Turkey - Poverty and coping after crises (Vol. 1 of 2) : Main Report

Report Report No. 24185-TR No. Turkey 24185-TR Poverty and Coping After Crises (In Two Volumes) Volume I: Main Report July 28, 2003 Human Development Unit Europe and Central Asia Region Turke y Poverty and Coping After Crises Volume I Document of the World Bank  Turkey: Poverty and Coping After Crises CurrencyEquivalents ExchangeRate Effective= September 1,2001 CurrencyUnit =Turkish Lira (TL) US$ 1=TL 1,500,000 GovernmentFiscalYear January 1-December31 Weights and Measures Metric System Abbreviationsand Acronyms BK Bag-Kur CPI ConsumerPrice Index DIE StateInstitute of Statistics ES Emekli Sandigi FA0 (UN)Food andAgnculturalOrganization HCIS Household Consumptionand Income Survey HIES Household Income and Expenditure Survey LSA Living StandardsAssessment PPP PurchasingPowerParity SRMP Social Risk Mitigation Project SSK Sosyal SigortalarKurumu SHCEK Social Services and Child Protection Organization SYDTF Social SolidarityFund SYDV(s) Social SolidarityFoundation(s) Vice President: Johannes Linn (ECAVP) CountryDirector: Ajay Chhibber(ECCU6) Sector Director: Annette Dixon (ECSHD) Sector Manager: Michal Rutkowski (ECSHD) Team Members: Jeanine Braithwaite (Team Leader), John Innes, Jennifer Manghinang (ECSHD); Mathew Verghis (ECSPE); Diane Steele (DECRG); Ayge Ayata, Sencer Ayata, Tahire Erman, Kivilcim Ozcan, Y. Ziya Ozcan (Consultants).  Prefaceand Acknowledgements The team i s gratehl for comments received from Laura Rawlings, EmilTesliuc, Sally Zeijlon, State Institute of Statistics, State PlanningOrganization, Social Solidarity Fund, and Undersecretariat of the Treasury. The research on which this report i s based was not undertaken by any Turkish Government agency, and thus the Government does not confirm the findings of this report, although the Government has agreed to the publication o f the report. Government agencies have commented on this report, expressingconcerns about the validity o fthe findings, particularly for rural areas, and concerns about the sample and especially the method used to reduce the sample (see Annex 1to the main report). The World Bank will publishthe questionnaire, data document including description of created variables, and the data on its webpage: http://www.worldbank.org/lsms/lsmshome.html   Table of contents Abbreviations and Accronyms Preface and Acknowledgements Executive Summary............................................................................................................. i 1. Data andMethodology................................................................................................... 1 2. Macroeconomic analysis ................................................................................................ 5 3. Extreme poverty............................................................................................................ 11 4.Urban food poverty....................................................................................................... 13 5.Vulnerability ................................................................................................................. 23 6.Inequality ...................................................................................................................... 25 7.Riskmitigation, risk coping, andrisk reduction among the poor................................. 26 8. Conclusions and Recommendations ............................................................................. 45 Bibliography ..................................................................................................................... 49 Annex 1. Onthe HCIS sample......................................................................................... 50 Annex 2. Text Tables....................................................................................................... 52   Executive Summary Turkey experienced severe losses o f life and infrastructure in 1999 caused by the August earthquake. The earthquake was followed by a period o f economic and financial crisis, culminating in a major currency devaluation inFebruary2001. What has beenthe social impact o f these crises? In order to answer that question, the World Bank and the Government o f Japan co-financed a household survey duringthe summer o f 2001, which consisted o f surveying 4200 households on their consumption and income, and interviewing 120 respondents in depth for case studies. There are more than the usual problems o f comparing data taken from two very different surveys in 1994 and 2001. This study seeks to answer three mainquestions: how many arepoor inTurkeyin 2001; who are the poor and why are they poor?; andhow do the poor cope with risk and poverty?. The main findings are: Extreme poverty ($1 per person per day) is basically unchanged since 1994 and remains quite low by international standards. Urbanpoverty (food consumption standard) increased since 1994. Inequalitywas unchanged, remainingat highlevels by international standards. Coping mechanisms o f the poor, especially relying on relatives and neighbors for in-kindandcash assistance, came under stress andthe poor reported a decrease in assistance from these traditional channels. Informal employment, an important mainstay o f the poor, was reduced as a consequence o f the crises. In spite of their reluctance, some of the poor were forced to either pull their children from school or have them attend less, for lack o f resources to cover out- of-pocket expenditures andinformal earnings from child labor. Idiosyncratic shocks such as major illnesses, were the least prepared for and most difficult for the poor to handle, after the daily task o f feeding the family was met. People ingeneral felt that they were worse o f fin2001. The major effect o f the crises has been an increase in poverty in urban areas o f Turkey from 1994 to 2001. Extreme poverty in all o f Turkey has not changed, and remains at low levels, but inequality i s also unchanged at quite high levels. A relatively large share (nearly one-fifth) of the urban population has consumption below a food standard, and qualitative evidence indicates that poverty has worsened in rural areas as well. There are two main definitions of poverty usedin this report. Extreme poverty i s defined for the portion o f the population with per capita consumption underneath the World Bank's extreme purchasingpowerparity poverty line o fUS$ 1per personper day. Urbanfood poverty is defined as those inurban areaswith equivalent consumptionbelow the cost of a food basket. The poor have been particularly impactedby a reduction in seasonal and informal employment opportunities in the urbanized areas, and some menare returning to their rural villages because they can not earn enough in the cities to cover their costs. The i  primary coping strategy o f the poor has been to reduce consumption, particularly consumption o f food and quality o f food consumed, but there are also indications that the poor may have to cut back on education expenses and withdraw children from school. The poor rely strongly on networks o f extended family, friends, neighbors, and hemjeri (people o f same place o f origin), but these networks are strained to the limit by the covariate macroeconomic shocks experienced in 1999 (earthquake) and 2000-2001 (financial crisis). Other important coping strategies o f the poor have also come under stress as a result o f the crises. Multiplejob holding has been curtailed by the reduction in seasonal and informal employment. The poor have been much less able to invest in physical assets or in their own human capital. Social capital, which i s an extremely important aspect o f t raditional Turkishs ociety, has c ome under s train as t hepoor c an no 1onger afford to attend traditional reciprocity events such as weddings. The poor are not positioned to sell their assets--nor is there much demand for them. Borrowing from neighbors i s a strategy usedacross the income spectrum-50 percent o f urban households and nearly 60 percent o f rural households reportedthat they had borrowed inthe previous nine months. Religious charity helps some o f the poor, but it i s episodic and does not cover needs sufficiently. As a last resort, the poor have sent their children out to work. Although the Government does finance ad hoc social assistance for the poor, this assistance i s too partial to meet the needs and i s not that well-targeted to the poorest. In recognition o f the impact o f the crises, in August 2001, the Turkish government transferred substantial resources to the Social Solidarity Fund (SYDTF) to finance back- to-school packs for 1.05 million poor children (TL 50 million per child). Additionally, the SYDTF expanded its food and fuel assistance for the winter. These measures, while important, were limited to a single payment. Recognizing that the poor needed more systematic assistance, the Government decided to adopt a new social assistance benefit- conditional cash transfers (CCT) which would be paid on a regular basis. The Government is also seeking to expand the traditional activities o f the SYDTF interms ofmicro-projects andadult literacy efforts as it recognizes that demandfor these activities exceeded supply. Finally, the Government intends to monitor poverty with more frequently household surveys, recognizing that one-off efforts such as the household consumption and income survey (HCIS) analyzed herein are not adequate to provide policy makers with the data needed to understand and therefore improve social protection and other sectoral efforts. Inview o f this substantial effort to reform the safety net in Turkey and to attack poverty, the World Bank has been able to respond by supporting the reform process with financial resources. The major vehicle o f Bank support is the Turkey Social Risk Mitigation Project/Loan (SRMP). The SRMP i s designed to support Turkey's ongoing efforts to reform, improve, and expand the social protection system and to address some o f the negative coping strategies that the poor have been forced to adopt in response to the impact o fthe earthquake and financial crises. 11  The report concludes with the following policy recommendations: 0 Macroeconomic management to resume broad-based growth, which should reverse the poverty trend since the vast majority o f the newly poor are not extremely poor 0 Counter negative coping strategies o f the poor by providing conditional cash transfers 0 Expand job opportunities for the newly poor through micro-projects and community development 0 Improve targeting and coverage of the extreme poor and outreach to them through institutional strengthening 0 Institute regular poverty monitoring through household surveys and the development of a poverty map. ... 111   1. Data and Methodology Turkey does not yet have a well-developed system o f annual household surveys that can be used to track poverty and social protection program utilization, and this lacuna is especially noticeable when questions about trends in poverty are raised. An important part o f the institutional development component o f the Turkey: Social Risk Mitigation Project (SRMP), a U S $500 million loan from the World Bank, seeks to rectify this gap by providing technical assistance and co-financing to an annual program o f household surveys and for producing a poverty map. However, the SRMP will co- finance surveys beginning in 2003. For retrospective analysis, other surveys must be utilized. Turkey did conduct major (and very large sample) household surveys in 1987 and 1994. These two household income and expenditure surveys (HIES) were analyzed by the World Bank in its Turkey: Living Standards Assessment (2000). The 1994 HIES surveyed more than 24,000 households using a diary and collecting detailed data on expenditures by individual type and on unit value prices. The HIES were designed primarily to provide weights for the consumer price index, not for poverty monitoring, although the HIES did provide data on household consumption and income that could be compared to various poverty lines. Turkey does not have an official poverty line. The Living Standards Assessment (LSA 2000) comparedhousehold consumption and income to a wide variety o f poverty lines (more than a dozen) for the 1994 HIES. The three main poverty lines used were: (a) the World Bank's U S $1 per person per day line; (b) a food consumption standard, based on minimum caloric intake and the FA0 equivalence scale for urbanareas; and (c) a vulnerability line equal to twice the food line. The L S A found for 1994 that the rate o f extreme poverty was quite low (2.5 percent), that food poverty was 7.3 percent, and vulnerability was 36.3 percent. The natural question is then raised, how did poverty change from 1994 through 2001, after Turkey had experienced the twin blows of the 1999 earthquake and the prolonged financial crisis o f 2000-2001. This report can answer this question to a certain extent, but there are important caveats about the 2001 survey that must be taken into account. First,the 2001 household consumption and income survey (HCIS) was not fielded by Turkey's State Institute o f Statistics (DIE), for practical and logistical considerations. The HCIS had to be undertaken rapidly to provide the Turkish government with sufficient data for developing a targeting mechanism to be used in the conditional cash transfer and local initiatives components o f the SRMP. Due to the press o f its regular reporting operations and methodological concerns, DIE was not able to field a HIES in 2001. As a matter of fact, DIE began an HIES in 2001 but stopped collecting data after the first three months, out o f concerns that the data taken in the immediateaftermath o f the February 2001 devaluation of the Turkish lira would not be reliable enough for constructing new weights for the consumer price index (CPI). As a result of these decisions, the World Bank (with partial financing from the Government of Japan) moved 1  forward to conduct a survey usingindependent consultants. So it i s immediately evident that comparisons between the 1994 survey, conductedby DIE with its methodology, and a survey conducted by independent consultants, are likely to vary significantly interms o f scope, sample, and methodology, which was indeed the case for the 1994 HIES and the 2001 HCIS. The goal o f the 1994 HIES was to provide weights for the CPI. The goal o f the 2001 HCIS was to obtain data on consumption to be usedto create a scoring formula for targeting the SRMP. The surveys were o f considerably different scope, with the 1994 HIES beingmuchlarger (25,000 households) than the 2001 HCIS (4,200 households). Next, the sample for the 2001 HCIS was completely different than the sample for the 1994 HIES. The 1994 HIES sample was drawn from the 1990 census sampling frame. The 2001 HCIS sample was drawn from a sample provided to the consultants by SIS. Unfortunately, the 2001 HCIS sample varies somewhat from the 2000 census results in one critical aspect. According to the preliminary figures for the 2000 census, the population in Turkey is 35 percent rural and 65 percent urban (where rural i s defined as those living in villages) whereas the HCIS sample has a much lower share o f rural (21 percent) init than does the population (Table 1.2)' Although the consultants maintain that the 2001 HCIS sample was representative for bothrural and urban areas since rural areas are muchmore homogeneous and thus can be under-sampled without loss o f representativity, the discrepancy between the HCIS sample and the census i s worrisome. It is o f course possible to re-weight the HCIS sample using population weights, but if the HCIS sample i s not representative, re- weightingon its ownwill not provide a solution. During consultations, questions were raised about how the 2001 HCIS sample was reduced from the original sample of 7,000 provided by the State Institute of Statistics. A short appendix was prepared by the consultants in response to this concern and the appendix also includes the technical objections raised by State Institute o f Statistics andby State PlanningOrganization duringthe consultations. It is very important to understand this caveat. If the 2001 HCIS sample is not representative for whatever reason, then the findings inthis report would not be accurate. Certainly, more than the usual degree o f caution i s necessary and these findings should not be taken to be definitive o f levels or trends, but rather indicative only. This is a problem specific to the circumstances surrounding the 2001 HCIS and should not be an issue inthe future, since the State Institute o f Statistics i s undertaking a 'In this table and throughout the study, when HCIS data are presented, they are on population basis. This means that household-level variables such as location are weighted by household size to generate population basis figures. Individual-level variables o f course are not weighted. Since responses vary by question and since not all individuals in households with more than 10 members were interviewed, there are slight discrepancies between parameter estimates from the two sources, household and individual data. 2  HIES for 2002 that will be used incooperationbetween the Government and the Bank to generate poverty estimates that will be comparable to the 1994 data. On methodology, it is important to understand the financial and other constraints faced by the 2001 HCIS. To keep within a reasonable budget for the HCIS, some difficult decisions had to be taken. For example, it was too prohibitive in cost to plan to visit each household more than once. The 1994 HIES enumerators visited households each month, and collected the consumption and income information from written diaries kept by the households. For the 2001 HCIS, the recall method was used and households were visited only once. The recall method in comparison to written record-keeping can lead to "telescoping" whereby respondents forget older expenditures, and can lead to under-estimation o f consumption and income (Deaton 1997). However, telescoping does not appear to have been that much of a problem for the 2001 HCIS-a comparison was undertaken for the 1987 and 1994 HIES and the 2001 HCIS consumption and income aggregates as compared to data from the national accounts (Table 1.3). Here the 2001 HCIS accounts for less than the share o f the national accounts aggregates than did the 1994 HIES, but not catastrophically SO.