Evidence from Migration Data

Viewer
Transcript

Economica (2005) 72, 39–67

Why Do the Young and Educated in LDCs Concentrate in Large Cities? Evidence from Migration Data By BARRY MCCORMICK and JACKLINE WAHBA University of Southampton Final version received 4 November 2003. Do the young and educated in LDCs have a greater preference to locate in big cities? If so, this may help to explain how cities spatially concentrate the educated and young, and why the rising share of these workers in many LDCs may contribute to city growth. This paper explores migration ﬂows into and out of Egypt’s three largest cities. We study whether the higher shares of such workers in cities arise because these workers perceive relatively greater beneﬁts from living in cities, given relative urban/rural wage rates, or because the relative demand for these workers rises with city size.

INTRODUCTION Explanations of LDC city growth and the changing composition of LDC city workforces place considerable emphasis on models of non-competitive wagesetting in which artiﬁcially high urban wages induce low-skill workers to migrate to low-productivity city jobs and unemployment.1 This approach helps to explain why cities are sometimes surrounded by ‘shanty towns’, whose inhabitants might be more productive in the rural sector. A different emphasis to understanding LDC city workforce growth and composition is provided by Henderson (1986), who seeks to explain the positive correlation between educational attainment and city size in Brazil. Henderson ﬁnds that the demand for city amenities may be greater from educated workers, and also that the relative demand for educated workers may rise with city size. Thus, as a result of both demand and supply inﬂuences, educated labour is conjectured by Henderson to be disproportionately drawn into cities. Recent empirical analysis of the role of human capital in city formation suggests that educated workers may particularly beneﬁt from skill spillovers in urban areas, and that a greater share of educated workers in cities may accelerate human capital accumulation, thereby increasing productivity growth (e.g. Glaeser 1999; Glaeser and Mare´ 2001).2 Thus, if the educated are particularly inclined to migrate to citiesFas Henderson conjecturesFurban population growth not only may provide a one-off productivity gain, but also may have a dynamic inﬂuence on productivity as a result of increased skill spillovers. This may induce a virtuous cycle of urban and productivity growth. Although the LDC migration literature is very large, there exist no microlevel studies comparing the structure of migration into and out of a major LDC city with, in particular, Henderson’s hypothesis of the inﬂuence of location preference on migration ﬂows and city educational composition. Instead, the literature on migration in LDCs is focused, as the survey by Lucas (1997) observes, on rural–urban and interregional migration. While it is recognized r The London School of Economics and Political Science 2005

40

[FEBRUARY

ECONOMICA

that migration to and from cities, as opposed to other urban areas, is likely to have distinctive characteristics (e.g. Mazumdar 1987), apart from evidence that migrants into cities are young,3 little is known about whether such migration contributes towards altering the composition of a city’s labour force, rather than merely the scale.4 There are several studies of the structure of migration into large cities in developing countriesFfor example, Fuller et al. (1985) and Mazumdar (1981) study Bangkok and Kuala Lumpur, respectivelyFand at least one study of migration from a large city: that by Pessino (1991), who studies Lima. However, there do not appear to be any studies using individual data to test hypotheses concerning the comparative structure of ﬂows into and out of an LDC city. The primary purpose of this paper is to explore Henderson’s hypothesis in an analysis of these migration ﬂows for Cairo, Giza and Alexandria, the three largest cities in Egypt. In Section I we draw together various hypotheses concerning the inﬂuence of education and age on migratory ﬂows between cities and both rural and urban areas. Human capital models of how cities draw certain types of worker into close proximity may be categorized into those theories arising from labour demand reasons and those that arise from the consumption or investment beneﬁts that workers derive from proximate location. Sections II and III examine these hypotheses using Egyptian data, exploring the supply-side arguments that assert that the young and educated are attracted into cities. To examine the potential contribution of supply-induced migration ﬂows to labour force change in a city, we explore the relative migration rates of young and educated workers into and out of Cairo, Alexandria and Giza, holding constant age- and education-speciﬁc relative wage levels. In this way we explore the labour supply inﬂuences that arise from individual age, education and other effects separately from the demand effects that inﬂuence relative wages. Since we are concerned with large cities, and most studies of migration concern rural–urban ﬂows, we explore a three-way distinction between large cities, other urban areas (small cities and towns), and rural areas. This enables us to examine separately hypotheses that distinguish between (a) rural and (b) urban migration into and out of large cities. Consequently, in Section III our econometric work explores, separately, ﬂows between the sprawling urban conurbation in northern Egypt, comprising Cairo, Giza and Alexandria, and (a) rural areas and (b) less dense urban areas. Section IV provides a brief summary of the ﬁndings, and comments on the extent to which the rising proportions of educated and young workers in LDC workforces may help to explain the growth of large LDC cities, and to complement the inﬂuence of high-wage public-sector jobs discussed in McCormick and Wahba (2003).5

I. THE STRUCTURE

OF

MIGRATION

AND

LARGE CITIES

Hypothesis 1. Large cities attract educated workers from other urban and rural areas. Beginning no later than Marshall (1890; cited in Glaeser 1999), economists have argued that large cities attract educated labour out of both rural areas and smaller cities.6 Several reasons have been given to explain why this might r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

41

occur. First, on the demand side it is hypothesized that ﬁrms with high-skill requirements beneﬁt most from the efﬁciency gains resulting from high-density locations. These efﬁciency gains for high-skill ﬁrms may arise in terms of the lower costs incurred in matching heterogeneous skilled workers to the tasks to be done, and the more rapid ﬂow of information about competitor products and production methods (Lucas 1997; Rauch 1993). High-skill ﬁrms are also frequently in the service sector, where face-to-face customer contact is of particular value, and is inexpensive to provide at a city centre. Large cities generally possess a broader skill range and so can offer a more diverse service infrastructure than other urban areas, and this may be of particular value to high-skill technology industries. Another reason why the demand for educated workers tends to be greater in big cities is the political economy argument given by Ades and Glaeser (1995), who show that non-democratic regimes tend to locate the bulk of public jobs, which are ﬁlled mainly by educated workers, in big cities, where rents from public employment can be more easily appropriated, in order to reinforce political support for their regime. Second, supply-side arguments also suggest that as a city becomes larger it will become increasingly attractive to skilled workers. High-skill workers are thought to be attracted to large cities because of the more extensive consumption amenities and servicesFsometimes called the ‘bright lights’ hypothesis (e.g. Henderson 1986). The relevant amenity could also be the high ratio of skilled to unskilled workers. Differences in skill composition across cities have various further implications. For example, if large cities have a relatively educated voting population, this may enhance the quality of local public decision-making and the local quality of life. Glaeser (1999) discusses a second supply-side reason why skilled individuals are attracted to large cities: dense urban agglomerations provide spillover learning opportunities. Thus, if skilled workers reap the highest returns from these learning opportunities, they are more likely to move to large cities as a form of human capital investment.7 Both of these supply-side arguments suggest that workers are willing to accept a city job at a lower wage to other jobs, all else equal. To structure the discussion of our testing of the ‘bright lights’ hypothesis, we summarize these ideas in the following simple model. Consider the familiar trade model with two goods, two factors of production and factor price ﬂexibility. Assume that good F (food) is produced only in a separate (rural) region, while good M (manufacturing) is produced in the urban region. The two factors are educated and uneducated labour, and are mobile between the industries and these regions. F is intensive in uneducated labour and M is intensive in educated labour. Also, assume that urban areas offer a positive amenity value for educated workers. This will reduce the supply price of educated labour to the city below the opportunity cost, given by the rural sector marginal productivity of educated labour. Educated labour migrates to the urban M sector until the urban–rural wage differential is just offset by the urban amenity value. In equilibrium this increases the share of educated labour in urban areas to above that reﬂecting only the effect of a higher relative demand for educated labour in the city. Our concern in the empirical section is to provide evidence from migration data for this amenity effect. Using US Census Data for 1990, Glaeser (1999) and Glaeser and Mare´ (2001) show that larger cities have a greater share of high-skilled persons than smaller cities. r The London School of Economics and Political Science 2005

42

ECONOMICA

[FEBRUARY

Henderson (1988) shows that the correlation between the ratio of high to lowskill population and city size in Brazil is 0.69. What are the implications of the hypothesized city amenity effect for migration? First, it is important to recognize that the demand and supply effects that we have discussed do not imply that net migration of educated workers into cities will necessarily be observed: in equilibrium, the employment share of high-skill workers and relative city wages would be expected to adjust to eliminate net migration ﬂows that follow shocks to demand and supply; but this will be in evidence only until compensating adjustments occur in wages, unemployment, amenities, etc., to re-establish an equilibrium. Thus, the supplyside arguments imply that in competitive markets, if within-skill wage rates are equal over space, skilled workers will be more likely than unskilled workers to migrate into a city in order to beneﬁt from either consumer amenities or the greater learning opportunities that city work offers the skilled.8 By itself, however, this empirical prediction is not a very powerful test of the supply-side explanations of city size: there are other reasons, unrelated to city amenities, why, all else equal, the more educated have higher migration rates into a city (see e.g. Schwartz 1976). However, these other arguments apply to the inﬂuence of education on migration to all destinations; in Section III, therefore, we test the ‘bright lights’ supply-side hypothesis by comparing the inﬂuence of education on migration from (a) urban, and (b) rural areas into a large city with that of education on migration from a large city to these areas. We shall reject the hypothesis that the greater preference of the educated for city locations has raised the share of educated employment in cities if increased education fails to have a signiﬁcantly greater effect on the probability of migration to the city, all else equal, than it has on migration from the city to (a) urban, or (b) rural areas. Hypothesis 2. Large cities achieve a large share of young workers by (a) attracting the net in-migration of young persons and (b) encouraging the net outmigration of older individuals. It is widely recognized that the young comprise a disproportionate share of those moving to cities (e.g. Todaro 1976), but only recently have economists considered why migration might reduce the mean age of city workers below that obtaining elsewhere. Young persons have long been recognized as being more willing to invest in learning, given that they can expect more years over which to realize the beneﬁts. If cities offer learning spillover beneﬁts from living near large numbers of skilled workers, young people will become net migrants to cities, while older workers with fewer years to enjoy the learning externalities from cities, and similar costs of living in built-up areas, will become net migrants from cities. Moreover, if young workers seek to experience a range of jobs, then, since cities spatially compress jobs, sampling need not require commuting or migration costs. Glaeser (1999), using Census data on stocks of employment by US city size, shows that in the United States younger persons have a greater propensity to live in cities, especially large ones.9 As with Hypothesis 1, the force of this argument would appear stronger for those migrating from rural areas to cities than for urban migrants to larger cities, since rural areas are less likely than small urban centres to provide learning opportunities for young people. r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

