. cSaldru'WorkinJ Paper No. 43 - openSALDRU

·,

,I

cape Town

Southern Africa Labour and DeveloplEnt Research Unit

THEEC<:NCMIC IMPLICATIONS OF AFRICAN RESEl'TLEMENT

. cSaldru'WorkinJ Paper No. 43

Septenber 1981

I II

1

", I' I,'

I ~

"

This paper was presented at a conference on resettlement held by the South African Institute of Race Relations in Johannesburg between 24 _ 26 June 1981. We are very grateful for their permission to republish it as a Saldru Working Paper.

1. Official estimates of Africans resettled 1960-79 between 3 and 4 2. Summary tctals of removals from rural areas 1960-1979 3

3. Net migration between region types, 1960-1980 4

4. Estimate of number of people in closer settlements,

1980 7

5. ' Expected' and actual sectoral shares of employment in South Africa, 1960, 1970 and 1980 9 6. Dates of arrival, reasons for coming and facilities

found by sample of households in eight areas, 1980-1 12 7. Labour force participation, errployment and unerrployrnent

in eight areas, 1981 13 - 17

8. Migrancy and de facto headship in eight settlements 19 9. Households with no earners (resident or migrant) by

headship and reliability of remittances from migrant

workers in eight settlements 20

10. Skill composition of the male workforce in eight

closer settlements 23

11. Homeland public expenditure and its financing, 1960-

78 24 - 25

12. Expenditure on resettlement by the S.A. Developnent

Trust, 1970/71 - 1976/77 27

The economic implications of African resettlement It would be a difficult task to give anything like a:

complete account of the economic implications of African resettlement for two main reasons:

(a) we lack key pieces of information at the macroeconomic and demographic levels. We do not have gross migration figures or a complete enumeration of people resettled; nor do we have an estimate of the number of people living in closer settlements. In the absence of macrodata, what one has by way of microdata becomes more difficult to interpret;

without macro ,bench marks, .one ~lways runs the, r~sk ,of placing inappropriate weight on the pieces of information one does have.

(b) understanding the economics of resettlement involves understanding a good deal else beside, including the determinants of African population movement (of which resettlement forms a part), homeland settlement patterns

(in particular, the rise of the closer settlement since, say, 1960), the evolution of the migrant labour system and the ability of the homeland public finance system to cope with rapidly rising homeland populations. Most of these topics (forming collectively many of the implications of separate development) have themselves been little studied.

Under these circumstances, i t would be rash to attempt to do more than sketch out some preliminary approaches to the topic in hand. These approaches will move between the issue of resettlement narrowly construed and related issues, partly because tracing out the links is important and partly (it must be admitted) because of informational gaps. The first section will attempt to piece together the little we do know about the demography of resettlement. The second will look at two important features of the South African economy against trends in semi-developed countries as a whole:

urbanisation and the structure of employment. The third will present information collected from eight closer settle- ments by the Surplus People's Project, whose permission to use the material is gratefully acknowledged. This information

unemployment; the fourth section will consider the question of whether there is a 'surplus' or 'marginalised' population in closer settlements. The fifth will talfe a look at home-

lan~ public finance in the light of rapidly rising- homeland populations and the last will draw such conclusions as are possible, given the diverse and fragmentary nature of the evidence available.

I

How many Africans have been resettled in- the last twenty years? Various answers have been given to this question; all accounts are ultimately founded on two sources.::--

(a) information supplied by the responsible Minister to Parliament 1-

(b)-- information contained in the annual- report.s of the responsible Department.

This information is compiled in Table T. Table I is a statistician's nightmare; the following defects are to be found in i t :

- there are a great many missing ¥alues;

usually information is given in terms of individuals;

sometimes i t is given in terms or families. Sometimes one does not know with what one- is dealing;- if 3 430-labour tenants were removed in I~63, does this mean 3 430 individuals or- 3 430 households of whom the head was. a labour tenant?

- figures are reported for different categories at different times and at the same time-._ It is not always clear what is included in each category; i f categories overlap, simple addition will involve an element of double-counting. The problem is particularly acute in the case of black spot removal and homeland consolidation, which, as one minister pOinted out, "must be seen as one interlocked undertaking"

(P.Q.I5 of I3 August I974J.

Despite these problems, i t is possible to make an estimate of the number of rural removals between 1960 and I979.

~---~----~---~--

3.

Year-by-year estimates and their derivation are reported at the foot of Table I and summary totals are presented in Table 2.

Table 2 Summary totals of removals from rural areas 1960-1979 (thousands)

I960-69 1970-79 I960-79 Labour tenants

and squatters 363 305 668

Black spot removals

and homeland consolidation 149 305 454

5I2 610 1 122

The situation as regards urban removals allows only more tentative comment. Removals from urban areas involve:

(a) Removals under sections I4 and 29 of the Urban Areas Act.

These constitute resettlement only~in rather an extended sense of the term. Figures in row (I6) of Table I reflect these for some, but not all, of the metropolitan areas in which one would expect the bulk (but not all) of such removals to take place;

(b) Removals following the deproclamation of African townships.

These deproclamations have gone on steadily since the I950s, but I know of no source of information revealing how many Afri- cans have been moved as a result. These deproclamations have occurred in non-metropolitan areas (small towns), sometimes involving removal to places far away, sometimes to homelands a,relatively short distance off, from which workers commute in order to continue their jobs. These type (b) removals are more like rural removals than type (a); they involve the movement of whole households who had previously enjoyed a certain security of tenure.

It is possible to provide some sort of context for these resettlement figures by comparing them with indirect estimates cif net migration of domestic Africans between 1960 and 1980.

