Livestock production economics on communal land in Botswana: effects of tenure, scale and subsidies

Abstract

This study looked at the economic and financial characteristics of three range-based livestock systems on communal land in Botswana. Small-scale traditional livestock keeping is inherently efficient and provides important household income. However, these private returns are attributable to subsidies, and economic efficiency is very low due to open access. Low input, unfenced, cattle post production is economically efficient, but the returns to land are low. Fenced commercial ranching is not privately or economically viable in the more remote communal land. The loss of favoured European market access for beef could make all the systems studied economically unsound. The current subsidies to livestock production might be more economically efficient if they were redirected, from input costs reduction, to directly support initiatives that enhance the average herd productivity.

1. INTRODUCTION

This paper reports on an analysis of financial and economic efficiency in three range-based livestock production systems on communal land in northern Botswana. The authors used all available relevant empirical data on these systems, from Botswana and the region, to develop typical 2001 enterprise models, for comparison and hypothesis testing. Of particular interest in the comparative analysis were the effects of tenure differences, scale of production, and subsidies and taxes on the financial and economic efficiency measures. The objective was to learn lessons and to inform land use and livestock development policy in Botswana. The study formed part of a broader one on returns to both livestock-based and wildlife-based land uses (Barnes et al., 2001, 2003).

1.1 Study area and livestock systems

Botswana is a semi-arid landlocked country in southern Africa. Agriculture is important, and is dominated by livestock. Indeed, livestock, especially extensive cattle keeping on natural range, dominates land use in Botswana. Land in Botswana is roughly shared between traditional livestock keeping on communal lands (60 per cent), commercial livestock production on private or leasehold ranches (10 per cent), and wildlife-based activities on state and communal land (30 per cent). On the communal land, where most of Botswana's livestock is found, three basic production systems can be identified: small-scale traditional livestock keeping, medium-scale unfenced cattle post livestock keeping (‘cattle post’ farming is a rangeland ranching system similar to the ‘stations’ used in Australia and the US, which are in most cases characterised by pastoral leases), and medium-scale to large-scale fenced commercial livestock ranching. These three systems form the basis of the analysis in the present study.

Agricultural policy has placed considerable emphasis on promoting beef production from the vast communal land cattle population. This has involved attempts to enhance beef production from small-scale cattle posts, as well as the development of commercial ranching under the Tribal Grazing Land Policy (TGLP) (Edwards et al., 1989; Ministry of Agriculture, 1990). Arntzen and Veenendaal (1986) illustrate the way the communal land livestock herds have grown since 1900 – at an overall rate of about 4 per cent per annum.

1.2 Conceptual model

The three livestock systems can be described in terms of their bio-economic characteristics and the theoretical and empirical underpinnings of these. The basic conceptual bio-economic model resembles that described for fisheries by Clark (1985), and adapted for range systems by Bell (1984). In this model, the biological production function of rangeland fodder, in a non-grazed system, follows a sigmoid (S-shaped) pattern. Maximum sustainable yield, in terms of milk and meat, coincides with the maximum sustainable plant production at the mid-levels of plant and animal biomass.

Tenure systems influence this model in that ‘open access’ to grazing results in more and more graziers and animals entering the system, increasing grazing pressure, and pushing plant biomass lower. Here biological, and therefore economic, production is low, and unfettered competition between grazers causes the economic net benefits to be dissipated.

The three livestock systems we studied represent variants of the theoretical bio-economic model described above. The small-scale livestock system exhibits almost total open access to grazing, allowing stocking levels to reach the biological limit or ‘ecological’ carrying capacity, while commercial livestock ranching involves exclusive access to grazing, restricting stocking levels to the ‘economic’ carrying capacity, or those where animal productivity is maximised. The cattle post system falls between these two extremes. Our empirical analysis provides an opportunity to assess the model's validity. In the study area, the ‘economic’ carrying capacity was that recommended by the Ministry of Agriculture, or 12 hectares per large stock unit (LSU), while the ‘ecological’ carrying capacity was an average of empirical rates recorded for open access areas in Botswana, or 6.5 hectares per LSU.

Stocking rates heavier that the recommended ‘economic’ rate should, according to conventional wisdom, be associated with environmental costs. However, evidence in Botswana suggests that permanent damage to rangeland is negligible (Perkins, 1991; Abel, 1993; White, 1993), particularly in the very flat, sandy setting of this study. Environmental costs have thus not been considered.

Behnke and Scoones (1993), Ellis et al. (1993), De Leeuw and Tothill (1993) and Scoones (1993) consider that fixed carrying capacities have little utility as goals in range management. Nonetheless, as De Leeuw and Tothill (1993) point out, long-term average values for carrying capacity are useful for planning.

The model describes technical aspects of livestock production and should be interpreted in the context of the broader political and economic environment of the country. This is attempted in the discussion below.

