THE DEVELOPMENT AND STRUCTURE OF POVERTY IN MONTEVIDEO, URUGUAY, 1983 TO 1992

Transcription

1 The Developing Economies, XXXIV-2 (June 1996) THE DEVELOPMENT AND STRUCTURE OF POVERTY IN MONTEVIDEO, URUGUAY, 1983 TO 1992 MARISA BUCHELI BJÖRN GUSTAFSSON P I. INTRODUCTION OVERTY assessments are of central importance when forming policies worldwide. In many countries cross-section data makes it possible to say something about the structure of poverty. Repeated surveys, less often available, make it possible to say more. Using such data changes as well as reasons for the changes can be investigated. However, analyzing time-series data raises the question of how a poverty line is updated. In this paper we extract from micro data a time series on poverty using different assumption of how the poverty line is updated. The territory under study is Montevideo, the capital of Uruguay which is an upper-middle-income country. The period under study is , during which Uruguay went from dictatorship to democracy. The paper is organized as follows. In the next section we briefly describe the scene. Discussion of issues in conceptualizing and measuring poverty are found in Section III. Section IV reports on the development of poverty under two different assumptions of updating the poverty line. For one of the series we link the extent of poverty to the macroeconomic performance. In Section V we disaggregate to look at risks of poverty in different subgroups. There we also look at the development over time and its causes. A concluding section ends the paper. II. THE SCENE Uruguay is a country having an ethnically rather homogeneous population. The population size, 3 million inhabitants, is small compared to its neighbors Argentina and Brazil. The rate of population increase has also been comparatively low. This is mainly due to a comparatively low birth rate and during some years also a sizable emigration. Uruguay is rather urbanized; 75 per cent of the people live in locations with more than 5,000 inhabitants. Of the total population, 43 per cent live in the capital Montevideo.

2 POVERTY IN MONTEVIDEO 187 TABLE I ANNUAL MACROECONOMIC INDICATORS FOR URUGUAY, GDP per Capital GDP Growth Unemployment Rate a Inflation Rate (U.S.$) (%) (%) (%) , , , , , , , , , , , , , , Source: Banco Central del Uruguay. a Average yearly values. Uruguay was long known as an advanced country in terms of living standards and social programs. Programs for education and health expanded at the beginning of the century when also systems of old-age pensions became institutionalized. At that time the economy was growing faster than in many other Latin American countries. Since the middle of the 1950s the reverse holds. Democracy in Uruguay was overthrown by a dictatorship when the country was hard hit by first oil crisis in The economic policy of the new dictatorial government aimed at opening up Uruguay to the rest of the world and making credit and capital markets as well as labor markets more flexible. In line with this, all labor unions were banned. New severe economic crises took place between 1982 and 1984 when GDP fell by a total of 16 per cent which culminated in ending the dictatorship. Unemployment jumped from 6.7 per cent in 1981 to 15.3 per cent in 1983 (see Table I). The crises were the result of policy failures at least to some extent. Starting in October 1978 the dictatorship tried to stop inflation by pre-announcing the exchange rate for a period of six months. This meant that the national currency was devalued at a known rate. However, when Argentina, one of Uruguay s major trade partners, devalued its currency, Uruguay lost competitiveness and internal as well as external demand dropped rapidly. The exchange rate policy was abandoned in November 1982 when the peso was allowed to float. After twelve years of dictatorship in Uruguay, a new democratic government came into office in Parallel with this, labor unions began to organize and they were made legal. Economic growth during the first years of democracy (especially

