4 . 4.4 4 u. 4.4 .h4::.:.4:4.4 ., ‘ -:\,~ .4 4 .‘1 4V1“) ‘:}.‘4 u. \ ,..44 .14‘4.44( .4-,:- -44- ,, ..,.,4......,.14 :, .44.4-_._...4 4- ‘44 uw , 4- mm: ‘ . ”14:4”. 44,4” ~44 ‘ :44. $2”:- .' .._,.. §§~ 444. wr‘r- - . '1444- an: . ~14- -",_: zit—‘2' :1" 4 . . I ‘ ’ ‘ 1 ~ _ 4 . '2. ' .‘17‘. 34 _ 4 . 4 4x— 4 ‘ _ . 4 4 r-w'n’r 2:173: .4... 4"- . - a 4 "‘ r r: r" 441;'__‘1» I 4-414f-v‘l," k— |'4K'l' ['1 VI ‘ ,1 .441” 4.44 , . . 4‘ 4 4 :14- . .‘ula-vwr I ,... . .Hrrrl‘v 4 4'4 4 , 13,4.” 4 4.44 4:. Inn]: 4 4 ., . ., 4. lgrm-wu .4. v 7" '4 .v.< r." 7V ‘r ' N - . ~ ' ’ 4aéer;;E $1000 1819 Total 27874 CHAPTER III VARIABLES AND METHODOLOGY This chapter is concerned with variables that are used to measure development. It also addresses the methodology used to obtain measures of inequalities with the purpose of studying the relationship between income distribution and development. Measures of Development For this study certain variables were chosen to represent the level of development within the country. Each one of these variables is considered a measure of development. These variables are level of income, median years of education, percentage of population engaged in agricultural activities, population growth and land ownership. These variables were chosen based on the criteria that they are considered to be good indicators of the socio- economic conditions of the population. In addition, most of these variables have been used in one way or another in the literature and they are considered as appropriate variables used to measure development. Finally, they are available from the sources presented in the previous chapter. 21 22 Income In this study the level of family income is used as a measure of development. Although this is one of the most accessible variables used as a measure of development, there is debate among authors about what measure or measures of income is best utilized as a measure of development. Lecaillon et a1. (1984), for example, recommended that the following types of income data should be used: data on income for one country and one year collected either by households, economic active population or individuals. Berry and Soligo (1980) consider individual income as the natural unit. But, they also emphasized that because individuals live together in families, or households, most information on income is obtained at the household level. Therefore, in this study median family income will be used. Education The relationship between the level of development and education represents an important feature which has been emphasized throughout the years. The level of development a country has can be measured by the level of education of its population. As education spreads throughout a country there are more opportunities for the population to be more educated, a situation that will help them to find better jobs as well as learn new technology to improve all branches of the economy. In addition, a well educated population will 23 have more knowledge about health and other social aspects. Thus, a country where most of its population is educated will achieve higher levels of development than ea country whose population is not educated. Since education plays an important role in development, it is expected that the level of education will be also related to differences in inequalities of income. Several studies have shown that differences in income inequalities can be related to differences in the level of education. Ahluwalia (1976a), for example, used the literacy rate as a measure of the basic education level of the population, and secondary school enrollment rate as a measure of the degree of educational achievement. He found a positive relationship between per capita GNP and these two variables. His results showed that secondary school enrollment affects positively the income share of the middle groups (middle forty percent of income share), whereas literacy affects positively the lower income group in the population (lowest forty percent). He explained this situation by arguing that lower income groups, principally in developing countries, are excluded from secondary schooling. An important fact about education is also mentioned by Berry and Soligo (1980). They affirmed that the education students receive in the present will affect the distribution of income and therefore the labor force in the future. Although this factor plays an important. role in the 24 distribution of income, this lagged relationship is beyond the scope of this thesis. In a country like Panama education constitutes an important factor that needs to be examined. Therefore, in this study the percentage of literate and the percentage of children between 6 and 15 years old enrolled in school will be used as indicators of development. Specifically, literacy is used as a measure of development because it reveals how minimally educated the population within each district is. In addition, school enrollment is used because it reveals how important education is viewed to be since this factor is considered more of a characteristic in high developed areas than in lower developed ones. Population in Agricultural Activities The distribution of the population in agricultural activities is another variable commonly used as a measure of development. Since the discovery of agriculture, this activity has constituted one of the most important economic activities in human life. As other economic activities came to be performed, agricultural activities were also affected because the population started to move from this "traditional" activity to the more "modern" activities. Thus, in this case, as the level of development increases there will be less population engaged in agricultural activities. 25 Even though agricultural activities tend to decrease with development this activity can not totally disappear, since the base of the subsistence of the population is usually in the amount of food that is produced for consumption. An increase in development also implies an improvement in technology for the agricultural sector and an increase in food production. Therefore, the amount of population engaged in agricultural activities may decrease with development but also development will affect the way this economic activity is performed. In most developing countries, a decrease in the population engaged in agricultural activities will be an indicator of development. Thus, the relationship between this factor of development and income distribution inequalities is an important aspect that needs to be addressed. According to Sundrum (1990), the traditional or agricultural sector usually has a lower level of income than the modern sector. Whereas the modern sector is affected by market forces, in the traditional sector income distribution is more affected by social forces which tend to keep income differences within a narrow range. As a result of this situation, income distribution is likely to be less unequal in the traditional sector than in the modern sector. Sundrum observed. that the agricultural share of the labor force Changed inversely with per capita income in each decade, and also that this share declined. Sundrum expressed that: 26 An interesting feature is that the decline of this share [agricultural share] over time has followed a logistic pattern, being slow in countries at a low level of development, then becoming faster at the middle stages of development and again becoming slow in the most developed countries. (1990, 178) Sundrum also mentioned that these two sectors of production (the traditional and modern sector) in developing countries have a low labor productivity and therefore, lower levels of income than those found in developed countries. For this study the percentage of population employed in agricultural activities will be used as one of the independent variables used to test the relationship between income distribution inequalities and development. This variable was chosen because the amount of people engaged in agricultural activities is inversely related to development, a situation that necessitates looking at how income distribution inequality is affected by a decrease in agriculture activities. Population Growth Population growth is another variable mentioned in the literature related to development. The argument for the use of this variable is that populations have had a tendency to grow where there is a greater variety of economic activities that can be performed and where there are better jobs and more opportunities to improve the economic status of the population. In addition, the discovery of new technology and an improvement in the health conditions and other social 27 conditions help people to live longer and, more important to the present argument, to have more children with a resulting increase in the growth of population. But, as a country reaches higher levels of development and more new technology is discovered related to population controls, there is also a tendency to have less children. This is most often explained by the fact that families will strive for a higher standard of living. Since population growth is considered a measure of development, and development is related to the distribution of income in a country, this variable will be used to establish the relationship between development and inequalities of income. Using regression results, Ahluwalia asserted that, "Our estimated results unambiguously show that high growth rates of population [were] systematically associated with greater income inequality" (1976a, 325-326). He based his hypothesis on. two explanations suggested. by the literature. First, higher growth rates occur in low income strata. This situation