections of society and improving the quality of public service delivery to the poor, particularly in rural areas. The recent poverty update for the country has concluded that though poverty is widespread nationally, it is pervasive in rural areas and concentrated in a few governorates. Public expenditures in education and health sectors were mildly pro-poor but did not address the magnitude of rural-urban and gender gaps. Almost all social programs were urban-biased and tended to benefit the non-poor. Benefit-incidence analysis of the social safety nets showed that their coverage was extremely limited, failed to address short-term downturns and vulnerability for the able-bodied poor and did not reach the poorest and most needy, especially children33. In this context, the effective targeting of public expenditures to the poor assumes critical importance. 31 The figures refer to poverty estimates for 1998 reported in World Bank Poverty Update (2002a). 32 See World Bank (2002b) ­ World Bank's most recent Country Assistance Strategy (CAS). 33 Programs under the second phase of the Social Development Fund (SFD) were found to be pro-poor, but the inter-governorate distribution of both Public Work Programs(PWP) and Social Welfare Fund (SWF) allocations showed no signs of pro-poor targeting. 198 Box A.8. 1: Decentralization and the Promise of Equity Yemen launched one of the most ambitious decentralization programs in the Middle East and North Africa when parliament passed the Local Authority Law in 2000. The law provides a mechanism for formalizing traditional democratic practices that have served Yemeni society well and can also help mitigate the trend in power concentration among a handful of shaykhs. Effective in May 2002, local councilors serve a five-year term; they represent 332 districts in 22 governorates. Teachers account for nearly 40 percent of those elected to district councils and shaykhs and civil servants each account for roughly 7 percent of office holders.1 The overwhelming number of teachers among the ranks of council members suggests the electorate votes on the basis of perceived qualifications, rather than social status. Decentralized governance presents a potential for improved public delivery systems, is anchored in traditional systems of governance and has an important precedent in the popular Local Development Association movement that was active in the highlands during the 70s. The LDAs represented popularly elected councils established throughout the YAR. They collected funds and in-kind support from residents, non-resident migrant laborers, external donors and built schools, water projects, roads, health facilities with much greater reach than the state. Much of their funding came from locally collected zakat payments (5 percent of net resources calculated annually). The LDAs lost their earlier community based character and weakened as central control increased and zakat funds were transferred to the central government Comparison of State and LDA Sponsored Development Projects1 Development Projects Development Projects Development Projects Rural Projects, 1976 Rural Projects, 1976 Rural Projects, 1976 Roads, schools, water, 1981 Roads, schools, water, 1981 Roads, schools, water, 1981 Projects, 1986 Projects, 1986 Projects, 1986 6. Successful implementation of national poverty alleviation strategies is contingent on the effectiveness of regional governments. Even when a national government manages to redistribute its resources to relatively poor provinces, the capacity of regional governments to target the poor is an important factor. The outcome will depend on the behavior of provincial governments and the political economy they are confronted with, which will differ in relevant ways. Some governments will care more about the poor than others or will face different constraints in their efforts to reach the poor; indeed, simply having a high incidence of poverty can result in worse targeting by a local government as poorer provinces usually lack information about the identity of the poor (Ravallion 1999a). Not all regional governments are able to exploit the local information that decentralized decision making affords, while some will be better able to secure the gains. 7. In an effort to reform the public administration system, the Yemeni parliament approved the Local Authority Law (LAL) in 2000. LAL envisages local governance as one of the pillars of the state and provides the legal foundations for the Yemeni inter-governmental system. This shift from a centralized system that is being attempted, with ex ante controls, to a more decentralized system, with an emphasis on ex post monitoring, presents challenges as well as opportunities to fight poverty and improve the quality of public services. In this context, it is useful to understand the structure of inter-governmental fiscal system and the targeting efficiency of various levels of government and individual governorates. Such understanding would help better design incentive structures and exploit the opportunities offered by decentralization reforms in the near future for a stronger impact on poverty. 199 8. The potential benefits of fiscal decentralization in Yemen are considerable. 34 In a decentralized system, where locally elected governments have the power to pursue the agenda mandated by voters, citizen participation in decision-making processes cultivates a culture of democracy and transparency in public management system. Decentralized service delivery has the potential of reaching vulnerable groups and therefore reducing poverty. In order to address poverty and quality of service delivery, education and health are of vital importance. While social indicators in Yemen are improving, they still remain at low levels35. For instance, only 33 percent of rural girls are enrolled in school compared to 77 percent of rural boys and 78 percent of urban girls. 36 Improvement of rural girls enrollment is best addressed by local authorities ­ given that this is dependent not only on provision of schools but provision of sanitation as well. Rather than depending on the central ministries to coordinate these services together, sub-national governments are better able to identify these problems and the inter- linked services that are required to address these problems. Box 1 below explains the current status of fiscal decentralization in Yemen. 37,38,39 9. The rest of this paper is organized in three sections. Section 2 presents the methodology used in this paper and derives from Ravallion's earlier work and provides the details of the decomposition of the national targeting differential into between and within governorate components. Section 3 describes the database created and used for the current analysis and explains the construction of the Unmet Basic Needs (UBN) Index that is used as a proxy to a 34 Allen et al 2005 states a variety of reasons for which decentralization holds the promise of poverty alleviation and improved service delivery ­ a) Poverty is largely a rural phenomenon (in 1998 ­ half the rural population was poor compared to a third of urban population. 83 percent of Yemen's poor live in rural areas. b) The level of education has a strong correlation with poverty incidence, depth and severity. 87 percent of the poor are either illiterate or did not complete primary school. c) Geographic location significantly affects the risk of being poor. d) Children and women living in rural areas without access to education and health services rank highest among those people vulnerable to poverty. e) Almost all social programs are urban-biased and tend to benefit the better off. 35 The case for fiscal decentralization in Yemen draws upon Allen et al, 2005. 36 The rural-urban divide holds in other social sectors as well ­ While 80 percent of the urban population has access to health care services, only 25 percent of rural population is provided with health care. About 33 percent of the rural population has access to safe drinking water compared to 87 percent in urban areas. 37 This box is based on Allen et al, 2005. We gratefully acknowledge the comments and clarifications provided by Richard Allen, Lead Public Sector Specialist and Monali Chowdhurie-Aziz, Senior Public Sector Specialist, MNSED, World Bank. 38 In contrast to devolution , which is a transfer of authority for decision-making, finance and management to quasi-autonomous units of local government, deconcentration is a redistribution of decision making authority among different levels of the central government and is often considered the weakest form of decentralization and is used most frequently in unitary states. Within this structure, however, policies and opportunities for local input vary: deconcentration can merely shift responsibilities from central government officials in the capital city to those working in regions, provinces or districts, or it can create strong field administration or local administrative capacity under the supervision of central government ministries. 39 The deconcentrated units are regarded as local organs and act as the administrative, technical, and executive organs of the local council and operate under the councils' supervision. 200 district level poverty index. Finally, section 4 concludes with the estimates of various targeting differentials and findings of the analysis. Box A.8. 