43

The higher price of land in cities may also explain the lower mean age in cities. Older workers, perhaps with a pension income or other ﬁnancial wealth, usually choose to work fewer hours. Since income from savings is the same at all household locations but the cost of living is not, workers supplying fewer hours will have a lower beneﬁt from, and thus lower demand for, the greater employment opportunities at a high-cost city location. This mechanism would provide a second explanation of how migration enables the share of the young to rise in cities, in order to allocate scarce city accommodation to those working more hours. As with education, there are several reasons why young workers are more likely to migrate that are unconnected with spatial density. Hence we adopt an econometric strategy similar to that outlined for education in order to test whether the inﬂuence of being young on the probability of migration to a big city exceeds that of the inﬂuence of being young on migration from a big city, ceteris paribus.

II. THE EGYPTIAN CONTEXT

AND

DATA

Background to city growth in Egypt With approximately 45% of the population living in towns and cities, mostly near the Nile, Egypt is a comparatively urbanized LDC. Although Cairo continues to dominate the urban population, small and medium-sized towns have grown strongly during the last three decades.10 These towns provide various service activitiesFadministration, education and healthFas well as manufacturing centres. In contrast, the 1986 and 1996 Census evidence shows that the population of the three largest cities, Cairo, Giza and Alexandria, has increased in absolute terms from 11.14 to 12.71 million, while declining as a proportion of both the total and the urban populations.11 Also, the 1986 and 1996 Censuses show that the proportions of the population between 15 and 64 years of age have increased in the largest cities, while the youngest age group, those less than 6 years old, have fallen (see Table 1). Thus, although rural areas have higher fertility rates, there is still evidence that the three largest cities have on average higher shares of the working-age population. More importantly, Cairo, Alexandria and Giza have experienced an increase in the proportions of their educated population holding university degrees (Table 1). In addition, these three cities have seen an increase in the proportion of professional and technical staff in the labour force over the last two decades. Thus, there is evidence to support the view that the largest cities in Egypt have experienced an increase in the shares of educated workers. While the data studied here conﬁrm the low net migration to Cairo/Alexandria in the 1980s and 1990s, we shall explore how far migration is changing the composition of the labour force in these cities. The data The analysis is based on two data-sets: the 1988 Egyptian Labour Force Sample Survey (1988 LFSS) and the 1998 Egypt Labour Market Survey (1998 ELMS) which provide detailed information on approximately 20,000 r The London School of Economics and Political Science 2005

44

[FEBRUARY

ECONOMICA

TABLE 1

Population Composition of Big Cities 1986 o 6 years

1996 15–59 years

Share of the population by age group (%) Cairo 12.2 59.1 Alexandria 12.5 59.5 Giza 14.5 56.2 All Urban 13.5 57.3 Rural 16.7 42.7 Total 15.2 51.7 1986

o 6 years

15–59 years

11.0 11.9 12.6 12.6 17.0 15.1

66.7 65.0 63.1 63.7 57.0 59.9 1996

Share of the population with university degree (%) Cairo 7.3 Alexandria 5.4 Giza 6.8 All Urban 5.6 Rural 1.0 Total 3.1

13.4 9.5 12.4 10.1 2.3 5.8

Professional/technical staff as % of the labour force Cairo 18.8 Alexandria 15.9 Giza 13.9 All Urban 17.7 Rural 6.1 Total 11.8

31.6 25.9 24.4 29.6 13.9 21.2

Source: Population Census, 1986 and 1996.

individuals. (See Appendix for details on the data-sets.) The 1988 LFSS reports residential and work locations in October 1988 (the time of the survey) and retrospective information concerning October 1981, while the 1998 ELMS reports locations in November 1998 and retrospective information concerning August 1990.12 We examine migration to and from the biggest three cities in Egypt: Cairo, Giza and Alexandria. These three cities account for around half of the total urban population and are located within around 200 kilometres of each other in northern Egypt. The analysis is conﬁned to male employed workers who are aged 15–64 years.13 The total sample size used is 7783 male employed workers,14 of whom 208 are migrants.15 Our sample comprises: (i) 2658 rural workers, of whom 50 have moved from rural areas to Cairo, Giza or Alexandria (CGA);16 (ii) 2785 urban workers (other urban areas excluding CGA), of whom 38 have moved from other urban areas to CGA; and (iii) 2340 CGA workers, of whom 82 have moved out of these three big cities and 38 have moved between these large cities. Table 2 displays the distribution of the working population between large cities (CGA), urban areas (small cities and towns) and rural areas in Egypt, by level of education (secondary and university degree-holders), and by age category (ages 15–34, 35–54 and 55–64).17 Consistent with other studies, large r The London School of Economics and Political Science 2005

r The London School of Economics and Political Science 2005

38 49 13 2340

19 28 19 –

Secondary/higherb (23.2% of area total)

Large citiesa

37 49 14 2785

Total sample

b

18 25 13 –

Secondary/higherb (20.6% of area total)

Other urban areas

Cairo, Giza and Alexandria. Percentage of area population with secondary or higher education in given age category.

a

% aged 15–34 35–54 55–64 Sample size

Total sample

TABLE 2

47 41 13 2658

Total sample

Distribution of the Working Male Population by Age and Education

7 7 2 –

Secondary/higherb (6.1% of area total)

Rural

2005] LDC CITY WORKFORCE GROWTH AND COMPOSITION

45

46

ECONOMICA

[FEBRUARY

cities have a higher share of educated workers than other urban areas: 23.2% of the large cities working male population have secondary or higher level of education, compared with 20.6% in other urban areas and 6.1% in rural areas. Furthermore, 52% of the labour force with secondary or higher education (at least 12 years of schooling) works in the ‘large cities’FCairo, Giza or Alexandria. However, it is striking that, of those aged under 35, the proportion of educated workers in the large cities is only slightly greater (by 1 percentage point) than in all other urban areas. The proportion of young (15–34) in big cities is 1 percentage point greater than in other urban areas and 9 percentage points less than in rural areas, where fertility is high. The proportion of older workers (55–64) in the cities (13%) is slightly less than in other urban areas (14%), and Table 3 also shows that the mean age of the workforce is marginally lower in the large cities than in other urban areas. Overall, the descriptive evidence provides only slight support for the view that large cities have higher shares of young workers, but stronger evidence that the large cities have a larger educated labour force than other urban or rural areas. The ﬂow of migrants in our sample between the rural, urban and large city areas is given in Table 3. There is an overall net inﬂow of both illiterate and well educated rural migrants into the large cities, as well as a net outﬂow from large cities to other urban areas: there are 38 migrants into large cities from urban areas, and 55 migrants from large cities to other urban areas.18 Given that the urban–city ﬂows of educated workers are about equal in each direction, it follows that the net ﬂow from large cities to other urban areas comprises a net ﬂow of illiterates. Thus, perhaps surprisingly, for ﬂows between big cities and other urban areas, ongoing migration contributes to the higher mean educational levels in the large cities by ‘draining’ less educated workers from the large cities, rather than by ‘draining’ highly educated workers from other urban areas. The large cities appear to be exporting workers to manufacturing jobs rather than attracting them to such jobs from other urban areas, which is consistent with manufacturing growth sectors being based outside CGA. The educational composition of migration from rural areas to large cities is very similar to that of migration from large cities to rural areas (Table 3, columns (2) and (10)). However, since the scale of rural migration to big cities is about double that from big cities, migration ﬂows raise educational levels in the cities only if migrants are more educated than the mean city levels. Since the education levels of rural–city migrants are marginally lower than city inhabitants (the ‘stayers’), we cannot conclude that rural migration is directly contributing to the comparatively high educational levels that we have found in CGA. Overall, the only mechanism whereby migration directly helps to raise educational levels in large cities is through net outﬂows of less educated workers to other urban areas. Tables 3 and 4 provide descriptive evidence concerning the contribution of migration selectivity to the lower mean age in large cities. Table 4 shows that 39% of migrants to large cities are under 27 years of age, while less than 21% of migrants from large cities are under 27 years. In other words, while migration is concentrated among the young, this is particularly so for migrants to CGA, and less so for migrants out of these densely populated areas. The r The London School of Economics and Political Science 2005

TABLE 3 Urban

Large cities

r The London School of Economics and Political Science 2005

20.21 62.45 8.69 3.32 25.54

14.89

43.43 19.15 14.89 22.53

10.67 21.53 4.46 63.34

50.74

26.45 53.32 20.23

39.7

(4)

59.07 – 40.98 – – 9.57 – 51.71 – 38.72 – – – – – – 205.86 233.60 2658 2747

55.43 38.43 6.41

21.79 51.06 19.15

46.81 53.19 – – – – – – 170.63 50

37.1

(3)

29.2

(2)

– – 13.51 62.16 24.32 – – – 189.30 38

5.26 23.68 18.42 52.64

55.26

8.11 45.95 45.95

33.8

(5)

– – 9.62 51.86 38.52 – – – 232.98 2785

10.59 21.56 4.66 63.19

50.81

26.11 53.22 20.59

39.6

(6)

– – – – – 57.53 14.46 28.01 – 2220

3.50 31.45 8.45 56.61

53.95

22.72 55.61 22.67

39.6

(7)

– – – – – 34.21 60.53 5.26 – 38

– 26.33 13.16 60.51

31.58

31.58 31.58 36.84

35.1

(8)

– – – – – 70.37 11.11 18.52 176.83 55

3.64 34.55 18.18 43.63

54.55

21.82 40.00 38.18

33.3

(9)

– – – – – 61.54 19.23 19.23 172.38 27

3.70 33.33 14.81 48.16

55.56

26.92 53.85 19.23

36.5

(10)

– – – – – 57.49 15.20 27.31 – 2340

3.41 31.48 8.84 56.27

53.64

21.94 54.83 23.23

39.4

(11)

LDC CITY WORKFORCE GROWTH AND COMPOSITION

a Average distance is in kilometres between the main city in the governorate of origin and Cairo. For migrants from big cities, distance is measured between Cairo and the main city in the governorate of destination.

