Volume 17
Issue 4
Economics
JOURNAL OF
POLISH
AGRICULTURAL
UNIVERSITIES
Available Online: http://www.ejpau.media.pl/volume17/issue4/art09.html
DETERMINANTS OF MONETARY POVERTY IN THE MAZOWIECKIE VOIVODSHIP
Iga Lisicka
Department of Econometrics and Statistic, Warsaw University of Life Sciences  SGGW, Poland
The aim of this paper is to determine the demographic and socioeconomic characteristics which influence the risk of monetary poverty of households from the Mazowieckie Voivodship. For this purpose, the binary logit model is applied. To identify poor households, the relative poverty line is used. Thus, each household is ascribed to one of two subsets: poor or nonpoor. The empirical analysis is based on the Household Budget Survey carried out by Central Statistical Office of Poland in 2010. In addition, a comparison of the risk of poverty of households from the Mazowieckie Voivodship to the situation of households from other Polish voivodships is made. The analysis does not include nonmonetary dimensions of poverty.
Key words: poverty, logit model, demographic, socioeconomic characteristics.
INTRODUCTION
Poverty is a great economic and social problem. It applies to both individual units (persons, households) and the whole community. In Poland, study of the phenomenon is carried out among others by the Institute of Labour and Social Welfare and the Central Statistical Office, which examine the scale of poverty and its forms. It needs to be stressed that the study should not be limited only to the determination of the measurement of poverty extent and depth. The important aspect is to determine what factors influence this phenomenon. Therefore, the article attempts to define the demographic and socioeconomic characteristics that influence poverty.
This article is limited to the monetary poverty analysis, using a traditional approach. Households are divided into two subsets – poor and nonpoor. For this purpose a relative poverty line is used. Due to the fact that dichotomous variable indicates a set of poor, the binary logit model is applied.
Pioneering work in the application of binary models in the analysis of poverty is the article concerning the situation in Canada made by Phipps [12].Other researches about the determinants of poverty by using binary choice models include: Seeth and al. [15], Elmelech, Lu [5], Coromaldi, Zoli [2]. It needs to be stressed that there is no reason to believe that the root causes of poverty are the same everywhere in the world. Poverty may be due to national, sectorspecific, community, household, or individual characteristics [8].
PURPOSE
The distribution of poverty in the voivodship is unequal and depends on the economic potential. This article investigates the poverty of households in the Mazowieckie Voivodship. Mazowieckie is regarded as the richest voivodship in Poland. The presence of a large agglomeration in the area may be the reason of this situation. Therefore, the article examines how household poverty is distributed depending on the size of the place of residence. Apart from that, it also examines other demographic and socioeconomic attributes which may influence the risk of monetary poverty.
DATA
The research analyzes the Polish household budgets survey (HBS) data carried out by the Central Statistical Office of Poland in 2010. This data covers information on households, such as incomes, expenditures and socioeconomic attributes. Household budgets survey is based on the representative method that allows for generalization of the results to the whole population of households within an error margin [9]. Due to the monthly rotation of households, a different group of households participates in the survey each month of the year. The number of households participating in the survey was 37 412, including 5388 in the Mazowieckie Voivodship. Households from the Mazowieckie Voivodship account for 14.4% of the whole sample.
METHODS
The article focuses on monetary poverty, usually represented by income or expenditures. The monthly rotation of households in the survey causes households with negative income. There are 176 such households and they account for 0.47% of the whole sample. Therefore, expenditures are considered as the indicator of poverty, as better approximating the “real income” <footnote_1>. Moreover, households have different demographic composition. For this reason, equivalence scales are used. They allow to measure the impact of the demographic structure of the unit on its needs represented by expenditures. It is relative. It means that concerned units’ equivalent expenditures are compared to the expenditures of the reference unit, which is usually a single adult or a couple of adults [4, 13, 17].
In this study the OECD70/50 scale (also called original of Oxford scale) is used. It is one of the most popular equivalence scales, applied in the study of poverty by Central Statistical Office of Poland [18]. OECD70/50 is created by assigning the value 1 to the first adult member in the household, 0.7 to each additional adult member and value 0.5 to each child. The equivalent income of a household is obtained by dividing the nominal income by the value of the equivalence scale.