2: The Status of Fiscal Decentralization and Sub-National Expenditures in Yemen The current Yemeni public administration system can be characterized as a from of deconcentration rather than one of devolved local self-government. Nevertheless, the system remains highly centralized, albeit equipped with an elaborate system of deconcentrated field offices of line agencies and democratically elected local councils. At this time, budget decisions for the most part are made by the central government and the role of sub-national authorities is largely confined to executing them and enjoy very limited fiscal autonomy. The central government in Yemen is the senior partner in the intergovernmental relationship. The share of subnational government spending in Yemen compares favorably with other countries. However, this comparison is misleading because, although a large share of the expenditures is disbursed through subnational government, they have little decision-making power over current expenditures. Subnational government capital expenditures, in which they have a significant degree of autonomy, represents less than 1 percent of the GDP, which is very low by international standards. Table 1 below shows the share of subnational expenditures in GDP in Yemen. Sub-national expenditures in Yemen amounted for about 6.4 percent of GDP in 2004. Current expenditure is the largest item in local budgets. At 80 percent of current expenditures, and 70 percent of total expenditures, wages and salaries consume most of the budget 2002* 2003** 2004*** Current Capital Total Current Capital Total Current Expenditures 5.94 0.76 6.70 Expenditures 5.94 0.76 6.70 Expenditures 5.94 The Yemeni public administration system is divided into three levels: the center and sub national units--governorates and local districts. Following the LAL, Yemen was divided into local administrative units that include the Sana'a council, the governorates and the districts. As of today, there are 22 governorates and 332 districts both of which are called administrative units. Each administrative unit has its own local authority, which consists of the centrally appointed administrative head of the unit (either the governor at governorate level or the director at the district level), the elected local council at both governorate and district levels, and the "executive organs" (branch offices of the ministries and other government agencies). As with the budget of central ministries, the MOF's budget circular provides general instructions to sub-national governments about their allocations using a process of incremental budgeting with a de-facto ceiling for wages and salaries, which are, by far, the largest share of current expenditures. These instructions are restrictive in that recurrent expenditures are given no flexibility to account for shifting sub-national government priorities. For example, they do not leave room for essential operation and maintenance (O&M) expenditures that has been historically under-budgeted. 201 III METHODOLOGY 10. Monitoring the performance of sub-national governments can provide the information base for the national government to design an incentive structure that encourages more equitable outcomes on poverty and provision of public services. However, the household level data necessary to examine the incidence and targeting effectiveness of public expenditures is not often available40. Ravallion (2000) addresses this problem by suggesting a method that allows an assessment of the degree to which spending tends to be targeted to the poor on an average. Targeting performance can be measured by exploiting the spatial variances in both public spending and poverty incidence across geographic areas. 11. The inter-regional targeting differential is estimated by regressing expenditure allocations across regions on the regional poverty measure. If a program is effectively reaching the poor, with little leakage to the non-poor, then the overall expenditure allocation across geographic areas will be highly correlated with the poverty rates across the same areas. Following Ravallion (2000), this property can be used to devise a measure of how well program allocations match the spatial poverty map in the form of an estimated mean difference in spending between the poor and non-poor. This national measure of targeting performance can also be decomposed into subgroups ­ between-region and within-region components - and thus help policymakers understand the sources of national targeting failures - between regions and within regions and further identifying the relative contribution of different provinces to the national targeting failure. Ravallion (2000) applied the method to assess Argentina's anti- poverty program's performance before and after reforms. Van de Walle (2005) has performed a similar analysis for Morocco on the basis of a provincial level database. 12. This paper applies the decomposition technique in the context of Yemen's public expenditure against the poverty map at the district level. The paper examines the distribution of spending across districts of Yemen and how well the poor are reached by public expenditure. This technique can be further extended to distinguish between differences in mean spending targeted to urban and rural areas as well as between North and South Yemen if one can demarcate these categories for all districts. 13. The equations to be estimated are: (1) to estimate inter-district or national targeting differential: Gij ­ G = TD (Hij ­ H) + Vij (2) to estimate inter-governorate targeting differential: Gj ­ G = TP(Hj ­ H) 40 See Alderman (2002) for an analysis of distributional and targeting outcomes of social expenditures in Albania on the basis of a household survey. 202 (3) to estimate intra-governorate targeting differentials (one equation for each governorate) Gij ­ Gj = Tj(Hij ­ Hj) + Vij Where Gij = percapita allocation to district i in governorate j Gj = percapita allocation to governorate j G = national percapita allocation Hij = head count ratio in districti i in governorate j Hj = head count ratio in governorate j H = national headcount ratio Vij = error term Tj = absolute difference between the average allocation to the poor and the the average allocation to the nonpoor in governorate j. Tj is also referred to as the intra-governorate targeting differential for governorate j. P T = Inter-governorate targeting differential TD = national targeting differential A Decomposing the National Targeting Differential 14. We can estimate a national (inter-district) targeting differential, TD, by regressing the values of Gij on Hij across all districts, irrespective of their governorate. The OLS estimate of the national targeting differential can be decomposed exactly into between- governorate and within-governorate components: ^ ^ ^ T D = S P T P + S j T j 15. Where SP is the between-governorate share of the total (inter-district) variance in poverty rates, and Sj is the governorate-specific share. The first term on the right side of the equation is the "between-provinces" component, and the second term is the "within- province" component. Annex 1 provides the details including the calculation of the respective weights. 41 IV DATA ISSUES 16. This method requires a disaggregated poverty map that predates and corresponds to expenditure disbursements for the same disaggregated geographic units. Since the available household budget survey data precludes analysis of public expenditure on poverty at the district 41 All the public expenditure data used in the paper refer to fiscal allocations and not actual expenditures. 203 level, we infer expenditure incidence on poverty indirectly by juxtaposing the geographic distribution of public spending and the corresponding poverty map based on the Unmet Basic Needs (UBN) Index. 17. The empirical analysis draws on budget data from the Ministry of Finance (AFMIS Project Unit) for 200442. Since further disaggregated data is not available on the expenditure side, the UBN index has been constructed using information from 1994 census data for 2126 sub-districts. From this data, a district level database (with 289 districts in all distributed across 20 governorates)43 was created44. A concordance between the expenditure data for 2004 and UBN index for 1994 was constructed. 45 The population data used to obtain per-capita allocations was from 1994 census instead of 2004 census since a usable mapping of districts of the two censuses was not available46. The district classifications for 2004 budget data and the 2004 population census districts is not the same as they were created by different government agencies which were perhaps not coordinated. B Construction of the UBN Index 18. The analysis requires data on disbursements by local government area and a corresponding poverty map. There are 289 districts in Yemen administered under 20 governorates. We do not have head count ratios (of consumption poverty) of districts based on household surveys that are designed to be representative at the district level. Hence we resort to 42 Under the Civil Service Modernization Project being financed by a credit of $11.3 million from the World Bank, the Ministry of Finance has embarked on a project to design and implement and Accounting and Financial Management Information System (AFMIS). The AFMIS is expected to provide the full range of functionalities for budget preparation, execution, accounting and financial reporting. This is a tool and its effectiveness is dependent on a clear and coherent strategy for budget reform and fiscal decentralization. See, Allen et al.2005. 43 However, the regressions were based on 287 districts only since expenditure data for two districts was missing. The districts with missing budget data are Attur in Hajja governorate and Khawlann in Sanaa governorate. 