3 188 THE DEVELOPING ECONOMIES during 1986 and 1987) was sizable. This is remarkable in a Latin American context because almost no other country on the continent experienced growth during those years. Remarkable also because the return to democracy in the mid-1980s in Argentina and Brazil were not followed by periods of growth. The last years of the 1980s were characterized by modest growth and rising inflation. A new democratic government committed to stopping inflation came into office in March At present other priorities are to deregulate the public sector and to push for economic integration with Argentina, Brazil, and Paraguay. Let us now briefly discuss the role of public transfers in Uruguay. A particular feature of the public sector in Uruguay is the complete lack of an income tax. Instead there is a rather heavy reliance on value-added taxes amounting presently to about 20 per cent; in addition wage taxes paid by the employer and social security contributions exist. Several observers seem to agree that evasion of social security contribution payments is of quantitative importance. Public transfers in Uruguay are mainly for old-age security. 1 Originally instituted in an environment of low inflation, benefits were defined in nominal terms. But in 1979 pension benefits were indexed to wages annually, and since 1989 the adjustment appears in the same month as public payments adjustment. As a rule pension rights are linked to work histories but there are exceptions. One can earn pension rights from being a widow, orphan, or being older than sixty-five. Among policymakers it is perceived as a problem that many people receive a pension without having a work history. Payments received in programs compensating for loss of income due to invalidity, sickness, and unemployment are of lesser importance. To become eligible for an unemployment benefit one must have a work history of at least six months in the formal sector, and such benefits are limited to an equally long period. This leads to only a narrow minority (perhaps not more than 5 per cent) of the unemployed being covered by unemployment compensation payments. There are programs of maternity benefits, housing benefits, and child allowances, but the total expenditures are small. The family seems to be an important institution for risk-sharing and for support in Uruguay. Sharing a household with persons from more than two generations is not unusual. It is common that children remain with their parents until they form a new family. Living alone is a rare (although increasing) phenomenon. III. CONCEPTUALIZING AND MEASURING POVERTY There is very little earlier work trying to quantify poverty in Uruguay. The main exception is Katzman [7] which is restricted to two years in the middle of the 1 See Davrieux [4], Centro de Investigaciones Económicas [3], or Lacurcia [8].

4 POVERTY IN MONTEVIDEO s. Several conceptual and methodological issues arise when trying to assess the development and structure of poverty. 2 A central issue is how to define the poverty line. In the empirical literature several approaches have been used. The alternative in this study is to set the poverty line at 50 per cent of mean disposable equivalent income. This type of approach has spread and is, for example, used in documents from the European Union when assessing poverty in the various member-countries. When calculating equivalent income, we have applied an often-used equivalent scale according to which the needs of the first adult is set equal to 1.0, of other adults equal to 0.7, and the needs of a child is set equal to 0.5. An analytic question distinguished from defining a poverty line at one point in time, is how to update the poverty line. Also here several approaches are possible. One possibility is letting only the change in consumer prices affect the poverty line. This is the approach taken for official estimates of poverty in the United States. A major criticism of this approach is that during periods of considerable economic growth, such a poverty line risks becoming less relevant for political decision making. Another possibility is to update the poverty line for each point of measurement. In this manner poverty becomes only an aspect of the lower tail of the distribution of income. A major criticism when applying a 50 per cent of mean poverty line is that the poverty line becomes much like a moving target. When trying to say something about poverty in Uruguay, we do not take sides on this important issue but work with both possibilities. That is in the relative series where the living standard of the poverty lines is continuously redefined. This is done by computing 50 per cent of mean at every quarterly point of observation and apply this as the poverty line. In the absolute series we start off from an income level at the beginning of the period under observation (50 per cent of mean for the entire year of 1983) and update it with the consumer price index. This means that the poverty lines are fairly close for the first year of observation. Given a poverty line, the extent of poverty can be expressed in various ways. We choose to use a family of indices suggested by Foster et al. [5]. For each poor unit this family of indices uses its normalized poverty gap which is a number indicating how far below the poverty line income falls on a scale bounded by 0 and (when no negative income is observed) 1. Those gaps are raised by a positive parameter (α, here set to 0, 1, 2) before the average is taken and then multiplied with the head count ratio. Higher parameter numbers give increasing weight to large poverty gaps, and thus indicate greater poverty aversion. Using all three indices gives a fuller view of poverty than if one were limited to a single alternative. If z is a family s poverty line and y i is a disposable income of ith poor family, where i = 1, 2,..., q (for whom y i < z), the class can be written as: 2 For fuller discussion, see, for example, Atkinson [1], Hagenaars [6], or Callan and Nolan [2].

5 190 THE DEVELOPING ECONOMIES TABLE II THE POVERTY LINE FOR THE RELATIVE SERIES: MONTHLY VALUES AVERAGED OVER YEARS, Year Poverty Line (Pesos) a Index , , , , , , , , , , a Constant pesos (1983 prices). ( q α 1 q α 1 y i g FGT(α) = 1 ) =, N ( i N i=1 z i=1 z ) where g i (= z y i ) is the poverty gap, and N the size of the entire population under study. Data comes from the Household Survey [9, various issues]. This data-gathering by means of interviewing was originally designed to monitor the labor market on a monthly basis. Since the beginning of the 1980s, questions on income have also been asked. The intention has been to ask each household member about his or her income. The sample of households has been drawn from a recent census and consists of about seven hundred households each month. A household is defined as a unit sharing food expenditures. Income consists of wages, earnings from self-employment, pensions, other transfers, interests, and dividends. Income can be obtained as money or in kind, but it does not include imputed rents of owner-occupied housing. One can assume that information on wages and pensions are of relatively high quality, while there might be more problems with the information for some other income components. To increase efficiency in the estimates, we aggregate surveys to cover quarters of a year. As inflation generally has been high, we use a series of monthly consumer price indices. And to ensure comparability between surveys, we start the period under study with the first quarter of 1983, although older surveys are available. In Table II we show annual averages for the poverty lines we are working with. As average income has increased considerably during the period under study, the