Mean age in reference year 37.2 Educational level (%) Illiterate 55.90 Less than secondary 38.20 Secondary/higher 5.90 Origin sector (%) Public 20.30 Origin industry (%) Agriculture 62.84 Manufacturing 8.50 Construction 3.10 Services 25.56 Origin region (%) Rural lower 59.30 Rural upper 44.70 Canal cities – Urban lower – Urban upper – Cairo – Giza – Alexandria – 206.50 Average distancea Sample size 2608

(1)

Rural Migrants to Total Urban Migrants to Total Large cities Migrants between Migrants Migrants Total stayers large cities sample stayers large cities sample stayers large cities to urban to rural sample

Rural

The Structure of Migration into and out of Large Cities

2005] 47

48

[FEBRUARY

ECONOMICA

TABLE 4

Age Composition of Migrants to and from Large Cities (%) Rural and other urban Age 15–21 22–26 27–34 35–44 45–54 55–64

Large cities

Migrants to large cities

Total sample

Migrants from large cities

Total sample

12.94 25.88 27.07 20.00 11.76 2.35

10.56 8.26 25.18 24.34 18.80 12.86

9.92 10.74 38.02 23.97 9.09 8.26

7.06 8.08 23.96 26.64 21.65 12.61

mean age of the migrant categories are given in the ﬁrst row of Table 3, and suggest that migration contributes to reducing the mean age of those living in Egypt’s largest cities. However, this effect comes almost entirely from migration to and from rural areas, rather than to and from less dense urban areas, with the mean age of rural migrants to the city over seven years below that of migrants from the city to rural areas. It is noteworthy that rural migrants to CGA are disproportionately drawn from Upper Egypt (53%)F the SouthFwhereas urban migrants to CGA are from Lower Egypt and the Suez Canal area (62% þ 14%). While these gross ﬂow data may suggest that the educated and young accumulate in CGA, we now explore how far this reﬂects relative wages and unemployment, the intervening effect of location, public-sector employment or the quality of public services, rather than the underlying greater preference, or non-wage advantages, of the educated and young to be in large cities.

III. HOW DOES MIGRATION ALTER THE LABOUR SUPPLY IN LARGE CITIES? ECONOMETRIC EVIDENCE The evidence described in Section II gives qualiﬁed support for the view that structure of migration between the Cairo/Giza/Alexandria (CGA) conurbation and other areas helps enable LDC large cities to bring into close proximity those groups most likely to productively beneﬁtFthe young and educatedFand that net migration to these cities does not merely increase the scale of the city labour force. In this section we explore migration probabilities rather than ﬂows, allowing for various other inﬂuences on migration into and out of CGA, and investigate the hypotheses described in Section I. Since our arguments distinguish CGA from both other urban and rural areas, the econometric analysis ﬁrst estimates models of the probability of individual migration into CGA from both rural and other urban areas. We then estimate a model of an individual in CGA choosing between migrating to a rural area, to another segment of CGA, to an urban area beyond CGA, or not migrating. By contrasting the parameter estimates of these models of migration into and out of CGA, we are able to test the implications for migration of the various hypotheses. r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

49

To implement this, in each location choice model we assume that the indirect utility of location choice j for individual i can be written Uij ¼ bj Xi þ eij ; where Uij is the utility from choice j by individual i, bj is a vector of parameters, which may vary between location choices, and Xi is a vector of attributes of individual i, which includes characteristics of the origin region in which the individual was working. The regional characteristics included are the relative wage and unemployment rates, and measures of the region’s distance from Cairo. If the errors eij are independently and identically distributed with the type I extreme value distribution, this gives the multinomial logit model. The probability of individual i choosing location j is given by 0

0

Pij ¼ expðbj Xi Þ=ð1 þ Sj expðbj Xi ÞÞ: Thus, we begin by estimating two binomial logit models of the probabilities of migrating (i) from rural areas to Cairo/Giza/Alexandria, and (ii) from urban areas to Cairo/Giza/Alexandria. Next, we study the determinants of migration choice for individuals previously working in CGA, using a multinomial model ( j ¼ 0, 1, 2 and 3). Let j be the regional location in the survey period (1988 or 1998). For CGA workers in the reference year (1981 or 1990), j ¼ 0 is staying in the origin part of CGA; j ¼ 1 is migrating to another part of CGA; j ¼ 2 is migrating to another urban area and j ¼ 3 is migrating to a rural area. To estimate the preceding models, we combine two data-sets, the 1988 LFSS and the 1998 ELMS, to study migration over 1981–88 and 1990–98, respectively.19 Since these two periods are seven and eight years, respectively, we include a dummy variable in the migration models to allow for this implied higher probability of migration in the second sample.20 The data used are described in Section II and the Appendix. We now discuss the testing of the hypotheses outlined in Section I and the various control variables used in the analysisFsee Appendix Table A1 for a summary of the deﬁnition of variables. Education. Evidence from both LDCs and DCs (e.g. Mazumdar 1987; Lucas 1997) shows that, both before and after controlling for other inﬂuences, and before and after the inclusion of other explanatory variables, the propensity to migrate is generally higher for the better educated. There are several possible explanations for this. First, formal qualiﬁcations may reduce the uncertainty associated with migration by contributing a person’s ability to collect and process information. Also, the higher migration rate of the educated may partly reﬂect their responsiveness to larger spatial wage differentials than exist for the uneducated (see Fields 1982; Schultz 1982), as well as the higher growth rate of jobs requiring educational qualiﬁcations in the urban economy. Support for Hypothesis 1 requires that the incremental inﬂuence of education on migration to large cities is not only positive but in excess of that found for migration from large cities. To examine this, we distinguish between three educational groups: those with no education, those with less than secondary degree (less than primary, primary and preparatory degrees), and those with secondary and higher (secondary, university and postgraduate degrees).21 r The London School of Economics and Political Science 2005

50

[FEBRUARY

ECONOMICA

Age. Younger persons are usually thought to be more mobile than older persons because the lifetime income gains from moving are larger for the young. In contrast, older individuals have higher moving costs, from having greater locational family ties. In most developing countries migration is concentrated in the 15–30 age group, with a substantial portion in the 15–24 sub-group. Support for Hypothesis 2 requires not only that being young has a positive effect on migration to large cities, but also that the effect exceeds that estimated for migration from big cities. We use parameter estimates to construct proﬁles of the predicted probability of migration by age, holding other inﬂuences constant, for migration both into and out of large cities. We also differentiate by ﬂows from rural and ‘other urban’ locations. Three age group dummies, 15–26, 27–34 and 35–64 years, are used to provide age differentiation. We also control for other individual and regional characteristics that might inﬂuence the probabilities of migration to and from large cities. Origin employment sector. The public sector has been a major employer for the educated during the last few decades in Egypt. McCormick and Wahba (2003) develop and test a model of how the growth of public jobs with wage premiums may help to explain the high, and potentially inefﬁcient, level of urbanization in LDCs. They ﬁnd that the growth of public jobs has altered the pattern of regional mobility in Egypt substantially. However, their focus was on interprovincial migration and not on migration to and from large cities. To examine the role of public-sector employment on migration, we distinguish between being a public-sector employee in the origin. One would expect that private-sector workers would be more likely to migrate to big cities than public-sector workers. This would be true especially of rural areas, where there are fewer public-sector jobs. However, private-sector workers in urban areas may not necessarily be more likely to migrate to a big city for public-sector jobs, since those public jobs are disproportionately allocated in urban areas but not necessarily in the three largest cities. Table 5 shows the origin and destination sector of employment of migrants. Around 36% of the rural private-sector migrants have moved to the public sector in the big city, compared to 27% of the urban private-sector migrants. This is not surprising, since the allocation of public-sector jobs and the appointment of public employment is centralized by the government. TABLE 5

Migrants’ Sector of Employment Origin sector of employment (%)

Rural origin Other urban origin Large cities origin

Destination sector of employment (%)

Private sector

Public sector

Private sector

Public sector

87 61 52

13 39 48

59 53 61

41 47 39

a

Total Private to number public (%)a of migrants

The percentage of private-sector migrants who moved to public sector.

r The London School of Economics and Political Science 2005

36 27 18

50 38 120

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

51

We also differentiate by sector of employment: agriculture, manufacturing, construction and services. We expect agricultural workers to be least mobile, given that their sector-speciﬁc skills are not easily used in urban areasFfor recent evidence see Tunali (1996). Manufacturing workers are often found to be less mobile than service and construction workers in developed countries studies, but this may not hold in countries where manufacturing skills are scarce.

