Small area estimates of income poverty in Croatia: methodological report 1 Introduction The At-Risk-of-Poverty (AROP) rate indicates the percentage of individuals within a country who live on less than 60 percent of the median national equivalized disposable income after social transfers. It is one of the main indicators derived from the European Union Statistics on Income and Living Conditions Survey (EU- SILC). In Croatia the EU-SILC is representative at the NUTS1-1 level as well as at the NUTS-2. The National at-risk-of poverty rate for 20122 in Croatia is 20.4 percent. While regional poverty rates are considerably different between Continental and Adriatic Croatia, 22 and 17 percent respectively. Nevertheless it is possible that poverty levels within NUTS-23 spatial units, differ considerably. Figure 1: EU-SILC poverty map at level of representativeness 1 Nomenclature of territory units for statistics (NUTS) based on Regulation (EC) No 1059/2003 of the European Parliament and of the Council of 26 May 2003 on the establishment of a common classification of territorial units for statistics 2 In the EU-SILC survey income information is gathered on the previously completed calendar year. 3 Presently there are two regions under NUTS-2 level, Adriatic and Continental Croatia. During the pre-accession period time of the 2012 EU-SILC there were three statistical regions corresponding to NUTS-2 level in Croatia: Northwest, Central and Eastern, and Adriatic Croatia. The 2012 EU-SILC is representative for the three statistical regions corresponding to NUTS 2 level. Continental Croatia is composed of the Northwest, and the Central and Eastern statistical regions. 1 Poverty figures at lower levels of aggregation (for example NUTS-3, LAU-1, or LAU-2) for Croatia are not possible with the EU-SILC. Geographical levels at which direct estimates lack the required precision are referred to as small areas (Guadarrama et al., 2015). Small area estimation (SAE) methods are those which seek to overcome the lack of precision. SAE methods achieve this by incorporating data sources with larger coverage. These methods present a way to circumvent the low representativeness of household survey methods by taking advantage of larger coverage surveys such as a census. In practice household surveys provide a satisfactory measure of welfare but possess low coverage, while the census has the coverage but lacks a suitable welfare measure. SAE methods take advantage of the best attributes of each data source in order to obtain welfare measures at levels of aggregation below those of the household survey’s representativeness. The use of SAE methods provides estimates of higher precision for small areas than those obtained using a household survey alone. Higher precision of welfare for smaller areas allows policy makers to better target assistance and interventions to the most disadvantaged communities. The Census of Population, Households and Dwellings of 2011 for the Republic of Croatia when combined with the 2012 EU-SILC facilitates the estimation of welfare at the household level. This makes obtaining poverty rates for areas below those of the EU-SILC’s representativeness possible. The small area estimation methodology used to obtain the estimates follows the one proposed by Elbers, Lanjouw, and Lanjouw (ELL) (2003).4 The methodology is perhaps the most widely used for small area estimation, and has been applied to develop poverty maps in numerous countries across the globe. Through the application of the analysis predicted poverty rates at the NUTS-3,5 as well as at the LAU-26 levels are obtained. 2 Modeling approach The ELL method is conducted in 2 stages. The first stage consists in fitting a welfare model using the 2012 EU-SILC data via ordinary least squares (OLS), and correcting for various shortcomings of this approach to arrive at generalized least squares estimates (GLS). It should be noted that the variables included in the welfare model of the 2012 EU-SILC must be restricted to those variables that are also found on the 2011 Census. This allows us to generate the welfare distribution for any sub-population in the 2011 Census, conditional on the sub-population’s observed characteristics (ELL, 2002). 4 The methodology is implemented via the World Bank developed software PovMap (accessed on August 1, 2016) 5 There are currently 21 NUTS-3 spatial units (Counties) in Croatia 6 There are 556 Local Administrative Units at level 2 (LAU-2). In Croatia LAU-2 level corresponds to municipalities and cities. Additionally, for the purposes of the analysis, the city of Zagreb is sub-divided into 19 districts. 2 After correcting for shortcomings, the estimated regression parameters, standard errors, and variance components from the EU-SILC model provide the necessary inputs for the second phase of the analysis. The second stage of the poverty mapping exercise consists in using the estimated parameters from the first stage, and applying these to the 2011 Census data in order to predict welfare at the household level. Finally, the predicted welfare measure is converted into a poverty indicator which is then aggregated in order to obtain poverty measures at the desired level of aggregation (NUTS-2, NUTS-3, or LAU-2). Before fitting the welfare model, a comparison between the observable household characteristics from the EU-SILC and the Census is necessary. The purpose of the comparison is to ensure that variables have similar distributions, and that these have similar definitions across data sources. Because the exercise consists in predicting welfare in the census data using parameters obtained from EU-SILC observed characteristics, it is imperative that the observed characteristics across surveys are comparable. The next step in the ELL methodology consists in estimating a log adult equivalized household income model which is estimated via OLS. The transformation to log income is done because income tends to not be symmetrically distributed (graph 1), taking the logarithm of income is done to make the data more symmetrical. Figure 2: Adult equivalized income and natural logarithm of equivalized income Kernel density estimate Kernel density estimate .00015 .8 .6 .0001 Density Density .4 .00005 .2 0 0 0 10000 20000 30000 40000 50000 4 6 8 10 12 Equivalised disposable income Nat. log of eq. income kernel = epanechnikov, bandwidth = 471.2075 kernel = epanechnikov, bandwidth = 0.0889 The household income model is: 𝑙𝑛 í µí±¦í µí±?â„Ž = í µí±‹â€²í µí±?â„Ž 𝛽 + í µí±¢í µí±?â„Ž (1) where í µí±¦í µí±?â„Ž is the adult equivalized income of household h in municipality c, í µí±‹í µí±?â„Ž are the household and locality7 characteristics, and í µí±¢í µí±?â„Ž is the residual. In the specified model the use of Households within a same municipality are usually not independent from one another and the following specification is used to account for this: 7 As mentioned above, the locality in the case of the Republic of Croatia refers to LAU-2, and districts of Zagreb 3 í µí±¢í µí±?â„Ž = í µí¼‚í µí±? + í µí¼€í µí±?â„Ž (2) where 𝜂 and 𝜀 are assumed to be independent from each other and uncorrelated with the observables, í µí±‹í µí±?â„Ž . 2 2 2 Households in the same location share the same 𝜂 , and it is expected that 𝐸[í µí±¢í µí±?â„Ž ] = 𝜎𝜂 + 𝜎𝜀 the larger the variance of 𝜂 the less precise the estimates of welfare will be when the spatial correlation of the residuals is ignored. 2 2 The estimation of 𝜎𝜂 and 𝜎𝜀 is done utilizing Henderson’s method III (Henderson, 1953).8 In the case 2 where the variance of the household specific error, 𝜎𝜀 , is assumed to vary across households a parametric form of heteroscedasticity is assumed and modeled as: 2 í µí¼€Ì‚í µí±?â„Ž 𝑙𝑛 [ 2 ] = í µí±?â€²í µí±?â„Ž 𝛼 + í µí±Ÿí µí±?â„Ž (3) 𝐴 − í µí¼€Ì‚í µí±?â„Ž 2 2 where 𝐴 = 1.05max(í µí¼€Ì‚í µí±?â„Ž ). 9 Making use of these estimates it is possible to obtain an estimate for 𝜎𝜀,í µí±?â„Ž . The existence of the variance parameters require a re-estimation of the welfare model given that the OLS assumptions are unlikely to hold. The variance covariance matrix utilized for the GLS estimates is household cluster specific, and where the interrelatedness between households within a cluster is also allowed.10 Once GLS estimates are obtained it is possible to move on to the second stage of the exercise. Small area estimates of welfare (and standard errors) are obtained by applying the parameter and error estimates from the survey to the census data. In order to do this we must simulate welfare. Since poverty indices are based on non-linear forms of log adult equivalized income simulations are ideally suited for obtaining estimates of Ìƒí µí±?â„Ž for each household is simulated making use these measures. A value of log income per adult equivalent 𝑦 of the 𝛽, 𝜂 , and the 𝜀 parameters from the first stage, where each simulation 𝑟 is equal to: 𝑦 ̃ 𝑟 + 𝜂 ̃ 𝑟 í µí±?â„Ž = í µí±’í µí±¥í µí±?(í µí±‹â€²í µí±?â„Ž 𝛽 Ìƒí µí±?𝑟 𝑟 + í µí¼€Ìƒí µí±?â„Ž ) (4) ̃ 𝑟 are drawn from bootstrapped versions of the EU-SILC sample.11 On the For each simulation a set of 𝛽 other hand for the location and household disturbance terms we obtain their variance parameters, 2 𝑟 2 𝑟 𝑟 (𝜎𝜀,í µí±?â„Ž ) and (𝜎𝜂 ) , from the 𝑟 𝑡ℎ bootstrapped version of the EU-SILC. 