To determine the monetary poverty indicator, absolute or relative poverty lines are used. Minimum subsistence is the absolute poverty line used most frequently. Its value is determined by the Institute of Labor and Social Welfare. The advantages of absolute poverty lines may include their relationship with the definition of poverty defined as the inability to obtain sufficient funds to live. The disadvantage is arbitrariness, because their values depend only on the opinion of the experts. The relative poverty line is calculated as a percentage of the average value of the measure of wealth. For example, 50% of the mean of equivalent expenditures or 60% of median equivalent income. The advantages of such lines include the ease of calculation, while the disadvantages – the fact that their values depend only on the distribution of income. In this article the relative poverty line is used. It accounts for 50% of the mean of equivalent expenditure in the whole sample. It means that a household is considered poor if its equivalent expenditures are lower than 50% of the mean of equivalent expenditure for all households from all Polish voivodships.
The aim of this paper is to determine demographic and socioeconomic characteristics which have an influence on the risk of monetary poverty of households from the Mazowieckie Voivodship. For this purpose, the following characteristics of households and a household head <footnote_2> are selected:
 the class of the place of residence,
 socioeconomic group,
 the number of children in the household,
 the occurrence of disabled persons in the household,
 the level of education of the household head,
 the gender of the household head.
In order to determine which of these attributes have a significant impact on the risk of poverty, an econometric model is used. The numerical values of these characteristics are converted to binary independent variables. The dependent variable is a variable that specifies whether the household belongs to a subset of the poor or not. Therefore, the dependent variable takes two values – 0 or 1. In such a case, binary choice models should be applied. One of the most commonly used models of this type are logit and probit. In this article the logit model is used.
The general form of the logit model is as follows [6]:
(1) 
where
i = 1, 2, ..., n,
j = 1, 2, ..., k,
x_{i}^{T}β = β_{0} + β_{1}x_{1i} + β_{2}x_{2i }+
...+ β_{k}x_{ki } ,
P_{i}^{} – probability,
F – distribution
function,
β_{i} – parameters
to be estimated,
x_{ji} – value of explanatory variable x_{j} for ith
household,
k – number of explanatory variables,
n – sample size.
The logit model is usually estimated by the maximum likelihood method. In order to determine the goodness of fit of this type of model to the empirical data, an indicator based on the maximum value of the loglikelihood function may be used. One of the most popular is McFadden’s likelihoodratio index. It is expressed as [7]:
(2) 
where:
ln_{UR} – maximized
likelihood for the model with all predictors,
ln_{R} – maximized
likelihood for the model without any predictor.
This measure of goodness can take values from the interval [0,1]. The model is better fitted the closer to 1 the value of the index is.
Another frequently used measure is the Count Rsquared. It transforms the predicted continuous probabilities into a binary variable on the same scale as the outcome variable (0 or 1), and then it assesses the predictions as correct or incorrect. The Count Rsquare is this correct count divided by the total count.
RESULTS
The value of poverty line is 685.77 zł. It is calculated on the basis of 50% of the mean of equivalent expenditure in the whole sample consisting of households of all Polish voivodships. The percentage distribution of poor households in different voivodships are shown in Table 1.
Table 1. Poverty rates in Poland’s voivodships in 2010 
of households 
of poor households 
of poor households 

Source: Own calculations
based on Household Budget Survey Data 
To estimate logit model, the Gretl program is used. Initially, the model consisted of 17 independent variables. Parameter for variable determining the gender of the household head is statistically insignificant at level α = 0.05. For this reason, this variable is eliminated from the model. The final form of the model is presented in Table 2.