^ A second serious area o f methodological difference between the 1994 HIES and the 2001 HCIS i s that the latter survey lacks the unit value prices that were calculated for the former in setting most o f the poverty lines for the LSA analysis. Unit value prices are calculated from consumption surveys whereby the amount spent on an item (very narrowly defined) is divided by the quantity reported as purchased. These unit value prices are not typically 100 percent o f the level o f prices collected in the CPI survey. In particular, survey unit value prices reflect more advantageous prices offered to households by their local grocer or small-scale distributor as well as bulk discounts and buyingon credit. As a result, survey unit value prices are typically somewhat lower than prices collected in large stores and urban markets for the CPI. Importantly, unit value prices for rural areas are much below those for urban areas, while the CPI prices are not collected in rural areas. As a result, poverty will be measured as lower when consumption or income i s compared to the lower survey unit value prices than when comparedt o (urban) CP Iprices and conversely, usingurbanCPIprices even for only urbanareasis likely to exaggerate somewhat the true extent o furbanpoverty. Again, owing to the constraints, it was not possible to field a full-blown HIES in 2001, and so the HCIS used a truncated consumption module that did not collect data on unit value prices for individual items. As a result, when consumption is compared to the food basket (even for urban areas only), the HCIS is 1ikely t o show s omewhat higher poverty than if survey unit values could have been used. Thus, the findings inthis report about food consumption poverty in urban areas o f Turkey probably represent the outer range for poverty and actual food poverty may have been somewhat less than estimated herein. At the same time, survey unit values and urban CPI prices move together and are closely related, so this caveat should not be interpreted to mean a sign change or 2Additionally, the average share of food expendituresintotal consumption inthe HCIS was 34.4 percent, which compares well to the 1994data of 36.3 percent, or the share of food expenditure inGDP of 35 percentin2000 (data for 1994and GDP share fromthe SYDTF). 3  significant change in magnitude, but rather should be understood as to suggest that the exact figure i s more in the nature o f a bound for the true underlying poverty it seeks to measure. Table 1.1 shows the food quantities and the regional CPI prices for the 7 regions o f Turkey that were used to price out the urban food poverty line. Finally, the consumption estimates used herein did include the imputed value of food consumed from own-production for those households (both urban and rural) which reported such con~umption.~HCIS broad category unit values (e.g. bread and bread products not by specific types o f bread as in the HIES) were used to impute food consumption. Since there were few observations, no attempt was made to differentiate this imputation by region. As a result, consumption may be overstated in less expensive areas such as the Southeast and understated inmajor metropolitan areas like I~tanbul.~ Another technical limitation o f the 2001 HCIS i s that space was made available to collect individual information for 10 family members, but a small number o f households reported larger family sizes. Inthese few cases (1.8 percent), individual information was not available for every household member, but only for the first 10 recorded. Additionally, in some cases, individual data on age (14.5 percent o f individuals) and gender ( 1.7 percent) w ere missing for s ome household members. T he consultants are working on collecting this informationretrospectively and this should be corrected inthe final version o f the database. Inthe interim, when age and gender data were needed for the food poverty line (which used the same UN Food and Agriculture Organization (FAO) equivalence sc ale as LSA 2 0005),the missingdataw ere imputedbased o nthe sample average. Finally, the report draws on three sources o f qualitative information: (i) Ayata and Ayata (2002, a background paper in Volume Two) summarized the finding of the 120 case studies undertaken in conjunction with the quantitative HCIS; (ii) field visits and informal focus groups conductedby SRMP team members duringthe preparation and initial supervision o f the SRMP (cited as SRMP Field Notes); and (iii)an interim beneficiary assessment on the rapidresponse component o f the SRMP. 3 Consumption i s a much better indicator of household welfare than money income-it includes the imputedvalue of food consumedfrom own productionand avoidsthe under-reporting of informal earnings so prevalent inmost countries (Deaton and Zaidi 2002, World Bank 2001, World Bank 2000, World Bank 1993, World Bank 1990, Deaton 1997, Hentschel and Lanjouw 1999). Poverty line methodology from Ravallion 1992, 2000). 4 Consequently,underlying poverty rates for the Southeast might be higher than shownbelow and lower for Marmara region. 'The F A 0 equivalencescale was the same one usedinLSA 2000, namely: childrenunder 5 0.64 children 5-11 1.oo male adolescent 12-17 1.00 female adolescent 12-17 0.84 prime age male 18-39 1.00 primeage female 18-39 0.84 older male 40+ 0.88 older female 40+ 0.76. 4  2. Macroeconomicanalysis Introduction This chapter seeks to asses the cumulative macroeconomic impact on the Turkish people of the series o f internal and external shocks that have hit the country since 1999. When the Marmara earthquake hit inAugust 1999, Turkey was already suffering from an economic slowdown caused by the Russian crisis and political turmoil which had led to early elections in mid year. Following a short lived consumption boom under the crawling peg based disinflation program in early 2000, Turkey experienced financial turmoil inNovember 2000, and then a full blown currency crisis in February 2001, Just as the strengthened economic program put in place in response to the February crisis began to show some promising results, the attack on the World Trade Center on September 11,2001 shook Turkey with another external shock with a loss o f tourism and export revenue, and concerns by investors about the possibility o f broader conflict in the Middle East. The economic crisis o f 2001 was exacerbated by a serious drought as well. The new macroeconomic framework and renewed program o f structural reforms pursued by the Government since December 2001, backed with exceptional support from the international financial institutions, has achieved financial stability and there are encouraging signs o f recovery in 2002. However, the economic and social cost o f the crises has been significant interms o f lost growth and highunemployment. The impactof the Marmaraearthquake6 The 1999 earthquake struck at Turkey's densely populated industrial heartland along the coastline o f the Marmara sea south o f Istanbul. The damage was severe and the economic consequences were felt throughout Turkey. The affected area contributes This section draws upon andupdates the assessment presentedin"Turkey: Marmara Earthquake Assessment", World Bank, 1999. 5  about 7 percent o f Turkey's GDP. The outlying suburbs of Istanbul, which accounts for about a quarter o f national output, were also affected. The World Bank assessment team estimated that output as a whole would be negatively impacted in 1999, but then reconstruction expenses in 2000 would lead to a pickup in growth. As can be seen from Table 1.1, this is what happened, although the impact of the earthquake on 1999 output appears to have been more severe than expected. Output recovered strongly in 2000, as the positive impact o f reconstruction expenditures was multiplied by a consumption boom driven by the fall in interest rates under the disinflation program introduced in December 1999. T he recovery was industry 1ed, but the other sectors w ere recording positive growth by the second quarter o f 2000. Source: State Institute of Statistics (SIS) " - Impact on employment T he 1999 earthquake had a marked impact o n employment. Employment fell by 10.5 percent betweenOctober 1999 and April 2000 (Table 1.2). The decline was particularly marked in agriculture and construction, sectors where the poor are concentrated. Only services employment rose during this period. Employment then 6  quickly recovered in the second quarter o f 2000, rising to levels seen prior to the earthquake, ledby strong recovery in agriculture and construction. Table 1.2: QuarterlyEmployment 1998-2000 1998 1999 2000 I ( I11 I I11 Employment (inthousands) 20,351 21,393 21,590 21,236 19,006 Emdovmentinamiculture 8.145 II I I1 8,777 I I 9.148 1II 8.595 III I 6.284 3,661 3,614 3,495 3,664 3,449 3,814 3,482 3,436 3,337 3,543 3,295 3,645 Employmentinservices 7,319 7,647 7,705 7,631 8,304 8,367 Unemploymentrate 6.94 6.70 7.93 7.37 8.30 6.23 Source: State Institute of Statistics (SIS) FinancialTurmoil and Crisis ............................................................................................................................................... With the task o f providing immediate relief to the earthquake victims well underway, in December 1999, the Government launched an ambitious exchange rate based disinflation program with support from the IMF, together with a series o f structural reforms backed by the World Bank. The program got off to a fast start with a sharp fall in interest rates fueling a strong recovery during the first three quarters of 2000. However, the economy began to overheat and was hit by severe financial turmoil in November 2000 when a medium sized private bank ran into a liquidity crunch. The Government responded quickly and the immediate turmoil subsided. However, confidence in the crawling peg exchange rate had been undermined. A public airing o f political tensions inFebruary2001 sparked a full fledgedcurrency crisis which forced the Government to abandon the disinflation program and float the Lira. A severe crisis ensued and the economy fell into a deep recession with high interest rates making it difficult to borrow, a volatile exchange rate and renewed surge in inflation making it difficult for firms to plan. In the meanwhile, overshadowed by financial market developments, Turkey was facing a drought that is likely to have particularly affected the poor. The macroeconomic consequences of these events i s assessedbelow. Impact on Growth The impact o f the crisis was most immediately felt in the construction sector as interest rates spiked to levels over 100 percent (Table 1.3). However, the malaise soon hit industry with bank credit drying up and the exchange rate becoming very volatile. Normally, the agriculture sector would not be expected to be highly correlated with a banking-led crisis, but 2001 was also a drought year, and after the first quarter, agriculture was in increasing decline relative to the same quarter o f the previous year. T he economy had enteredinto a s erious recession which resulted ina record decline inGNP o f 9.4 percent for 2001 as a whole. 7  Source: SIS Impact on employment By mid 2000, unemployment had started to decline as the economy was heating up. However, after the February crisis hit, the unemployment rate increased once again. The severity o f the recession was reflected inlargejob losses. By the thirdquarter o f2001,the unemployment rates had begunto climb dramatically, as employment started falling, notably inagriculture and construction. Construction was hit particularly hard, with over ?4millionjobs lost from the third quarter o f 2000 to the first quarter o f 2002. By the first quarter o f 2002, the unemployment rate had doubled since the lows recorded right before the November 2000 financial turmoil. Only employment inservices remained buoyant. The unemployment picture would have been significantly worse if the service sector had not been able to pick up some o f the slack. More generally, it is important to note that the employment impact o f the crisis on the poor i s likely to have been disproportionately large as the poor depend more heavily on informal employment which is unlikelyt o be fully captured by the o fficial s tatistics. Informal employment was probably hit especially hard by the crisis as indicated by the heavy impact on agriculture and construction where much o f the informal employment is concentrated. As in most countries, informal workers in Turkey do not benefit from unemployment insurance and other similar social insurance mechanisms. Employmentinservices 8,277 8,341 8,298 8,138 8,298 8,511 8,414 Unemploymentrate 5.63 6.33 8.60 6.90 8.02 10.58 11.76 The macro picture o f the severe impact o f the 2001 crisis and accompanying shocks on growth and employment i s confirmed by micro data collected by chambers o f commerce and industry in Turkey at various times during 2001 to assess corporate distres~.~The Ankara Chamber o f Industry (ASO) survey in the fourth quarter o f 2001 These data are from the "The Republic of Turkey: Corporate Sector Impact Assessment", World Bank, draft, May 2002. 8  estimated the average decline inthe workforce inAnkara was 5.6 percent in the third quarter, a very sizeable reduction for one quarter. The more representative Union o f Chambers o f Commerce, Industry, Maritime Trade and Commodity Exchanges o f Turkey (TOBB) survey asked a question about workforce size over three intervals. Table 1.5 shows that, in each quarter, 55-60 percent o f the firms reduced their workforce relative to the previous quarter o f 2001. The surveys also found that small firms were especially affected which i s to be expected as they typically have less capacity to weather crises than large companies. In particular, smaller firms showed limitedability to avoid lay-offs. Firms o f all sizes typically resort to lay offs as the last measure to cut costs and, in general, large firms are more able to cope with resort to other, less drastic cost cutting measures. The TOBB survey found that 62 percent o f small firms surveyed decreased their workforce in the last quarter o f 2001 as opposed to 46 percent o f large firms. 1st Survey 2nd Survey 3rd Survey Responses (inpercent) Jan.-Mar 2001 Apr.-Sept. 2001 0ct.-Dec. 2001 - Increased 2.47 5.02 4.5 1 B-No difference 41.10 35.07 38.52 C- Decreased 56.43 59.91 56.97 Total 100.00 100.00 100.00 Source: Draft Corporate Sector Impact Assessment, World Bank, May 2002 Cumulative Social Impact The period under discussion was a difficult one for Turkey. Table 1.6 suggests that per capita GNP in constant terms fell by 13 percent between 1998 and 2001. The economy contracted sharply in 1999. While output recovered strongly in 2000, growing macro imbalances under the crawling peg set the stage for crisis in 2001. Hit by financial crisis and shocks from the drought and fall-out from September 11, Turkeyrecorded its worst economic performance in 2001 since independence. The unemployment rate decreased in 2000, although the total number o f people employed fell as well. Unemployment rose sharply in 2001 and this trend continued in early 2002. Over the period, agriculture and construction, the sectors most likely to employ the poor, shed large numbers of jobs. Inflation remained persistently high with high within-year fluctuations. To the extent that the increases in inflation during the period were unanticipated, they may have had an adverse impact on poverty.8 Unfortunately, wage data are not available for the sectors where the poor are likely to be concentrated: construction, some services and agriculture. Manufacturing wages remained relatively constant in real terms through 2000 as nominal wage increases kept pace with inflation, but then declined sharply through 2001. Real wages in manufacturing See "Turkey: Economic Reforms, Living Standards, and Social Welfare Study", World Bank, January 2000. 