2. METHODOLOGY

We extracted values for enterprise characteristics from empirical data, assembled from unpublished and published sources. We used these to subjectively derive typical budget/cost–benefit models, representing long-term average conditions, for the following enterprises in Ngamiland District, northern Botswana:

• (a) Small-scale livestock keeping with unfenced, open access tenure for milk, beef and transport service production as well as wealth generation, in core areas of human settlement, which is mostly along the southern and western edges of the Okavango Delta and Panhandle.

• (b) Cattle post livestock keeping with de facto partially exclusive access for beef production and wealth generation, in the less settled sandveld areas of the district.

• (c) Commercial livestock production, with exclusive access, for beef production, typified by the TGLP ranches of the Hainaveld.

Money values are presented in Botswana Pula. In 2000, P1.00 was equal to US$0.20. Extracted money values were converted and inflated/deflated accordingly. Our measure of welfare or utility was ‘direct use value’ as defined by Pearce and Turner (1990), and primarily the economic measure of ‘national income’ as defined by Gittinger (1982). Gross national income is a measure of income earned, whether domestically or abroad, by the factors of production owned by Botswana residents. Net national income is gross national income less capital asset depreciation. National income corresponds fairly closely to national product. Our measures of national income represented opportunity cost.

A secondary measure of direct use value was private profitability, as it affected the livestock owner or investor. In this case, all private costs and benefits, both monetary and in kind, were used to measure the private incentive for investment in the activity. The extent to which private returns differed from the economic ones was taken to indicate the influence of policy and/or market imperfections, as described by Jansen et al. (1992). The basic empirical data were generally recorded private values for income and expenditures.

Adjustments included the use of shadow prices when private values differed significantly from the true opportunity cost to society (economic values), following criteria developed by the Central Economic Advisory Service (1989) and Gittinger (1982) and described by the Ministry of Finance and Development Planning (1986), Matambo (1988) and Barnes (1998). Specifically, shadow prices are necessary for labour in the face of unemployment and social pressure for higher wages, since the market price of labour would tend to be higher than its scarcity value. Also, in the case of excess demand for traded and tradable goods and services, economic analysis should include a premium for foreign exchange. The pula has been considered to be overvalued in the short term (Matambo, 1988), and a foreign exchange premium of 10 per cent was added to the prices of all tradable items in the models.

The main shadow pricing adjustments were made with respect to costs of unskilled labour (50 per cent of the market price), excess demand for foreign exchange (10 per cent premium on prices of all tradable items), and domestic transfers (taxes and subsidies removed). Domestic transfers, although they are costs or benefits in financial terms, do not change national income, and therefore must be removed in the economic analysis. Those removed included sales tax, Botswana Meat Commission (BMC) levies, land rentals, input subsidies on veterinary costs, supplements and other ranch inputs. Important adjustments were also made for cross-subsidisation in the BMC carcass grade-price system as well as for subsidies on transport costs inherent in the Commission's freight equalisation scheme. The grade-price and transport subsidies were calculated using the findings of McGowan International and Coopers & Lybrand (1987).

Capital was accounted for in discounted net-benefit flows as direct expenditures and residual values. While private return estimates include land rental costs, the economic returns reflected land use before, rather than after, the inclusion of land opportunity costs. This allowed direct comparison between models regarding returns to land. The economic models also did not include central government expenditures in support of the agricultural sector.

All models contained wildlife or livestock herd/flock projections, incorporating birth rates, mortality rates, off-takes and purchases, within the constraint of prevailing range carrying capacity. ‘Economic’ carrying capacity was defined as the area of habitat required to support 1 LSU, while maximum sustainable yield was possible. ‘Ecological’ carrying capacity was the area that could support the maximum possible number of LSUs sustainably, regardless of yield. Livestock density was calibrated in LSU equivalents, where 1 LSU was the metabolic mass equivalent of a 450-kilogram bovine steer or ox, as determined for various species and intra-specific age groups by Meissner (1982a, 1982b).

3. EMPIRICAL DATA ANALYSIS AND MODEL FORMULATION

The results of the work of Vierich (1979), Bailey (1982), Hubbard (1982), Bekure (1982), Behnke (1982, 1985), Phuti (1984, 1985), Flint (1986), McGowan International and Coopers & Lybrand (1987), McGowan International (1988), Arntzen (1989, 1998), Van der Sluis (1992), Abel (1993) and Townsend and Sigwele (1998) contributed to the development of models for typical Ngamiland livestock systems.

Model parameters were corroborated from the results of work carried out in similar conditions in Namibia – for example by Ashley and LaFranchi (1997) and Turpie et al. (2000) – and in less similar conditions in Zimbabwe and South Africa – for example by Barrett (1992), Scoones (1993), Tapson (1993) and Campbell et al. (2000). The analysis of the macro-elements of the livestock sector was assisted by the results of Townsend and Sigwele (1998), Metroeconomica Economic Consultants (1996) and Sigwele and Khupe (1996). The models are described in detail below. Selected key parameters for the three enterprise models, derived from this analysis, are presented in Table 1.