6 POVERTY IN MONTEVIDEO 191 choice of when to update the poverty line can obviously have large effects on estimates for the later part of the period. 3 IV. THE DEVELOPMENT AND REASONS FOR THE DEVELOPMENT The time series of poverty in Montevideo are documented in Appendix Table I. Figure 1 shows the absolute series using all three indices, and Figure 2 depicts the relative series. Direct comparisons of the two can be made in Figure 3 which reports estimated head count ratios. Several interesting conclusions follow from the results. Let us start with the question of whether the choice of poverty index affects assessment of how poverty has developed. The estimated indices for the absolute series show very high correlations, but for the relative series the correlations are smaller. 4 Fig ) Index (1983. I Evolution of Poverty Indices, I IV: FGT for Absolute Series I I I I I I I I I I Source: Appendix Table I. FGT(0) FGT(1) FGT(2) 3 Katzman [7] using a subsistence basket of goods estimated poverty for households in Montevideo in 1984 and 1986 using the same data. For those years his poverty line is lower than ours. 4 For the absolute series all correlation coefficients are equal to In the relative series FGT(0) and FGT(1) have a correlation coefficient of 0.85, and FGT(0) and FGT(2) a correlation coefficient of The correlation coefficient of FGT(1) and FGT(2) is 0.96.

7 192 THE DEVELOPING ECONOMIES Fig. 2. Evolution of Poverty Indices, I IV: FGT for Relative Series ) Index (1983. I I I I I I I I I I I Source: Appendix Table I. FGT(0) FGT(1) FGT(2) According to the absolute series the extent of poverty increased during the last years of the dictatorship and reached its highest value ever just when the democratic government came into power. Then it decreased very rapidly until Compared to this period more recent reductions in the extent of poverty have been small. While the changes in the absolute series seem to be driven by the development of average income, the relative series shows much smaller changes. Changes in the shape of income distribution at the lower tail have thus been relatively small during the period of observation. Actually the increases visible at the beginning of the period in the absolute series is not found in the relative series. When income decreased during the last year of dictatorship, the shape of the distribution (at the lower tail) remained rather unchanged. The decrease during the end of the period in the absolute series has a counterpart of increases in the relative series. Persons at the lower tail of the income distribution thus benefited less than others. Only for the first years of democracy (after 1985) do the two time series develop in the same manner. Then growth in average income went hand in hand with decreased inequality. To closer investigate how the absolute series is affected by macroeconomic

8 POVERTY IN MONTEVIDEO 193 Fig. 3. Poverty Rates [FGT(0)] for Montevideo Using Various Approaches to Update the Poverty Line, I IV ( ) I I I I I I I I I I Relative line Source: Appendix Table I. Absolute line development, we have run some regression analyses using the head count ratio as the dependent variable. In the first specifications we use as explanatory variables the lagged dependent variable, the log of GDP index, and alternatively the unemployment rate. In other specifications we add the rate of inflation during the quarter and alternatively the change in inflation since the last quarter. In Table III the model estimates are reported. 5 The coefficient of the lagged dependent variable appears stable across specifications. In interpreting the value slightly less than 0.8, one should remember that we are analyzing quarterly data. Thus while most of a temporary increase in the poverty rate during one quarter continues into the next, after one year most has disappeared. The estimated coefficients of the GDP index and the unemployment rate variable both have high t- values. The size of the coefficients implies that a permanent increase in the unemployment rate by 1 per cent of the labor force raises the poverty rate in steady state by almost 4 per cent. The corresponding effect of GDP is 0.8 per cent. Coefficients of the inflation variables have low t-values. 5 Reestimating the models for different sub-periods gave similar results.

