Migration and development in Mozambique: poverty, inequality and survival

Abstract

Migration is many Mozambicans' preferred employment option (and sometimes last resort). Rural southern Mozambique, short of resources and traditionally less productive agriculturally than other regions of Mozambique, is now more developed and better off than other rural areas. An inter-regional analysis of the South, Centre and North of Mozambique demonstrates developmental differences largely attributable to labour migration (mainly to South Africa) and remittances. However, although migrant worker households, usually deficit agricultural producers largely dependent on migrant remittances, are often better off than non-migrant ones, many are still vulnerable to poverty. Migration has changed significantly over the last 15 years, with the eclipsing of mine migration and the increasing scarcity of jobs available to young Mozambicans. Despite the overall positive economic impact of migrant labour in southern Mozambique, as remittances decrease because earnings are lower and mechanisms for transfer are limited, its benefit may diminish.

¹Independent Consultant, Mozambique.

1. INTRODUCTION

Southern Mozambique has been a significant labour exporting area for more than 150 years (Katzenellenbogen, 1982; Jeeves, 1985; Harries, 1995). However, apart from migration occurring from Tete Province to Southern Rhodesia in the early 1900s, the rest of Mozambique has supplied almost no external labour migration and experienced comparatively little internal labour migration. Such a regional dichotomy allows for interesting comparisons, especially in regard to the impact of migration on household accumulation and wealth. In the 1980s, household differentiation was clearly evident in southern Mozambique, largely because of the fairly significant wage differentials between relatively skilled and unskilled mineworkers (First, 1983). Although many migrant-sending households clearly benefited from migration, the majority of these remained impoverished and became wage-dependent as their capacity to produce subsistence crops diminished. External work opportunities and conditions for migrants, especially since the abolition of apartheid, have become much more varied, leading to a much higher degree of household differentiation than prevailed from the mid-1800s to 1990 (de Vletter, 2000).

This article demonstrates that rural southern Mozambique, an area relatively bereft of resources and traditionally less productive agriculturally than other regions of Mozambique (due to poorer soils and erratic weather patterns), is now more developed and better off (at least in terms of average income and levels of wealth) than other rural areas. This difference is largely attributed to labour migration and the transfer of significant volumes of remittances.

Although migrant worker households are often seen as better off than non-migrant supplying households there are, indeed, significant numbers of the former that are vulnerable to poverty. These households are usually deficit agricultural producers, being largely dependent on migrant remittances. In turn, with the increasingly harsh attitude to unskilled illegal migrants in South Africa, their employment situation has become less and less secure (Crush, 1999). Although external migration is the preferred employment option for many Mozambicans, it represents the option of last resort for many others simply because of the limited employment absorption capacity of Mozambique's formal economy. In South Africa, employment is available for almost anyone willing to risk the consequences of illegal entry and prepared to be exploited, meaning that, for such workers, the remittances or accumulated wages brought home are likely to be minimal.

This article undertakes an inter-regional analysis (based on the results of a national survey of some 4000 rural households) of the South, Centre and North of Mozambique, demonstrating clear developmental differences that are largely attributable to many years of remittances channelled to the mainly rural areas of southern Mozambique.

2. METHODOLOGY

This study draws mainly from the results of two surveys: the National Roads Administration (ANE)/Austral Survey of Rural Households (1999–2001) and the SAMP Migration and Remittances Survey (MARS) (Pendleton et al., 2006) described below. Supporting data was drawn from the 1966 SAMP Survey of Mozambican Miners (de Vletter, 1998). The rural household study was used because it incorporated detailed questions on migrant labour and looked at a broad range of variables to determine household wealth. The results of the recent MARS provide important new revelations on migrant remittance patterns which help us better understand the influence of migration on development and household differentiation. Details of the three surveys are provided below.

ANE/Austral Survey of Rural Households (1999–2001)

The National Roads Administration (ANE), in collaboration with the consulting company—Austral Consultants (Austral Consultoria e Projectos)—conducted a comprehensive rural household survey covering all regions of Mozambique along selected sections of rehabilitated secondary roads. The sample consisted of approximately 4000 households. These households were visited annually during a period of three years (1999–2001) with the primary objective of measuring the socio-economic impact of road rehabilitation. The survey provided an excellent opportunity to also collect detailed economic data for rural households, including comprehensive information on migrant labour. Regions were defined as follows:

• South: Provinces of Maputo, Gaza and Inhambane as well as Maputo City

• Centre: Sofala, Manica, Zambezia and Tete

• North: Nampula, Niassa and Cabo Delgado.