leads to a slower growth of per-capita income for the poorer population as compared to the high income population, with a consequent increase of inequalities. Second, high growth rates may increase income inequalities. These high growth rates lead to an excess of labor supply in an area where the population is already engaged in a restricted number of low income jobs. In addition, higher growth rates will create more burdens on the head of households because there will be more individuals dependent 28 on the income that is brought into the household with a consequent diminishing of savings. For this study population growth was chosen as an independent variable based on the assumption that ‘high population growth will increase income inequalities and that those districts with high growth are usually the less developed ones. Land Ownership Land ownership constitutes an important resource for the development of a country. The way this land is used and distributed within the population can affect development either in a positive or negative way. For example, high concentration of land ownership in a few hands or the fragmentation of land in very small portions could provoke serious imbalances which will affect the economic and social structure of a country. In addition, the form of the tenancy of the land may be an indicator of development because as a country develops it is expected that people will try to legally own the land, instead of only renting or having the right of the land generally transmitted from generation to generation by simple possession. Thus, land ownership can be considered as a measure of development. The relationship between land ownership as a measure of development and inequalities of income distribution has been suggested by Lecaillon et a1. (1984) and Quan and Koo (1985) 29 among others. Lecaillon, et a1. (1984) discussed how the surplus of population in rural areas provokes a fragmentation of the land. They argued that, with time, this fragmentation would result in two different groups: landless peasants and big land owners. Peasants with very small pieces of land will be forced to sell their lands and usually land owners are the people who get them. Due to the fact that in agricultural areas accessibility to land also means accessibility to employment and income, Lecaillon et al. also divided the economically active population in those areas into two groups: farmers and agricultural laborers. He described farmers as people who, whether they are landowners or tenants, are responsible for a piece of land which they work all year around. On the other hand, agricultural laborers do not have or own the land, and they work according to the demand of labor. This means that they will be employed in agriculture only during certain periods of time. At other periods they may be unemployed. Therefore, their income depends on the days they work and these may be very few. This situation of higher concentration of land holdings will bring more income inequalities and land distribution problems. Because data for the concentration of land among the population was not available at the district level, the variable of agricultural land under private ownership will be used instead in an attempt to analyze the influence that this variable has on income inequalities. Thus, in this 30 study the percentage of agricultural land under private ownership will be used as a measure of development. Table 3 summarizes the definition of each of these independent variables. Measures of Inequalities As ‘mentioned. earlier, in order to develop the three indexes of inequalities used in this study a pueliminary step was necessary. This step was to construct a Lorenz Curve for each of the districts of Panama. The Lorenz Curve is defined as: A simple graphical method used for comparing a given DISTRIBUTION with a perfectly even one, with a view to establishing the degree of concentration or segregation shown by the distribution. (Small and Whiterick 1990, 137-138) Another similar but more general definition of the Lorenz curve was mentioned by Gastwirth: "The Lorenz curve plots the percentage of total income earned by various portions of the population when the population is ordered by the size of their incomes" (1971, 1037). To obtain a Lorenz Curve, for each district, it was necessary to plot the cumulative percentage of families on the y axis and the cumulative percentage of income on the x axis in an ascending order. This curve is compared to the 45° line which represents equality and is reached only if the distribution of income is perfectly even with the distribution of families. Thus, the nearer the curve is to 31 Table 3.--Independent variables used in this study Variables Definition Median family income Population employed in Agricultural Activities Literacy School enrollment Population growth Agricultural land under private ownership Median household income by district in 1980 Percentage of population employed in agriculture, cattle raising, hunting, fishing and wood extraction in 1980 Percentage of population able to read and write Percentage of the population between 6 and 15 years enrolled in elementary and junior high school Growth of population from 1970 to 1980 as a percentage of 1970 Percentage of agricultural land under private ownership 32 this equality line, the diagonal line, the less income inequalities there will be within a district. A Lorenz curve was obtained for each of the sixty six districts of the country. An example of the construction of a Lorenz curve can be developed from the data displayed in Table 4 for the district of San Miguelito. First, it was necessary to obtain the middle for each one of the range distribution. In Table 4.-- Income distribution data used to obtain the Gini coefficient, I820 and 1840 for the district of San Miguelito Range Middle Families Income Cumulative Cumulative % of Fami- % of lies income < $75 70 1519 106330 4.757 10.415 $75 - 99 87 528 45936 11.283 30.282 $100-124 112 1240 138880 19.954 45.043 $125-174 150 2712 406800 37.562 66.441 $175-249 212 4931 1045372 60.788 84.768 $250-399 324 6474 2097576 78.478 93.902 $400-599 499 4908 2449092 88.208 97.456 $600-799 699 2417 1689483 92.656 98.670 $800-999 899 1326 1192074 94.550 99.071 > $1000 1250 1819 22737500 100.000 100.000 addition, the number of families distributed within the different range of income also was obtained. In 1980 there were 1,519 families in the district of San Miguelito with an income of less than seventy five dollars. The number of families within each income range was multiplied by the 33 middle income value of each interval of income; this give us the total income per each category. For example, in the category of less than seventy five dollars there was 106,330 of income which is the result of the multiplication between the numbers of families within that category and the middle income. The mid-point of the "less than $75" range was arbitrarily set to $70. (Similarly the "over $1000" mid- point was set to $1250.) After that the percentage of families existent in each category was obtained throughout the division of families within each category and the total of families multiplied by 100. In the case of San Miguelito the percentage of families that earn less than $75.00 was obtained by dividing 1,519 by the total families in the district that was 27,874 in 1980 and multiplied by 100 giving us 5.450%. In other words 5.5% of families in the district of San Miguelito in 1980 earned less than $75.00. To obtain the percentage of income for each category the same procedure used for the percentage of families was performed. For example, 0.929% of the total income fell into the category of less than $75.00. Finally, a Lorenz curve for the district of San Miguelito was drawn using the cumulative percent of families and the cumulative percent of income. (See Figure 3). It is important to recall that such a curve is produced for each of the sixty-six districts. These Lorenz curves were then used to derive three measures of inequality: Gini Percent of Families Cumulative 34 100 10— Cumulative Percent of Income Figure 3 Lorenz Curve for the district of San Miguelito 35 coefficient, income share of the poorest 20% of the households (I820) and income share of the poorest 40% of the households (I840) for each one of the sixty six districts. The Gini Coefficient As introduced earlier, the Gini coefficient is a measure of inequality derived from the Lorenz curve. The Gini coefficient is defined as, "The ratio of the 'area' between the diagonal and the Lorenz curve divided by the total area of the half-square in which the curve lies" (Todaro 1985, 145). This measure goes from 0 (perfect equality) to 100 (perfect inequality). In other words, the greater the value is for the Gini coefficient, the more unequal and the higher is the concentration of a phenomenon. This measure in conjunction with the Lorenz curve is considered "...the most appropriate methods to measure and illustrate inequality" (Swindell 1989, 67). According to Todaro, countries with a highly unequal distribution of income will usually present Gini coefficients between 0.50 and 0.70; and countries with distributions that are less unequal will have distribution between 0.20 to 0.35. To obtain the Gini coefficient the following formula can be applied: 36 Where A corresponds to the area between the line of equality and the Lorenz curve; and B is the area below the Lorenz curve. The area B is measured with a series of polygons that are added. The area A+B is obtained by computation of the area of a right triangle. And therefore area A may be obtained by subtraction. For the example district of San Miguelito this Gini was 37.98. For this study all the Gini coefficients for each district of the country were obtained with this polygon summation using Lotus 123. The 66 Gini coefficients were plotted on a map with the purpose of showing the degree of inequalities in income that was present in 1980. (See Figure 4). A limitation of the Gini coefficient is that this measure does not show the distribution of inequalities for specific segments of the population; for example the poor population, which are usually the most affected by unequal distribution of income. In other words, differently shaped Lorenz curves can give rise to the same "areas" and therefore the same Gini coefficients. Therefore, the Gini coefficient. will be used as only one measure that shows inequalities that exists among the entire population of households within a district. Two other measures, the income share of the poorest 20% of the population (I820) and the income share of the poorest 40% of the population (1840), are used to focus upon specific portions of the household population. 