9 194 THE DEVELOPING ECONOMIES TABLE III ESTIMATES OF MODELS EXPLAINING POVERTY IN THE TOTAL POPULATION OF MONTEVIDEO MEASURED BY THE HEAD COUNT RATIO: ABSOLUTE SERIES, I IV (a) (b) (c) (d) Intercept ( 3.524) (3.727) (3.347) (2.994) Lagged index (10.443) (10.377) (10.038) (10.46) Unemployment rate (4.064) Log GDP index ( 3.729) ( 3.436) ( 2.996) Inflation rate (0.774) Inflation variation (1.477) R 2 (adjusted) Steady state effect on the poverty index of an increase of: One point of the unemployment rate One per cent of GDP Note: Forty observations. Let us summarize the substantial findings. During the last year of the dictatorship in Uruguay, real income decreased pushing more people under an absolute poverty line. However, those worse off were not hit by decreased income more than other groups of the population. During the first years of democracy income grew, and particularly those who were worse off benefited from this. However, while income continued to grow after 1988, those worse off have benefited less than others. The proportion of people under a poverty line indicating a constant living standard is closely related to GDP, or alternatively to the unemployment rate, but not to measures of inflation. V. THE STRUCTURE OF POVERTY What is the composition of the poor in Montevideo? Has the structure of poverty changed? Out of various possible breakdowns of the population we present results for breakdowns according to age of the individual and according to education of household head. It is interesting to note that we can report little relation between sex of household head and poverty status in Montevideo as documented in Appendix Table II. A very low proportion of the total population living in a distinct single

10 POVERTY IN MONTEVIDEO 195 Fig. 4. The Extent of Poverty in Montevideo by Age, 1983, 1986, 1989, and 1992: Relative Series ] Index [FGT (0) for all population Age parent household contributes to this result. It might thus be mistaken to talk about feminization of poverty in Montevideo. 6 In Figure 4 we illustrate how the poverty indices ( relative series) vary with age of the individual relative to the index for the entire population. 7 The pattern seen in the figure is quite clear and stable in respect to variation in index used as well as of year of observation. Poverty is most extensive among children. 8 Children have poverty rates at least 50 per cent higher than for the population as a whole. The risk of poverty decreases with age for almost the rest of the course of life. There is a plateau in the risk for age-group 35 44, which is an age when children often are found in the household. Above that age poverty rates are smaller than for the population as a whole. Strikingly little extent of poverty is found among elderly persons. To some extent the low extent of poverty among the elderly can be traced back to 6 However, it should be remembered that we have assumed that income of household members are fully pooled, and in several cases single parents and their children are members of larger households. 7 These figures are based on averages for the last two quarters of each year under observation. 8 Generally in our data children are members of households having another head.

11 196 THE DEVELOPING ECONOMIES Fig. 5. Simulated Effects of Transfers on Poverty in Montevideo by Age, 1983, 1986, 1989, and 1992 ( ) Age Notes: Simulations are based on the relative series. the fact that some are still working and to the fact that aged persons are members of extended families. Nevertheless, a considerable proportion are brought out of poverty by transfers from the public sector. This is illustrated in Figure 5 which shows the result of an accounting exercise. We have simulated alternative poverty estimates under the assumption of all transfers being set equal to zero and everything else unchanged. These numbers were then compared with those of the real situation. A limitation with the exercise it that we do not take behavioral responses into account. According to Figure 5, during all years under study, transfers have taken 30 to 40 per cent of the aged out of poverty. The figure is highest for the last year under study, smallest for the first year. The effect of transfer payments is also important among persons years. Below 55 years of age the transfer-reducing effect does not vary much with age and is rather similar for all the years under study. Figure 6 shows that poverty status is very strongly linked to the educational status of the household head. Among households where the head has an education

12 POVERTY IN MONTEVIDEO 197 Fig. 6. The Extent of Poverty in Montevideo by Education of Household Head, 1983, 1986, 1989, and 1992: Relative Series 100] Index [FGT(0) for all population Years of schooling TABLE IV POVERTY RATES FOR PERSONS LIVING IN FAMILIES WITH CHILDREN: EDUCATION OF HOUSEHOLD AND NUMBER OF CHILDREN RELATIVE SERIES, 1983 AND 1992 Population of All Persons in Families with Children (%) The Head Count Ratio: FGT(0) Education of household head not more than nine years At least three children Not more than two children Education of household head at least ten years At least three children Not more than two children All families with children Note: Third and fourth quarter averages for 1983 and 1992.