SAMP Migration and Remittances Survey (2004)

The SAMP MARS was conducted in Botswana, Lesotho, Malawi, Mozambique, Swaziland and Zimbabwe. The survey interviewed only households with external migrants and focused on remittance patterns and migration history. The Mozambique survey was conducted in early 2004 and consisted of 726 households located only in the South. The survey areas were randomly sampled and included households in rural and urban areas.

SAMP Survey of Mozambican Miners (1996)

An earlier survey of mineworkers was undertaken by SAMP in Mozambique (de Vletter, 1998). The study interviewed 455 miners during the months of August and September 1996. Interviews were conducted at the Teba/Wenela depots at Ressano Garcia and Johannesburg as well as the depots of the recruiting agency Algos, which recruits mine and farm labour. In addition a separate survey instrument was used for interviewing 160 miners' wives in the provinces of Inhambane, Gaza and Maputo.

3. WAGE EMPLOYMENT AND MIGRATION PATTERNS IN RURAL MOZAMBIQUE

The ANE study found that one-quarter of all households have at least one member engaged in wage employment, but the distribution of wage-worker-supplying households is highly skewed. More than half (55 per cent) of the rural households of the South have members engaged in wage employment, compared with only 18 per cent in the Centre and 7 per cent in the North. Furthermore, the wage employment opportunities available for the South are overwhelmingly located at a considerable distance from the households. In the North and Centre more than half the wage workers are employed in the same district as their households (73 per cent and 58 per cent, respectively). Of the households with wage workers, 75 per cent in the South have absentee workers (compared with 32 per cent in the Centre and 41 per cent in the North). Reflecting the relative imbalance of male adults, the percentage of female-headed households with wage workers is only 42 per cent, compared with 60 per cent for male-headed households.

Of households with a wage worker, most of those in the South have more than one worker in wage employment (averaging 1.6) while almost all in the Centre and North have around one (1.1 and 1.3, respectively). Households in the South are therefore more dependent in two ways: in terms of the proportion of households involved in migrant labour and the extent to which households with wage workers participate in wage employment. Of all the wage workers, 42.5 per cent could be considered commuter workers, i.e. workers who normally sleep at the household.

Almost one-tenth (9.6 per cent) of all households were found to have seasonal workers (usually employed according to agricultural needs) who worked an average of five months a year. Seasonal work opportunities were concentrated in the South and Centre (with 14.3 per cent and 11.6 per cent of households providing seasonal workers respectively, compared with barely more than 1 per cent in the North).

Reflecting the higher wages in South Africa (and to a certain extent wages in Maputo), the ANE survey found that more than two-thirds (67.6 per cent) of the households in the South have workers earning the equivalent of more than US$60 per month compared to only 13.7 per cent in the Centre, while more than half of the households of the Centre have wage earnings of less than US$12, or considerably less than the minimum wage.

Historical and other factors have ensured that large numbers of workers from the rural South are employed outside Mozambique (de Vletter, 2000). More than half (53 per cent) of the wage employees coming from households of the South work outside the country. In very sharp contrast, both the Centre and North have almost no workers migrating abroad having, respectively, only 3 per cent and less than 1 per cent of their workers outside Mozambique. Despite the strong dependency on South African employment, economic development within southern Mozambique has managed to absorb almost half (47 per cent) of the workers coming from the rural areas of the South. Most of these workers are located outside their districts, in contrast to the large majority of workers who are located near their households in the Centre and North.

The MARS study found that 66 per cent of the adult population earn a cash income of some sort and more than half of this 66 per cent (55 per cent) are external migrants. Virtually all (94.1 per cent) of the external migrants work in South Africa and the vast majority (93.1 per cent) are male. Almost half (47.3 per cent) of the external migrants are aged between 25 and 40. More than half (54.9 per cent) of the external migrants are married, with 15 per cent cohabitating and 26 per cent unmarried. Significantly, more than half (50.7 per cent) of the migrants are sons (and in a few cases, daughters) of household heads, while just over a third (34.9 per cent) of the migrants are actually household heads (coming from just over half of the households). Almost half (46.1 per cent) of the household population were either students (22.8 per cent) or too young to work (23.3 per cent).