Table 1: Comparative key parameters used in base case models for the three livestock systems (Ngamiland, 2000, per annum)

	(a) Traditional	(b) Cattle post	(c) Commercial
Land used (hectares)	180	6400	10 000
Number of cattle (head)	35	774	922
Number of goats (head)	3	59	125
‘Economic’ carrying capacity (hectares/LSU)	12	12	12
Stocking rate (hectares/LSU)	6.5	8.3	12.9
Calving rate (per cent cows)	60	63	65
Calving rate (per cent heifers)	60	63	65
Mortality rate (per cent calves)	18	9	5
Mortality rate (per cent others)	11	5	3
Bull rate (per cent herd)	5	5	5
Goat reproductive rate (per cent)	20	25	30
Average total herd growth rate (per cent)	0.77	8	4.8
Cattle off-take rate (per cent herd)	9	12	18
Goat off-take rate (per cent flock)	20	25	30
Milk yield (litres/lactating cow/annum)	158	−^a	−^a
Transport/draft (days/span of four/annum)	55	−^a	−^a
Cattle price variation due to grading (per cent)	−15	−5	0
Transport subsidy (per cent stock sales value)	22	32	32
Grade price subsidy (per cent stock sales value)	−10	5	14
Long-term borrowing (per cent initial capital)	0	5	25
Short-term borrowing (per cent recurrent costs)	0	10	20
Net subsidies/taxes (per cent production costs)^b	33	52	40
Notes: ^aMilk harvested from small proportion of herd for consumption by labour only, and no use of stock for transport/draft. ^bTotal transport, grade price and input subsidies, net of all indirect taxes, as proportion of production costs at full production.

3.1 Small-scale traditional livestock keeping

The base case small-scale traditional livestock keeping model involved a household unit with an average size herd of 35 cattle and three goats, amounting in total to 28 LSU equivalents, situated on the western edge of the Okavango Delta. Vegetation was transitional between the northern Kalahari tree savanna and the Okavango Delta mixed Acacia woodland. The recommended ‘economic’ carrying capacity was 12 hectares per LSU. Actual livestock stocking rates (near ecological carrying capacity) were 6.5 hectares per LSU. Use was made of one communal borehole, with the costs of this shared among 20 households. Grazing land was unfenced and effectively open access. Livestock have been routinely excluded from the localised areas of cropland, but we could find no empirical evidence of community-based, or other, efforts to control access of livestock to the common property grazing lands in Ngamiland.

In the base model, milk (52 per cent of gross income), meat (33 per cent), draft (15 per cent) and manure (0.1 per cent) were produced. Other key assumptions included herd appreciation at 0.77 per cent, reflecting fully stocked land; and milk production at 158 litres per lactating cow per annum, based on interpolation of data from, among others, Flint (1986), Townsend and Sigwele (1998), Arntzen (1998) and Campbell et al. (2000). Prices were obtained, and appropriately inflated, for milk, manure use and draught use from Townsend and Sigwele (1998); for livestock grades from McGowan International and Coopers & Lybrand (1987); and for the BMC agent's fee, amounting to 2.3 per cent of turnover. Only draught use that did not serve as an input to the livestock enterprise itself was included in the model. No loans were involved in the enterprise.

The calving rates, mortality rates and bulling rates for cattle are presented in Table 1. Labour requirements for the 38 animals were 1.3 full-time labourer equivalents per annum, calculated from the data of Bailey (1982) and Flint (1986).

For the economic analysis, subsidies and taxes were removed, including the transport subsidy inherent in the BMC freight equalisation scheme, the effective tax on the lower carcass grades produced, subsidies to veterinary costs and other ranch inputs, and indirect taxes, mainly sales tax, but also licence fees and levies. The subsidies net of indirect taxes made up 33 per cent of the annual production costs.

Botswana's access to the EU market has resulted in a price estimated to be some 40 per cent higher than world prices (Metroeconomica Economic Consultants, 1996; Sigwele & Khupe, 1996; Townsend & Sigwele, 1998). This international transfer had a positive effect on the national income. It is unlikely that Botswana would be compensated if it was removed, and it therefore has an opportunity cost. The rationale for removing domestic transfers, as costs or benefits, from the economic analysis is because they merely shift income within the economy from one party to another and do not change the national income. The EU price effect was treated as an extraneous economic benefit or windfall, and it was not eliminated as a subsidy in the economic model.

3.2 Cattle post livestock production

The base case cattle post livestock model involved an unfenced grazing area on previously unused communal land, with a herd of mostly cattle, but with a small number of goats, situated on sandveld in western Ngamiland. The livestock belonged to a single owner, or a small syndicate, resident off-site, somewhere in Ngamiland. Vegetation was northern Kalahari tree savanna, dominated by short Terminalia sericea and Acacia spp. The recommended ‘economic’ carrying capacity was 12 hectares per LSU. Actual livestock stocking rates (between ecological carrying capacity – about 6.5 hectares per LSU – and the recommended ‘economic’ carrying capacity) were around 8.3 hectares per LSU. Use was made of one borehole, developed privately for the cattle post. Grazing land was unfenced, but access to it by other graziers was limited by remoteness and custom.