9  1998 1999 2000 2001 GNP per capita (constant 1987 TL) 1,836,704 1,694,912 1,766,124 1,572,820 GNP growth (%) 3.9 -6.1 6.3 -9.4 Agriculture 8.4 -5.0 3.9 -6.1 Industry 2.0 -5.0 6.0 -7.5 Construction -0.3 -12.7 4.4 -5.9 It is difficult to quantify the cumulative social impact o f the crisis and shocks. For the 15,000 people killed inthe earthquake and their friends and families, the cost i s limitless. Even simply calculating the economic impact is difficult because the counterfactual is unkown--what would have happened if the crisis had not happened? Nevertheless a rough indicator o f the wealth loss can be obtained by comparing projections made before the earthquake with actual outcomes. In Table 1.7, actual GNP per capita i s compared with potential GDP per capita computed by applying the growth rates projected prior to the crisis to the 1998 actual values. These projections were o f course themselves subject to considerable uncertainty. They assumed no major macroeconomic crises, and incorporated assumptions about the impact of expected reforms on growth as well as assumptions about the direction of the world economy. With these caveats in mind, the indicative results suggest that GNP per capita would be about 23 percent higher had the economy progressed consistent with the December 1998 projections. InU S dollar terms, GNP per capita in 2001 would have been higher by $703 usingthe Bank's Atlas method. Table 1.7: Actual and Proiected GNP Der caDita 1998-2001 1998 1999 2000 2001 Actual GNP (per capita constant TL) 1,909,918 1,766,837 1,859,910 1,659,199 1998 Projections GNP (per capita constant TL) 1,909,918 1,909,918 1,967,215 2,038,035 Actual GNP per capita (US$,Atlas method) 3,170 2,880 3,080 2,680 1998 projections (US$, Atlas method) 3,170 3,170 3,265 3,383 10  3. Extreme poverty. As was found to be the case for the 1994 data, in 2001, very few Turkish households were found to be below the World Bank's absolute poverty line o f U S $1 per person per day.' In 2001, less than 2 percent of the population had per capita consumption underU S $1 per day, and only 3 percent had per capita income underU S $1 per day. These results are basically unchanged from 1994, where 2.5 percent consumed under U S $1 per person per day. There i s a decline in measured consumption-based extreme poverty (from 2.5 percent in 1994 to 1.8 in2001) but these numbersare so small, this difference is well within the standard error o f the samples. If census population weights are used, the extreme poverty rate moves from 1.8 to 1.9 percent. It is difficult to meaningfully characterize such a small share o f the sample population. Four dimensions are highlightedhere: number of children, location, assets, and education. However, since only 2 percent o f the sample is being studied, these findings shouldbeunderstoodas indicative oftrends only, andnot definitive. As t o be expectedwhen using a per c apita s tandard, 1arger families are poorer (Lanjouw and Ravallion 1995, Lanjouw, Milanovic and Patemostro 1998). The average household size i s 3.9 members, but the average household size o f the poor (consumption under $1per person per day) is 6.3 members (for income, 6.5 members). Most of these extra dependents are children, andthe poverty rate g enerally increases w ithadditional children (Table 3.1), although there is a discontinuity between four and five or more children. Absolute poverty in Turkey is concentrated in the South-East Anatolia region (Table 3.2). Overall, 1.8 percent were poor when per capita household consumptionwas compared to the World Bank poverty line, but in Southeast Anatolia, the rate o f extreme poverty was nearly 5 times the national average. Of the less than 2 percent extremely poor, nearly half(46 percent) ofthem lived in Southeast Anatolia. Extreme poverty i s basically the same in rural and urban areas-nly about 2 percent o f the population is extremely poor (1.8 percent for urban areas and 2.1 percent for rural areas, difference not significant). Education (or lack thereof) i s correlated with extreme poverty. Overall, about 6 percent o f the household heads surveyed reported that they could not read or write. The poverty rate (as a percent o f the population) o f such households was 4.4 percent (as opposed to the average poverty rate o f 1.8 percent). The poverty rate of households with primary-only education was 3.0 percent, while the poverty rate o f households with tertiary educationwas 0.5 percent. To summarize these results, a multivariate probit regression was run to measure the impact of these extreme povertycorrelates on the probabilityo fbeingextremelypoor. Not many variables were found to be estimated reliably, this obviously related to the very PPP line discussedinWorld Bank 2001, World Bank 2000, World Bank 1993, and World Bank 1990. 11  small number o f observations. However, there were four variables found to be significant at the five percent level: 0 location in the Southeast which increased the probability o f being extremely poor by 3 percent location inBlack Sea which reducedthe probability o fbeing poor infinitesimally 0 children o f both genders, also increasing the probability o f being extremely poor infinitesimally. 12  4. Urban food poverty It is not possible to exactly replicate the LSA findings for food-based poverty as explainedinthe data and methodology s ection, s o the findings here about urbanfood poverty should be understood as an outer bound on what poverty inTurkey "really was" in 2001. Even with this caveat, it is clear that such poverty inurban areas in Turkey in 2001 was much higher than in 1994. In 1994, for the country as a whole (urban and rural), 7.3 percent o f the population had per equivalent expenditures below the imputed value o f a food basket (LSA 2000) while the poverty rate was 6.2 percent inurban areas only. In 2001, 17.2 percent o f the urban population had per equivalent expenditures below the imputed value o f a food basket. Even taking this 17.2 percent as an upper bound on poverty in Turkey, it i s clear that the combined effects of the earthquake and financial crises caused urbanpoverty to widen in2001. In 1994, rural and urban food poverty rates were quite close (8.5 and 6.2 percent respectively) and the process o f rural to urban migration noted in the LSA 2000 has continued, althoughthere i s qualitative evidence o f some reverse migration in2001. It is difficult to argue a priori what the effect of the crises has had on rural poverty-an one hand, income inequality has not declined while real income has, so rural food poverty maybe expected to increase, but not necessarily more than urban food poverty has, since consumption o f self-produced food i s likely to be relatively unaffected by the crises and forms the bulwark o f rural consumption expenditure on food (and therefore on consumption as a whole). Furthermore, more than four-fifths o f the sample was drawn from urbanareas, so the weight o f rural poverty intotal poverty for the HCIS is relatively low. This section presents a profile o f the food poor inurban areas in Turkey in2001, Food poverty is analyzed along many dimensions o f poverty, but the most striking correlate o f urban food poverty is region, paralleling the findings for $1 per person per day poverty, but showing an even more extreme effect o f regional location. Household size is also strongly correlated with urbanfood poverty, even though an equivalence scale was used (which should reduce the effect o f the per capita scale used for extreme poverty). Aside from subjective indicators, few correlates o f urban food poverty produce such striking effects on poverty rates as location and household demographics, other than importantvariables such as educationo fhouseholdhead or unemployment. Location The most obvious correlate o f urban food poverty in Turkey i s regional location (Table 4.1). To some extent this could be an artifact o f the sample, which was not designedto be representative for the seven regions o f Turkey. However, these findings are so striking that they are included herein, albeit with a very strong caveat. The Southeast accounted for less than 7 percent o f the population but for just over one-quarter o f the urban food poor (Figure 4.1). The Southeast i s a predominately rural area and there i s little local industry, so it i s not surprisingthat the region would account for such a large share o f urban poverty. As discussed in section 8, the Southeast has been the major 13  source for rural to urban migration inTurkey, but the migrants tendto leave the Southeast entirely for the Marmara area, whichhas the greatest number ofjob opportunities." Figure 4.1 Turkey: Region and Urban Food Poverty 40 35 30 25 ' 0 Total 20 15 Poor 10 05 10There is somemigrationto the larger cities ofthe Southeast such as Diyarbikir (whichhas a street childrenproblem) andGaziantep,but this is dwarfedby migrationto Istanbuland Marmararegions. 14  Demographic Variables As was the case with extreme poverty, urban food poverty is clearly associated with having a larger household (Table 4.2)," but there is a difference in the type o f dependents. Having children increases the risk o f food poverty in the urban population much more than having an elderly household member, and the elderly (aged 60 and older) have a much lower rate o f poverty than do children (Tables 4.3 and 4.4). There i s no major effect o f gender. Urban households with four or fewer members are less likely to be poor than average, and single person households are virtually never poor (Table 4.2). This latter point relates to the social fabric o f Turkey-only 1percent o f the urban population lives ina singlepersonhousehold, and livingthis sort oflifestyle is clearly achoice enabled by income, not a necessity. Larger urban households are poorer than smaller households, and it can be demonstrated that the type o f dependent matters-children are poorer than the elderly, and contribute more to urban food poverty. For household size, the break-point where the average i s exceeded is four members, but for children, this effect happens after the first child (Table 4.3). There are two points o f discontinuity in this correlation (more children, higher poverty): the poverty rates for zero children and only one child are practically the same, as i s the case for four or five or more children (Figure 4.2). Figure 4.2 , Turkey: Number of ChildrenandUrbanFoodPoverty ~ 50.0 40.0 L 0 * 30.0 2g 20.0 10.0 0.0 0 1 2 3 4 5+ Number ofChildren However, there i s essentially no difference in poverty rates between urban households with elderly members or without (Table 4.4), nor between having one or more elderly members. Thus having an elderly dependent does not put an urban household at any higher risk for poverty, but having more than one child clearly does. " This was the case even though the FA0equivalence scale was used, which should reduce the effect of the per capita scale used for extreme poverty. 15  Another way to look at the relative contribution to poverty o f children and the elderly is to look at the poverty rates by age group (Table 4.5). There is a correlation between age and urban food poverty that speaks to the elderly's relative insulation from poverty-the poverty rate o f individuals aged more than 60 is less than half the rate for children aged 0-9. There is a negligible gender differential, meaning that there was no effect o f gender on the risk of food poverty in the urban population (18 percent o f urban men were poor versus 17 percent of urban women). Further, there are too few single elderly females (only 0.5 percent o f the urban population) to reliably generalize, but it is quite indicative that only 1 o f the 56 single female elderly surveyed was poor. Other female-headed urban households (adult females with at least one child who did not report an adult male member) do not seem at risk for poverty, their poverty rate was only 20 percent as compared to the average o f 17.2 percent, and there are few such households- only 1.9 percent o f the urban population. Many o f these households are still likely to be supported by males-nearly half o f them were married (the husband probably guestworking elsewhere), and the other half were widowed. 16  Education As is true in virtually every country, education in Turkey is closely correlated with welfare. Those few lucky enough to receive a university education inTurkey (about 11 percent o f the urban population) are not very likely to be poor-their food poverty rates are all below 8 percent (Table 4.6). Interestingly enough, neither primary education nor lack o f literacy has a huge impact on poverty rates. Of course household heads with primary education are poorer than average (Table 4.6), and those limited to primary school graduate have a urban food poverty rate nearly 60 percent higher than total. For all urban individuals reporting literacy inthe sample (includes children), the poverty rate was 16 percent poor while it was only 26 percent for those who were illiterate. This i s not as sharp a distinction as the poverty rates between families without children and those with four children, for example. UnemploymentandEmployment Unemployment i s o f course associated in Turkey with poverty, but as inthe case o f education, not as strongly as one might expect. The poverty rate for urban households with employed heads was 16.2 percent (as opposed to the average o f 17.2 percent, essentially the same) while the poverty rate for households who reported that their head was unemployed was 30.6 percent. This i s o f course a noticeable difference-the poverty rate i s almost twice as highfor unemployed heads, but it i s not as sharp a poverty premium as observed from having many children where the relative poverty risk o f 5 or more children is nearly 4 times that o f not having any children. The finding on unemployment is likely to be somewhat understated for several reasons: (i)it is shameful to admit unemployment-only 7 percent of the urban population reported that the head was unemployed; (ii) the fluid nature o f casual given and t emporary employment inthe informal sector inTurkey many respondents w ould consider themselves to be employed even ifthey were not earning income at the particular moment of the interview; and (iii)in urban areas near agricultural areas, respondents would report that they are employed ifthey were working on their land plot. 17  Interms of employment or "main activity" as queried in the HCIS, predictably enough heads who worked as merchants or wage workers had poverty rates substantially below those o f farmers, stockbreeders, or interestingly enough, in the only aspect which shows any noticeable gender differentiation-housewives (Table 4.7). It is important to note the large share of the urbanpopulation with retired heads-as discussed under risk mitigation, Turkey has a generous retirement system for the formally employed and civil servants so it i s possible for many to retire early. Also, as noted above, urban poverty is correlated with youth, and well-being with age as reflected inthe poverty rate for retired heads. Housingand HousingAttributes Home ownership in urban areas o f Turkey is associated with a lower rate o f poverty (but not that significantly, a full 15 percent o f owners were poor versus 21 percent o f tenants, Table 4.8). Predictably enough, living in a gecekondu (slum or shantytown area) has a relative risk o f urban food poverty three times higher than that o f apartment dwellers (Table 4.9). The presence o f amenities such as private tap or sewage hook-up i s associated with a lower poverty rate than the absence o f the amenity. Against the general background o f some effect on poverty rates but not large effects, a few observations from the annex are noteworthy. M ost urbanhouseholds w ere 1ivingindwellings that w ere constructed o f concrete, but the 10 percent in wood, earth brick, or stone houses were poorer (Table 4.10). Urban households with latrines were poorer than households with flush toilets (Table 4.11). Less than 2 percent o f the urban households lacked an indoor 18  tap (Table 4.12), and only 0.4 percent had no electricity (Table 4.13). Respondents indicated more than one source o f fuel for both cooking and heating, there is consequently little difference in poverty rates by source (Tables 16 and 4.15). Only 5 percent o f urbanhouseholds lack garbage collection (Table 4.16) and only 8 percent lack public sewerage (Table 4.17). ConsumerDurablesandAssets The urban Turkish population is relatively affluent-31 percent of urban households have a dishwasher, 20 percent have a VCR, 26 percent have a car, and 11 percent have a computer (Table 4.18). For these four durables, the poverty rate for those inpossession ofeachwas 5-8 percent-which means that interestingly enough, a few urban poor have been able to obtain these items as well. Four other durables are possessedby 80-96 percent or more o f urban households, including radio, tape recorder, refrigerator, and carpet. The poverty rates o f those who do not possess these are above the rates for those who do, but again, on the range o f 10-15 percentage points which i s not a huge difference. The most significant asset for the urban population i s a summer house (Table 4.19)-nearly 3 out o f 5 urban households have a summer house, and these are not the poor (their poverty rate is only 11 percent). The other assets are possessed by 10 percent or less o f the urban population, and the strongest predictor o f wealth i s owning a foreign 19  currency account-only 2 percent of foreign currency holders are poor, but only 6 percent o f urbanhouseholds are lucky enough to possess one. CommunityInfrastructure The urban areas o f Turkey seem to be relatively well supplied with infrastructure-99 percent have a retail shop in their community, 76 percent a health clinic, 88 percent a primary school and the same percent have telephone access, and 94 percent have both garbage service and an asphalt road (Table 4.20). However, there seems to be inequality in the distribution o f these amenities-in every case, the food poverty rate i s lower for those urban households with the infrastructure than for those who lacked the infrastructure. 