External migrants are generally poorly educated: only 15.2 per cent of the external migrants have secondary education, most (70.5 per cent) have primary education, while 8.2 per cent have no education at all. Sectorally, mine work still dominates. Of the 860 migrants whose place of work is known, 38 per cent work on the mines. The informal sector is the second most important (at 14 per cent), followed by the manufacturing industry (7.8 per cent) and working in a shop (3.6 per cent). A further 3.6 per cent are professionals. The remainder occupy a variety of different jobs (Table 1). The agricultural sector is surprisingly under-represented (at less than 2 per cent), since many irregular migrants entering South Africa are thought to get caught up in agricultural work in Mpumalanga Province before moving on to other work (Mather, 2000). It is very likely that many of those listed as “don't know” (18.2% of the responses) are engaged in agriculture as the activities of illegal migrants (many of whom start off in agriculture) are not known by household members.

Table 1: Sectoral distribution of external migrants

Main workplace	No.	%
Factory	67	6.4
Mine	329	31.3
Shop	31	2.9
Office	14	1.3
Government	5	0.5
Informal	120	11.4
Domestic	13	1.2
Farm	15	1.4
Profession	31	2.9
School	5	0.5
Other	230	21.9
Don't know	191	18.2
Source: MARS.

The rate of internal migration for domestic employment from rural households in the South of Mozambique is almost as high as that of external migration. However, the number of household members from external migrant-sending households working as migrants within Mozambique is very low. The MARS data shows that 19.1 per cent of the total household population lives outside Mozambique and only 3.7 per cent of the household population lives away from the household in another part of Mozambique. This suggests that migrant-sending households tend to send migrants either abroad or within Mozambique, but rarely in combination.

4. MIGRATION AND HOUSEHOLD WEALTH PATTERNS

Evidence collected from the ANE survey shows stark disparities between the three regions in terms of household income and assets (wealth). The South, poor in natural resources and more prone to drought and floods than the other regions, has labour migration (both external and internal) as its most important economic feature, distinguishing it from the Centre and North. On average, households from the South are better off materially than those in the rest of the country. However, there are still many poor and vulnerable migrant-sending households in the South. Remittances are often low and unpredictable, and subsistence production is marginal and susceptible to an unpredictable climate.

One of the best indicators of household wealth (and some would argue welfare) in Mozambique is the material used for building houses. The ANE survey found that one of the biggest contrasts between the regions was in housing (Table 2). In the South almost a quarter (23.8 per cent) of the households have constructed their principal houses with cement blocks. In the Centre this drops to 5.4 per cent and in the North to less than 1 per cent. The type of roofing shows even greater variation: in the South almost two-thirds (64.6 per cent) of the houses have non-thatch roofing (mainly zinc sheets). In the Centre 86.9 per cent of the houses have thatch roofs and in the North almost all (98.8 per cent). Other notable regional differences were the source of lighting and the use of rivers or springs as the principal source of water.

Table 2: Indicators of living conditions

Variable	South (%)	Centre (%)	North (%)
Cement walls	23.8	5.4	0.6
Thatch roof	34.1	86.9	98.1
Cement floor	52.0	10.1	1.9
Source of lighting – kerosene	85.9	52.8	39.6
Cooking fuel – firewood	94.9	90.5	94.0
River/spring as principal source of water	17.9	49.9	61.1
Source: ANE.

Households of the South own significantly more livestock than those in the other regions, with the greatest disparities being in cattle ownership (Table 3). Much higher percentages of the households in the South also own common household goods and assets (Table 4). Households in the South are also much more inclined to make investments than those in the other regions (Table 5). More than half the male-headed households in the South had invested more than the equivalent of US$40 in the previous year, compared with considerably lower percentages in the Centre (39.1 per cent) and North (17.7 per cent). Female-headed households tended to invest less but in similar regional proportions to male-headed households. Most of the investment went into construction or renovation of houses and relatively little into agricultural activities or transport. Not only is the proportion of households undertaking investments significantly greater in the South, the value of the investments is also higher. More than a quarter of the households in the South invested values in excess of US$400 in housing, agricultural equipment, transport and animals. In contrast, virtually none of the investing households of the Centre and North had reached such levels.