In the base case model, cattle were mainly sold for meat via the BMC, but some milk and meat were consumed on site by hired labour and their families. Herd appreciation occurred at average rates of 8 per cent, reflecting the rate of expansion of cattle post herds. The amount of milk taken for local consumption was assumed to be the same as that for the small-scale traditional model, above. The livestock off-take rate was 12 per cent of the herd, by number of head. The prices for livestock off-take were assumed to equal those of the BMC. Off-take was trekked to BMC agents buying sites, and sent to the BMC facility in Maun (presumed to have been re-opened). A BMC agent's fee amounting to 2.3 per cent of turnover was assumed.

Table 1 presents herd characteristics including calving rates, mortality rates and herd constitution. Prices were those for upper grades, but were slightly lower in aggregate (5 per cent) than those for commercial beef production, described below (from data of McGowan International and Coopers & Lybrand, 1987). An equity of 95 per cent was assumed, and a loan of 10 per cent of working capital. Labour requirements were four persons full time, constituting one skilled and three unskilled labourers.

For the economic analysis, subsidies and taxes were removed, as described above for the small-scale model. After subtraction of indirect taxes (sales tax, fees and duties), subsidies made up 52 per cent of annual production costs. The higher-grade prices received (a blend of grades 1 and 2) resulted in prices that were 10 per cent above the real value.

For the reason given above, no account was taken of the loss to Botswana, which would occur if, as is likely, price support in beef importing countries (such as those in the EU protocol) was reduced. This support was considered to have no opportunity cost, and was not subtracted in the economic model. The economic net national income estimate included the value of herd appreciation, at 8 per cent per annum.

3.3 Commercial beef production

The base case beef production model involved breeding for production of 3.5-year-old slaughter steers on a 10 000 hectare ranch in south-east Ngamiland. Vegetation was northern Kalahari tree savanna, the recommended ‘economic’ carrying capacity was 12 hectares per LSU, and the actual stocking rate was 12.9 hectares per LSU.

Steers were marketed directly to the BMC facility in Maun, which was assumed to have been re-opened. From empirical data presented by McGowan International and Coopers & Lybrand (1987), it was calculated that the price was some 15 per cent higher than that received by traditional livestock keepers. A BMC agent's fee amounting to 2.3 per cent of turnover was assumed. Six staff members were required, comprising three unskilled labourers, two skilled labourers and one manager.

Table 1 presents rates of calving and mortality. Further, it was assumed that at full production 20 per cent of cows were replaced annually from heifers bulled at 2 years, and that bulls, at 5 per cent of the herd, were replaced every 3 years.

For economic analysis, subsidies and taxes were removed as for the other models. The net subsidy effect amounted to 40 per cent of production costs. In common with the previous livestock models, no account was taken of the loss to Botswana, which would arise if price support in beef importing countries was reduced. Again, the support was considered to have no opportunity cost, and was not subtracted in the economic model.

4. COMPARATIVE RESULTS

Table 2 sets out selected private results from the base case models for the three livestock systems, showing the income generated by these activities from the point of view of both investor and local community. Generally, private profitability and private rate of return are highest with the small-scale traditional system, and lowest with commercial production.

Table 2: Comparative financial results in three livestock systems' base-case models (Ngamiland, 2000, per annum)

		(a) Traditional	(b) Cattle post	(c) Commercial
Land used	Hectares	180	6400	10 000
Livestock	LSU	28	774	922
Initial capital	Pula	44 000	535 000	1 624 100
	P/hectare	247	84	162
	P/LSU	1605	898	2101
Financial turnover	Pula	15 500	170 300	366 600
	P/hectare	86	27	37
	P/LSU	560	286	474
Variable financial costs	Pula	3312	91 630	152 100
Fixed financial costs	Pula	7483	53 344	250 400
Net cash income	Pula	4709	25 299	−35 920
	P/hectare	26	4	−4
	P/LSU	170	42	−46
Local community income^a	Pula	10 560	22 500	45 000
	P/hectare	59	4	5
	P/LSU	381	38	58
Financial rate of return^b	Per cent	11.5	6.8	2.9
Financial net present value^b	Pula	381	−52 846	−526 984
Land rental^c	Pula	0	0	600
	P/hectare	0	0	0. 06
Notes: ^aIncludes profits, wages and salaries accruing to local communities. ^bMeasured over 10 years with an 8 per cent discount rate. ^cRental paid to district Land Board.

The economic characteristics of the three base case livestock models are presented in Table 3. Here, the economic rate of return is highest for the small-scale system, and is lowest for the cattle post system. However, in contrast to the private results in Table 2, the small-scale system generates less national income per unit of land than the cattle post system – only P0.26 compared with P1.86 in net national income per hectare. The commercial production system appears to be economically inefficient, generating negative values for all economic measures.

