Revista economică

Corneliu BOLBOCEAN, IEFS PhD student, lecturer, State University of , Department of Social Work

REGIONAL DISTRIBUTION OF IN MOLDOVA Although poverty is the most stringent problem in Moldova, the existing research had been done primarily by the WB and the UNDP. Measurement of poverty rates, , and socio-economic characteristics of poor people had been the major focuses of the existing research, however very little information exists on the distribution of poverty. This research explains how poverty is distributed in Moldova using the minimum of existence level as poverty estimate.

Poverty assessment me-thod- The minimum of existence ology. level, reported quarterly by the The research used consump- NBS, had been used as poverty tion approach to determine the dis- estimator. We did not find a spe- tribution of poverty in Moldova, cific definition of the minimum of since it has several advantages existence although we requested it comparing with the income ap- several times through NBS inter- proach1 : net based communication system. Consumption is a theoretically However, it could be concluded more satisfactory measure of from the official NBS statistical well-being. reports that the minimum subsist- Income is used in industrial ence level is the strict, minimum countries where self-employ- level of consumption, which in- ment is relatively rare so that cludes food expenditures as well most household income comes as non-food items. from a few sources, where an- Why is the minimum of exist- nual income variation is low, ence level a better poverty estima- and consumption data are rela- tor compared with poverty line tively costly to gather. estimates? Consumption is less variable 1.The World Bank estimate, i.e. over the period of a year, much absolute poverty line, is currently more stable than income in ag- severely scrutinized in the eco- ricultural economies and makes nomic literature for many reasons it more reasonable to extrapolate such as: the Bank uses an arbi- from two weeks to a year for a trary international poverty line survey household. that is not adequately anchored

27 Revista economică

in any specification of the real (HBS) for 2004 (December sub- requirements of human being. sample only), available on the Also, the poverty line employs NBS internet web page www. a concept of purchasing power statistica.md. HBS is a national “equivalence” that is neither survey of 6121 observations, well defined nor appropriate for conducted monthly by the NBS poverty assessment2 . in 45 primary sampling units, 2.The minimum of existence representative for the whole level is a better poverty estimator country. since it captures the full extent The research used prima- and complexity of poverty phe- rily parametric statistical methods nomenon in Moldova, because it such as: Paired t-tests; Chi-square takes into account both food and test, Mantel-Haenszel Chi-Square, non-food items a person requires and Likelihood Ratio Chi-Square; to consume in a given time pe- Contingency tables; Hypothesis riod. Moreover, the minimum of testing; Power analysis; Analysis existence is reported separately of variance between groups (one for urban and rural areas, it is also way ANOVA). However, the disaggregated by gender and age. non-parametric methods had been The World Bank and the Ministry used as well, specifically during of Economy poverty lines do not the analysis of variance between have these comparative advantag- groups (one way ANOVA) proce- es which mean that a significant dure: Wilcoxon, Kruskal-Wallis, bias is committed while research- and Van der Waerden One-Way ing poverty using them. Analysis tests. Expenditures per capita vari- Poverty within Moldavian able is the main variable of interest regions that had been used in research. Poverty is a major character- Thus, both measures, the mini- istic for both urban and rural areas mum of existence and consump- in Moldova, except Chisinau3 . tion expenditures per capita, are However an effective national expressed in money value and strategy aimed to fight poverty refer to the individual consump- would need to correctly allocate tion. Because of this, it is legiti- limited public resources according mate to use the consumption per to some well established priorities. capita variable and the minimum That is why to identify the most of existence level to assess pov- needful geographical areas that erty level. would be targeted by poverty The research used data from alleviation initiatives represents the Household Budget Survey one of the main purposes of any