Table 3: Ownership of livestock

	South		Centre		North
Type of livestock	%	Average number	%	Average number	%	Average number
Cattle	18.5	6.4	4.8	7.7	0.1	2.0
Goats	52.7	5.2	38.8	9.3	21.7	6.1
Pigs	42.3	2.9	22.8	4.9	6.3	3.3
Chickens	72.6	11.4	68.1	10.2	46.5	8.3
Source: ANE.

Table 4: Selected inanimate asset holdings

Type	South	Centre	North
Solar panel	17.6	0.6	0.2
Bicycle	35.1	51.7	40.1
Pick-up van	20.0	0.8	0
Car	7.8	0.6	0.1
Tractor	1.9	0.3	0
Refrigerator	14.2	1.5	0
Television	11.7	1.9	0
Video	10.6	0.6	0
Music system	35.9	14.3	3.6
Generator	3.9	0.1	0
Grain mill	12.6	0.6	0.1
Plough	17.6	2.5	0.1
Watch	66.7	34.0	10.1
Water tank	12.8	1.6	0.1
Large water drum	59.1	9.2	2.1
Source: ANE.

Table 5: Investments (above 500 000 MT) made in last year (% of households)

	South	Centre	North
Male-headed	54.1	39.1	17.7
Female-headed	35.4	24.8	10.0
Residence	35.8	16.0	6.7
Transport	6.9	12.0	5.6
Agricultural equipment	4.8	2.7	3.6
Animals	8.6	2.4	1.1
Source: ANE.

At attempt was made to consolidate numerous variables (including wage income, investments, housing materials, assets, farming techniques, size of farm, and so on) and to convert them into ‘wealth points’ (see the Appendix for details of the variables that were selected and their point distribution). As expected, the distribution of wealth points is highly skewed towards poverty, with the vast majority of households considered poor. More than three-quarters (77–3 per cent) have less than 50 points and half have less than 25 points (the average being 36.8). Only 5.5 per cent had more than 100 (Table 6).

Table 6: Distribution of wealth points by region (%)

	Region
Quartiles	South	Centre	North	Total
1st quartile (lowest)	11.6	24.4	48.0	26.7
2nd quartile	12.8	27.9	28.9	24.1
3rd quartile	23.7	28.6	17.4	24.6
4th quartile (highest)	51.9	19.1	5.7	24.6
Total	100	100	100	100
Source: ANE.

Levels of wealth vary considerably amongst the regions. More than half the households in the South are in the highest quartile (Table 6). By contrast, almost half the households of the North are in the lowest quartile and barely 5 per cent in the highest quartile. Although there are indeed stark differences between the regions, the wealth point distribution is distorted by the fact that self-consumed production is not included. Because cash income is a substitute for self-consumption for many households in the South and is relatively easily measured and incorporated in the wealth point calculation, a more realistic depiction would show a downward adjustment for the South and an upward one for the North and Centre.

There is a greater concentration of female-headed households in the lowest quartile. More than a third (35.7 per cent) of female-headed households fall into the lowest quartile, compared to just under a quarter of the male-headed households. Somewhat surprisingly, however, there appears to be an almost equal chance among female- and male-headed households of being in the highest quartile (25.3 per cent for male-headed and 21.6 per cent for female-headed).

5. REMITTANCES AND POVERTY IN THE SOUTH

As this article has shown, inter-regional household welfare comparisons provide convincing evidence of a significant wealth disparity between the households of the South and those of the Centre and North. Within the South itself, however, this does not mean that there is not significant differentiation between households. Migrant-sending households are probably better off than non-migrant-sending households, a hypothesis currently being tested in another SAMP survey (the Migration and Poverty Survey or MAPS). However, the MARS survey shows that even among migrant-sending households there is considerable differentiation. Levels of remuneration vary considerably from household to household, a function of the income-generating capacity of the migrant, level of education and years of experience. Earning capacity and remittances are also affected by the sector of employment, the privilege certain households enjoy in gaining access to the mining sector and the legal status of the migrant. Differentiation is also strongly influenced by the degree of commitment of migrant workers to remit money or goods to their households. Many migrants are working under such exploitative conditions that they do not have sufficient surpluses to remit.