Table 3: Comparative economic results in base case models for the three livestock systems (Ngamiland, 2000, per annum)

		(a) Traditional	(b) Cattle post	(c) Commercial
Land used	Hectares	180	6400	10 000
Livestock	LSU	28	774	922
Initial capital	Pula	40 600	501 300	1 570 300
	P/hectare	225	78	157
	P/LSU	1465	841	2031
Economic gross output	Pula	13 008	194 084	301 576
	P/ha	72	30	30
	P/LSU	470	325	390
Annual economic costs	Pula	12 085	167 686	345 154
	P/ha	67	26	35
	P/LSU	436	281	446
Gross national income	Pula	922	26 397	−43 579
	P/ha	5.12	4.12	−4.36
	P/LSU	33.29	44.31	−56.37
Net national income	Pula	47	11 885	−132 014
	P/ha	0.26	1.86	−13.20
	P/LSU	1.69	19.95	−170.77
Economic rate of return^a	Per cent	10.1	2.0	−^b
Economic net present value^a	Pula	4679	−235 621	−895 013
Capital cost per job	Pula	31 214	125 323	224 323
Notes: ^aMeasured over 10 years with an 8 per cent discount rate. ^bNegative economic rate of return.

4.1 Small-scale traditional livestock keeping

The results in Tables 2 and 3 show that traditional livestock keeping is fairly intensive in the use of both capital and land. This is partly because of the open-access system of grazing that predominates, resulting in high stocking rates, at the ecological carrying capacity of the range. Gross income or turnover per unit of land, and even per unit of stock biomass (LSU), is highest for the small-scale system, because intensive use is made of a range of products, besides meat. Similarly, the 10-year financial and economic rates of return and net present values are higher than for the other systems, indicating that this system is inherently efficient in the use of capital.

The net economic values (national income) for traditional livestock keeping tend to be lower than the private ones (net cash income), illustrating the strong effect of input and transport subsidies, which outweigh the effect of the carcass-grade price tax (see Table 1). Yet the financial inputs per unit of land and stock tend to be more intensive in the small-scale system than in the other systems.

Table 4 presents the results of sensitivity analysis on the small-scale traditional livestock keeping model. Key parameters were varied, to show the effect on key economic and financial values.

Table 4: Results of sensitivity analysis on the base case assumptions for the small-scale traditional livestock-keeping model (Ngamiland, Pula, 2000)

Calving rate (per cent)	40%	45%	50%	55%	60%	65%	70%
Net national income/hectare	−25.71	−20.15	−13.68	−7.11	0.26	9.57	18.80
Net cash income/hectare	−2.49	3.89	11.10	18.26	26.16	35.99	45.62
Community income/hectare	30.46	36.39	43.04	50.44	58.66	67.73	77.70
Mortality rate (calves) (per cent)	25.5%	23.0%	20.5%	18.0%	15.5%	13.0%	10.5%
Net national income/hectare	−22.40	−16.59	−9.06	0.26	11.70	25.61	41.78
Net cash income/hectare	1.82	8.42	16.54	26.16	37.52	50.89	65.99
Community income/hectare	34.24	41.04	49.11	58.66	69.92	83.16	98.69
Beef prices (variation) (per cent)	70%	80%	90%	100%	110%	120%	130%
Net national income/hectare	−3.89	−2.51	−1.12	0.26	1.64	3.03	4.41
Net cash income/hectare	19.40	21.65	23.91	26.16	28.41	30.66	32.92
Community income/hectare	51.90	54.15	56.41	58.66	60.91	63.16	65.42
Capital costs (variation) (per cent)	130%	120%	110%	100%	90%	80%	70%
Net national income/hectare	−3.27	−2.09	−0.92	0.26	1.44	2.61	3.79
Net cash income/hectare	24.27	24.90	25.53	26.16	26.79	27.92	28.05
Community income/hectare	56.77	57.40	58.03	58.66	59.29	59.92	60.55
Milk yield (litres/lactating cow)	128 litres	138 litres	148 litres	158 litres	168 litres	178 litres	188 litres
Net national income/hectare	−5.89	−3.84	−1.79	0.26	2.31	4.36	6.41
Net cash income/hectare	18.93	21.34	23.75	26.16	28.57	30.98	33.38
Community income/hectare	51.43	53.84	56.25	58.66	61.07	63.48	65.88
Transport/draft (days/span)	35 days	45 days	55 days	65 days	75 days	85 days	95 days
Net national income/hectare	−3.10	−1.42	0.26	1.94	3.62	5.30	6.98
Net cash income/hectare	22.21	24.19	26.16	28.13	30.11	32.08	34.06
Community income/hectare	54.71	56.69	58.66	60.63	62.61	64.58	66.56
Note: Base cases in italics.

The results in Table 4 provide several inferences. First, large increases in economic rents would be possible with only small improvements in average livestock productivity (lower mortality rates and higher calving rates). Second, the open access grazing system, which drives down herd production – and causes high mortality, low calving rates, and low average herd growth – also tends to drive down net national income per hectare to low levels. This would fit with the conceptual model described above, where open access tends to drive the system towards ecological carrying capacity and dissipated net benefits. Third, the diversity of products (milk, meat and transport) reduces the vulnerability of the net benefit to value changes in any one of these.

