rural-to-urban and urban-to-urban migration patterns in colombia

18
IfABITINTL. Vol. I?. No. I. pp. 133-150. I993 Printed in Great Britain 0197-3Y75/93 $6.00 + 0.00 @ 19Y3 Pergamon Press Ltd Rural-to-Urban and Urban-to-Urban Migration Patterns in Colombia DANIEL SHEFER and LUIS STEINVORTZ Center for Urban and Regional Studies, Technion - Israel Institute of Technology, Israel The object of this study was to identify the factors that may explain, and help to predict, the direction and intensity of migration flows from rural to urban and from urban to urban areas in Colombia. For this purpose, statistical models were used with a view to obtaining a better insight into the push-and-pull causes of the migration patterns and a better understanding of their consequences. Whereas the principal movement of most migrants in the past was to leave the rural sector for the cities, a large proportion of the migration in Colombia today originates in small urban areas. This relatively recent phenomenon of urban-to-urban migration deserves special attention. INTRODUCTION In the past few decades, the Colombian economy has undergone a remarkable structural change, manifested in a shift in the labour force from agriculture to other sectors, primarily the services. 1 During this period, Colombia also experienced one of the most accelerated urbanisation processes among the Latin America countries. This phenomenon is attributed more to a net inflow of migrants into the cities than to a rise in the rate of natural population growth. From 1938 to 1985, the proportion of the urban population more than doubled, from 29.7% to 67.3%. At the same time, the rural population rate of growth decreased (see Appendix A). Between 1973 and 1985, moreover, it turned into a negative annual growth rate, of -0.2%.2 Local and municipal governments in Colombia have experienced severe difficulties in providing their residents with the necessary employment oppor- tunities, housing, pubfic infrastructure and services. The consequences, of the kind often associated with a rapid, uncontrolled urbanisation process, were undesirable socio-economic problems leading to social and political unrest and inhibiting development and economic growth. GEOGRAPHICAL LOCATION Colombia, situated at the northern point of South America, serves as the link between South America and Central America. The country has an area of about 1,142,OOOkm2 and a population of about 28 million. Colombia is divided by the Andes mountains into two portions: the West, characterised by the Andes highlands and the coastal plains of the Pacific w@ 17:1-J 133

Upload: daniel-shefer

Post on 22-Nov-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Rural-to-urban and urban-to-urban migration patterns in Colombia

IfABITINTL. Vol. I?. No. I. pp. 133-150. I993 Printed in Great Britain

0197-3Y75/93 $6.00 + 0.00 @ 19Y3 Pergamon Press Ltd

Rural-to-Urban and Urban-to-Urban Migration Patterns in Colombia

DANIEL SHEFER and LUIS STEINVORTZ Center for Urban and Regional Studies, Technion - Israel Institute of

Technology, Israel

The object of this study was to identify the factors that may explain, and help to predict, the direction and intensity of migration flows from rural to urban and from urban to urban areas in Colombia. For this purpose, statistical models were used with a view to obtaining a better insight into the push-and-pull causes of the migration patterns and a better understanding of their consequences.

Whereas the principal movement of most migrants in the past was to leave the rural sector for the cities, a large proportion of the migration in Colombia today originates in small urban areas. This relatively recent phenomenon of urban-to-urban migration deserves special attention.

INTRODUCTION

In the past few decades, the Colombian economy has undergone a remarkable structural change, manifested in a shift in the labour force from agriculture to other sectors, primarily the services. 1 During this period, Colombia also experienced one of the most accelerated urbanisation processes among the Latin America countries. This phenomenon is attributed more to a net inflow of migrants into the cities than to a rise in the rate of natural population growth. From 1938 to 1985, the proportion of the urban population more than doubled, from 29.7% to 67.3%. At the same time, the rural population rate of growth decreased (see Appendix A). Between 1973 and 1985, moreover, it turned into a negative annual growth rate, of -0.2%.2

Local and municipal governments in Colombia have experienced severe difficulties in providing their residents with the necessary employment oppor- tunities, housing, pubfic infrastructure and services. The consequences, of the kind often associated with a rapid, uncontrolled urbanisation process, were undesirable socio-economic problems leading to social and political unrest and inhibiting development and economic growth.

GEOGRAPHICAL LOCATION

Colombia, situated at the northern point of South America, serves as the link between South America and Central America. The country has an area of about 1,142,OOO km2 and a population of about 28 million.

Colombia is divided by the Andes mountains into two portions: the West, characterised by the Andes highlands and the coastal plains of the Pacific

w@ 17:1-J 133

Page 2: Rural-to-urban and urban-to-urban migration patterns in Colombia

and Caribbean; and the East, consisting of jungle and grasslands comprising approximately one-half of thcs country’s area and very sparsely populated, with only 0.6% of the country’s total population. The rest af CoXcrmbia’s population (99,4%$ is concentrated mainly in the cities. Owing to this fact, the present study deals solely with the western portion of the ccru~rtry~ This part of the state is divided i&o four natural admir&rakz regions and 23 sub-regions, all of which are quite different from one another with respect to ~o~~ation~ physical c~~r~c~er~st~c~ ~~o~gr~~~~~ geography, &mate), and ethnic b~~~~r~~~~ of g&e ~~~~~~~~~~~ (see Fig. 1 and Appendix ES).