44 Ideally, one would prefer to use a poverty index closer in time to the expenditure data as the assumption is that the current spending allocations are determined by the most recent information available regarding the poverty status of geographic areas. However, the Republic of Yemen came into existence as recently as 1991 and 1994 is the only year for which census data is available. The poverty estimates based on the household budget survey of 1998 can not be used since they are not representative at the district level. 45 There have been several reclassifications and reassignment of territory in the intermittent period 1994-04. The district level concordance has been constructed after getting the original Arabic data files translated with some assistance from the Department of Statistics. However, for some governorates like Sanaa city, a near-perfect concordance was created, for several other governorates, the mapping may not be perfect. In cases, where a district that retained its name over the decade, may have lost territory in which case the expenditure is overstated. On the other hand, for districts that have gained territory expenditure would be understated. 46 It is however, not possible to construct an exact concordance between the country classification in 1994 and 2004 without further assistance from GOY ­ There were 20 governorates and 289 districts in 1994 compared to 22 governorates and 332 districts in 2004. 204 a poverty measure that is possible to construct at the district level - the proportion of households with unmet basic needs (UBN), based on the 1994 census. Figure A.8. 1: UBN Index by Governorate Source: Staff estimates based on Republic of Yemen Census, 1994. Note: The numbers indicate the percentage of population that does not have basic needs met. 19. The UBN index is constructed as a composite of housing quality, access to safe water, infant mortality, and educational attainment - literacy (of adults), school enrollment (of children). As opposed to a consumption measure of poverty, the UBN index measures the actual deprivation in select dimensions of quality of life (Box 2). All the four components are given equal weight age (with the subcomponents of education ­ literacy and enrollment sharing equal weights within educational attainment). Unmet basic need is measured against the benchmark need of 100 percent fulfillment. For example, the benchmark for safe water is that 100 percent of population should have safe water. 20. Since it is based on the census, the unmet basic needs index covers the whole population and is representative at the district level. (By contrast, none of the household surveys for 205 Yemen is representative at that level.)47 The UBN index is the main poverty index we use as a proxy to the head count ratio of poverty for our analysis. This index has the advantage that one can safely treat it as exogenous to the public spending. While the composition and weighting of the component indicators are not beyond question, Ravallion (2000) has used this method for the analysis of Argentina's Trabajar program. Box A.8. 3: Measuring Poverty Poverty is pronounced deprivation in well-being. The commonly used consumption measure of poverty measures deprivation in the material (money-metric) dimension. Measuring deprivation in key social dimensions like health and education (as captured in the Unmet Basic Needs Index here) is an alternative and complementary measure of poverty. WDR (2000-01) extends the concept of deprivation beyond the aforementioned dimensions to include vulnerability to risk and exposure. 21. The relative positions of governorates on the basis of UBN index thus constructed (for 1994) and head count ratios (based on consumption poverty in 1998) do not exactly match. According the estimates of World Bank's 1998 poverty update for Yemen, the number of poor people as a percentage of the governorate population is highest in Taiz (56 percent), Ibb (55 percent), Abyan (53 percent), and Lahj (52 percent), but is also high in Dhamar (49 percent), Hadramout, Al-Mahrah and Shabwah (43 percent). The incidence of poverty is lowest in Al- Baida (15 percent) and Saddah (27 percent), and in the two major urban centers, Sana'a city (23 percent) and Aden (30 percent). The ranks match for some governorates like Sanaa city, Al- Mahrah, Sanaa and Dhamar. However, the classification of governorates for both these measures are not exactly comparable and the data do not belong to the same year.48 22. Expenditure data show that Al-Mahrah, Aden and Abyan have the highest expenditure allocations, Al-Baida Al- Jawf, and Al-Hodeidah seem to be allocated the least amount on a percapita basis (Table 3). In terms of the UBN index, Sanaa city and Aden are the best performers while Al-Hodeidah, Hajjah, Al-Jawf and Al-Mahwit have the highest UBN indices in the country49 (Figure 1). V FINDINGS 23. This section interprets the results of the regression analysis presented in Table 3 and summarizes the findings of the paper. 47 Since the time of Yemen unification, the Central Statistical Organization (CSO) has implemented three household surveys: (i) the 1992 Household Budget Survey (HBS-92), (ii) the 1998 Household Budget Survey (HBS-98), and (iii) the 1999 National Poverty Phenomenon Survey (NPS-99). 48 The head count ratios for consumption poverty in 1998 have been based on a classification of Yemen into 15 governorates and the UBN index for 1994 is constructed for a classification of Yemen into 20 governorates. 49 See Annex 2 for exact estimates at the governorate level. 206 24. The analysis finds evidence of anti-poor bias in public expenditure allocations. The national targeting differential (TD) for Yemen in 2004 is about 6290 Rials per person, i.e, the poor receive YR 6290 (US $33) less than the non-poor in per-capita terms which is almost the same as the per-capita public expenditure on all Yemenis50 (Figure 3). The absolute level of targeting failure is substantial and there is a significant bias at the national level in government expenditure allocations against the poor. (t-ratio=-3.09) With a large of sample of 287 districts, the coefficient of targeting differential is negative and significant. Figure A.8. 2: Public Expenditure per capita (2004) and UBN Index (1994) 18 AL-M ahrah 16 Aden Government Exp per-capita in 2004, YR 14 12 Abyan Lahj 10 Hadramout Ad-dala 8 Shabwh M areb AL-M ahwit Sana'a 6 Sana'a Cit y Taiz Ibb Amran Sa'adah Hajjah Dhamar AL--Hodeidah AL-Jawf 4 AL- Baida 2 10 15 20 25 30 35 40 45 50 55 Poverty Index, 1994 Source: Department of Statistics (Census 1994) and Ministry of Finance (AFMIS Project Unit) 50 This refers to public expenditure at the decentralized level and accounts for about 10 percent of all public expenditure in Yemen. 207 Table A.8. 1: Contributions to the National Targeting Differential (in percent) 51 Governorate Contribution Inter-governorate 87.5 Intra-governorate 12.5 Hadramout 6 Shabwh 4.3 Sana'a 4.2 Abyan 2.3 Lahj 2.3 AL-Jawf 1.9 Amran 1.8 AL-Mahwit 1.5 Aden 1.3 Mareb 1.3 SanaaCity 0.6 Ad-dala 0.4 AL- Baida 0.4 Ibb 0.2 Sa'adah 0.2 Hajjah 0 AL-Mahrah -0.4 Dhamar -2.1 AL--Hodeidah -5.8 Taiz -7.8 Yemen 100 Source: Staff estimates 25. The anti-poor bias at the governorate level can be evaluated through the inter- governorate regression52,53.The results show that most of the anti-poor bias occurs at the level of governorates - the poor get YR 10,060 (US $52) per-capita less than the non- poor. The inter-governorate targeting differential (TP) is significantly different from zero 51 The decomposition of the national targeting differential into inter-governorate and intra-governorate components is exact and the contributions here are reported in percentages to offer magnitudes of relative importance and add up to 100 percent. 52 This refers to equation 2 listed in the earlier section on methodology (Section 2). 53 The estimate of TP is weighted by the number of districts in each governorate. The weighting is done by multiplying all variables by the square root of the number of districts prior to running the regression of the Gjs on the Hjs across governorates. See Ravallion (1999) for details. 208 at the 10 percent level (t-ratio=-1.78)54. A scatter plot of expenditure per-capita (2004) and poverty index (1994) at governorate level (Figure 2) confirms the negative correlation (Correlation coefficient = -0.4553). Table A.8. 2: Summary Results of the Regression Analysis by Govornate Targeting Per-capita expenditure allocations Differential t-ratio Total Poor Non-poor N National -6.29 -3.09 5.94 2.12 8.41 287 Inter-governorate -10.06 -1.78 20 Intra-governorate AL-Mahrah 9.02 0.71 17.24 21.89 12.87 8 Taiz 7.25 1.34 5.1 9.94 2.69 18 Dhamar 5.05 0.74 4.79 7.42 2.37 9 AL--Hodeidah 3.1 1.46 4.03 5.49 2.39 22 Hajjah 0.09 0.03 4.67 4.72 4.62 28 Ibb -0.57 -0.08 5.11 4.76 5.33 18 Sa'adah -1.42 -0.29 4.79 4.06 5.48 14 AL- Baida -2.11 -0.51 3.08 1.65 3.76 10 Ad-dala -6 -0.52 8.22 4.71 10.71 9 Hadramout -7.85 -2.02 8.92 2.98 10.83 28 Lahj -11.48 -1.63 9.96 2.5 13.98 14 Sana'a -12 -2.04 6.04 -0.29 11.71 18 AL-Mahwit -13.34 -2.66 6.74 0.08 13.42 8 Amran -16.35 -2.96 5.46 -4.13 12.22 19 Aden -20.4 -1.59 14.34 -3.82 16.59 8 Mareb -20.94 -1.07 6.63 -5 15.94 12 AL-Jawf -21.38 -1.7 3.62 -6.9 14.48 12 Abyan -23.31 -3.65 11.28 -4.86 18.45 10 Shabwh -27.28 -3.69 7.47 -10 17.29 16 Sana'a City -87.88 -1.3 5.17 -69.04 18.83 6 Source: Staff estimates Notes: 1. Variables are regressed as deviations from relevant group means. 