Table 3 sets put these net migration estimates:

me_

~

~ A. h'hite fams (1) L~ur te. ... ^.mts (1) - cumulative figuresa

(from 1. 9.56) l2} - irrplicit annual figures (3) ~i~~;~r~igureSa

Official estimates of Africans resettled, 1960-1979

,-"" :~ ::~ :: :: = :~: :: I,: n. Dn I "n I "n I m'l D~ I"~ "" I "" I Dn

1810 976 44 154 533 5920 1844

(from 1.9.56)

(4) - i.rrplicit a.l'1llual figures b (5) (iii) Tenunts and squatters (6) (iv) hl1ite rural areas as

a :,.,11010a

14087 5833 1312 16082 2812 7301 10904 40763 54850 60733 62045 78127 80939 88240 199144

I

I

I

B. Black sp:lts and harela.~

cc:;nsol':dation (i)~~

(7) - indiVi6.:alsu.

(8) - families a

(9) Mir .... isterial estirratesd , iv (9)

(10) (II) (n) (13) (14)

(I5) (16) iI7)

(li) Ectrl21and cor.solidaticn - inw. vidualsd

- i.:.rnilie::;a

Ninisteria: esti.rrates~

Cl.lITllllati,,'e figures fran 19489 Irrplicl t .::m.nua.l figures C. U~:n aL~

(i) Urban areas as a wholea .\11 ;~:t~~centres e,v

S0UrCes:

(;li)

5Il23

I

5453 5453 5453 5453 888

a. Reports of the Departnent of Bantu Administration and Develq:rrent (and successors) ::'. H.,UlSi1.rd P.Q.S45 22.4.60

L

c. I:uns.::..rd P.Q.S46 22.4.80

d. Ha:'.Oard P.Q. 10 20.2.68; P.Q.537 22.4.80

e. ;bns~rd P.Q.ro1.4I67 15.9.70; P.Q. 15.2.77; P.Q.39S 30.3.78; P.Q.245 28.2.79;

P.Q.I38 20.2.80;

f. Han..c;ard P.O.

g. Har.sard P.Q.I35 11. lIans~rd P.Q.466

Nc·te::;:

14.2.75; P.Q. I.4.76.

I6.2.73; P.Q.62 I3.2.74; P.Q.2I.2.75; P.Q. 1.4.76; P.Q. 9.3.77.

I6.3.73.

(i)-'1972 , r~fp.rs to the period 1.1.72 to 31.3.73. 1hereafter period runs fzan I.3. to 2&.2.

(ii) 1r. 15172 'black: spots' refers to hcrreland consolidation as well.

(iii) FrOi:'. I960.

(i v) ~..t..ere a nove was spread CNer several years, a oonstant nunber was assumed to have been rrovecl in each year.

(v) Urder Sections I4 & 29 of the U.tban Areas Act.

A.::!<nON"lecgEIrent:

I am gr~teful to Laur.!ne P~1;1:.%y for her help in drawing up the table.

';1!q2)

17937 23264

2090 IOO94

19882 52128 222II 33257 62459 66683

6151 IlOS3 9809

17900 II260 17018 16570 I5560 I8):)2 29680 59557 57355 61778 8054 32178 86743 41709 15688 74276 770 lI49 4667 1659 1294 5350 12401 5919 2157 10956 17032 II432 lI7I7 16832 13127 18721 29400 62379 57881 60437

5226 24612 I75788 I8I788 I90794 2II626 258632

6000 9006 20832 47006

37219 2):)26

I

33851 I II095 I I 7337 I I7383 1 9836 I .. 818 57957 45397

"""" -~~

Estim.l+:~ {thOl .. ;s.c".

""'" ~.= - , ";~I

" .. 1. .. 1._.

B:i<,ck spots & hmeland ^cons 12,8 12,8 12,8

bib I

I S.5~

Assurrptions be.~ind

~

a. 3 other people on average were ncved with each labour tenant and squatter.

b. aggregate figure apportioned equally """"9 years.

c. figure for tenants and squatters taken from P;Q.54S of 22.4.80.

d. tMen fran roo (9) e. rough average

f. ':a'<cn from reM (10) assuming 5 persons per fam11y.

g •. taken from (9)

13,4 ~

(20.0(-

Note cu:w1ative black spot/honeland consolidation figure COI1ES to 194 9al in 1972.

a little over the supplied figure. . .

• I •

I

,.," I,.,.! , .. "! ".,"! """1 """1 """1 ~.,'

52,9 51.,0

10.Si

I

II.3~ I7,0~

Table 3 Net migration between region types, I960-I980 (domestic Africans)

(a) I960-I970

age in Urban Rural Homelands

I970

~ F M F M

r

0-I4 -2567I6 -234798 - 29692 - 507I2 +286408 +2855IO I5-29 +36I9I8 + 74739 -I859II -209730 -I76007 +I3499I 30-59 - 5526I - 60720 - 32I56 -I24456 + 874I7 +I85I76 60+ - I28I8 - I096I - 24084 - 5I86I + 36902 + 62822 Total + 37I23 -23I740 -27I483 -436759 +234720 +668499

(b) I970-I980

age in Urban Rural Homelands

I980

~

r

0-I4 -543678 -404004 +I33843 + 60356 +409835 +343648 I5-29 +460932 + 70293 -2553I5 -222884 -2056I7 +I5259I 30-59 - 92848 +I77756 -IIII69 -I25098 +2040I7 +I07342 60+ - 49307 - I3999 + 2784 - I5729 + 46433 + 29728 Total -2248II -329954 -229857 -303355 +454668 +633309 Source: C.E.W. Simkins, The distribution of the African popu- lation of South Africa by age, sex and region type I960, I970 and I980, Saldru Working Paper No.32, Cape Town, I98I, Table 3.

Note: A + denotes net immigration, a - net emigration.

Two points need to be made in comparing Tables 2 and 3:

(a) One is co~paring gross emigration figures (Table 2) with net migration figures (Table 3), not an entirely satisfactory situation. Perhaps in the light of the increasing labour surplus in the rural areas since I960 (on this, see my African population, employment and incomes on farms outside the reserves, I923-I969) we may assume that gross immigration into the rural areas has been small. A second problem arises in the inter- pretation of the 'black spot removals and homeland consolida- tion' figures. If there is a removal from a privately owned non-scheduled black spot to a homeland, this will show up in

5.

