URBANIZATION, PERSONAL INCOME, AND PHYSICAL QUALITY OF LIFE: T H E CASE OF KENYA
YORK W. BRADSHAW The Ohio State University
This paper addresses a variety of methodological, theoretical, and historical problems associated with previous research on urbanization and development in Kenya. The first part of the paper discusses several general theories of Third World urbanization and development.Next, these perspectivesare examined within the context of recent historical circumstances in Kenya. The final part of the paper presents an entirely new quantitative study of urbanization and development in Kenya. It improves on earlier research by using (l) data from all urban regions of the country, (2) a statistical model that tests change over time, and (3) several new variables. Overall, the analysis shows that both the causes and effectsof urbanization are more complex than what was indicated in previous studies. The quantitative findings can be explained by reference to various theoretical and historical concerns discussed in the paper.
frica's level of urbanization is increasing rapidly, although it still lags behind many areas of the world (Hay, 1977). Unfortunately, the study of urbanization in Africa has been severely inhibited by inadequate data. Few countries on the continent have regular census and/or survey data necessary for determining the causes and effects of urbanization. One exception to this generalization is Kenya, a nation that has maintained relatively good data on urban expansion and other characteristics which affect economic development. Kenya has completed two rather extensive post-independence censuses which show that the nation's urban population increased from 9.9 percent in 1969 to 15.1 percent in 1979. The Nairobi Urban Study Group estimates that the level of urbanization by the year 2000 will be 25-30 percent (Ominde, 1984:61-62), a figure slightly lower than that projected by Kenya's Ministry of Finance and Planning (Republic of Kenya, 1985:28). Some urban residents find wage employment in manufacturing, commercial, or service jobs, but many remain unemployed or work in
A
16
Studies in Comparative International Research / Winter 1988
the informal sector (Werlin, 1981). Not surprisingly, many urban dwellers are unable to afford decent housing and thus are forced to live in the slums and squatter settlements that surround the city. Increased urbanization has created an obvious problem in Kenya: Cities cannot provide adequate employment and services for the urban population. Urban growth has exacerbated unemployment and underemployment, expanded squatter settlements, and raised feelings of frustration and despair. This situation underscores the importance of examining the causes of urban expansion in an effort to facilitate balanced development. Despite the importance of studying urbanization in Kenya and the availability of relatively good urbanization data, there have only been a few quantitative attempts to analyze its city growth. Several studies have used multiple regression techniques to examine the causes of rural-to-urban migration in Kenya (Rempel, 1979, 1981a, 1981b; Anker and Knowles, 1983). However, these studies suffer from four major shortcomings. First, despite the fact that Kenya had 47 urban centers in 1969 (and 90 in 1979), these studies are based on information from only 11 cities. Rempel (1981b:43) notes, therefore, that "considerable caution must be exercised in generalizing to the total country unless one has an objective basis for determining the relative importance of each urban center in the urbanization process." Second, all of these studies use a cross-sectional statistical design that provides only a "snap shot" view of the urbanization process in Kenya. This design does not tell us anything about the effects of certain variables over time. Moreover, as discussed in several global studies of underdevelopment, crosssectional models may generate results based on the reciprocal causation of one or more variables (e.g., Bornschier et al., 1978; Rubinson and Holtzman, 1981; Bornschier and Chase-Dunn, 1985). Third, these studies do not investigate the relationship among urbanization, different segments of the modern sector, education, personal income, and physical quality of life. As shown below, this omission paints an incomplete picture of urbanization and development in Kenya. Finally, these studies do not link their quantitative findings to several theoretical and historical concerns relevant to city expansion and underdevelopment. This study corrects these methodological, theoretical, and historical shortcomings. The first part discusses several general theories of Third World urbanization and development. Next, these perspectives are examined within the context of recent historical circumstances in Kenya. The final part of the paper presents an entirely new quantitative study
Bradshaw
17
of urbanization and development in Kenya. It improves on previous research by using (I) data from all urban regions of Kenya, (2) a statistical model that tests change over time, and (3) several new variables. Overall, the analysis shows that both the causes and effects of urbanization are more complex than what was indicated in earlier research. The quantitative findings can be explained by reference to various theoretical and historical concerns addressed in the paper. THEORETICAL ISSUES Economists and some sociologists have examined Third World urbanization from the standpoint of modernization theory. This view typically argues that rural migrants are pulled to urban areas by relatively high industrial wages. Kelley and Williamson (1984:179), for example, state clearly that "industrialization (and manufacturing employment growth) has been the 'engine of urbanization' in the past and will continue to be so in the future?' Continued urban growth in the face of urban unemployment is explained by Todaro's argument that migrants do not respond simply to the actual wage differential between country and city, but rather to the expected differential (Todaro, 1985: 258-61; see also Rogers and Williamson, 1982:472-73; Stark, 1982:66). Thus, migrants weigh their earnings potential in rural and urban areas and choose the region that offers the highest "expected" gain. Todaro (1985:253) acknowledges that rural-to-urban migration is the fundamental cause of surplus labor and unemployment in Third World cities, problems that increase "structural imbalances between urban and rural areas?' Despite the problem of urban unemployment, economists generally perceive rural-to-urban migration as a positive feature. Townward migration supposedly enhances total national output because citizens move from areas of "low marginal productivity and low wages to regions with higher marginal productivity and wages" (Berliner, 1977:448). Moreover, surplus urban labor--particularly in the form of informal-sector work--may facilitate economic expansion by decreasing the cost of labor and materials for formal-sector industries (Todaro, 1985:282-83). Informal labor provides a cheap subsidy to formal labor, preserving scarce Third World capital and increasing national productivity. Sociologists have also argued that several noneconomic factors are crucial when making the societal transition from "traditional" to "modern." Inkeles and Smith (1974), for instance, claim that modern
1.8
Studies in Comparative International Research / Winter 1988