37 O.Nm m.m¢ 8m. . SE55 E <2m <2 0 b2 < 0 b1 < 0 b2 > 0 income v Population em- ployed in agri- b1 > 0 b2 < 0 b1 < 0 b2 > 0 cultural acti— vities Literacy b1 > 0 b2 < 0 b1 < 0 b2 > 0 SChOOl enroll- b1 > 0 b2 < 0 131 < 0 b2 > 0 ment Rate of popu- b1 > 0 b2 < 0 b1 < 0 b2 > 0 lation growth Agricultural b1 > 0 b2 < 0 b1 < 0 b2 > 0 land privately owned Measure of Inequalities 100 80 60 40 20 O 47 F'— 1820 or 1840 GINI 1 I l I I .1 100 200 300 400 500 600 Least < > Most Developed Developed Figure 7 An example of a U-shaped and inverted U-shaped curves 48 For the percentage share of income received by the poorest 20% (1820) and 40% (1840) of households within each district, it is hypothesized that a U-shape curve will be found. These will be a U-shape instead of an inverted U- shape because high relative values of 1820 and 1840 represent lower levels of inequality; low values of 1820 and 1840 represent higher levels of inequality. For methodological purposes, in this study the variables that represent development were defined in a manner such that all of them will behave from "worse" to "better". That is, lower levels of development is seen as "worse" and higher levels of development is seen as "better". For example, low median family income is representative of low development, while high median family income is representative of high development. In the case of percentage of population employed in agricultural activities, a high percentage of population engaged in this sector is representative of low development and a low percentage of population in agricultural activities is seen as high development. Hence, for this variable the percentage of population which is NOT employed in agricultural activities will be used to have all the variables behaving from "worse" to "better". For the percentage of literacy a low percentage of literate population represents low development, and a high percentage of literate population represents high development. In the same way, for school enrollment a low 49 percentage of population that goes to school represents low development, and a high percentage of population enrolled in school represents high development. On the other hand, a high rate of population growth is representative of low development and a low rate is representative of high development. Finally, for agricultural land under private ownership, a low percentage of land under private ownership is representative of low development, while a high percentage is representative of high development. Although the quadratic equation is used, it is necessary to mention that a statistical problem for the regression equations in this study has been the relatively low tolerance values presented by the variables. This indicates that there is a degree of correlation between the independent variables. This "tolerance" problem may have been caused by the fact that, in order to use the quadratic equation, the original variables were squared. It is not unusual for independent variables differing only in their exponents, to exhibit some collinearly. These equations show this problem, but to obtain the curvilinear relationship it was necessary to use them. I am assuming the degree of collinearly is not a problem. The next chapter will discuss the results. CHAPTER IV ANALYSIS AND RESULTS Multiple regression analyses between each index of inequality and each index of development were obtained. All hypotheses were tested at 95% confidence level using a one tail test. The results of these statistical analyses are addressed below. General Results Table 6 shows the results of the regression analyses. Overall, eighteen sub-hypotheses were tested. Twelve of theSe were significant and six were not. Within the twelve significant sub-hypotheses six of them were significant for the two terms (coefficients b1 and b2), thus the null hypothesis was completely rejected for this group; one sub- hypothesis had only coefficient b2 as significant, thus, the null hypothesis was partially rejected. In addition, five sub-hypotheses were significant but with opposite signs to what was hypothesized, thus, the null hypothesis was accepted. It is important to mention at this point that the entire set of 66 districts was used for the regression analyses. However, from this. entire set eighteen districts 50 51 Table 6.-- Expected and actual signs of the relationships Independent Expected Actual variables Gini 1820 I840 Gini I820 I840 Median fami- b1>0 b1<0 b1<0 b1>0+ b1<0+ b1<0+ ly income b2<0 b2>0 b2>0 b2<0+ b2>0+ b2>0+ Agricultural b1>0 b1<0 b1<0 b1>0+ b1<0 b1<0+ employment b2<0 b2>0 b2>0 b2<0+ b2>0@ b2>0+ Literacy b1>0 b1<0 b1<0 b1>0* b1>0& b1>0& b2<0 b2>0 b2>0 b2<0* b2<0& b2<0& SChOOl b1>0 b1<0 b1<0 b1>0* b1>0& b1>0& enrollment b2<0 b2>0 b2>0 b2<0* b2<0& b2<0& Population b1>0 b1<0 b1<0 b1<0& b1<0+ b1<0* growth b2<0 b2>0 b2>0 b2>0& b2>0+ b2>0* Agricultural b1>0 b1<0 b1<0 b1>0* b1<0* b1<0* land under b2<0 b2>0 b2>0 b2<0* b2>0* b2>0* private pro- perty One tail test + Significant & Significant @ Significant * Not significant at 95% confidential level. with expected signs with opposite to expected signs with only coefficient b2 with expected sign 52 were chosen to serve as examples for conclusions derived from the functions. The reason why all 66 districts are not shown in the following graphs is that the sheer number of districts (66) makes the graphs difficult to read. It is not possible to discriminate the position of the individual districts in the different graphs. Therefore, eighteen districts were chosen because they seem to be the most representative of the relationship between levels of development. and income inequalities. These: districts are listed in Table 7. Analysis of the Results As was mentioned above 'the results of this study presented two different groups: a group of significant relationships where the null hypotheses were rejected and another group of relationships where the null hypotheses were accepted. Sub-Hypotheses with Expected Signs The results obtained from the regression analysis showed support for the inverted U—shaped and the U-shaped curve for the whole country. The relationship between the indexes of development and the indexes of income inequalities (Gini coefficient,°1820 and 1840) that support the hypotheses for both coefficients b1 and b2 using these three measurements of inequalities were: median family 53 Table 7.-- Sample districts of Panama with the dependent and independent variables District No Gini I820 I840 Coefficient Changuinola 2 38.96 6.124 16.379 Ch. Grande 3 48.04 4.795 11.155 Aguadulce 4 42.34 4.728 14.050 Colon 10 43.10 4.200 13.250 Boquete 19 46.11 5.452 12.654 David 21 43.45 4.293 13.214 San Lorenzo 27 45.39 7.307 15.179 Chitre 31 43.52 4.643 13.447 Las Minas 32 41.30 9.917 19.833 Los Santos 40 47.23 . 5.699 12.643 Pedasi 42 49.57 6.438 13.048 Panama 52 40.71 4.281 14.006 S.Miguelito 54 37.98 5.542 15.854 Canazas 58 36.68 ~ 11.107 22.213 La Mesa 59 39.82 10.650 20.997 Santa Fe 64 42.55 9.122 18.243 Santiago 65 46.92 4.038 11.144 Sona 66 ‘ 46.56 _ 7.609 15.218 No Median Liter- School Agricul- Popu- ' Family acy enroll- tural lation Income ment employment growth 2 224.9 76.3 80.3 74.2 29.5 3 167.5 27.6 34.5 84.6 15.1 4 262.3 92.1 '90.3 16.1 29.4 10 286.5 95.4 91.5 6.6 22.9 19 180.0 87.6 82.8 53.1 17.4 21 280.3 92.6 89.0 13.5 35.7 27 105.9 50.1 49.9 78.7 21.3 31 249.8 92.8 91.4 11.1 33.4 32 61.9 55.5 66.8 86.7 8.1 40 144.8 84.2 82.0 45.2 15.4 42 102.9 79.9 82.2 68.9 -20.2 52 392.7 96.2 93.0 1.8 29.1 54 319.7 95.9 93.2 l.8 129.0 58 63.1 50.6 68.3 84.9 14.0 59 55.5 68.6 81.9 82.4 1.6 64 72.5 59.4 78.4 82.2 24.1 65 230.4 86.8 90.8' 32.1 34.2 66 92.8 66.2 72.9 71.3 4.4 54. income, and percentage of population employed in agricultural activities. The relationship between the rate of population growth and 1820 also show an indication of the U-shaped curve (See Tables 8, 9 and 10 for more detail). The results shown in Table 8 and represented in Figure 8' support the second order relationship between median family income. and. the. Gini coefficient. In general the districts were dispersed throughout the curve, with most of them concentrated at the "top" of the curve. The functions also showed that although the districts with high median family income had lower income distribution inequalities than the districts located at the top of the curve, some of these districts with high median family income still have more income inequalities than the districts with the lowest median family income. For example, the district of Canazas (58) with a low median family income presented a lower degree of income inequality than the district of Panama (52), which has the highest median family income in the country. This situation suggests that even though the district of Panama has a high median family income, this district. has not "reduced" its income inequalities with development to the same level 'as the district of Canazas. But the reality is that the district of Canazas has not reduced its income inequalities but is in reality an undifferentiated or early economy, .