Population Census based migration figures as a rural to home- land movement. If the area of origin, on the other hand, is a scheduled or trust-owned area, the movement will be an intra- homeland one and will not be reflected at all in Table 3. Let us assume that 2/3 of the black spot/homeland consolidation removals were of the former type in 1960-69, and that 1/3 were in 1970-79, a decade during which ever bolder redrawings of homeland boundaries took place. This would make resettlement 65% of net emigration from rural areas in 1960-69 and 76~ in 1970-79. The corresponding proportions of gross emigration would be smaller, the balance being made up of:

- farm evictions and movement to the homelands;

- 'voluntary' migration, i~ that there was neither forced removal nor transportation to a particular homeland site.

Resettlement, then, accounts for a significant (probably the greater) part of emigration from rural areas from 1960 to 1979, a mixture (in unknown proportions) of farm evictions and voluntary moves accounting for the rest.

Note that in the case of the urban areas, emigration of children (mostly born to migrant workers) is very important, net immi- gration of people of IS and above being positive in the period 1960-1979. This emigration, rather than removal of households, is the dominant reason for slow urban African population growth.

Resettled people are transported to 'resettlement camps'. After a time (as will be shown below) the term 'resettlement camp' becomes something of a'misnomer in that a significant number of non-resettled people come to live there. A more inclus.ive term would be 'closer settlement'. How maya closer settlement . be defined? One has to mark i t off from t\~O other types of

settlement found in the homelands:

- urban areas, developed' with S.A. Development Trust Funds.

and built to the specification of townships outside the home- lands. In 1960, 1,2% of the homeland population lived in such areas. By 1970, this had risen to 8,0% and the 1980 Population Census preliminary results indicate that 17,1%

of the non-independent homeland population is urban;

- traditional (and bettered) rural areas, where many households

have rights to fields and grazing and there is no absolute prohibition on any household acquiring such rights.

By contrast, people in closer settlements live either in self- erected housing or in rudimentary state-provided structures, have urban-sized (or only slightly bigger) plots and can under no circumstances acquire rights to fields and grazing.

'Closer settlements' on this definition encompass three

different types of settlement as envisaged by General Circular No.25 of 1967 from the Secretary for Bantu Administration and Development:

- towns with rudimentary services and housing;

- more densely populated residential areas with only a rudi- mentary layout on the basis of agricultural residential areas;

- suitable Trust Lands where families are settled in accordance with a system of controlled squatting.

How many people live in closer settlements? The basis for a very rough estimate lies in the 1960 Agricultural Census in which kraal heads were classified according to their access to land. Table 4 sets out the results of the enumeration; from i t can be seen that the great majority of rural people in the homelands had land rights of some kind in 1960. It is also clear that increases in these rights were even then hard to grant; arguably the homelands had reached the ceiling of their stock-carrying capacity as long ago as the late twenties and i t is difficult to see where new tracts of arable land could have been opened up. If one assumes that no more people could have lived under rural conditions in 1980 than in 1960, then the number of people living under closer settlement conditions can be obtained by deduction and turns out to be 4,7 million - rather too high to be plausible. One may alternatively assume that i t was possible for the rural population to grow slowly

(say at 1% p.a.) and then the closer settlement population would turn out at 3,7 million. It is hard to see how the

rural population could be any greater than this estimate implies.

7.

Table 4 Estimate of number of people in closer settlements, I980

(a) I960 Agricultural Census (kraalheads) Residential plots only

Residential plots and arable land Residential plots and grazing land Residential plots and arable land

and grazing land

(b) I960 Homeland population Urban (I, 2%)

Rural (98,8%) Total

56878 4682977 4739855

(c) I980 Homeland population _A Urban (I7,I%)

Rural (4I,3%)

Closer settlemen.t (4I,6%) Total

I93885I 4682977 47I6480 II338308

II

I0044I I97807 73558 339I50 770956

B 1938851 5714122 3685335 II338308

(17,1%) (50,4%) (32,5%)

Closer settlements containing resettled people as well as others, then, are a substantial part of the contemporary South African scene. How may they be understood against the back- drop of the country's growth as a semi-developed economy?

Gross domestic product per capita at I970 prices in South Africa was R380 in I960 and R559 in I979, (S.A. Statistics I980, p.2I.4) or $4!3 in I960 and $608 in I979 in I964 prices.

This places us squarely in the 'semi-develope~ country 'cate- gory ,for the whole of the period under consideration. It is in the $IOO to $I 000 per capi\a range that 75 to 80 percent of the total structural change in an economy takes place according to a study by Chenery and Syrquin (Patterns of ,Deyelopment I950-I970); assuming we maintain a 2% p.'a. real

growth in per capita income, we shall reach the upper limit of that range in the year 2004.

The structural changes of most interest here are two: urbani- sation and employment patterns.

(a) Urbanisation. From the 1960 and 1970 Population Censuses one readily finds that the percentage of all people urbanised in all parts of South Africa was 46,7% and 47,8% respectively

(see S.A. Statistics 1980, p.I.7). The preliminary results of the 1980 Population Census have the disadvantage of not including Transkei, Bophuthatswana and Venda; using raw figures would result in an upward bias since these areas are not as urbanised as South Africa as a whole. Assuming (i) that they are as urbanised as the rest of the homelands taken

together and (ii) that the ratio of their populations to home- land population as a whole was the same in 1980 as in 1970, the 1980 urbanisation figure works out at 47,3%. In other words, there has been no increase in the proportion of people living in urban areas in the past twenty years. This contrasts with the known tendency to increased urbanisation with develop- ment. Chenery and Syrquin regress urbanisation on income per capita and population using observations for many countries;

given South Africa's income and population their urbanisation regression equation predicts 49,6% urbanisation in 1960, 55,3% in 1970 and 56,7% in 1980. Using these predictions as norms, one could say that South Africa was underurbanised by about 9,4% of total population (i.e. about 2,7 million people) in 1980. This provides a justification (one among others) for seeing closer settlements as a form of displaced urbani- sation. They exist to contain people, who, without influx control, would have settled in South Africa's cities, perhaps under similarly rudimentary conditions, since General Circular 25 of 1967 makes i t clear. that closer settlements (as opposed to 'self-contained Bantu towns') are for the poorer classes in African society.