13 198 THE DEVELOPING ECONOMIES TABLE V ESTIMATES OF MODELS EXPLAINING POVERTY BY AGE OF THE INDIVIDUAL IN MONTEVIDEO MEASURED BY THE HEAD COUNT RATIO: ABSOLUTE SERIES, I IV Age of the Individual Intercept (4.080) (4.569) (4.129) (2.714) (4.184) (2.741) (1.103) Lagged index (9.5875) (7.694) (7.494) (9.451) (6.031) (10.525) (9.891) Lag GDP ( 4.072) ( 4.553) ( 4.115) ( 2.710) ( 4.165) ( 2.741) ( 1.100) R 2 (adjusted) Steady state effect on poverty of an increase of 1 per cent of: GDP Age of the Individual Intercept ( 3.037) ( 3.082) ( 3.683) ( 2.907) ( 3.893) ( 3.552) ( 2.485) Lagged index (9.141) (6.420) (5.980) (9.543) (6.109) (11.940) (11.469) Unemployment rate (4.211) (3.943) (4.384) (3.376) (4.621) (3.908) (2.767) R 2 (adjusted) Steady state effect on poverty of an increase of 1 point of the: Unemployment rate of not more than five years, poverty is up to two times as great as for the population as a whole. On the other hand, the extent of poverty among those with the longest education is very small, less than one-fifth compared to the population as a whole. High risk of poverty among the less educated and among children means that one would expect to find high poverty rates in families with many children, headed by a person with a low education. This is also what is revealed when making crosscalculations, according to education of household head and number of children (see Table IV). There very high poverty rates, around 60 per cent, are reported for families where the head has a low education and there are at least three children. 9 9 Applying an equivalent scale in which needs of a child is set equal to 0.3 does not affect this conclusion.

14 POVERTY IN MONTEVIDEO 199 TABLE VI ESTIMATES OF MODELS EXPLAINING POVERTY BY EDUCATION OF HOUSEHOLD HEAD IN MONTEVIDEO MEASURED BY THE HEAD COUNT RATIO: ABSOLUTE SERIES, I IV Years of Schooling of Household Head Intercept (4.014) (3.703) (3.366) (3.379) Lagged index (7.923) (8.820) (4.223) (1.482) Log GDP ( 4.012) ( 3.696) ( 3.322) ( 3.277) R 2 (adjusted) Steady state effect on poverty of an increase of 1 per cent of: GDP Years of Schooling of Household Head Intercept ( 2.984) ( 2.886) ( 3.099) ( 1.317) Lagged index (9.128) (9.163) (3.809) (1.565) Unemployment rate (3.974) (3.705) (3.972) (2.681) R 2 (adjusted) Steady state effect on poverty of an increase of 1 point of the: Unemployment rate The other extreme is families where the head has a high education, and there are not more than two children. For such families the poverty rate is much smaller than for the population as a whole. How does macroeconomic performance affect poverty in various subgroups? In order to investigate this we have run regression models using the absolute series as the dependent variable and specifications as above. Table V shows the results for the different age-groups. There is some structure in the size of the estimates for the lagged dependent variable. The highest value is found among the aged. Also for the age-group the coefficient for the lagged variable is relatively high. Those estimates can be interpreted as poverty being more persistent for older persons than for others. On the other hand, coefficients for the macroeconomic variables are much lower among the elderly, and in the case of the GDP index has a low degree of statistical

15 200 THE DEVELOPING ECONOMIES significance. This is quite reasonable as many elderly depend on transfers which are not directly affected by changes in the macroeconomy. What can be said about patterns between age and the steady state effects of macroeconomic variables? In all of the specifications it appears that poverty in the category years is less affected. This is quite reasonable as such persons are closely attached to working life. On the other hand, poverty among children is the most affected by macroeconomic variables. The estimates for groups formed according to the education level of the household head are shown in Table VI. Here the pattern is clearer than the breakdown according to the age of individual. The coefficients for the lagged dependent variable are the lowest for the two groups having the highest education; and the coefficient for the group with the highest education is very low and has a low degree of statistical significance. Poverty among low-educated groups thus appears as more persistent than among those with a high education. Table VI also reports a very clear relation between education of household head and the steady state effect of macroeconomic variables. These results indicate that a recession has much higher effects on poverty rates among persons with the lowest education, while increases in poverty rates among highly educated are small. VI. CONCLUSIONS In this paper we have derived time series on the extent of poverty in Montevideo, Uruguay for a period starting in the first quarter of 1983 and ending in the last quarter of 1992 using different approaches for updating a poverty line and different poverty indices. In this application, which covers a period of relatively large changes in average income, the picture of how poverty develops is much affected by how the poverty line is updated. By comparison, the choice of poverty index is of lesser importance. What do the results say about the changed situation of those worse off in Montevideo from 1983 to 1992? During the last years of the dictatorship real income decreased generally which pushed more people under a poverty line representing a constant living standard. However, those worse off were not hit more than the average population. In the first years of democracy, income grew and it grew particularly among those worse off. But after 1988 little income growth has trickled down to those worse off. The proportion of the population under a poverty line indicating a constant living standard is closely related to GDP and alternatively the unemployment rate, but not to measures of inflation. Poverty in Montevideo has had a very pronounced age profile for the entire period under study. Young persons, especially children, are poverty prone. On the other hand, poverty rates for persons forty-five years of age and older are comparatively low. This is also true for the elderly. Transfers from the public sector help to