societies are comprised of people who possess certain social-psychological characteristics. The authors conducted a study that interviewed about 1000 men aged 20-30 in six underdeveloped countries. They attempted to define a "modern man" in terms of his attitudes and values; specifically, they formulated a measure of"overall modernity" based on over 20 "themes" ranging from a person's acceptance of new ideas to his attitude toward work (Inkeles and Smith, 1974:15-35). After testing various arguments, the authors conclude that a modern person is "an informed participant citizen; he has a marked sense of personal efficacy; he is highly independent and autonomous in his relations to traditional sources of influence.., and he is ready for new experiences and ideas, that is, he is relatively open-minded and cognitively flexible" (Inkeles and Smith, 1974:290). How does a person become modern and what impact does individual modernity have on society as a whole? Inkeles and Smith (1974; see also Delacroix and Ragin, 1978) tested the effects of "modernizing institutions" like the school, the factory, and the mass media on individual modernity scores. Education was by far the most powerful institution in shaping individual modernity, followed by factory experience and exposure to the mass media. The authors also found that nonindustrial (primarily informal) jobs have a modest impact on modernization, presumably because these occupations expose people to a diversity of situations and provide workers with an element of autonomy in arranging their work. A somewhat surprising result concerned the finding that urbanization does not have a direct positive effect on individual modernity. It does have an indirect effect, however, because modernizing institutions are more heavily concentrated in urban areas than in rural regions. The process of individual modernization is especially important, argue Inkeles and Smith (1974:315-16), because "neither rapid economic growth nor effective government can develop, or, if introduced, will be long sustained, without the widespread diffusion in the rank and file of the population of those qualities we have identified as those of the modern man." Delacroix and Ragin (1978:125) summarize Inkeles and Smith's entire theory as a process whereby modernizing institutions create modernized individuals who then staff the modern institutions that stimulate economic development. Thus, economic underdevelopment is caused by deficient modernizing institutions that cannot inculcate modern (primarily Western) values which supposedly stimulate national economic expansion. In sharp contrast to this strand of modernization theory, dependency
Bradshaw
19
and world-system arguments attribute both urbanization and underdevelopment to structural characteristics in the global economic system. Several quantitative, cross-national studies have demonstrated that foreign investment increases urbanization, expands the service and informal labor sectors, and retards economic development/growth (see Evans and Timberlake, 1980; Kentor, 1981; Timberlake and Kentor, 1983; Bradshaw, 1985, 1987; London, 1987). These results are congruent with theoretical expectations concerning the effects of external capital. Foreign investment in large-scale agricultural production may push rural inhabitants to urban areas (Ledogar, 1975; Walton, 1977), while investment in urban manufacturing pulls people to cities (Bradshaw, 1987). Unfortunately, foreign capital is concentrated in capital-intensive industry and therefore does not appreciably increase urban employment. Without adequate urban employment, increased urbanization expands unemployment and underemployment (i.e., informal labor) in Third World cities. According to dependency and world-system arguments, the informal sector provides cheap goods and services for formal-sector enterprises within the periphery. The benefits of this subsidy to the formal sector also extend beyond the borders of the Third World. Multinational corporations, for example, take advantage of low wages in the periphery by establishing factories that make products at a low cost. These products are either exported back to the core or sold to Third World elites, thus enhancing multinational profits. This transfer of capital and other resources out of the Third World tends to retard its development while further enriching core countries (see Portes and Walton, 1981; Portes, 1985; Timberlake and Lunday, 1985). This view directly opposes arguments advanced by modernization theory which assert that informal labor enhances economic expansion because it provides a cheap subsidy to the formal sector. In agreement with the dependency/world-system perspective, the urban bias theory also asserts that rapid urbanization has a harmful effect on Third World development. However, the latter view does not attribute this situation to the harmful effects of foreign capital; instead, it cites evidence showing that Third World states allocate a disproportionate share of domestic resources to urban-based enterprises (Lipton, 1977, 1984). This bias in favor of urban areas translates into a better quality of life for urban residents, who enjoy higher levels of consumption, better wages, and greater productivity than rural inhabitants. The disparity in welfare between urban and rural areas tends to draw citizens to cities. Rural-to-urban migration also inhibits national eco-
20
Studies in Comparative International Research / Winter 1988
nomic expansion because it draws people away from agriculture, the sector that would provide a sufficiently high per capita income so that "the major sacrifices of consumption, required for heavy industrialization, can be undertaken without intolerable hardship and repression" (Lipton, 1977:23). In the absence of mass agricultural development, industrial efforts cannot be sustained and economic development will falter. Both dependency/world-systemand urban bias theories suggest that rapid urbanization diminishes the ability of poor countries to provide adequate services for their citizens. Migrants usually are forced to live in slums and squatter settlements, areas that seldom have adequate housing or basic sanitation facilities like flush toilets and clean water. Moreover, the rate of urban expansion has exceeded state capacity to educate children in cities. This is simply another example of the negative association between city growth and the distribution of basic services to urban residents. Although the urban bias perspective acknowledges that cities have more services than rural areas, it emphasizes that such services are monopolized by urban elites (Lipton, 1984). THE CASE OF KENYA The post-independence Kenyan state has implemented numerous policies relevant to modernization and urban growth. One important policy was created tc increase African commercial employment, a segment of the economy formerly dominated by Asians. In fact, before independence Asians owned and controlled most activities in wholesale and retail trade, real estate, and manufacturing. This wealthy minority may have owned as much as three-quarters of all private nonagricultural assets in Kenya at the time of independence in 1963 (Leys, 1975:45). Participation in commerce and manufacturing proved lucrative and, in the words of Leys (1975:44), meant that "the Indian trading community . . . had produced the prototype of a national industrial bourgeoisie?' Thus, it is not surprising that a tiny African capitalist class sought government assistance in achieving exclusive access to commercial activities after independence. This movement had its origins in 1965 with the formation of the Kenya National Trading Corporation (KNTC). Although the KNTC was established to take over all import-export trade in Kenya, it eventually became the vehicle for the Africanization of distribution throughout Kenya (Swainson, t980:187). The first state action in favor of local capitalists occurred in 1967 when the government passed the
Bradshaw
21