(which by definition should not exhibit much income inequalities) whereas the district of Panama has already begun in its development. 55 Table 8.-- Regression results for Gini coefficient Gini coefficient Independent variables a b1 b2 Adjusted R2 Median family 41.97 0.0527 -0.0002 0.187 income (0.000) (0.007) (0.001) Agricultural '39.27 0.3042 -0.0030 0.296 employment (0.000) (0.000) (0.000) Literacy 42.84 0.0470 -0.0002 0.000 (0.000) (0.408) (0.440) School 41.22 0.1622 -0.0014 0.000 enrollment (0.000) (0.281) (0.236) Population 46.29 —0.0779 0.0003 0.142 growth (0.000) (0.001) (0.0175) Agricultural 45.59 0.0135 -0.0007 0.013 land under (0.000) (0.441) (0.2475) private property One-tail probability levels are in parentheses Table 9.-- Regression results for 1820 56 I820 Independent variables Adjusted a b1 by R2 Median 12.09 -0.0566 0.0001 0.689 family income (0.000) (0.000) (0.000) Agricultural 4.68 -0.0064 0.0006 0.528 employment (0.000) (0.397) (0.010) Literacy -2.24 0.3390 -0.0028 0.430 (0.415) (0.000) (0.000) School -9.76 0.5436 -0.0042 0.330 enrollment .(0.023) (0.000) (0.000) Population 7.19 -0.0408 0.0002 0.114 grthh (0.000) (0.002) (0.010) Agricultural 7.487 -0.0001 0.0001 0.065 land under (0.000) (0.179) (0.456) private property One tail probability levels are in parentheses 57 Table 10.-- Regression results for 1840 I840 Independent Variables a b1 b2 Adjusted R2 Median family 21.94 -0.0849 0.0002 0.377 income (0.000) (0.000) (0.000) Agricultural 15.06 -0.1352 0.0019 0.358 employment (0.000) (0.000) (0.000) Literacy 4.66 0.3777 -0.0031 0.219 (0.166) (0.004) (0.001) School -1.84 0.5211 -0.0039 0.087 enrollment (0.401) (0.008) (0.005) Population 14.79 -0.0151 0.0001 0.000 growth (0.000) (0.239) (0.246) Agricultural 15.29 -o.0377 0.0003 0.015 land under (0.000) (0.248) (0.336) private property One-tail probability levels are in parentheses. 58 55 52— 51* SOF 42 49* 48H '3 4 7.. 40° .65 66° 46— 45— 44— coefficient 43— 42— 41— 40" 39— 38— Gini 36 Median Family Income (in dollars) Figure 8 Relationship between median family income and Gini coefficient. 59 As shown in Figures 9 and 10 the relationship between median family income and both 1820 and 1840 showed that overall, the curvilinear pattern is present. A significant aspect in these relationships is that the population within 1820 received less than 12% of the total income, and for 1840 they received less than 23% of the total, a situation that indicates a disproportional distribution of income. For 1820 the districts with the lowest median family income were those located at the downward slope of the function, whereas the districts with higher median family income were located at the bottom of the curve, where it begins to move upward. This situation suggests that districts with the lowest median family income present a more proportionate share of income per family than districts with higher median family income. A possible reason for this situation is that once the process of development starts ‘it is expected that inequalities ‘will. raise. Therefore, since. districts ‘with higher median family income are considered more developed, they will have less income share than the less developed districts. Later'on with increasing development the share should be more evenly distributed. In Figure 9 the districts with the highest median family income, with the exception of San Miguelito (54), had the lowest income share for 1820. A significant example is the district of Panama (52). This district had the highest median family income in the country, but its population within the 1820 received almost the same income share as the I820 60 12 58. .59 10b" .32 Median Family Income (percent) Figure 9 Relationship between median family income and 1820 61 districts of Colon (10) and Santiago (65) which have much lower incomes and levels of development. A possible reason for this situation is the fact that the district of Panama and the other districts located within the "Metropolitan Region" (Colon and San Miguelito) received an unusually high percentage of migrants during the 1970's and 80's, principally people from the lowest stratum of the population, from other less developed districts. This large uneducated and unskilled population had to compete with those already there for the income that was available, driving down wages and thus driving down the income share of this portion of the population. For 1840 the results were similar to those obtained for 1820. Some districts with high incomes have higher inequalities than others with lower incomes. For example, even though the districts of Panama (52), Colon (10), David (21) and San Miguelito (54) had low income share for 1840, their income share for this dependent variable was less unequally distributed than for districts such as Boquete (19) and Los Santos (40). These four districts mentioned above were located further along the upward slope of the curve (Figure 10). Comparing the position of the districts in the graphs that show the relationship between median family income and the three dependent variables it is observed that the Gini coefficient showed low income distribution inequalities, principally for those districts with a high median family I840 62 58 52- Median Family Income (in dollars) Figure 10 Relationship between median family income and 1840 63 income (eg. Panama and San Miguelito). While for 1820 and 1840 it was noticed that inequalities of income within these sections of the population for the same districts remain high. The results also suggest that although the curvilinear pattern is present, even the most developed districts in Panama have not yet achieved the "best" distribution of income. The inverted U-shaped pattern is also confirmed for the relationship between Gini coefficient and the percentage of population employed in agricultural activities (Figure 11). The. districts ‘with. the Jhighest percentage of population employed. in. agricultural activities. had the lowest Gini coefficients and most of them were clustered on the upward slope of the curve. On the other hand, the districts with the lowest percentage of population employed in agricultural activities had higher Gini coefficients and they are located on the downward slope of the curve. Finally, districts with intermediate levels of population employed in agricultural activities are located at the top of the curve. As an example, the districts of Santiago (65) and Los Santos (40) had less percentage of population employed in agricultural activities than the districts located on the upward slope of the curve, but each also had a higher Gini coefficient, with consequently higher income inequalities. The districts of Panama (52), Colon (10), David (21) and San Miguelito (54) according to the population employed in agricultural activities had the lowest percentage of Coefficient Gini 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 64 Non Agricultural Population “ 42, r r” )— _ 54o _ .58 I I I I 0 20 40 60 80 100 10 30 50 70 90 (percent) Figure 11 Relationship between population not employed in agricultural activities and Gini coefficient 65 population employed in these activities. Following the hypothesis it can be implied that they are the most developed districts in the country. But at the same time, their Gini coefficient were still similar and, for. the district of Panama it was even higher than that for the district of Canazas (58), which is one of the least developed districts, with a high percentage of population employed in agricultural activities. The districts of Chitre (31) and Aguadulce (4), for example, had less percentage of agricultural population and less income distribution inequalities for the Gini coefficient than would be expected after looking at the variable median family income. This situation may indicate that these districts, at least 'in this aspect of development, may be moving toward a greater level of development. The low income distribution inequalities found in districts where a high percentage' of the population is employed in agricultural activities may be related to the economic structure of the districts. In those districts, where most of the population work in agricultural activities, there is a very low level of income but the differences in income among them may be not that great. With the presence of other activities not related to agricultural activities there is a shift of employed population from agriculture to these other activities, principally commerce, industry and service. This situation may create a gap in 66 income between the population employed in agricultural activities and the rest of the population. This gap in income within the population engaged in agricultural activities and the other economic activities is also affected by other factors such as education, migration and fertility. With an increase of secondary and tertiary activities the population employed in agricultural activities may suffer a sharp decrease, especially in areas close to urban centers. The results of the relationship between the percentage of population employed in agricultural activities and the 1820 was significant only for coefficient b2. In other words, only the downward slope of the curve is shown to be significant. The results also seem to indicate that most of the districts in the Republic. of Panama had high income inequalities in the relationship between median family income and the distribution of income share for 1820. For example, the districts of Panama, Colon and David had the lowest percentage of population employed in agricultural activities, but at the same time these districts also had the lowest share income for the 1820 (Figure 12). The income share for 1820 in these three districts was less than 5%. The districts of Los Santos (40), Boquete (19), Pedasi (42), Changuinola (2) and Sona (66) had, in fact, higher income shares for 1820 than for the districts with the lowest percentage of population employed in agricultural activities. I820 67 12 .58 ’59 10— 4_. 