28 Revista economică poverty research. rat; Comrat, Beshlama; Ştefan Moldavian territory had been Vodă, Feşteliţa. traditionally divided into three Observation: It appears that regions, specifically North, South, central region is overrepresented and Central region. HBS includes compared with northern and south- 46 primary sampling units how- ern regions. Thus, 24 psm within ever the 36th unit had not been the sample are drawn from central defined at all in the survey and region (53%), 14 psm drawn from did not appear in the dataset. Con- Northern region (30%), Southern sequently the 36th territorial unit had been omitted in the analysis. region represented by the remain- We grouped the territorial ing 8 psm (17%). However, the variable according to Northern, overrepresentation of the central Central and Southern region of region could be explained by the Moldova as follows, listing by fact that significant part of the cities, raions (counties) and vil- population lives in Chisinau and lages: its suburbia. (In 2004, almost 50% Northern region represented by of the total urban population lived the following primary sampling in Chisinau, while the over half of units: Briceni, Grimancauţi; urban population lived in other cit- Donduşeni; Drochia – Chet- ies. According to the NBS the total rosu and Sofia; Edineţ; Edineţ population in Moldova on January - Gloria and Bratuseni; Făleşti, 1, 2005 constituted 3386 thousand Calugar; Glodeni, Hоjdieni; people from which, 1308.8 or Ocniţa; Ocniţa - Ocniţa and Grinăuţ; Bălţi 2 psm;. 38.7% lived in urban areas, and Central region, represented by 2077.2 or 61.3% lived in rural the following psm: Anenii Noi; areas. The total urban population Călăraş, Hоrjauca and Volocineţ; constituted 1308.8 thousand and Hоncesti, Bujor and Boghiceni; 647.7 thousand live in the capital, Nisporeni, Ciuciuleşti; Orhei; 49.5% of the total4 ). Orhei, Cucuruzeni; Straşeni; To determine whether there is Straşeni - Vorniceni; Teleneşti, a difference between poverty level Negureni; Ungheni; Ungheni, in Northern, Central and Southern Pоirliţa; Chişinău units from 1 regions of Moldova, the analysis to 9; Ialoveni, Bardar. of variance (ANOVA) had been Southern region, included the performed. Since the dataset had following primary sampling a large sample size (510 observa- units: Cahul, Cahul - Rosu; tions), the central limit theorem Cantemir, Carpeşti; Căinari, Carbuna and Caşcalia; Com- justifies the use of normality as-

29 Revista economică

NOTE: Calculations performed in SAS software. Where: TERRIT_recoded 1 – Northern regions. TERRIT_recoded 2 – Central region. TERRIT_recoded 3 – Southern region. sumption performing parametric poverty rate must be different in ANOVA5 . at least two regions. Central region The insignificantly small obtained the highest consumption p-value of the F test, less than per capita level 643 lei per month, 0.0001 suggested that there is suf- the mean of consumption expen- ficient statistical evidence to state ditures per capita within Northern that - at least two means of ex- region was 532 lei per month, and penditures per capita within Mol- the smallest level of consumption davian regions are different, i.e. expenditures per capita were achieved within Southern region

NOTE: Calculations done in SAS software.

30 Revista economică

- 415 lei per month. - Chisinau compared with the Moreover, the non-parametric rest of the country, the Central techniques6 version of ANOVA region must be less affected by confirmed the validity of our con- poverty compared with Northern clusion (p-value less then 0.001). and Southern parts of the country, As we had seen, all three this would mean that the mean of consumption per capita in Central regional consumption expendi- regions would be significantly tures per capita means were less higher than the means of the rest than the minimum of existence of the country. Paired t-test had level, which for 2004 constituted revealed (p-value less than 0.003) 679.9 lei per month. However, that the a priori assumption was to find out if regional means are valid and that Central region statistically significant different indeed obtained a higher expen- from each other, we performed the ditures per capita level compared paired t-test. Testing the difference with the rest of the country. between means of paired samples However, we would also ex- is used when at least one of the pect that Northern region would following assumptions is not satis- have a higher mean of consump- fied: each sample is independent tion expenditures per capita com- of the other; both samples are from pared with Southern region, which normally distributed populations; could be explained due to the agri- the variances of both samples are cultural foundation and traditional equal.7 economic underdevelopment of The a priori assumption - due Southern part of Moldova. to comparatively higher living Below is the paired t-test standards in Moldavian capital output that confirmed our previous

NOTE: Calculations done in SAS software.