2. Negative sign means that targeting differential (TD) is against the poor. The poor get less than the non-poor. 3. Coefficient values (TDs) are in absolute units of 1000 Rials. For example, the second data row says that at the governorate level, the poor receive 10,060 Rials less than the non-poor. 4. N is the number of observations (districts/governorates) available for the regression. 26. Intra-governorate allocation of expenditures across districts offsets some of the anti-poor bias observed at the governorate level. Four of the 20 governorates (Taiz, Al-Hodeidah, Ibb and Hajjaah), accounting for about 40 percent of Yemen's population, allocate more to the poor 54 The regression is weighted for the purposes of decomposition of the total targeting failure into between and within governorate components. 209 than the non-poor, or, allocate equal amount. The worst five governorates in this respect are Shabwah, Amran, Sanaa, Lahj and Hadramout55. 27. Analyses of public transfer programs in other countries also show that inter-regional targeting is often less pro-poor than intra-regional targeting (Alderman 2002 for Albania; Galasso and Ravallion 2005 for Bangladesh) which turns out to be the case in our analysis. Van de Walle (2002) assessed the incidence and targeting effectiveness of Yemen's safety net and poverty programs and arrived at a similar conclusion. Figure A.8. 3: Targeting Differentials by Governorate (in thousand Rials) Source: Staff estimates 55 However, these results need to be used with caution because of the statistical significance of the results for some governorates. The coefficients for the inter-governorate targeting differential and those for Amran, Sanaa, Hadramout and Abyan are statistically significant (see Table 3). 210 Box A.8. 4: Explaining the Targeting Differentials: North-South and Urban-Rural Dimensions Interesting insights emerge when the targeting differentials are analyzed from the North-South and rural-urban frames of reference. The Targeting differential is far more pronounced across the North-South dimension than the rural-urban dimension. While the North has bias neither for nor against the poor, the South has significantly anti-poor bias in per-capita allocation of expenditures. North-South Dimension Historical factors could help explain this pronounced targeting differential along the North-South dimension. North and south Yemen were united into a single state - the Republic of Yemen - on 22 May 1990. This replaced the Yemen Arab Republic (YAR) in the north and the People's Democratic Republic of Yemen (PDRY) in the south. The South had lesser poverty at the time of unification compared to the North. We find that only 28.2 percent of the erstwhile PDRY's population was poor on the basis of the UBN index constructed for 1994 compared to 41.6 percent of the erstwhile YAR. However, public expenditure allocations for 2004 reveal that per-capita allocation to the South (YR 10,270) is nearly twice as much as that of the North (YR 4970). This could be due to the higher number of civil servants in the South that were inherited from the PDRY. Moreover, population is more dispersed raising the cost of public service delivery. Only about 2.64 million Yemenis (18 percent of total) live in the South (Table 4). Figure 4 - UBN Index: North and South Yemen Figure 5 - UBN Index: Urban and Rural Yemen 80 80 60 60 UBN Index UBN Index 40 40 20 20 0 0 North Yemen South Yemen Urban Yemen Rural Yemen Source: Staff estimates based on Dept. of Statistics (Census 1994) Data. Note: The boxes contain the middle 50% of the data ­ the upper and lower hinges of the box indicates the 75th and 25th percentiles of the data. The line in a box indicates the median value and when it is not equidistant from the hinges, the data is skewed. The whiskers at the ends of the vertical lines indicate upper and lower adjacent values that are the extreme data points within 1.5 times the Inter- quartile Range of the nearer quartile.The points outside the ends of the whiskers are outliers. The UBN index across districts here is population-weighted. Per-capita allocation of expenditure is significantly anti-poor in the south. The poor get YR 2020 per capita as against YR 13,510 that the non-poor receive. In the North, however, where 87 % of the total poor live, per-capita allocation is neutral between the poor and the non-poor. The main reason for the observed anti-poor bias of expenditures at the national level can be traced to the anti-poor allocations in the South. Table A.8. 3: Summary Results of the Regression Analysis by North-South and Rural-Urban Districts Targeting Per-capita expenditure allocations Population Differential t-ratio Total Poor Non-poor N HCR (in millions) National -6.29 -3.09 5.94 2.12 8.41 287 39.2 14.38 Inter-group -39.64 2 Intra-group North Yemen -0.19 -0.11 4.97 4.86 5.05 198 41.6 11.74 South Yemen -11.49 -3.39 10.27 2.02 13.51 89 28.2 2.64 Inter-group 2.17 2 Intra-group Urban -9.50 -3.20 5.73 -0.98 8.51 51 29.3 5.07 Rural -7.97 -5.25 6.06 1.64 9.61 236 44.5 9.31 Source: Staff estimates Note: Targeting differential and per capita allocations in thousands of Rials. 211 28. Kanbur (2003) discusses the empirical and normative significance of such decompositions. It is suggested, based on a literature review, that whichever decomposition is done, it turns out that empirically, for gender (two groups) and race (usually less than five groups), the between group component is less than 15 percent. For space, it depends on how disaggregated a grouping is possible.56 However, our results, based on spatial disaggregations, conclude otherwise 57 . Table 2 presents the relative contributions of between and within governorate components to the national targeting differential. The between-governorate component accounts for 87.5 percent of the national targeting differential. Thus, most of the anti-poor bias occurs at the level of allocations to the governorates. 29. It is instructive to look at the targeting differential along North-South and Urban-Rural dimensions. The North (capitalist) and South (socialist) Yemen had very different economic systems which were merged at the time of unification. Further, since two-thirds of Yemenis live in rural districts, it is also useful to investigate whether targeting differential has any urban- rural bias (Box 3). 30. While the results clearly indicate an anti-poor bias at the governorate level, it is not obvious why this is so. It will be worthwhile to know the decision making process behind the allocation of government expenditures to governorates and districts. In Yemen's case, there are several reasons that potentially explain the seemingly anti-poor bias of inter-governorate allocations. Big city infrastructure costs, lobbying, high cost of provision of public services in hilly governorates with dispersed population and bogus wage bills are among the factors that may explain some of this bias. 31. At the intra-governorate level, it is not surprising that some governorates are better at pro - poor allocation, since there is no uniformity and capacity varies across governorates. Since most of the spending at the governorate level is the de-concentrated line ministry spending, analysis and decomposition at the ministry level expenditures may offer better explanations. 32. The national government presumably has more control on redirecting resources to the poor governorates (relative to influencing targeting within governorates) and hence possesses greater leverage on influencing the national targeting differential through the inter-governorate targeting differential. To the extent that the observed inequity in governorate level expenditure allocation captures more than the higher cost of provision of services in inaccessible regions, Yemen could improve targeting efficiency of public expenditures. 56 For example, for rural Peru, it requires going below Region, below Province, and below Canton to the Parroquia level (there are 915 of these units), for the between group component still to rise only to 15 percent. See Elbers et al. (2002) 57 There is also a technical question of the extent to which the number of groups in any given classification affects the between groups component. Clearly, if groups are subdivided into further subgroups, the between group component will increase for this reason alone. See Kanbur (2003). 212 REFERENCES Alderman, H., 2002. Do local officials know something we don't? Decentralization of targeted transfers in Albania" Journal of Public Economics 83: 375-404. Allen, R. et al, 2004. Moving Forward With Budget Reform And Fiscal Decentralization, Draft Report, The World Bank, Washington DC, September. Elbers, Chris, Peter Lanjouw, Johan Mistiaen, Berk Ozler and Ken Simler. 