𝜂 Ìƒí µí±?𝑟 and í µí¼€Ìƒí µí±?â„Ž are thus drawn from a 2 2 𝑟 𝑟 normal distribution assuming mean zero and variances equal to (𝜎𝜀,í µí±?â„Ž ) and (𝜎𝜂 ) , respectively. If we ̃ 𝑟 í µí±?â„Ž ) define 𝑓(𝑦 as a function that maps the estimated adult equivalized income measure to a poverty 8 An additional method is the one proposed by ELL (2003) 9 For a more detailed description, interested readers should refer to Elbers, Lanjouw and Lanjouw (2003) as well as Van der Weide (2014) 10 For details on the structure of the variance covariance matrix refer to Van der Weide (2014). 11 An alternative option is to draw the 𝛽 from a multivariate normal distribution 𝛽~í µí±?(𝛽 𝑔𝑙𝑠 , í µí±£í µí±?𝑜𝑣(𝛽𝑔𝑙𝑠 )) 4 measure such as the at-risk of poverty head-count-rate (FGT 0) then the estimated mean poverty rate for a municipality í µí±? is equal to: 𝑅 𝐻 1 ̃ 𝑟 í µí±?â„Ž )í µí±¤í µí±?â„Ž 𝐹𝐺𝑇0í µí±? = ∑ ∑ 𝑓(𝑦 (5) 𝑅 𝑟=1 â„Ž=1 where í µí±¤í µí±?â„Ž is the population expansion factor (number of household members in household â„Ž divided by the total population of Croatia in the census). An alternative for the estimation of 𝜂 is to use the information from the survey, Empirical-Best estimation (EB). The best estimate available to us of 𝜂 , for a particular municipality is that which comes from the survey (ln í µí±¦í µí±?â„Ž − í µí±‹â€²í µí±?â„Ž 𝛽 = í µí±¢í µí±?â„Ž ). Therefore making use of this information the estimates for the municipalities, cities, and districts of Zagreb that are present in the EU-SILC are tighter since more information is included into their drawing. For all locations that are not present in the EU-SILC, the use of EB makes no difference, since for these localities there is no additional information and thus their data generation process is still 2 𝑟 normal with mean zero and variance (𝜎𝜂 ) . Within the estimated measures there are three main sources of error: model error, error due to the disturbance, and due to computation error. These three sources of error, as noted by ELL (2003) are not correlated. The error in the welfare measure within a municipality due to the disturbance arises as a result of unobserved components of income within a particular locality. The smaller the population of the targeted municipality the larger this error will be, and thus limits the degree of disaggregation possible. The exact point at which this becomes unacceptable depends on how well the model fits the data. The model error depends entirely on the properties of the first stage estimators it is independent from the population size of the municipality. Within a given municipality the magnitude of this error component will also depend on how different the 𝑋 variables are in that municipality from those of the EU-SILC data. Finally, computation error is due to the method used for computation. This error can be made as small as possible depending on computational resources at hand. Because often simulations are a finite number, the larger the number of simulations, the smaller the error due to computation will be. 3 Data description The poverty mapping analysis requires two sources of data. In this instance the Croatian EU-SILC for 2012, and the Census of Population, Households and Dwellings of 2011 for the Republic of Croatia. The EU-SILC 5 for 2012 is an ideal household survey for the SAE analysis because incomes reported in the 2012 EU-SILC correspond to 2011 calendar year, and thus are for the same time period as the census. Small area estimation is done under the assumption that the same underlying population is being captured by the survey and the census. This last assumption will be valid if both datasets are from the same time frame. Nevertheless, the inclusion or the use of datasets that are from differing time periods, or if the survey is not representative of the population, will break down this assumption. This last remark is more salient in instances where there have been considerable shocks in between the collection of the survey and the collection of the census (Bedi et al. 2007). 3.1 EU-SILC 2012, Croatia The EU-SILC data is the EU reference source for comparative statistics on income and social exclusion. The 2012 EU-SILC for Croatia was made up of 5,853 households and is representative at the NUTS-2 level. The at risk of poverty threshold12 in Croatia for 2012 (income year 2011) is 24,000 HRK. Using this poverty threshold, the at-risk-of-poverty head count rate is 20.4 percent. The 2012 EU-SILC uses the 2001 Census as a sampling frame. The survey is performed as a stratified two- stage sample. The at-risk-of-poverty threshold is obtained by including all households, among these 2 have reported negative net disposable incomes. For purposes of the analysis done these households are no longer included. The households included in the EU-SILC dataset come from 370 municipalities. Finally, all municipalities with less than 3 households in the EU-SILC must be removed for the analysis.13 The final sample for the EU- SILC is made up of 5,618 households. 3.2 Census of Population, Households and Dwellings 2011, Population by Sex and Age The 2011 Census for Croatia was provided by the Croatian Bureau of Statistics.14 The census includes key information on demographics of the household, education, labor force status, economic activity, occupation type, and labor status in main job. Along with these characteristics, the census also has information on the type of dwelling, the status of the dwelling, number of rooms in the dwelling, living area of the dwelling, and the construction year. 12 60 percent of the median household equivalized income 2 13 This is necessary in order to estimate the variance of the location effect, í µí¼Ží µí±?â„Ž , for every municipality. 14 Access to the Census, as well as the EU-SILC (with excluded direct identifiers of persons and households) was provided in the Croatian Bureau of Statistics’ safe room according to the Agreement and inclusion of this exercise in the Annual Implementation Plan 2016. 6 3.3 Variable comparison between EU-SILC and Census Because small area methods require an estimation of a welfare model in the first stage which will then be applied to the census it is necessary that the choice of correlates matches across surveys. This not only requires variables to be similar, but requires that these have similar distributions. The selection of candidate variable is done in a two stage process: 1. Comparison of questionnaires between the EU-SILC and the Census. The comparison yields a first set of candidate variables for the estimation. Candidate variables must come from similar questions. 2. Comparison of the distribution of the candidate variables across datasets. The comparison is undertaken at the level of Republic of Croatia and at the NUTS-2 level. The comparability of the variables across surveys ensures that the welfare model from the 2012 EU-SILC can be applied to the Census such that reliable income estimates for the population can be derived. Making use of all variables that meet the above criteria several welfare models are estimated via OLS. Unlike most of econometrics, the purpose of the model is not to find any causal relationships but to find a model that best reflects the income level of a household. The income of a household is assumed to be a function of the number of household members present in the household, and the age composition of the household members. Additionally, income is assumed to be a function of the marital status of individuals aged 15 and over, their level of education, their occupation, and the sector in which they are employed in. In addition, and while likely not a determinant of income, we include a variable which reports the area of the dwelling in square meters. This variable is expected to have reasonable correlation with welfare. Finally, the use of location means of household level variables are included.15 This is done in order to explain the variation in welfare due to location as much as possible and thus improve precision of the welfare estimates. Table 1 contains a listing of the candidate variables for use in the model. The EU-SILC and the Census contain a comprehensive set of variables which match the criteria for modelling income at the household level. Both datasets contain information on the number of household members present in a household. Given that the sampling frame for the 2012 EU-SILC is the previous Census (Census of Population, Households and Dwellings 2001) it is not unexpected that the first moments of the EU-SILC and Census are somewhat different. Nevertheless, at the national level the means of the candidate variables match up considerably well. 15This is recommended by ELL (2002). Variable means at the municipal level are included and come from the Census. These are the share of households in the municipality that were built between 1990 and 2000, share of household that have sewerage access, share of individuals that receive pension income, and the share of employed individuals in the municipality. 