Page 3: Rural-to-urban and urban-to-urban migration patterns in Colombia

Migration Patterns in Colombia 135

REVIEW OF MIGRATION STUDIES

Migration research has concentrated primarily on examining why people migrate and which factors influence their decision to migrate. Although both economic and non-economic (psychological) factors are recognised as playing a role in migrants’ decision-making process, the refative influence of these factors varies from person to person, since migrants comprise an assortment of individuals of different backgrounds. The desire to translate the migration phenomenon into concrete theoretical and empirical models becomes necessary in order to provide a better understanding of migrant behaviour and to address important public-policy issues.

The celebrated model of rural-urban labour migration espoused by Todaros postulates that potential migrants consider various labour-market opportunities available to them in both rural and urban areas and compare their expected income in each location for a given time period in order to choose the one that is perceived to maximise their expected net gains.4 These expected gains, defined as the difference between the expected returns and the expected costs of migration, are measured as the difference in real income between rural and urban jobs and the probability of a new migrant’s obtaining an urban job. Labour will continue to migrate from rural to urban areas if their perceived expected income in the urban area, net of the cost of migration, exceeds their perceived expected income in the rural area.

The level of urban unemployment is determined by an equilibrium condition that states that excessive migration in the presence of rising rates of un- employment leads to a reduction in the urban-rural wage differential; the result is that urban incomes fall to a point at which no further migration takes place. An alternative equilibrating mechanism was presented by Stiglitz,s who suggests that as the number of unemployed increases, the expected income of job seekers in the urban sector falls.6

In view of the fact that many Third-World countries have witnessed high migration rates into high-unemployment urban areas, the efficiency of Todaro’s model for predicting migration was challenged. Harris and Todaro’ recognised that when making a decision to migrate, the individual balances the probabilities and risks of being underemployed, or even unemployed, for a given period of time until one succeeds in finding a permanent job in the formal sector.8 Studies by Hay,9 Barnum and Sabotio and Oberai,” among others, provide evidence that migrant urban incomes tend to rise rapidly, especially during the first few years in the city. l2 Following Harris and Todaro other studies subsequently expanded the model, utilising more sophisticated econometric techniques and complex migration functions. 13 Fields and Hosek proposed a framework for interpreting turnover that characterises the job-allocation mechanism as a first- order Markov process in which the probability of being hired if unemployed and fired if employed is constant over time. When this model is applied to migration, the expected earnings gained from migration become a function of W [P,,/(r + Pue + P,,)J(l -i- r)/r, rather than W (1 - r/)/r, where r is the discount rate, W the real wage gain in urban as compared with rural employment, U the urban unemployment rate, and Peu and P,, are the probabilities of being fired if employed and of being hired if unemployed during the reference period, respectively.

Todaro’s work dealt only with the rational behaviour of an individual who decides for himself if and when to migrate. A different approach will emerge from the rational behaviour of a family member when the decision is reached collectively by the entire family. The family provides financial support, thus reducing the adverse effects of being temporarily under- or unemployed; the result is a higher level of migration.14

Page 4: Rural-to-urban and urban-to-urban migration patterns in Colombia

136 Daniel Shefer and Luis Steinvorrz

In addition to the expected returns, other factors may influence the decision to migrate. Distances separating origin and alternative destinations and the potential migrants’ personal characteristics, such as age,15 sex and education level, are also very likely to affect the probability of migrating.16 Distance as a measurement of friction of space could serve as a proxy for both the transportation cost (direct monetary costs) and psychic costs (indirect or non- monetary costs) as well as the psychological costs of adjustment associated with moving away from family, friends and familiar surroundings.17 Distance, however, may affect potential migrants differently in accordance with each one’s own unique personal characteristics and circumstances.ls

Environmental conditions, like climate, housing inventory, and improved social services, are among the non-economic variables that were also found to affect the potential migrants’ decision-making process.*9

The process describing the multifa~ous aspects of the potential migrant’s decision-making process is vividly depicted in Fig. 2.

Expected present value of migration

r decision *

Fig. 2. Flow chart of factors influencing the decision to migrate.

Page 5: Rural-to-urban and urban-to-urban migration patterns in Colombia

Migration Patterns in Colombia 137

DATA SOURCES AND LIMITATIONS

Although micro-data are considered to be more appropriate to the study of the migration phenomenon, there is a severe shortage of this sort of data, particularly in developing countries, thus making it necessary to conduct migration analysis with macro-data. Whereas studies utilising micro-data focus on the role played by migrants’ personal characteristics, together with the attributes of their origin and destination, macro-studies emphasise the influence of the socio-economic and demographic attributes of the origin and destination on the decision to migrate. The present study utilises macro-data in the empirical analysis.

In view of the fact that approximately 50% of Colombia’s population resides in the country’s major cities ,20 it is important to identify the factors affecting the migration flow to these localities.

The study data were obtained from the 1985 census, which includes only in-migration figures to all the major cities in the western part of the country recorded for 1980 and 1985. The role played by socio-economic and demographic attributes of the origins and destinations on the direction and intensity of the migration flow in Colombia will be examined. It should be noted, however, that no socio-economic and demographic data are available for all the migrants who originated from all sub-regions, excluding the destination sub-region.

~ ,.’ Ll _I,- ’ - Major city II. (RU2)

Migration from

I. (RUl) other sub-regions

Migration from the same sub-region (RUl) Ezl

- Rural area (RZ)

I - Rural area (RI) - Other major cities (U21)

- Other cities (Ul) - Other cities (U22)

Fig. 3. Migration to the major cities, analysis of 1985 census data.