The MARS survey of external migrant-sending households found that remitted cash is the most widespread source of household income (76 per cent of households) followed by remitted goods (64 per cent) (Table 7). Income from wage work (largely internal) was cited by a third of the households (34 per cent). Migrant contributions provide the overwhelmingly largest share of household income (Table 8), although almost a quarter (22 per cent) of migrant-sending households do not receive cash income from migrant workers. Most migrants send money home either once a month or once a quarter (Table 9). The average value of remittances received is closely correlated with the frequency with which money is sent. Households receiving remittances on a monthly basis averaged about US$825, compared with US$123 for those receiving once a year.

Table 7: Household income from all sources (by frequency)

	Responses	% of households	% of responses
Wage work	243	33.5	13.6
Casual work	103	14.2	5.8
Remittances – money	548	75.5	30.6
Remittances – goods	466	64.2	26
Income from farm products	161	22.2	9
Income from formal business	30	4.1	1.7
Income from informal business	157	21.6	8.8
Pension/disability	23	3.2	1.3
Gifts	22	3.0	1.2
Other	23	3.2	1.3
Refused to answer	11	1.5	0.6
Don't know	4	0.5	0.2
Total	1791		100
Source: MARS.

Table 8: Household income by source and value

	No.	Mean	Median
Wage work	149	$1 016.19	$608.70
Casual work	44	$226.67	$65.22
Remittances – money	438	$523.99	$347.83
Remittances – goods	266	$393.79	$217.39
Income from farm products	115	$103.84	$39.13
Income from formal business	8	$779.89	$391.30
Income from informal business	84	$255.72	$130.43
Pension/disability	15	$263.45	$86.96
Gifts	12	$60.94	$26.09
Other income	11	$226.48	$52.17
Total income	579	$936.91	$528.26
Source: MARS.

Table 9: Frequency and value of money remittances

	No.	Mean	Median
Twice or more per month	13	$630.77	$365.22
Once a month	135	$862.39	$782.61
More than twice in 3 months	37	$327.09	$304.35
Once in 3 months	147	$303.14	$326.09
Once every 6 months	68	$201.62	$163.04
Once a year	82	$123.40	$65.22
At end of the contract	4	$240.22	$197.83
Other	67	$494.16	$304.35
Don't know	15	$435.65	$478.26
Source: MARS.

The earning potential of Mozambican migrants in South Africa is affected by three basic factors. First, wages vary considerably among sectors in South Africa, even for unskilled or semi-skilled migrants. Farmworkers are notoriously badly paid. In contrast, the mining sector is relatively well paid. Secondly, the migrants' education and experience affects their job prospects. Thirdly, migrant social networks influence labour market accessibility. Finally, the legal status of the migrant influences job access and security. Miners have a degree of security unknown to irregular migrants working in sectors such as construction and services.

About one-third of the MARS migrants work in the mining sector and earn substantially more than those engaged in other sectors. Miners also have better facilities for transferring wages and goods. In addition, they are subject to a system of deferred pay which ensures that the majority of their pay packet can only be drawn in Mozambique. In sum, households with mine migrants tend to be considerably better off than other households. As Table 10 shows, miners are second only to the small number of professional migrants when it comes to remitting home. At the other end of the scale, it is likely that many migrants from poorer households are working irregularly and under exploitative low-paid conditions.

Table 10: Annual cash remittances by level of education and sector

	No.	Mean	Median
(a) Education
None	40	$444.60	$304.35
Primary	424	$426.32	$269.13
Secondary	82	$582.36	$347.83
Don't know	28	$492.45	$428.26
(b) Main workplace
Factory	37	$390.09	$304.35
Mine	249	$537.22	$434.78
Shop	20	$462.15	$201.09
Office	5	$427.22	$243.48
Government	1	$434.78	$434.78
Informal	48	$312.57	$158.70
Domestic	6	$323.19	$304.35
Farm	5	$241.30	$239.13
Professional categories	16	$842.93	$304.35
School	2	$67.39	$67.39
Other	124	$402.84	$173.91
Don't know	43	$338.52	$152.17

Apart from sector and legal status, earning capacity is largely determined by the level of education and work experience. The vast majority (at least 70 per cent) of migrants have only primary education and 8 per cent have none at all. The survey showed distinct differences in the average annual amounts of cash remitted by level of education (Table 10). There is a big difference between the amounts remitted by those with only primary education and those with secondary education, the former sending an average of US$784 and the latter US$1072.