31 Revista economică assumption that tested the means erty rates and we conclude that of consumption expenditures poverty rate is higher in Southern per capita between Southern and part of Moldova, comparing with Northern regions of Moldova. Northern part. Indeed Southern part of At the same time, hypothesis Moldova obtained the smallest testing analysis revealed that all consumption expenditures per means of consumption expen- capita mean which would suggest ditures per capita by Northern, that on average people from this Central (observations from the part would achieve less consump- capital excluded), and Southern tion per capita than people from region are less than the minimum Northern and Central regions. of existence level, however the However consumption expen- situation was different in Molda- ditures per capita difference be- vian capital (p-values less than tween two regions is not very big. 0.0001). Thus at 5% level of significant this Conclusions and implica- difference is almost eighteen lei tions: per month, which is indeed a small The Southern part of Moldova difference between means in term is the poorest region in Moldova of consumption of expenditures compared with the Northern and per capita. This result would sug- Central regions. Although the gest that there should not be a very difference between means of con- high difference in terms of the sumption expenditures per capita consumption per capita standards within the Southern and Northern between Northern and Southern regions is small, the percentage regions, and thus, both regions of people who lived below the should not have a significant dif- minimum of existence in 2004 ference between poverty rates. within Southern part (90%) is sig- However, using the minimum nificantly higher compared with of existence for the whole country Northern region (76%). of 679.9, we found that in South- Moreover when the Central ern part of the country 90% of the part of Moldova is analyzed population achieved consumption without including the observa- expenditures per capita less than tions from Chisinau – the capital the minimum of existence, while of Moldova, the statistical tests 76% of the population from the proved that there is no statistical Northern region achieved con- significant difference between the sumption less than the minimum means of consumption expendi- of existence. 14% represents a tures per capita by regions, and visible difference between pov- that the means of consumption

32 Revista economică expenditures per capita by re- nance and Statistics, Moldavian gions are less than the minimum Academy of Sciences. of existence level for the whole 5. “Mathematical Statistics country of 679.9 lei per month, and Data Analysis”, Third Edition, however the situation was dif- John Rice, 485p. ferent in Chisinau, the capital of 6. www.statistica.md Moldova. References: Bibliography: 1 A. Deaton and S. Zaidi «Guidelines for Constructing Consumption Aggregates 1. “How not to count the for Analysis» 2002, Living poor”, Sanjay G. Reddy, Thomas Standards Measurement Study Working W. Pogge, www.worldbank.org. Paper: 135. v. 104, pp. xi, Washington, D.C.: The World Bank. 2. A. Deaton and S. Zaidi 2 “How not to count the poor”, Sanjay «Guidelines for Constructing G. Reddy, Thoman W. Pogge, www. Consumption Aggregates for worldbank.org 3 Corneliu Bolbocean «Poverty incidence Welfare Analysis» 2002, Living in rural and urban areas of Moldova”, Standards Measurement Study published in Economic Growth in Con- Working Paper: 135. v. 104, pp. ditions of Internationalization, Chisinau, September 2007, Institute of Econom- xi, Washington, D.C. The World ics, Finance and Statistics, Moldavian Bank. Academy of Sciences 3. “Nonparametric Statisti- 4 www.statistica.md cal Methods”, Myles Hollander, 5 “Mathematical Statistics and Data Analysis”, Third Edition, John Rice, Douglas A. Wolfe, Second Edi- 485p. tion, 1999. 6 “Nonparametric Statistical Methods”, 4. “Poverty incidence in ru- Myles Hollander, Douglas A. Wolfe, Second Edition, 1999. ral and urban areas of Moldova”, 7 Joseph G. Monks, Byron L Newton, by Corneliu Bolbocean September “Statistics for business” Second edi- 2007, Institute of Economics, Fi- tion. p.317

33