The 1985 census recorded the number of migrants who remained within the borders of their origin sub-region and the number of those who migrated from

Page 6: Rural-to-urban and urban-to-urban migration patterns in Colombia

138 Daniel Shefer and Luis Steinvortz

other sub-regions. In addition, it provided information about the rural or urban sector from which the migrants originated. The migration flow to the main city in the same sub-region, RUI , is indicated as R1 and UI when it is from the rural and the urban sector, respectively. When the migration flow is to the city of another sub-region, RU2, it is indicated accordingly as R2 and U2 (see Fig. 3). Thus, the migration flow to large cities, RU, may be represented as follows:

RU = RUl + RU2, (1)

where RUI represents migration from within the same sub-region as j, and RU2 represents migration from other sub-regions than j. Thus:

RUI = RI + Ul (59

and

RU2 = R2 + U2, (3)

where U2 includes people migrating from other major cities (U21) and from smaller urban localities of other sub-regions (U22), such that:

u2 = U2I + u22. (4)

EMPIRICAL MODELS

Owing to the nature of the 1985 data, this study will confine itself to data on in-migration flow to the main 23 cities in the most populated sub-regions of Colombia. This restraint limits the number of independent variables that can be included in the statistical models.

Based on demographic and socio-economic characteristics that were found in past migration studies to be significant predictors of migration flow as well as on data availability, the following independent variables were used:

POP - population at origin i and at destination j, WAGE - average wage or income level in i and j, HOUSE - rate of unoccupied housing in i and j, UNEMP - rate of unemployment in i and j, LITPOP - rate of literate population in i and j,

where i denotes the origin of both rural and urban sectors and j refers to the major cities as destination.

HYPOTHESES

Several fundamental hypotheses were postulated and tested in the present study:

Population

Population sizes as a surrogate for economic opportunities at both the urban and rural origins and at the destination are likely to affect the flow of migration. The larger and more populated a city, the greater will be the inflow of migration. This is so because potential migrants perceive fewer risks and greater opportunities in

Page 7: Rural-to-urban and urban-to-urban migration patterns in Colombia

Migration Patterns in Colombia 139

moving to a large urban centre. It is also postulated that population size at origin i will either encourage or discourage migration to i. Population size at i may encourage migration to i when i grows at a faster rate than do its employment opportunities, thus compelling its inhabitants to seek jobs in other localities. Alternatively, a large population size at i implies the availability of public and private services and an ever-growing supply of employment opportunities, thus reducing the desire to migrate. It should, however, be mentioned here that a wide variation in the size of the population of the observations in the samples could inadvertently bias the statistical estimates.

Income

Although a relatively higher income at destination j is expected to induce migration from either rural or urban localities, the effect of the relative income level at origin i, in both rural and urban sectors, may be looked at from two different viewpoints: (1) with regard to the effects of average (or median) income at i, it is hypothesised that a low flow of out-migration will ensue from an origin with a relatively low average income. The reason for this is the potential migrant’s lack of sufficient financial resources needed to accomplish the move. (2) A relatively high average income at origin i may, on the one hand, satisfy the prospective migrant’s aspirations and act as a deterrent to migration; on the other hand, income level could provide the potential migrant with the financial means required to pursue greater opportunities, i.e. income, in other locations.21

Rate of unoccupied housing

The rate of unoccupied housing of the total housing stock is expected to affect the migration decision in the following ways: a high rate of unoccupied housing in i could be associated with a low level of employment opportunities; thus a large out-migration is expected from those areas. Although unoccupied housing units are a consequence of a time-log in a difference equation model, it is a determinant variable in a no-time-log model.

The rate of unoccupied housing in J’ is hypothesised to affect migration in two opposing directions: (1) a low rate could act as an inducement factor to potential migrants because that market condition may suggest a viable, dynamic economy; (2) a high rate could be associated with a lack of employment opportunities, thus inhibiting the flow of in-migration to the locality.

Unemployment

Unemployment, at both the origin and the destination, is likely to affect the individual’s decision to migrate, and thus affect the migration flow. A high rate of unemployment in i, whether it is a rural or an urban sector, is expected to induce out-migration from i to i, provided that the prospective migrant has the financial means or support required for the move and, therefore, will be able to look for employment opportunities in another location. On the other hand, a migrant’s lack of financial resources, in spite of a high rate of unemployment in i, is expected to act as an obstacle to migration.

Unemployment in i is expected, prima facie, to discourage in-migration although there are migrants who, because of inadequate or insufficient informa- tion, may still decide to migrate toj. Furthermore job turnover at the destination could affect differentially the potential migrant’s perception of the rate of unemployment.

Page 8: Rural-to-urban and urban-to-urban migration patterns in Colombia

140 Daniel Shefer and Luis Steinvoriz

Education

In the case of migration to a major city, the level of education at the origin is expected to influence the migration flow positively. As the rate of the literate population in i grows, so will the flow of migration out of i and into i. Educated individuals tend to believe that there are higher-paying job opportunities for them in the major city.

The influence of the literacy rate at the destination will be such that as the rate increases, i will become more attractive because it is likely to offer more in the way of living and cultural amenities as well as higher-paying employment opportunities.

THE GENERAL MODEL

The basic migration model utilised in the present study links migration flow to the major cities, i, with the economic and demographic variables in both i and i, where i refers to both rural and urban sectors of the same sub-region as well as to rural and urban sectors of other sub-regions. With respect to the different origins, the possible statistical models may be divided into two main groups:

(1) The first group of models is concerned with migration to the main city of the sub-region from the same sub-region. These models were estimated independently for rural and urban migration flows as well as for the joint migration flow:

MODEL 1.1: RI = f (X,, X,)

MODEL 1.2: #VI = f (Xc, X,)

MODEL 1.3: RUI = f (X,, X,, X,, Dummy),

where the dependent variables denote the number of migrants (gross migration) moving into the sub-region’s main cities. It should be noted, however, that a more appropriate dependent variable is the rate of migration which indicates the probability of migrating from origin i to destination i. In the present study, it was impossible to compute the probability of migration between i and i because of insufficient data.