The MARS survey found that the most common method of remitting both money and goods is taking them back personally or through friends, despite significant changes in transfer technology. The most important changes relate to improved banking facilities for miners using TEBA Bank and the pre-paid delivery services of Kawena Distributors. Kawena, formerly limited to serving mineworkers, now offers facilities in various cities and towns in South Africa to anyone wishing to deliver goods to any accessible household in southern Mozambique or to one of a large network of warehouses. With time, more sophisticated transfer mechanisms will be increasingly used by Mozambicans working in South Africa.

Regardless of methods used, of major importance to the determination of remittance flows is still the degree of commitment by the migrant to transferring wages or goods. Commitment seems to be closely linked to gender, marital status, age and relationship to the household head. Annually, household heads send considerably more (US$1000) than their sons/daughters (US$625) or spouses (US$560), and males send more than females (US$840, compared with US$688) (Table 11). The older the migrant, the higher the remittances. Cash remittances also increase by age up to the 40–59 cohort, but decrease thereafter. Married migrants predictably send home much more money than non-married (almost twice as much).

Table 11: Money sent home: average amount over a year by migrant type

	No.	Mean	Median
Relationship
Head	285	$570.31	$417.39
Spouse/Partner	26	$316.91	$219.57
Son/Daughter	224	$341.41	$173.91
Father/Mother	3	$628.99	$608.70
Brother/Sister	27	$317.01	$173.91
Grandchild	3	$333.33	$434.78
Son/Daughter-in-law	2	$165.22	$165.22
Nephew/Niece	5	$456.96	$217.39
Other relative	1	$380.43	$380.43
Gender
Male	553	$457.65	$304.35
Female	23	$374.64	$152.17
Age (years)
15–24	55	$247.57	$152.17
25–39	292	$457.21	$304.35
40–59	119	$612.53	$434.78
60 and over	9	$472.71	$304.35
Don't know	99	$372.19	$217.39
Marital status
Unmarried	79	$264.92	$160.87
Married	367	$498.29	$347.83
Cohabiting	109	$474.51	$330.43
Divorced	7	$298.14	$260.87
Separated	3	$236.23	$304.35
Widowed	4	$196.20	$109.78
Don't know	1	$260.87	$260.87
Source: MARS.

Household expenditure estimates derived from the MARS survey are at best indicative. Table 12 shows the frequency of the types of expenditures incurred in the previous month. Food was by far the most important (89 per cent of households), followed by fuel (mainly wood and paraffin) (47 per cent), transport (45 per cent) and education (44 per cent). Other important categories included utilities (of particular importance to urban households), clothes, alcoholic drinks and medical expenses.

Table 12: Monthly household expenses

	No	% of households	% of responses
Food and groceries	648	89.3	20.4
Housing	5	0.6	0.2
Utilities	291	40.1	9.1
Clothes	262	36.1	8.2
Alcohol	240	32.6	7.5
Medical expenses	229	31.5	7.2
Transport	323	44.5	10.2
Cigarettes, tobacco, snuff	43	5.9	1.4
Education	319	43.9	10
Entertainment	15	2.1	0.5
Savings	84	11.7	2.6
Fuel	338	46.6	10.6
Farming	82	11.3	2.6
Building	97	13.4	3
Special events	80	11.0	2.5
Gifts	41	5.6	1.3
Other expenses	33	4.5	1
No expenses	8	1.1	0.3
Refused to answer question	43	5.9	1.4
Total	3181		100
Source: MARS.

In terms of value spent, the largest average amounts were spent on building activities (about US$150), although relatively few households (13 per cent) spent money in this category. The second highest expenditure was on food (US$70), which was also the most frequently cited expenditure. The third highest average value was for special events (US$50) such as weddings and funerals.

Food was by far the most dependent on remittances (78 per cent claiming that remittances were ‘very important’). Remittances were also considered to be ‘very important’ for cattle purchases, school fees, clothing, transport costs, vehicle purchase and maintenance, informal sector trading and farm labour costs. They were felt to be ‘important’ to the survival of the household in a significant majority of cases for food (73 per cent), medical treatment (64 per cent) and cash income (75 per cent).

The relevance of these findings for household differentiation is that expenditure on food and other basic needs overwhelmingly dominates the budget of external migrant-sending households. Comparatively few households (mainly those with miners and a handful of others with members in higher income jobs) therefore have the capacity to invest in housing, cattle or vehicles.