Table 2. The results of the estimation of logit model 
Source: Own calculations
based on Household Budget Survey Data 
The meaning of variables in Table 2 is the following:
 PLACE_{1} (reference variable) equals 1 if the household lives in a city of over 500 thousand residents, i.e. in Warsaw, or 0 – otherwise,
 PLACE2 equals 1 if the household lives in a city of 100–499 thousand residents, e.g. in Radom or Płock, or 0 – otherwise,
 PLACE3 equals 1 if the household lives in a town of 20–99 thousand residents, 0 – otherwise,
 PLACE4 equals 1 if the household lives in the town below 20 thousand residents, 0 – otherwise,
 PLACE5 equals 1, if the household lives in rural area, 0 – otherwise,
 GROUP_{1} equals 1 for employees' household, 0 – otherwise,
 GROUP_{2} equals 1 for farmers, 0 – otherwise,
 GROUP_{3} equals 1 for selfemployed, 0 – otherwise,
 GROUP_{4} equals 1 for pensioners, 0 – otherwise,
 GROUP_{5} equals 1 for retirees, 0 – otherwise,
 GROUP_{6} (reference variable) equals 1 for the household maintained on unearned sources, 0 – otherwise,
 EDU_{1} equals 1 if the household head has lower secondary education or less, 0 – otherwise,
 EDU_{2} equals 1 if the household head has secondary education or less, 0 – otherwise,
 EDU_{3} (reference variable) equals 1 if the household head has higher level education, 0 – otherwise,
 CHILD_{1} (reference variable) equals 1 for household without children, 0 – otherwise,
 CHILD_{2} equals 1 for the household with 1 child, 0 – otherwise,
 CHILD_{3} equals 1 for the household with 2 children, 0 – otherwise,
 CHILD_{4} equals 1 for the household with 3 children, 0 – otherwise,
 CHILD_{5} equals 1 for the household with 4 or more children, 0 – otherwise,
 DIS equals 1 if there is at least one disabled person in the household, 0 – otherwise. Disabled people were considered as those with current disability certificate.
The value of McFadden’s likelihoodratio index is 0.175047. The number of correct predictions calculated on the basis of the Count Rsquared is 88.5%.
DISCUSSION
Considering the calculations in Table 1, the lowest value of the proportion of poor households is for the Mazowieckie Voivodship. It amounts to 11.6%. The highest values are noted for the Lubelskie and Podlaskie Voivodships. For those voivodships the value exceeds 27%, therefore it is more than twice higher than for the Mazowieckie Voivodeship.
The positive estimate of parameters corresponding to variables PLACE_{2}, PLACE_{3}, PLACE_{4} and PLACE_{5} indicate that the smaller the place of residence the greater the probability that the household is poor. Households living in towns below 20 thousand residents and in rural areas are in the worst situation, ceteris paribus. The probability of the risk of poverty is compared to households from Warsaw – a city of over 1,700 thousand residents.
The negative estimate of parameters occurs at variables GROUP_{1}, GROUP_{2}, GROUP_{3}, GROUP_{4} and GROUP_{5}. It means that compared to households maintained on unearned sources the situation of households from other socioeconomic groups is better. Comparing households of pensioners and retirees, households of the first of this socioeconomic group is on average in worse situation, ceteris paribus.
Considering the level of education of the household head it is noted that with the increase in the level of education the risk of poverty decreases. Poverty is on average higher in households whose reference person has no more than lower secondary education, ceteris paribus.
The positive estimate of parameters corresponding to variables CHILD_{2}, CHILD_{3}, CHILD_{4} and CHILD_{5} indicates that the more dependent children in the household the higher the risk of poverty, ceteris paribus.
The positive estimate of parameter DIS means that the risk of poverty is also affected by the presence of at least one disabled person in the household.
The gender of the household head does not have any significant influence on the risk of poverty.
In the studies carried out by Szulc [16], Dudek [3], Rusnak [14] and Panek [11] similar results are obtained. The level of education of the household head, class of place of residence and the number of children in the household are important factors. Examination of the situation of polish households in 1993, 1996 and 1999 made by Szulc shows that the presence of people with disabilities has significant influence on the monetary poverty risk. However, in the Okrasa’s research the gender of the household head also has significant influence [10]. But this result is not confirmed in the other surveys.
CONCLUSIONS
In the Mazowieckie Voivodship the lowest value of the percentage of poor households is noted. The highest values are found for the Lubelskie and Podlaskie Voivodships.
The study shows that selected demographic and socioeconomic characteristics have a statistically significant influence on the monetary poverty risk in the Mazowieckie Voivodship. The greatest differences observed were due to the level of education of the reference person in a household. Households managed by a person with higher level of education are in the best situation. On average, the risk of poverty decreases with an increase in the level of education. The place of residence also has a significant influence. The smaller the place of residence of the household, the greater the possibility of it being poor. Considering socioeconomic groups, households living on unearned sources, farmers and pensioners are in the worst situation. The risk of poverty is also influenced by the number of children in the household. The highest risk of poverty is for households with four and more dependent children. Furthermore, the presence of at least one disabled person in the household increases the risk of it being poor. However, the gender of the reference person has no significant influence.