The independent variables here are the vectors, X,, X, and X,, which are, respectively, the characteristics of the major destination city of the sub-region, the rural origin sector, and the urban sector of the same sub-region.

These vectors are composed of the following characteristics:

(a) XC = (CPOPj, CWAGEj, CHOUSEj, CUNEMPj, CLITPOPj),

where: CPOPi = population of the main city, CWAGEi = average income in the manufacturing sector of the main city, CHOUSEj = rate of unoccupied housing in the main city, CUNEMPj = rate of unemployment in the main city, CLZTPOPi = rate of literate population of the main city.

(b) X, = (RPOPi, RWAGE, RHOUSE, RUNEMP, RLITPOP,),

Page 9: Rural-to-urban and urban-to-urban migration patterns in Colombia

where: RPOPi RWAGE, 1 RHOUSEi = RUNEMP, = RLITPOP, =

Migration Patterns in Colombia

rural population of the sub-region of the origin, average rural income in the agricultural sector, rate of unoccupied housing in the rural sector, rate of unemployment in the rural sector, rate of literate population in the rural sector.

141

(C) & = ( UPOPi, UWAGE, UHOUSE, UUNEMPi, ULZTPOPi),

where: UPOP, = urban population of the sub-region of the origin, USAGE, = average urban income in the manufacturing sector, UHOUSE, = rate of unoccupied housing in the urban sector, UUNEMP, = rate of unemployment in the urban sector, ULZTPOPi = rate of literate population in the urban sector.

Model 1.3 also includes a dummy variable. Thus when the migrant originates from rural localities, a value of 1 is assigned; from urban localities, a value of 0.

Based upon the hypotheses postulated above, it is expected that the values of the partial derivates of the migration flows will be of the following signs with respect to each of the specified variables:

aRu >. aRu >. aRU >

amop ‘aCWAGE aRU <o

i ’ aCHOUSE<” aCUNEMP

aRu

* acmpop > 0;

i i i i

aR1 >o aR1 > aR1 >o

aR1

aRPoP 20

aR1

’ > 0;

i ’ aRWAGE<O’ aRHOUSE

i i ’ aRUNEMP< aRLITPOP

i i

au1 >. aUPOT<

au1 >. au1 ’ aUWAGE<

au1 >o ’ aUHOUSE

‘0 au1 >().

i ’ aUUNEMP< ’ aULITPOP ’

i i

where the hypotheses concerning the characteristics of the major cities (X,) are valid for all the models dealing with these cities. R and U (RU) refer to all the different migration flows from rural, R, and urban, U, localities to the major cities.

If it is assumed, in recognition of the recent surge in urban-to-urban migration that most of the migrants to the major cities have migrated from other cities, then the sign associated with the dummy variable is expected to be statistically significant and positive; i.e. DUMMY > 0.

(2) The second group of models deals with the flow of migration to major cities from rural and urban sectors of other sub-regions thanj. The dependent variable in these models is the total number of migrants from the other sub-regions, and the independent variables pertain only to the characteristics of the main cities as destination.

Page 10: Rural-to-urban and urban-to-urban migration patterns in Colombia

142 Daniel Shefer and Luis Steinvortz

MODEL 2.1: RU2 = f(X,, Dummy).

This model also includes a dummy variable in order to identify the relative importance of rural and urban migration. The independent variables are those characteristics of the destination as was seen in the previous group of models.

Since urban migrants from other sub-regions may come from other major cities (U21) or simply from other small urban areas (U22), three additional models were included in the analyses in order to identify the effect of the characteristics of the cities on the migration flow. These models consider the two different sources of urban migration flows separately, and then jointly take the total number of in-migration flow from urban localities, U2, to the major cities:

MODEL 2.2: U21 = f (X,)

MODEL 2.3: U22 = f (X,)

MODEL 2.4: u2 = f(X,, L)ummy).

X, refers to a vector of independent variables characterising the destinations - the major cities. The dummy variable pertains to the migrants from other main cities in order to differentiate them from migrants originating from smaller urban areas.

FINDINGS

The findings presented in this section are based on the log-linear regression models used in the statistical estimation procedures. Tables 1 and 2 report the results obtained from the statistical estimations of the coefficients of the independent variables included in each of the models, along with their Ievel of significance and the percentage variance explained by the models (i.e. R*).

Population size at destination i was found to associate positively with in- migration. Therefore, the more populous a city, the more it seems to attract in-migration.

When i and i exist in the same sub-region, the higher the urban-rural income ratio, CWAGEj/RWAGEi, the greater is the flow of migration from i to i. This result confirms the hypothesis postulated by Todaro.22

The rate of unemployment in i, when i refers to the rural sector in the same sub-region asj, has a negative effect on the migration flow. That is, as the rate of unemployment in i increases, the flow of migrants toi decreases. In this instance, it would appear that the lack of financial means resulting from unemployment makes the act of migration from rural areas to the cities an impossibility. The rate of unemployment in i inhibits the migration flow from other major cities. This finding may be attributed to the fact that an individual considering migration from one major city to another is well informed about j, and thus not likely to move to a destination experiencing a high rate of unemployment.