One of the more interesting findings to come out of the MARS study relates to the borrowing patterns of the respondent households. The need to borrow can be seen as an important indicator of vulnerability. A significant 42 per cent of the households said they had borrowed money during the previous year. Of those borrowing, half borrowed from family, 37 per cent from friends and 2 per cent from employers. The main reasons for borrowing money were to buy food (33 per cent), for health care (22 per cent) and for funerals (5 per cent). Financing companies are hardly present, with the exception of a few microfinance operators (with almost no presence in the rural areas and usually lending according to small business needs).

Loans are for the most part used for ‘survival’ issues, i.e. food and health. These are typical periodic needs of poor households, especially for those who have to rely on irregular remittances. A big advantage for many remittance-receiving families is that they are probably seen as a lower risk for loans than subsistence households with less reliable cash flows. Schooling and business loans are also quite common and are possibly linked to the household's ability to repay.

6. HOUSEHOLD DIFFERENTIATION

Household economic differentiation can best be highlighted through poverty analysis. A common poverty indicator takes the percentage of food expenditure relative to overall expenditure. The MARS survey found that the average percentage of expenditure devoted to food is 57 per cent. ‘Relatively poor’ households were defined as those spending between 60–79 per cent of their total expenditure on food and ‘extremely poor’ as spending between 80–100 per cent. The results show that almost a quarter (25 per cent) of the households can be considered to be relatively poor, and slightly less households (22 per cent) extremely poor. These findings should be qualified for two reasons. The first is that many households depend to a significant extent on self-produced household consumption that is not measured in the above analysis. This would mean that food ‘expenditure’ is even higher than indicated and that the level of poverty is, in fact, worse than indicated. On the other hand, the proportion of expenditure devoted to food may have been exaggerated, given that most households were interviewed during the month following traditionally high consumption periods (Christmas and New Year).

Others such as the Lived Poverty Index, suggest a high level of poverty amongst a significant portion of migrant-sending households. The MARS survey found that 24 per cent of the households are often without food and 11 per cent are often without medical care. Although these households are very dependent on cash income, 37 per cent claim to have been often without cash and 33 per cent several times without cash.

Despite the difficulties of acquiring accurate figures relating to poverty indicators, the data suggests that many external migrant-sending households are indeed very poor. These findings underscore what earlier work on Mozambique demonstrated, i.e. that there is a high degree of economic differentiation among migrant-sending households, ranging from the elite who benefit from several migrants with relatively high mine wages or professional salaries to households that are forced to send members to work under poor conditions for lack of suitable employment in Mozambique (First, 1983).

7. CONCLUSION

The rural areas of southern Mozambique have fewer resources and are agriculturally poorer and more vulnerable to climatic instability than those of the Centre and North. Yet, as this study has demonstrated, the pool of economic assets of the average rural household in the South is far greater than that of other regions. This disparity can largely be explained by the phenomenon of wage migration. Although a significant number of households in the south have migrant and commuter members working for wages in the domestic economy (mainly in the industrial enclave of Maputo-Matola), the most significant flow of wage-seeking labour has been, and continues to be, to South Africa.

Early migration may have been largely influenced by push factors such as hut tax, chibalo labour (the colonial system of forced labour), drought and famine. Later, however, employment in South Africa, particularly on the mines, was the preferred income-generating choice of Mozambican men from the rural (and often also urban) areas. Free transport, board and lodging and a virtually quarantined life allowed miners to accumulate most of their wages. Compulsory deferred pay (a system of forced savings) further ensured that miners would return to their homes with comparatively large amounts of money and goods. Such remittances were generally used, at least initially, for improving the household's quality of life (through the construction or furnishing of cement walled homes. Remittances were also used for savings (normally in the form of livestock) or investment. Traditionally, one of the most common investment choices was to buy a pick-up truck (bakkie) for transport purposes (often hired out) or a pump for irrigated agriculture. Now, with the proliferation of vehicles in the rural areas and limited irrigable areas, there is a greater tendency to invest remittances in informal sector trade activities undertaken by resident family members. For many years, import duty exemptions for miners gave them further accumulative advantages over other Mozambicans. Although no longer enjoying such privileges, Mozambicans in South Africa can take advantage of distribution services that provide reliable and cost-effective delivery of a large variety of goods direct to their rural base.