7 The mean values for the EU-SILC and for the Census are presented. The final choice of variables for the model is not only dependent upon how well the variables match up, but on how well they explain the variation of income. As the numbers on Table 1 illustrate, the two datasets match up quite well. The age groups, proportion of males, and household size are very close to one another, even at the statistical area level the variables are comparable with one another (Table 1A). Comparison between labor market variables also reveal that the datasets are close to each other with some differences arising in some of the occupations. Similarly these slight differences are also reflected at the regional level comparisons. Given that the differences that arise are not considerable all of the variables are valid candidates for the welfare model to be estimated in the next stage. Variables that are highly correlated are not included simultaneously. Keeping this in mind the selected model is the one which maximizes the adjusted R-squared of the model, but at the same time conforms to prior beliefs of how should the variable be related to income. Table 1: Population weighted candidate variable means in Census and EU-SILC EU- Variable name Census SILC Male 0.483 0.482 Age [0,5) 0.050 0.045 Age [5,15) 0.103 0.106 Age [15,30) 0.186 0.186 Age [30,65) 0.486 0.490 Age [65+) 0.174 0.172 Household size (Share of individuals living in household type) Households size of 1 0.088 0.088 Households size of 2 0.183 0.183 Households size of 3 0.202 0.202 Households size of 4 0.248 0.247 Households size of 5 0.143 0.143 Households size of 6 0.076 0.073 Household size of 7 or more 0.060 0.063 Occupation (15+) (Share of individuals in households with at least one member) Manager 0.051 0.032 Professionals 0.150 0.142 Technicians 0.182 0.132 Clerical support 0.129 0.118 Service and sales 0.223 0.214 8 Skilled agriculture 0.041 0.051 Craft and trade 0.153 0.167 Machine operators 0.112 0.117 Elementary occupations 0.091 0.071 Labor status, age 15-64 (Share of individuals in households with at least one member) Employed 0.742 0.724 Retired 0.497 0.503 Student 0.220 0.213 Disabled 0.038 0.024 Other 0.749 0.726 Industry, age 15-64 (Share of individuals in households with at least one member) Agriculture, mining, and fishing 0.065 0.068 Manufacturing 0.189 0.195 Services and Sales 0.630 0.572 Share of members with education in HH (age 15-64) Primary education 0.086 0.071 Lower secondary 0.199 0.196 Upper secondary 0.547 0.595 Tertiary education 0.169 0.138 Dwelling characteristics Square meters 87.542 88.942 4 Model results The initial welfare model corresponding to equation (1) is presented in column 1 of Table 2. The adjusted R- Squared for the model is (0.52) reflecting that the chosen model explains the variation on income well. In addition to the variables present in both the Census and EU-SILC, variable means for municipalities, cities, and districts of Zagreb are obtained from the Census and introduced to the model; these variables are introduced to improve precision by reducing the unexplained variation in income due to location. With the inclusion of these variables the ratio of the variance of 𝜂 over the model’s MSE is 0.035. The low ratio illustrates the key role the variables play in improving precision of the estimates. 9 Table 2: Weighted OLS & GLS estimates for Income model: 2012 EU-SILC Coeff. WOLS Coeff. GLS Intercept 8.4124*** 8.5379*** No children under 5 -0.104*** -0.0781*** No children between 5 and 15 -0.1322*** -0.1294*** One child between 5 and 15 -0.0795** -0.0834** No indiv. with lower secondary 0.0433** 0.045** No indiv. with primary 0.2104*** 0.1671*** One individual with primary 0.1113 0.0943 One person with tertiary education 0.1123*** 0.0989*** Two people with tertiary education 0.1207*** 0.1299*** 1 member HH 0.8795*** 0.9324*** 2 member HH 0.7396*** 0.8062*** 3 member HH 0.533*** 0.5899*** 4 member HH 0.3815*** 0.4271*** 5 member HH 0.1972*** 0.2414*** 6 member HH 0.1801*** 0.2069*** Nat. log Sq. M 0.1091*** 0.0933*** No married ind. In HH -0.1337*** -0.134*** Proportion of dwellings built 1990-2000 0.3398** 0.3602** Proportion of dwellings with sewerage 0.0967*** 0.0891*** Proportion of HH with pension income 1.0688*** 0.994*** Municipal employment rates 0.9721*** 0.9221*** No ind. is a clerk -0.1071*** -0.1107*** No ind. is elementary teacher 0.0743* 0.0752** No ind. is a manager -0.2233*** -0.224*** No ind. is a professor -0.174*** -0.1781*** No ind. is a technician -0.1427*** -0.1298*** Northwest × no lower education 0.0966*** 0.074** Northwest × 2p retired 0.0101 0.0251 Central East × lnM2 0.1009** 0.1074*** Central East × 2p workers -0.0755* -0.0819** Central Eastern -0.3389* -0.3659** Adriatic 0.1142*** 0.1063*** 1 retiree 0.2299*** 0.1921*** 2 retirees 0.2733*** 0.2303*** 0 administrative workers 0.085* 0.0788** 0 public employees -0.1317*** -0.1248*** 1p working in HH 0.5493*** 0.5428*** 2p working in HH 0.3499*** 0.3463*** 3p working in HH 0.1464*** 0.1529*** Adjusted R-squared 0.52 10 Ratio of variance of η over Mean Sq. error 0.035 Number of observations 5,618 5,618 *, **, *** significant at the 10, 5, 1 percent level respectively. All households which have inconsistent labor information are removed. As noted in section 2, it is likely that income levels within a location are highly correlated and as a consequence 𝐸[í µí±¢í µí±?â„Ž í µí±¢í µí±?𝑖 |𝑋] ≠ 0. Additionally, error terms will likely have differing variances across 2 observations (𝐸[í µí±¢í µí±?â„Ž |𝑋] ≠ 𝜎 2 ). Due to these issues the model is re-estimated using Generalized Least Squares (GLS). The results for the GLS fitted model are presented in column 2 of Table 2.16 Equivalized income is positively correlated to household size. The omitted group is households with 7 or more individuals. Furthermore, equivalized income is negatively correlated to the absence of children in the household. Under the modified OECD scale, when comparing two households with equal household income, the household with lower adult equivalents will have greater adult equivalized income. Thus, all else equal, a household with 2 adults and a child will have greater adult equivalized income than one with 3 adults. Households with retirees also have greater equivalized incomes, this is most likely due to pensions being received by these individuals. After labor the most important source of labor income in Croatia is pension income. Education is also strongly correlated to equivalized income, households with members who have tertiary education have on average greater equivalized incomes. Also correlated to income is the presence of working members and most of the labor variables included are significantly correlated to equivalized income. Among these variables, the presence of working members have the greatest coefficients. Location, and location variable means are also correlated to equivalized income. Adult equivalized income is negatively correlated to being located in Central and Eastern Croatia as opposed to being in the Northwest. On the other hand residing in the Adriatic is positively and significantly correlated to adult equivalized income. In addition, equivalized income is positive and significantly correlated to localities with higher shares of households with pension incomes, households with sewerage, and dwellings built between 1990 and 2000. 16 The alpha model (equation 3) corresponding to the GLS is presented in Table 2A. 11 5 Poverty results The coefficients estimated in the previous section provide the necessary inputs in order to estimate the first Ì‚ ) by combining coefficients with the Census variables. The vectors of disturbances part of equation 4 (í µí±‹â€²í µí±?â„Ž 𝛽 for households are unknown, and must be estimated. As mentioned before, the error component is decomposed using Henderson’s method III, and the coefficients, 𝛽, are obtained by bootstrapped samples of the EU-SILC data. The model chosen is where 𝜂 and 𝜀 are drawn from a normal distribution, with their respective variance structures. Finally, empirical best methods are chosen since these incorporate more information and are thus expected to provide a better fit. The clustering used for estimations is at the municipal, city, and districts of Zagreb level, the resulting poverty map aggregated to the NUTS-3 level is presented in Figure 3 and at the municipal, city, and districts of Zagreb level in Figure 4. The resulting poverty rates used for validation of the small area estimation undertaken are presented in Table 3. These compare the poverty rates obtained from the small area estimation to the direct estimates from the EU-SILC at the statistical area level. This provides support to the quality of the estimates obtained. Table 3: Poverty rates from EU-SILC and from poverty map exercise AROP EU-SILC Statistical region EU-SILC 95% CI Predicted 95% CI Northwestern 16.7% 13.6% 20.4% 14.1% 12.8% 15.5% Central & Eastern 29.1% 26.2% 32.2% 28.0% 25.7% 30.2% Adriatic 17.0% 14.0% 20.6% 17.4% 15.8% 19.1% Total 20.4% 18.5% 22.4% 19.2% 18.0% 20.4% Note: Poverty line is at 24,000 HRK per adult equivalent Results at the NUTS-3 spatial unit level are presented in Table 4. These estimates illustrate the heterogeneity within the country. Within the Adriatic region poverty rates range from 11.9 to 25.2 percent, within Continental Croatia (composed of the Northwestern, and Central and Eastern statistical area) poverty ranges from 9.8 percent in Grad Zagreb, to 35.9 percent in