The sensitivity analysis also tested the private measures – net cash income (profit) and the community incomes (local rural incomes) – generated by small-scale production. The results here are more robust, with positive private profits and incomes being generated through large ranges of parameters. This illustrates the effects of the subsidies enjoyed by the small-scale livestock sector. It also explains the apparent strong motivation for investment in livestock by rural households, and the high demand for this activity.

4.2 Cattle post livestock production

In Tables 2 and 3 the medium-scale to large-scale cattle post system is shown to have involved very low inputs. Positive but moderately low annual private profits were made, while the private return on investment was marginal. The economic net national income generated per unit of land and per unit of stock was positive but lower than private profit, showing the effect of input and price subsidies, which amounted to 52 per cent of annual production costs.

Sensitivity analyses for the cattle post model are presented in Table 5. Both private and economic results were moderately sensitive to changes in herd production values and prices. The cattle post system described in the model had lower stock densities and thus better herd production than the small-scale livestock system. The results for this system suggest that the high reliance on beef production, which comes with the location and the larger scale, reduced private profitability. This is despite the price and input subsidies available, and the relatively low inputs. They also suggest that moderate but positive economic net benefits can be generated with this type of land use in the more remote sandveld areas.

Table 5: Results of sensitivity analysis on the base case assumptions for the cattle post livestock production model (Ngamiland, Pula, 2000)

Calving rate (per cent)	52.5%	57.5%	62.5%	67.5%	72.5%
Net national income/hectare	−2.90	−0.63	1.86	4.58	7.54
Net cash income/hectare	1.64	2.77	3.95	5.20	6.49
Mortality rate (calves) (per cent)	16.0%	13.5%	11.0%	8.5%	6.0%
Net national income/hectare	−7.14	−4.78	−1.75	1.86	6.40
Net cash income/hectare	−0.64	0.64	2.19	3.95	6.08
Beef prices (variation) (per cent)	70%	80%	90%	100%	110%
Net national income/hectare	−4.80	−2.58	−0.36	1.86	4.08
Net cash income/hectare	−1.91	0.05	2.00	3.95	5.91
Capital costs (variation) (per cent)	150%	125%	100%	75%	50%
Net national income/hectare	−0.58	0.64	1.86	3.08	4.29
Net cash income/hectare	2.14	3.05	3.95	4.86	5.76
Note: Base cases in italics.

4.3 Commercial beef production

In Tables 2 and 3 the model depicting large-scale commercial breeding and rearing of cattle for slaughter shows that the system has relatively high inputs. The relatively high intensity of capital and recurrent investments in fencing, water development and herd management have been aimed at achieving returns in enhanced herd productivity. Yet private returns were consistently negative, despite subsidies, amounting to 40 per cent of annual production costs. A sensitivity analysis, presented in Table 6, shows that this inherent poor profitability is robust in the face of large, somewhat unrealistic, changes in prices and herd productivity. For example, for a positive net cash income, the calving rate would need to rise from 65 per cent to higher than 75 per cent, the beef price would need to rise by about 20 per cent, or the capital costs would need to be reduced to some 60 per cent of those existing. Based on the evidence to hand, such changes are very unlikely. Empirical evidence from extensive commercial beef production in similar semi-arid environments in southern Africa showed average calving rates around 65 per cent (Division of Economics and Markets, 1952; Lange et al., 1998). This was lower than conventional wisdom would suggest, but the only evidence that could be found for higher calving rates, of 70 per cent to 80 per cent, refers to either controlled experimental data in Botswana (Buck, 1978), or survey data from more productive environments, such as in Zimbabwe.

Table 6: Results of sensitivity analysis on the base case assumptions for the commercial livestock production model (Ngamiland, Pula, 2000)

Calving rate (per cent)	60%	65%	70%	75%	80%	85%	90%
Net national income/hectare	−15.72	−13.20	−10.08	−6.70	−3.51	−0.10	4.06
Economic rate of return (per cent)	–^a	–^a	–^a	–^a	0.25	0.81	1.40
Net cash income/hectare	−5.53	−3.59	−1.27	1.17	3.40	5.73	8.49
Financial rate of return (per cent)	2.06	2.89	3.78	4.61	5.28	5.91	6.59
Mortality rate (calves) (per cent)	–	7.5%	5.0%	2.5%	0.0%	–	–
Net national income/hectare	–	−17.19	−13.20	−8.38	−2.60	–	–
Beef prices (variation) (per cent)	100%	110%	120%	130%	140%	150%	160%
Net national income/hectare	−13.20	−11.02	−8.84	−6.66	−4.47	−2.29	−0.11
Economic rate of return (per cent)	–^a	–^a	1.21	2.53	3.75	4.86	5.90
Net cash income/hectare	−3.59	−1.01	1.58	4.17	6.76	9.34	11.93
Financial rate of return (per cent)	2.89	4.46	5.89	7.20	8.41	9.52	10.6
Capital costs (variation) (per cent)	100%	87.5%	75.0%	62.5%	50.0%	37.5%	25.0%
Net national income/hectare	−13.20	−10.81	−8.41	−6.01	−3.62	−1.22	1.17
Net cash income/hectare	−3.59	−1.59	0.42	2.42	4.43	6.43	8.44
Notes: Base cases in italics. ^aNegative economic rate of return.