Trades Licensing Act. This legislation helped Kenyans in two ways: it excluded non-citizens (primarily Asians) from all trade in rural and non-central urban areas, and it barred non-citizens from trading a wide variety of products. These products first included basic goods like maize, rice, and sugar, but later covered manufactured items such as batteries, insecticides, hardware, cement, and other products (Swainson, 1977:41, 1980:187)? The Kenyanization of domestic trade was slowed by a feud between large-scale bourgeoisie traders and small-scale petty bourgeoisie traders. The tbrmer group used the KNTC to gain access to lucrative distributorship contracts with foreign firms. For example, by 1974 the KNTC had forced foreign cement companies to distribute their products through local citizens chosen by the KNTC. The local distributors in Nairobi included members of the Kenyan bourgeoisie, a member of the parliament, and the brother of the KNTC chairman (Swainson, 1977:41-42, 1980:187). Although many foreign firms did not yet use African distributors, it is clear that other companies like British American Tobacco, Cadbury Schweppes, and Bata used the dominant sector of the bourgeoisie to distribute products. In short, a powerful local economic and political elite maintained its position largely through cooperative efforts with foreign capital. This elite group even included the family of President Kenyatta (Swainson, 1978:366). As could be expected, small-scale traders resented the unofficial monopoly over lucrative distributorships enjoyed by wealthier Africans. Starting in 1969, small traders used the Nairobi Chamber of Commerce to pressure the Ministry of Commerce to force all foreign firms to Africanize distribution. Moreover, the Chamber of Commerce demanded that more traders be awarded distributorships through the KNTC which had previously favored a few elite traders. These elites vehemently opposed the Chamber's proposals because such action would seriously erode the monopoly they maintained over product distribution. Bourgeoisie traders and their organization, the Kenya Association of Manufacturers (KAM), were able to forestall any governm e n t l e g i s l a t i o n on a m o r e e q u i t a b l e s y s t e m of awarding distributorships until the mid-1970s. However, in 1975 the conflict reached a crisis point and, after personal intervention by President Kenyatta, an amendment to the Trades Licensing Act of 1967 was passed. It stated that all manufactured goods produced by foreign firms must use citizen--preferably African--distributors selected by the KNTC. This legislation opened up many commercial opportunities
22
Studies in Comparative International Research / Winter 1988
that had not been available to small-scale traders (Swainson, 1977:42, 1978:367, 1980:189). Despite government legislation to Africanize commerce, many noncitizen Asians were still involved in small-scale commercial activities in the early 1970s. This prompted the Kenyan government to use the provincial trade officers of the Ministry of Commerce and Industry to make it difficult for non-citizens to renew their trade licenses. Between 1972 and 1975, many non-citizens in the nation's most populous cities were forced to sell their shops to citizens. A substantial number of shops were sold to Asian citizens because there were not enough Africans with sufficient capital to purchase the businesses. By 1975, however, even citizen Asians were told to leave tlieir businesses in favor of Africans. During this year the Ministry of Commerce and Industry told 463 shops and businesses in Nairobi alone to close, including 69 owned by citizen Asians (Swainson, 1978:369, 1980:192-93). Increased commercial opportunity for Africans represents an important feature of post-independence development strategy in Kenya. Specifically, the government has linked commercialization to urbanization ever since the first development plan lamented that "[t]here are hardly any African-owned shops in the main shopping streets in the larger cities and towns, and even the sales personnel in these shopping streets are overwhelmingly non-African" (Republic of Kenya, 1965:269-70). In 1966, non-citizen businesses accounted for 70 percent of all wholesale and retail sales in urban areas (Republic of Kenya, 1970:413-14). Early development plans outlined a specific strategy to integrate commerce with urban expansion. Starting with the second development plan, the government stressed that "[m]edium size towns have been designated to serve as the main commercial centres for an entire district" (Republic of Kenya, 1970:87). The expansion of these towns was largely responsible for the rapid increase in urban areas throughout Kenya. As reported in Table 1, the total number of urban centers in Kenya expanded from 47 in 1969 (9.9 percent of the total population) to 90 in 1979 (15.1 percent of the total population) (Republic of Kenya, 1981).2 Growing urbanization underscores the inequality that exists within Kenyan society. On the one hand, a small group of government and business elites enjoys access to employment, education, and adequate housing in urban areas. This segment of the population enjoys a relatively large income and a good physical quality of life. On the other hand, an increasing number of rural-to-urban migrants are unemployed or "employed" in the informal sector. They usually live in the
23
Bradshaw
Table t. Urban Population in Kenya, 1969and 1979 SIZE O F U R B A N POPULATION
100,000+
NUMBER OF URBAN CENTERS 1969 1979
2
3
20,000-99,999
2
13
10,000-19,999
7
11
5,000-9,999
11
22
2,000-4,999
25
41
47
90
1.080
2.309
9.9
15.1
TOTAL NUI~F.ROFURJt~ CENTERS TOTAL URBAN POPUIAYIOR ( m i l l i o n s ) PERCENTAGE OF TOTAL POFULATION
Source: Republic of Kenya, 1981 [Vol. 2]:14.
slums and squatter settlements that surround urban areas. Though it is difficult to determine exactly bow many people live in urban slums, there have been several attempts to estimate this number. In 1971, one report stated that one-third of Nairobi's population resided in squatter settlements. The largest single settlement is Mathare Valley, which grew from 50,000 to 69,000 between 1971 and 1976 (Stren, 1984:239-41). The population of Mathare Valley has continued to increase in the 1980s and is now over 100,000 (Chege, 1981; Lamb, 1985:25-28). Another Nairobi slum--Dagoretti--has also expanded in recent years and now contains over 100,000 people. Thus, Mathare and Dagoretti together comprise about a quarter of Nairobi's total population (Chege, 1981). In the most recent development plan, the Kenyan government acknowledges its inability to provide adequate living conditions for all citizens. It states: "The objective of the Government's housing policy since Independence has been to provide adequate shelter for all, in urban and rural areas. Rapid urbanization since Independence has resulted in the problem of the proliferation of unplanned urban settlements lacking in essential services such as potable water and sanitation" (Republic of Kenya, 1984a:33-34). In addition to housing shortages, rural-to-urban migration has increased "overcrowding in educational, health and other facilities in urban centres" (Republic of Kenya, 1981 [Vol. 2]:15).
24
Studies in Comparative International Research / Winter 1988
Importantly, the state lacks the resources to build schools for the increasing number of children and teenagers in squatter settlements. This is significant because secondary education is one of the surest vehicles to economic and political power in Kenya. Todaro (1981; see also Bigsten, 1984) argues that the current educational system tends to increase inequality within Kenya because personal income is highly correlated with level of education. Secondary education is not free in Kenya and therefore many families are unable to send their children to secondary school. Further, since the majority of Kenyans do not attain an education at the secondary level, the educated minority is disproportionately represented in jobs that pay relatively high wages. Court (1984:285) also notes that wealthy and politically influential regions continue to provide their children with a superior education. This is done largely through construction of self-help harambee (a Swahili idiom meaning "Let us pull together") schools which are built with donations from citizens, especially political and economic elites. Kenya has a long history of harambee projects in education and other community activities (see Thomas, 1985). Although elites have heavily influenced these projects, nonelites have also contributed. For example, many rural-to-urban migrants work in the city and then remit earnings back to the countryside. Some of these earnings are used to pay school fees and contribute to fundraising drives which benefit education. But increased urbanization may inhibit the future contributions of nonelite citizens. Specifically, the growing underemployment and unemployment associated with rapid urbanization may reduce the amount of money remitted to rural areas. This possibility, combined with the increasing pressure on urban educational facilities, may inhibit the future rate of educational expansion in both rural and urban areas. The emergence of this possibility would be significant because, in addition to enhancing personal income and power, education is associated with a better physical quality of life. The most recent Kenyan census, for instance, provides data showing a strong negative relationship between mothers' level of education and child mortality (Republic of Kenya, 1981 [Vol. 2]:89-100). Education increases knowledge about sanitation, nutrition, health care, and other issues related to quality of life. Unfortunately, because education is disproportionately enjoyed by the rich, those with the worst living conditions have the least knowledge about improving their quality of life. This section has briefly sketched several factors that have influenced patterns of urbanization and development in Kenya since indepen-