The fact that the dummy variable was found to be statistically significant suggests that a larger proportion of migrants to the main cities have arrived from other urban areas. Similarly it was found that most urban migrants from other sub-regions come from the major cities of those sub-regions rather than from small cities.

Page 11: Rural-to-urban and urban-to-urban migration patterns in Colombia

MigFution Patterns in Colombia 143

* h h

Page 12: Rural-to-urban and urban-to-urban migration patterns in Colombia

Tab

le 2

. G

roup

2:

mod

els

of m

igra

tion

fro

m

othe

r su

b-re

gion

s th

an j

Mod

el 2

.1:

RV

2 =

f(

Xc,

Dum

my

) M

odel

2.

2: U

21 =

f(

X,

) M

odel

2.

3: U

22 =

f(

Xc

) D

epen

dent

M

odel

2.

4: V

2 =

f(

X,,

Dum

my

)

vari

able

I

I II

I

II

I In

depe

nden

t B

-val

ue

B-v

alue

B

-val

ue

B-v

alue

B

-val

ue

B-v

alue

va

riab

le

( r-

valu

e )

( t-

valu

e )

(r-v

alue

)

( t-

valu

e )

(t-v

alue

)

( t-

valu

e )

In C

POPi

0.

989

* 0.

944

* -0

.010

1.

010

* (

7.25

8)

In c

pop,

(

7.85

4 )

( -0

.087

)

In C

PO

Pi

1.08

1 *

0.13

0 In

CP

OP

, (

5.93

6 )

( 0.

668)

(

9.41

5 )

In C

WA

GE

j 0.

097

( 0.

269)

In C

HO

VSE

, -0

.507

(

-1.0

92 )

In C

VN

EM

Pj

-0.4

48

( -1

.343

)

In C

LITP

OP

, -0

.106

(

-0.0

61)

In D

UM

MY

1.75

7 *

( 10

.086

)

In C

WA

GE

j -

- -

-

In C

HO

USE

, -0

.003

0.

029

( -0

.008

)

( 0.

071

1

In C

VN

EM

P,

-0.6

30

* -0

.722

(

-2.2

06

) (-

2.51

1

In C

LITP

OP

i 2.

870

3.04

2 (

1.84

3)

( 1.

942

In C

WA

GE

j -0

.589

-0

.642

In

CW

AG

E,

-0.2

60

( -1

.213

)

( -0

.274

)

(-0.

~1

In C

HO

USE

,

1 In

RV

NE

MP

)

In C

LITP

OP

,

Con

stan

t -3

.584

C

onst

ant

-14.

478

-15.

570

Con

stan

t

-0.1

13

-0.1

83

)

-0.6

21

-1.3

95

)

-0.6

31

-0.2

70

)

-0.0

87

In C

HO

VSE

i

( -0

.135

)

-0.7

27

In C

UN

EM

P,

( -1

.575

)

-0.4

94

In C

LITP

OP

, (

-0.2

04

)

In D

UM

MY

6.20

4 5.

989

Con

stan

t

-0.0

55

(-0.

151

)

-0.6

16

* (

-2.3

48

)

1.14

2 (

0.82

8 )

0.70

4 *

( 5.

133)

-5.0

01

(-0.

406)

(

-2.3

16

) (

-2.4

75

) (

0.52

6)

( 0.

482

) (

-0.7

20

)

N

45

N

22

22

N

22

22

N

45

R*

0.86

2 R

2 0.

903

0.35

1 R

2 0.

882

0.16

0 R

2 0.

863

F

40.6

11

F

42.1

35

2.43

9 F

15.7

65

0.64

9 F

41

.104

Tw

o ve

rsio

ns

of e

ach

mod

el

wer

e te

sted

: I,

whe

re

the

depe

nden

t va

riab

le

is t

he g

ross

mig

ratio

n;

II,

whe

re

the

depe

nden

t va

riab

le

is t

he

rate

of

mig

ratio

n.

* R

egre

ssio

n co

effi

cien

ts

stat

istic

ally

si

gnif

ican

t at

the

95%

lev

el.

The

fig

ure

in p

aren

thes

is

is t

he t

-val

ue.

Page 13: Rural-to-urban and urban-to-urban migration patterns in Colombia

Migration Patterns in Colombia 145

A graphical representation of the migration statistics depicts the proportion of rural-to-urban migrants as well as urban-to-urban migrants (see Fig. 4); of the total number of migrants to the major cities, 75% arrived from an urban context; of these, 46.3% came from cities in other sub-regions. In contrast, only 28.7% moved to the major city of their own respective sub-region. It is possible to conclude from this result that most migrants eventually reach the large main cities of the country only after a process of first moving from a rural village to a small urban area.

Urban (75.0%

RU2

RU2 Ul (54.9%) Migration from other sub-regions

u21 (30.8%)

Fig. 4. Sources of migration to major cities.

Because the 1985 census data do not provide info~ation about the migrants’ place of origin, distance could not be used as an explanatory variable in the statistical analysis. Nevertheless, certain conclusions can be drawn from the statistics available. The data disclose that 45.3% of the migrants to major cities originate from the same sub-region as that of i; 54.1% originate from outside the sub-region, 42% of these coming from the Same region as that of j (see Fig. 5). A mere 0.6% of the migrants were found to originate from outside Colombia.

1. Migration from the

2. Migration from the

same sub-region

3. Migration 4. Migration from other from other countries regions

Fig. 5. Sources of migration to major cities, based on grouping by d&tame.