Brodsko-posavska. Poverty levels within the Central and Eastern statistical area are considerably greater than the country average. At the municipal, city, and districts of Zagreb level further heterogeneity is revealed. In the Continental NUTS-2 region certain pockets of high poverty levels are detected, particularly in the Central and Eastern statistical region. In the Adriatic region some municipalities with higher poverty rates are also observed. The 12 results of the poverty map suggest an overall spatial clustering of poverty; this is further analyzed in section 6, where basic analysis of the spatial association is undertaken. Figure 3: Poverty Map for Croatia (NUTS-3 poverty headcount) Finally, the distribution of the Republic of Croatia’s population that is at-risk-of-poverty is illustrated in Figure 5. The County with the lowest concentration of poor is in the Adriatic region, LiÄ?ko-senjska. The county is one of the least populated in the country, and although it has an at-risk-of-poverty rate which is close to 20 percent it has the fewest poor. On the other hand Grad-Zagreb which is the least poor county in the Republic of Croatia with an at-risk-of-poverty rate close to 10 percent has the third highest concentration of the country’s poor. 13 Figure 4: Poverty Map for the Republic of Croatia (poverty headcount for municipalities, cities, and districts of Zagreb) 14 Table 4: County level poverty estimates EU-SILC direct estimates H3-EB Model prediction Statistical Area AROP 95% CI NUTS-3 (counties) Population AROP 95% CI ZagrebaÄ?ka 311,918 16.7% 13.9% 19.5% Krapinsko-zagorska 129,393 18.8% 15.9% 21.7% Varaždinska 170,380 17.1% 14.6% 19.7% Northwestern 16.7% 13.6% 20.4% KoprivniÄ?ko-križevaÄ?ka 112,540 20.3% 17.4% 23.3% MeÄ‘imurska 110,888 20.8% 17.5% 24.0% Grad Zagreb 772,340 9.8% 8.0% 11.6% SisaÄ?ko-moslavaÄ?ka 168,534 23.7% 19.6% 27.8% KarlovaÄ?ka 125,722 23.2% 19.4% 27.1% Bjelovarsko-bilogorska 117,420 20.0% 15.6% 24.5% VirovitiÄ?ko-podravska 83,129 33.4% 28.7% 38.2% Central & Eastern 29.1% 26.2% 32.2% PožeÅ¡ko-slavonska 75,912 26.5% 21.1% 31.9% Brodsko-posavska 154,863 35.9% 31.6% 40.1% OsjeÄ?ko-baranjska 297,230 28.0% 24.8% 31.1% Vukovarsko-srijemska 174,324 31.9% 28.4% 35.3% Primorsko-goranska 290,446 11.9% 10.0% 13.8% LiÄ?ko-senjska 49,766 19.8% 15.7% 24.0% Zadarska 167,029 25.2% 20.9% 29.5% Adriatic 17.0% 14.0% 20.6% Å ibensko-kninska 107,345 24.7% 20.7% 28.8% Splitsko-dalmatinska 445,049 19.5% 16.9% 22.0% Istarska 204,025 11.9% 9.6% 14.1% DubrovaÄ?ko- neretvanska 118,707 14.5% 11.3% 17.8% Republic of Croatia 20.4% 18.5% 22.4% 4,186,960 19.2% 18.0% 20.4% Note: Poverty line is at 24,000 HRK per adult equivalent 15 Figure 5: Distribution of the poor by NUTS-3 spatial units for the Republic of Croatia 16 6 The use of poverty maps 6.1 Local indicators of spatial association of poverty Using the poverty map output we seek to determine if there is a pattern to how poverty rates of municipalities, cities, and districts of Zagreb are distributed within the Republic of Croatia. When analyzing geographical data it is assumed that things that are closer are more related to things that are farther away (Tobler, 1970). This supposes that two municipalities that are closer together will be more alike than municipalities which are farther away. As noted in Section 5 and in Figure 4, there appears to be some spatial clustering in the results from the poverty maps. In fact the Central and Eastern statistical area seems to be lagging behind the Adriatic and Northwest. This illustrates a divergence within the Continental NUTS-2 region. Poverty rates in Central and Eastern regions are considerably greater than the rest of the country, and the region appears to be a hotspot for poverty. Furthermore, there appears to be a clear demarcation of low versus high poverty areas. Insofar as determining if there is in fact spatial correlation we rely on Global Moran’s I as well as Local Moran’s I statistic. In order to obtain undertake analysis of spatial association it is necessary to establish a degree of spatial proximity between the locations in Croatia. In order to do this, a spatial weights matrix is used, which relies on the row-standardized inverse distances between the center of the municipalities and the surrounding municipalities. This ensures that nearer neighbors have a greater influence on the analyzed outcomes, in this instance poverty rates. The presence of spatial association is confirmed by a global Moran’s I index of 0.52 which is significant at the 1 percent level. Local Moran’s I can aide in identifying which localities have a statistically significant relationship with its neighbors. Spatial autocorrelation facilitates the identification of high poverty areas noted in the map presented in Figure 4 (particularly in the Central and Eastern statistical area within the Continental NUTS-2), as well as low poverty areas (around Zagreb and the surrounding areas of Istarska). These results bring to light the challenges that arise for regional development, and add a new layer to the discussion. Figure 6 presents the results for the Global and Local Moran’s I statistics. The significant Global Moran’s I of 0.52 suggest that there is spatial autocorrelation. Additionally, the map illustrates regions which are significantly different from their neighbors, and regions which are high-poverty areas and low poverty areas. All colored areas show a significant relationship to their neighbors. Those locations marked as “High – Highâ€? (“Low-Lowâ€?) are areas where poverty is significantly greater (lower) than the neighborhood’s poverty and are greater (lower) than the average poverty among municipalities, cities and districts of Zagreb. 17 A cluster of high poverty is clearly delineated in the Eastern Central statistical area (Figure 6 and 7). In Zagreb and surrounding areas a cluster of low poverty is highlighted, the same holds true for the north of the Adriatic region. Municipalities, cities, and/or districts of Zagreb marked as low-high outliers and the high-low outliers are particularly of interest. While poverty may be high (low) in particular areas, there are some municipalities that have a significantly lower (higher) level of poverty than its surroundings. These are mostly observed in the Adriatic and Eastern Central areas. The hot spot analysis in Figure 8, brings to light a demarcation and separation between regions. This was also evident in the results from the OLS and GLS (see Table 2). All three statistical areas are different. Independently from the NUTS-2 classification which aggregates the Northwestern statistical area and the Eastern and Central statistical area, when it comes to welfare these areas are considerably different. Figure 6: Poverty Map for the Republic of Croatia (Spatial association of headcount poverty) 18 Figure 7: Poverty Map for the Republic of Croatia: hot spot analysis (Getis-Ord Gi) 7 Concluding remarks Direct poverty estimates from the EU-SILC are only reliable at the statistical area level, and thus at the NUTS-2 level. This complicates the analysis of poverty at more disaggregated levels since the reliability of direct estimates are questionable. Data from the Census of Population, Households and Dwellings 2011 coupled with small area estimation techniques aide policy makers overcome the lack of precision at lower geographical levels. The results from the poverty mapping exercise, coupled with spatial analysis reveal the heterogeneity of poverty in Croatia. 19 Results from spatial analysis reveal that there is a cluster of high poverty in the Central and Eastern region of Croatia. There is a clear poverty demarcation in the country, where the Central and Eastern part of the country is clearly doing worse than the rest of the country. Results also reveal that while the Continental NUTS-2 spatial unit, may seem poorer than the Adriatic, the result is mainly driven by the aggregation of the two statistical regions (Northwest, and the Central and Eastern statistical regions). The use of the poverty map in order to assist in the guidance of resource allocation can help policy makers achieve considerable gains in poverty reduction. Additionally, the visual format of the maps is simple to understand which makes it easy for the population at large to take notice of where their community stands compared to the rest of the country. Moreover, because the maps are based on established data sets, these are objective. As a consequence the maps may help prevent subjective decision making. Given the mentioned uses of the poverty maps these are valuable component of the policy maker’s tool kit when trying to decide where limited funds can be distributed among the population which needs assistance. 