20  Summary To summarize these results, a multivariate probit regression was run to measure the impact of these poverty correlates on the probability o f being poor for the urban population (Table 4.21). Many variables in the specification were significant at the 5 percent level, including all the regional dummies, several o f the assets discussed above, andthe householddemographic variables. The single greatest impact on the probability o f the urbanhousehold's being poor was location in the Southeast, which increased the probability o f being poor by 34 percent. This i s a huge margin, much larger than those observed for poverty attributes o f some transition countries (Braithwaite, Grootaert and Milanovic 1999). Three other factors increased the probability of being poor by 10-15 percent, including location inthe East, gecekondu house, or living in room in a shared house. Household demographic variables (basically the number and kind o f dependents) increased the probability of being poor only slightly-2 to 4 percent. Assets such as dishwashers, cars, and computers also had a small correlation with lack o f poverty, reducing the probability of being poor by 2-7 percent. Perceptions of Poverty Most o f the information sources for perceptions o f poverty are qualitative in nature; however, the HCIS did include several questions on subjective evaluations o f poverty and changes inliving conditions and the results o f these questions for the urban population can be broken down by poverty status. The urban food poor are more likely to identify themselves as "poor" or "below average" but many non-poor also assessed themselves as "poor" or "below average" (Table 4.22). About one-quarter o f the urbanpopulation evaluated themselves as "poor" 21  and the same share for "below average". The poverty rate o f those who assessed themselves as poor was 29.6 percent, about 75 percent higher than the average poverty rate, but still leaving a sizeable chunk o f people whose self-evaluations are not matched by their revealed spending and food consumption from own production as captured in their household consumption levels. Nearly halfthe respondents considered themselves average and only 10 percent o f these were poor. There was greater congruence between those who assessed themselves as "above average'' or "rich" - no one who was poor assessed themselves as rich, but 5 percent o f the few who deemed themselves above average were objective poor by the urban food poverty criteria. The perceived impact o f the crises can also be examined from HCIS data (Table 4.23). A small share o f the urbanpopulation felt that they were better off in2001 than in the previous year, but 80 percent said they were "worse o f f ' or "much worse off." Interestingly enough, although only 3 percent said they were "much better off' more than two-fifths o f these w ere poor. P erhaps these 1ucky few w ere the recipient o f positive idiosyncratic shocks such as finding a job or reduced dependency burden because children have grown. A significant share o fthe urbanpopulation reportedthat they "often" or "always" experienced difficulties satisfying their household food needs, and o f these, more were poor than the average poverty rate (Table 4.24). About 15 percent "never" experienced problems, only 4 percent o f those who never hadproblems were poor. The poor appear to be slightly more pessimistic about their community than their own self-evaluation-an even higher share o f the urban population felt their community was "much worse" o f f or a "little worse" off than in the preceding year, but there was little variation between the poverty rate o f those who answered negatively and the average poverty rate (perhaps because most people responded negatively (Table 4.25). 22  5. Vulnerability In studies of panel data, vulnerability is usually defined dynamically, relating to the probability o f falling under the poverty line in the second or subsequent observations. In the LSA 2000, vulnerability was defined statically, as households which had per equivalent expenditure under a "vulnerability" line, which was set equal to twice the food poverty line. Lacking panel data, the L S A 2000 definition o f vulnerabilitywas replicated for the urban population o f Turkey inthe HCIS 2001. Using this definition, vulnerability increased from 36.3 percent of the population in 1994 to 56.1 percent o f the urban population in 2001. It should be noted that while vulnerability did noticeably increase from 1994 to 2001, its rate o f increase was markedly less than that for urban food poverty. At the same time, the absolute numbers o f those who were vulnerable in 2001 were muchmore than in 1994. As was the case for food poverty, vulnerability is sharply higher inone region of Turkey-the Southeast, where 93 percent o f the urban population had per equivalent consumptionbelow the vulnerability line (Table 5.1). At the same time, a plurality o f the vulnerable live in the most populated region, Marmara. Vulnerability was also elevated inthe East,but was about average for every other regionexcept the Black Sea whichwas the least vulnerable (this last finding was also markedly the case for food poverty). Vulnerability i s also associated strongly with household size (as was the case for food poverty) and the least vulnerable households were those o f only one member (Table 5.2). As noted above, it i s such an unusual occurrence in Turkey for a single person to live alone (single person households account for only 1.3 percent o f the population and slightly less than halfof these are elderly women-presumably wealthy widows) that the few cases in the HCIS sample are most likely living alone out of choice and not necessity. There i s basically a monotonic relationship12between vulnerability andhousehold size for the urbanpopulation--adding additional dependents means that the household is more likely to have per equivalent consumption under the vulnerability line' (Figure 5.1). Households with 4 or less members are less likely to be vulnerable, indicating in Turkey that the first two dependents do not seriously impact on the household's dependency burden, but having more dependents than two does present a clear burden to the household. There is also a difference between the kindof dependent(elderlyvs. children) and vulnerability. C hildrenare more 1ikely t o be associated w ithurbanv ulnerability-the vulnerability rate i s below the average for no children or one child, but at two children, 12The discontinuity at more than 10 members reflects the l o w percentage o f these kinds o f households-in the urbanpopulation there was only one household with 11 members and one with 13, both vulnerable. l3An equivalence scale was used so this is not likely to be the "per capita" effect demonstrated inthe literature (Lanjouw and Ravallion 1995) but this effect could be muted if stronger assumptions about scale effects were used (Lanjouw, Milanovic, and Paternostro 1998). 23  the vulnerability rate is above average and increases monotonically with additional children (Table 5.3). Figure 5.1 Turkey: Vulnerability andHouseholdSize 100.0 sE a, 80.0 2a, 60.0 P - -+- Series1 40.0 B 20.0 0.0 1 2 3 4 5 6 7 8 9 1W Number inHousehold Interestingly enough, the age o f child appears to be associated with increased vulnerability (Figure 5.2), which probably reflects the impact o f the number o f children more than anything else. Households with children aged 5-14 are more vulnerable than families with children aged 0-4, but households with older children are more likely to have younger children as well. The age effect noted inurban food poverty i s also a factor in urban vulnerability-individuals aged sixty and above had the lowest rates o f vulnerability inthe urbanpopulation (Figure 5.2). Figure 5.2 Turkey: UrbanVulnerability andAge 24  Vulnerability is not as closely associated with education as was food poverty (this probably reflects the fact that most o f those with inferior education are already poorer) and vulnerability rates for those with primary education are only about 20 percent higher than average (Table 5.4). Again as the case with food poverty, vulnerability was lowest for those with university education-those with advanced degrees were 84 percent less likely to be vulnerable than average. Other factors that were important for food poverty such as main activity o f the head and whether the head was unemployed or not were still important for vulnerability, but relatively less so. For example, the food poverty rate for households with self-reported unemployed heads was 77 percent higher than average, while the vulnerability rate for such households was only 32 percent higher than average. Heads employed in agricultural activities were more vulnerable than those in other activities (Table 5.5), but the marginfor vulnerability was not as wide as observed for food poverty. 6. Inequality Inequality as measured in the 1994 HIES and the 2001 HCIS i s basically unchanged-it was quite highin 1994 and remained at the same level in2001. Inequality was measured by calculating inequality measures on per capita consumption and per capita income, weighted by household size (Milanovic 1998). The Gini coefficient for income was 45 in 1994 and remained at 46 in 2001 while the Gini coefficient for consumption was 41 in 1994 and 40 in2001 (Table 6.1). Inequality was a major theme inthe LSA (2001) and income inequality in2001 i s discussed in one o f the backgroundpapers (Ozcan 2002). Basically, little has changed in the distribution o f income from 1994 to 2001. In 1994, urban-rural differentials disappeared when adjusted for price differences, in 2001,even without adjusting for price differences, there were no difference in Gini coefficients for urbanand rural areas. What 25  mattered in l! i s the same predominate factor in 2001-the >94 very sha~rpregional differentials noltedin 1994are also characteristic for 2001. 7. Risk mitigation,riskcoping, andriskreductionamongthe poor Turkey's poor use a combination o f tactics to deal with their poverty. Holtzman and Jorgensen (2000) have adopted a typology of methods that the poor use to handle their poverty-to manage their social risks as the authors term it. Their framework is presented below (Table 7,1), as adapted to Turkey and will serve to organize analysis of the qualitative data obtained from three sources: (i) and Ayata (2002, full text in Ayata Volume Two) summarized the finding o f the 120 case studies undertaken in conjunction with the quantitative HCIS; (ii)field visits and informal focus groups conducted by SRMP team members during the preparation and initial supervision o f the SRMP (cited as SRMP Field Notes); and (iii) an interim beneficiary assessment on the rapid response component of the SRMP. Most o f the qualitative data speak to the informal sector tactics utilized by the poor, although some o f the other topics in market-based and particularly public social risk management are also partially addressedhere. RiskReduction The most important aspects o f risk reduction in Turkey are migration, macroeconomic policies, public health, and child labor statutes. Formal sector in-service training i s not notably widespread inTurkey, most learning on thejob occurs as "learning by doing." 26  Migration O f all the tactics, arguably one o fthe most visible is rural to urban migration, seen inthe creation ofshanty-towns and slumscalledgecekondu ("built overnight" inTurkish) and the sharp reduction o f the share of the population that lives in rural areas of Turkey (Table 7.2). The major cities, particularly Istanbul, Ankara, and Izmir, have grown explosively, but other urban centers have also increased very rapidly. Urbanization increased steadily in Turkey's regions, except for a sharp acceleration in urbanization in Marmara from 1980 to 1990 (Figure 7.1). Figure 7.1 Turkey: Share Ruralby Region 100.0 `MARMARA m c 80.0 ---AEGEAN 60.0 - MEDITERRANEAN .c1 8C 40.0 CENTRAL 20.0 +BLACK SEA EASTERN 0.0 -+--SOUTHEASTERN 1960 1970 1980 1990 2000 Recent migrants t o urban areas t endt o 1ive in neighborhoods already settled by those from their native villages, and there i s an extremely important set o f informal connections among such neighbors (who are often at least distantly related to the newcomers) and fiends, as well as members o fvillage associations-hemgeri, discussed below. Besides rural to urban domestic migration, there is international migration in terms o f a very large group o f expatriate Turks (particularly in Germany), many o f whom help relatives in Turkey substantially. Inthe 2001 HCIS, 33 percent o f the urban population reported at least one relative abroad, with a poverty rate half o f that o f those who had no relatives abroad (Table 7.3). PublicHealth In the Southeast and East, midwives are an important source of health and nutrition advice for informal risk reduction and attend a "majority" o f births (Ayata and Ayata 2002), while the public health system i s available throughout the majority o f urban areas o f the country--three-quarters o f the urban population reported that there was a health clinic in their community (public risk reduction). However, rural areas are not so fortunate and most rural dwellers report significant travel time and cost to reach a clinic. The primary problem with public risk reduction in health i s that the poor do not have equal access to the health system. 27  Only 2.5 percent o f the urban population reported having a green card, which exempts the holder from payment for health services--but not medicine, the cost o f which is sometimes covered by the local Social Solidarity Foundation (see b e l o ~ ) . ' ~ Furthermore, only 30 percent o f those inurban areas who reported a green card were food poor. Qualitative sources suggest that fraud, evasion, and favoritism may all play a role in the granting o f green cards to the non-poor, while the poor find it quite difficult to access the process. The poor are less-informed, less likely to be literate, and often lack the resources o ftime and travel cost to get their documents inorder. Box 1.Green Cards inTurkey "Getting green card seems to be more easily available in the rural areas. Not only the rural poor has less access to other social security mechanisms but also they can prove their poverty more easily and they can persuade the muhtar to give them papers o f poverty. Urban muhtars are also elected but they will have less awareness o f the conditions o f people living in their neighborhood, and they may be less accessible. Besides, the rural poor may not only be poorer but they may have less access to health services. So green card may be their only solution." Ayata and Ayata 2002. "In Ankara, two women told us that when they applied for green cards they had to send a fax to their village to receive proof that they did not own anything even though they have been living in the gecekondu for 10 years. The fax costs 3 million TL ...The lack o f information about public social assistance seems, however, more important in rural areas. Inpoor villages o f east and south east Anatolia, only some people knew about the green cards. In several cases, even when people know about the existence o f social programs, they have a very little understanding o f the process o f getting assistance, including the Green card which is more widely known than the other assistance. In Van, none o f the three gecekondu families interviewed had Green Cards, and the women inthese families did not know how to get it. They thought it was difficult to get it. There was also a widespread belief among the poor that "some people are chosen, they are the lucky ones.. ." SRMP FieldNotes. ChildLabor Statutes InTurkey, schooling is mandatory for children for eight years, after which it is legal for children who are at least 14 to work full-time. However, informally, children (particularly boys) do work for low wages or inkindremuneration at much younger ages, particularly inthe agricultural sector and on the streets o f major cities (discussed below). Eventhough schooling is mandatory for eight years, many children do not attend regularly and in essence drop out informally. Turkey does not have official figures for drop-out rates or attendance. Unfortunately owing to non-response, the HCIS can not shed much light on attendance rates, but there i s a wealth o f qualitative evidence (SRMP field notes, Ayata andAyata 2000) that strongly indicates that: 0 Poor children attend less frequentlythan better-off students Poor children drop out muchmore than better-off students 0 Girls o f traditional families (typically the Southeast and East areas but also experienced in urban gecekondu areas populated by recent migrants) drop out after only a few years or are enrolled to comply with the law but never attend. 