Page 14: Rural-to-urban and urban-to-urban migration patterns in Colombia

146 Daniel Shefer and Luis Steinvortz

CONCLUSIONS

The findings obtained from the statistical analyses confirm some of the hypo- theses postulated in the present study and validate the economic models used in explaining inter-regional migration in Colombia. The following variables, in order of importance, were found to affect migration flow in Colombia: population size at the destination, relative income level at the destination, and unemployment at both origin and destination.

Population size at destination was found to affect migration flow positively. That is, the larger the city, the greater is the flow of in-migration. The fact that migrants make their way to the larger cities reveals that they consider population size to be an important criterion by which to estimate the expected returns at a particular destination.

The relative income variable, or income ratio, was found to be positively associated with migration flow. If the influence of the urban-rural wage differential is considered, it was found that when urban exceeds rural income, a greater flow of rural-urban migration ensued.

As far as the urban-urban migration flow is concerned, neither the level of income at the origin nor that at the destination was found to have any statistically significant effect on the pattern of migration flow.

When the origin is a rural sector, the rate of unemployment in an urban destination was not found to affect migration flow (see Appendix C). This fact reinforces Stark’s hypothesis,23 which states that despite a rate of high unemployment in the city, rural migrants who have family support are not discouraged from trying their luck in such a destination. Another explanation may be that rural migrants are simply uninformed as to what really awaits them in the city.

When the origin is an urban sector, the rate of unemployment in the destination was found to affect migration negatively. This result may suggest that, compared with a rural migrant, an urban migrant is less likely to take risks; or perhaps urban migrants are better informed of the unemployment situation at the alternative destinations.

In addition to showing the effects attributed to each of the variables discussed above, the present study contributes to an understanding of the pattern of migration currently taking place in Colombia. In the past, the direction of most migration flows was from the rural sector to the cities; in contrast, today’s migrants, although still directed toward the large cities, originate primarily from smaller cities. This relatively new phenomenon of urban-to-urban migration flow points to the more advanced stage of urbanisation that nowadays prevails in Colombia, where approximately 70% of the population is found to reside in cities. This finding is indeed significant and important, and therefore deserving of

1.

2.

3.

4.

further study.

REFERENCES

Departamento Administrativo National de Estadistica, Colombia Esfadistica 86 (Bogota, Colombia, 1985). Departamento Administrativo National de Estadistica, Boletin de Estadistica Abril-Junio (Bogota, Colombia, 1985). Departamento Administrativo National de Estadistica, XV Censo national de poblacion y IV de vivienda, Vols l-7 (Bogota, Colombia, 1986). L. Steinvortz and D. Shefer, “Determinants of Inter-regional Migration in Colombia”, Research Report No. 117 (Center for Urban and Regional Studies, Technion, Haifa, Israel, 1987). M.P. Todaro, “A Model of Labor Migration and Urban Unemployment in Less Developed Countries”, American Economic Review 59 (1969), pp. 138-148. S. Bowles, “Migration, as Investment: Empirical Tests of the Human Investment Approach to Geographical Mobility”, Review of Economics and Statistics, pp. 356-362 M. Lipton, “Migration from Rural Areas of Poor Countries”, in: Migration and the Labor Market in Developing Countries (Edited

Page 15: Rural-to-urban and urban-to-urban migration patterns in Colombia

Migration Patterns in Colombia 147

by R.H. Sabot) (Westview Press, Boulder, CO, 1982) pp. 191-228. A. Schwartz, “On the efficiency of Migration”, Journal of Human Resources 6, No.2 (1971), pp. 193-205. L.A. Sjaastad, “The Costs and Returns of Human Migration”, Journal of Political Economy (1962), No. 5, pp. S&93.

5. J.E. Stiglitz, “The Structure of Labor Markets and Shadow Prices in LDCs”, in: Migration and the Labor Market in Developing Countries (Edited by R.H. Sabot) (Westview Press, Boulder, CO, 1982). pp. l-3-63.

6. See also J.E. Stiglitz, “Rural-Urban Migration, Surplus Labor and the Relationship Between Urban and Rural Wages”, Eastern Africa Economic Review (1969), pp. l-27. J.E. Stiglitz, “Alternative Theories of Wage Determination and Unemployment in LDCs: the Labor Turnover Model”, Quarterly Journal of Economics 88 (1974). pp. 194-227.

7. J. Harris and M.P. Todaro, ‘*Migration, Unemployment and Development: a Two Sector Analysis”, American Economic Review 60 (1970), pp. 126-142.

8. G. Fields, “Rural-Urban Migration, Urban Unemployment and Underemployment and Job Search Activity in LDCs”, Journal of Development Economics 2 (1975), pp. 165-187.

9. M.J. Hay, “An Economic Analysis of Rural-Urban Migration in Tunisia”, Unpublished Ph.D. dissertation, University of Minnesota (1974).

10. H. Barnum and R.H. Sabot, “Education, Employment and Rural-Urban Migration in Tanzania”, Oxford Bulletin of Economics and Statistics 39, No. 2 (1977).

11. A.S. Oberai, “An Analysis of Migration to Greater Khartoum (Sudan)“, paper for restricted distribution only, mimeograph (ILO, World Employment Programme, Geneva, 1975).

12. M.P. Todaro, _‘* Internal Migration in Developing Countries: a Review of Theory, Evidence, Methodology, and Research Priorities” (International Labor Organisation, Geneva, 1976). p. 31.

13. G.S. Fields and J.R. Hosek, “Human Investment Decisions, Labor Market Choice, and Unemploy- ment”; paper presented at the December 1973 meeting of the Econometric Society, New York (1973). G. Fields, “Labor Force Migration, Unemployment and Job Turnover: Test of a Markovian Approach”, Review of Economics and Statistics 28, No. 3 (1976). pp. 407-415. G.E. Johnson, “The Structure of Rural-Urban Migration Models”, Eastern Africa Economic Review (1971), pp. 21-28.