(b) Structure of employment. It is a standard feature of development that, as income per capita rises, the proportion of the employed declines in primary sectors of the economy and rises in the secondary and tertiary sectors. Table 5 reports the 'expected' proportions in each sector on the basis of the Chenery and Syrquin cross-country study as well as the best estimates I can produce of the actual situation in South Africa in 1960, 1970 and 1980.

9.

Table 5 'Expected' and actual sectoral shares of employment in South Africa, 1960, 1970, 1980

Ca) 'ExEected' (b) Actual

60 70 80 60 70 80

Primary 0,431 0,369 0,357 0,445 0,356 0,311 Secondary 0,240 0,273 . 0,282 0,152 0,218 0,236 Tertiary 0,329 0,358 0,361 0,403 0,426 0,453 Note: Primary refers to agriculture and mining, secondary to manufacturing and construction and' tertiary to all other sectors. Columns add to one; rows indicate trends.

The 'actual' figures are rendered uncertain by the difficulty of knowing what the level of employment is in both homeland agriculture and domestic service; conceptual as well as informational problems abound here. My assumptions are that employment in subsistence agriculture has remained constant in absolute terms (though there is evidence that many black women growing maize or beans on a small piece of land noN regard them- selves as employed just as little as their white gardening counterparts) and that domestic service has grown a little in absolute terms but not as fast as the white population growth rate.

:1, What we can reasonably conclude from Table 5 is that the

I,'

Whereas the actual share was a little above the expected in 1960, i t was well below in 1980. This has placed an employment provision burden 'on the other two sectors. Secondary sector employment has made a considerable contribution; its share has been smaller than expected throughout the period but the actual-expected gap was less in 1980 than in 1960. Our large services sector has grown in relative terms as well.

The decline in the primary employment share can be linked to Table 2 which indicates part of the spatial redistribution

associated with relative extrusion from agriculture. To the extent the rapid decline continues we can expect further removals and evictions and a further rise in the closer settle- ment population.

III

In the context of this phase of South Africa's development when there is substantial unemployment, how do people in closer settlements fare? The Surplus People's Project has kindly consented to my presenting some of the information collected in eight closer settlements (four in KwaZulu and four in the Ciskei). A brief note on each settlement follows:

KWAZULU I. Compensation. This is a resettlement camp on Trust Land 73km from Pietermaritzburg in the Impendhle district. The first people came from a black spot in the Underberg district.

There are signs that the government means to move more people in soon.

(AFRA Report No.6 October 1980)

2. Ezakheni. Ezakheni is 2Ikm from Ladysmith. Dating from the I960s and forming one part of a chain of closer

settlements i t is constantly growing. Interviews were conduc- ted in an area where mainly people from Roosboom (a black spot near Ladysmith) were resettled.

(AFRA Report No.9 March 1981; SPP notes)

3. Qhudeni. Qhudeni is 70km from Kranskop and is inhabited mainly by ex-labour tenants. A new closer settle- ment is to be sited close by.

(SPP notes)

4. Sahlumbe. Sahlumbe is the local name for a large area of farms bought up by the S.A. Development Trust in the late I960s. Large numbers of labour tenants, from the Weenen area mainly, were settled here then. Because this land was farm land before, there were some families already living there - they had and still have grazing and field land. The settlement is squeezed between hills and a large, flat and arable piece of land that has been taken over by the KwaZulu Development Corporation as an agricultural scheme - Tugela

II.

Estates. Residents of Sahlurnbe have no land and are very bitter about the presence of the KDC on arable land.

(SPP notes)

CISKEI 5. Keiskamrnahoek (Elukl".anyweni). The people here ca'me mainly from Humansdorp. They live in small houses, each with a modest plot. On the steep hillside opposite is a freehold land area. A dry and isolated area.

(SPP notes)

6. Kamrnaskraal. (Peddie) This is the most recent resettlement area in the Eastern Cape. people were moved there from farms in Wooldridge and Alexandria. Homes have been built in the area but not for everyone. The people from Woold- ridge came first after they had appealed to the Ciskei govern- ment for land of their own; the people from Alexandria were removed from a chicory farm.

7. Dimbaza. Notorious in the early I970s, Dimbaza has been greatly upgraded and improved, to the point of being rebuilt. It has factories, shops, churches, a beerhall 'and a police station. Different grades of housing exist; telephones are relatively easily obtained. A toy-making project and

making of school uniforms takes place in the town hall. 'A steel foundry and new wool factory are being built.

(SPP notes) Dimbaza is, in fact, no longer a closer settle- ment but qualifies, in the phrase of Government Circular 25 of I967, for the title of 'self-contained Bantu town'.

8. Mdantsane. Mdantsane was started in the early sixties as a black dormitory settlement for East London.

People have been removed to i t from there and elsewhere.

Like Dirnbaza i t should more properly be counted as urban than as a closer settlement.

In each of these areas, 80 to I20 households were interviewed.

Table 6 reports the dates of arrival of these households, why they came and what they found by way of ' facilities when they arrived.