65¢ 10 I 3 I I I I I I 0 I 20 I 40 I 60 I 80 I 100 10 3O 50 70 90 Non Agricultural Population (percent) Figure 12 Relationship between population not employed in agricultural activities and 1820 68 The low share of income for the population within 1820 for the districts with the lowest percentage of population employed in agricultural activities may be related to migration patterns also. As is well known in developing countries like Panama, the pattern during the last three decades has been a high migration of population from rural areas to urban areas. Usually the population from these rural areas who migrate were engaged in activities related to agriculture. They are poor, landless and have a minimum level of education. Once they arrive to urban districts they have to compete with more skilled and educated people in the job market. and they have to deal with a higher cost of living. In addition, the most skilled individual may get the best jobs; these immigrants end up performing the lowest paid jobs and often becoming worse off than when they were in the rural areas. Another pattern of this migration from rural to urban areas is that the most educated population has the greatest tendency to leave. Thus, the rural areas are left without this important group of jpeople and suffer the negative economic consequences of an uneducated work force and population. I . Even though the curve for the relationship between the percentage of population employed in agricultural activities and Gini coefficient showed that in general the population with the highest non-agricultural population showed higher income inequalities, the relationship between population 69 employed in agricultural activities and 1820 indicates that the districts with the lowest percentage of people employed in agricultural activities have also the lowest share of income for 1820. In this case, the function showed that the most developed districts in Panama seem to have more unequal levels of income distribution. In the case of the relationship between the percentage of population employed in agricultural activities and 1840 the curvilinear pattern is more clear and it showed a clear upward slope in contrast to the relationship between this independent ‘variable and 1820. Both coefficients of the quadratic equation were significant (Figure 13). The district of Canazas (58) had the highest share of income for the 1840. On the curve, the districts with the lowest percentage of population employed in agricultural activities have a relative better position in relation to their income share than the position they had on the graph for the relationship between this population engaged in agricultural activities and 1820. I The relationship between the rate of population growth and 1820 also supported the U-shaped curve. In this case, most of the districts with low rate of population growth were located at the downward slope of the curve (Figure 14). Districts with a moderate rate of population growth were located at the bottom of the U-curve with the lowest income share for 1820. Districts with .the highest rate of population growth had higher share of 1820 than districts I840 70 25 22— 21” ‘ 18— 17— 16” 540 14“ ”5 19. .40 12— Non Agricultural Population (percent) Figure 13 Relationship between population not employed in agricultural activities and 1840 I820 71 12 11_ .58 I'59 10— .32 -20 20 60 100 140 180 220 0 40 80 120 160 200 240 3I IIIIT I II Population Growth (percent) Figure 14 Relationship between rate of population growth and 1820 72 with a moderate rate of population growth. But these same districts also had a lower income share than districts with relatively low rate of population growth. Thus, an increase in population growth can have negative effects on the share of income for the population within the 1820. The hypothesis for this relationship, that as the rate of population growth increases in a'district, the share of income for 1820 will decrease, with a subsequent increase once the population growth decreases, was confirmed. The results show that districts with the lowest rate of population growth had the highest income share for 1820. An important observation for this relationship is that these districts (Canazas (58), La Mesa (59), Las Minas (32) and Santa Fe (64)) are the same as thbse with the highest percentage of population employed in agricultural activities and the lowest median family income, variables that indicate that these districts are not very developed. On the contrary, districts that are considered more developed, according to the variables mentioned before, were the ones with the highest rate of population growth (eg. San Miguelito, Colon, David and Panama). Although migration is not directly addressed in this study, this high rate of population growth in districts that, according to their characteristics can be considered as developed, might be attributed to migration. A possible explanation for this pattern is that those districts with the lowest rates of population growth were affected by 73 migration trends which have drained out the young people, resulting in negative ’population growth in some districts (e.g. Pedasi (42)). An opposite situation may be true in districts that had a higher rate of population growth in which the income share within 1820 went down as a result of an increase of immigration, principally in those districts that are the capitals of provinces and that have greater urban characteristics. However, this possible relationship could not be examined more closely in this study because the information could not be found in the National Census to differentiate between migration and natural growth at the district level. Another factor that may be affecting the results of the relationship between the rate of population growth and 1820 is fertility. Usually the poorest people have more children because they do not have as much knowledge and accessibility to methods of birth control. Because migration usually involves the population within the fertile ages, creating a movement from rural areas to semi-urban and urban areas, it is expected that the number of children will rise in semi urban and urban locations. Thus, higher fertility rates are inversely related to 1820. 74 Significant Functions with Signs Opposite from Expected Within the group of sub-hypotheses, with t-values with absolute magnitude sufficiently ' large to consider "significant", five of them presented signs opposite to what was expected. Thus, for this group the null hypothesis was technically accepted because the significance of the b coefficient was opposite. However, they warrant discussion. Those sub-hypotheses were: the relationship between literacy and 1820 and 1840; the relationship between school enrollment and 1820 and 1840; and the relationship between the rate of population growth and Gini coefficient. ' For the relationship between the variables representing education (literacy and school enrollment) and 1820 and 1840 several districts were located close to the top of the downward slope of the curve, whereas a cluster representing most of the districts was located near the bottom of the downward slope of the curve (Figures 15, 16, 17 and 18). The same pattern observed in most of the relationships between these two dependent variables and the independent variables is also present in this relationship. The districts with the highest percentages of literacy and school enrollment had the lowest income share for the population within 1820 and 1840. The opposite relationship found between literacy and school enrollment on the one hand with 1820 and 1840 on the other indicates that the poorest population does not become I820 75 12 10" 58' 59 320 64o Figure Literacy (percent) 15 Relationship between literacy and 1820 IS20 76 12 58 11— ~59 10‘— 32. .64 9— 8.— (66 27, 7— 42 6— 2° 40. 19. ‘54 5...— 3' 31o. 21, .52 4__ 65:1 3 I 'l I 20 I 40 I 60 I 80 I 100 30 50 70 90 School Enrollment (percent) Figure 16 Relationship between school enrollment and I820 I840 77 25 24— 23* 20” 19— 18— 15— 14“ 13” 58, .59 .32 6? 1 l I Figure 17 Literacy (percent) Relationship between literacy and 1840 I840 78 25 24" 23* 22* 21* 19* 18* 17* 14* 32’ — 64. 