20 8 References Baric, M., & Williams, C. (2015). Tackling the undeclared economy in Croatia. South-Eastern Europe Journal of Economics, 11(1). Bedi, T., Coudouel, A., & Simler, K. (Eds.). (2007). More than a pretty picture: using poverty maps to design better policies and interventions. World Bank Publications. Elbers, C., Lanjouw, J. O., & Lanjouw, P. (2002). Micro-level estimation of welfare. World Bank Policy Research Working Paper, (2911) Elbers, C., Lanjouw, J. O., & Lanjouw, P. (2003). Micro–level estimation of poverty and inequality. Econometrica, 71(1), 355-364. Elbers, C., Fujii, T., Lanjouw, P., Özler, B., & Yin, W. (2007). Poverty alleviation through geographic targeting: How much does disaggregation help?. Journal of Development Economics, 83(1), 198-213. Guadarrama, M., Molina, I., & Rao, J. N. K. (2016). A Comparison of Small Area Estimation Methods for Poverty Mapping. STATISTICS IN TRANSITION new series and SURVEY METHODOLOGY, 41. Tobler, W. R. (1970). A computer movie simulating urban growth in the Detroit region. Economic geography, 46(sup1), 234-240. Van der Weide, R. (2014). GLS estimation and empirical bayes prediction for linear mixed models with Heteroskedasticity and sampling weights: a background study for the POVMAP project. World Bank Policy Research Working Paper, (7028). 21 9 Appendix 9.1 Mathematical appendix The discussion below presents the methodology detailed by ELL (2002 and 2003). Interested reader should refer to these documents for the full discussion. ˆch , and by defining u From the estimation of equation 1 we obtain the residuals u ˆc. as the weighted ˆch for a specific cluster we can obtain e average of u ˆch : The variance of the location effect (í µí¼‚í µí±? ) is given by: where 𝑢. . = ∑𝐶 í µí±¤í µí±? í µí±¢í µí±? . (where the í µí±¤í µí±? represents the cluster’s weight) and: ∑ℎ í µí±’í µí±?â„Ž where í µí±’í µí±?. = í µí±›í µí±? (í µí±›í µí±? is the number of households in the cluster). The parametric from of heteroscedasticity is presented as: 2 This is simplified by setting 𝐵 = 0 and 𝐴 = 1.05max(í µí±’í µí±?â„Ž ), which leads to the simpler form that can be estimated via regular OLS: By defining 𝐵 = exp(í µí±?í µí±?â„Ž 𝛼) and using the delta method the household specific variance for í µí±’í µí±?â„Ž is equal to: 22 The use of 𝜎2 2 𝜂 and 𝜎𝜀 allows us to get the variance covariance matrix used for the OLS estimates: The estimates for the GLS detailed by ELL (2003) are: and In response to criticisms of the methodology an extensive revision was made to the methods, including the addition of empirical best estimation, by Van der Weide (2014). For a detailed discussion on the EB approach and the other changes implemented readers are guided towards Van der Weide (2014). The revisions include an improved GLS estimator: and a new variance covariance matrix: These are the estimates used for the second stage of the estimation (detailed in the methods section). 23 9.2 Poverty mapping software One of the most common small area methods used for poverty mapping was proposed by Elbers, Lanjouw, and Lanjouw (2003). This methodology has been widely adopted by the World Bank and has been applied in numerous poverty maps conducted by the institution. In its efforts to make the implementation of the ELL methodology as simple as possible, the World Bank created a software package that could be easily used by anyone. The software, PovMap (Zhao, 2006), has proven to be an invaluable resource for the World Bank as well as for many statistical agencies seeking to create their own poverty maps. The software is freely available and has a graphical user interface which simplifies its use. Poverty map results produced in this document have all made use of the PovMap software. The PovMap software can be downloaded, free of charge, at http://iresearch.worldbank.org/PovMap/PovMap2/. 24 9.3 Additional tables and graphs Table 1A: Population weighted candidate variable means in Census and EU-SILC at the Statistical Area levels Northwest Central & Eastern Adriatic Variable name Census EU-SILC Census EU-SILC Census EU-SILC Male 0.4777 0.4771 0.4843 0.4832 0.4873 0.4870 Age [0,5) 0.0515 0.0442 0.0476 0.0512 0.0483 0.0400 Age [5,15) 0.1021 0.1079 0.1082 0.1050 0.0992 0.1059 Age [15,30) 0.1872 0.1873 0.1897 0.1897 0.1817 0.1817 Age [30,65) 0.4937 0.4964 0.4764 0.4801 0.4899 0.4920 Age [65+) 0.1655 0.1642 0.1782 0.1740 0.1810 0.1805 Household size (Share of individuals living in household type) Households size of 1 0.086 0.087 0.086 0.087 0.088 0.090 Households size of 2 0.175 0.173 0.181 0.183 0.195 0.196 Households size of 3 0.200 0.199 0.189 0.189 0.215 0.217 Households size of 4 0.243 0.244 0.237 0.238 0.260 0.257 Households size of 5 0.144 0.143 0.154 0.147 0.133 0.140 Households size of 6 0.083 0.089 0.085 0.081 0.061 0.046 Household size of 7 or more 0.070 0.065 0.067 0.074 0.047 0.053 Occupation (15-64) (Share of individuals in households with at least one member) Manager 0.066 0.032 0.031 0.015 0.052 0.048 Professionals 0.188 0.173 0.107 0.103 0.145 0.140 Technicians 0.214 0.151 0.140 0.095 0.183 0.140 Clerical support 0.150 0.129 0.103 0.072 0.127 0.145 Service and sales 0.220 0.192 0.192 0.187 0.254 0.263 Skilled agriculture 0.035 0.037 0.064 0.106 0.025 0.021 Craft and trade 0.169 0.202 0.145 0.151 0.140 0.141 Machine operators 0.122 0.135 0.118 0.112 0.093 0.099 Elementary occs. 0.090 0.067 0.103 0.069 0.081 0.080 Labor status, age 15-64 (Share of individuals in households with at least one member) Employed 0.793 0.762 0.689 0.671 0.732 0.727 Retired 0.497 0.513 0.515 0.527 0.492 0.470 Student 0.223 0.226 0.220 0.192 0.221 0.216 Disabled 0.036 0.016 0.052 0.045 0.030 0.016 Other 0.727 0.725 0.794 0.754 0.745 0.703 Industry, age 15-64 (Share of individuals in households with at least one member) Agriculture, mining, and fishing 0.052 0.047 0.112 0.130 0.041 0.040 Manufacturing 0.225 0.241 0.191 0.177 0.147 0.158 Services and Sales 0.684 0.605 0.532 0.469 0.655 0.624 25 Share of members with education in HH (age 15-64) Primary education 0.075 0.067 0.107 0.074 0.081 0.074 Lower secondary 0.184 0.195 0.263 0.252 0.162 0.149 Upper secondary 0.536 0.569 0.521 0.580 0.578 0.639 Tertiary education 0.206 0.170 0.110 0.093 0.179 0.139 Dwelling characteristics Square meters 90.711 87.120 92.523 95.296 83.187 85.564 Table A2: Alpha model Coeff. Std Err. 1 Retiree -0.2663** 0.1066 No service sector workers 0.3921*** 0.1407 1 working person -0.289** 0.147 2 working persons -0.2543** 0.1208 Constant -5.5976*** 0.1786 Adj. R2 0.0019 Observations 2,229 26 Figure A1: NUTS 3 Poverty estimates and 95% confidence intervals 1 .8 Head count poverty .6 .4 .2 0 0 200 400 600 Municipalities, Cities, and Districts of Zagreb Head count poverty 95% CI 27 Figure A2: Poverty in the districts of Zagreb 28 Table 3A: Poverty indicators by LAU-2 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Donji Grad 35,609 6.90 1.60 1.60 0.40 0.50 0.20 0.30 Gornji Grad-Medvešèak 29,750 5.50 1.80 1.20 0.40 0.40 0.20 0.20 Trnje 41,021 7.30 1.60 1.70 0.40 0.60 0.20 0.30 Maksimir 47,362 7.50 2.40 1.70 0.60 0.60 0.20 0.40 Pešæenica-Žitnjak 55,057 16.00 3.20 4.40 1.00 1.80 0.40 1.00 Novi Zagreb-istok 58,052 6.60 1.70 1.40 0.40 0.50 0.20 0.40 Novi Zagreb-zapad 56,647 10.40 2.30 2.50 0.60 0.90 0.30 0.70 TreÅ¡njevka-sjever 54,197 9.90 2.60 2.40 0.70 0.90 0.30 0.60 TreÅ¡njevka-jug 65,555 6.80 1.70 1.50 0.40 0.50 0.20 0.50 Èrnomerec 37,577 6.80 2.20 1.50 0.60 0.50 0.20 0.30 Gornja Dubrava 60,882 16.10 3.90 4.20 1.20 1.70 0.50 1.10 Donja Dubrava 35,871 16.30 3.50 4.30 1.10 1.80 0.50 0.70 Stenjevec 50,678 8.70 2.20 2.10 0.60 0.80 0.20 0.50 Podsused-Vrapèe 44,580 6.80 1.40 1.50 0.40 0.50 0.10 0.30 Podsljeme 18,858 4.90 1.50 1.10 0.40 0.40 0.10 0.10 Sesvete 68,924 12.70 6.80 3.30 2.00 1.30 0.90 1.00 Brezovica 11,720 12.30 3.80 2.90 1.10 1.10 0.40 0.20 Grad Zagreb 772,340 9.80 0.90 2.40 0.30 0.90 0.10 8.60 AndrijaÅ¡evci 4,020 37.50 8.90 11.10 3.20 4.80 1.60 0.20 Antunovac 3,610 21.30 7.80 5.70 2.50 2.30 1.10 0.10 Babina Greda 3,516 42.60 10.90 13.10 4.20 5.70 2.10 0.20 Bakar 8,211 16.00 4.80 4.00 1.40 1.50 0.60 0.10 Bale - Valle 1,125 13.80 4.80 3.30 1.30 1.20 0.50 0.00 Barban 2,688 10.70 5.80 2.50 1.70 0.90 0.70 0.00 Barilović 2,967 23.90 8.60 6.60 2.80 2.70 1.30 0.10 BaÅ¡ka 1,658 12.60 4.90 2.90 1.40 1.00 0.60 0.00 BaÅ¡ka Voda 2,773 21.60 6.30 5.70 1.90 2.20 0.80 0.10 Bebrina 3,185 40.30 10.70 12.40 4.30 5.50 2.20 0.10 BedekovÄ?ina 7,759 20.00 5.50 5.30 1.70 2.10 0.80 0.20 Bednja 3,954 31.60 7.30 9.30 2.70 4.00 1.30 0.10 Beli Manastir 9,459 32.50 6.40 10.50 2.60 4.80 1.40 0.30 Belica 3,150 12.30 5.10 2.90 1.30 1.00 0.50 0.00 Belišće 10,509 36.20 10.20 11.60 4.00 5.30 2.10 0.40 Benkovac 10,934 42.30 8.60 13.20 3.50 5.80 1.80 0.50 Berek 1,437 39.90 10.50 13.10 4.20 6.10 2.20 0.10 Beretinec 2,117 18.30 7.50 4.40 2.10 1.70 0.90 0.00 Bibinje 3,969 30.30 8.50 8.50 3.00 3.50 1.50 0.10 Bilje 5,590 23.00 6.40 6.50 2.10 2.70 1.00 0.10 Biograd Na Moru 5,501 17.00 6.30 4.30 1.90 1.60 0.80 0.10 Bizovac 4,456 23.00 7.00 6.00 2.20 2.40 1.00 0.10 Bjelovar 39,061 15.80 5.00 4.20 1.60 1.70 0.70 0.70 Blato 3,460 6.00 3.10 1.10 0.70 0.40 0.20 0.00 Bogdanovci 1,877 24.20 8.40 6.30 2.70 2.40 1.20 0.10 Bol 1,576 16.50 5.90 4.00 1.60 1.50 0.70 0.00 29 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Borovo 4,857 41.80 7.80 13.00 3.30 5.80 1.80 0.20 Bosiljevo 1,253 24.70 6.30 7.00 2.20 2.90 1.10 0.00 BoÅ¡njaci 3,748 43.00 9.90 14.20 4.40 6.50 2.40 0.20 Brckovljani 6,432 26.20 7.20 7.40 2.40 3.10 1.10 0.20 Brdovec 11,048 13.70 4.00 3.30 1.10 1.20 0.40 0.20 Brestovac 3,691 40.20 11.60 12.20 4.50 5.20 2.20 0.20 Breznica 2,188 27.70 9.40 7.60 3.10 3.10 1.40 0.10 Brinje 3,180 33.30 7.30 9.70 2.70 4.10 1.40 0.10 Brod Moravice 849 20.30 5.60 7.00 2.10 3.50 1.20 0.00 Brodski Stupnik 2,950 47.20 15.10 15.40 6.60 6.90 3.50 