Distance and regional location. Distance between origin and destination is usually found to have a deterrent effect on migration for both developed and developing countries (e.g. Greenwood 1997; Lucas 1997). However, various authors have asserted, without providing evidence, that distance has a less systematic role in explaining migration to a very large city than that in explaining interregional migration, with large centres developing their own, perhaps distant, catchment areas. For example, Mazumdar (1987) observes: ‘Large cities typically develop their individual ‘‘catchment areas’’ from which migrants are drawnFand these areas are not necessarily concentrically distributed in terms of distance.’ These ‘catchment areas’ could be determined by the historical accident by which family/ethnic groups ﬁrst established communities in the city, and thereby acquired the widest network of contacts for potential migrants.22 Thus, the deterrent effect of distance on migration to large cities may differ from that for interregional migration.23 In our analysis location concerns the workplace (not residence), so that ‘migrating into a city’ may reﬂect no change in the location of residence but instead a decision to commute. For households living within commuting distances to the city ‘migration’ is therefore plainly easier, and so we introduce a variable capturing whether a worker lived and worked within commuting distance at the base year. Our ‘contiguous location’ dummy variable is intended to capture this and takes the value 1 if origin work location is less than or equal to 50 kilometres from Cairo. In Table 3 we found that Rural Upper Egypt (the South) provides a disproportionate share of rural migration to Cairo, Giza and Alexandria, and we ask how far this occurs despite a signiﬁcant distance variable effect, or whether it is facilitated by the unimportance of distance to destination. We also wish to account more generally for the factors behind extensive migration from the rural southern areas to the large cities in the north. Regional ﬁxed effects are used to control for regional-speciﬁc characteristics of region of origin that may inﬂuence migration. Two regional dummies are used: Upper is the much more remote south of Egypt; Lower is the north. Work by Katz and Stark (1986) suggests a further reason why remote rural areas may be more likely to provide migrants to a large city than rural areas nearer to the city. Rural migrants may be thought of as choosing whether to migrate to a nearby urban area or to a more distant city. If the region’s farfrom-large cities have a greater share of agricultural output, so that the prosperity of nearby urban areas is more highly correlated with that of the surrounding rural area, then rural families in such regions seeking to diversify their sources of income will be more likely, ceteris paribus, to send workers to a distant city. r The London School of Economics and Political Science 2005

52

ECONOMICA

[FEBRUARY

Origin local labour market conditions. Various studies show that migrants respond to economic circumstances in local labour markets. However, relatively few of these studies allow for both earnings and employment probabilities at origin and destination. The available results show that some combination of economic incentives do matter for migration, but there is little consensus emerging from LDC studies that one of these inﬂuences is especially important.24 Lucas (1985) formulates a fairly general speciﬁcation using micro data to study migration in Botswana and ﬁnds that both origin earnings and employment probabilities have an equal and opposite effect to the same variables at destination. We explore how the relative wage and unemployment rates in urban and rural areas explain the likelihood of migration to the large city. Since the latter is the sole destination, we are able to eliminate the destination condition from our models and express both urban and rural origin log wages and log unemployment relative to the national average.25 We assume that ‘expected’ wages and unemployment are potentially important determinants of migration and use wages and unemployment in 1988 and 1998; also, to add speciﬁcity to these variables, we (a) differentiate the hourly wage at origin by educational category and express this relative to the national average hourly wage, and (b) differentiate the unemployment rate at origin by six age categories. Finally, to help us to understand the circumstances in which certain areas supply migrants to the city when exploring the inﬂuence of local wage and unemployment, we shall consider how far migration from rural areas to large cities is more sensitive to local labour market conditions than migration from urban areas to large cities. In LDCs, public goods and subsidized services are often concentrated into urban areas, with rural areas receiving comparatively low levels of services. Thus, in times of adverse labour market conditions, when families look to social provision for income/food/health support, the rural areas are less likely than urban areas to be provided with social resources. Hence for any given adverse movement in relative regional wages or unemployment, rural workers are more likely to migrate for better long-term opportunities than urban workers. We assume that some of these rural migrants seeking social protection go to CGA. Local amenities. Urban bias in the allocation of public services and amenities is seen by many to be an important determinant of rural–urban migration.26 To capture this effect, we use the proportion of the rural population with access to piped water as a measure of the local provision of amenities in rural areas of origin.

Empirical results The estimates of the models of the probability of migration to CGA are given in Tables 6 and 7. Table 6 provides the odds ratio of urban migration to CGA, while Table 7 provides the odds ratio of rural migration to CGA. Estimates of the multinomial model of migration from CGA to (i) another area of the conurbation, (ii) another urban area and (iii) a rural area are given in Table 8. More speciﬁcally, Table 8 gives the relative odds ratio of migration to another large city (column (1)), migration to another urban area (column (2)) or r The London School of Economics and Political Science 2005

2005]

53

LDC CITY WORKFORCE GROWTH AND COMPOSITION

TABLE 6

Logit Models of Probability of Migration from Urban to Large Cities: Odds Ratios a,b Urban to large cities Variables

(1)

Educational levels Less than secondary Secondary/higher Educated Age group dummies 15–26 years old 35–64 years old Distance variables Distance (km) ‘Contiguous’ area dummy Origin occupational dummies Public-sector ¼ 1 Public-sector

n

secondary/higher

Origin labour market Log relative unemployment rate Log relative wage rate Origin industry dummies Agriculture ¼ 1 Manufacturing ¼ 1 Construction ¼ 1 Region of origin Canal cities ¼ 1 Upper urban ¼ 1 1998 Round Log-likelihood Sample size (no of migrants)

(2)

2.24 (1.35) 8.04 (3.45) –

– – 3.45 (2.06)

(3)

(4)

2.43 (1.62) 9.45 (4.35) –

2.46 (1.49) 4.69 (1.93) –

3.38 (3.65) 0.78 ( 0.63)

2.84 (3.30) 0.76 ( 0.66)

3.10 (3.27) 0.82 ( 0.52)

3.17 (3.24) 0.78 ( 0.63)

1.00 ( 0.19) 4.38 (2.30)

1.00 ( 0.22) 5.24 (2.36)

1.00 ( 0.17) 4.31 (2.29)

1.00 ( 0.16) 4.66 (2.58)

1.46 (0.90) –

1.82 (1.41) –

0.60 ( 1.20) 0.29 ( 3.17)

0.62 ( 1.05) 0.15 ( 4.81)

0.59 ( 1.22) 0.31 ( 2.96)

0.61 ( 1.16) 0.28 ( 3.41)

1.17 (0.18) 1.35 (0.83) 8.71 (7.30)

0.84 ( 0.19) 1.12 (0.32) 7.36 (6.66)

1.06 (0.06) 1.32 (0.77) 7.81 (8.81)

1.12 (0.13) 1.47 (1.03) 8.82 (7.80)

1.95 (1.00) 0.72 ( 0.53) 0.45 ( 2.16) 162.54 2785 (38)

2.51 (1.35) 0.74 ( 0.46) 0.52 ( 1.88) 166.80 2785 (38)

1.91 (0.95) 0.72 ( 0.53) 0.43 ( 2.39) 162.95 2785 (38)

2.01 (1.06) 0.73 ( 0.51) 0.46 ( 2.16) 161.88 2785 (38)

a

– –

0.99 ( 0.02) 2.62 (1.24)

Robust t-statistics are in parentheses. Reference group: 27–34 years old, illiterate, employed in the services industry and working in Lower Egypt. b

r The London School of Economics and Political Science 2005

54

[FEBRUARY

ECONOMICA

TABLE 7

Logit Models of Probability of Migration from Rural to Large Cities: Odds Ratios a,b Rural to large cities Variables

(1)

Educational levels Less than secondary Secondary/higher Educated Age group dummies 15–26 years old 35–64 years old Distance variables Distance (km) ‘Contiguous’ area dummy Origin occupational dummies Public-sector ¼ 1 Public-sector

n

secondary/higher

Origin labour market Log relative unemployment rate Log relative wage rate Origin local amenities Water supply Origin industry dummies Agriculture ¼ 1 Manufacturing ¼ 1 Construction ¼ 1 Region of origin Upper rural ¼ 1 1998 Round Log–likelihood Sample size (no. of migrants)

(2)

2.55 (3.66) 11.48 (4.90) –

– – 2.91 (4.10)

(3)

(4)

2.32 (2.99) 7.69 (3.52) –

2.44 (3.64) 15.65 (5.70) –

1.09 (0.19) 0.35 ( 4.91)

0.96 (0.11) 0.32 ( 5.45)

1.20 (0.43) 0.33 ( 5.31)

1.12 (0.25) 0.34 ( 4.95)

1.00 ( 0.79) 2.31 (1.60)

1.00 ( 0.73) 2.82 (2.02)

1.00 ( 0.84) 2.47 (1.71)

1.00 ( 0.77) 2.28 (1.59)

0.32 ( 3.07) –

0.47 ( 2.30) –

1.24 (0.20) 1.07 (0.41)

1.03 (0.11) 1.11 (0.20)

1.09 (0.27) 1.25 (0.42)

1.07 (0.20) 1.24 (0.40)

0.97 ( 2.21)

0.97 ( 2.11)

0.97 ( 2.16)

0.97 ( 2.17)

0.55 ( 1.53) 1.83 (1.17) 3.65 (2.04)

0.50 ( 2.12) 1.58 (0.97) 3.09 (1.97)

0.89 ( 0.32) 2.46 (1.76) 5.52 (2.74)