25
Bradshaw
dence. The next section uses a statistical analysis to test some of these propositions. STATISTICAL D E S I G N
Because previous research on urbanization in Kenya has relied on cross-sectional designs, it is important to use a model that tests change in certain variables over time. Several dozen quantitative, cross-national studies have examined change by using a panel regression model. This is an effective technique that regresses each dependent variable at a later point in time on its value at an earlier point in time and on the independent variables at the earlier time point (see Bornschier et al., 1978; Rubinson and Holtzman, 1981; Bornschier and Chase-Dunn, 1985, for a review of these studies). The following equation illustrates a typical panel model: Y2 = a + BIY I -k BzX l -{- B3X t -{- e,
where Y: = the dependent variable at a later point in time; a = the constant term; YL = the dependent variable at the earlier point in time; X~ and X, = independent variables at the earlier point in time; and e = the error term. This design has two important advantages over a strict cross-sectional model. First, a panel model controls for the prior effect of the dependent variable (denoted by Y,) and thus prevents false inferences due to reciprocal causation. Second, a panel model has another purpose, especially when all variables are logarithmically transformed, a common correction for skewness and/or heteroscedasticity. Specifically, as noted by Hanushek and Jackson (1977:96-101), when the natural logarithm of a variable is regressed on the natural logarithm of another variable, the equation actually evaluates the percentage change in the dependent variable caused by every percentage change in the independent variable. In other words, the regression of log(Y2) on log(Y,) would measure the rate of change in the dependent variable over the time period specified. If log(X,) and log(Xt) were added to the equation, then we could assess the rate of change in the dependent variable produced by every percentage change in the independent variables. Again, this is a common strategy in cross-national studies of underdevelopment. Following the logic of cross-national research, this study uses a panel regression model that tests data across 32 districts of Kenya at two points in time, 1969 and 1979. 3 These two dates represent the only
26
Studies in Comparative International Research / Winter 1988
post-independence years for which census data have been collected. All variables are logarithmically transformed to correct for skewness and to facilitate an analysis based on variables' rate of change. Each variable in the analysis is described below, and the Appendix reports the data sources for the indicators. It should be noted that all money is measured in Kenyan currency, thus avoiding the usual problems associated with converting foreign currency into U.S. dollars. Urbanization. Because this study uses district-level data, the urbanization indicator is measured as the proportion of the entire district that resides in urban areas. The 1969 census listed 47 cities with a population of 2,000 or more people, the government criterion for a city. Nine of Kenya's 41 districts did not have any urban areas in 1969 and, consequently, they were eliminated from the analysis. 4 The urbanization indicator is a dependent variable for part ofthe analysis and therefore is calculated at two points in time, 1969 (URBAN69) and 1979 (URBAN79). When urbanization is a dependent variable, the panel model assesses the rate of change (i.e., rate of growth) in the urban population between 1969 and 1979 (see Table 1). Employment Variables. This study uses four different employment indicators in 1969 as independent variables. These variables include the proportion of the population employed during 1969 in modernsector agriculture (AEMPL69), manufacturing (MEMPL69), commerce (CEMPL69), and services (SEMPL69). Agricultural labor refers primarily to wage employment on large-scale farms and plantations (e.g., picking coffee or tea), while the nonagricultural indicators include mainly urban occupations (see Republic of Kenya, 1972:iii). Manufacturing labor refers to numerous activities from meat slaughtering to factory work to handicraft production. Commercial labor refers to employment in wholesale and retail trade as well as in banks and financial institutions. And service labor refers to a wide variety of service activities in health, tourism, government, and other areas. Earlier studies of urbanization in Kenya have not examined the separate effects of different types of employment. This strategy may be incorrect, however, because the previous section indicated that the Kenyan government has endeavored to increase commercial opportunities for Africans. Available data show that, on balance, commercial employment is consistently the highest paying type of labor in Kenya. In 1969, average annual (modern-sector) agricultural earnings were only 14.9 percent of average commercial earnings; average service earnings were 52.6 percent of average commercial earnings; and average manufacturing earnings were still only 69.3 percent of average corn-
Bradshaw
27
mercial earnings (Republic of Kenya, 1972). These figures do not represent an aberration. In 1984, an examination of the private sector shows that average agricultural earnings were 14.8 percent of average commercial earnings; average service earnings were 36.4 percent of average commercial earnings; and average manufacturing earnings were 54.9 percent of average commercial earnings (Republic of Kenya, 1985). Secondary Education. This study is interested in the effects of secondary school enrollment on both urbanization and development. Following cross-national research, secondary school enrollment is measured as the proportion of potential secondary school students in each district who attended secondary school. Since this indicator is a dependent variable for part of the analysis, it is calculated in both 1969 (SECED69) and 1979 (SECED79). When secondary education is a dependent variable, the panel model evaluates the rate of change (increase) in secondary school enrollment between the two points in time. Amenity Index. The analysis is interested in whether the availability of urban amenities will facilitate urbanization and economic development. To examine these issues, this study includes an amenity index that was constructed by the Kenyan government and is summarized by Rempel (1981b:128-131). The index assigns a point value of 0-3 for 21 different amenities that were present in towns and villages in 1969, with 0 indicating the absence of a particular amenity and 3 indicating the highest availability of an amenity. Only 10 amenities have complete data for each district and these are included in the index. The amenities evaluate the availability of administrative offices, judicial services, police services, health facilities, education facilities, postal services, auto services, air transport, retail services, and open markets. The amenity index is an independent variable and thus is calculated in 1969 (AMEN69). Income Disparity Between Urban and Rural Areas. The urban bias perspective is especially interested in factors that contribute to an overall disparity between country and city (Lipton, 1977, 1984). In this paper, the disparity is measured as the ratio of average earnings in modern-sector nonagricultural (i.e., urban) enterprises to average earnings in modern-sector agricultural activities (average nonagricultural earnings/average agricultural earnings). 5 The strongest proponent of the urban bias theory also uses the non-agricultural sector as a proxy for the urban sector and the agricultural sector as a proxy for the rural sector (Lipton, 1977, 1984). The income disparity between urban and rural areas is an independent variable that is calculated in 1969 (DISP69).