54- 100 30 50 70 90 School Enrollment (percent) Figure 18 Relationship between school enrollment and 1840 79 better off with education, but, on the contrary, they seem to become worse off. Specifically, it was observed that districts of Canazas (58), Las Minas (32) and Santa Fe (64) among others had a low percentage of literacy and school enrollment and a high share of income for 1820 and 1840. This may be related to the situation that people from rural areas who are more educated are more likely to leave their places of origin. Other districts, including Sona (66), Changuinola (2), Pedasi (42) and Los Santos (40), presented higher literacy and school enrollment than the districts mentioned above but less income share for 1820 and 1840. The districts with one of the highest percentage of literate population and school enrollment (Colon (10), David (21), Chitre (31) and Panama (52) among others) also had the lowest share of income for the two measures of inequalities mentioned above (1820 and 1840). Among these districts Santiago (65), David (21), Panama (52), Colon (10), Chitre (31) and Aguadulce (4) were the lowest. They are located on the downward slope, toward the bottom of the curve (Figures 15 and 16). The district of Chiriqui Grande (3) had the lowest literacy and school enrollment of the country and also very low income share for 1820 and 1840, a situation which indicates that the population in this district is still very far away from having a more equal distribution of income as compared to districts that present similar levels of education. 80 The results obtained in .this relationship seems to indicate that even though literacy and school enrollment may play an important role in development, at least within a single country, these variables are not good indicators of development once a district has become more developed. If education affects the income share of 1820 and 1840 as was expressed before, it is expected that the income among districts with similar levels of education will be very similar. In districts where the level of development is more advanced there is a need for a better educated population. As the needs for more skillful and educated people increase in a district, to be literate or to have a minimum level of school enrollment is no longer enough in itself. Of course, migration is a natural result of education and this in turn may- distort the. relationship between education and development. Thus, in the future, measures of higher education are needed together with measures of the creation of job opportunities to assess the migration and educational situation. The level of education may also contribute to an increase in income inequalities between the population that migrates from. the less developed districts to the most developed districts and the population that already resides in the most developed districts. This is due to the fact that in more developed areas the competition for high skilled jobs is stronger than that for less developed areas. The people who are more educated will usually have more 81 opportunities to get better jobs and salaries. Therefore, disadvantages in the level and quality of education are going to be more important in more developed areas than in less developed ones. Thus, the income share of. the population within 1820 and 1840 will be more negatively affected by the degree of education in more developed than in less developed areas. The relationship between the rate of population growth and the Gini coefficient also had significant coefficients which had signs opposite to what was expected (Figure 19). The observations are not dispersed along the entire curve but most of them are clustered along the downward slope of the curve. Another small cluster was located close to the upward slope of the curve. For example, the district of Pedasi (42) had a very high Gini coefficient but negative rate of population growth. In the literature it has been suggested that a low rate of population growth will reduce income inequalities. But in the case of a district with a negative rate of population growth, as the case of Pedasi, it is an indicator that migration. has taken jplace jpossibly ‘with an accompanying decrease in births. This situation may reflect and cause negative effects on the economy of the district because it is likely' the "best” of the :population. are leaving and businesses within the secondary and tertiary sector will not establish themselves there. Thus, a high negative rate of coefficient Gini 82 55 54* 52* 51* 5 0'— 42‘ 49* 48* 47* 46* 45%- 44* 43* 42* 41* 40* 39* 371 QM 36 I I I I I I I -40I 0 I40I80I120I160I200I240 -20 20 60 100 140 180 220 Pop Figure 19 ulation Growth (percent) Relationship between rate of population growth and Gini coefficient 83 population growth may also lead to income inequalities within the population as was registered by Gini coefficient. Two clusters of districts can be found with a rate of population growth between 0 and around 30, one with a high Gini coefficient (between 42.5 and 50) and a smaller one with a moderately high Gini coefficient (between 35 and 41.5). The fact that districts with similar rates of population growth have different Gini coefficients (for example La Mesa (59) and Sona (66) as well as, Canazas (58) and Boquete (19), and Changuinola (2) and Chitre (31)) leads to the suspicion that migration is affecting the results of the Gini at. a national level (within the districts of Panama). Migration is more dynamic within the country, and people have a greater tendency to move from place to place to 'find better jobs, more land, and better social and economic conditions than originally assumed in this research. The district of San Miguelito (54) had one of the highest rate of population growth in the country (129.0%) but its Gini coefficient was one of the lowest (37.98). This district received a significant amount of migration from other areas of the country during the 1970's and 1980's, a situation that contributed to its high population growth. Comparing the results obtained between the rate of population growth and Gini coefficient and between the rate of population growth and 1820 it may appear that although 84 this district has a high population growth, the population in general in this district is doing much better than the rest of the districts. But in reality this district had a lower share of income for 1820 than other districts with smaller rate of population growth. Thus, we can conclude that the Gini coefficient may present problems of interpretation because it does not capture details about the situatiOn of the lowest income levels of the population. In general, it can be concluded that for certain relationships the function and the estimated curve for the U-shape and the inverted U-shaped pattern can be clearly seen with its corresponding increase and decrease of income inequalities with development. But also it was observed that there was no decrease at all in income inequalities for certain districts even though they have reached a degree of development. Another important feature is that for other relationships (e.g. 1820 and Gini coefficient and the rate of population growth) although the function shows a significant relationship for both coefficients of the quadratic equation, the expected signs for the relationship were not obtained and the districts were not evenly distributed throughout the curve; on the contrary, most of the districts were concentrated in one area of the curve, possibly due to the influence of migration. 85 Insignificant Relationships Six relationships were not significant in this study. These relationships are: the relationship between literacy and school enrollment and Gini coefficient; the rate of population growth and 1840; and the percentage of agricultural land. under’ private ownership and the three measures of inequality (Gini coefficient, 1820 and 1840). The results of the regression analysis were insignificant at alpha 0.05 level. Thus, the relationship between the percentage of agricultural land under private ownership and Gini coefficient indicates that they are not interrelated. One reason why this relationship does not show the expected results may be related to the measure used. Even though the fact that the legality of the land ownership can be considered an indicator of development, in this case it seems that there are other factors related to land ownership that need to be considered to establish whether or not this variable as a measure of development is related to income inequalities. Therefore, it would be expedient to find another measurement of distribution of land ownership and to regress it against Gini coefficient. On the other hand, the percentage of agricultural land under private ownership presents a significant negative relationship with 1820, and no relationship with 1840. In the case of 1820 an increase in the amount of farmland privately owned also increases the share of income of this percentage of the population. In addition the weakness of 86 the relationship for both measures of inequalities indicates that data on farmland privately owned are not robust enough to support the relationship between income distribution inequalities and whether and/or how the land is owned. To summarize, the results showed that there is an indication of an inverted U-shaped curve between income distribution measured by Gini and development in Panama. In addition, the relationship between. 