0.54 ( 1.64) 1.75 (1.15) 3.63 (2.07)

1.79 (1.30) 1.10 (0.33) 193.46 2658 (50)

1.68 (1.13) 1.24 (0.74) 193.46 2658 (50)

1.77 (1.28) 1.12 (0.42) 195.93 2658 (50)

1.80 (1.32) 1.10 (0.34) 192.86 2658 (50)

a

– –

0.47 ( 1.62) 0.35 ( 2.33)

Robust t-statistics are in parentheses. Reference group: 27–34 years old, illiterate, employed in the services industry and working in Lower Egypt. b

r The London School of Economics and Political Science 2005

2005]

55

LDC CITY WORKFORCE GROWTH AND COMPOSITION

TABLE 8

Multinomial Models of Migration Probability from Large Cities: Relative Risk Ratiosa,b Large cities to

Variables Educational levels Less than secondary Secondary/higher Educated Age group dummies 15–26 years old 35–64 years old

Large cities to

Large cities

Urban

Rural

Large cities

Urban

Rural

(1)

(2)

(3)

(4)

(5)

(6)

–

–

–

–

–

–

0.68 ( 3.35) 3.37 (4.89) –

0.86 0.59 ( 0.38) ( 0.68) 2.25 0.61 (2.19) ( 0.44) – –

1.02 (0.04) 0.61 ( 1.12)

0.76 0.97 ( 0.76) ( 0.04) 0.32 0.64 ( 4.54) ( 0.82)

1.20 (1.24)

1.24 0.60 (0.59) ( 0.62)

0.91 ( 0.27) 0.57 ( 1.20)

0.68 0.97 ( 1.03) ( 0.04) 0.32 0.64 ( 4.76) ( 0.82)

Origin occupational dummies Public-sector ¼ 1 0.40 ( 4.03)

1.33 (0.76)

1.51 (0.73)

0.51 ( 3.35)

Origin local market Log relative wage rate

1.52 0.21 (0.39) ( 1.23)

7.64 (3.54)

5.69 (2.72)

Origin industry dummies Manufacturing ¼ 1 1.24 (2.16) Construction ¼ 1 1.65 (0.58) Region of origin Alexandria ¼ 1 Giza ¼ 1 1998 Round Log-likelihood Sample size (no. of migrants)

1.57 (1.00) 2.53 (2.09)

1.46 (1.14)

1.51 (0.74)

1.87 0.22 (0.56) ( 1.22)

1.39 (0.87) 2.23 (1.56)

0.96 ( 0.22) 1.32 (0.33)

0.57 ( 0.29) 14.41 (2.56)

0.64 0.42 ( 0.55) ( 0.65) 0.90 0.71 ( 0.18) ( 0.27)

0.57 ( 0.27) 13.97 (2.45)

0.62 0.42 ( 0.59) ( 0.65) 0.87 0.71 ( 0.24) ( 0.27)

0.45 ( 3.35)

0.84 0.27 ( 0.35) ( 2.16)

0.60 ( 2.64)

0.94 0.27 ( 0.14) ( 2.16)

536.86 2340 (38, 55, 27)

1.35 (0.72) 2.27 (1.87)

1.39 (0.93) 2.23 (1.59)

547.18 2340 (38, 55, 27)

a

Robust t-statistics are in parentheses. Reference group: 27–34 years old, illiterate, employed in the services industry and working in Cairo. b

migration to a rural area (column (3)) relative to not migrating from a large city. Although our results describe the implications of Cairo, Giza and Alexandria as the dense conurbation in north Egypt, we have also re-estimated our models for CGA and Kalyuobia, and for Cairo, Giza and Kalyuobia only r The London School of Economics and Political Science 2005

56

ECONOMICA

[FEBRUARY

(excluding Alexandria). In each case our conclusions are identical to those reported. We also run reduced-form estimation with similar regressors for all types of migration (to and from CGA) for the sake of comparison (see Table A2 in the Appendix). The results are similar to those obtained from the nonreduced form, which we discuss below. Consider ﬁrst implications of the supply-side version of Hypothesis 1, that educated labour is ‘draining’ into the cities, thereby enabling cities to acquire a higher share of educated workers. This requires that, ceteris paribus, not only does having a higher level of education increase the migration probability of both rural and urban workers to large cities, but also that this effect is greater for migration to the large city than from it. We ﬁnd in Tables 6 and 7 that, for both rural and urban workers, the estimates give a consistent picture of a positive effect of increasing levels of education on migration to large cities.27 Table 8 gives the contrasting picture of the effects of education on migration from CGA. We ﬁnd no tendency for additional education to increase outmigration from large cities to rural areas (Table 8, columns (3) and (6)). The odds ratio of migration by the highly educated group from rural areas to large cities is greaterFand statistically signiﬁcant at the 5% levelFthan the relative risk ratio of migration from large cities to rural areas.28 Thus, the structure of ﬂows between rural and large cities supports the supply-side arguments of Hypothesis 1. Education increases the likelihood of migration from other urban areas to large cities. However, the effect of additional education on migration from large cities to other urban areas (Table 8, column (2)) is more ambiguous, although weaker than that on migration to large cities from other urban areas. The odds ratio of the secondary and higher education group migrating from other urban to large cities is signiﬁcantly greater, at the 1% level, than migrating in the opposite direction. A little education, though, reduces the probability that a large city worker will move to another urban area. So in columns (4)–(6) of Tables 8 we present a model of migration from large cities but distinguish only those who are literate, described as ‘educated’. We ﬁnd that the migration patterns of ‘educated’ largecity workers are not signiﬁcantly different from those who are not ‘educated’. In other words, the ﬁndings suggest that the educated are more likely than those with no education to move from other urban areas to large cities, but not from large cities to other urban areas. Turning now to Hypothesis 2, and using again Tables 6–8, we ﬁnd that workers younger than 35 years are more likely to migrate to big cities than those aged 35 or over. However, migration from other urban areas peaks at a very early age (15–26 years), while that from rural areas peaks between 27 and 34 years of age.29 Older workersFthose 35 years or moreFare less likely to migrate from rural areas to big cities (Table 7, column (1)), but not from big cities to rural areas (Table 8, column (3)).30 The implications of the effects for probabilities are given in Table 9(a),31 while Table 9(b) shows the signiﬁcance of selected predicted differences in migration probabilities. First, examining the effect of education on migration without distinguishing between age groups (Table 9(a), panel C), we ﬁnd that the probabilities of those with secondary and higher education migrating to large cities from both rural and urban areas are higher than of their migrating from large cities, thus supporting our Hypothesis 1.32 Second, if we do not r The London School of Economics and Political Science 2005

0.014 0.004 0.003 0.005

0.034 0.011 0.006 0.011

0.099 0.035 0.026 0.031

0.032 0.016 0.010 0.013

0.026 0.000 0.004 0.006

0.052 0.033 0.008 0.017

0.163 0.073 0.035 0.049

0.059 0.037 0.013 0.022

0.017 0.013 0.004 0.010

0.047 0.030 0.009 0.024

0.100 0.082 0.032 0.059

0.031 0.029 0.007 0.029

Pred.

Actual

Actual

Pred.

Urban–large cities

Rural–large cities

A: No education Age Less than 26 0.023 27–34 0.007 35 or more 0.006 All age groups 0.010 B: Less than secondary education Age Less than 26 0.043 27–34 0.038 35 or more 0.007 All age groups 0.022 C: Secondary/higher education Age Less than 26 0.001 27–34 0.048 35 or more 0.037 All age groups 0.041 D: All educational levels Age Less than 26 0.031 27–34 0.026 35 or more 0.008 All age groups 0.017

TABLE 9(a)

r The London School of Economics and Political Science 2005

0.026 0.022 0.011 0.016

0.063 0.040 0.017 0.026

0.017 0.011 0.006 0.009

0.034 0.031 0.016 0.023

Pred.

0.025 0.041 0.014 0.021

0.001 0.059 0.032 0.036

0.016 0.008 0.009 0.001

0.037 0.048 0.008 0.021

Actual

0.029 0.043 0.015 0.024

0.056 0.070 0.025 0.039

0.023 0.031 0.010 0.017

0.031 0.042 0.013 0.023

Pred.

Large cities–urban

0.014 0.021 0.012 0.015

0.037 0.019 0.006 0.011

0.025 0.031 0.007 0.015

0.005 0.012 0.023 0.017

Actual

0.014 0.014 0.009 0.011

0.014 0.012 0.006 0.009

0.014 0.014 0.009 0.011

0.015 0.018 0.011 0.013

Pred.

