Evaluating Equity Impacts of Animal Disease Control: The Case of Foot and Mouth Disease in Zimbabwe†
This paper was presented at the session “Analyzing International Public Goods and Trade: Methodological and Policy Implications from Studies of Food and Mouth Disease,” organized at the Allied Social Sciences Association annual meeting in Philadelphia, January 7–9, 2005.
The articles in these sessions are not subject to the journal's standard refereeing process.
As in other countries in southern Africa, foot and mouth disease (FMD) is endemic in certain parts of Zimbabwe inhabited by African buffalo, which harbor the virus. Despite this, Zimbabwe has invested for many years in maintaining freedom from FMD over much of its territory, permitting it to take advantage of favorable tariff arrangements for export of boneless beef and other livestock products to high-value markets in Europe. To maintain freedom from FMD, a zonation system implemented with fencing and movement controls has been used to protect a central export zone where much of the large-scale commercial farming is concentrated. Since 2001, Zimbabwe has suffered a series of FMD outbreaks, leading to a ban on the export of many livestock products, including beef. A high level of investment would now be required to re-establish FMD freedom and permit renewed export.
Traditionally, assessing such investment in animal disease control involves a straightforward application of benefit-cost analysis (BCA) to see if the investment is justified on efficiency grounds as contributing to economic growth (Dijkhuizen and Morris). Policy objectives in developing countries, however, increasingly feature poverty alleviation in addition to economic growth, as evidenced in the national Poverty Reduction Strategy Papers (Swallow). In this paper, we report extensions to the traditional BCA framework to address these equity impacts as part of an evaluation of improved FMD control in Zimbabwe recently undertaken by Perry et al.
Extending the Benefit-Cost Analysis Framework
BCA provided the general framework in the study by Perry et al. for evaluating and comparing alternative FMD control policies for Zimbabwe. The BCA follows the standard approach of identifying and valuing the full range of direct and indirect impacts for a baseline scenario. These are compared to see how incremental investment costs and their benefits accrue over a given time horizon—twenty-five years in this case—under alternative scenarios, including one which assumes disinvestment in FMD control. The economic impact of the various scenarios can be evaluated in terms of their net returns, with income streams appropriately discounted. Uncertainty about key parameters, particularly those regarding risk of disease outbreaks and the evolution of the international trade environment, required auxiliary analyses. Integrating a stochastic epidemiological model into the BCA framework permitted simulating future FMD outbreaks, while an analysis of trade trends and likely changes in trade arrangements over time generated three best-guess scenarios for the future evolution of prices earned by Zimbabwe beef exports. A Monte Carlo approach was applied, simulating the BCA 5,000 times to predict the epidemiological model and other stochastic elements. The following three extensions to the BCA were introduced to address the distributional issues.
Disaggregated Costs and Benefits
The first extension involved disaggregating the analysis by sub-sector. Costs and benefits were distinguished in terms of public versus private sector. The private sector was further broken down into communal and commercial cattle production, beef industry, and the rest of the economy. Changes in income were allocated to five distinct sub-populations: communal area households, lower- and upper-income households on large-scale commercial farms, and lower- and upper-income urban households. In addition, the numbers in the three categories of rural households located in the various FMD-control zones, as well as numbers of households in six livelihood subcategories within the communal areas, were tracked across scenarios as the definitions of these zones were varied.
Macroeconomic Modeling
The single most important impact associated with a change in FMD control in Zimbabwe is the loss of access to high-value beef export markets. A single FMD outbreak in one of the control zones triggers a year-long export ban on livestock products with immediate income and employment repercussions within the beef sub-sector that generate multiplier effects transmitted across other sectors of the economy. Garner and Lack demonstrated that such sectoral impacts of an FMD outbreak in Australia could be quantified using an input-output (I-O) model. A similar I-O approach was developed for Zimbabwe, for which a Social Accounting Matrix (SAM) based on 1991 data was already available that divided the population into the five sub-populations adopted for the disaggregated BCA (Thomas and Bautista). The SAM was adapted to distinguish separate beef-processing and export activities, thus permitting the impact of eliminating the revenues from beef-export sales to be traced using a Computable General Equilibrium (CGE) algorithm.