0 Communications difficulties also affect attendance The poor have great difficulties in meeting the out-of-pocket expenditures for school and to forgo income from child labor. j 4Green cards are given by the Ministry o f Health to those who meet an income criteria. 28  lBox 2. Education and Gender in Turkey. "Among traditional and uneducated families, the traditional separation of roles by gender is an additional factor that may prevent grls from attending school. Some evidence show that when school i s no longer compulsory, gender emerges as an even more important issue. In some traditional families many girls do not continue in non- compulsory education, regardless o f their economic situation. Girls would generally have to be trained for becoming housewives and mothers inthese families see no needs for their daughters to continue their education. The situation i s somewhat different for boys at this level, since for them the economic pressure i s often the main factor for leaving secondary education. In a small village in eastem Anatolia, a young women said she did not know whether her daughter will go to school. She said: "the husband will decide on this matter". And she also added: "the village schools are not good to leam anything; my daughter knows how to read and write, but she i s not graduated from the primary school". Traditional families are also very reluctant to send their daughters out o f the village or the neighborhood. This is even more true when girls begin to get matured physically at age 13-14 and when a new sets o f restrictive rules and norms begin to regulate their lives. At this age, girls from traditional families would generally not go anywhere without having a trustworthy company with them. This has obviously a strong negative impact on school attendance, inDarticular when schools are far away from residence." SRMP FieldNotes. ~BOXEducation andRelatedCosts 3. "The financial difficulties encountered in meeting the expenses the for school dress, shoes, books, stationery, school fees and school meals urge families to take their children from school. As most schools are in walking distance and as the govemment provides free transport for village children who go to town schools, transport costs do not appear as an important source o f complaint. Families unanimously emphasize that worsening economic crises and the decline o f regular sources o f income in the family seriously threaten the education o f their children. Finally, it should be emphasized that factors such as taking girls from school so that they would help their mothers at home in housework, the absence of schools in the vicinity as in the case of some Eastem , Anatolian provinces and the recently established gecekondu areas and, the absence o f children's birth certificates e also prevent the schooling o f the children. The poor families have great difficulty in providing proper wear, school bags books and stationery, meals, and school fees for their children. Hence, many have reportedthat they were unable to provide at least some o f these items during the school year. For instance, for primary school children in advanced years books cost a minimum o f twenty-thirty dollars and provided that there are two children inthe family, the cost would double to become a major financial burden on the household. Thus, frequently, the parents would let their children go to school without books. Although what is often at stake i s very little money, the extremely poor families cannot afford all the items o f school expenses as they have very low and irregular incomes and as they suffer chronically from shortage o f cash." Ayata and Ayata (2002) . . . . 1 . 29  RiskMitigation Several risk mitigation strategies are well-represented by Turkey's poor. These include multiplejobs, investing in physical assets, investing in human capital, investing insocial capital, marriageand extended family. Market-basedtactics such as investingin multiple financial assets are not practiced by the poor, and disability insurance is almost non-existent in Turkey. O n public risk mitigation, Turkey does have formal pension programs and the self-employed can enroll inthe Bag-Kur program, although virtually no poor are able to afford the premiums. Unemployment insurance is a recent innovation in Turkey, it was only inaugurated in 2001 and was not operational during the HCIS fieldwork Health insurance i s provided through the three public pension programs. Private health insurance i s available but only a small share o f the total population has it (1.5 percent), and these are not the poor. Multiplejobs It is quite common to have multiple "jobs" in Turkey-a better characterization would be multiple economic activities. For example, many rural families both farm their own land and send extended family members to urban areas (or even abroad) to earn seasonal or informal income w hile urban families keep gardens and 1ivestock inareas immediately surrounding their homes or back inthe village from which they migrated to the city. Inthe HCIS, 13 percent o f the total population reported having land outside o f their own village or settlement area, and 68 percent o f these were urban households. Urban households also produce some o f their own food-14 percent o f the population reportedhaving a v egetable garden for their o w n c onsumption and h a l f o f t hese w ere urban(HCIS data). Informal sector employment is both the mainstay o f many poor and the proximate cause o f their poverty. Working in a very low-paid informaljob sustains poor families, but typically at such a low level that they can not escape poverty or invest in their children. 30  lBox 4. Informal Sector inTurkey "Casual workers change jobs frequently shifting from one employer to another even during very short periods o f time. Thus, many report having worked for tens or even hundreds o f employers and being engaged in numerous trades in the past. During the frequent periods o f unemployment between changing employers and jobs the casual worker himself and often the whole family w h o depends on him lose an income. The casual poor typically wait for days in the coffee houses, bus stops or other identified spots to be picked by an employer on a daily basis. Even in the case o f working for the seasonal trades such as the garment, brick making and construction industries that offer the highest pay, jobs are often temporary and workers are arbitrarily laid off. As the latest wave o f migrants to the cities the majority o f the extremely poor have a background in farming, agricultural wage work and animal husbandry. When they were still in the village as well as after coming to the city almost all casual laborers have worked for the building industrythough many were not able to develop new skills there. Factory work however appears much less common as it was generally limitedto the labor-intensive industries. Only four household heads inIstanbul have mentioned that they worked in the garment industry at one time in the past. In addition, the casual workers have almost all attempted doing various street occupations working as shoe-shiners, porters, food sellers etc. Finally, those most desperate for income work as garbage pickers (bottle collectors, scrap metal collectors, and paper sorting) and they wash dishes in the coffee houses and worst o f all as sewer men. Incomes varied significantly in such casual trades; shoe-shining and garbage picking brings only one or two, street selling two and three, and working inthe building industry four-five dollars per day. Almost all these jobs, except factory work, offers no social security for the work force and cases where workers were abused, for instance being refused to pay what was initially promised, were quite common." Ayata and Avata (2002). 31  Investmentin PhysicalAssets To the limited extent provided by their limited means, the poor in Turkey invest in physical capital-particularly housing and livestock. The majority o f the urban population own their home (63 percent) and the poverty rates are only slightly different between urban owners and renters (Table 4.9). House ownership is basically similar in rural areas (69 percent o f the rural population own their house) where families typically buildtheir own dwelling or addrooms to an existing one (HCIS data). InvestmentinHumanCapital--Health The poor in Turkey invest as much possible intheir humancapital, particularly in education for children. The poor are not as able to invest in health because o f the prohibitive cost o f health care and the scarcity o f the green card. In the case studies, it was demonstrated that a serious illness was a major idiosyncratic shock to the affected poor household and strongly associated with poverty for the entire household (particularly when the adult male breadwinner was affected). The poor inTurkey are less likely to report that they were illin a reference period than the non-poor (HCIS data). This does not mean that the prevalence o f illness i s necessarily worse for the non-poor, but probably indicates that the poor have fewer options to pay for treatment, and thus are forced to ignore or minimize conditions that would result ina visit to the doctor by those who could afford it. InvestmentinHumanCapital--Education Education is an important investment for many poor families, even though it can be quite difficult for families to pay for the out-of-pocket costs and to lose potential income from child labor. Some poor families are forced to keep their children home from secondary school or even to pull one or more children out, leaving the best student studying. Box 5. Parental Commitment to Education most families however, are highly enthusiastic about the education o f their children. The education o f the children i s seen as investment into the future, the most important source o f security not only for the children themselves but for the parents and indeed the whole family. Thus, many have mentioned their children's school expenses among the most fundamental needs o f the household. The parents' willingness however primarily depends on the school performance o f the child. Hence, even the most desperate families make great effort to keep their successful children at school. Some can achieve this until the end o f the High School but for further stages they are almost totally helpless. Since parents cannot send them to good schools and to the expensive preparation courses the poor children's chances o f getting to the university i s very low and those few who do, find it very difficult to meet the expenses o f living in a different city. In such cases the parents would ask for support from relatives, neighbors, employers, charity associations and the government authorities." Avata and Avata (2002). 32  Social capital. Turks have a high degree o f social solidarity (Ayata and Ayata 2002) and most invest as much as they can afford into reciprocity for important social capital building events, such as attending weddings and ritual circumcisions. Hospitality i s particularly valued in Turkish culture, and extremely commonly interviewers (and SRMP team members) were served tea with sugar or ayran (a yoghurt drink) even by extremely poor families who would have been shamed not to offer honored guests some refreshment. Box 6. Social Capital and Reciprocity inTurkey. One important neighborly or kinship help with the expectation o f reciprocity i s during traditional ceremonies for significant life passages such as marriage and sunnet (circumcision) These ceremonies are the key for the reproduction o f social relations. Therefore there is a rule o f "reciprocity" at these ceremonies and they are a way to receive money or gold from the kinand neighbors." SFWP FieldNotes. Turkish society has produced an important informal expression o f social capital- the "village associations" o f hemjeri (people o f same place o f origin). These hemjeri networks provide a very important resource o f private, informal assistance to the poor, Iwho utilize them as a risk coping device (discussed below). Box 7. Social Capital and Networks inTurkey "In the gecekondus where the entire population migrated at some point from different regions, hemgeri networks act as the basis o f s olidarity and patronage relationships. Aside f r o m helping the poorest with some inkindassistance, these associations have an impact o n the integrationto their members into the city life by articulating local identities with the necessities o f everyday city life. People from the same village or town have meetings through these a ssociations, where they share their problems they face intheir new environment such as their health problems, unemployment, political party affiliations and so on." SRMP FieldNotes. Marriage, Family, and Extended Family. It is difficult to overstate the importance in Turkey o f marriage, family and extended family ties. B o x 8. Marriage and Extended Family inTurkey "As a respondent has expressed so well, "breaking with ones family is committing economic suicide". Although the support networks are based primarily on patrilineal ties, solidarity among married sisters seem particularly strong while women also emphasize that they do get help frequently from their own fathers brothers. On the other hand, the widowed women report receiving considerable help from their husbands relatives too. Disputes among members o f the extended family are also common and many tend to cut off their relations with close kin almost completely. Sometimes support for needy relatives continue although the parties involved do not speak to each other. On other occasions, parents cross with some o f their children would get support from others and in a similar vein relations with some close relatives would be smooth while others not. Mothers may continue to help children refused by their husbands. In summary, despite the growing tensions and conflict, total isolation from the extended family i s a very rare occurrence." Ayata and Ayata 2002. 