1820 and 1840 and development also presents the expected results of a U-shaped curve for most of the independent variables. The relationship between median family income and agricultural employment and the three measures of inequality had, overall, the highest adjusted R2. Even though for the other relationships between the independent variables and the dependent variables the adjusted R2 was low, the results of the relationships were significant for Gini and population growth, as well as for 1820 and literacy, school enrollment, and population growth. For 1840 the relationship between the dependent variable and, literacy and school enrollment were also significant (See Tables 8, 9 and 10). It is also important to mention that the best results for this study were obtained for the relationship between 1820 and the independent variables. From six regressions performed two were significant with the expected signs (median family income and population growth), one had the expected signs and coefficient b2 was significant (agricultural employment) and two were significant with 87 opposite signs to what was expected (literacy and school enrollment). The second best results were shown for the relationship between 1840 and four of the independent. variables. In. this case, two of the relationships ,were significant with‘the expected signs (median family income and agricultural employment) while two were significant but with opposite signs. to ‘what was expected (literacy and school enrollment). For the Gini coefficient two were significant with expected signs (median family income and agricultural employment), and one was significant with the opposite sign (population growth) (See Table 6). CHAPTER V CONCLUSIONS This study concludes that there are significant regional income disparities in the Republic of Panama at the district level. The relationships Abetween this income distribution inequality and some factors of development were analyzed. Several conclusions can be drawn. First, different facets of development such as the level of median family income and population in agricultural activities, in general, showed that income inequalities were less in districts that were either underdeveloped or the most developed according to these two variables. Therefore, the hypothesized relationship between development and income inequality is confirmed for these variables. An important finding was that districts with very low development showed less inequalities in the distribution of income than districts with the highest development. This situation is an indication that even the most developed districts in the country still are not fully developed or developed enough where inequalities had decreased to the point of showing less inequalities than many less developed areas. Thus, for the case of a developing country like Panama there is a curvilinear pattern. But it has not yet reached its complete maturity. An important feature is the fact that 88 89 the range of difference in the levels of income between underdeveloped and developed districts is very wide. A situation is observed for the position of the districts on the curve: districts with low and high median family income also have lower inequalities than district with moderate income. This is due to the fact that the most developed district areas usually are located within the Metropolitan Region, where the base of the economy is more related to the secondary and tertiary sector, whereas in the less developed districts the base is related to the primary sector. Thus, the level of income received by the population living in the Metropolitan area is sometimes more than two times the amount of income the population in the less developed districts receives, resulting in great levels of income inequalities between districts. Looking at the range of the level of income for the different districts and the level of development, it is observed that as development increases the income level increases and in general, inequalities in the distribution of income decreases. Second, the relationship between income distribution and the variables of development chosen for this study seem to show differences according to the measure of inequality used. The Gini coefficient, for example, showed that the most developed districts have a lower degree of inequalities than less developed districts. But it also showed some developed districts like Panama with a slightly greater 90 inequality of distribution of income than that of the least developed districts (e.g. Canazas). Of course, these two study areas have very different dynamics in their economic and social structure, a situation that is reflected in the results of the Gini. The results obtained by the Gini, in general, do not seem to present a high level of income disparities when development is present, but the income share for the poorest 20% and 40% of the population gave more specific results. Although the expectation that districts with a higher level of development would have low income distribution inequalities was statistically confirmed, these results showed that the share of income for these two segments of the population was more unequal in those districts with a medium high and high development (with the exception of the district of San Miguelito) as compared to the less developed districts. This situation implies that development is very relative in a developing country like Panama, and that even though the general tendency is toward a decrease of inequalities with development, still there are factors that may affect the distribution share of income for certain groups of the population, principally the poor ones. In addition, the relationship between the population employed in agricultural activities and the share of income for the poorest 20% and 40% also showed that districts with the lowest percentage of population employed in agricultural 91 activities also had the lowest share of income for the poorest 20%. This implies that other forces may be at work. In addition, districts with the highest levels of literacy rate and school enrollment had the lowest income share for ‘the poorest 20% and 40%. This may due to a concentration of rural and uneducated population in the more developed and urban districts, due .to migration, and with the consequent effects on the income share of these districts. Third, of the three measures of inequalities, 1820 and 1840 more clearly showed the income distribution inequalities. They produced better results for the median family income and population engaged in agricultural activities when used in the regressions. Therefore, it is necessary for future studies to select an additional whole range of variables that can help better reveal the relationship between income distribution inequalities and development, principally when one wants to focus upon the situation of the poorest segments of the population. In particular, the variables of literacy and percentage of school enrollment seem to be extremely deficient as variables. Literacy is no longer an adequate measure of development; being able to read and write is assumed in the modern world and can no longer differentiate a developed and a developing country. The percentage of school enrollment is also deficient because this variable looks to the future, and thus, has no significance to the work force under 92 study. A better measure for future studies may be the percentage of the labor force which has either a high school and/or a university degree. This measure stresses the need that exists in a modern industrial/computerized world for an educated labor force as well as being aplicable to the work force under study. Thus, there is a need for countries to recognize this fact and develop statistics of this kind rather than the literacy rate which has, for all predictable purposes, ceased to be relevant. The interaction of intrinsic forces present in developing countries like Panama such as migration patterns (principally between districts) also plays an important role in the study of the relationship between income distribution inequalities and development. These forces can affect the economy of the more developed districts and, as a consequence, their distribution of income. Both, high skill laborers and those with no skill tend to migrate principally to larger metropolitan areas. This is due to the fact that there is a need for more skilled laborers but, at the same time, there is also a need for cheap labor. The result would be a higher proportion of inequalities in more developed areas when compared to less developed areas since the former areas receive the unskilled laborers from the latter. This situation 'may explain ‘why :most of the highly developed districts also showed less income share for the poorest 20% and 40% of the population as compared to other less developed districts. 93 Limitations Although this study addresses the relationship between development and income inequalities using several variables to represent development, other variables such as migration and fertility, which can affect development, could not be obtained. Another limitation in this study was a lack of adequate data for land ownership. Even though a surrogate variable, percentage of farmland under private ownership, was used, the results were not very promising. A further limitation can be seen in the results obtained from the education data. As noted above, the variables used (literacy and school enrollment) were inadequate in determining the level of development of a district due to the fact that the higher levels of development. in a country are probably better reflected in higher education measures. Recommendations For further study it is recommended that another variable for education should 2 be included to see if the results are different from those obtained with literacy and school enrollment. A good measure may be the percentage of people in college. This suggestion is based on the consideration that the two variables mentioned above did not really show what the level of education is; therefore, it seems likely that using college level will be a better variable to measure development. 