28
Studies in Comparative International Research [ Winter 1988
Quality of Agricultural Land. It is also important to control for the quality of agricultural land when testing potential causes of rural-tourban migration. It is reasonable to speculate that the availability of high-quality agricultural land would reduce the amount of migration to cities. To examine this possibility, an independent variable measuring the thousands of hectares of "high potential" land in 1969 is included in the analysis (LAND69). Land is considered "high potential" by the Kenyan government if it receives a large amount of annual rainfall. This indicator is standardized on total district population because the effect of land quality on urbanization may vary according to the district's population. If high-quality land is extremely crowded, then more people may migrate townward than if such land is not densely populated. Personal Income. Most studies measure individual economic development in terms of per capita income, even though this measure obviously does not assess income distribution. Because district-level data on GDP per capita do not exist in Kenya, this study uses total per capita earnings as a measure of personal income. This indicator is a dependent variable for part of the analysis and therefore is measured in 1969 (INCOM69) and 1979 (INCOM79). When individual income is a dependent variable, the panel model evaluates the rate of change (increase) in personal income over the specified 10 year period. Physical Quality of Life. Few underdeveloped countries have detailed district- or local-level data on indicators of physical quality of life. In fact, many countries do not have good national-level data on such indicators as calorie and protein consumption, total number of physicians, literacy, and infant/child mortality. To help compensate for the absence of mortality data, William Brass developed a technique in the 1960s to estimate child mortality based on the reported survival rates of children (Brass and Coale, 1968; see also United Nations, 1967). He shows that the proportion of children dying before age 1 is close to the proportion of children reported dead by women aged 15 to 19; the proportion of children dying before age 2 is close to the proportion of children reported dead by women 20 to 24; and the proportion of children dying before age 3 is close to the proportion of children reported dead by women aged 25 to 29. This technique also adjusts for societies that have an early or late fertility schedule by establishing multipliers to correct for unusual rates of fertility. Brass's technique and variations of his method have been used by the Republic of Kenya and independent sources to estimate mortality rates from post-independence census data (Republic of Kenya, 1970-1977,
Bradshaw
29
1981; Ankers and Knowles, 1983). Brass shows that his technique is more accurate when estimating child mortality than infant mortality; consequently, this study measures physical quality of life as the proportion of children dying before age 3, a variable estimated from the proportion of children reported dead by women aged 25-29. Child mortality is also a better measure than infant mortality for Kenya because many children die in the first several years of life (not just infancy) from a variety of diseases (see Brass and Coale, 1968). The child mortality indicator is a dependent variable and thus is calculated in 1969 (MORT69) and 1979 (MORT79). Again, the panel model examines the rate of change in child mortality between 1969 and 1979.6 RESULTS OF THE PANEL ANALYSIS To preserve already scarce degrees of freedom, it is essential that unnecessary variables are excluded from each equation. This analysis runs a variety of equations to ensure that results are not simply statistical artifacts. The level of statistical significance used throughout the analysis is P < .10 (.05 for a one-tailed test) since only 32 cases are tested. This number is reduced to 29 when equations include the quality of land variable, which is missing data on 3 cases. 7 The analysis begins with an attempt to evaluate the effects of modern-sector employment and several other variables on rate of change in urbanization. The employment variables are not entered simultaneously so that each will have a maximum opportunity to explain variation in rate of change in urbanization. Equation 1 in Table 2 regresses rate of growth in urbanization on manufacturing employment, urban amenities, urban-rural income disparity, and secondary education. The results show that only the urban amenities variable (AMEN69) has a significant positive effect on rate of change in urbanization. Equation 2 substitutes commercial employment for manufacturing employment, finding that both urban amenities (AMEN69) and commercial employment (CEMPL69) have a positive association with rate of change in urbanization. Equation 3 substitutes service employment for commercial employment, and finds again that only urban amenities (AMEN69) has a significant effect on the dependent variable. Finally, equation 4 substitutes modern-sector agricultural employment for service employment. Further, this equation drops urbanrural disparity because it is highly correlated with agricultural labor (r = .908). In place of this income disparity is the variable measuring quality of agricultural land. As before, the only variable showing a
30
Studies in Comparative International Research / Winter 1988 Table 2. Panel regression of rate of change in urbanization on independent variables Devendent Variables
IQ~_~LL~. UEB/~69 b s.e.b. B ~4PL69
URBAN79
URBAN79
URBAN79
UE,BAN79
.342 .137 .397"*
.&76 .214 .553**
.500 .104 .592***
.5A3 ,lOA .630"**
b
.00~
.
.
.
.
.
s.e.b_
.027 ,018
.
.
.
.
.
.
.
.
.
....
,311
. . . . . . . .
---....
.151 .396**
. .
B CE/IPL69
b s.e.b. B $~4PL69 b
. . . . . . - . . . . . . . . . . .
s,e.b. m
. . .
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
. .
.
. .
. .
. .
.171
.
. .
. .
. .
. .
. .
. . .... .... ....
.469 .099
AEIIPL69 b
. . . . . . . - . . . . . .
s.e.b. B Al'g~69
.
.
.
.
.
.
1.163
b
.
.
B DISP69 b
.014
.
.
.
. .
.
.
. .
.
.
.002 .030
.006
.
1.094
1.167
.542
.590
.257*
,274"
.587 .269"
.015
....
,611 .273*
s.e,b.
.
.
-.082
s.e.b.
.132
E
.011 .142 .226 .099
,131
1.108
.132
....
-,063
.012
....
-,076
.iii .229 .078
.192
SECED69 b s.e.b. $ LAND69 b
.
s.e.b.
-.
B
.
Adj.
.
.204 -.053 .
.
.
.
.
.
.
.
.
.
.
.
.
.782
R2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.813
.
.
.202
.133 .
.
.783
.039 ,05&
.
.081
.798
NoKe: b - u n s t a n d a r d i z e d regression coefficient; s.e.b. -scandard of the b; B - s t a n d a r d i z e d regression coefficient; * - P < .i0 tailed
tes~);
** - P <
.05;
*** - P <
error (t~o-
.01.
statistically significant impact is urban amenities, which has a positive relationship with rate of expansion in the urban population, g The findings in Table 2 are interesting from both a substantive and a theoretical standpoint. In terms of substance, the positive effect of commercial employment on urbanization is not surprising because the state has implemented policies that have expanded commercial opportunities in urban areas. Moreover, the consistently strong positive effect
Bradshaw
31
of urban amenities suggests that rural inhabitants do not respond simply to urban employment, but to various amenities offered by the city. Rempel (1981b:129) notes, however, that "the availability of amenities in the urban centers does not necessarily mean that the incoming migrants can afford to enjoy these amenities." Instead, amenities tend to attract rural inhabitants even though new arrivals lack the financial resources to gain access to many of these services. In terms of theory, the results indicate that modern-sector employment in commerce facilitates urban expansioin, but other types of employment (most notably in manufacturing) do not influence such growth. Because secondary education has been linked to various types of development in the Third World, it is important to examine factors that enhance education. The next equation tests the effect of urbanization, individual income, and urban educational amenities (EAMEN69) on rate of change in secondary school enrollment.9 Perhaps the most interesting variable from a theoretical perspective is urbanization, as modernization proponents argue that cities tend to increase the rate of education relative to that found in rural areas. By contrast, urban bias and dependency/world-system theorists see Third World urbanization as a degenerative process that does not promote education or meaningful employment. Results reported in Table 3 support the latter view, Table 3. Panel regressionof rate of changein secondaryschool enrollment on independentvariables De~en~nt
Ind, Vars.
SECED79
SECED69 b s.e.b. B URBAN69 b
.637 .150 .692*** -,369
s.e.b.
.093
B INCOM69 b s.e.b. B EAMEN69 b s.e.b. B
- .052 .124 -. 043
AdJ. R 2
.730
Note:
Variable
Sea n o t e
-. 669*** .353
.162 .511"*
r
Table 2.