33  MultipleFinancialAssets In the large cities/developed Marmara region, financial infrastructure is well developed (there are many banks offering a full range o f financial services), but not inthe rural areas and particularly not in S outheast A natolia. Investmentinthese assets is a possibility only for the non-poor. For example, as noted previously, less than 6 percent o f the urban population has a foreign currency account. Admittedly these lucky few are not poor (their poverty rate is less than 2 percent), but they are not typical for Turkey. Rather more of the urban population has a credit card (30 percent) and somewhat surprisingly, 7 percent o fthese are urbanfood poor (Table 7.4) FormalSectorPensions Turkey's formal sector pensions are analyzed in the 2000 Country Economic Memorandum for Turkey (World Bank 2000 CEM). There are three public pension programs-- Sosyal Sigortalar Kurumu (SSK), Emekli Sandigi (ES), and Bag-Kur (BK)- covering different areas o f the labor market. SSK covers public and private sector workers excluding civil servants, while ES covers civil servants and BK covers the self- employed and farmers. Prior to the 1999 reforms, approximately 2.8 million people were beneficiaries o f SSK, 1.1million o f ES, and nearly 0.9 million o f BK (CEM 2001, Table 2.1). Inthe HCIS, 21 percent of the total population reported receipt of a pension of any kind in the preceding month.15 Of these, more than four-fifths were urban households. Of the urban population reporting pension receipt, only 11 percent were poor. Thus, it seems that pensions are not an important weapon inthe arsenal o f the poor l5Which programprovidedthe pensionwas not capturedinthe HCIS. 34  to mitigate risk, since so few poor households rely on them because so few poor households receive them. HealthInsurance Health insurance is provided by SSK, ES, and BK for their enrolled members. A scant few employers offer private health insurance. Turkey's health insurance i s described inthe Turkey Health Sector Review (forthcoming). As was the case for formal sector pensions, most households with health insurance are urban and few o f the urban population are poor (Table 7.5). Less than 3 percent o f the urban population have green cards. RiskCoping Market-based tactics such as selling o f financial assets or borrowing from banks are not accessible to most o f the population-only 5 percent o f the total population reported having a hard currency bank account and the same amount (5 percent o f total population) reported that they had succeeded in obtaining a bank loan. These tactics are not discussed further. Public risk coping interventions include disaster relief, Social Solidarity Fundad hoc transfers, and Social Solidarity Fund income-generating activities and literacy efforts. SellingofAssets Inthe HCIS, almost one-fifth of the total population reported that they had sold household assets or valuables (there was no difference between the urban and rural population). Asset sales do not seem to be strongly associated with poverty-the poverty rate was only a little lower for those inthe urbanpopulation who sold assets (20 percent) thanthose who didnot (16percent, Table 7.6). Q ualitative information suggests that there is little market for the assets owned by the poor-if they have a TV, chances are excellent that it either does not work or is ancient, and nearly 100 percent o f households already have a carpet andwould not be interested inpurchasing an old one from the poor. Borrowingfromneighbors,informalsupportnetworksand charity. After migration, the next most important informal tactic for the poor i s the strong sense o f social solidarity and the high extent o f informal assistance and help among relatives, friends, neighbors, and hemgeri. Of course it i s not possible to quantify this factor in a formal international comparison, but there i s a strong sense from the qualitative work and the HCIS that inTurkey there i s muchmore direct assistance incash and kind between households than in the neighboring countries o f the former Soviet Union and Europe. For example, in the 2001 HCIS, one-fifth o f the urban population reported that they assisted other households with cash or kinddonations16and the poverty rate o f those who did assist was nearly 10 percentage points below average (Table 7.7). l6The same percentage- 20-of rural householdsreportedthat they hadassistedanotherhouseholdwith cash or kind. 35  Nearlyhalf o f the urban population reported that they had borrowed from friends or relatives and the poverty rate o f those who borrowed was 7 percentage points higher than those who did not borrow (Table 7.8). A slightly higher share o f the rural population (57 percent) reportedthat they hadborrowed. There also seems to be a high level o f religiously-motivated private charitable work, sometimes directly from a wealthy individual to the needy (such as poor interviewedin field visits in Sultanbeyli (a gecekondu area o f Istanbul) who were living rent-free in somewhat dilapidated housing provided by a local business magnate, SFWP Field Notes) as well as intermediated through the imam (religious leader) and mosque such as thefitre and z ekut (alms which are obligatory for the w ell-off) as w ell as the traditional gift o fmutton to poor families duringthe Buyrum holidays. The culmination o f these flows o f private and informal assistance are extremely important for the poor. lBox 8. Importance o f InformalNetworks inTurkey. I The poor emphasize that they are able to send their children to school only with the help they receive from their associates; these would include relatives, neighbors and hemgeri as well as middle-class charity associations. Often, help i s in kind, in the form o f books, stationary, school bags, school dress, shoes and contributions to the registration fees. The family strategy is generally one o f combining these sources Close relatives such as brothers and sisters, parents and first cousins respond to children's various needs and sometimes this takes the form o f giving second hand school materials to the needy children. Afflueni relatives inEurope are particularly helpful. On the other hand, especially the teachers but also the headmer inthe quarter mobilize and organize help for the poor students inorder to keep them at school. Finally fem families have mentioned that they send their children to relatives in other towns and cities in order tc provide them better educational opportunities." Ayata and Ayata (2002). B o x 8. KinshpNetworks inTurkey. InTurgali village inVan a villager who worked as a construction worker inthe past and earned 6 to 8 million TL a day, was unable to find work (in the past month 5 months he only worked 2 months). Right now his family o f 12 lives o n 25 million TL a month and have 300 million TL debt. H e could not afford his electricity bill so it was cut off. The last time they had meat was ina marriage ceremony and they were not much helped by their family nor their neighbors, not because o f the lack o f will but because they were also poor and helped them as much as they could which was n o t much. But regardless o f the level of poverty, especially in the traditional setting in eastern Anatolia, kinship relations played a major role in surviving strategies o f the poor. In Hasbey village in Gevas, an extremely poor household o f 12 (two brothers with their wives, children, step mother and a sister), lived in a house given to them free o f rent by the villagers. They hadno landandthe two male adult male members hadnot beenable to findwork inthe past couple o f months. Regardless, they received their widowed sister and her son from Istanbul just recently to live with them since her only son, a 17 year old, could not work anymore to support her due to a kidnevillness. SRMP FieldNotes. However, such informal and private assistance i s not infinite, and can be particularly difficult to sustain inthe event o f a large covariate shock such as the financial crisis. Indeed, some qualitative evidence points to reductions in the ability o f people to assist each other since all inthe community are suffering from the economic downturn. 36  Table 9. Limits o f Informal Assistance inTurkey. "A widowed woman who hadlost her husband 4 years received 30 million TL from a vakif(foundation) in Istanbul for three years continuously. She lived with her only son who was working since age 7 and in recent years, at the age o f 17, was eaming 55million TL a month when their rent inIstanbul was 30 million. T o supplement their income the vakifin their neighborhood paid her 30 million TL a month. But inrecent month, she was told by the vakifthat as a result o f economic crisis, they had to stop paying her and since his child also was sick she hadto return to her brothers intheir village inthe Gevas district." SRMP Field Notes. Child Labor Child labor i s relatively visible in Turkey, both in the informal sector in urban areas and inrural areas. For example, inthe district o f Duragan, Sinop Province, landless villagers reported that they "sold" their under-age boys for the summer to work for landowners inthe valley, while the menworked in seasonal informal construction jobs in Istanbul, Ankara or other urban areas. There i s also a small but distinct population o f street children, particularly visible inIstanbul, Ankara and Diyarbikir (see below). Children o f school age are sent out to work, and ifthere are no family business or quasi-formal sector opportunities (working under the age o f 14 i s illegal) then the only alternative is informal employment, and in urban areas, this invariably means on the 37  street (in rural areas most child 1abor is family labor related t o agriculture, o r cottage industries such as carpet weaving). Such informal sector activities range from petty trading, vehicle guarding, porterage, through to the illegal such as under-age prostitution, drugdealing and theft. Although providing family income, the costs are highfor the child concerned interms o f foregone education, lack o f development o f marketable skills, and exposure to illegal activities. Street-working children (and child labor ingeneral) i s a problem confronted most by the poorest quartile o f the population. There are significant seasonal variations - with an increased demand for child labor, or at least opportunities to gain income, and hence number o f children working on the street inthe summer months (especially intourist areas). Table 10. Child Labor inTurkey "Families are in general willing to findjobs for their school aged children at least during the vacations but because o f the scarcity o f employment opportunities not all o f them are able to do so. Inthe rural areas the children join their parents in agricultural work in case that there is a family farm but more frequently they work with their parents as agricultural laborers. Inthe urban areas they undertake various informal jobs or work as casual laborers. Especially where the children have access to relatively stable and well-paid jobs work becomes a substitute for education, as needy families do n o t w ant t o be deprived o f an important income source by keeping children at school. In other words, poverty itself becomes the most serious obstacle to the education o f children. Especially where children are not among the brightest in class, the threat o f parents taking them from school becomes more immanent as income from work, however meager, becomes the priority o f the family. The children themselves often want too leave the school in order to work. Sometimes the parents have to make a choice betweentwo or more children as to educate only one o f them and let the other work; in such cases their preferences for the school would first be the brighter ones and secondly the boys." Ayata and Ayata 2002. Street Children Children living on the street is more o f a social problem, although there are severe economic and welfare consequences. Most such cases are as a result o f family problems (with children running away from home) - lesser causes are children leaving institutions or fleeing the consequences o f crime. These children are truly marginal and live right on the margins o f the law - with 70% estimated to be male, and 30% female (SRMP Field Notes). They are extremely vulnerable, with no fixed abode, usually no identification, and exposed t o every vice imaginable. M any c ases c an be resolved through e ffective social work, if they are caught in the early stages. It i s often possible to work out some family reconciliation, resolve the problem, place the children back in school or at least in a more secure environment. With the passage o f time, all o f these actions become increasingly difficult and such children can become consolidated into the street culture with few, if any, employment opportunities, on the margins o f society and the law. While there are no longitudinal studies on such children inTurkey, anecdotally it i s clear that their situation i s bleak and often they end up incrime, prison or inpremature death. 38  39  ConsumptionReduction A major tactic widely used by the poor inresponse to the crisis has been the risk coping device o f simply reducing their consumption, particularly on non-food items, but also the poor reported significant reduction in food consumption inthe HCIS (Table 7.9) and in the qualitative interviews. For example, more than three-fifths o f urban households reportedthat they cut down or "stopped" food consumptionwhile predictably enough, the poverty rate o f those who did not change their food consumption was significantly lower than average (Table 7.10). Nearly three-quarters o f rural households reported that they had cut down or "stopped" food consumption. B o x 11. Food Consumption o f the Poor inTurkey "The basic food items that the poor consume most frequently as part o f their staple diet include the following; flour and its derivatives (bread, bulgur-cracked wheat, macaroni), rice, cooking oil, salt, sugar, tomato paste, potatoes, eggs and tea. The benefit dependent poor, and especially those poorest among them, cannot always secure even a minimum supply o f these items in their kitchen. The extremely poor families who try to survive on the edge o f the starvation line would most frequently have tea with bread for breakfast and bread, whether home made or bought from outside, and either with `flour soup' or one o f bulgur, macaroni, rice or boiled potatoes for dinner. Tomato paste would almost always be used as an ingredient to cook these dishes. In the summer months the majority o f the poor, including those who are better off, would have bread and uncooked vegetables i.e. tomatoes, green pepper and cucumber for breakfast and very frequently for lunch too. Even the most desperate in their most difficult times would maintain a minimum quantity o f flour, cooking oil, sugar and tomato paste. Occasionally the poorest families may have difficulty inmaintaining this bare minimumregularly but in such cases the possibility of receiving food from neighbors is particularly high. At times when the poor family really becomes desperate, the very close neighbors, however poor they themselves would be, will tend to offer help in the form o f cooked dishes (a pot o f bulgur, macaroni, soup, boiled potatoes). The destitute however have also reported cases o f almost absolute hunger for a day or even two. The very poor people generally suffer from frequent shortages o f cash and credit inachieving this level of food consumption and for many o f them conditions have become even worse during the economic crisis. In the majority o f the interviews it was observed that the poor families who could regularly provide the standard meals described above, considered themselves as lucky. In the worst cases witnessed during the fieldwork, families who had no regular income were unable to maintain even the minimum food stock for cooking some o f the dishes mentioned above. In such situations the family survived o n having bread and cooking oil, bread and tomato paste, bread and potatoes and sometimes only bread. It was observed at least in two cases that the families were living on the edge o f starvation as their children occasionally slept hungry. The unemployedfather o f seven children in a squatter settlement o f Istanbul expressed his sorrow inthe following words, `Ido not go home, Ifind myself a place to hide inthe darkness untilthe children sleep so that Iavoid the tragic sight o f my hungry children crying for food"'. Ayata and Ayata (2002). Inthe HCIS, about the same share ofurbanhouseholds reportedthat they had cut down on non-food items (59 percent) and a significant portion (15 percent) reported that they had stopped altogether spending on non-food items (50 and 26 percent for rural households respectively, Table 7.11). Interestingly enough, the human capital investment items of education and health spending were less affected by reductions-35 percent o f the urbanpopulation and 4 4 percent o f the ruralpopulation reported spending 1ess on both health and education (Tables 7.12 and 7.13). About two-thirds o f the urban populationreported cutting down on food quality (Table 7.14). 40  Malnutrition Visible malnourishment o f children can be observed inTurkey, but it is relatively infrequent. Only about 2 percent o f children under 5 were wasted (low weight for age), which i s usually an indicator o f acute malnutrition (UNICEF 2000). Interviewers for the HCIS qualitative study and a very few times during SRMP (Field Notes) did report some acute malnourishment, but in most cases, the poorest households did get food assistance directly from neighbors, helping to alleviate this problem. Rather more children were stunted (low height for age) - about 20 percent in 1998 (UNICEF 2000). Stunting reflects chronic malnourishment and i s much more difficult for a casual observer to note than acute malnourishment. "Stunting i s more prevalent in rural areas, in the East, and among children o f mothers with no education. Stunting occurs more frequently among children who are o f higher birthorder, and among those born after an interval o f less than 24 months" (UNICEF 2000). PublicRiskCoping There are two institutions for publicly-fundedrisk coping mechanisms inTurkey: the Social Solidarity Fundand SHCEK, the Social Services and Child Protection Organization. These are three kinds o f programs run by the Social Solidarity Fund and its affiliated Foundations which pertain to risk coping: disaster relief, ad hoc transfers, and micro-projects and literacy training. SHCEK runs 381 social service provincial organizations such as community centers, special houses for women, elderly and disabled people, orphanages, family information centers and centers for street children and youth. SHCEK i s primarily responsible for institutionalized children and other vulnerable groups and does not provide cash transfers, that i s the purview o f the Social Solidarity Fund. 41  Box12. SYDTF The Social Assistance and Solidarity Encouragement Fund (Social Solidarity Fund, SYDTF) was established in 1986 as an umbrella organization and financing entity for 931 regional affiliate foundations (SYDVs). The purpose o f the SYDTF and the SYDVs was defined as "To aid poor and destitute citizens in circumstances o f need and, as necessary, those who have been accepted inTurkey or have travelled here by whatever means, to ensure the distribution o f wealth in an equitable fashion by taking measures to improve social justice and to encourage social assistance and solidarity." More than 4 million people benefit annually from the SYDTF and S Y D V activities which are concentrated o n various types o f ad hoc assistance to needy individuals: 1. Hot food program: Provision o f free meals (soup kitchens) to 64,000 poor people who are registered at the S Y D V and canprovide the registration card . 