94 In the case of the relationship between the rate of population growth and the three measures of inequalities, more studies should be done over time. This thesis uses only one year as a reference, but several years might reveal.more of development dynamics. In addition, future studies should also include a variable that measures natural growth (e.g. fertility) in order to analyze the influence of this variable. Another ‘variable that needs to be restructured. for future studies is the percentage of farmland privately owned. It would be instructive to find out how much land is owned by the different strata of the population over different years and how the distribution of land affects the income share, particularly within the two poorest categories of the distribution (income share for the poorest 20% and 40% of the population). I Finally, the key to obtaining a better distribution of income among the population in a developing country like Panama may not only be present in the level of development that a district has, but in the way this development is achieved. By giving more opportunities to the population to have access to a better education, employment and other social conditions these inequalities may be diminished in the long run. Therefore, other studies regarding development policy, poverty in urban and rural areas as well as economic and social regional disparities are recommended. BIBLIOGRAPHY BIBLIOGRAPY Adelman, Irma., and C.T. Morris. Economic growth and social equity in developing countries. Stanford, California: Stanford University Press, 1973. Adelman, Irma., and Sherman Robinson. Income Distribution Policy in Developing Countries: A case Study of Korea. Stanford: Stanford University Press, 1978. Ahluwalia, Montek S. "Inequality, Poverty and Development." Journal of Development Economic 3 (1976a): 307-342. . "Income Distribution and Development: Some Stylized Facts." American Economic Review 66, N0 2 (1976b): 128-135. Amos, Orley M. "Unbalanced Regional Growth and Regional Income Inequality in the Latter Stages of Development." Regional Science and Urban Economics 18 (1988): 549-566. Berry, Albert., and Ronald Soligo., eds. Economic Policy and Income Distribution in Colombia. Boulder, Colorado: Westview Press, 1980. Bornschier, Volker. "World Economy, Level Development and Income Distribution: An Integration of Different Approaches to the Explanation of Income Inequality." World Development 11, No 1 (1983): 11-20. Browman, Mary J. "A Graphical Analysis of Personal Income Distribution in The United States." The American Economic Review 35, No 4 (Sep, 1945): 607-628. Brown, Lawrence A. "Reflections on Third World Development: Ground Level Reality, Exogenous Forces and Conventional Paradigms." Economic Geography 64, N0 3 (July, 1988): 255-277. Chakravarty, Satya R. "Extended Gini Indices of Inequality" International Economic Review 29, N9 1 (February, 1988): 147-156. 95 96‘ Chilcote, Ronald H. Theories of Development and Underdevelopment. Boulder: Westview Press, 1984. Cline, William R. "Distribution and Development. A survey of Literature." Journal of Development Economics 1 (1975): 359-400. Cook, Cristopher J. "Commodity Price Distortions and Intraagricultural Income Distribution in Colombia." The Journal of Developing Areas 22 (January, 1988): 219-238. Contraloria General de la Republica de Panama. Direccion de Estadistica y Censo. Censos Nacionales de 1980. Volumenes I, 111 y V. [National Census of 1980. Volumes I, III and IV]. PanamazDepartamento de Informacion y Divulgacion, 1980. . Censos Nacionales de 1980. Cuarto Censo Nacional Agropecuario. 17 al 24 de mayo de 1981. Volumen I. Produccion Agricola.[Census of Agricultural Production. Volume 1]. Panama, 1980. Darden, J.T., and A.S. Tabachneck. "Algorithm 8: Graphic and mathematical descriptions of inequality, dissimilarity, segregation, or concentration." Environment and Planning 12 (1980): 227-234. Dookeran, Winston. "The Distribution of Income in Trinidad y Tobago." Review of Income and Wealth 27 (June, 1981): 195-206. Gastwirth, Joseph L. "Notes and Comments. A General Definition of The Lorenz Curve." Econometrica 39, No 6 (November, 1971): 1037-1039. Gerstle, Mack. The Land Divided. A History of THE PANAMA CANAL and other Isthmian Canal Projects. New York: Alfred A. Knopf, 1944. Ginneken, Vouter van., and Jong-goo Park. Generating ' Internationally Comparable Income Distribution Estimates. Geneva: International Labour Office, 1984. Hicks, J.R. Essays in World Economics. Oxford: Oxford University Press, 1959. 97 Higgins, Benjamin. Economic Development. Principles, Problems and Policies. New York: W.W. Norton & Company Inc, 1959. Hirschman, Albert 0. The Strategy of Economic Development. 10th printing. New Heaven and London: Yale University Press, 1966. Instituto Geografico Nacional "Tommy Guardia". Atlas Nacional de la Republica de Panama. Tercera Edicion [National Atlas of the Republic of Panama]. Panama: Impreso Instituto Geografico Nacional Tommy Guardia, 1988. Jaen Suarez, Omar. Geografia de Panama. Tomo 1. [Geography of Panama. Volume 1]. Panama: Imprenta del Banco Nacional de Panama, 1985. Jones Richard C. "Regional Income Inequalities and Government Investment in Venezuela." The Journal of Developing Areas 16, (April, 1982), 373-390. Kuznets, Simon. "Economic Growth and Income Inequality." The American Economic Review 45, No 1 (March, 1955): 1-28. Lecaillon, Jacques; Paukert, Felix; Morrison, Christian; and Germidis, Dimitri. Income Distribution and Economic Development an Analytical Survey. Geneva: International Labour Office, 1984. Lewis, W.A. "Economic development with unlimited supplies of labor." The Manchester School 22, 139-191. . Mahler, Vincent A. "Income Distribution within Nations. Problems of Cross-National Comparison." Comparative Political Studies 22, N9 1 (April, 1989):3-32. Meier, Gerald M. Leading Issues in Development Economies. Selected Materials and Commentary. New York: University Press, 1964. Meier, Gerald M., and Dudley Seers. eds. Pioneers in Development: A World Bank Publication. New York: Oxford University Press, 1985. Mera, Koichi. Income Distribution and Regional Development. Japan: University of Tokyo Press, 1975. 98 Murphy, D. C. "Calculation of Gini and Theil Inequality Coefficients for Irish Households Incomes in 1973 and 1980." The Economic and Social Review 16, No 3 (April, 1985): 225-249. Myrdal, Gunnar. Rich Lands and Poor. The Road to World Prosperity. New York: Harper & Brothers- Publishers, 1957. Nugent, Jeffrey B. "An Alternative Source of Measurement Error as an Explanation for the Inverted-U Hypothesis." Economic Development and Cultural Change 31, N9 2 (Jan, 1983): 385-396. Nugent, Jeffrey B., and Fayez A. Tarawneh. "The Anatomy of Changes in Income Distribution and Poverty among Mexico's Economically Active Population between 1950 and 1970." The Journal of Developing Areas 17 (January, 1983): 197-226. Papanek, Gustav F., and Oldrich Kyn. "The Effect on Income Distribution of Development. The Growth Rate and Economic Strategy." Journal of Development Economics 23 (1986): 55-65. Pryor, Federic L. "Changes in Income Distribution in Poor Agricultural Nations: Malawi and Madagascar." Economic Development and Cultural Change 39, N0 1 (Oct, 1990): 23-45. Quan, Nguyen T., and Anthony Y.C. Koo. "Concentration of Land Holdings: An Empirical Exploration of Kuznets' conjecture." Journal of Development Economics 18, (1985): 101-117 Ram, Rati. "Economic Development and Income Inequality: Further Evidence on the U-Curve Hypothesis." World Development 16, No 11 (1988): 1371-1376. Randolph, Susan. "The Kuznets Process in Malaysia." The Journal of Developing Areas 25,(Oct, 1990): 15-32. Robinson, Sherman. "Income Distribution in Developing Countries: Toward an Adequate Long- Run Model of Income Distribution and Economic Development." American Economic Association 66, N0 2 (May, 1976): 122-127. . "A Note on the U-Hypothesis Relating Income Inequality and Economic Development." The American Economic Review 66, No 3 (June, 1976): 437-440. 99 Rossi, Jose W. "Income Distribution in Brazil: A Regional Approach." The Journal of Development Studies 17, N9 2 (January, 1981): 226-234. Sahota, Gian Singh. Poverty Theory and Policy. A study of Panama. Baltimore and London: The Johnsons Hopkins University Press, 1990. Seers, Dudley. "The Meaning of Development." International Development Review 2 (1977): 2-7. Slater, David. "Underdevelopment and Spatial Inequality. Approaches to the Problems of Regional Planning in the Third World." Progress in Planning 4, No (1975): 97-167. Slottje, D. J.; R. L. Basmann; and M. Nieswiadomy. "On the empirical relationship between several well known inequality measures." Journal of Econometrics 42, (1989): 49-66. Small, John., and Michael Witherick. A Modern Dictionary of Geography. Second Edition. London: Edward Arnold, a division of Hodder and Stoughton Limited, 1990. Smith, Courtland L. "Patterns of Wealth Concentration." Human Organization 50, N0 1 (1991): 50-60. Smith, David M. Where the Grass is Greener. Living in an unequal world. Baltimore: The Johns Hopkins University Press, 1982. Sopher, David E. "A Measure of Disparity." The Professional Geographer 26, N0 4 (Nov, 1974): 389-392. Stewart, Charles T. Jr., and Jin-Hsia Lee. "Urban Concentration and Sectoral Income Distribution." The Journal of Developing Areas 20 (April, 1986): 357-368. Sundrum, R.M. Income distribution in less developed countries. London and New York: Routledge, 1990. Swindell, K.; Baba, J.M.; and Mortimore, M.J., eds. Inequality and Development: Case Studies from the Third World. London: Macmilla, 1989. Todaro, Michael F. Economic Development in the Third World 3rd ed. New York: Longman: 1985. 100 Weisskoff, Richard, and Edward N. Wolff. "Structure of Income Inequality in Puerto Rico." Economics 9, (1981): 205-228. Williamson, J.G. "Regional Inequality and the Process of National Development: A description of Patterns." Economic Development and Cultural Change 13, (July, 1965): 3-44. MICHIGQN STQTE UN “III IIIIIIIIIIIIIIIIIIIIIIIII IIES , II1I2I9IIII