16 POVERTY IN MONTEVIDEO 201 keep the poverty rates low among persons fifty-five years of age and older. The extent of poverty among the elderly is relatively loosely connected to macroeconomic conditions which is in contrast to conditions of poverty among children. While it might be misleading to talk about feminization of poverty in Montevideo, there is a very clear relation between the education of the household head and poverty status. A high education means a low risk of becoming poor and the extent of poverty among highly educated is little affected by macroeconomic conditions. On the other hand, a low education means a high risk of poverty and macroeconomic conditions have large effects on the poor. A combination of a low education and many children leads to an extremely high risk of poverty. Such results indicate that policies targeted at households with many children and headed by a person with a low education are promising alternatives to alleviate poverty. REFERENCES 1. ATKINSON, A. Poverty and Social Security (Hemel Hempstead: Harvester Wheatsheaf, 1989). 2. CALLAN, T., and NOLAN, B. Concepts of Poverty and the Poverty Line: A Critical Survey of Approaches to Measuring Poverty, Journal of Economic Surveys, Vol. 5, No. 3 (1991). 3. Centro de Investigaciones Económicas. Introducción al Uruguay de los 90 (Montevideo: Centro de Investigaciones Económicas / Ediciones de la Banda Oriental, 1990). 4. DAVRIEUX, H. Papel de los gastos públicos en el Uruguay , Estudios CINVE 9 (Montevideo: Centro de Investigaciones Económicas / Ediciones de la Banda Oriental, 1987). 5. FOSTER, J.; GREER, J.; and THORBECKE, E. A Class of Decomposable Poverty Measures, Econometrica, Vol. 52, No. 3 (May 1984). 6. HAGENAARS, A. J. M. The Definition and Measurement of Poverty, in Economic Inequality and Poverty: International Perspectives, ed. L. Osberg (Armonk: M. E. Sharpe, 1991). 7. KATZMAN, R. The Heterogeneity of Poverty: The Case of Montevideo, CEPAL Review, No. 37 (April 1989). 8. LACURCIA, H. El sistema de pensiones en Uruguay, in Sistema de pensiones en América Latina: diagnóstico y alternativas de reforma, ed. A. Uthoff and R. Szalachman (Santiago, Chile: Economic Commission for Latin America and the Caribbean / United Nations Development Program, 1991). 9. Uruguay, Instituto National de Estadística. Encuesta de hogares (Montevideo).

17 202 THE DEVELOPING ECONOMIES APPENDIX TABLE I ESTIMATES OF POVERTY IN MONTEVIDEO: ABSOLUTE AND RELATIVE SERIES, I IV Estimation Using a Relative Estimation Using an Absolute Poverty Line Poverty Line FGT(0) FGT(1) FGT(2) FGT(0) FGT(1) FGT(2) 1983 I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV

18 POVERTY IN MONTEVIDEO 203 APPENDIX TABLE II ESTIMATES OF POVERTY: VARIOUS BREAKDOWNS OF RELATIVE SERIES, 1983 AND % in Pop- % in Population FGT(0) FGT(1) FGT(2) ulation FGT(0) FGT(1) FGT(2) All Montevideo Years of schooling of the household head: Other Sex of the household head: Male Female Individual age: People who belong to households with at least one younger than 18: All Household head with 9 or less years of schooling and: One or two children Three children and more Household head with 10 or more years of schooling and: One or two children Three children and more Notes: 1. The size of the sample for 1983 is 15,522 persons; that for 1992 is 15,635 persons. 2. Third and fourth quarter averages for 1983 and 1992.