2002. "Are Neighbours Unequal? Estimating Local Inequality in Three Developing Countries,." Paper presented at the Cornell/LSE/WIDER Conference on Spatial Inequality and Development. Galasso, E. and M. Ravallion, 2005. "Decentralized Targeting Of An Antipoverty Program", Journal of Public Economics, Vol 89, Issue 4. Kanbur, R., 2003. "The Policy Significance Of Inequality Decompositions;" Cornell University, Mimeo, August. Ravallion, M,, 1999a. "Are Poorer States Worse at Targeting Their Poor?" Economics Letters 65(3):373­77. Ravallion, M,, 1999. Monitoring targeting performance when decentralized allocations to the poor are unobserved, Policy Research Working Paper 2080, The Development Research Group, The World Bank, Washington DC, March. Ravallion, M, 2000. Monitoring Targeting Performance When Decentralized Allocations To The Poor Are Unobserved, The World Bank Economic Review; Vol 14, No.2, May 2000. Sen, A.K, 1999. "Development As Freedom;" Oxford University Press, UK. Van de Walle, D., 2002. "Poverty and Transfers in Yemen," MENA Working Paper Series No. 30, Office of the Chief Economist of Middle East and North Africa Region, World Bank, Washington DC, December. Van de Walle, D., 2005. "Do Services And Transfers Reach Morocco's Poor? Evidence from Poverty and Spending Map;," Policy Research Working Paper 3478, The Development Research Group, The World Bank, Washington DC, January. World Bank, 2000-01. "Attacking Poverty;" World Development Report, Washington DC. World Bank, 2002a. "Republic of Yemen: Poverty Update," Report No. 24422-YEM (In Two Volumes), Middle East and North Africa, Social and Economic Development Group (MNSED), December 11. World Bank, 2002b. "Republic of Yemen ­ Country Assistance Strategy (CAS)," Report No. 24372-YEM, August 6. 213 ANNEX 1: DECOMPOSITION OF THE NATIONAL POOR-AREA TARGETING DIFFERENTIAL Decomposition: ^ ^ ^ T D = S P T P + S j T j Targeting differentialsa Weightsb (Gij - G)(Hij - H) fij ^ TD Inter-district (Hij - H)2 fij -1 M j (Gj - G)(H j - H) f j Mj (Hj - H)2 f j ^ TP SP Inter-governorate M j (H j - H)2 f j Hij - H)2 fij ( Mj Mj ^ (Gij - G j )(Hij - H j ) fij (Hij - H j )2 fij Tj i =1 Mj Sj i=1 Intra-governorate (Hij - H)2 fij (H i =1 ij 2 - H j ) f ij Notes: Gij is public expenditure percapita in the ith district of the jth governorate. The mean for that governorate is Gj, and the national mean is G. Governorate j contains Mj districts. Hij is the Unmet Basic Needs (UBN) Index (used as a proxy for the head count ratio of consumption poverty) in the ith district of governorate j, with governorate mean Hj and national mean H. fij refers to the population of the ith district of the jth governorate and fj refers to the population of governorate j. Indexing of the summations is only given when there is any ambiguity. a. Regression coefficients of public spending on the UBN indices across geographic areas b. Shares of the geographic variance of unmet basic needs. Based on: Ravallion 2000. 214 ANNEX 2: UBN INDEX BY GOVERNORATE Governorate UBN Index Abyan 30.8 Ad-dala 41.6 Aden 11.0 AL- Baida 31.9 AL--Hodeidah 53.0 AL-Jawf 50.8 AL-Mahrah 48.5 AL-Mahwit 50.1 Amran 41.4 Dhamar 47.8 Hadramout 24.3 Hajjah 51.2 Ibb 38.5 Lahj 35.0 Mareb 44.5 Sa'adah 48.5 Sana'a 47.2 Shabwh 36.0 Taiz 33.2 Sana'a City 15.5 YEMEN 39.2 Source: Department of Statistics (Census 1994) 215 ANNEX 9: CONSTRUCTION OF SOCIAL ACCOUNTING MATRIX Social Accounting Matrix for Yemen Activities Millions Rial, 2005 1 2 3 4 5 6 Receipts \ Payments A-Agriculture Mining and quarry A-Food ther and its produ everage and tobactextile and clothes Activities 1 A-Agriculture 2 A-Mining and quarrying 3 A-Food 4 A-beverage and tobacco 5 A-textile and clothes 6 A-leather and its products 7 A-wood and its product 8 A-paper and printing 9 A-oil refineries 10 A-chemical and fertilizer 11 A-rubber industry 12 A-non-metallic 13 A-metal and iron products 14 A-electrical and medical equipments 15 A-transportation 16 A-furniture 17 A-Electricity, Water and Gas 18 A-Construction 19 A-trade 20 A-Restaurants and Hotels 21 A-Transport, Storage & Communications 22 A-Financial Institutions & Real Estate 23 A-Real Estate & Business Serv. 24 A-Other services 25 A-Public administration Commodit 1 C-Agriculture 106,345 32 34,926 490 382 231 2 C-Mining and quarrying 33,571 92 0 5 3 C-Food 341 22,208 2,929 8 1 4 C-beverage and tobacco 998 3,773 5 C-textile and clothes 1,039 25 203 18 1,126 0 6 C-leather and its products 0 692 7 C-wood and its product 57 3 1 2 0 8 C-paper and printing 144 1,758 3,982 913 95 15 9 C-oil refineries 38,458 98,113 6,705 686 546 26 10 C-chemical and fertilizer 16,887 56 318 663 86 484 11 C-rubber industry 498 1,070 345 126 35 2 12 C-non-metallic 28 1,995 289 500 13 C-metal and iron products 14 14 C-electrical and medical equipments 533 341 358 43 10 2 15 C-transportation 110 1,625 39 17 11 3 16 C-furniture 1,451 2,147 459 114 171 3 17 C-Electricity, Water and Gas 170 164 361 95 130 25 18 C-Construction 10,267 7,518 6,686 496 431 24 19 C-trade 51,013 78,942 10,369 4,247 2,568 726 20 C-Restaurants and Hotels 5,605 958 28 21 C-Transport, Storage & Communications 19,813 32,849 13,871 1,261 890 134 22 C-Financial Institutions & Real Estate 11,931 33,320 4,969 350 260 20 23 C-Real Estate & Business Serv. 874 993 497 127 65 29 24 C-Other services 3,376 1,938 890 14 36 1 216 7 8 9 10 11 12 13 A-wood and its produc A-rubber industry A-non-metallic -paper and printingA-oil refineries hemical and fertiliz tal and iron produ 1 A-Agriculture 2 A-Mining and quarrying 3 A-Food 4 A-beverage and tobacco 5 A-textile and clothes 6 A-leather and its products 7 A-wood and its product 8 A-paper and printing 9 A-oil refineries 10 A-chemical and fertilizer 11 A-rubber industry 12 A-non-metallic 13 A-metal and iron products 14 A-electrical and medical equipments 15 A-transportation 16 A-furniture 17 A-Electricity, Water and Gas 18 A-Construction 19 A-trade 20 A-Restaurants and Hotels 21 A-Transport, Storage & Communications 22 A-Financial Institutions & Real Estate 23 A-Real Estate & Business Serv. 24 A-Other services 25 A-Public administration 1 C-Agriculture 44 47 243 119 185 2 C-Mining and quarrying 0 0 288,413 60 95 935 667 3 C-Food 4 26 11 51 2 4 C-beverage and tobacco 5 C-textile and clothes 4 4 2 142 1 0 6 C-leather and its products 0 17 7 C-wood and its product 2,395 57 22 0 3 5 8 C-paper and printing 2 5,379 82 51 31 875 126 9 C-oil refineries 84 32 76,248 118 98 512 184 10 C-chemical and fertilizer 34 165 3,724 1,277 1,489 33 65 11 C-rubber industry 276 2 255 59 780 62 36 12 C-non-metallic 16 9 5,405 56 6,426 1,150 13 C-metal and iron products 15 0 134 132 4,920 14 C-electrical and medical equipments 5 4 65 12 12 27 39 15 C-transportation 3 2 115 3 12 55 18 16 C-furniture 23 0 6,428 16 7 197 8 17 C-Electricity, Water and Gas 6 50 8,336 35 30 332 429 18 C-Construction 31 11 3,178 60 306 1,536 686 19 C-trade 446 447 53,184 1,629 2,984 1,585 1,287 20 C-Restaurants and Hotels 3,037 21 21 C-Transport, Storage & Communications 204 161 23,895 176 548 2,474 1,447 22 C-Financial Institutions & Real Estate 9 7 1,348 30 115 1,371 815 23 C-Real Estate & Business Serv. 9 31 1,300 5 159 31 146 24 C-Other services 3 2 84 3 20 46 94 25 C-Public administration 1 Paid-Public-Urban 21 141 1,893 28 715 217 21 22 23 24 25 A-Transport, Storage & Comm Institutions & R Estate & Business al A-Other servicesPublic administrat A-Agriculture A-Mining and quarrying A-Food A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing A-oil refineries A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas A-Construction A-trade A-Restaurants and Hotels A-Transport, Storage & Communications A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services A-Public administration C-Agriculture 153 586 14,033 C-Mining and quarrying 0 C-Food 315 9,134 C-beverage and tobacco 224 3 5 127 C-textile and clothes 19 1 313 4,096 C-leather and its products 211 954 C-wood and its product 15 5 24 1,596 C-paper and printing 2,426 581 527 1,685 4,274 C-oil refineries 58,900 36 332 7,052 21,561 C-chemical and fertilizer 1,040 433 6,534 1,869 C-rubber industry 13,991 16 70 308 41 C-non-metallic 1 1,931 59 18,097 C-metal and iron products 165 6,961 C-electrical and medical equipments 1,141 139 967 124 C-transportation 7,547 4 142 935 413 C-furniture 241 8 12 791 455 C-Electricity, Water and Gas 1,067 46 40 804 2,266 C-Construction 12,698 583 232 6,225 15,666 C-trade 68,250 73 1,594 22,507 11,326 C-Restaurants and Hotels 6,994 86 174 12,641 C-Transport, Storage & Communications 71,164 848 5,960 9,358 177,737 C-Financial Institutions & Real Estate 2,314 5,766 339 1,149 4,097 C-Real Estate & Business Serv. 8,261 196 1,116 2,355 14,390 C-Other services 484 51 2,649 5,341 1,238 C-Public administration 17,977 Paid-Public-Urban 6,864 3,332 68 3,308 139,747 218 or Yemen Commodities 1 2 3 4 5 6 7 C-Agriculture Mining and quarryi C-Food ther and its produ everage and tobactextile and clothes wood and its produc A-Agriculture 597,291 A-Mining and quarrying 1,454,669 A-Food 185,882 A-beverage and tobacco 30,427 A-textile and clothes 19,703 A-leather and its products 4,160 A-wood and its product 10,216 A-paper and printing A-oil refineries A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas A-Construction A-trade A-Restaurants and Hotels A-Transport, Storage & Communications A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services A-Public administration C-Agriculture C-Mining and quarrying C-Food C-beverage and tobacco C-textile and clothes C-leather and its products C-wood and its product C-paper and printing C-oil refineries C-chemical and fertilizer C-rubber industry C-non-metallic C-metal and iron products C-electrical and medical equipments C-transportation C-furniture C-Electricity, Water and Gas C-Construction C-trade C-Restaurants and Hotels C-Transport, Storage & Communications C-Financial Institutions & Real Estate C-Real Estate & Business Serv. C-Other services 219 8 9 10 11 12 13 C-paper and printing C-oil refineries C-rubber industry emical and fertiliz C-non-metallic tal and iron produ A-Agriculture A-Mining and quarrying A-Food A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing 16,827 A-oil refineries 276,358 A-chemical and fertilizer 6,369 A-rubber industry 12,486 A-non-metallic 59,914 A-metal and iron products 38,093 A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas A-Construction A-trade