Much of southern Mozambique's external migration history took place during periods when black workers, especially foreigners, were subjected to the most exploitative of conditions. South Africa's migration system was the economic modus operandi of the apartheid system. Yet despite the degradation and oppression of such work Mozambican men streamed into South Africa, usually offering a supply much greater than the absorption capacity of a mining industry wary of excessive dependence on one source. Mine work still offered the best of the economic options for the majority of rural work seekers from the South and allowed them to build up their rural home base, but at considerable social cost.

Mozambican miners may collectively be seen as a wage elite. Households with several generations of miners are likely to have built up assets and a home-based production capacity that would put them well above the economic status of other households with a more recent involvement in mine migration. Households with miners with greater skills, longer service or more than one miner may have relatively high earnings. However, a significant proportion of mine-sending households could be considered to be poor. Differentiation between households is even more poignant when looked at across the entire range of migrant-sending households.

Despite Mozambique's economic growth rate being one of the highest in Africa over the past few years, much of the growth is linked to the development of highly capital-intensive ‘mega’ projects with limited absorption of unskilled workers. The urban informal sector, which has hitherto absorbed considerable numbers of the unemployed, has become less attractive for the rural labour surpluses as increasing competition makes economic survival more difficult. Such limitations within the domestic economy, recently exacerbated by the current drought in the South, have forced many rural dwellers to seek employment in South Africa.

This article has demonstrated that the overall economic impact of migrant labour has been positive in southern Mozambique. It has also shown that the nature of migration has changed significantly over the last 15 years with the eclipsing of mine migration and the increasing numbers of young Mozambican men chasing a limited number of jobs. It is therefore likely that, in the coming years, as the amounts of wages remitted are reduced (because of lower earnings) and the mechanisms available for doing so become more limited than they are for miners and workers in other, more privileged, wage sectors, the economic benefit for Mozambicans of working as migrant labourers in South Africa may diminish quite substantially.

The author would like to express his gratitude to the National Roads Administration (ANE) and the consultancy company Austral Consultoria e Projectos, Lda, for allowing the use of household interview data collected during a survey of rehabilitated roads during the period 1999–2001.

Notes

¹Independent Consultant, Mozambique.

Appendix: Wealth point determinants

Table

Wealth point determinants for ANE survey
Determinant	Cohorts	Point allocation
1. Size of Machamba (fields)	< 1 ha	0
	1–2 ha	5
	2–5 ha	10
	5 ha +	20
2. Value of crop (000 MT per annum)	< 200 (but greater than 0)	1
	200–500	2
	500–1000	4
	1–2000	8
	2–5000	16
	5–10 000	30
	10 000–20 000	40
	> 20 000	50
3. Improved seeds	Yes	10
	No	0
4. Fertilisers	Yes	10
	No	0
5. Pesticides	Yes	10
	No	0
6. Livestock	Per goat owned	1
	Per head of cattle	5
7. Monthly wages (000 MT)	Don't know but > 0	5
	< 200	5
	200–500	10
	500–1000	20
	> 1000	30
8. Monthly non-agricultural income (000 MT).	Don't know but > 0	2
	< 50	2
	50–100	4
	100–200	8
	200–500	16
	> 500	20
9. Animals sold during past 12 months (000 MT)	< 200	1
	200–500	2
	500–1000	4
	1000–2000	8
	2000–5000	16
	5000–10000	30
	10 000–20 000	40
	> 20 000	50
10. Investments (annual) (MT)	0.5–1 m	5
	1–3 m	10
	3–5 m	20
	> 5 m	30
11. Housing materials	Cement block	40
	Other	0
	Cement/tiled floors	5
	Other categories	0
	Water piped in or out of house/ well in yard	10
	Other categories	0
	Electricity supplied	30
	Electricity not supplied	0
12. Other assets	Solar panel	5
	Generator	5
	Water pump	20
	Grain mill	20
	Plough	5
	Refrigerator	5
	Radio	2
	Music system	3
	Television	5
	Video	5
	Watch	1
	Boat	15
	Fishing net	3
	Bicycle	3
	Motor cycle	20
	Pick-up van	50
	Car	50
	Truck	80
	Tractor	50
	Water tank	3
	Drums (200 litres)	1
	Other large water containers	1
	Cool box	1