0.20 Brtonigla - Verteneglio 1,622 14.60 5.90 3.30 1.50 1.20 0.60 0.00 BudinÅ¡Ä?ina 2,390 36.10 10.70 10.50 3.90 4.40 1.90 0.10 Buje - Buie 5,102 10.70 4.40 2.50 1.20 0.90 0.50 0.10 Buzet 6,048 6.90 3.40 1.50 0.90 0.50 0.30 0.00 Cerna 4,489 37.30 8.00 11.20 3.10 4.80 1.50 0.20 Cernik 3,562 40.10 9.40 12.40 3.90 5.40 2.00 0.20 Cerovlje 1,650 12.20 5.50 2.70 1.30 1.00 0.50 0.00 Cestica 5,504 34.90 6.90 10.70 2.40 4.90 1.20 0.20 Cetingrad 1,921 32.10 11.00 9.40 4.20 3.90 2.10 0.10 Cista Provo 2,310 42.40 11.40 13.10 4.70 5.70 2.40 0.10 Civljane 226 64.00 13.30 22.50 7.00 10.60 4.00 0.00 Cres 2,777 10.70 4.60 2.40 1.20 0.80 0.50 0.00 Crikvenica 10,947 13.00 2.80 3.10 0.80 1.20 0.30 0.20 Crnac 1,445 41.80 8.80 12.80 3.70 5.50 1.90 0.10 ÄŒabar 3,748 4.70 3.70 0.90 0.90 0.30 0.30 0.00 ÄŒaÄ?inci 2,758 37.90 8.80 11.30 3.30 4.80 1.60 0.10 ÄŒaÄ‘avica 1,983 33.90 10.60 9.70 3.80 4.00 1.80 0.10 ÄŒaglin 2,363 46.30 9.80 15.20 4.40 6.90 2.40 0.10 ÄŒakovec 26,422 17.20 3.10 5.30 1.00 2.50 0.50 0.50 ÄŒavle 7,071 12.20 4.10 2.90 1.10 1.00 0.50 0.10 ÄŒazma 7,926 13.20 4.20 3.20 1.10 1.20 0.40 0.10 ÄŒeminac 2,780 27.40 6.80 7.30 2.20 2.90 1.00 0.10 ÄŒepin 11,299 19.50 6.50 5.10 2.00 2.00 0.90 0.30 Darda 6,746 45.50 8.40 16.00 3.70 7.80 2.10 0.30 Daruvar 11,482 10.80 3.40 2.50 0.90 0.90 0.30 0.10 Davor 2,967 33.70 10.20 9.60 3.70 3.90 1.80 0.10 Delnice 5,747 12.90 3.70 3.40 1.10 1.40 0.40 0.10 Desinić 2,604 26.40 9.30 7.00 2.90 2.80 1.30 0.10 Dežanovac 2,706 37.80 13.90 11.30 5.80 4.90 3.00 0.10 Dicmo 2,753 29.90 8.50 8.50 3.00 3.50 1.40 0.10 Dobrinj 2,051 14.00 5.30 3.20 1.50 1.10 0.60 0.00 DomaÅ¡inec 2,217 24.70 7.60 7.40 2.50 3.30 1.20 0.10 Brela 1,698 14.50 5.30 3.50 1.50 1.30 0.60 0.00 Donja Dubrava 1,895 17.60 6.20 4.30 1.80 1.60 0.80 0.00 Donja Stubica 5,375 15.00 5.10 3.70 1.40 1.40 0.60 0.10 Donja Voća 2,392 44.60 7.20 14.30 3.00 6.40 1.60 0.10 30 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Donji Andrijevci 3,666 32.30 7.70 9.50 2.90 4.00 1.40 0.10 Donji Kraljevec 4,527 12.90 4.80 3.00 1.30 1.10 0.50 0.10 Donji Kukuruzari 1,634 61.20 8.80 21.90 5.00 10.50 3.00 0.10 Donji Lapac 2,028 47.20 11.70 15.70 5.30 7.20 2.90 0.10 Martijanec 3,788 16.60 6.60 3.90 1.80 1.40 0.80 0.10 Donji Miholjac 9,275 29.30 5.70 8.20 1.90 3.40 0.90 0.30 Muć 3,838 25.50 7.10 6.60 2.30 2.50 1.00 0.10 Proložac 3,491 38.30 8.70 11.70 3.40 5.10 1.70 0.20 Donji Vidovec 1,378 21.10 6.00 6.10 1.90 2.60 0.90 0.00 Draganić 2,665 23.10 6.70 7.00 2.30 3.10 1.10 0.10 Draž 2,681 47.90 10.40 16.10 4.70 7.50 2.60 0.10 Drenovci 4,969 44.60 8.90 14.60 4.00 6.60 2.10 0.30 Drenje 2,592 51.60 10.80 17.30 4.90 8.00 2.70 0.20 DrniÅ¡ 7,422 22.80 6.20 5.90 2.10 2.30 0.90 0.20 Drnje 1,832 19.20 5.80 5.90 1.90 2.70 1.00 0.00 Dubrava 5,023 31.80 9.60 8.80 3.40 3.50 1.60 0.20 Dubrovnik 41,417 7.80 2.30 1.80 0.60 0.60 0.20 0.40 Duga Resa 11,120 19.00 7.00 4.90 2.30 1.90 1.00 0.20 Dugi Rat 6,982 26.00 7.10 7.10 2.30 2.80 1.00 0.20 Dugo Selo 17,201 16.80 4.90 4.30 1.50 1.70 0.60 0.30 Dvor 5,478 45.20 8.10 14.80 3.70 6.70 2.00 0.30 Ä?akovo 26,790 30.20 6.00 8.70 2.10 3.70 1.00 0.90 Ä?elekovec 1,490 18.70 5.40 4.90 1.70 1.90 0.80 0.00 Ä?ulovac 3,171 43.50 12.40 14.10 5.10 6.50 2.70 0.20 Ä?urÄ‘enovac 6,598 36.50 7.00 10.80 2.50 4.60 1.20 0.30 Ä?urÄ‘evac 8,090 23.90 5.30 7.70 1.90 3.60 1.00 0.20 Ä?urmanec 4,150 17.80 6.90 4.20 2.00 1.50 0.80 0.10 Erdut 7,108 48.30 11.70 16.00 5.20 7.30 2.80 0.40 Ernestinovo 2,064 14.40 6.00 3.30 1.60 1.10 0.60 0.00 Ervenik 1,098 62.80 11.00 22.70 6.00 10.80 3.50 0.10 FarkaÅ¡evac 1,889 30.90 11.30 9.40 4.10 4.20 2.00 0.10 Ferdinandovac 1,739 22.40 9.20 6.30 2.90 2.60 1.40 0.00 FeriÄ?anci 2,093 39.00 9.10 12.10 3.70 5.30 1.90 0.10 Fužine 1,570 10.40 4.20 2.30 1.10 0.80 0.40 0.00 GarÄ?in 4,729 41.70 10.30 13.30 4.10 5.90 2.10 0.20 GareÅ¡nica 10,258 26.70 5.70 7.90 2.00 3.40 1.00 0.30 Generalski Stol 2,586 23.90 7.10 6.10 2.10 2.40 0.90 0.10 Glina 8,757 28.10 6.30 8.10 2.20 3.40 1.10 0.30 Gola 2,389 22.90 6.80 6.00 2.00 2.40 0.90 0.10 GoriÄ?an 2,777 17.80 5.40 4.30 1.50 1.60 0.60 0.10 Gorjani 1,564 40.10 11.00 12.10 4.20 5.20 2.00 0.10 Gornja Stubica 5,258 23.30 6.70 6.00 2.00 2.30 0.90 0.10 Gornji Bogićevci 1,957 52.60 7.50 18.70 3.70 9.00 2.20 0.10 Gornji Kneginec 5,252 20.70 6.10 5.30 1.80 2.00 0.70 0.10 Gospić 12,320 14.10 3.60 3.50 1.00 1.30 0.40 0.20 GraÄ?ac 4,661 43.40 8.40 13.80 3.60 6.10 1.80 0.20 31 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. GraÄ?išće 1,416 11.50 4.70 2.60 1.20 0.90 0.50 0.00 Gradac 3,237 25.80 9.00 7.30 3.10 3.00 1.50 0.10 Gradec 3,601 25.70 7.80 7.10 2.60 2.90 1.20 0.10 Gradina 3,799 55.60 9.20 19.20 4.60 9.00 2.60 0.20 GradiÅ¡te 2,627 34.20 8.00 10.00 3.00 4.20 1.50 0.10 Grožnjan - Grisignana 733 19.10 5.40 4.60 1.60 1.70 0.70 0.00 GrubiÅ¡no Polje 6,383 19.40 4.20 5.30 1.30 2.10 0.60 0.10 Gundinci 2,013 58.50 11.40 20.50 5.80 9.70 3.30 0.10 Gunja 3,637 60.30 8.20 23.20 4.50 11.80 2.70 0.20 Hercegovac 2,378 15.90 6.20 4.00 1.80 1.50 0.80 0.00 Hlebine 1,271 23.20 6.90 6.60 2.30 2.90 1.10 0.00 Hrašćina 1,535 22.10 6.80 5.30 2.00 1.90 0.80 0.00 Hrvace 3,595 39.60 10.80 11.80 4.20 5.00 2.10 0.20 Hrvatska Dubica 2,070 47.60 8.10 15.60 3.60 7.00 2.00 0.10 Hrvatska Kostajnica 2,734 27.40 7.80 7.40 2.70 2.90 1.30 0.10 BrezniÄ?ki Hum 1,314 25.00 9.20 6.70 2.90 2.60 1.30 0.00 Hum Na Sutli 4,851 11.80 5.70 2.80 1.60 1.00 0.70 0.10 Hvar 4,218 12.10 4.00 2.80 1.00 1.00 0.40 0.10 Ilok 6,500 19.30 5.80 5.00 1.80 1.90 0.80 0.10 Imotski 10,671 39.20 9.20 12.70 3.80 5.70 2.00 0.50 Ivanec 13,447 16.90 3.20 4.20 0.90 1.60 0.40 0.30 Ivanić-Grad 14,292 20.60 4.40 5.60 1.40 2.30 0.60 0.30 Ivankovo 7,762 36.70 6.90 10.50 2.60 4.40 1.20 0.30 Ivanska 2,908 24.50 8.40 7.00 2.70 3.00 1.30 0.10 Jakovlje 3,813 15.00 5.40 3.60 1.50 1.30 0.60 0.10 JakÅ¡ić 3,986 26.70 7.50 7.50 2.60 3.10 1.20 0.10 Jalžabet 3,120 23.40 6.50 6.20 2.00 2.50 0.90 0.10 Jarmina 2,440 31.10 9.80 8.50 3.30 3.40 1.50 0.10 Jasenice 1,395 25.60 9.00 6.60 2.80 2.50 1.20 0.00 Jasenovac 1,987 34.40 10.10 10.00 3.70 4.10 1.80 0.10 Jastrebarsko 15,625 13.10 3.90 3.20 1.10 1.20 0.40 0.20 Jelenje 5,277 19.20 6.00 4.70 1.70 1.80 0.70 0.10 Jelsa 3,556 16.10 6.90 4.00 2.10 1.50 0.90 0.10 Josipdol 3,723 30.00 8.80 9.10 3.10 4.10 1.50 0.10 Kali 1,628 18.90 9.00 4.50 2.80 1.60 1.20 0.00 Kanfanar 1,541 8.10 3.60 1.80 0.90 0.60 0.40 0.00 Kapela 2,939 37.50 10.20 11.50 4.00 5.00 2.00 0.10 Kaptol 3,446 40.20 10.00 12.70 4.00 5.60 2.00 0.20 Karlobag 915 25.90 10.30 7.00 3.70 2.80 1.70 0.00 Karlovac 54,120 18.00 2.80 4.80 0.90 1.90 0.40 1.10 Kastav 10,346 9.20 3.40 2.10 0.90 0.70 0.30 0.10 KaÅ¡tela 38,044 20.30 5.20 5.20 1.60 2.00 0.70 0.90 Kijevo 415 24.40 8.40 5.90 2.50 2.10 1.00 0.00 Kistanje 3,429 74.80 8.60 32.50 6.40 17.80 4.40 0.30 Klakar 2,251 29.60 8.30 8.10 2.90 3.30 1.40 0.10 Klana 1,966 9.70 4.00 2.20 1.00 0.80 0.40 0.00 32 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Klanjec 2,911 8.90 4.00 2.00 1.00 0.70 0.40 0.00 Klenovnik 2,006 20.30 7.20 5.20 2.20 2.00 0.90 0.00 KlinÄ?a Sela 5,108 14.50 6.30 3.50 1.80 1.30 0.70 0.10 Klis 4,738 23.10 5.20 6.00 1.60 2.30 0.70 0.10 KloÅ¡tar Ivanić 5,990 27.50 7.70 7.70 2.70 3.20 1.30 0.20 KloÅ¡tar Podravski 3,200 41.00 8.30 15.40 3.70 8.00 2.10 0.10 Kneževi Vinogradi 4,517 41.50 9.10 13.30 3.80 6.00 2.00 0.20 Knin 15,011 42.70 7.70 14.00 3.40 6.30 1.80 0.70 Komiža 1,519 16.30 5.40 3.90 1.50 1.40 0.60 0.00 Konavle 8,549 10.40 4.60 2.40 1.20 0.90 0.50 0.10 KonÄ?anica 2,340 11.20 6.20 2.70 1.70 1.00 0.70 0.00 KonjÅ¡Ä?ina 3,658 18.60 8.00 4.80 2.50 1.80 1.10 0.10 Koprivnica 29,930 14.70 2.30 3.80 0.70 1.50 0.30 0.50 KoprivniÄ?ki Bregi 2,270 20.50 4.90 5.20 1.50 2.00 0.70 0.10 KoprivniÄ?ki Ivanec 1,972 19.70 7.60 5.00 2.30 1.90 1.00 0.00 KorÄ?ula 5,585 12.70 5.70 2.90 1.60 1.10 0.60 0.10 KoÅ¡ka 3,889 34.80 8.40 10.30 3.20 4.40 1.60 0.20 Kotoriba 3,080 25.80 5.70 9.40 2.20 4.80 1.30 0.10 Kraljevec Na Sutli 1,727 10.30 4.20 2.10 1.00 0.70 0.40 0.00 Kraljevica 4,490 11.50 3.90 2.60 1.00 0.90 0.40 0.10 Krapina 12,105 13.00 3.90 3.10 1.00 1.20 0.40 0.20 Krapinske Toplice 5,249 14.00 5.60 3.50 1.60 1.30 0.70 0.10 Križ 6,794 26.90 6.20 7.30 2.00 2.90 0.90 0.20 Križevci 20,631 15.10 4.60 3.70 1.30 1.40 0.60 0.40 Krk 5,951 10.50 5.20 2.30 1.30 0.80 0.50 0.10 Krnjak 1,826 48.20 10.50 16.20 4.80 7.50 2.70 0.10 KrÅ¡an 2,913 15.90 5.40 4.00 1.60 1.50 0.70 0.10 Kula Norinska 1,608 37.70 9.60 11.60 3.80 5.10 2.00 0.10 Kutina 22,337 19.70 4.00 5.50 1.30 2.30 0.60 0.50 Kutjevo 6,165 30.70 8.50 8.70 3.00 3.60 1.40 0.20 Labin 11,497 6.70 3.10 1.40 0.80 0.50 0.30 0.10 Lanišće 328 17.80 6.90 4.00 2.00 1.40 0.90 0.00 Lasinja 1,612 15.00 6.60 3.80 1.90 1.50 0.80 0.00 Lastovo 792 16.50 7.20 4.00 2.10 1.50 0.90 0.00 Legrad 2,185 11.80 4.60 3.00 1.30 1.10 0.50 0.00 Lekenik 5,885 22.90 6.20 6.10 1.90 2.50 0.90 0.20 Lepoglava 7,437 22.70 6.40 6.10 2.10 2.40 1.00 0.20 Levanjska VaroÅ¡ 1,016 60.50 9.50 23.40 5.60 11.90 3.60 0.10 Lipik 6,002 22.50 6.40 6.10 2.10 2.40 0.90 0.20 Lipovljani 3,450 17.50 6.30 4.30 1.80 1.60 0.80 0.10 LiÅ¡ane OstroviÄ?ke 686 32.30 10.00 9.70 3.90 4.20 2.00 0.00 Ližnjan - Lisignano 3,806 14.10 4.60 3.40 1.30 1.30 0.50 0.10 Lobor 2,818 25.50 6.10 6.60 1.90 2.50 0.80 0.10 Lokve 1,004 15.60 5.40 3.60 1.50 1.30 0.60 0.00 Lovas 1,207 15.70 7.50 3.80 2.10 1.40 0.90 0.00 Lovinac 995 13.20 6.30 3.30 1.80 1.30 0.80 0.00 33 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Lovran 4,033 9.50 3.80 2.20 1.00 0.80 0.40 0.00 Lovreć 1,691 35.10 9.80 10.50 3.80 4.50 1.90 0.10 Ludbreg 8,223 10.70 4.20 2.60 1.10 1.00 0.50 0.10 LukaÄ? 