Large cities–rural LDC CITY WORKFORCE GROWTH AND COMPOSITION

0.025 0.022 0.014 0.018

0.030 0.026 0.026 0.027

0.016 0.008 0.009 0.010

0.047 0.062 0.011 0.027

Actual

Large cities–large cities

Actual and Predicted Probabilities of Migration, by Age and Education

2005] 57

58

ECONOMICA

[FEBRUARY

TABLE 9(b)

Significance of Selected Predicted Differences in Migration Probabilities Migration between Rural–large cities and urban–large cities

Rural–large cities and large cities–rural

Urban–large cities and large cities–urban

Secondary/higher education o26 Insigniﬁcanta 27–34 Signiﬁcant at 6%b All Signiﬁcant at 3%

Signiﬁcant at 8% Signiﬁcant at 1% Signiﬁcant at 1%

Insigniﬁcant Signiﬁcant at 9% Signiﬁcant at 9%

All educational levels o26 Insigniﬁcant 27–34 Signiﬁcant at 6%

Signiﬁcant at 5% Signiﬁcant at 4%

insigniﬁcant Signiﬁcant at 1%

a This means that for those 26 years old or less with secondary and higher degrees, the predicted probability of migrating from rural areas to large cities is not statistically different from that of migrating from urban areas to large cities. b This means that for those 27–34 years old with secondary and higher degrees, the predicted probability of migrating from rural to large cities is statistically different (at 6% level) from that of migrating from urban areas to large cities.

distinguish between educational levels and examine the impact of age alone (Table 9(a), panel D), we ﬁnd that the probability of the youngest group of workers (those less than 27 years old) moving from rural areas to large cities is higher than the probability of their migrating out of large cities into rural areas (3% compared with 1.4%).33 In addition, the probability of those aged 27–34 moving from rural areas to large cities is higher than of their moving in the opposite direction.34 So for those less than 35 years old the probability of their migrating from rural to large cities is greater and more signiﬁcant than their moving in the opposite direction. However, for very young urban individuals (less than 26 years old), the probability of migrating to large cities is similar to that of migrating from large cities. On the other hand, those between 27 and 34 years old are more likely to move out of large cities to urban areas than in the other direction.35 Finally, for those with higher education and less than 26 years old, we ﬁnd that the probability of their migrating from rural areas to large cities is 10%, while in the opposite direction it is only 1.4%. However, for this same group, the probability of their migrating from urban areas to large cities is similar to that of their migrating in the opposite direction (3%). To sum up, we ﬁnd evidence to support our Hypothesis 2 in the case of rural ﬂows, but not in the case of urban ﬂows, to and from big cities. The likelihood of migration from rural areas to large cities by a privatesector worker is higher than that for a public-sector worker. However, the probability of a public-sector worker moving from an urban area to a large city is not signiﬁcantly different from that of a private-sector worker. Being employed in the public sector has no signiﬁcant impact on large-city workers migrating to either urban or rural areas. One explanation for this may be the fact that it is not uncommon to transfer public-sector workers from large cities to smaller cities and rural areas. In order to disentangle the effect of publicr The London School of Economics and Political Science 2005

2005]

59

LDC CITY WORKFORCE GROWTH AND COMPOSITION

TABLE 10

Multinomial Models of Migration Probability from Large Cities and the Effect of the Public sector: Relative Risk Ratiosa,b Large cities to Variables Educational levels Less than secondary Secondary/higher Age group dummies 15–26 years old 35–64 years old

Large cities

Urban

0.58 ( 4.35) 2.41 (3.32)

0.92 ( 0.20) 2.48 (2.44)

1.23 (0.61) 0.51 ( 1.63)

0.70 ( 1.16) 0.34 ( 4.85)

Origin occupational dummies Public-sector ¼ 1 – Public-sectorn Secondary/higher Origin local market Log relative wage rate

–

Large cities to Rural

Large cities

Urban

Rural

0.65 0.70 ( 0.62) ( 3.35) 0.71 2.68 ( 0.33) (3.79)

0.83 ( 0.45) 2.79 (2.41)

0.63 ( 0.58) 0.00 (0.00)

0.86 1.00 ( 0.23) (0.00) 0.67 0.61 ( 0.72) ( 1.15)

0.78 ( 0.72) 0.32 ( 4.57)

0.93 ( 0.12) 0.64 ( 0.82)

0.30 ( 3.32) 1.86 ( 2.37)

1.54 (1.52) 0.68 ( 0.75)

1.19 (0.25) 0.00 (0.00)

–

–

–

–

5.62 (2.67)

1.50 (0.37)

0.21 ( 1.25)

5.65 (2.55)

1.50 (0.37)

0.22 ( 1.21)

Origin industry dummies Manufacturing ¼ 1 1.29 (2.27) Construction ¼ 1 1.92 (0.70)

1.57 (1.01) 2.42 (2.18)

1.42 (0.96) 2.09 (1.36)

1.32 (2.68) 1.62 (0.57)

1.55 (0.98) 2.53 (2.12)

1.43 (0.90) 2.22 (1.54)

Region of origin Alexandria ¼ 1 Giza ¼ 1 1998 Round Log–likelihood Sample size (no. of migrants)

0.55 ( 0.30) 14.36 (2.68)

0.64 ( 0.55) 0.89 ( 0.20)

0.42 0.57 ( 0.64) ( 0.28) 0.70 14.32 ( 0.28) (2.61)

0.63 ( 0.56) 0.90 ( 0.19)

0.43 ( 0.63) 0.72 ( 0.27)

0.49 ( 2.87)

0.81 ( 0.44)

0.26 0.46 ( 2.17) ( 3.32)

0.83 ( 0.38)

0.27 ( 2.11)

540.50 2340 (38, 55, 27)

535.02 2340 (38, 55, 27)

a

Robust t-statistics are in parentheses. Reference group: 27–34 years old, illiterate, employed in the services industry and working in Cairo. b

sector employment from education, since a big proportion of the educated are employed in the public sector, we ﬁrst omit the public-sector dummy; i.e. we examine the effect of education without controlling for public-sector employment (Table 6, column (3); Table 7, column (3); and Table 10, columns (1)–(3)). We ﬁnd that the effect of education is stronger in the case of migration from r The London School of Economics and Political Science 2005

60

ECONOMICA

[FEBRUARY

urban to large cities, and weaker in the case of migration from rural to large cities. Second, we analyse the impact of public-sector employment on the educated by controlling for being highly educated and employed in the public sector (Table 6, column (4); Table 7, column (4); and Table 10, columns (4)– (6)). The results are consistent with the previous ﬁndings: there appears to be no simple inﬂuence of public-sector employment on the effects of age and education on migration. The evidence concerning the inﬂuence of distance consistently supports the view that, provided areas near to CGA are treated separately, then, using the commuting distance dummy variable, the probability of migration to CGA is not signiﬁcantly affected by the potential migrant’s distance from the large cities. This supports our a priori discussion that distance has a less systematic role in explaining migration to a very large city than that of interregional migration. Various nonlinear speciﬁcations for the role of distance yield similar conclusions. We also ﬁnd that households already living and working within commuting distances are more likely to migrate to work in CGA from both urban and rural areas, although this effect is stronger for urban workers. Turning to the inﬂuence of local labour markets, migration from rural areas to large cities does not appear to be sensitive to local labour market conditions. The ﬁndings suggest that neither relative unemployment rates nor wage rates in rural areas affect rural migration to big cities. Although economic factors such as income and employment opportunities do not seem to affect rural migration, access to local amenities does appear to have an important inﬂuence. However, improved access to water supply, our measure of local amenities, has a negative and signiﬁcant impact on migration, indicating that, the greater the provision of rural amenities, the less is the probability of rural migration to large cities. The relative wage rate at origin appears to have the conventional effect in the case of urban migration to large cities: a lower-than-average origin area wage rate induces urban migration to large cities. However, the impact of unemployment rates on urban migration to large cities is not signiﬁcant. There is little evidence that local labour market conditions have a greater inﬂuence on migration from rural areas. We can now summarize the contribution of the remaining control variables. The industry to which the individual is attached plays a conventional role in determining migration into and out of large cities. Workers in the construction sector are more mobile than workers in the services sector, regardless of their origin and destination. Rural workers attached to the agriculture sector are less likely to migrate to large cities. This perhaps reﬂects their comparative lack of skills for work in urban areas or, alternatively, the existence of nonmarketable rents, which are earned in agriculture sector work.

IV. CONCLUSION This paper considers how locational preference among educated and young workers helps to concentrate these groups in large cities, and suggests how migration data can be used to test for such preferences. The paper begins by contrasting the employment shares of the young and educated in Egyptian large cities, other urban, and rural areas, and then explores whether the relative r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

61

preference of the educated and young for living in the largest cities raises the equilibrium employment share of the educated and young in these densely populated urban areas, as predicted by the ‘bright lights’ and ‘spillover learning amenities’ hypotheses. This requires that educated workers living in rural areas (or other urban areas) are more likely to migrate to the big cities than the less educated, holding constant rural (or other urban areas) wages relative to big city wages, and that this differential between educational groups is larger than that between these same groups migrating from the big cities. We ﬁnd evidence consistent with the view that the high share of educated workers in the big cities’ labour forces is at least in part a result of a supply-side effect, reﬂecting either the perceived amenities (‘bright lights’) of a large city or the greater learning opportunities that the city provides. The evidence is especially compelling for rural–large city migration, and indicates that rural– large city migration in Egypt is more consistent with the older (Marshallian) view that cities disproportionately attract the educated rather than the uneducated. Migration into and out of the big city among the young enables the big city to experience a signiﬁcantly lower mean age of worker, but migration among the old is not sufﬁciently differentiated to signiﬁcantly reinforce this effect. The results suggest that rural–big city migration patterns are broadly consistent with models in which LDC large cities have grown partially in order to concentrate young workers and educated workers spatially. The migration patterns between other urban areas and large city are suggestive of this effect, but insigniﬁcant. We have considered evidence regarding the comparative willingness of the educated to work in Egyptian large cities. However, in the past forty years Egypt has also experienced a rising share of the educated in the labour force: between 1960 and 1986 the share of the population holding university degrees rose from 1.7% to 7.5%; over the same period the percentage of illiterates dropped from 63.8% to 45.6%. What are the implications of this? In our model an increase in the stock of educated labour will increase the relative size of the sector intensive in educated labourFthe urban sector. Policies to increase the share of educated workers may prompt net migration ﬂows to the city until equilibrium is reached. Thus, whereas the preference of the educated workers for cities can help explain both the high share of educated workers in cities and the existence of large cities, a steady increase in the share of the educated workers in the labour force can interact with this preference to help explain persistent net migration of the educated into cities, and the relative growth of such cities. Given the evidence that we have discussed regarding preference for city location, it appears plausible that policies to increase education may have contributed, through net migration, to increasing the share of LDC employment in large cities.