Table 6 Dates of arrival, reasons for coming and facilities found by samples of households in eight areas, 1980/1

I

Compen- Ezak- Qhude- Sahlurn- Keis- Kannas- Dim- ~klant-

sation heni ni be kamma kraal baza sane

--- ---

---

- - -

---

ARlUVAL DATES (year)

-first quintile 1978 1976 1976 1969 1976 1980 1969 1967 -median 1978 1976 1977 1969 1977 1980 1971 1977 -fourth quintile 1978 1977 1978 1970 1977 1980 1976 1980

REASON FOR COMING (%)

-resettled 97 80 60 64 92 55 21 32

-evicted 3 3 21 32 I

-

-voluntary nove

-

FACILITIES FOUND (%)

-water 96 100 39 100 99 100 95 92

-latrines 94 100 55 3 ?I 72 94 ?I

-roads 4 70 50 98 99 94 98

-buses 43 I 14 28 77 90 77

-taxis 5 14 2 57 42

-shops 5 97 2 15 20 39 69 46

-schools I3 I3 78 ?I 77 62

-clinics I 9 3 20 30 77 44

-churches I3 14 9 9 70 36

-local authority

and/or p:llice 88 65 85 75 91

Notice the difference between Dimbaza and Mdantsane on the one hand, and all the rest on the other. In both Dimbaza and Mdantsane, about two-thirds of the households interviewed had moved to their present place without being resettled or evicted; in all others

(except Kammaskraal) the corresponding proportion is less than one- fifth. Note too the consistently high proportion of households finding various amenities on arrival, especially in Dimbaza.

Elsewhere the majority of households were resettled, with a

substantial minority in Qhudeni and Sahlumbe (areas to which labour tenants were resettled) of evicted households who chose to move there. As far as amenities go, Ezakheni, Keiskaw~ahoek and

J

13.

Kammaskraal have a range not to be found in either the older Sahlumbe or newer Compensation and Qhudeni.

Table 7 extends the picture by considering labour force partici~

pation, employment and unemployment in the settlements.

Table 7 Labour force participation, employment and unemployment in eight areas, 1981

Nale Female

(%) (%)

.NEA AR Ep Ec Emt U ^{NEA AR} Ep Ec E m KWAZULU

I. COllq:Ensation

i

< 15 101 2 I 95 2

15 - 24 35 26 5 2: 6 5 45 2 4

25 - 44 4 87 21 3; 19 3 36 25 2 7

45 - 64 2 50 I ^! 8 I I6 16 6

\65 5 ^: 2 10 I

I I Missing age

! : ^I

I I

30

!

'IDrnL 147 ; i 6 9 203 5 I7

1

I

I

I

!'I.asculinity ratio resident ^I

all 86 94

15 - 64 41 29

I

- _% enployed P, C, r.1 42 8 50 23 77

'Unemployment rate:

U

,

Ep+Ec+U Ep+Ec+Em+U

U

I 10

3 2

16

14 ,

Male Female

(% (%)

NEA AR E E E U NEA AR E E Em U

2. Ezakheni P c m p c

< 15 98 I 121

15 - 24 33 43 3 15 I3 7 53 38 4 I3 3 16 25 - 44 6 86 2 30 15 6 22 64 3 24 14 12 45 - 64 4 84 4 15 6 2 18 42 I I I 5 I

~65 8 I2

Missing age I I

TOl'AL ISO IO 60 34 I6 226 8 48 22 29

Masculinity ratio

resident, all 76 81

15 - 64 43 9 60 34 15 93 8 48 22 29

% errployed I', C,M 9 58 33 10 62 28

Unemployment rate:

U

,

E +E +U

P c E +E +E +ti p c m

3. Qhudeni

< 15 140 I I 128

15 - 24 32 18 I I 19 5 37 36 5 8 6 8 25 - 44 2 88 I 8 54 5 39 37 3 I6 II 4

45 - 64 8 53 2 4 14 3 27 23 I 6 I

~ 65 6 I 5

Missing age 21 I 2 7 I 22 6

'IUI'AL 209 6 16 95 14 258 9 36 I7 I3

Masculinity ratio

resident, all 78 I02

15 - 64 42 5 I3 87 I3 103 8 30 I7 I3

7, errployed P,C,H 5 14 81 15 58 27

Unemployment rate: ,

U U (%) 42 ·II 25 19

E +E +U E+E+E+U _i

p c P c m

IS.

Male Fenale

(%) (%)

NEA AR _Ep E _c E _m U NEA AR _{Ep E c} E _m U 4. SahluniJe

< IS 92 84 I

IS - 24 31 14

I

25 - 44 100

I

45 - 64 5 . 17

I

~65 Missing age 140 II

I I I

I

I I

TOl'AL 279 I 3 II 74 24 446 3 10 II 4

Masculinity ratio

I

resident, all 168 104

i

I

IS - 64 36

I

I

% employed p,e,M

!

I

Unanployrrent rate:

I

1

I

u ,

171

Ep+Ec+U Ep+Ec+Em+U I I

i I

I

j

I

!

1

I

CISKEI i

,

:

5. Keiskamrehoek ^!

I

;

i'

I

<

i

I

I

IS - 24 42

i

I

25 - 44 5 169

I

I

45 - 64 9 i53 I 9 9 19 34 8 9 2

~65 13 1

I

I

I I

I

Missing age

I

'IOI'AL 120 2 22 164 9 149 6 17 41 19

Masculinity ratio

561

i I ,

I i

resident, all 80 I 194

I

IS - 64 ,

I

2 22 64 8 77 6 16 41 19

I

i, I,

% employed P ,e,M _{2 25 73} !O 25 65

I

Unemployrrent rate:

---11-, _II (%) 25 8 46 23

E +E +U E +E +E +U

P c P c m

1-

6. Kanmaskraal

I I

.:: 15 70

i

15 - 24 I

34 15 I! 15 5 38 22

I

12~

25 - 44 7 59 I 2

:

45 - 64 5 37 3

i

!

I

~65 15

I

Missing age I _{I 2}

I

6

!

TOl'AL 131

!

i72

!

I

Nascu1inity ratio

resident, all 90

I

15 - 64 46 I 6 60 12 83 3 7

I

% errployed P,C,11 I 9 90 8 19 73

Unerrployment rate:

__ U __ , U (%) 63 15 72 41

Ep+Ec+U E +E +E +U P c m

i

I

;

I

:

I

I

7. Dirrbaza

I

<15 34

~I

15 - 24 9 50 5 5 4 25 39 I 9 6 6

25 - 44 3 85 3

I

45 - 64 I 90

I

~65 ^I I 2

Missing age 2 3 3 2 I 2

TOl'AL 49

i

Masculinity ratio resident, all

I

I

15 - 64 13 4 36 21 6 30 8 24 9 18

% employed P,C,M 7 58 35 20 58 22

Unemployment rate:

I

U

,

E +E +U E+E+E+U p c p c m

I

I

17.