3,568 41.30 6.90 12.80 2.70 5.60 1.40 0.20 Lupoglav 918 13.70 6.20 3.10 1.60 1.10 0.60 0.00 Ljubešćica 1,837 21.80 6.20 5.60 1.90 2.20 0.80 0.00 MaÄ?e 2,511 30.60 8.00 8.20 2.80 3.30 1.30 0.10 Makarska 13,684 11.60 3.40 2.80 1.00 1.10 0.40 0.20 Mala Subotica 5,274 24.80 4.60 9.40 1.80 5.00 1.10 0.10 Mali Bukovec 2,185 21.40 7.10 5.80 2.20 2.40 1.00 0.10 Mali LoÅ¡inj 7,916 14.70 4.50 3.40 1.20 1.20 0.50 0.10 Malinska-DubaÅ¡nica 3,050 13.40 5.20 3.10 1.40 1.10 0.60 0.00 MarÄ?ana 4,199 13.70 4.00 3.30 1.10 1.20 0.50 0.10 Marija Bistrica 5,889 18.30 4.80 4.60 1.40 1.70 0.60 0.10 Marijanci 2,358 28.60 8.10 7.50 2.50 2.90 1.10 0.10 Marina 4,496 24.00 5.90 6.20 1.90 2.40 0.80 0.10 Martinska Ves 3,393 26.30 7.50 7.10 2.50 2.80 1.10 0.10 MaruÅ¡evec 6,275 15.00 4.30 3.70 1.10 1.40 0.50 0.10 Matulji 11,121 11.10 4.10 2.60 1.10 1.00 0.50 0.10 Medulin 6,374 6.20 3.20 1.40 0.80 0.50 0.30 0.00 Metković 15,956 29.00 7.20 8.40 2.50 3.50 1.20 0.50 Mihovljan 1,921 35.00 8.10 10.10 3.00 4.20 1.40 0.10 MikleuÅ¡ 1,449 47.60 10.30 15.40 4.60 6.90 2.50 0.10 Milna 1,022 14.50 6.30 3.40 1.80 1.20 0.70 0.00 Mljet 1,061 20.10 6.40 5.30 2.10 2.10 0.90 0.00 Molve 2,147 23.70 8.10 6.10 2.50 2.40 1.10 0.10 Podravska Moslavina 1,153 35.10 9.40 10.20 3.40 4.30 1.60 0.00 MošćeniÄ?ka Draga 1,526 10.10 4.30 2.30 1.10 0.80 0.40 0.00 Motovun - Montona 916 19.60 6.90 5.10 2.10 1.90 0.90 0.00 Mrkopalj 1,205 12.80 5.50 2.90 1.40 1.00 0.60 0.00 Mursko-Središće 6,209 24.90 7.00 7.90 2.40 3.70 1.20 0.20 NaÅ¡ice 15,912 24.30 5.80 7.00 1.90 3.00 0.90 0.40 Nedelišće 11,700 23.90 4.10 8.40 1.50 4.20 0.80 0.30 Nerežišća 845 13.80 5.80 3.00 1.50 1.00 0.50 0.00 Netretić 2,791 22.20 7.30 5.70 2.20 2.20 0.90 0.10 Nin 2,710 23.00 6.90 6.00 2.40 2.30 1.10 0.10 Nova Bukovica 1,769 50.50 9.70 17.00 4.50 7.80 2.50 0.10 Nova GradiÅ¡ka 13,880 26.70 6.10 7.90 2.10 3.40 1.00 0.40 Nova Kapela 4,108 35.20 9.70 10.00 3.50 4.00 1.70 0.20 Nova RaÄ?a 3,391 20.20 7.20 5.20 2.10 2.00 0.90 0.10 Novalja 3,613 16.20 5.30 3.80 1.40 1.40 0.60 0.10 Novi Marof 13,103 14.20 3.80 3.40 1.00 1.30 0.40 0.20 Novi Vinodolski 4,976 13.90 4.30 3.40 1.20 1.30 0.50 0.10 Novigrad - Cittanova 4,145 9.30 3.50 2.10 0.90 0.70 0.40 0.00 Novigrad Podravski 2,758 32.90 7.50 10.10 2.70 4.60 1.30 0.10 Novska 13,404 25.20 7.80 7.10 2.70 2.90 1.30 0.40 34 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. NuÅ¡tar 5,486 25.00 6.90 7.00 2.30 2.90 1.00 0.20 Nijemci 4,643 38.30 12.30 11.80 4.80 5.20 2.40 0.20 Obrovac 4,254 43.70 9.30 14.50 4.10 6.70 2.30 0.20 Ogulin 13,687 19.60 5.30 5.20 1.60 2.10 0.70 0.30 Promina 1,048 27.20 9.70 6.90 3.10 2.60 1.30 0.00 OkuÄ?ani 3,362 63.10 10.90 24.00 6.60 12.10 4.20 0.20 OmiÅ¡ 14,654 27.10 6.70 7.50 2.30 3.00 1.00 0.50 OmiÅ¡alj 2,973 14.00 4.90 3.70 1.50 1.50 0.70 0.00 Opatija 11,369 12.40 4.00 2.90 1.10 1.10 0.40 0.20 Oprisavci 2,481 24.70 7.30 6.50 2.20 2.60 1.00 0.10 Oprtalj - Portole 850 19.30 7.80 5.00 2.40 1.90 1.00 0.00 Opuzen 3,133 18.60 6.50 4.70 2.00 1.80 0.90 0.10 Orahovica 5,090 25.40 6.70 6.90 2.30 2.80 1.00 0.10 Orebić 4,031 9.00 5.00 2.00 1.30 0.70 0.50 0.00 Oriovac 5,719 33.50 7.80 9.80 2.90 4.20 1.40 0.20 Biskupija 1,688 56.70 11.40 18.90 5.60 8.50 3.10 0.10 Oroslavje 6,039 14.20 4.00 3.50 1.10 1.30 0.50 0.10 Osijek 105,841 18.30 3.20 4.90 1.00 1.90 0.40 2.20 OtoÄ?ac 9,516 17.30 4.00 4.50 1.20 1.80 0.50 0.20 Otok 5,401 41.70 11.50 12.90 4.70 5.70 2.40 0.30 Ozalj 6,537 27.00 10.40 7.40 3.30 3.00 1.50 0.20 Pag 3,802 11.30 4.60 2.50 1.20 0.90 0.40 0.00 PakoÅ¡tane 4,090 39.90 10.50 12.50 4.40 5.50 2.30 0.20 Pakrac 8,345 24.10 5.90 6.60 2.00 2.60 0.90 0.20 PaÅ¡man 2,069 29.00 9.60 7.80 3.30 3.10 1.50 0.10 Pazin 8,570 18.40 10.20 4.60 3.00 1.80 1.30 0.20 PeruÅ¡ić 2,636 25.00 8.30 7.00 2.80 2.90 1.30 0.10 Peteranec 2,648 29.50 6.70 10.10 2.50 5.00 1.30 0.10 Petlovac 2,350 45.70 9.00 14.60 3.90 6.50 2.00 0.10 Petrijanec 4,695 24.10 7.20 8.40 2.50 4.30 1.40 0.10 Petrijevci 2,761 30.20 8.30 8.50 2.80 3.50 1.30 0.10 Petrinja 23,896 19.00 4.50 5.10 1.50 2.00 0.70 0.50 Petrovsko 2,643 25.20 8.00 6.70 2.40 2.70 1.10 0.10 Pićan 1,805 12.60 5.40 2.80 1.40 0.90 0.50 0.00 Pisarovina 3,661 10.40 4.70 2.40 1.20 0.90 0.50 0.00 PitomaÄ?a 9,782 40.80 6.20 13.50 2.50 6.30 1.40 0.50 PlaÅ¡ki 2,057 52.40 10.20 17.10 4.80 7.70 2.60 0.10 Pleternica 11,115 28.70 8.10 8.00 2.90 3.20 1.30 0.40 PloÄ?e 9,776 21.00 6.20 5.50 2.00 2.10 0.90 0.20 Podbablje 4,679 35.30 6.70 10.90 2.60 4.80 1.30 0.20 Podcrkavlje 2,544 33.80 8.30 10.20 3.20 4.40 1.60 0.10 Podgora 2,505 25.10 6.70 6.80 2.20 2.70 1.00 0.10 PodgoraÄ? 2,834 53.80 9.10 19.40 4.20 9.70 2.40 0.20 Podstrana 8,932 11.40 3.40 2.80 0.90 1.10 0.40 0.10 Podturen 3,810 29.20 8.30 8.80 2.70 4.00 1.30 0.10 Pojezerje 896 38.00 11.70 10.90 4.40 4.50 2.10 0.00 35 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. PolaÄ?a 1,452 31.50 9.30 8.70 3.30 3.50 1.50 0.10 PoliÄ?nik 4,454 29.60 8.80 8.00 3.00 3.10 1.30 0.10 Popovac 2,044 43.00 9.50 14.00 4.30 6.30 2.30 0.10 PopovaÄ?a 11,394 25.70 6.00 7.70 2.10 3.40 1.00 0.30 PoreÄ? - Parenzo 16,438 11.50 3.50 2.80 1.00 1.00 0.40 0.20 Posedarje 3,565 32.50 8.70 9.20 3.10 3.80 1.40 0.10 Postira 1,542 11.80 4.40 2.70 1.20 1.00 0.50 0.00 Požega 25,406 18.80 3.80 4.90 1.20 1.90 0.50 0.50 Pregrada 6,485 24.70 6.50 6.30 2.00 2.40 0.80 0.20 Preko 3,339 17.40 5.90 4.10 1.70 1.50 0.70 0.10 Prelog 7,638 14.60 4.60 3.50 1.30 1.30 0.50 0.10 Preseka 1,413 11.80 5.50 2.50 1.30 0.80 0.50 0.00 PrimoÅ¡ten 2,794 18.40 5.80 4.40 1.70 1.60 0.70 0.10 PuÄ?išća 2,144 14.90 5.00 3.50 1.30 1.20 0.50 0.00 Pula - Pola 55,918 11.20 2.00 2.60 0.50 0.90 0.20 0.70 Punat 1,907 10.50 4.30 2.30 1.10 0.80 0.40 0.00 Punitovci 1,750 36.60 9.50 10.40 3.40 4.30 1.60 0.10 Pušća 2,615 13.40 5.30 3.30 1.50 1.30 0.60 0.00 Rab 7,942 15.20 6.10 3.60 1.70 1.30 0.70 0.10 Radoboj 3,339 25.30 6.00 6.60 1.80 2.50 0.80 0.10 Rakovica 2,368 23.00 8.20 6.10 2.60 2.30 1.20 0.10 Rasinja 3,171 40.50 7.00 13.10 2.80 6.00 1.40 0.10 RaÅ¡a 3,074 14.90 4.90 3.50 1.40 1.30 0.50 0.10 Ravna Gora 2,426 8.10 4.00 1.70 1.00 0.50 0.40 0.00 Ražanac 2,900 32.70 10.10 9.20 3.60 3.80 1.70 0.10 ReÅ¡etari 4,653 52.90 17.10 18.80 8.80 9.00 5.20 0.30 Rijeka 125,857 10.90 1.50 2.60 0.40 0.90 0.20 1.60 Rovinj 13,942 12.90 4.00 3.00 1.10 1.10 0.50 0.20 Rovišće 4,749 30.20 6.70 8.90 2.30 3.90 1.10 0.20 Rugvica 7,661 25.30 7.10 6.90 2.20 2.80 1.00 0.20 Ružić 1,559 22.60 8.40 5.60 2.60 2.10 1.10 0.00 Saborsko 626 33.60 12.70 10.10 4.80 4.30 2.40 0.00 Sali 1,672 14.00 5.90 3.00 1.60 1.00 0.60 0.00 Samobor 37,186 13.90 3.60 3.40 1.00 1.30 0.40 0.60 Satnica Ä?akovaÄ?ka 2,082 44.70 10.70 14.10 4.50 6.30 2.40 0.10 Seget 4,787 26.00 7.30 6.90 2.30 2.70 1.00 0.10 Selca 1,786 17.80 5.70 4.30 1.70 1.60 0.70 0.00 Selnica 2,885 26.10 6.10 6.90 2.00 2.70 0.90 0.10 Semeljci 4,219 44.20 9.80 15.20 4.20 7.30 2.30 0.20 Senj 7,095 13.50 3.70 3.20 1.00 1.10 0.40 0.10 Sibinj 6,815 35.90 9.20 10.60 3.60 4.50 1.80 0.30 Sinj 24,471 24.30 7.70 6.70 2.60 2.70 1.20 0.70 SiraÄ? 2,201 23.40 8.60 6.10 2.70 2.40 1.20 0.10 Sisak 46,762 17.00 3.70 4.50 1.20 1.80 0.50 0.90 Skrad 1,054 8.60 4.70 1.70 1.10 0.50 0.40 0.00 Skradin 3,701 25.00 7.30 6.70 2.40 2.60 1.10 0.10 36 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Slatina 13,529 25.90 5.30 7.40 1.80 3.10 0.90 0.40 Slavonski Brod 57,296 30.30 4.40 9.10 1.60 4.00 0.80 2.00 Slavonski Å amac 2,112 41.50 10.10 13.30 4.20 5.90 2.20 0.10 Slivno 1,906 22.80 7.50 6.00 2.20 2.40 1.00 0.00 Slunj 5,012 36.00 9.30 10.70 3.60 4.50 1.80 0.20 Smokvica 874 8.00 3.70 1.60 0.90 0.50 0.30 0.00 Sokolovac 3,346 34.00 9.00 10.10 3.40 4.30 1.70 0.10 Solin 23,670 12.00 4.00 2.90 1.10 1.10 0.40 0.30 Sopje 2,242 49.50 11.90 15.70 5.40 6.90 2.90 0.10 Split 173,163 13.40 1.80 3.30 0.50 1.20 0.20 2.60 SraÄ?inec 4,689 18.50 5.90 4.80 1.70 1.90 0.70 0.10 Stankovci 1,982 31.90 10.00 8.60 3.50 3.40 1.60 0.10 Stara GradiÅ¡ka 1,349 42.10 11.20 13.20 4.60 5.80 2.40 0.10 Stari Grad 2,744 15.60 6.20 3.70 1.80 1.30 0.70 0.00 Stari Jankovci 4,322 40.90 9.40 12.80 3.80 5.60 1.90 0.20 Stari Mikanovci 2,864 38.10 11.70 11.70 4.80 5.10 2.50 0.10 Starigrad 1,869 29.30 8.10 8.00 2.80 3.10 1.30 0.10 Staro Petrovo Selo 5,090 47.40 8.70 15.70 3.90 7.20 2.10 0.30 Ston 2,287 24.90 8.80 6.80 3.00 2.70 1.40 0.10 Strizivojna 2,494 42.00 7.90 12.90 3.00 5.60 1.50 0.10 StubiÄ?ke Toplice 2,736 14.10 5.20 3.50 1.50 1.30 0.60 0.00 Sućuraj 458 21.40 8.50 5.00 2.50 1.80 1.00 0.00 Suhopolje 6,477 36.00 10.50 11.50 4.30 5.10 2.20 0.30 SukoÅ¡an 4,533 31.80 7.90 8.80 2.80 3.60 1.30 0.20 Sunja 5,709 44.50 9.80 14.30 4.30 6.40 2.30 0.30 Supetar 3,997 12.60 5.00 2.90 1.40 1.10 0.50 0.10 Sveti Filip I Jakov 4,434 30.70 7.30 8.70 2.50 3.60 1.20 0.20 Sveti Ivan Zelina 15,623 19.90 4.90 5.10 1.50 2.00 0.70 0.40 Sveti Križ ZaÄ?retje 6,037 19.40 5.50 4.80 1.70 1.80 0.70 0.10 Sveti LovreÄ? 