APPENDIX: THE DATA This paper uses data from two rich surveys: the Egypt Labour Market Survey 1998 (1998 ELMS) and the 1988 Labour Force Sample Survey (1988 LFSS), carried out by the Central Agency of Public Mobilization and Statistics (CAPMAS) in Egypt. The 1988 LFSS was the ﬁrst survey in Egypt to collect detailed data on employment r The London School of Economics and Political Science 2005

62

[FEBRUARY

ECONOMICA

TABLE A1

Definition of Variables Variable

Deﬁnition

Individual characteristics Educational levels: illiterate is the reference group Less than secondary ¼ 1 if individual has less than secondary degree (less than primary, primary or preparatory certiﬁcate) Secondary/higher ¼ 1 if individual has secondary or university degree Educated ¼ 1 if individual has less than secondary degree or higher than secondary degree Age group dummies: 27–34 years old is the reference group 15–26 years old ¼ 1 if individual was 26 years old or less 35–64 years old ¼ 1 if individual was 35 years old or more Distance variables Distance (km) Distance in kilometres between the main city in the governorate of origin work location, and Cairo ‘Contiguous’ area Dummy ¼ 1 if origin work location is less than or equal to 50 kilometres from Cairo Origin occupational characteristics Public-sector ¼ 1 if employed in the public sector before migration Public-sectorn secondary/higher Interactive dummy: ¼ 1 if employed in the public sector before migration and if individual has secondary or university degree Origin industry dummies: services are the reference group Agriculture ¼ 1 if employed in the agriculture sector before migration Manufacturing ¼ 1 if employed in the manufacturing sector before migration Construction ¼ 1 if employed in the construction sector before migration Origin labour market Log relative unemployment rate Log unemployment rate in origin governorate relative to national average Log relative wage rate Log hourly wage rate in (rural or urban) origin governorate relative to national average, by educational level Origin local amenities Water supply The proportion of the population in origin governorate with access to piped water. Region of origin: ‘Lower urban’ is the reference group for urban areas; ‘Lower rural’ is the reference group for rural areas; ‘Cairo’ is the reference group for big cities Canal cities ¼ 1 if working in ‘Canal cities’ (Port Said, Suez or Ismalia) before migration Upper urban ¼ 1 if working in ‘Upper urban’ before migration Upper rural ¼ 1 if working in ‘Upper rural’ before migration Alexandria ¼ 1 if working in Alexandria before migration Giza ¼ 1 if working in Giza before migration Round: 1988 Round is the reference group 1998 Round Dummy ¼ 1 if Round is 1998; ¼ 0 if Round is 1988

r The London School of Economics and Political Science 2005

2005]

63

LDC CITY WORKFORCE GROWTH AND COMPOSITION

TABLE A2

Reduced-Form Models of Probability of Migration to and from Large Cities: Odds Ratios and RRR a,b Large cities to Variables

Urban to Rural to large cities large cities

Educational levels Less than secondary

Large cities

Urban

Rural

2.26 (1.35) 9.12 (3.31)

2.81 (3.36) 17.27 (4.77)

0.65 ( 1.32) 3.21 (3.18)

0.83 ( 0.52) 2.14 (2.20)

0.44 ( 1.18) 0.43 ( 0.75)

3.58 (3.78) 0.81 ( 0.51)

1.23 (0.45) 0.38 ( 4.25)

1.01 (0.03) 0.62 ( 1.11)

0.75 ( 0.82) 0.33 ( 4.52)

0.95 ( 0.08) 0.66 ( 0.78)

Origin occupational dummies Public-sector ¼ 1 1.37 (0.76)

0.35 ( 3.24)

0.40 ( 3.98)

1.33 (0.75)

1.51 (0.75)

0.54 ( 1.49) 0.36 ( 2.39)

1.19 (0.85) 1.17 (0.39)

0.611 ( 0.21) 4.39 (1.33)

0.30 ( 1.56) 1.21 (0.21)

0.02 ( 2.49) 0.061 ( 2.84)

1.79 (1.95) 8.52 (8.43)

3.05 (2.52) 5.89 (4.50)

1.29 (2.16) 1.65 (0.59)

1.57 (1.02) 2.54 (2.09)

1.42 (0.91) 2.32 (1.64)

–

–

–

–

–

–

–

–

–

–

–

Secondary/higher Age group dummies 15–26 years old 35–64 years old

Origin labour market Log relative unemployment rate Log relative wage rate Origin industry dummies Manufacturing ¼ 1 Construction ¼ 1 Region of origin Canal cities ¼ 1 Upper urban ¼ 1 Upper rural ¼ 1

1.17 (0.26) 0.46 ( 1.49) –

Alexandria ¼ 1

–

1.88 (1.19) –

Giza ¼ 1

–

–

1998 Round Log–likelihood Sample size (no. of migrants)

0.42 ( 2.29) 167.28 2785 (38)

0.78 ( 0.67)

0.58 ( 0.28) 11.46 (1.15)

0.81 ( 0.25) 0.52 ( 0.79)

0.48 ( 0.55) 0.18 ( 1.44)

0.31 ( 0.66)

0.52 ( 1.66)

0.04 ( 3.95)

204.72 2658 (50)

a

533.98 2340 (38, 55, 27)

Robust t-statistics are in parentheses. Reference group: 27–34 years old, illiterate, employed in the services industry.

b

r The London School of Economics and Political Science 2005

64

ECONOMICA

[FEBRUARY

characteristics, labour mobility, earnings and other labour market variables. Taking the 1988 LFSS as a baseline, the 1998 ELMS has replicated the design and methodology used in the 1988 LFSS to ensure comparability of the two data-sets. Both surveys are based on a multi-stage stratiﬁed random sampling.36 Weighting schemes were designed to correct for bias in the selection process, and expansion weights were based on the population estimates. Both data-sets are nationally representative household surveys that gathered data on a wide range of labour market variables at the household and individual level. The 1988 LFSS covered 10,000 households, while the 1998 ELMS covered 5,000 households. Each data-set is comprised of three questionnaires: (1) the household questionnaire; (2) the individual questionnaire; (3) the family enterprise questionnaire. Each household had at least one household questionnaire and one individual questionnaire. If any of the members of the household were self-employed or employers, a family enterprise questionnaire for this household was administrated. Data for the household questionnaire were collected from the head of the household and included the roster of members of the household, each individual’s relationship to the head of the household and demographic characteristics of the household. The individual questionnaire applies to individuals 6 years old and over. A battery of individual modules was designed to collect data on individual characteristics, employment characteristics, unemployment, mobility and career history, and earnings. Data for the individual questionnaire were collected from the individuals themselves except for individuals less than 15 years old. This paper uses the labour mobility modules in both surveys. The labour mobility module of 1988 was carried on half the total sample of respondents to LFSS1988, covering around 5000 households. The labour mobility module of 1998 also covered around 5000 households. Both modules were administered to individuals who were, or had been, at some of time, in the labour force. Both modules collected data on a vector of employment characteristics and employment location for the reference period (October 1988 for the 1988 LFSS and November/December 1998 for the 1998 ELMS) and retrospective employment history using memorable events as markers to help people in their recollection; for example, the 1988 LFSS chose the assassination of President Sadat in 1981, and the 1998 ELMS focused on the time of Iraq’s invasion to Kuwait, in August 1990. Given our interest in the changing composition of the labour force, we deﬁne migration with reference to work location rather than residential location. The sample for analysis is conﬁned to males who were employed at the beginning and at the end of the period under study, since women have different participation patterns. The total sample size used is 7783 male employed workers, of whom 208 are migrants, i.e. have changed work location. Our analysis is based on three sub-samples. (i) For the rural–big cities migration, we had 2658 rural workers, of whom 50 had moved from rural areas to Cairo, Giza or Alexandria (CGA). (ii) For the other urban areas–big cities migration, we had 2785 urban workers (other urban areas excluding CGA), of whom 38 had moved from other urban areas to CGA. (iii) For migration out of the three largest cities, we had 2340 CGA workers, of whom 27 had moved into rural areas, 55 had moved into other urban areas and 38 had moved between these large cities.

NOTES 1. See e.g. Lucas (1997) for a survey of the evidence that artiﬁcially high urban wages may prompt city growth. In this framework, workers leave rural jobs with positive productivity to become unemployed in cities, so that city growth is inefﬁcient. 2. Other major contributions to this literature include Jacobs (1969), Kelly and Williamson (1984), Glaeser et al. (1992) and Rauch (1993). 3. For example Todaro (1976) and Yap (1977). 4. The importance of migration for city growth is widely analysed and found to explain in the order of 50%–70% of LDC city growth (see e.g. Mazumdar 1987; Eastwood and Lipton 2000). 5. McCormick and Wahba (2003) use inter-provincial migration and employment data to study how governorates (provinces) with large shares of high-wage public jobs may attract migrant workers. Our analysis here focuses on migration into and out of a very large city, r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