Male Ferrale

(%) ^- (%)

NEA AR

-

I

J

8. Mdantsane

I

<I5 86

I

I5 - 24 45 26

I

I

25 - 44 4 89

7 II8 2I 7 7 87 4 27 9 I7 45 - 64 I 94 4 8 7 4 7 76 4 I I I 7,

~65 4

\ I

I I I I I

Missing age

'lUl'AL I40 I I 13I 32 23 ISO 9 4I I2 39

Masculinity ratio

l~

resident, all 86 94

I5 - 64 SO I I 32 23 70 8 40 I2 38

% employed P,C,M I5 42 43 I5 66 I9

Unemployment rate:

__ U __ ' U (%) 36 24 44 39

E +E +U

p c E+E+E+U p c m

~

Notes: All figures in raw counts except where percentages indicated.

NEA - not economically active.

Ep - employment near to place of residence in homeland.

Ec - employment of daily or weekly commuters to ~ihite areas

- employment of migrants (who visit home once a month or less) - unemployed.

- local activity rate i.e. U+Ep:l-Ec (expressed as a U+E +E +NEA percentage).

p c

Masculinity ratio resident excludes migrants, all includes them.

% employed P,C,M diVides workers into residents, commuters and migrants.

Unemployment rate Unemployment rate

U E +E +U

P c U

calculates local unemployment rate.

calculates degree of unemploy- E +E +E +U

P c m

ment among household members, whether present or absent.

There are several things to be noted from Table 7 :

(a) The median local unemployment rate is 3I% for men and 3I% for women. This rate counts residents only and is comparable with the

Current Population Survey estimates for the homelands as a whole (excluding Transkei and Bophuthatswana) of 8% for men and 15% for women in January 1981. The median unemployment rates when migrants are included drop to 13% for men and 29% for women.

(b) The median activity rates (labour force participation rates) for men 15 - 24, 25 - 44, and 45 - 64 are 23%, 87%, and 53%

respectively. The corresponding rates for women are 22%, 36%, and 35%. The rates for men are broadly similar to those found by the CPS in March 1980 (lower for the two outer age groups, higher for the inner) and for women are slightly higher than the CPS rates.

(c) The median proportion of migrants among the male employed was found to be 65%, and among the female employed 37%. Four of the areas (Dimbaza, Ezakheni, Mdantsane and Compensation) have no more than 50% of employed men as migrants; three have over 80%.

(d) One can distinguish three criteria for ranking employment conditions in the eight areas. The first is the activity rate, a high rate indicating most active engagement with the labour market. The second is the proportion of resident workers and ccmmuters, indicating the labour absorption capacity of the respective regions and the third is the unemployment rate. The ranks of the areas by these criteria are broadly similar as inspection of Table 7 will reveal; the best area is Dimbaza.

Ezakheni and Mdantsane follow with Compensation, Qhudeni and Keiskammahoek some distance behind. Sahlumbe and Kammaskraal are worst of all. There is again a broad correspondence between

level of amenities and economic conditions.

The picture can be made more detailed by considering the effect of migration on household composition as well as the numbers of households with zero earners (residents, commuters or migrants) together with information on frequency of remittances.

Table 8 ,demonstrates the relationship between migrancy and de facto household headship.

19.

Table 8 Migrancy and de facto household headship in eight settlements

Household De facto

heads (resi- % heads household

dent and migrant heads

migrant)

-

Compensation 32 43 24 47

Ezakheni 29 16 27 40

Qhudeni 14 54 I3 60

Sahlumbe 22 61 28 69

Keiskammahoek 30 34 2I 54

Dimbaza I3 I3 6 24

Mdantsane 38 25 32 53

Kammaskraal 22 30 16 45

Median 25 32 23 50

Note: The de facto household headship rate is derived by assuming a-wGman always takes charge when either a male or female household head becomes a migrant. The estimate is therefore upwardly

biased, but probably only slightly.

The median proportion of female heads (resident and migrant) is close to the average for Africans in urban areas and on white farms and reflects a common process arising from African nuptiality and mortality. The median proportion of· de facto female heads

(migrants removed) is double the earlier figure; this proportion is also similar to the homelands-wide figure.

Female headship has an effect on poverty as· can be seen from Table 9.

Table 9

20.

Households with no earners (resident or migrant) by headship and reliability of remittances from migrant workers in eight settlements

% households % migrants sending remittances

wi th no earners male feJ1'ale

headed by regu- irregu- regu- irregu-

male feJ1'ale larly ~ never larly larly never

Collpensation IS

Ezakheni 5

Qhudeni 10

Sahlumbe 9

Keiskarmahoek 7 Kanrraskraal 20 Dimbaza

Mdantsane 7

Median 9

52 I7 33 31 19 36

6 31

78 39

75 68

21 59 23 21

I 2

2 I I

87 45 89 100

13 45

7

Note: Imperfectly collected information prevents the publication of a full table.

Table 9 reflects a world-wide phenomenon: that female-headed households are more prone to poverty than their male-headed counterparts. The difference in this case is quite striking.

Given the level of rnigrancy, however, the fact that there is at least one earner in a household does not necessarily save the homeland residents from severe poverty: the frequency and magni- tude of migrant remittances must also be taken into account. In three of the four areas from which information is available, i t seems that more than two-thirds of both male and female migrants remit income regularly. How much they remit is not known, but the pattern of Table 10 suggests behaviour that is far from irresponsible and which, more than anything else, helps to keep crushing social distress at bay.

IV

Are people in closer settlements being marginalised? Or, to ask the question in another way, in what sense are the discarded people discarded? These are important, but large questions; I

IO

4

21.

shall limit myself to two conceptual and four empirical pOints.