Economic returns were also consistently negative for this system, reflecting inherent economic inefficiency. For example, calving rates would have needed to rise to 90 per cent, beef prices would have needed to rise by 60 per cent, or capital costs would have needed to drop by 60 per cent for a positive net national income contribution to be generated. Again, such conditions are clearly unrealistic. The economic inefficiency of this system is exacerbated by the location, remote from the main markets, which reduces real product value to some 32 per cent of that in southeastern Botswana, close to the markets. However, even without this the net value added is negative, confirming that this type of production is generally economically inefficient.

5. DISCUSSION

In semi-arid Botswana, climatic and environmental variation causes livestock systems to be in an almost constant state of change, and our models depict long-term average conditions that are rarely in evidence. Yet, biophysical and economic forces constantly push toward these average conditions. We consider that our use of sensitivity analysis, despite being linear, would have captured much of the variation likely to be encountered.

A comparison between the financial and economic results shows the degree to which subsidies and taxes benefited or hurt investors in the three systems. Producers in all systems received positive net subsidisation effects, ranging from 33 per cent of production costs in the small-scale traditional livestock keeping system to 52 per cent in the cattle post system. The economic results suggest that, even with removal of these effects, certain forms of livestock production in Ngamiland generate positive returns.

As discussed above, if Botswana's favoured access to the European beef market should be phased out, Botswana's beef export prices might drop by some 40 per cent. Our results indicate this would severely affect the economic viability of all livestock production activities in Ngamiland.

We have no picture of the economic efficiencies, or otherwise, of the beef processing sector, so we cannot pronounce on the viability of the livestock sector as a whole. Our results also cannot directly provide the means to determine the economically optimal allocation of land uses in the study area. Nonetheless, in the context in which we have used them, our models appear robust and illustrate several key characteristics of the systems, pointing to how livestock land uses could best be promoted in the study area, and other more remote communal land areas, to meet development objectives.

5.1 Small-scale traditional livestock keeping

Research by Abel et al. (1987), Abel and Blaikie (1989), Perkins (1991), Biot (1993), Scoones (1993), Tapson (1993), White (1993) and Abel (1993, 1997) has failed to establish that the high stocking rates prevalent in small-scale traditional livestock systems result in irreversible ecological degradation. Further, Abel (1993, 1997) – using crude financial gross margin measures – was unable to detect any financial incentive for small-scale households to reduce stocking rates in eastern Botswana.

Recent literature (Abel, 1993, 1997; Behnke & Scoones, 1993; White, 1993) suggests that little can be gained from reducing stocking rates in small-scale livestock systems. In contrast, Campbell et al. (2000), using detailed, dynamic cost–benefit models of different management scenarios in Zimbabwe, measured a strong private incentive for communal farmers to collectively reduce stocking rates.

Our results suggest that small-scale traditional livestock keeping is an inherently efficient system. The private evidence in Table 2 suggests that the small-scale system is efficient in contributing a wide range of household livelihood requirements, through several products, and that it is efficient in the use of capital. This is in agreement with the findings of Scoones (1992), Barrett (1992), Abel (1993, 1997) and others. However, none of these researchers examined the complete array of costs involved, or the effects of subsidies in the system.

Comparison between the private (net cash income) and economic (net national income) measures in Table 4 is interesting. The private measure shows positive returns, even under very adverse conditions. The economic results are consistent with those of Campbell et al. (2000) in that they tend to confirm the theoretical premise that de facto open access to grazing results in dissipation of net benefits. Positive returns in good years are cancelled out by losses in poor years. Small negative or positive changes in herd productivity would result in large negative or positive changes, respectively, in the net national income contribution.

The results suggest that traditional small-scale livestock keeping in Botswana has the potential to generate high economic values, with small increases in the average herd productivity. Such increases might be achieved simply by maintaining lower stocking rates, or through refined tracking and buffering systems, as proposed by Behnke and Kerven (1994) and Toulmin (1994), although these systems introduce new costs and may be economically inefficient (Campbell et al., 2002).

Increased herd productivity might result from implementing community grazing programmes, which restrict access and bring down the average stocking rates. However, traditional common property ownership systems seem unable to effectively restrict access to grazing in our study area, and elsewhere. Even with donor support, only a few specific cases of common property management (Danckwerts, 1974; Bennett & Barrett, 2003) have achieved success. In most situations, high transaction costs and institutional failures have apparently precluded it (Devitt, 1982; Sylla, 1994; Fitter et al., 2001; Birch & Shuria, 2002; Bruce & Mearns, 2002; Campbell et al., 2002; Turner, 2003). It appears that the incentives for household participation have not been high enough (Luckert & Campbell, 2003).