32
Studies in Comparative International Research / Winter 1988
with urbanization (URBAN69) demonstrating a strong negative effect on rate of growth in secondary school enrollment. ~~The results also indicate a strong positive relationship between individual income (INCOM69) and rate of change in secondary schooling. This finding is quite logical because secondary school is not free in Kenya and therefore enrollment requires a certain level of family income. The third independent variable in the equation--urban educational amenities-has a nonsignificant effect on the dependent variable. The strong negative relationship between urbanization and rate of expansion in secondary school enrollment has profound implications for future development in Kenya. It suggests that the rate of secondary education will decline as cities continue to grow, thereby placing greater pressure on the government to allocate scarce resources to numerous services. Although some urban areas educate more children in absolute terms, the actual rate of growth for education will decline because cities simply cannot accommodate the increasing number of young people needing education. Moreover, increasing urbanization will create greater unemployment and underemployment, possibly reducing the flow of remittances back to rural areas. The potential seriousness of this situation can be understood fully by examining the link between education and development in Kenya. To study this issue, the analysis turns to an explicit test of variables that have an impact on rate of change in personal income. Equation 1 in Table 4 regresses rate of expansion in individual income on manufacturing employment, urban amenities, urban-rural income disparity, and secondary school enrollment. The results show that urbanrural income disparity (DISP69) has a significant negative effect on the dependent variable while secondary school enrollment (SECEN69) demonstrates a positive impact. The first finding supports urban bias arguments which assert that an income disparity in favor of urban areas will, in the long run, decrease national economic growth. The second finding supports modernization arguments which state that education is essential ifa nation is going to develop in terms of individual wealth and quality of life. Equation 2 shows that commercial employment (CEMPL69) and secondary education (SECED69) have a positive effect on rate of growth in personal income while urban-rural disparity (DISP69) continues to demonstrate a significant negative impact. Equation 3 follows the same pattern, with secondary education (SECED69) and urban-rural disparity (DISP69) showing a positive and negative effect, respectively. Finally, equation 4 eliminates urban-rural disparity from the equation because, as noted before, it is highly corre-
Bradshaw
33
Table 4. Panel regression of rate of change in personal income on independent variables Dependent Ind.
Vats.
INCOM79
INCOM69 b s .e .b, B MF21PL69 b s .e.b,
.559 .065 .733***
INCOM79
.377 .085 .495"**
Variables
INCOM79
.539 .122 .708"**
INCOM79
.537 .i18 9
.015 .013 .090
B
CEMPL69 b s.e.b,
.... ---....
B
SEMPL69 b s.e.b, B AEMPL69 b s.e.b.
.2i8
.070 93 9 2 * * *
.074 .200 .061
.... ---.... .... ----
- .022 .012 -. 129"
AMEN69
b s.e.b. B
.091
.243 .030
.314 .224 .104
-.050 .267 -.017
.158 .053 .173 *~r*
-.i15 .064 -9
-. 1 2 0 9 301 - .040
DISP69
b s.e.b.
B
- .119
.059 -.
$ECED69 b s.e.b.
129" .230 .ii0
.226**
B
Lrs~AN69 b s.e,b. B
.... ---....
Adj.
.909
R2
.263 .086 .259***
.294 .I01 .289***
9 9 93 2 0 * * * .050 90 8 7
.082 9931
,905
9
~gce: See n o t e ~o T a b l e 2 9
lated with agricultural employment. In place of this disparity is the variable measuring urbanization. Modernization scholars assert that urbanization should increase development while dependency/worldsystem and urban bias theorists claim that city growth will inhibit development. Neither view is supported here as urbanization has a nonsignificant impact on rate of growth in individual income. However, agricultural employment (AEMPL69) has a significant negative impact on this growth, and secondary education (SECED69) still has a strong positive impact. '~
34
Studies in Comparative International Research / Winter 1988
The results demonstrated in Table 4 are especially significant for two reasons. First, commercial employment is the only type of modernsector employment that demonstrates a significant effect on individual income. This finding is congruent with post-independence history in Kenya, since the state has acted to increase lucrative opportunities for Kenyan commercial interests. Results from Table 2 showed that commercial employment was also instrumental in attracting people to urban areas where much of this employment was located. Second, urbanization has an indirect negative impact on rate of increase in individual wealth. Urbanization has a negative impact on rate of expansion in secondary school enrollment (see Table 3) which, in turn, has a positive relationship with individual income (see Table 4). Will per capita income and other factors associated with "modernization" improve physical quality of life? To investigate this issue, the analysis examines the impact of individual income, secondary education, urban health amenities (HAMEN69), and urbanization on rate of change in child mortality. 12 If any indicator is negatively associated with child mortality rates, then it increases physical quality of life. Results from equation 1 in Table 5 show that secondary education (SECED69) has a significant negative effect on rate of change in child mortality while all other variables fail to demonstrate statistical significance. Because individual income has a moderate correlation with urbanization (r = .788) and secondary education ( r = .741), some might argue that multicollinearity could obscure the true impact of several coefficients in the equation. Thus, the remaining equations test the separate effects of individual income, secondary education, and urbanization after controlling for urban health amenities. Equations 2-4 again show that only secondary education (SECED69) has a statistically significant impact on rate of change in child mortality, which is used here as the indicator of physical quality of life. These findings again underscore the importance of secondary education in Kenya's development process. Not only does education enhance personal income, it also decreases child mortality rates and thereby increases physical quality of life. Unfortunately, these findings also suggest that increased urbanization will indirectly impede physical quality of life. Urbanization has a negative impact on rate of change in secondary education which, of course, is positively related to rate of change in physical quality of life. This is an ominous finding because the city population is expanding throughout Kenya.
Bradshaw
35
Table 5. Panel regression of rate of change in child mortality on independent variables De~endent Variables Ind. Vars.
MORT79
MORT79
MORT79
MORT79
MORT69 b s.e.b.
1.016 .094
1.010 .I01
.986 .090
1.034 .099
B
.879"**
.873***
.853*4*
.894***
INCOM69 b s.e.b. B
.058 .067 .181
SECED69 b s.e.b B
-.138 .055
. . . . . ....
.085 .033
.... ....
-. 325**
.
.199"*
....
HAMF2~6 9 b
s.e.b. B URBAN69
-.022 .028 -.069
. . . . . . . . .
.
.
.
.
. . .
. . .
. . .
. . .
. . .
.021
.002
.012
.003
.050 .038
.048 .004
.044 .022
.049 .005
b s.e.b.
-.003 .039
. .
. .
.022
B
-.010
. . . . . . . .
.002
.782
.777
Adj. R 2
.820
. .
. .
. .
. .
. .
. .
.819
.000
Note: See note Co Table 2.