The two conceptual points are:

(a) one's concept of marginalisation needs to be specified not only at an abstract theoretical level but should also contain rules about deciding whether a specific group of people has been marginalised or not. The empirical aspect of the issue has not been sufficiently clarified, at least not in South Africa. If, to speak fairly simply, one understands marginalisation as a process of reducing a group's life chances (above all in relation to the labour market) to a level significantly below those in another comparable group, then i t seems that one should focus on access to the labour market (and the indicator of its con- verse - unemployment). Access can be considered in institutional terms or in terms of the.skill content of jobs available; i t must also be seen in relation to the aggregate level of demand for labour. This specification will be used here.

(b) one needs to ask constantly whether factors which apply to closer settlements do not apply to homeland populations in gene- ral, i.e. whether, i f marginalisation is going on, whether homeland rather than just closer settlement people are being marginalised. As the closer settlement population rises in relation to total homeland population, then there will be a purely arithmetic tendency to convergence of conditions in the part and in the whole.

The empirical points are these:

.(a) i t has already been pointed out the median local unemploy- ,ment rate for the closer settlements studied is higher than the

homeland average, suggesting a differentiation between the part and the whole at this level. On the other hand',' the· median unemployment rates for households as a whole were found to be 13% for men and 29% for wo~en. This can be compared with rates for settled households in metropolitan areas: about sixty percent of men and eiqhty ~cent of women in metropolitan areas live in

settled households. Virtually all the metropolitan unemployment is concentrated in settled households, very little, of course, being found among migrants. So to get the settled rates, one needs to multiply the measured rates by 1,67for men and

1,25 for women leading to estimates of 13% and 23% respectively.

so the median rate of household unemployment is very l~ttle different for

result, not proving ~hat there is not substantial unemployment but that its distribution is different from what is sometimes thought.

(b) the question of institutional access to the labour market i., tied up with the distribution of both homeland labour bureaux a of private recruiting agencies. This is a very difficult matte) to analyse because, following the Riekert report, the homeland labour bureaux network is under reconsideration and little is known about the private recruiting network, part of which is very mobile. A couple of general observations can be made: at.

any point in time, a network of agencies will define a set of easy access nodes and a set of remote regions. There seems no reason to suppose that closer settlements will always find themselves in either nodes or remote areas; indeed, the settle- ments conSidered so far are in both.

(c) the relative extent of nodes and remote areas depends, of course, not on the spatial distribution of labour bureaux and recruiting agencies but on the aggregate demand for homeland labour which is in turn related to the demand for all labour.

In other words, the marginalisation question is related to unemployment in South Africa as a whole. I believe that there are two conditions for the presently high level of unemployment to be reduced:

(i) the long-term growth rate must exceed 5% p.a.;

(ii) the urban labour market must not become so isolated from tl::e h~meland market that homeland workseekers are unable effectively to bid on i t .

i

There was cause for concern on both fronts in The growth rate dropped well below 5% p.a. for while we saw increasing barriers being erected

the late

seventie~

a number of year~

around the urbani' I

i

labour market behind which urban-rural differentials w e r e i becoming greater. At present, optimism has increased about thel growth rate, but there seems no reversal of the trend towards

I

the creation of a privileged but limited urban labour market.

i

Why this matters can be illustrated by two numbers: from 1960 I

to 1979, employment in the non-primary, non-government and -domestic service sectors of the economy (predominantly urban) grew at a compound rate of 3,5% p.a. The rate dropped from

23:

4,5% p.a. in the sixties to 2,4% p.a. in the seventies; the rate of growth of real product dropped from 7,3% p.a. to 3,4%

p.a., so the drop in employment growth reflected the drop in output growth. A rate of 2,4% p.a. is very close to the labour force growth rate of 2,2% p.a.; should the former fall below the latter, unemployment will deteriorate without the hope of a long-term improvement. The modern sector, then, has not contributed to marginalisation in the past twenty years (though i t has made a negligible contribution to absorbing labour extruded from primary production and domestic service) but i t has come perilously close to doing so.

Also, the rise in urban wages relative to rural earnings in the early seventies followed by the institutional change which has been the practical outcome of the Riekert Commission report has served to isolate the urban labour market. Conditions on this market have improved rapidly over the last decade; one possible future is that they will continue to improve with a corresponding further move towards capital-intensity: a welcome development for urban insiders, but not for members of the outsider workforce.

The possibilities of marginalisation in such a course are

obvious; at present, however, this.is a potential threat rather than an advanced process.

(d) Are workers from closer settlements incorporated into the labour market at inferior levels compared with African workers in other sectors (particularly the modern sector)? Table 10 compares the skill levels of male workers in eight settlements taken together (a few farm workers 'excluded) with the distribution of skills recorded by the 1979 National'Manpower Survey (which dealt with all sectors of the economy except agriculture and domestic service) •

Table 10 Skill composition of .the male workforce in eight closer settlements (%)

Eight settlements 1979 Manpower

Resident Commuter· Mi9: rant Total Survey

White-collar I 4 5 IO 12

Skilled and

semi-skilled I 3 8 12 22

Unskilled 6 19 53 78 66

8 26 66 100 100

Comparing the total column with the I979 Manpower Survey, one notes that the difference is not great; the big discrepancy is between the proportions of skilled and semi-skilled workers.

This may reflect differences in average levels of education;

i t may also reflect differences in the sectoral composition of employment. Yet i t seems likely that settled workers are in general regarded as more favourable candidates for on-the-job training for the increasing number of skilled jobs opening up for Africans; to the extent that they are, migrants will be discriminated against. Yet in the light of Table 10 as a whole, i t would be overstating the case to refer to this discrimination as a major element in a process of marginalisation.

v

Has resettlement imposed major new claims on scarce public funds in the homelands? Not only does the South African Development Trust bear the costs of removal but people, once removed, require educational, health, pension services etc. How have homeland authorities been able to cope?

Table 11 sets out the broad picture of expenditure in the home- lands and its financing.