Our findings suggest that the subsidies that have accrued to small-scale livestock enterprises may indeed have reduced the private incentives for households to engage in common property management activities. While households continue to benefit even under adverse conditions, the incentive at national level for improving herd productivity appears strong. A case might thus be made for the redirection of subsidies, which provide income support (derived through inputs, transport and pricing), toward the promotion of mechanisms that reduce long-term average stocking pressures. The effective tax on lower carcass grades, inherent in the BMC grade-price structure, is already designed with this objective.

5.2 Medium-scale to large-scale livestock systems

There is no private incentive in the study area for either small-scale livestock-owning households or cattle post producers to switch to more intensive, fenced, commercial beef ranching. Our results indicate that capital-intensive commercial livestock ranching is privately unprofitable and economically inefficient, and it should not be promoted in the more remote parts of Botswana. This finding confirms those of Behnke (1982), Barnes (1994, 1998) and, more broadly, Scoones (1994). If Botswana's access to the high priced European beef market, through the EU Protocol, was removed, beef export prices would be likely to drop by some 40 per cent (Metroeconomica Economic Consultants, 1996; Sigwele & Khupe, 1996), and this would even further increase the inefficiency of ranching.

The question arises as to why commercial beef ranching still exists on freehold land in Botswana, if there is no financial incentive. The answer may lie in the fact that most capital on existing freehold ranches is already sunk. Much ranch infrastructure would have been developed when the terms of trade in beef production were significantly better than in recent years. In addition, speculation involving the buying and finishing of slaughter stock appears to be important in freehold areas (Barnes, 1994). This activity appears to be more privately profitable than standard breeding and rearing for slaughter (Loxton, Venn & Associates & Rural Development Services, 1985), due to cross-subsidy in the BMC grade-price structure. The freehold areas are closer to primary markets than our study area, and this would also enhance the viability of beef ranching.

Attempts to promote new fenced commercial ranch development on the more remote communal land areas in Botswana have been problematic. Many of the commercial TGLP ranches, despite subsidies, tend to be replaced by low-input production, more like that on cattle posts (McGowan International, 1988). Our findings suggest that cattle post livestock production is a relatively privately profitable and economically efficient land use for sandveld areas that are moderately remote from human settlement. Returns per unit of land are low, but in this specific setting alternative land uses (involving wildlife) tend to have even lower returns (Barnes et al., 2001, 2003). Adoption of more livestock speculation and finishing activities, instead of breeding and rearing, on TGLP commercial ranches, would be likely to enhance financial attractiveness, but the profitability of this is due to the grade price subsidy, and it would not enhance economic viability.

6. POLICY IMPLICATIONS

Although our study cannot confirm the economic efficiency of the livestock sector as a whole, the indications are that livestock production can contribute positively to development in the moderately remote parts of Botswana. Small-scale and low-input large-scale livestock production activities can have positive effects on livelihoods, and make positive contributions to national income. However, current policy appears to be suboptimal, and certain policy changes might significantly improve the economic efficiency and potential contribution of the sector.

Small-scale livestock production could potentially play a more important role in the more densely settled areas. Large economic gains could be expected to follow any success in improving herd productivity. Current input and transport subsidies, which increase farmers' net incomes, appear to reduce the private incentives for participation in community management activities to improve herd productivity. These subsidies might be more economically efficient if redirected to increase these incentives, and promote appropriate institutional development.

Fenced commercial beef production on ranches should not be promoted in Botswana's more remote communal lands. Instead, expansion of the livestock industry, into undeveloped and less densely settled areas, should focus on promoting low-input beef production systems, such as occurs at cattle posts. This is not to say that commercial systems should not be promoted in the less remote parts of the country.

Current policy permits the development of fencing to enclose grazing in communal lands. While fencing can strengthen and delimit property rights, it introduces high capital costs and reduces the mobility of stock and wildlife, severely reducing economic returns. Policies promoting fencing are therefore unlikely to be economically efficient.

Livestock development needs to be dovetailed with wildlife development initiatives in an economically efficient way. Livestock production occupies ecological, social and economic niches that complement those of wildlife-based land uses, which have been shown to have very significant economic development potential in certain parts of Botswana (Barnes, 2002; Barnes et al., 2001, 2003). Possible future loss of favoured access to European beef markets will shift the comparative advantage away from livestock to wildlife activities. Policy on land allocation should be flexible and sensitive to these forces.

The findings in this study need substantiation through further ongoing research into the economic values associated with the different livestock land uses and development options.

Acknowledgments

The present research was funded by Conservation International. The authors wish to thank Karen Ross and Karl Morrison of Conservation International for initiating and supporting the work, and Charlotte Boyd, Ann Gollifer, Gary Mullins, Beth Terry, Jeremy Perkins and Jan Isaksen for assisting with advice, assembly of documents and data. The authors also wish to thank the Swedish government for funding part of Jonathan Barnes's time on this study, through Sida (Swedish International Development Cooperation Agency).