DISCUSSION AND CONCLUSION In contrast to previous research on urbanization in Kenya, this study uses a panel regression model that examines the rate of change in particular variables between 1969 and 1979. By contrast, a simple cross-sectional design does not test change over time; further, it often makes it difficult to separate cause and effect in an equation, since results may be due to reciprocal causation. Future studies of Kenya should continue to test variables over time because urbanization and several other indicators are changing rapidly. A cross-sectional design will not capture such change. The analysis produces several important findings relevant to urbanization and development in Kenya. First, commercial employment and urban amenities are positively associated with rate of growth in the urban population. Second, urbanization has a very strong negative impact on rate of expansion in secondary school enrollment. Third,
36
Studies in Comparative International Research / Winter 1988
commercial employment and secondary education have a positive effect on rate of growth in individual income while the urban-rural income disparity and agricultural labor demonstrate a negative effect on such growth. Finally, secondary education has a consistently strong negative impact on child mortality rates, showing that education enhances physical quality of life. Overall, the results pertaining to education may be the most significant finding of the paper. These findings run counter to modernization arguments which assert that increased urbanization has a positive impact on education and other "modern" aspects of society. In contrast to such ideas, however, Kenya and many other Third World countries have experienced (and will continue to experience) a negative association between increased urbanization and rate of educational expansion. Growing cities simply cannot accommodate the rapid influx of school-age citizens, many of whom are extremely poor. Furthermore, by creating higher rates of unemployment and underemployment, rapid urbanization may decrease the amount of money earned in cities and remitted to rural areas. Fewer citizens will find meaningful employment in cities, making it more difficult to sustain themselves, much less remit money to the countryside. The adverse impact of urbanization on Third World development has long been predicted by the dependency/world-system perspective and, more recently, by the urban bias theory. The findings concerning urbanization and education are especially significant in Kenya because the present analysis shows that education has a positive effect on both personal income and physical quality of life. In conclusion, this study has extended and refined research on urbanization and development in Kenya. In addition to conducting a new quantitative analysis, this paper has explained its findings within the context of several general theories and recent Kenyan history. In fact, the results provide a unique combination of support for theories of modernization, economic dependency, and urban bias. Modernization theory is substantiated by the positive association between commercial employment and both urbanization and personal income, and by the positive effect of secondary education on both personal income and physical quality of life. Urban bias and dependency/world-system arguments are supported (and modernization theory is contradicted) by the negative relationship between urbanization and secondary education. Moreover, the urban bias perspective is further supported by the negative impact of the urban-rural income disparity on personal income. This theoretical discussion should facilitate comparison with
Bradshaw
37
other underdeveloped countries currently facing increased urbanization. Future studies of Kenya and other Third World countries should examine a variety of development issues by combining theoretical, historical, and quantitative research.
APPENDIX Variable Urbanization, 1969 Urbanization, 1979 All four employment variables, 1969 Secondary education, 1969 Secondary education, 1979 Amenity index, 1969 Rural-urban income disparity, 1969 Quality of agricultural land, 1969 Personal income, 1969 Personal income, 1979 Child mortality, 1969 Child mortality, 1979
Source Republic of Kenya, 1970-1977 Republic of K~nya, 198i Republic of Kenya, 1972 Republic of Kenya, 1970-1977 Republic of Kenya, 1981 Rempel, 1981b Republic of Kenya, 1972 Republic of Kenya, 1971 Republic of I~nya, 1972 Republic of Kenya, 1984b Calculated from child survival data in Republic of Kenya, 1970-1977 Calculated from child survival data in Republic of Kenya, 1981
NOTES I would like to thank Janet Abu-Lughod, Jack Goldstone, Charles Ragin, and an anonymous SCID referee for very helpful comments on previous drafts ofthis paper, I am also grateful to the Institute for Development Studies at the University of Nairobi for providing me with helpful information while I conducted field research in Kenya. I. Modernization theory usually focuses on manufacturing employment as a modernizing agent. In Kenya, however, local citizens often are denied access to the manufacturing sector because this area is largely controlled by foreign capital. 2. It is clear that urban population growth is primarily attributable to townward migration, not to natural increases in the city population. Urban population growth has continued to exceed overall population growth in Kenya (Republic of Kenya, 1981). 3. All districts of Kenya have maintained the same boundaries between 1969-1979 and therefore geographic inconsistency is not a problem. 4. The nine districts include Busia, Elgeyo Marakwet, Gadssa, Kajiado, Kirinyaga, Mandera, Siaya, Wajir, and West Pokot. It should also be noted that a few small- and intermediatesized urban areas extended their boundaries between 1969 and 1979. To ensure that this extension did not alter the results, the analysis was rerun after eliminating districts that contained these cases. The results from the two samples did not differ appreciably. 5. Some may criticize the income disparity indicator because it does not take into account subsistence farmers outside the wage-earning (modern) sector. Unfortunately, this criticism cannot be avoided because district-level data on subsistence agriculture are unavailable for the years under study. Although this disparity does not measure subsistence incomes, it does, however, provide a good representation of rural incomes. Agricultural wage laborers are not wealthy members of the modern sector. Instead, they often are landle~ peasants who avoid squatter settlements by accepting wage labor on large farms or plantations (House and Killick, 1981). Surveys of income distribution in Kenya place formal-sector agricultural workers toward the bottom of the income scale, even lower than some smallholders (see International Labor Office, 1972:74).
38
Studies in Comparative International Research / Winter 1988
6. Ideally, an additional variable would measure foreign investment in each district of Kenya. Unfortunately, foreign investment data are not available at the district level. 7. With only one exception (noted below), multicollinearity is not a problem in the analysis. There is only one occasion where two explanatory variables are correlated at a level greater than .80. This equation is rerun to ensure that multicollinearity does not obscure the true effect of certain variables. 8. In an earlier analysis, the employment variables (MEMPL69, CEMPL69, SEMPL69, and AEMPL69) were replaced by variables measuring per capita earnings in these sectors of the economy. In other words, per capita earnings in manufacturing was substituted for employment in manufacturing and so on. The only earnings variable to show a statistically significant effect was commercial earnings which had a strong positive impact on rate of'change in urbanization. This finding lends further support to the argument that the commerdal sector draws people to urban areas. Moreover, the only other independent variable to demonstrate a statistically significant effect in all four earnings equations was urban amenities which, again, had a positive effect on rate of change in urbanization. 9. Urban educational amenities (EAMEN69) represents one ofthe ten indicators that comprise the amenity index in 1969 (AMEN69). 10. This finding is remarkably strong and consistent. It obtains regardless of what variables are included in the equation. 11. As before, the employment variables were replaced by variables representing per capita earnings in these sectors of the economy. Again, the only earnings variable to show a statistically significant effect was commercial earnings, which had a strong positive association with rate of change in personal income. Further, like in Table 4, secondary education had a significant positive effect, while urban-rural disparity and agricultural employment demonstrated a significant negative impact. 12. Urban health amenities (HAMEN69) represents one of the ten indicators that comprise the amenity index in 1969 (AMEN69). REFERENCES Anker, Richard and James C. Knowles 1983 Population Growth, EmploymenL and Economic-Demographic Interactions in Kenya: Bachue-Kenya. New York: St. Martin's. Berliner, Joseph 1977 "Internal Migration: A Comparative Disciplinary ViewY In Alan Brown and Egon Neuberger (eds.), Internal Migration: A Comparative Perspective. New York: Academic. Bigsten, Arne 1984 Education and Income Determination in Kenya. Aldershot, Hampshire, England: Gower. Bornschier, Volker and Christopher Chase-Dunn 1985 Transnational Corporations and Underdevelopment. New York: Praeger. Bradshaw, York W. 1985 "Overurbanization and Underdevelopment in Sub-Saharan Africa: A Cross-National Study." Studies in Comparative International Development 20(3):74-101. 1987 "Urbanization and Underdevelopment: A Global Study of Modernization, Urban Bias, and Economic DependencyY American Sociological Review 52:224-39. Brass, William and Ansley J. Coale 1968 "Methods of Analysis and Estimation7 In William Brass et al., The Demography of Tropical Africa. Princeton: Princeton University Press. Chege, Michael 1981 "A Tale of Two Slums: Electoral Politics in Mathare and Dagnretti7 Review of African Pofitical Economy 20:74-88. Court, David 1984 ~The Educational System as a Response to Inequality." In Joel D. Barkan (ed)., Pofitics and Pubfic Policy in Kenya and Tanzania. Rev. ed. New York: Praeger. Delacroix, Jacques and Charles Ragin 1978 "Modernizing Institutions, Mobilization, and Third World Development: A Cross-National StudyY Amcrican Journal of Sociology 84:123-50.