2. Lunch for school children from remote villages: Provision o f free lunch to 540,000 students coming from villages with no primary schools and that have to be transferred to a central school by buses. 3. Educationprograms: (i) In-kindbenefits conditional on school enrolment (textbooks, uniforms, waiver o f dormitory fees) provided to 153,000 poor students; (ii) 345 student hostels provides accommodation to 53,000 students; (iii) scholarships for 202,549 poor students enrolled in universities conditional o n highgrades. 4. Ramazan and Bayram assistance. Food and clothing assistance provided to 1,000,000 poor people in the month o f Ramazan and duringBayram holidays or episodically. 5. Winter (heating) assistance: Provision o f coal and fuel for heating before winter to one million poor people. 6. Health assistance: covering out-patient treatment and medicine expenses o f poor people with green card. 7. Assistance for the disabled. Support programs for 13,500 disabled. 8. Natural disasters. Cash or in-kindbenefits (shelter, compensation to families for deaths o f relatives; damage to SME businesses) received by 817,949 individual victims o f natural disasters (floods, earthquakes). 9. Damage from terrorist activities. Cash or in-kind benefits received by 841 individual victims of terrorist activities. The SYDVs also administer income-generation programs: (i) projects creating direct employment for 89,234 poor and unemployed citizens (trout production, beekeeping, carpentry rug weaving, cow raising, business, knitting, sheep raising, small handicraft, mushroom growing, greenhouses); and (ii) training for 186,3 12 poor and unemployed (vines, hit-walnut-olive raising, vegetables, projects for the disabled, poultry, fodder, etc.) as well as literacy programs. SRMP Field Notes. EarthquakeBenefits The SYDTF and SYDVs geared up for a major effort to provide cash benefits to earthquake victims with financing from the World Bank-the US$ 250 million Emergency Earthquake Recovery Loan (EERL) Four types o f benefits were provided: accommodation (or rental) allowance, repair allowance, and death and disability benefits. The Government was so satisfied with the rental allowance that they extended it for an additional three months (it had been intendedto stop after one year o f receipt) from their own financing. The SYDTF and SYDVs performed admirably, under very difficult conditions in the earthquake zone. Several very positive evaluations o f the EERL were conducted, including an operational audit, a beneficiary assessment, extensive supervision, and a World Bank internal review. 42  Box 13. Earthquake and SYDTF "The Adapazari center was staffed entirely by volunteers (both management and application processors) and was a showcase o f organization and private-public partnerships. A local businessman was working himself as a data entry operator and general staff manager and had recruited the entire staff from the local community. There were 10 computers each with three staff-one person to talk withthe client, one person to physically search through the file and read out loud the information to the third person, the data entry operator. The local business community had been mobilized, and the food for all the staff was donated by local business. Applicants were moved quickly through the application process and n o queues that lasted longer than 20-30 minutes were observed. The Adapazari center was set up in a basketball gymnasium. Clients picked up their application forms at the entrance and when they were filled out, were ushered into seats o f the gymnasium and offered ajuice or tea (donated by the local business community) while a staff member explained the intake procedure to them. The clients then proceeded to one o f the 10 computers and were present while their information was entered into the computer, allowing the clients to verify their information directly." EERL Field Notes. By June 2000, 325,000 households were inregular receipt o f the accommodation allowance and credited it with helping them reintegrate into society and productive employment after the earthquake. Ad Hoc Social Assistance The SYDTF provides financing to the 931 SYDVs to provide ad hoc social assistance to "needy" families. Formal criteria for determining eligibility are that applicants must produce documentation o f their lack o f property and social insurance (see box), but the actual selection o f applicants is done by a local committee. Applicants are often recommended to the local SYDV by the muhtar (who i s typically consulted regardless) and a local committee (of various government officials but also three representatives o f civil society) decides on the applications individually. Resources are limitedandmany applicants receivenothingor only very limitedassistance. Box 14. DocumentationRequired by SYDV for social assistance applicants: 1. Directorate o f Census, for confirmation o f the family status. 2. TAPU(State Property Authority), to confirm lack o f ownership o fland and house. 3. Social security agencies (SSK, ES, BK),to confirm that the person i s not covered by social insurance. 4. Ministryo fInterior, for security clearance. 5. Ministryo f Finance directorate, to assess the tax record. 6. Municipality, to assess the taxes paid o n properties. Having the form stamped by different government offices can be a difficult task for some o f the respondents who usually have to take the application inperson to have it stamped. The SYDV also requires that an evaluation form be completed and sent to the S Y D V Board recommending the person for assistance. Moreover, in many cases the SYDV assigns someone (often a policeman or a gendarme) to conduct a visit at the application's house to further investigate if the person really requires assistance. SRMP Field Notes 43  "Social Solidarity Fund(SYDTF and SYDVs) gives a food package generally consisting o f staples such as cooking oil, macaroni, sugar, flour, tea and tomato paste. Food is distributed by the provincial government administration to the needy families periodically, for instance, once in every six months. The families who benefit from food aid report that the contents o f the package meet their basic food requirements for almost one month, depending of course on the size o f the family." Ayata and Ayata (2002). Less than two percent o f the population (1 percent o f the urban population and 0.5 percent o f the rural population) reported that they received cash income from the SYDTF (HCIS data). Part o f this could be terminology, inthe qualitative interviews, it was clear that very few respondents could identify the source o f the assistance that they reported- usually they said that the assistance was from the muhtar even when it was clearly from the SYDTF (SRMP Field Notes). An additional two percent o f the total population reported receiving "other state allowances," whatever this meant to the respondent i s not clear (HCIS data). These numbers are too small for any meaningful analysis, although it i s indicative that 70 percent o f this 1 percent o f urban recipients reporting SYDV assistance were not food- poor (Table 7.15). Micro-Projects and Literacy A major focus o f SYDV activity is the provision o f financing for micro-projects and for employability training (particularly adult literacy programs). However, access to these programs is limited. Only 22 percent o f urban households and 17 percent o f rural households surveyed reported that there were micro-project or training programs intheir residential area (HCIS data). Of the urban households which reported access, only 9 percent o f them were food poor (Table 7.15), suggesting that in urban areas, these courses and programs are not offered in the poorest areas, or that the poor simply lack information about the programs, or that the associated costs (time, travel) are too high. 44  8. ConclusionsandRecommendations This report has documented an increase inpoverty in urban areas o f Turkey from 1994 to 2001, related to the impact o f the 1999 earthquake and 2000-2001 financial crisis. Extreme poverty in all o f Turkey has not changed, and remains at low levels, but inequality i s also unchanged at quite high levels. A relatively large share (nearly one- fifth)oftheurbanpopulationhas consumptionbelow a food standard, andqualitative evidence indicates that poverty has worsened inrural areas as well. The poor have beenparticularly impacted by a reduction in seasonal and informal employment opportunities in the urbanized areas, and some menare returning to their rural villages because they can not earn enough in the cities to cover their costs. The primary coping strategy o f the poor has been to reduce consumption, particularly consumption o f food and quality o f food consumed, but there are also indications that the poor may have to cut back on education expenses and withdraw children from school. The poor rely strongly on networks o f extended family, friends, neighbors, and hemjeri, but these networks are strained to the limit by the covariate macroeconomic shocks experienced in 1999 (earthquake) and 2000-2001 (financial crisis). Other important coping strategies o f the poor have also come under stress as a result o f the crises. Multiple job holding has been curtailed by the reduction in seasonal and informal employment. The poor have been much less able to invest in physical assets or in their own human capital. Social capital, which i s an extremely important aspect o f traditional Turkish s ociety, has c ome under s train as the poor c an n o 1onger afford to attend traditional reciprocity events such as weddings. The poor are not positioned to sell their assets--nor is there much demand for them. Borrowing from neighbors i s a strategy used across the income spectrum-50 percent o f urban households and nearly 60 percent of rural households reported that they hadborrowed inthe previous nine months. Religious charity helps some o f the poor, but it i s episodic and does not cover needs sufficiently. As a last resort, the poor have sent their children out to work. Although the Government does finance ad hoc social assistance for the poor through the system o f the SYDTF and its 931 affiliated SYDVs, this assistance i s too partial to meet the needs and i s not that well-targeted to the poorest. Additionally, Turkey's social indicators do not compare favorably with other middle income countries (Table 8.1). In particular, infant and matemal mortality in Turkey i s quite highfor a middle-income country and female literacy is noticeably lower than the comparator countries (except for Tunisia). Life expectancy at birth i s equal to Tunisia and lower than the other countries. These facts point out the importance o f interventions in health and education, especially education o f women. The World Bank has been working in partnership with Turkey on the supply side o f these interventions, with several projects in education (including basic education) and health. However, to date, there have been limited efforts on the demand side, mostly in the form o f public information campaigns, which have not had much effect on these social indicators. 45  Against this background, the Turkish Government has taken decisive action to help mitigate the effects o fthese shocks on the poorest and to help insure against adverse coping strategies (removing children from school, underinvestingin health) and to break the v icious cycle of poverty perpetuated as children are w ithdrawn from school today only to become the poor o f tomorrow. InAugust 2001, the Turkish government transferred substantial resources to the Social Solidarity Fund (SYDTF) to finance back-to-school packs for 1.05 million poor children (TL 50 million per child). Additionally, the SYDTF expanded its food and fuel assistance for the winter. These measures, while important, were limited to a single payment. Recognizing that the poor needed more systematic assistance, the Government decided to adopt a new social assistance benefitdonditional cash transfers (CCT) which would be paid on a regular basis. The CCT are a highly targeted social assistance transfer to families with children, requiring positive family behavioral change with respect to health and education. Through the SYDTF and the SYDVs, the Government envisions an expanded social safety net targeted to the poorest families linked to certain positive behavioral changes such as keeping children in school and ensuring children receive adequate immunization coverage in a timely manner, basic health care and nutrition. This would require the introduction o f an improved and systematic targeting system using "points" for household characteristics linked to poverty. The CCT are modeled on highly successful programs in Latin America such as Mexico's program Oportunidades (formerly ~rogresa). ' The Government is also seeking to expand the traditional activities o f the SYDVs interms o fmicro-projects and adult literacy efforts as it recognizes that demand for these activities exceeded supply. Finally, the Government intends to monitor poverty with more frequently household surveys, recognizing that one-off efforts such as the HCIS are not adequate to provide policy makers with the data needed to understand and therefore improve social protection and other sectoral efforts. Inview of this substantial effort to reform the safety net in Turkey and to attack poverty, the World Bank has been able to respond by supporting the reform process with financial resources. The major vehicle o f Bank support i s the Turkey Social Risk Mitigation Project/Loan (US$ 500 million). The development objective o f the SRMP i s to mitigate the impact o f the crises on poor households (social risk mitigation) and to improve their capacity to cope with similar risks in the future (social risk management). The SRMP will achieve these objectives through: (i)an adjustment portion, providing immediate support to the poorest affected by the crisis (social risk mitigation); and (ii) an investment portion, which consists of three components (a) building up the capacity o f 17ProgresaiOportunidadeshas been extensively reviewed by the International Food Policy Research Institute, and the papers are available at: http:/lwww.ifpri.org/themes/progresa/progresa reporthtm The program is described in Spanish at httu://www.progresa.gob.mxi It i s an integrated program which provides educational, health, and nutritional support. Its major component is the payment o f cashtransfers to the mothers of poor children who attend school and health clinics regularly. 