65

and, while the same argument applies, we have only one destination, and thus too few degrees of freedom to test the same hypothesis. Nevertheless, we investigate how far public employment might moderate the inﬂuence of education and age on migration in Section III. 6. Alfred Marshall’s (1890) observations are cited in Glaeser (1999, pp. 2–3). 7. Marshall (1890) argues that skilled workers beneﬁt from being near to one another in dense areas because they learn new skillsFor, as he calls it, learn about ‘the mysteries of trade’. 8. Pessino (1991) shows that for Peru the highly educated are not more likely to migrate from Lima, and the same result applies to other urban areas. However, migration from rural areas, albeit to any destination, is more likely for the highly educated. 9. This evidence is presented in a table analogous to our Table 2Fsee Glaeser (1999). 10. Between 1976 and 1986, the inter-census period, the urban population grew by 10% and the number of towns having more than 100,000 inhabitants rose from 20 to 24. Egypt certainly conforms to Lucas’s conjecture (1997, p. 727), that it is ‘not true that the largest cities are out-growing the medium size cities’. 11. The proportion of the total population living in these three cities declined from 22.94% to 21.44%, and the proportion of the urban population fell from 52.50% to 50.29%, between 1986 and 1996. 12. October 1981 was chosen to benchmark retrospective information since President Sadat was assassinated in that month; August 1990 was chosen because it was the time of Kuwait’s invasion. 13. Since we examine migration during 1981–88, and in the second survey during 1990–98, we measure age at the middle of the period, since workers could have moved either at the beginning of the period or at the end of it. In addition, since we are interested in how migration alters the labour force composition, we include in our analysis individuals as young as 12 years old at the base year (15 years at the middle of the period) as long as they are employed. We do not question whether children can make an independent location decision, though Iversen (2002) ﬁnds empirical evidence in support of child workers’ exhibiting autonomous migration behaviour in India. We also test for the robustness of our results given the way age is measured, by using age at the beginning of the period and considering workers aged 15–64 at the base year, and ﬁnd that the results are robust. 14. Employed workers in the formal, informal and agricultural sector are included. We do not include unemployed individuals in the sample, since the majority of those unemployed are new entrants to the labour market whose migration behaviour is different from those employed. 15. Although 3% of workers have migrated to the largest cities, which might appear a low migration rate over seven or eight-year span, we are examining migration into and out of only three cities. 16. We are interested in migration to big cities; thus, we deﬁne migration as a movement to any of those three large cities, and we also consider any other type of migration as a non-move, e.g. rural to rural or rural to other urban. 17. Sampling weights based on population estimates are used. However, the results are not sensitive to these weights. All descriptive statistics reported in this section are signiﬁcant at the 10% level or better. 18. This difference is signiﬁcant at the 10% level. 19. We pool both data-sets to increase sample size. Both data-sets have similar design and sampling methodologyFsee the Appendix for details. 20. The estimation was also carried out on each data-set separately to check for the robustness of the results across both time periods. The estimates were similar, but because of the small sample sizes it was felt that combining both data-sets would provide better statistical signiﬁcance. 21. We have data on educational levels only at the time of survey, i.e. in 1988 or 1998. However, this should not be a major concern, since our analysis is conﬁned to employed individuals at the base year and at the end of the period (time of survey). 22. In the case of Egypt, such an incident occurred when the construction of the High Dam in Aswan in 1972 caused Nubian communities to migrate to Cairo following loss of their land. 23. In this paper, the distance variable measures the distance between the area of origin rural or urban governorates and Cairo. 24. See the surveys by Lucas (1997) and Mazumdar (1987). 25. Hourly wage rates are calculated from the earnings modules of the 1988 LFSS and the 1998 ELMS. Since using grouped (aggregated) data in individual-level regressions can result in the standard errors being biased because of the correlation of the error term across individuals in a region or industry, we correct for the correlation of error terms across individuals in each province and report the robust estimates. The Robust (Huber/White/ Sandwich) estimator of the variance was used in place of the conventional Maximum

r The London School of Economics and Political Science 2005

66

ECONOMICA

[FEBRUARY

Likelihood Estimation variance estimator, and observations were allowed to be dependent within governorate/province. 26. See Eastwood and Lipton (2000) for a recent discussion. 27. Although the coefﬁcient estimates of secondary and higher education are not signiﬁcantly different for rural–large city migration from those for other urban–large city migration. 28. Pessino (1991) shows that for Peru the highly educated are not more likely to migrate from Lima, and the same result applies to other urban areas. However, she does not distinguish between urban and rural destinations, and does not consider migration into Lima. 29. This difference is signiﬁcant at the 1% level. 30. The odds ratio of those 35 years old or over migrating from other urban areas to large cities is greater than (and signiﬁcant at 1% level) the relative risk ratio of migrating from large city to other urban areas. 31. Both predicted (simulated) and actual (raw) probabilities are presented to show how raw probabilities can be misleading. Predicted probabilities are based on Table 6, column (1), Table 7, column (1) and Table 8, columns (1)–(3). 32. These differences are signiﬁcant at 10% level or better; see Table 9(b). 33. This difference is signiﬁcant at 5% level, as shown in Table 9(b). 34. Table 9(b) shows that this difference is signiﬁcant at 5% level. 35. This difference is signiﬁcant at 1% level, as in Table 9(b). 36. For details on the sampling design and methodology of the 1988 LFSS, see Fergany (1991), and for the 1998 ELMS see Assaad and Barsoum (1999).

REFERENCES ADES, A. and GLAESER, E. (1995). Trade and circuses: explaining urban giants. Quarterly Journal of Economics, 110, 195–227. ASSAAD, R. and BARSOUM, G. (1999). Egypt Labour Market Survey 1998: Report on the Data Collection and Preparation. Cairo: Economic Research Forum. CAPMAS (1986). Population, Housing and Establishment Census. Cairo: CAPMAS. FFF (1996). Population, Housing and Establishment Census. Cairo: CAPMAS. EASTWOOD, R. and LIPTON, M. (2000). Rural–urban dimensions of inequality change. Working Paper no. 200, United Nations University, WIDER, September. FERGANY, N. (1991). Overview and General Features of Employment in the Domestic Economy: Final Report. Cairo: CAPMAS. FIELDS, G. (1982). Place-to-place migration in Colombia. Economic Development and Cultural Change, 30, 539–58. FULLER, T. D. et al. (1985). Rural–urban mobility in Thailand: a decision-making approach. Demography, 22, 565–79. GLAESER, E. (1999). Learning in cities. Journal of Urban Economics, 46, 254–7. FFF and MARE´, D. (2001). Cities and skills. Journal of Labor Economics, 19, 316–42. FFF, KALLAL, H., SCHEINKMAN, J. and SHLEIFER, A. (1992). Growth in cities. Journal of Political Economy, 100, 1126–52. GREENWOOD, M. J. (1997). Internal migration in developed countries. In M. Rosenzweig and O. Stark (eds.), Handbook of Population and Family Economics, vol. 1B, Amsterdam: Elsevier Science. HENDERSON, J. V. (1986). Urbanisation in a developing country: city size and population composition. Journal of Development Economics, 22, 269–93. FFF (1988). Urban Development: Theory, Fact and Illusion. Oxford: Oxford University Press. IVERSEN, V. (2002). Autonomy in child labour migrants. World Development, 30, 817–34. JACOBS, J. (1969). The Economy of Cities. New York: Random House. KATZ, E. and STARK, O. (1986). Labour migration and risk aversion in less developed countries. Journal of Labor Economics, 4, 134–49. KELLY, A. and WILLIAMSON, J. (1984). What Drives Third World City Growth? Princeton, NJ: Princeton University Press. LUCAS, R. E. B. (1985). Migration amongst the Botswana. Economics Journal, 95, 358–82. FFF (1997). Internal migration in developing countries. In M. Rosenzweig and O. Stark (eds.), Handbook of Population and Family Economics, vol. 1B. Amsterdam: Elsevier Science. MAZUMDAR, D. (1981). The Urban Labour Market and Income Distribution: a study of Malaysia. New York: Oxford University Press. r The London School of Economics and Political Science 2005

2005]

LDC CITY WORKFORCE GROWTH AND COMPOSITION

67

FFF (1987). Rural–urban migration in developing countries. In E. S. Mills (ed.), Handbook of Regional and Urban Economics, vol. II. Amsterdam: Elsevier Science. MCCORMICK, B. and WAHBA, J. (2003). Did public wage premiums fuel agglomeration in LDCs? Journal of Development Economics, 70, 349–79. PESSINO, C. (1991). Sequential migration theory and evidence from Peru. Journal of Development Economics, 36, 55–87. RAUCH, J. (1993). Productivity gains from geographic concentration of human capital: evidence from the cities. Journal of Urban Economics, 34, 380–400. SCHULTZ, P. (1982). Lifetime migration within educational strata in Venezuela: estimates of a logistic model. Economic Development and Cultural Change, 30, 559–93. SCHWARTZ, A. (1976). Migration, age, and education. Journal of Political Economy, 84, 701–19. TODARO, M. (1976). Urban job expansion, induced migration and rising unemployment: a formulation and simpliﬁed empirical test for LDCs. Journal of Development Economics, 3, 211–25. TUNALI, I. (1996). Migration and remigration of male household heads in Turkey, 1963–1973. Economic Development and Cultural Change, 45, 31–67. YAP, L. (1977). The attraction of cities: a review of migration literature. Journal of Development Economics, 4, 239–64.

r The London School of Economics and Political Science 2005

Return migration: evidence from academic scientists

Data Migration from QnE Delphi to Optimum.pdf

Exploiting evidence from unstructured data to enhance master data ...

Evidence from Head Start

Who Migrates and Why? Evidence from Italian Administrative Data

Data Migration System in Heterogeneous Database - International ...

From eurocrisis to asylum and migration crisis

Evidence from Goa

Evidence from Ethiopia

Dynamic Data Migration Policies for* Query-Intensive Distributed Data ...

GSTR-1 Data Migration to Tally from Other Software.pdf

The Economics of Migration from Mexico

International Migration from Indonesia: Stylized Facts ...

Evidence from Diversified Conglomerates - Chicago

Evidence from Head Start - Harvard University

Striking Evidence from the London Underground Network

Evidence from evaluating mathematical strategies

Domestic Gains from Offshoring? Evidence from TAA ...