Bradshaw
39
Evans, Peter and Michael Timberlake 1980 "Dependence, Inequality and the Growth of the Tertiary: A Comparative Analysis of Less Developed Countries7 American Sociological Review 45:531-52. Hanushek, Eric A. and John E. Jackson 1977 Statistical Methods for Social Scientists. New York: Academic. Hay, Richard Jr. 1977 "Patterns of Urbanization and Socio-Economic Development in the Third World: An Overview." In Janet Abu-Lughod and Richard Hay Jr. (eds.), Third World Urbanization. New York: Methuen. House, William J. and Tony Kiilick 1981 "Inequality and Poverty in the Rural Economy, and the Influence of Some Aspects of Policy." In Tony Killick (ed.), Papers on the Kenyan Economy: Performance, Problems and Policies. Nairobi: Heinemann. lnkeles, Alex and David H. Smith 1974 Becoming Modern: Individual Change in Six Developing Countries. Cambridge: Harvard University Press. International Labor Office 1972 Employment, Incomes and Equality: A Strategy for Increasing Productive Employment in Kenya. Geneva, Switzerland: International Labor Office. Kelley, Allen C. and Jeffrey G. Williamson 1984 What Drives Third World City Growth? A Dynamic General Equilibrium Approach. Princeton: Princeton University Press. Kentor, Jeffrey 1981 "Structural Determinants of Peripheral Urbanization: The Effects of International Dependence." American Sociological Review 46:201-I I. Lamb, David 1985 The Africans, Rev. ed. New York: Vintage. Ledogar, Robert 1975 Hungry for Profits: U.S. Food and Drug Multinationals in Latin America. New York: IDOC/North America. Leys, Colin 1975 Underdevelopment in Kenya: The Political Economy of Neo-Colonialism, 1964-1971. London: Heinemann. Lipton, Michael 1977 Why Poor People Stay Poor. A Study of Urban Bias in World Development. Cambridge: Harvard University Press. 1984 "Urban Bias Revisited." Journal of Development Studies 20(3):139-66. London, Bruce 1987 "Structural Determinants of Third World Urban Change: An Ecological and Political Economic Analysis." American Sociological Review 52:28-43. Ominde, S.H. 1984 "population Distribution and Urbanization." In S.H. Ominde (ed.~ Population and Development in Kenya. Nairobi: Heinemann. Portes, Alejandro 1985 "The Informal Sector and the World-Economy: Notes on the Structure of Subsidized Labor." In Michael Timberlake (ed.), Urbanization in the World-Economy. Orlando: Academic. Portes, Alejandro and John Walton 1981 Labor, Class, and the International System. New York: Academic. Rempel, Henry 1979 Labor Migration Into Urban Centres and Urban Unemployment in KenyaY In Dharham Ghai and Martin Godfrey (eds.), Essays on Employment in Kenya. Nairobi: Kenya Literature Bureau. 1981a "The Extent and Nature of the Population Movement Into Towns: In R.A. Obudho (ed.), Urbanization and Development Planning in Kenya. Nairobi: Kenya Literature Bureau. 198 Ib Rural-Urban Labor Migration and Urban Unemployment in Kenya. Laxenburg, Austria: International Institute for Applied Systems Analysis.
40
Studies in Comparative International Research / Winter 1988
Republic of Kenya 1965 Development Plan 1965-1970. Nairobi: Government Printer. 1970 Development Plan 1970-1974. Nairobi: Government Printer. 1970-77 Kenya Population Census 1969 (4 Vols.). Nalrobi: Government Printer. 1971 Statistical Abstract 1971. Nalrobi: Government Printer. 1972 Employment and Earnings in the Modern Sector 1968-1970. Nairobi: Government Printer. 1981 Kenya Population Census 1979 (2 Vols.g Nairobi: Government Printer. 1984a Development Plan 1984-1988. Nalrobi: Government Printer. 1984b Statistical Abstract 1984. Nalrobi: Government Printer. 1985 Economic Survey 1985. Nairobi: Government Printer. Rogers, Andrei and Jeffrey Williamson 1982 "Migration, Urbanization, and Third World Development: An Overview." Economic Development and Cultural Change 30:463-82. Rubinson, Richard and Deborah Holtzman 1981 "Comparative Dependence and Economic DevelopmentY International Journal of Comparative Sociology 22:86-101. Stark, Oded 1982 "Research on Rural-to-Urban Migration in LDCs: The Confusion Frontier and Why We Should Pause to Rethink AfreshY World Development 10:63-70. Stren, Richard 1984 "Urban Policy."In Joel D. Barkan (ed.),Politicsand Public Policy in Kenya and Tanzania. Rev. ed. New York: Praeger. Swalnson, Nicola 1977 "The Rise of a National Bourgeoisie in Kenya." Review of African Political Economy 8:39-55. 1978 "State and Economy in post-Colonial Kenya, 1963-1978." Canadian Journal of African Studies 12:357-81. 1980 The Development of Corporate Capitalism in Kenya, 1918-1977. London: Heinemann. Thomas, Barbara P. 1985 Politics, Participation, and Poverty: Development Through Self-Help in Kenya. Boulder and London: Westview. Timbedake, Michael and Jeffrey Kentor 1983 "Economic Dependence, Overurbanization, and Economic Growth: A Study of Less Developed Countries." Sociological Quarterly 24:489-507. Timberlake, Michael and James Lunday 1985 ~Labor Force Structure in the Zones of the World-Economy, 1950-1970." In Michael Timberlake {ed.), Urbanization in the World Economy. Orlando: Academic. Todaro, Michael P. 198l ~Education and National Economic Development in KenyaY In Tony Killick (ed.), Papers on the Kenyan Economy: Performance, Problems, and Policies. Nairobi: Heinemann. 1985 Economic Development in the Third World. 3rd ed. New York: Longman. United Nations 1967 Methods of Estimating Basic Demographic Measures from Incomplete Data. Manual IV, Population Studies No. 42. New York: United Nations. Walton, John 1977 "Accumulation and Comparative Urban Systems: Theory and Some Tentative Contrasts of Latin America and Africa." Comparative Urban Research 5( I ):5-18. Werlin, Herbert H. i 981 "The Hawkers of Nairobi: The Politics of the Informal Sector." In R.A. Obudho (ed.), Urbanization and Development Planning in Kenya. Nairobi: Kenya Literature Bureau.