The value of goodness of fit of the model is low. In this type of research the value of measure takes a similar level [3]. Furthermore, the number of cases of the correct prediction is at 88.5%. It would be appropriate to take into account other demographic and socioeconomic factors. Their inclusion in the model could improve the goodness of fit.
It needs to be stressed that the standard of living is affected by both the level of income and nonmonetary factors, and the two need to be taken into account in an analysis of poverty.
 Bywalec C., Rudnicki R., 2002. Konsumpcja [Consumption], PWE, Warszawa [in Polish].
 Coromaldi M., Zoli M., 2011. Deriving Multidimensional Poverty Indicators: Methodological Issues and an Empirical Analysis for Italy, Social Indicators Research, 107, 1.
 Dudek H., 2006. Determinants of Poverty in Polish Farmers’ Households – Binary Choice Models Approach, Wydawnictwo Akademii Rolniczej we Wrocławiu, Wrocław, 9, 1.
 Dudek H., 2011. Skale ekwiwalentności – estymacja na podstawie kompletnych modeli popytu [Equivalence Scales – Estimation on the Base of Complete Demand Systems], Warsaw University of Life Sciences, Warsaw [in Polish].
 Elmelech Y., Lu H., 2004. Race, Ethnicity, and the Gender Poverty Gap, Social Science Research, 33, 1.
 Greene W.H., 2012. Econometric Analysis, Prentice Hall, New Jersey.
 Hardin J.W., Hilbe J.M., 2007. Generalized Linear Models and Extensions, Texas.
 Haughton J., Khandke S.R., 2009. Handbook on Poverty and Inequality, The World Bank Press, Washington.
 Household Budget Surveys in 2010, 2011. CSO Social Surveys and Living Conditions Statistics Department, Warsaw.
 Okrasa W., 1999. Who Avoids and Who Escapes From Poverty During Transition? – Evidence From Polish Panel Data, 199396, Policy Research Working Papers Series, The World Bank.
 Panek T., 2013. Ubóstwo [w:] Diagnoza społeczna 2013 [Poverty [in:] Social Diagnosis 2013], Council for Social Monitoring, Warsaw [in Polish].
 Phipps S., 1991. Regional Differences in the Incidence of Poverty in Canada, Canadian Journal of Regional Science, 14, 1.
 Rusnak Z., 2007. Statystyczna analiza dobrobytu ekonomicznego gospodarstw domowych [Statistical Analysis of Welfare of Households], Wydawnictwo Akademii Ekonomicznej im. Oskara Langego we Wrocławiu, Wrocław [in Polish].
 Rusnak Z., 2012. Logistic Regression Model in Poverty Analyses, Econometrics, Publishing House of Wrocław University of Economics, Wrocław, 35, 1.
 Seeth H., Chachnov S. Surinov A., Von Braun J., 1998. Russian Poverty: Muddling Through Economic Transition with Garden Plots, World Development, 26, 9.
 Szulc A., 2006. Poverty in Poland During the 1990s: Are the Results Robust?, Review of Income and Wealth, 52, 3.
 Szulc A., 2007. Dochód i konsumpcja [w:] Statystyka społeczna [Income and Consumption [in:] Social Statistics], PWE, Warszawa, 131–161 [in Polish].
 Ubóstwo w Polsce w 2010 r. [Poverty in Poland in 2010], 2011. Departament Badań Społecznych i Warunków Życia GUS, Warszawa [in Polish].
FOOTNOTES
 The term "real income" corresponds to the concept of permanent income , implemented by Milton Friedman [1].
 The head of household is defined as the person who provides all or most of the financial resources needed to maintain the household. When such a person cannot be determined, the head of household is considered to be the person who manages most of these resources.
Accepted for print: 9.12.2014
Iga Lisicka
Department of Econometrics and Statistic, Warsaw University of Life Sciences  SGGW, Poland
ul. Nowoursynowska 166
02787 Warszawa
Poland
email: iga_lisicka@sggw.pl
Responses to this article, comments are invited and should be submitted within three months of the publication of the article. If accepted for publication, they will be published in the chapter headed 'Discussions' and hyperlinked to the article.