A-Restaurants and Hotels A-Transport, Storage & Communications A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services A-Public administration C-Agriculture C-Mining and quarrying C-Food C-beverage and tobacco C-textile and clothes C-leather and its products C-wood and its product C-paper and printing C-oil refineries C-chemical and fertilizer C-rubber industry C-non-metallic C-metal and iron products C-electrical and medical equipments C-transportation C-furniture C-Electricity, Water and Gas C-Construction C-trade C-Restaurants and Hotels C-Transport, Storage & Communications C-Financial Institutions & Real Estate C-Real Estate & Business Serv. C-Other services C-Public administration Paid-Public-Urban 220 14 15 16 17 18 19 20 C-electrical and medical equC-transportation C-furniture ctricity, Water an C-Construction C-trade estaurants and Ho A-Agriculture A-Mining and quarrying A-Food A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing A-oil refineries A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments 2,340 A-transportation 956 A-furniture 8,390 A-Electricity, Water and Gas 51,274 A-Construction 367,714 A-trade 462,004 A-Restaurants and Hotels 82,704 A-Transport, Storage & Communications A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services A-Public administration C-Agriculture C-Mining and quarrying C-Food C-beverage and tobacco C-textile and clothes C-leather and its products C-wood and its product C-paper and printing C-oil refineries C-chemical and fertilizer C-rubber industry C-non-metallic C-metal and iron products C-electrical and medical equipments C-transportation C-furniture C-Electricity, Water and Gas C-Construction C-trade C-Restaurants and Hotels C-Transport, Storage & Communications C-Financial Institutions & Real Estate C-Real Estate & Business Serv. C-Other services C-Public administration Paid-Public-Urban 221 21 22 23 24 25 C-Transport, Storage & Comm Institutions & Re al Estate & BusinessC-Other servicesPublic administrat A-Agriculture A-Mining and quarrying A-Food A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing A-oil refineries A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas A-Construction A-trade A-Restaurants and Hotels A-Transport, Storage & Communications 607,754 A-Financial Institutions & Real Estate 95,374 A-Real Estate & Business Serv. 133,059 A-Other services 70,908 A-Public administration 656,813 C-Agriculture C-Mining and quarrying C-Food C-beverage and tobacco C-textile and clothes C-leather and its products C-wood and its product C-paper and printing C-oil refineries C-chemical and fertilizer C-rubber industry C-non-metallic C-metal and iron products C-electrical and medical equipments C-transportation C-furniture C-Electricity, Water and Gas C-Construction C-trade C-Restaurants and Hotels C-Transport, Storage & Communications C-Financial Institutions & Real Estate C-Real Estate & Business Serv. C-Other services C-Public administration Paid-Public-Urban 222 or Yemen Labor Capital Households 1 2 3 4 1 1 Paid-Public-Urban Paid-Public-RuralPaid-Private-Urban Paid-Private-Rural Capital Urban Households A-Agriculture A-Mining and quarrying A-Food A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing A-oil refineries A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas A-Construction A-trade A-Restaurants and Hotels A-Transport, Storage & Communications A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services A-Public administration C-Agriculture 220,525 C-Mining and quarrying C-Food 55,181 C-beverage and tobacco 13,267 C-textile and clothes 17,837 C-leather and its products 1,038 C-wood and its product C-paper and printing 2,125 C-oil refineries 13,100 C-chemical and fertilizer 14,797 C-rubber industry 5,038 C-non-metallic 3,304 C-metal and iron products 13,072 C-electrical and medical equipments 490 C-transportation 3,910 C-furniture 1,101 C-Electricity, Water and Gas 17,084 C-Construction C-trade 43,931 C-Restaurants and Hotels 28,795 C-Transport, Storage & Communications 164,161 C-Financial Institutions & Real Estate C-Real Estate & Business Serv. 50,271 C-Other services 17,915 223 or Yemen Government 2 1 2 3 4 Rural-Households Government Indirect Taxes Subsidies Duties A-Agriculture 4,423 A-Mining and quarrying A-Food 2,739 A-beverage and tobacco A-textile and clothes A-leather and its products A-wood and its product A-paper and printing A-oil refineries 213,935 A-chemical and fertilizer A-rubber industry A-non-metallic A-metal and iron products A-electrical and medical equipments A-transportation A-furniture A-Electricity, Water and Gas 17,497 A-Construction 4,023 A-trade A-Restaurants and Hotels A-Transport, Storage & Communications 1,691 A-Financial Institutions & Real Estate A-Real Estate & Business Serv. A-Other services 41,691 A-Public administration C-Agriculture 490,255 C-Mining and quarrying C-Food 150,165 C-beverage and tobacco 21,536 C-textile and clothes 28,233 C-leather and its products 2,112 C-wood and its product 11,061 C-paper and printing 937 C-oil refineries 21,800 C-chemical and fertilizer 23,109 C-rubber industry 10,307 C-non-metallic 3,876 C-metal and iron products 8,868 C-electrical and medical equipments 366 C-transportation 2,822 C-furniture 421 C-Electricity, Water and Gas 18,075 C-Construction C-trade 77,709 C-Restaurants and Hotels 21,754 C-Transport, Storage & Communications 150,350 C-Financial Institutions & Real Estate C-Real Estate & Business Serv. 44,408 C-Other services 26,142 8,956 224 or Yemen Capital account Rest of World Residual Total 1 2 1 1 Saving-Investment Stock-variations Row A-Agriculture 601,714 A-Mining and quarrying 1,454,669 A-Food 188,621 A-beverage and tobacco 30,427 A-textile and clothes 19,703 A-leather and its products 4,160 A-wood and its product 10,216 A-paper and printing 16,827 A-oil refineries 490,293 A-chemical and fertilizer 6,369 A-rubber industry 12,486 A-non-metallic 59,914 A-metal and iron products 38,093 A-electrical and medical equipments 2,340 A-transportation 956 A-furniture 8,390 A-Electricity, Water and Gas 68,771 A-Construction 371,738 A-trade 462,004 A-Restaurants and Hotels 82,704 A-Transport, Storage & Communications 609,445 A-Financial Institutions & Real Estate 95,374 A-Real Estate & Business Serv. 133,059 A-Other services 0 112,600 A-Public administration 656,813 C-Agriculture 28,351 7,248 916,602 C-Mining and quarrying 75,538 1,051,730 1,456,102 C-Food 52,172 24,256 2,106 324,324 C-beverage and tobacco 4,585 351 47,171 C-textile and clothes 267 465 54,799 C-leather and its products 2,072 1,377 32 8,510 C-wood and its product 26 118 31,750 C-paper and printing 23,115 784 30 60,548 C-oil refineries 48,799 78,789 -16,072 477,331 C-chemical and fertilizer 80 4,472 382 80,255 C-rubber industry 547 156 39,974 C-non-metallic 793 72 98,019 C-metal and iron products 120,000 9,510 30,449 215 217,161 C-electrical and medical equipments 97,208 1,215 8 103,817 C-transportation 71,116 455 66 90,513 C-furniture 14,782 678 15 29,801 C-Electricity, Water and Gas 350 51,274 C-Construction 275,346 0 367,714 C-trade -1,012 462,004 C-Restaurants and Hotels 497 82,704 C-Transport, Storage & Communications 75,057 3,113 832,899 C-Financial Institutions & Real Estate 7,201 116,975 C-Real Estate & Business Serv. 936 133,059 C-Other services 443 70,908 225 ANNEX 10. UPDATES OF THE INPUT/OUTPUT TABLE FOR 2005 I IO AND SAM TABLES FOR THE YEMENI ECONOMY I TECHNICAL NOTE 1. Properties and advantages of IO and SAM tables are well established in the recent literature on policy simulation modeling. They provide a comprehensive and consistent data foundation. An IO table describes quantitatively the sectoral transactions taking place in an economy during a specified period of time, generally one year. It consists of row and column accounts that represent the inter-industry transactions, payments to factors of production, expenditures of households and government, investment by commodity, and transactions with the rest of the world. Analytically, IO tables are widely used in the analysis of production, employment, trade, as well as issues of more recent interest, such as energy and the environment. Statistically, IO tables function as frameworks for data compilation that permit the statistician to check for quality and consistency. 2. On the other side, a SAM is a comprehensive, economy-wide data framework, typically representing the economy of a nation. A SAM is formatted as a square matrix in which each account is represented by a row and a column. Each cell shows the payment from the account of its column to the account of its row. Thus, the incomes of an account appear along its row and its expenditures along its column. The underlying principle of double-entry accounting requires that, for each account in the SAM, total revenue (row total) equals total expenditure (column total). A SAM may be viewed as an input-output table that has been extended to cover the full circular flow of incomes, linking GDP on the supply side, represented by incomes accruing to factors and the government (indirect taxes net of subsidies), to GDP on the demand side, defined as the sum of domestic and foreign final demands for the nation's outputs net of imports. This requires that the database include comprehensive budgets for domestic institutions (government and non- government) and the rest of the world (the current account of the balance of payments). In addition, compared to what is implied by the IO structure, a SAM typically has a more disaggregated treatment of factors, domestic non-government institutions (households and enterprises), indirect taxes, and subsidies. For each institution, these budgets cover all current revenues and expenditures, including savings. Given the requirements of the SAM structure, it is necessary that the institutional budgets be consistent in terms of disaggregations and values, both in their interface with the accounts that appear in the IO table (for example, the sum of household consumption demands for any commodity in the more detailed SAM must equal the value for aggregate household consumption demand for the same commodity in the IO table) and internally (for example, a transfer payment from the government to a household must appear with the same value and account name in the government and household budgets). 