14. 0. Stark. “On the Ootimal Choice of Canital intensitv in LDCs with Miaration”. Journal of Development Economics 9 No. 1 (1981). pp. ‘123-132. 0. Stark. “Towards a Theory of Remittances in LDCs” (Harvard Institute of Economic Research, Discussion Paper Series, 1982). 0. Stark, “A Note on Modelling Labor Migration in LDCs”, Journal of Development Economics 19, No. 4 (1983). pp. 539543.

15. Sjaastad, 1962 (see note 4). 16. Schwartz, 1971; Lipton, 1982 (see note 4). G. Fields, “Place-to-place Migration in Colombia”,

Economic Development and Cultural Change 30, No. 3 (1982), pp. 539-558. M.I. Greenwood, “Research on International Migration in the United States: a Survey”, Journal of Economic Literature 13 (1975), 397-433. M.B. Levy and W.J. Wadicky, “The Influence of Family and Friends on Geographic Labor Mobility: an International Comparison”, Review of Economics and Statistics (1973) pp. 198-203. T.P. Schultz, “Notes on the Estimation of Migration Decision Functions”, in: Migration and the Labor Market in Developing Countries (Edited by R.H. Sabot) (Westview Press, Boulder, CO, 1982a) pp. 91-126. T.P. Schultz, “Lifetime Migration within Educational Strata in Venezuela: Estimates of a Logistic Model”, Economic Development and Cultural Change 30 (1982b), pp. 559-593. A. Schwartz, “Interpreting the Effect of Distance on Migration”, Journal of Political Economy, (1973), pp. 1153-l 169.

17. See Schwartz, 1973; Greenwood, 1975; Schultz, 1982b (ibid.); B. Deaton, L. Morgan and K. Anschel, “The Influence of Psychic Costs on Rural-Urban Migration”, American Journal of Agricultural Economics (1982), pp. 177-187.

18. See Swartz, 1973; Schultz, 1982b (ibid.); 0. Stark and D. Bloom, “The New Economics of Labor Migration”, American Economics Review, Papers and Proceedings 75 No. 2 (1985), pp. 173-178.

19. R.J. Cebula and R.K. Vedder, “A Note on Migration, Economic Opportunity, and the Quality of Life”, Journal of Regional Science 13 No. 2 (1973), pp. 205-211. R.J. Cebula and R.K. Vedder, “A note on Migration, Economic Opportunity, and the Quality of Life: Reply and Extension”, Journal of Regional Science 16 (1976). no. 113-115. P. Graves, “A Reexamination of Migration. Economic Opportunity, and Quality’ of ‘Life”, Journal of Regional Science 16, No. 1 -(1976), pp. 107-112. P. Graves, “Migration and Climate”, Journal of Regional Science 20 (1980), pp. 227-237. F.W. Porell, “Intermetropolitan Migration and Quality of Life”, Journal of Regional Science 22 (1982). pp. 137-158.

20. Major cities refer to the country’s regional capitals. 21. G. Feder, “On the Relations between Origin, Income and Migration”, Annals of Regional Science

16, No. 2 (1982). pp. 46-61. D. Shefer, “The Effect of Price Support Policies on Inter-regional and Rural-Urban Migration in Korea, 1967-1980”, Journal of Regional Science and Urban Economics (1987) No. 3, pp. 333-344.

22. Todaro, 1969 (see note 3). 23. Stark, 1981-1983 (see note 14).

Page 16: Rural-to-urban and urban-to-urban migration patterns in Colombia

148 Daniel Shefer and Luis Steinvortz

APPENDIX A

Rural-urban population distribution, 1964, 1973 and 1985

Sub-region 1964 1973 1985

(& (2) (& (& (& (&

1.1. La Guajira 29.9 70.1 41.7 58.3 69.1 30.9 1.2. Magdalena 44.3 55.7 50.1 49.9 53.1 46.9 1.3. Cesar 37.4 62.6 54.2 45.8 63.2 36.8 1.4. Atlantic0 90.9 9.1 93.9 6.1 94.3 5.7 1.5. Bolivar 57.8 42.2 63.7 36.3 67.2 32.8 1.6. Cordoba 30.7 69.3 38.5 61.5 45.2 54.8 1.7. Sucre 41.1 58.9 51.9 48.1 55.8 44.2

2.1. N. de Santander 49.2 50.8 52.9 47.1 67.4 32.1 2.2. Santander 43.9 56.1 53.5 46.5 62.0 38.0 2.3. Boyaca 24.1 75.9 30.3 69.7 37.3 62.7 2.4. Cundinamarca 28.8 71.2 36.4 63.6 46.2 53.8

Bogota 97.9 2.1 99.5 0.5 99.7 0.3 2.5. Meta 47.0 53.0 57.7 42.3 67.5 32.5

3.1. Antioquia 3.2. Caldas 3.3. Risaralda 3.4. Quindio 3.5. Tolima 3.6. Huila 3.7. Caqueta

4.1. Choco 4.2. Valle 4.3. Cauca 4.4. Narino

53.4 46.6 62.8 31.2 68.0 32.0 50.1 49.9 55.5 44.5 66.6 33.4 57.0 43.0 65.0 35.0 72.5 21.5 68.2 31.8 70.0 30.0 82.8 17.2 42.1 57.9 51.4 48.6 57.6 42.4 43.1 56.9 49.7 50.3 55.6 44.4 23.6 16.4 28.1 71.9 26.8 73.2

23.4 76.6 28.6 71.4 35.3 70.4 29.6 77.3 22.7 82.5 23.2 76.8 32.4 67.6 39.7 30.4 69.6 37.0 63.0 42.4

52.0 48.0 59.0 41.0 67.3

64.7 17.5 60.3 57.6

Total 32.7

Source: The Colombian Central Bureau of Statistics (DANE, Bogota, 1986).