46  state institutions providing basic social services and social assistance to the poor (social risk management); (b) implementing a social assistance system (CCT) targeted to the poorest conditional on improved use o f basic health and education services (social risk mitigation and prevention); and (c) increasing the income generating and employment opportunities o f the poor (social riskprevention). Thus, the SRMP is designed to support Turkey's ongoing efforts to reform, improve, and expand the social protection system and to address some o f the negative coping strategies that the poor have adopted in response to the impact o f the earthquake and financial crises. Clearly, the most pressingtask for Turkey to undertake for the goal o f poverty reduction i s to improve macroeconomic management along the lines already agreed in programs with the International Monetary Fund and the World Bank. Duringthe drafting o f this report, Turkey experienced two political crises (February 2001 and August 2002) and early elections are scheduled for November 2002. The February 2001 crisis had significant negative reverberations on the economy, which lead to the sharp increase in poverty documented inthis report. Although poverty did sharply increase in 2001, most o f the poverty was shallow- extreme poverty was unchanged. This means that with good macroeconomic 47  management, it should be possible to restart growth and roll back the food poverty increase registered in2001. Most o f those who became poor in 2001 had a stock o f assets and a level o f human capital that would enable them to respond positively to the new opportunities that will come when broad-based labor intensive growth i s resumed in Turkey. The poverty reduction strategy for Turkey must therefore focus on the renewal o f growth driven by key structural reforms and strengtheningo fmarket institutions. At the same time, it is important to ensure that the newly poor are not forced to adopt negative coping strategies such as pulling their children from school or under-investing in their health. Here, the Government's innovation o f conditional cash transfers (CCT) with Bank support i s key for ensuring that poverty in 2001 i s not perpetuated in future generations. Additionally, the Government's expansion o f micro-projects in the Local Initiatives (LI) component o f the SRMP will help poor individuals to create income-producing activities and will enable communities to create needed infrastructure and foster community development. This component also includes adult literacy and other training programs. These efforts are necessary to assist the poor to take advantage o f already-existing opportunities as well as the new opportunities that will come with growth. Evenwith growth, however, there is a small portion o fthe Turkish population (2 percent) who comprise a hard core group o f the extreme poor, and growth will do little to ameliorate their situation. The evidence from the HCIS i s that this core group o f extreme poor is not always reached by the existing SYDTF and SYDV programs. Outreach needs to be improved, and the current system o f allocating funds mechanically across SYDVs means that in poorer areas, there is less scope for the SYDV to provide adequate social assistance inkind or cash. Institutional development o f the SYDTF and SYDVs as well as SHCEK i s needed to improve outreach and operating procedures, and the Government's efforts in this area are also being supported by the SRMP (in the institutional development component). A new targeting mechanism, a scoring formula, is being developed for both CCT and LI. This formula could be usedto identifythe extreme poor, which would improve targeting. There were clear indications from the HCIS and qualitative sources that targeting i s a problem under the current social assistance program. The Government's capacity to monitor poverty also requires bolstering. The last household income and expenditure survey was in 1994, and the 2001 HCIS provides only a snapshot and can not fulfil the needs o f the Government for a consumption survey that can generate weights for the consumer price index as well as monitor poverty. Under the institutional development component o f the SRMP, the Bank will cofinance four annual HIES and will also assist in the development o f a poverty map for Turkey. 48  Bibliography Braithwaite, Grootaert and Milanovic (1999). Braithwaite, Jeanine, Christiaan Grootaert and Branko Milanovic. Poverty and Social Assistance in Transition Countries. New York: St. Martin's Press. 1999. Deaton (1997). Deaton, Angus. The Analysis of Household Surveys: A Microeconometric Approach to Development Policy. Washington, DC: The World Bank and Johns Hopkins UniversityPress, 1997. Deaton and Zaidi (2002). Deaton, Angus and Salman Zaidi. "Guidelines for Constructing Consumption Aggregates for Welfare Analysis. http://poverty.worldbank.org/library/view/4361 3 Lanjouw, Milanovic and Patemostro ( 1998). L anjouw, Peter, Branko Milanovic, and Stefano Paternostro. "Poverty and the Economic Transition: How Do Changes in Economies o f Scale Affect Poverty Rates for DifferentHouseholds?" Policy ResearchWorking Papers. No. 2009. Washington, DC: The World Bank, 1998. Lanjouw and Ravallion (1995). Lanjouw, Peter and Martin Ravallion. "Poverty and Household Size." The Economic Journal. Vol. 105. pp. 1415-1434. LSA (2000). World Bank. Turkey: Economic Reforms. Living Standards, and Social Welfare Study. Report No. 20029-TU. January 2000. Milanovic (1998). Milanovic, Branko. Income, Inequality, and Poverty during the Transition from Plannedto Market Economy. Washington, DC: The World Bank, 1998. Ravallion (1992). Ravallion, Martin. "Poverty Comparisons: A Guide to Concepts and Methods." Living StandardsMeasurement Study Working Papers. No. 88. Washington, DC: The World Bank, 1992. Ravallion (2000). Ravallion, Martin. "Poverty Lines in Theory and Practice. Living Standards Measurement Study Working Papers. No. 133. Washington, DC: The World Bank, 2000. UNICEF (2000). UNICEF. The State o f Children and Women inTurkey. June 2000. World Bank (2002). "A Sourcebook for Poverty Reduction Strategies." http://www.worldbank.org/poverty/strategies/sourctoc.htm World Bank (2001). World Development Report 2000/2001: Attacking Poverty. Washington, DC: The World Bank and Oxford University Press, 2001 World Bank (1993). Poverty ReductionHandbook. Washington, DC: The World Bank, 1993. World Bank (1990). World Development Report 1990: Poverty. Washington, DC: The World Bank, 1990. 49   Annex 1. On the HCIS sample. Questions Raised During Consultations: The State Institute o f Statistics and the State Planning Organization questioned the method used by the consultants to reduce the sample from 7,000 households as provided by SIS to 4,200. "Using random selection o f the necessary number o f blocs by tearing blocs from layers. If such a decreasing operation were to be done, it should have been within layers. Contrary implementationwill damage both the regional and urbadrural distribution." Furthermore, both institutions noted that the original sample was not designed to be representative on the regional level for Turkey, although the report does present findings on the regional level, which caveatswill be included. Response to These Questions: The original sample was drawn by State Institute o f Statistics (SIS) and the sample size was 8000 households. Due to the limited resources available for the project, the sample size was decided to be set to 4000 and additional 300 households were selected to be on the safer side. The original sample drawn represents Turkey. It is a multi-stage stratified cluster sample which allows comparisons by region and rural-urban places. The original sampling design required stratifying the country into 7 regions at first stage. At the second stage, residential units in each region was divided into population strata such as places with 0-2000 population, 2001-5000, 5001-10000, 10001-20000, 20001-50000, 50001-100000, 100001-150000 and places with more than 150000population. Clusters were formed by combining 30 households within each population strata. At final stage, clusters were selected within each population stratum independently by using random selection technique. Given the budget of the research, the sample size reduced to 4000 by way o f a technique which i s akin to subsampling which was done the following way. For urban sample, households in each province are separated. Cluster identification i s removed from each cluster and assigned random numbers. 50 percent o f them were selected randomly with the help o f computer. This is done proportionately for each province. Later, they were identified by combining with their respective stratum information. Considering the possible loses due to non-response and unusable questionnaires, additional 300 households were also selected proportionately which increased the sample size to 4300. For rural population a similar technique was employed to identify the sample units. 50 percent o f the households were selected randomly. The employed technique resulted changes in the composition o f clusters. However, there i s evidence indicating that if the random selection technique in the identification o f sample unitswithin each subunits (provinces), y i s an unbiased estimate o f Y. For calculation 50  o f variance for y and sample estimation o f the variance for two-stage samples, refer to Cochran 1977: 277-278. The resulting sample size did not reduce the number o f provinces included in the sample. The reduced sample included 63 provinces as in the original sample. There were 7 provinces in Mediterranean, 8 in Aegean, 10 in Marmara, 7 in Southeast, 8 in East, 11 in Central and 12 inBlacksea. Urban-rural division in the original sample was 17-83% which was considerably different than the division reported in SIS publication. As it became clear later that SIS oversampled in urban areas due to higher heteregoneity and undersampled inrural areas due to higher homogeneoity which was not acknowledged and given in the form o f weights. Another factor increasing the discrepancy between the population and the sample percentages i s the desire o f the researchers to define only villages as rural which was decided inthe field. Interviewers were instructed to classify villages only as rural and other places which were not village as urban. 51  Turkey: Poverty and CopingAfter Crises Annex 2 to Volume One: MainReport List of Tables Table 1.1Foodquantities andthe regional CPIprices for the 7 regions of Turkey ...................... 54 Table 1.2. Turkey: RegionalandUrban-RuralHCISandCensusPopulation........................................ 58 Table 1.3. Turkey: ComparingSurveyandNationalAccounts ConsumptionandIncome...................... 58 Table 2.1: EconomicIndicatorsinMostImpactedEarthquakeAreas, 1998........................................... 59 Table 2.2: Impact of EarthquakeonManufacturingEstablishments....................................................... 59 Table 2.3: Real GDP Growthby Province..................................................................................... 59 Table 2.4: Quarterly Real GNP growth rates (1987 TurkishLira) ............................................... 60 Table 2.5: Turkey: Employment 1998-2000Quarter* .................................................................. 60 Table 2.6: Direct fiscal costs of the earthquake............................................................................ Table 2.7: Inflationafterthe earthquake............................................................................................ -61 60 Table 2.8: RealGNP SectoralGrowthRates 2000-2001...................................................................... 61 Table 2.9: Employmentby Sector...................................................................................................... 61 Table 2.10: Reponsesto TOBB Questiononthe size of the workforce duringthe quarter....................... Table 2.11: EconomicIndicators(1998-2001)..................................................................................... 61 62 62 Table 3.1. Turkey: ExtremePovertyandNumber of Children............................................................. Table 2.12: Actual and 1998Projectionsfor per capitaGNP (1998-2001)............................................. 63 Table 3.2. Turkey: ExtremePovertyandRegion................................................................................ 63 Table 3.3. Turkey: ProbitResultsfor ExtremePoverty........................................................................ 64 Table4.1. Turkey. RegionandUrbanFoodPovertyRates.................................................................. 65 Table 4.2. Turkey: HouseholdSize andUrbanFoodPovertyRates ..................................................... 65 Table 4.3. Turkey: Number of ChildrenandUrbanFoodPovertyRates............................................... 66 Table 4.4. Turkey: Number ofElderlyandUrbanFoodPovertyRates................................................. 66 Table 4.5. Turkey: Age andUrbanFoodPovertyRates...................................................................... Table 4.6. Turkey: Educationof HeadandUrbanFoodPoverty Rtes.................................................. 67 67 68 Table 4.8 Turkey: Ownership Status ofHouseandUrbanFoodPovertyRates..................................... Table 4.7. Turkey: MainActivity of HeadandUrbanFoodPovertyRates ........................................... 69 Table4.9 Turkey: HousingType andUrbanFoodPoverty Rates ........................................................ 69 Table4.10 Turkey: Wall Type andUrbanFoodPovertyRates............................................................ 70 Table 4.11 Turkey: TolietType andUrbanFoodPoverty Rates .......................................................... 70 Table 4.12 Turkey: Water Source andUrbanFoodPovertyRates ....................................................... 71 Table 4.13 Turkey: ElectrictyandUrbanFoodPovertyRates............................................................. 71 Table 4.14 Turkey: Cooking Sources andUrbanFoodPovertyRates .................................................. 72 Table 4.15. Turkey: HeatingSources andUrbanFoodPovertyRates .................................................. Table 4.16 Turkey: Garbage DisposalandUrbanFoodPovertyRates ................................................. 72 73 Table 4.17 Turkey: Sewage andUrbanFoodPovertyRates................................................................ 73 Table 4.18 Turkey: ConsumerDurables andUrbanFoodPovertyRates .............................................. 74 Table 4.19 Turkey: Assets andUrbanFoodPovertyRates+A69 ......................................................... 75 Table 4.20. Turkey: UrbanFoodPovertyProbitResults....................................................................... 76 Table 4.21 Turkey: Self-EvaluationandUrbanFoodPovertyRates..................................................... 77 Table 4.22. Turkey: Self-ComparisonwithPreviousYear andUrbanFoodPovertyRates..................... 77 Table 4.23 Turkey: Problemswith Satisfying FoodNeedsandUrbanFoodPovertyRates ..................... 78 Table 4.24 Turkey: CommunityEconomic Situation andUrbanFoodPovertyRates ............................ 78 Table 5.1. Turkey. RegionandUrbanVulnerability Rates.................................................................. 79 Table 5.2. Turkey. HouseholdSize andUrbanVulnerabilityRates...................................................... 79 Table 5.3. Turkey. Number of andUrbanVulnerabilityRates............................................................. 80 52  Table 5.4. Turkey. EducationofHeadandUrbanVulnerability Rates ................................................. 80 Table 5.5. Turkey. MainActivity of HeadandUrbanVulnerabilityRates............................................ 81 Table 6.1. Turkey: Inequality Measures 2001................................................................................... -82 Table 7.1. SocialRiskManagementinTurkey:............................................................................ 83 Table 7.2. Turkey: UrbanandRuralPopulation, 1960-2000....................................................... 83 Table 7.3 Turkey: RelativesAbroadandUrbanFoodPovertyRates.................................................... 84 Table 7.4 Turkey: Credit CardUse andUrbanFoodPovertyRates..................................................... 84 Table 7.5 Turkey: HealthInsurance andUrbanFoodPovertyRates .................................................... 84 Table 7.6 Turkey: SellingAssets or ValuablesandUrbanFoodPoverty Rates ..................................... 85 Table 7.7 Turkey: Cashor In-Kindto Other HouseholdsandUrbanFoodPovertyRates....................... 85 Table 7.8 Turkey: BorrowingfiomRelativesiFriendsandUrbanFoodPovertyRates........................... 86 Table 7.9. Turkey: ReportedConsumption andUrbanFoodPovertyRates................................ 86 Table 7.10 Turkey: FoodCopingandUrbanFoodPovertyRates ........................................................ 87 Table 7.11 Turkey: Non-FoodCopingandUrbanFoodPovertyRates ................................................. 87 Table 7.12 Turkey: EducationCopingandUrbanFoodPoverty Rates.................................................. 88 Table 7.13 Turkey: HealthCoping andUrbanFoodPovertyRates....................................................... 88 Table 7.14 Turkey: FoodQualityReductionandUrbanFoodPovertyRates ........................................ 89 89 Table 8.1. SelectedSocialIndicators-Country Comparison................................................................ Table 7.15. Turkey: Social SolidarityFundandUrbanFoodPoverty Rates.......................................... 90 53  K m - - m YI m t t o 3: w o + N e W Y) r m n m m w h w w w h o h w o m o w ~ m ~ o r - m o w o o h o w-0 m m ~ - w w 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 m w w N m m m m w m N 0 0 0 m m m m o r o - N r r .. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 m w W N m m m o w m N 0 0 0 m m m m 0 - 0 r w - r r i3 a, W .. c d P L jn 0 2. r0 c  z 2 W W I 5 4 0 c 4 z 4 2 4 I z C W 0 - w m w r our-0 b m m - z "NN P P r 2 0 m m w m W m o r - m I oar-* m m o - z I - W O O 4 "" W r z 4 I I D W P W W m m m m cn IC+" m - m o "NN- W I m r - m w N r - m N w o w m m"N- c 0 z 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 m m w ~ " m m m w m N 0 0 0 m m m m m m w w " m m m w m N 0 O O m m m m r m r N r r c N v c.7 - m r N r r r U c W W n I a, m c v-3 a n a A - V I m 21 z m - m L 8 0 0) y ?C > L 0 - 3 Y a, a Q VI -- m E m m C L O U L n n VI m m C : .^ : 0 W c a, 0 z a, x L 3 !-  z 2 0 W L b o o m i O P P P -1 0 c a 0 m z a z L Y W I- v) a W c c m P z 0 -mw oo m - o m m o ~ ~ m m omm- m w m c mu m a w m m m m m m m m m a a - w a m o w m w ~m m m b 2 ~ o m m h m ~ w m ~w ~u -co - w m m m m o 0 P P W - b ~ m m w w m m ~ r 0 I m a W v) 1 Y 0a -1 m - 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 w a a N m m m m a m N 0 0 0 m m m m m a a ~ m m m m a m ~ o o o m w w m - N m - m - N r - r 0 - m r N r r E W - J W > i W 0 a n m v) -- E m 2 0 3 L U V v) m m c W c 0 W /" 0 2. W x 3 c a a a a  > Y Y L a c z 0 W Y L 5 J 0 + a z a z L Y c v) a 0