3. The construction of SAMs is driven by three motivations. First, it displays information in a manner that exhibits the structure of an economy in an illuminating way. 226 Secondly, by exposing inconsistencies between data from different sources, it contributes to improvements in the database. Thirdly, it provides all or at least a major part of the data needed for different types of models, most importantly fixed-price SAM-multiplier models and Computable General Equilibrium (CGE) models (Round 2003, pp. 301-302). 4. This technical note describes the estimation methods and the data used in the updating of the IO table and the development of a Social Accounting Matrix (SAM) for Yemen for the year 2005. II THE IO TABLE 5. Drawing on the IO table for 2002 for Yemen and available data for 2005, we estimated a new IO table for the year 2005. The method used for the updating consists of simultaneous applications of RAS for account balancing. The data used for the updating has been drown from various sources including national accounts, government budget, household survey, industrial survey, trade statistics, and many other non-published documents produced by the Central Statistics Organization. The following list presents some of the key data sources used in the updating of the IO table for Yemen. Central Statistical Organization (2006). "Household Budget Survey April 2005 ­ March 2006". Central Statistical Organization (2006). "Results of the Labor Force Survey for the year 2001", Central Statistical Organization (2006). "Statistical Year Book" Central Bank of Yemen (2006). "Annual Report 2005" T.G. . « An Input Output Table for Yemen 2002 ». IMF (2006). "Republic of Yemen: Statistical Appendix". COMTRADE DATABASE 2006 6. The final version of the Yemeni IO table developed in this study includes 25 activities. Table 1 presents a listing of these sectors. The methodology for updating the IO table for the year 2005 is documented in the next sections. 227 Table A.10. 2: Sectoral Dimension of the Yemeni IO Table for the year 2004 Sector Sector identification Classification Sector 1 Agriculture Sector 2 Mining and inquiring Sector 3 Food Sector 4 Beverage and tobacco Sector 5 Textile and clothes Sector 6 Leather and its products Sector 7 Wood and its products Sector 8 Paper and printing Sector 9 Oil refineries Sector 10 Chemical and fertilizer Sector 11 Rubber industry Sector 12 Non-metallic Sector 13 Metal and iron products Sector 14 Electrical and medical equipments Sector 15 Transportation Sector 16 Furniture Sector 17 Electricity, Water and Gas Sector 18 Construction Sector 19 Trade Sector 20 Restaurants and Hotels Sector 21 Transport, Storage & Communications Sector 22 Financial Institutions Sector 23 Real Estate & Business Service Sector 24 Other services Sector 25 Public administration The updating of the IO table for Yemen has been realized following three major steps. In the first step, we have used aggregated data related to the national accounts to build an aggregated IO table with only one production sector. The data covers total public and private final consumption, gross output, GDP, intermediate consumption, wages, exports and imports, indirect taxes and subsidies, investment and changes in stocks, and import duties. These data are provided by the CSO. The aggregated IO table has served as a coherent accounting framework for the sectoral disaggregation carried out throughout this work. At a first stage, we have used the sectoral data on the production value, the intermediates consumption and the value added provided by the CSO for a given number of sectors. These sectors are the following: agriculture, oil and gas, other mining and quarrying, oil refining, other manufacturing, electricity water and gas, construction, wholesale and retail sale, restaurants and hotels, maintenance, transport and storage, communications, financial institutions, real estate and business services, social and personal services, government services, private non-profit services, and household services. For sectors listed in the original IO table for the year 2002 (manufacturing sectors) and no data is available on their respective GDP in 2005, we used the sectoral shares in the total manufacturing sector (excluding oil refining) drawn from the IO 2002. Once a table on sectoral GDP and intermediate consumption is established, we update the IO 2002 by imposing the new vectors on sectoral GDP and intermediate consumption in addition to data covering sectoral exports and imports, final private and public demand by commodity, investment by commodity, and taxes and subsidies. Data on sectoral exports and imports are drawn directly from COMTRADE database. Private and public final demand vectors as well as sectoral investment are calculated in two steps. In the first one, we used initial IO table for 2002 to derive coefficients on sectoral repartition of private and public final consumption as well as sectoral investment. In the second step, these coefficients are imposed to the respective value of total final consumptions and investment for 2005. A new vector of sectoral final demand and investment are than estimated. For indirect taxes, subsidies and tariff, the corresponding rates are calculated using the IO 2002. These rates are than imposed on total output at basic prices (for indirect taxes and subsidies) and on imports for tariffs. The total of indirect taxes, subsidies, and duties revenues calculated using these coefficients are different than those figuring in the national accounts. An adjustment was made on these rates to produce the exact values as reported in the national accounts. When the new IO table is estimated, many inconsistencies appears, mainly in the form of high variations (growth or decline) in stock variations. To avoid this shortcoming, an additional adjustment was applied on the estimated IO table. The adjustment of the IO table is a continuous process meant to ensure a greater representation of the Yemeni economy at the level of sectoral technologies as well as macro-economic balances. In this context, since the sectoral technologies for the 2005 IO table are, in the first place, estimated on the basis of the 2002 IO table, some adjustments have been introduced to improve this table given the structural changes in the economy, mainly as result of higher oil prices, reduction of subsidies on energy and food products... Thus, all the detailed information provided by the CSO on the cost 228 composition of certain activities, or even some enterprises have been used to refine the values of the technical coefficients, estimated earlier. Some balancing procedures have been applied whenever new specific data have been used, with the aim to maintain the equilibrium of the table, especially at the macro-economic level. In fact, the estimation of the value added and its distribution between its different components for the manufacturing sectors as well as the estimation of the total of intermediate consumption and its distribution between the different products and services of the sectors of the IO table has been enriched by additional sectoral data as well as at firm level. Similarly, we have proceeded to a new estimation of the commodity-structure of the household consumption using the results of the latest survey on income and expenses of Yemeni households. III THE SOCIAL ACCOUNTING MATRIX Table A.10. 3: The Accounts in our SAM for Yemen and its Structure Account acronym Description ACT Production activities. Paying (in its column) for inputs used and paid (in its row) for outputs produced. COM Commodities. Paying the supply side (including production activities and imports) and paid by the demand side (domestically and for exports) LAB* Labor. Paying the institutions to which the labor belongs and paid by the activities in which it is employed). CAP* Capital (other than capital owned abroad). Paying to the institutions that own the capital and paid by the activities in which it is employed Households** Domestic non-government institutions (including households, enterprises, and private non-profit institutions). Payments made cover consumption, transfers, direct taxes, and savings. Payments received consist of factor incomes and transfers from other institutions. GOV** Government. Payments made cover consumption, transfers, and savings. Payments received consist of factor incomes, transfers, and taxes. Duties Duties on imports from the rest of the world I-TAX Indirect taxes paid by activities and commodities and forwarded to the government. SUBSIDIES Treated as indirect taxes with negative values. Paid by activities and commodities and forwarded to the government. ROW** Rest of the world. Paid when Yemen imports and makes transfers to the rest of the world. Makes payments when Yemen exports and receives capital income from abroad and in the form of foreign savings (the current account deficit, which may be negative). S-I Savings-investment account. Collects savings from domestic institutions and the rest of the world and allocates these to domestic investment (gross fixed capital formation and stocks) *Note: Production factors. **Institutions. 