Page 17: Rural-to-urban and urban-to-urban migration patterns in Colombia

AP

PE

ND

IX

B

Pop

ulat

ion

of su

b-re

gion

s of

w

este

rn C

olom

biu

und

thei

r m

ajor

ci

ties,

19

73 a

nd

1985

Reg

ion

Sub-

regi

on

Are

a (k

m’)

19

73

1985

M

ajor

ci

ty

--...

““_

_ _...,

~ 19

73

1’38

5

I. T

he

Car

ibbe

an

regi

an

2. T

he

Eas

tern

re

gion

The

And

es

regi

on

3. T

he

Cen

trat

re

gion

4. T

he

Paci

fic

regi

on

Tot

al

1.1.

La

Gua

jira

20

,848

18

1,77

1 1.

2. M

agda

lene

23

,188

54

0,25

8 I.

3 C

esar

22

,905

34

0,65

7 f-

4. A

tfan

tico

3&8&

96

4,08

7 1.

5. B

ofiv

ar

25,3

78

8 f7

,838

1.

6. C

ordo

ba

25$x

%

649,

462

1.7.

Suc

re

10,9

27

352.

369

2.1.

N.

de S

anta

nder

21

,458

70

3,04

1 2.

2. S

aman

der

30,6

37

1,12

7,99

9 2.

3. B

oyac

a 23

.189

99

2,17

7 2.

4. C

undi

nam

arca

24

,210

3,

697,

190

2.5.

Met

a 85

,635

24

2.66

4

3.1.

Aot

ioqu

ia

3.2.

Cal

das

3.3.

Ris

arai

da

3.4.

Qui

ndio

3.

5. T

ofim

a 3.

6. H

uila

3.

7. C

aque

ta

4.1.

Cho

co

4.2.

Vaf

fe

4.3.

Cau

ca

4.4.

Nar

ino

63,6

12

2,96

5,11

6 78

88

698,

042

4140

45

5,66

7 I&

54

322,

825

23,6

52

fros

&@

9 19

,890

46

7,65

1

88,9

65

180,

291

Med

ellin

M

ar&

ales

Pe

reir

a A

rmen

ia

ibag

ue

Nei

va

Pfor

enci

a

46,5

30

203,

635

242,

768

Qui

bdo

22,1

40

2,18

6,80

1 2,

847,

087

Caf

i 29

,308

58

2,70

9 79

5,83

8 33

,268

Po

paya

n 80

9,17

8 1,

019,

098

Past

0

20.3

87,0

37

27,2

S6,

008

Rio

hach

a Sa

nta

Mar

ta

Vaf

fedu

par

Bar

r~qu

~ffa

C

arta

geG

a M

onte

&i

Sinc

efej

o

Cuc

uta

Buc

aram

anga

T

unja

B

ogot

a D

. E

. V

iffa

vice

ncio

39,5

08

76,9

43

i27,

755

218,

205

t12,

057

192,

044

731,

OJt

*

X,f

J7,f

50

* 31

2,55

7 53

1,42

6 I5

4,59

9 22

4,14

7 36

,190

I%

,@7

310,

426

* 44

3,0%

3 *

361,

799

* 59

5.O

Q6 *

79

,391

93

,792

25

71,5

48

3,9&

2,94

1 91

,559

17

8,68

5

1,51

7,94

4 *

2,09

5,14

7 *

245,

887

* 32

7,77

8 *

253,

736

* 38

9,47

9 *

145,

341

1g7,

L%

f 20

8,69

9 29

2‘96

5 12

1,11

0 19

4,55

6 49

,101

79

,525

49,6

37

75,5

24

952,

121

* 1,

400.

828

* 91

,124

15

8,33

6 14

7,77

9 24

4,70

0

8,75

0,70

9 13

,255

,252

Tot

al p

opul

atio

n of

Col

ombi

a 22

,915

,229

27

,837

,9X

&xw

ce:

Tfz

e C

ofam

b~aR

Cen

traf

B

urea

u of

Sta

tistic

s (D

AN

E,

Bog

ota,

19

86)

* Po

pufa

t~o~

of

~~et

ropo

i~ta

~ A

reas

,

Page 18: Rural-to-urban and urban-to-urban migration patterns in Colombia

150 Daniel Shefer and Luis Steinvortz

APPENDIX C

Group 2: models of migration from other sub-regions than j

Model R2 = f (Xc) Dependent I II

variable

independent B-value variable ( t-value )

B-value ( t-value )

In CPOP,

In CWAGE,

In CHOUSE,

In CUNEMP,

In CLITPOP,

Constant

N R’ F

0.993 * 0.042 ( 4.119) ( 0.177)

0.355 0.303 ( 0.553 ) ( 0.475 )

-0.996 -0.969 (-1.211 ) (-1.188 )

-0.298 a.404 ( -0.505 ) ( -0.691 )

-1.831 -1.694 ( -0.591 ) ( -0.551 )

1.623 1.317 ( 0.104) ( 0.085)

22 22 0.746 0.162 9.9% 0.660