Table 11 Haneland public expenditure and its financing, I960 - I978 (a) Expenditure

Expenditure in Expenditure in b Haneland Expenditure current prices a I978 prices populationc

fer capita (I978 prices)

(R 'CXX) (R 'CXX) (R)

I960/I 27003d 87106 4739855 I8,38

I970/I I784I3d 426825 7385954 57,79

I978/9 IOI63ISe IOI63I8 II338308 89,64

25.

(b) Financillil

();m Indirect Pretorlah _ Total O.S. LT. Pl'A CUrrent

sources taxation revenue as

% of e»-

Current pri~ - R ^'cxx) (%) :eemtw:e

Ig74"j5 6686-r 20-3384 270245 24",7 75,3 63,4

I976/7 II1OO7 18874 336926 466807 23,8 4,0 12",2 58,g- 1978/9 I5I38"6g- ^I42564g~ 389704 683654 22,1 20,9- 57,0 60,5 Sources and notes:

a. E"xc-l udes grants-_ to development corporations..

b •. Conversion_ us-es implicit GOP deflator from the Sou~h A·frican national accounts. -

c. TaRen from my ~The distribution of ehe African popuration- of SQuth Kfrfca by age,_ sex- and region-type', SALDRU Working Paper

32 ^t 19·81 ~

d. BENSa, Blaek Development, Pretoria, 197&, Tab.le. B~I4-.2.

e. BENSO, Statistical Survey of Black Development, Pretorra;- 1979,- Table 90.

f. BENSO Survey, Table 88.

g .. All .indirect. taxation was- collected by Pretoria and handed over to home Iand- authorities except R60 million in I978/9- collected by the Transkei. BENSO' shows this as- ⁰ own sources ^{0 •} I have trans·ferred i t to indirect taxation for: consistency.

h. S.A.D.~. grants incltided.

Table 11 shows two interes.ting things. .The first is that despite_

rapid population growtiT, homelands >:wUc-exrenditure

!'Eir

-Whatever else reset.tlement has.. done, it has not- caused a sta_gna- tion in the pro-visron of homeland public services. Information on financing, unfortunately, is available ove-r a much. shorter period, but i t appears that from I97-4_ to I~na, about. 60% of expendi ture was- financed by current revenue, the other 407- presumably being provided by loans. Homeland sources provided just less than a quarter of current revenue, a slight decline being noticeable. The rest comes from Pretoria and has since 1975/6 been divided into two parts: 'indrrect toaxation° and-

o statutory and additional grants o. The. 0 indirect taxation 0

component is presumably calculated according to some notio ll of

the homelands share of indirect taxation (of which now the most important part is the general sales tax); i t has risen very steeply since its introduction. If one regards this component of revenue as a homelands source merely collected on their behalf by Pretoria, then the financial dependence of homelands authori- ties on Pretoria declined noticeably between 1974 and 1978;

if not, then i t increased slightly.

How does public expenditure per capita on Africans in the home- lands compare with that elsewhere? A comparison at the aggregate level would take a great deal of study since there is no consoli- dated set of public accounts for Africans outside the homelands.

Yet two partial comparisons are possible: (a) in 1978/79, expenditure on education per pupil in the homelands was R58 as compared with R72 in the rest of South Africa (BENSO Survey, Table 70), the pupil-teacher ratio was 50 compared with 47 (BENSO Survey, Table 67) and the number of pupils as a percentage of children aged 5 - 19 was 69 as opposed to 45 (the 'white area' average being lowered by the rural areas);

(b) in 1978/79, 350361 pensioners in the homelands received an average of R245 each; in the same year 190000 pensioners outside the homelands received an average of R247 each. (BENSO Survey, Tables 83 and 84) Pensioners in the homelands were 58% of all persons over 60;

outside the homelands the corresponding figure was 45%.

In these two fields, then, there seems to be parity of treatment in relation to the level of service (not surprising since the rules and delivery system have a cornmon origin; i t is too early to say yet what increasing homeland autonomy will bring) and there seems to be rather better coverage in the homelands than outside, principally because social services are so poorly developed in 'white' rural areas. In the short term, certainly, there is evidence that resettlement disrupts receipt of both pensions and education; in the longer term i t could easily be the case that most closer settlements enjoy services close to the homeland norm.

Table 12 presents some information on the direct costs of

resettlement. Taken from material on vote K of the South African

I " I

1 )

--~_J

27.

Development Trust, i t shows amounts allocated and spent and reasons for underspending. It covers the years 1970/71 to 1976/77; after the latter date, the accounts were reorganised and expenditure specifically on resettlement is no longer identifiable.

Two pOints are worthy of notice: (a) the very low amount spent on compensation and infrastructure in relation to the number of people moved. From 1971 to 1974, the total expenditure was RI,I8 million; in the same period about 78 000 people were moved (see Table I) so per capita cost of removal was as little as RI5;

(b) the substantial increase in spending after 1975; from 1975 to 1976 total expenditure was R10,49 million and removals were 85 000 so per capita cost jumped to RI23. I t is not clear why this jump should have taken place but i t probably reflects both increased amounts payable in compensation and increas,ed development costs for the resettlement sites. Whether the latter represents increasing average stand- ards or not would form an interesting'topic for further study.

Table 12 Expenditure on resettlement by the S.A. Development Trust, 1970/71 - 1976/77

Date Allocated (R) 70/71 310000

71/72 220000

72/73 151000 73/74 392000 74/75 813000 :]5/76 7374500

76/77 6660000

Spent 163051

174721

37536 371480 595884 4438267

6052528

Reasons for under spending

Delays in provision of services and adequate housing for the resettlement of Bantu, so less spent on compensation.

Certain basic services could not be provided, which resulted in delays in the settlement of Bantu

As 70/71

Compensation paid for the removal of Bantu less than expected.

Unavoidable delays in development of farms for settlement

Development of newly purchased farms and removal of black spots delayed as a result of economy measures

Claims in connection with removals not submitted timeously

VI

Economists often assess the implications of a policy by com- paring its consequences with the situation that would have arisen had an alternative been in force. In the case of resettlement there are two alternatives worth considering:

(a) no resettlement but continued influx control. Here the differences are quite easy to enumerate. A number of small town locations