7. In addition to the IO table, building the SAM requires the following additional data: transfers between domestic institutions and the rest of the world (balance of payments), taxes paid by households to government, and transfers from government to households. Savings of the different institutions is residual. Various data sources are used for building the SAM. They include the balance of payments and government budget provided by CSO. The most important task in building the SAM was the disaggregaqtion of labor into four categories and households into two categories. For households, final consumption by commodity and household category are estimated using the result of the latest household survey on income and expenditures. Concerning labor disaggregation, the same survey is 229 used to estimate the number of workers by category and sector as well as their respective yearly wages. Table A.10. 4: Price Vectors for the Following Simulations Price vector with new SAM Price vector with new SAM Price vector with new SAM Price vector with new SAM for Civil Service Wage for oil subsidy shock for tariff reduction for GST implementation increase Assumptions: overall average Assumptions: weighted % Assumptions: flat increase of tariff reduction for Assumptions: Paid Public increase applied, 250% to oil 10% on commodities was commodities of -7.4% was Urban and Paid Public Rural refineries and 33% of Gas implemented (excluding implemented (excluding sectors were shocked by 50% increase (100%) public administration) public administration) C-Agriculture 21.8% -16.0% 21.6% 2.0% C-Mining and quarrying 28.8% -16.2% 21.9% 2.6% C-Food 13.9% -14.3% 19.3% 1.2% C-beverage and tobacco 11.0% -14.1% 19.1% 1.1% C-textile and clothes 7.7% -11.2% 15.1% 1.2% C-leather and its products 6.7% -12.7% 17.2% 0.9% C-wood and its product 4.0% -10.2% 13.8% 0.6% C-paper and printing 2.7% -9.7% 13.1% 0.6% C-oil refineries 250.0% -16.5% 22.2% 1.5% C-chemical and fertilizer 1.4% -8.3% 11.2% 0.2% C-rubber industry 4.5% -10.9% 14.7% 0.6% C-non-metallic 9.5% -13.2% 17.8% 1.9% C-metal and iron products 2.5% -9.0% 12.2% 0.4% C-electrical and medical equipments 0.6% -7.7% 10.4% 0.1% C-transportation 0.3% -7.5% 10.2% 0.2% C-furniture 5.3% -10.2% 13.8% 0.5% C-Electricity, Water and Gas 99.6% -20.9% 28.3% 7.6% C-Construction 15.3% -18.6% 25.1% 2.3% C-trade 12.5% -14.4% 19.5% 1.8% C-Restaurants and Hotels 17.1% -18.3% 24.7% 2.1% C-Transport, Storage & Communications 27.2% -15.2% 20.5% 2.1% C-Financial Institutions & Real Estate 10.4% -13.9% 18.8% 3.6% C-Real Estate & Business Serv. 12.7% -14.9% 20.2% 2.1% C-Other services 29.7% -19.5% 26.3% 5.3% C-Public administration 29.3% -13.8% 18.7% 25.3% Paid-Public-Urban 8.8% -13.2% 17.9% 54.4% Paid-Public-Rural 8.2% -12.7% 17.2% 53.6% Paid-Private-Urban 8.8% -13.2% 17.9% 4.4% Paid-Private-Rural 8.2% -12.7% 17.2% 3.6% Capital 4.2% -6.5% 8.7% 1.9% Urban Households 8.8% -13.2% 17.9% 4.4% Rural Households 8.2% -12.7% 17.1% 3.5% 230 ANNEX 11. DEMAND SYSTEM ESTIMATION 231 232 233 234 235 236 237 238 239 240 241 ANNEX 12: NATIONAL ACCOUNTS DATA Yemen: Nominal GDP, Expenditure Side, millions of Yemeni Rials ITEM 1990 1991 1992 1993 1994 1995 1996 1997 1998* 1-Final Consumption Expenditure 113,237 160,683 191,237 257,695 306,016 508,353 640,681 748,736 838,855 public final consumption 23,003 29,418 37,685 46,048 58,847 74,865 104,177 137,563 140,173 private final consumption 90,234 131,265 153,552 211,647 247,169 433,488 536,504 611,173 698,682 2-Gross Investment 18,406 24,334 43,026 48,249 64,390 112,713 170,879 217,786 203,181 Gross Fixed capital formation 15,074 20,955 38,157 41,627 58,267 106,227 158,016 188,237 194,526 Change in stock 3,332 3,379 4,869 6,622 6,123 6,486 12,863 29,549 8,655 3-Balance of goods & services -7,330 -33,860 -42,062 -66,927 -56,127 -101,493 -66,208 -77,864 -177,127 Exports of good & services 18,060 19,416 22,513 32,833 42,091 115,957 285,587 320,822 228,025 exports of goods 16,197 16,861 18,164 26,218 34,002 99,947 262,407 293,983 204,327 Exports of services 1,863 2,555 4,349 6,615 8,089 16,010 23,180 26,839 23,698 Imports of goods & services 25,390 53,276 64,575 99,760 98,218 217,450 351,795 398,686 405,152 imports of goods 17,400 40,997 48,184 77,975 77,102 172,660 285,035 311,112 311,002 Imports of services 7,990 12,279 16,391 21,785 21,116 44,790 66,760 87,574 94,150 4-GDP at market prices (1+2+3) 124,313 151,157 192,201 239,017 314,279 519,573 745,352 888,658 864,909 Non - Oil GDP 106,903 136,036 178,835 226,389 296,544 450,461 552,581 641,884 723,859 5-Consumption of fixed capital 5,276 8,615 11,927 15,245 19,884 28,845 42,080 57,346 78,691 6-indirect Taxes (net) 8,496 11,452 12,211 14,374 14,299 31,118 7,550 -19,705 -1,364 indirect Taxes 8,496 11,452 12,211 14,374 14,299 31,118 48,192 55,096 50,905 Subsidies 40,642 74,801 52,269 7-GDP at factor cost (4-6) 115,817 139,705 179,990 224,643 299,980 488,455 737,802 908,363 866,273 8-Domestic Demand (1+2) 131,643 185,017 234,263 305,944 370,406 621,066 811,560 966,522 1,042,036 9-Domestic Saving (4-1) 11,076 -9,526 964 -18,678 8,263 11,220 104,671 139,922 26,054 10-Net Factor income from abroad -940 -5,832 -5,011 -4,901 -6,388 -22,535 -65,067 -69,775 -37,763 Labor income from abroad 4,380 166 146 85 85 908 7,688 8,201 9,358 Labor income to abroad -960 -318 -273 -258 -333 -1,269 -390 -279 -324 Investment income from abroad 440 354 311 180 180 1,929 5,519 8,998 9,376 Investment income to abroad -4,800 -6,034 -5,195 -4,908 -6,320 -24,103 -77,884 -86,695 -56,173 11-GNP at market prices (4+10) 123,373 145,325 187,190 234,116 307,891 497,038 680,285 818,883 827,146 12-Net current transfers from abroad 16,900 26,992 33,260 49,114 84,569 128,774 138,664 150,524 171,646 Received 17,180 29,100 34,972 51,168 86,441 132,458 146,080 155,817 177,803 Payment -280 -2,108 -1,712 -2,054 -1,872 -3,684 -7,416 -5,293 -6,157 13-National Disposable Incom (11+12 - 5) 134,997 163,702 208,523 267,985 372,576 596,967 776,869 912,061 920,101 14-National Saving from NDI (13-1) 21,760 3,019 17,286 10,290 66,560 88,614 136,188 163,325 81,246 15-National Saving from GNP (11-1) 10,136 -15,358 -4,047 -23,579 1,875 -11,315 39,604 70,147 -11,709 * Provisional Actual ** Provisional *** Comp 242 GDP at Producers Prices, Nominal Prices, millions of Yemeni rials 1990 1991 1992 1993 1994 1995 1996 1997 1998* 1-Agriculture, Forestry and Fishing 29,877 33,017 44,126 50,550 69,888 100,840 122,492 140,839 168,677 Agriculture & Foresty (with out qat) 18,483 19,542 28,175 31,173 42,755 64,818 76,996 88,347 109,875 Qat 10,598 12,092 14,050 17,312 21,319 27,190 34,796 39,236 44,188 Fishing 796 1,383 1,901 2,065 5,814 8,832 10,700 13,256 14,614 2-Mining and Quarrying 17,660 15,393 13,753 13,162 18,488 70,166 194,046 248,088 142,461 Mining and Quarrying 250 272 387 534 753 1,054 1,275 1,314 1,411 Oil and Gas 17,410 15,121 13,366 12,628 17,735 69,112 192,771 246,774 141,050 3-Manufacturing 9,795 13,114 19,005 26,360 39,798 65,860 63,406 67,182 68,293 Manufacturing 9,292 12,576 18,461 25,814 39,300 64,095 58,677 63,762 62,948 Oil Refining 503 538 544 546 498 1,765 4,729 3,420 5,345 4-Electricity, Water and Gas 1,378 1,916 1,988 2,102 2,028 3,117 4,858 5,903 7,839 5-Construction 3,437 4,790 7,268 8,137 12,571 20,224 32,647 46,164 50,703 6-Wholesale and Retail Trade, Rest. & Hot 10,418 15,285 19,702 27,743 39,773 65,298 79,829 91,982 107,644 Wholesale and Retail Trade 7,884 11,369 14,706 19,986 27,723 45,593 56,585 64,692 77,973 Restaurants and Hotels 1,166 2,017 2,650 4,192 7,007 11,568 13,727 17,105 18,518 Maintenance 1,368 1,899 2,346 3,565 5,043 8,137 9,517 10,185 11,153 7-Transport, Storage & Communications 16,951 20,923 27,962 38,148 44,781 63,986 80,801 103,709 115,148 Transportand Storage 16,003 19,755 26,404 36,037 42,294 60,444 76,196 98,003 108,611 Communications 948 1,168 1,558 2,111 2,487 3,542 4,605 5,706 6,537 8-Financial Institutions & Real Estate 11,876 18,222 22,836 28,984 38,745 58,156 74,566 80,540 97,958 Financial Institutions 4,174 7,858 8,832 11,164 16,078 21,541 21,429 23,536 33,602 Real Estate & Business Serv. 7,702 10,364 14,004 17,820 22,667 36,615 53,137 57,004 64,356 9-Community Social & Personal serv. 1,390 1,821 2,179 3,026 4,141 6,173 6,987 8,032 9,054 Total Of Industries 102,782 124,481 158,819 198,212 270,213 453,820 659,632 792,439 767,777 B-Producers Of Government Services 19,934 25,445 32,762 41,286 49,184 66,951 77,708 90,755 101,966 C-Household Sector ( houses's Servecies ) 300 310 320 340 360 380 390 400 410 D-Producers Of Private Non -Profit serv. 73 101 132 165 203 243 287 335 388 E- Import Duties 4,122 6,119 6,769 7,680 7,475 16,989 26,251 25,785 24,080 Less: Imputed Bank Services Charge -2,898 -5,299 -6,601 -8,666 -13,156 -18,810 -18,916 -21,056 -29,712 G D P At Market Prices 124,313 151,157 192,201 239,017 314,279 519,573 745,352 888,658 864,909 Non -Oil GDP 106,903 136,036 178,835 226,389 296,544 450,461 552,581 641,884 723,859 * Provisional Actual ** Provisional 243 ANNEX 13: MARKET SHARE ANALYSIS Constant Market Share Analysis: Yemen Manufacturing Non-Oil Total Exports 98-05 98-05 98-05 Total Increase 100.0 100.0 100.0 World Trade Effect 13.1 32.1 43.6 Composition Effect -0.8 -13.6 -41.9 Competitiveness 87.6 81.5 98.3 244 ANNEX 14: CALCULATION OF WELFARE GAINS 245