1,014 10.10 4.90 2.10 1.20 0.70 0.50 0.00 Sveta Nedelja 2,880 8.60 4.80 1.90 1.20 0.60 0.50 0.00 Sveti Petar U Å umi 1,052 8.10 4.40 1.60 1.00 0.50 0.40 0.00 SvetvinÄ?enat 2,184 13.20 5.40 3.40 1.60 1.30 0.70 0.00 Sveta Nedelja 17,785 11.00 5.00 2.60 1.30 0.90 0.50 0.20 Sveti Ä?urÄ‘ 3,763 27.20 7.90 7.80 2.60 3.40 1.20 0.10 Sveti Ilija 3,357 15.50 6.30 3.80 1.80 1.40 0.70 0.10 Sveti Ivan Žabno 5,086 21.20 7.30 5.30 2.10 2.00 0.90 0.10 Sveti Juraj Na Bregu 4,909 31.90 13.20 9.10 4.70 3.80 2.20 0.20 Sveti Martin Na Muri 2,586 21.40 5.00 5.50 1.50 2.10 0.60 0.10 Sveti Petar Orehovec 4,449 12.50 5.30 2.80 1.30 1.00 0.50 0.10 Å estanovac 1,849 38.70 10.50 11.50 4.10 4.80 2.00 0.10 Å ibenik 45,426 13.90 3.00 3.40 0.90 1.20 0.40 0.70 Å kabrnja 1,770 23.90 8.10 6.40 2.60 2.60 1.20 0.00 Å olta 1,668 20.40 7.60 5.00 2.30 1.80 0.90 0.00 Å piÅ¡ić Bukovica 4,171 41.90 8.60 13.20 3.50 5.90 1.80 0.20 Å tefanje 1,988 23.60 8.10 7.40 2.90 3.40 1.50 0.10 37 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Å trigova 2,526 24.90 6.80 6.70 2.10 2.70 1.00 0.10 Tinjan 1,660 11.30 4.90 2.60 1.30 0.90 0.50 0.00 Tisno 3,089 22.80 7.50 5.70 2.30 2.10 0.90 0.10 PlitviÄ?ka Jezera 4,299 15.40 5.20 3.70 1.50 1.40 0.60 0.10 Tompojevci 1,523 37.40 10.70 11.00 4.20 4.60 2.10 0.10 Topusko 2,956 23.70 7.40 6.70 2.60 2.70 1.20 0.10 Tordinci 2,004 33.50 10.30 9.60 3.70 4.00 1.70 0.10 Tovarnik 2,736 26.10 7.80 7.20 2.60 2.90 1.20 0.10 Trilj 8,801 42.30 8.40 13.00 3.40 5.60 1.70 0.40 Trnava 1,568 53.70 10.80 18.50 5.00 8.80 2.80 0.10 Trnovec BartoloveÄ?ki 6,470 11.70 4.10 2.70 1.10 0.90 0.40 0.10 Trogir 12,784 20.10 5.60 5.10 1.70 2.00 0.70 0.30 Trpinja 5,386 41.60 8.40 12.80 3.40 5.60 1.80 0.30 Tuhelj 1,973 18.20 5.50 4.40 1.60 1.70 0.60 0.00 Udbina 1,791 23.90 9.20 6.10 2.90 2.30 1.20 0.00 Umag 13,383 13.00 4.00 3.10 1.10 1.20 0.40 0.20 UneÅ¡ić 1,637 24.10 8.00 5.90 2.40 2.10 1.00 0.00 Valpovo 11,216 21.50 5.30 5.70 1.70 2.30 0.80 0.30 Varaždin 45,378 10.20 2.70 2.40 0.70 0.90 0.30 0.50 Varaždinske Toplice 6,316 17.30 6.20 4.30 1.80 1.60 0.80 0.10 Vela Luka 4,059 13.00 5.50 3.00 1.50 1.10 0.60 0.10 Velika 5,393 34.80 8.00 10.40 3.10 4.50 1.50 0.20 Velika Kopanica 3,258 47.90 10.50 15.40 4.60 6.90 2.40 0.20 Velika Ludina 2,614 27.00 8.00 7.80 2.70 3.30 1.30 0.10 Velika Pisanica 1,775 11.30 4.90 2.50 1.20 0.80 0.40 0.00 Veliki GrÄ‘evac 2,808 18.40 7.10 4.90 2.10 1.90 0.90 0.10 Veliko Trgovišće 4,856 26.90 8.70 7.20 2.80 2.80 1.30 0.10 Veliko Trojstvo 2,687 29.90 8.20 8.40 2.70 3.40 1.20 0.10 Vidovec 5,325 16.60 5.50 4.00 1.50 1.50 0.60 0.10 Viljevo 2,038 61.10 10.40 22.30 5.20 11.00 3.00 0.10 Vinica 3,336 15.90 5.50 3.90 1.60 1.50 0.70 0.10 Vinkovci 34,453 21.50 3.10 5.90 1.00 2.40 0.50 0.80 Vinodolska Općina 3,539 13.80 4.10 3.20 1.10 1.20 0.40 0.10 Vir 2,972 26.60 8.50 7.20 2.80 2.90 1.30 0.10 Virje 4,451 30.90 7.80 9.00 2.80 3.80 1.40 0.20 Virovitica 20,924 18.20 4.30 4.70 1.30 1.80 0.60 0.40 Vis 1,842 14.90 5.80 3.40 1.60 1.20 0.70 0.00 Visoko 1,498 35.30 7.90 9.40 2.70 3.60 1.30 0.10 ViÅ¡kovci 1,885 36.70 13.80 11.70 5.70 5.30 3.00 0.10 ViÅ¡kovo 14,235 12.20 3.80 2.90 1.00 1.10 0.40 0.20 ViÅ¡njan - Visignano 2,261 11.80 4.70 2.60 1.20 0.90 0.50 0.00 Vižinada - Visinada 1,146 10.80 4.80 2.40 1.20 0.80 0.50 0.00 Voćin 2,274 74.30 8.40 31.20 6.00 16.70 4.10 0.20 Vodice 8,784 24.60 4.90 6.50 1.60 2.50 0.70 0.20 Vodnjan - Dignano 5,943 23.90 7.10 6.70 2.30 2.80 1.10 0.20 Vojnić 4,524 57.20 9.40 20.50 4.90 9.90 2.90 0.30 38 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. VratiÅ¡inec 1,953 20.20 7.00 4.80 2.00 1.70 0.80 0.00 Vrbanja 3,815 34.40 8.70 9.80 3.10 4.00 1.50 0.10 Vrbje 2,162 60.70 9.50 22.10 5.00 10.80 2.90 0.10 Vrbnik 1,244 9.00 4.70 2.00 1.20 0.70 0.50 0.00 Vrbovec 14,406 22.40 5.40 6.00 1.70 2.40 0.80 0.40 Vrbovsko 5,025 17.60 5.60 4.50 1.70 1.70 0.70 0.10 Gvozd 2,889 42.10 9.80 12.80 4.20 5.50 2.10 0.10 Vrgorac 6,336 34.10 7.90 10.10 2.90 4.30 1.40 0.20 Vrhovine 1,378 57.50 10.10 20.30 5.20 9.60 3.00 0.10 Vrlika 1,968 15.80 5.60 3.90 1.60 1.40 0.70 0.00 Vrpolje 3,457 41.60 9.70 13.10 4.00 5.80 2.00 0.20 Vrsar - Orsera 2,152 9.80 4.30 2.20 1.10 0.80 0.40 0.00 Vuka 1,145 29.40 8.70 8.00 2.90 3.20 1.30 0.00 Vukovar 26,975 25.80 5.10 7.20 1.80 2.90 0.80 0.80 Zabok 8,938 12.60 5.00 3.10 1.40 1.10 0.60 0.10 Zadar 73,680 19.60 3.80 5.10 1.20 2.00 0.50 1.60 Zagorska Sela 990 12.50 7.10 2.80 1.80 0.90 0.70 0.00 Zagvozd 1,186 30.70 8.40 8.50 2.90 3.40 1.40 0.00 Zažablje 720 38.60 9.20 12.50 3.90 5.60 2.10 0.00 Zdenci 1,869 44.90 9.90 13.90 4.00 6.00 2.00 0.10 Zemunik Donji 1,885 19.80 6.90 5.00 2.10 1.90 0.90 0.00 Zlatar 6,014 20.10 5.00 5.20 1.50 2.00 0.70 0.10 Zlatar Bistrica 2,562 13.40 4.10 3.30 1.10 1.20 0.40 0.00 Zmijavci 2,038 29.10 8.40 8.00 2.80 3.20 1.30 0.10 Žakanje 1,856 13.10 4.90 3.10 1.30 1.10 0.50 0.00 Žminj 3,462 7.90 4.10 1.70 1.00 0.60 0.40 0.00 KraÅ¡ić 2,511 21.30 7.00 5.50 2.20 2.10 1.00 0.10 Županja 11,622 34.70 9.70 11.00 3.90 5.00 2.10 0.50 Otok 6,218 35.90 10.90 10.70 4.20 4.50 2.10 0.30 Rakovec 1,238 15.50 7.60 3.50 2.10 1.20 0.80 0.00 Novigrad 2,365 25.80 5.80 6.80 1.80 2.70 0.80 0.10 Kostrena 4,152 10.70 4.10 2.60 1.10 0.90 0.50 0.10 Marija Gorica 2,214 16.90 6.10 4.40 1.80 1.70 0.80 0.00 Žumberak 830 24.40 7.10 6.00 2.10 2.30 0.90 0.00 Velika Gorica 62,711 13.80 3.90 3.50 1.10 1.30 0.50 1.00 Orle 1,924 28.10 6.80 8.10 2.30 3.50 1.10 0.10 ZapreÅ¡ić 24,935 10.30 3.10 2.50 0.80 0.90 0.30 0.30 Pokupsko 2,210 40.50 8.90 12.60 3.50 5.60 1.80 0.10 Kravarsko 1,966 34.20 9.00 9.90 3.30 4.10 1.60 0.10 Bistra 6,389 15.30 6.50 3.70 1.80 1.40 0.80 0.10 Luka 1,323 20.10 6.70 5.10 2.00 2.00 0.90 0.00 Dubravica 1,425 18.80 6.50 4.80 2.00 1.90 0.90 0.00 Bedenica 1,424 17.70 7.70 4.30 2.30 1.60 1.00 0.00 Stupnik 3,652 12.10 5.20 3.00 1.50 1.20 0.60 0.10 Jesenje 1,512 21.50 7.90 5.40 2.40 2.00 1.00 0.00 Kumrovec 1,587 16.20 5.60 4.00 1.60 1.50 0.70 0.00 39 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Novi Golubovec 971 31.90 10.00 9.00 3.50 3.70 1.60 0.00 Majur 1,185 33.90 8.80 10.00 3.30 4.20 1.60 0.00 Ribnik 473 18.40 8.40 4.40 2.60 1.60 1.10 0.00 Tounj 1,143 38.80 9.80 11.60 3.90 5.00 2.00 0.10 Veliki Bukovec 1,411 22.60 8.30 6.10 2.60 2.50 1.20 0.00 Kalinovac 1,596 13.30 4.90 3.40 1.50 1.30 0.60 0.00 Kalnik 1,351 28.80 8.60 8.20 2.90 3.40 1.40 0.00 Novo Virje 1,169 18.40 7.60 4.30 2.10 1.60 0.80 0.00 Severin 873 21.20 8.70 5.40 2.60 2.10 1.10 0.00 Å androvac 1,742 14.40 5.00 3.70 1.50 1.50 0.70 0.00 Velika Trnovitica 1,356 27.50 8.30 7.90 2.90 3.30 1.40 0.00 Zrinski Topolovac 861 27.00 8.40 7.70 2.70 3.30 1.30 0.00 Bukovlje 3,018 34.80 7.60 10.50 2.80 4.50 1.30 0.10 Dragalić 1,340 30.30 9.60 8.90 3.50 3.80 1.70 0.00 Gornja Vrba 2,478 34.50 8.70 10.10 3.20 4.20 1.60 0.10 Sikirevci 2,461 41.60 11.30 12.30 4.30 5.20 2.10 0.10 Galovac 1,226 25.30 8.60 6.60 2.70 2.50 1.20 0.00 Kukljica 686 16.20 7.30 3.90 2.20 1.40 0.90 0.00 Povljana 756 17.00 7.00 4.10 2.00 1.50 0.80 0.00 Privlaka 2,211 25.10 8.70 6.70 2.70 2.60 1.20 0.10 Tkon 754 27.90 8.70 7.50 2.90 3.00 1.30 0.00 Donja MotiÄ?ina 1,637 42.70 11.90 12.90 5.00 5.50 2.50 0.10 Magadenovac 1,904 26.60 10.90 7.60 3.60 3.20 1.70 0.10 Vladislavci 1,836 40.20 9.50 11.90 3.50 5.00 1.70 0.10 Pirovac 1,850 26.60 7.40 7.00 2.50 2.70 1.10 0.10 Rogoznica 2,339 31.10 8.50 8.90 3.00 3.70 1.50 0.10 Privlaka 2,754 33.60 9.60 9.60 3.40 4.00 1.60 0.10 VoÄ‘inci 1,931 34.80 9.20 9.90 3.30 4.10 1.50 0.10 Dugopolje 3,439 24.80 8.60 6.30 2.60 2.40 1.10 0.10 Lećevica 577 34.10 9.70 9.80 3.60 4.00 1.80 0.00 LokviÄ?ići 783 50.80 8.80 16.30 4.20 7.20 2.30 0.00 Okrug 3,326 26.70 6.40 7.30 2.10 2.90 1.00 0.10 Prgomet 665 14.40 6.10 3.40 1.80 1.20 0.70 0.00 Primorski Dolac 769 19.30 7.30 4.80 2.10 1.70 0.90 0.00 Runovići 2,373 28.50 9.60 8.40 3.50 3.60 1.80 0.10 Sutivan 800 11.60 4.90 2.50 1.30 0.80 0.50 0.00 TuÄ?epi 1,925 20.20 7.00 5.40 2.30 2.10 1.00 0.00 Zadvarje 250 15.00 5.90 3.80 1.70 1.40 0.80 0.00 Karojba 1,427 12.90 4.60 2.90 1.20 1.00 0.50 0.00 KaÅ¡telir-Labinci - 1,463 17.30 6.80 4.30 2.00 1.60 0.90 0.00 Castelliere-S. Domenica DubrovaÄ?ko Primorje 2,081 11.30 4.50 2.70 1.20 1.00 0.50 0.00 Janjina 544 8.10 4.30 1.70 1.10 0.50 0.40 0.00 Lumbarda 1,211 11.40 5.70 2.60 1.40 0.90 0.60 0.00 Trpanj 705 13.20 6.50 3.00 1.70 1.00 0.70 0.00 Župa DubrovaÄ?ka 8,056 10.90 4.70 2.50 1.20 0.90 0.40 0.10 40 Std. Std. Std. Head Err. Poverty Err. Poverty Err. Share Location Population count Head Gap Poverty Gap Poverty of poor poverty count Sq. Gap Gap poverty Sq. Dekanovec 735 18.40 7.10 4.50 2.00 1.60 0.80 0.00 Gornji Mihaljevec 1,911 24.90 8.30 6.50 2.70 2.50 1.20 0.10 Orehovica 2,478 39.90 7.00 16.30 3.50 8.90 2.30 0.10 Strahoninec 2,653 10.30 4.70 2.30 1.20 0.80 0.50 0.00 Sveta Marija 2,284 11.20 4.60 2.40 1.20 0.80 0.40 0.00 Å enkovec 2,795 6.80 3.80 1.50 0.90 0.50 0.40 0.00 Jagodnjak 1,969 62.20 9.40 24.30 5.60 12.60 3.50 0.10 MarkuÅ¡ica 2,524 49.30 8.90 16.70 4.00 7.70 2.10 0.10 Negoslavci 1,370 40.20 11.20 12.30 4.30 5.30 2.20 0.10 Å odolovci 1,598 31.80 10.30 9.30 3.80 3.90 1.80 0.10 Podravske Sesvete 1,616 20.40 6.20 5.30 1.90 2.10 0.80 0.00 Murter - Kornati 2,040 20.80 6.80 5.20 2.10 1.90 0.90 0.00 Gornja Rijeka 1,753 22.40 7.80 5.40 2.20 2.00 0.90 0.00 Fažana - Fasana 3,491 11.50 4.10 2.70 1.10 1.00 0.40 0.00 Pribislavec 3,096 32.00 6.10 13.10 2.70 7.20 1.70 0.10 Bilice 2,255 18.20 6.90 4.70 2.10 1.80 0.90 0.00 Kolan 789 10.10 4.80 2.10 1.20 0.70 0.40 0.00 Kamanje 855 17.00 6.30 3.90 1.70 1.40 0.70 0.00 Lopar 1,233 22.70 7.60 6.00 2.40 2.30 1.10 0.00 Vrsi 2,036 26.10 8.40 6.60 2.60 2.50 1.10 0.10 Tribunj 1,534 19.00 7.00 4.50 2.00 1.60 0.80 0.00 Å titar 2,049 41.80 10.70 12.60 4.30 5.30 2.10 0.10 Funtana - Fontane 907 15.50 5.90 3.70 1.60 1.40 0.60 0.00 Tar-Vabriga - Torre-Abrega 1,982 9.10 3.60 2.20 0.90 0.80 0.40 0.00 41