A limitation of using 1991 data for the analysis is that much of that year, the Zimbabwean beef industry was still under a European Union (EU) trade ban from an FMD outbreak in 1989; only an estimated 4,000 tonnes of fresh and frozen beef were reported exported that year, filling less than half of Zimbabwe's quota of 9,100 tonnes for beef exports to the EU. Cattle and beef sector activity levels in the SAM, therefore, represent a constrained, rather than free market, equilibrium, necessitating caution when interpreting the analytical results. To simulate the impact of varying degrees of FMD control for a given year, the baseline scenario for 1991 of 4,000 tonnes of exports was compared to scenarios of:
- Simulation 1: Ban (a total ban);
- Simulation 2: Normal trade, doubling the amount to 8,000 tonnes to approximate filling its EU quota; and
- Simulation 3: Expanded trade, tripling the amount to 12,000 tonnes to represent an aggressive export strategy.
Analysis of Price Transmission within Cattle Markets
In addition to generating differential economic activity and nominal income by sub-sector and population group as estimated by the combination of the SAM structure and the CGE modeling, beef-export trade may also affect real incomes in Zimbabwe through its impact on prices. Although the CGE model provides estimates of changes in equilibrium prices within each sub-sector, the model is not sufficiently disaggregated to generate meaningful conclusions regarding the differential impact of price shocks when FMD-prompted export bans occur. Cattle are a common feature across smallholder farming systems in Zimbabwe, so differential price impacts within the cattle sub-sector might be expected to have important implications for the rural poor.
To explore this possibility, the macroeconomic modeling was complemented by an in-depth analysis of the transmission of price shocks from the beef-export market back to and within cattle markets. A preliminary analysis examined the correlation between the export market and domestic cattle and meat markets based on monthly price series for 1998–2000. Much longer monthly price series were also assembled for the most common grades of meat and animals sold in commercial and communal cattle markets in representative areas from different FMD-control zones, covering 1993–2000.
These price series permitted examining the transmission of price changes in the FMD-free zone between the market in which primarily commercial producers sell their cattle for meat export and auction markets where communal producers sell their stock; between auction markets for communal producers in the FMD-free and the other FMD-control zones; and between superior and inferior grades of animals sold in auction markets for communal producers within each type of FMD-control zone. The various price series examined were first evaluated for non-stationarity using the Augmented Dickey Fuller (ADF) and Phillips-Perron (PP) tests before undergoing co-integration analysis. Price transmission between each pair of price series was tested using the following procedure:
- Selection of the appropriate order of vector autoregression representation based on Schwartz-Bayes and Akaike information criteria;
- Test for Granger non-causality, as a measure of short-run price transmission;
- Test for imposition of 1:1 restrictions, as a measure of transmission of price shocks;
- Construction of error correction models (ECM), as a measure of long-run price transmission;
- Impulse response (IR) and forecast error decomposition (FED), as measures of the timescale over which shocks in one market are transmitted to another; and
- Co-dependence analysis.
Analytical Results
Macroeconomic Modeling
Simulating changes in beef-export volumes in the SAM/CGE model to represent different FMD-control scenarios provides results generally consistent with expectations. A ban on beef-export trade in a given year, for example, is associated with a contraction in the value of exports and gross domestic product (GDP). However, in order for the model to solve, it must vary export prices according to the given elasticity parameters to stimulate the targeted change in export volume. This model requirement leads to counter-intuitive shifts in equilibrium prices across the simulations, as well as significant crowding out of export value from other sectors of the economy. This effect, together with the unusually low level of beef-export activity in the model base year noted earlier and inconsistencies in the national-level data, was suspected to introduce bias into the analytical results, warranting caution in their interpretation.
Table 1 reports the changes in nominal values of overall GDP and domestic activities by sector associated with the baseline and three simulations. Based on the average change associated with the first two simulations (having less confidence in the predicted parameters associated with the much larger change in trade volume assumed under the third simulation), a marginal change of Z$1 in beef-export trade is estimated to contribute to a Z$1.10 nominal change in Zimbabwe's GDP.1 Within the economy, however, exports generate proportionally an even greater increase in the value of economic activity. The overall multiplier effect of beef exports is estimated to range Z$1.67–1.86, for a weighted average Z$1.79. Most of this effect is captured by the beef sub-sector, with a Z$1 increase (decrease) in beef exports leading to a Z$0.50 increase (decrease) in activity for cattle producers, Z$0.79 in the beef-processing industry, and Z$0.50 in the rest of the economy. Commercial farmers bear 61% of the change in activity among cattle producers, and communal-sector cattle keepers, the remaining 39%.
| Change from Base (Z$ million) | ||||
|---|---|---|---|---|
| Baseline (Z$ million) | Total Ban | Normal Trade | Expanded Trade | |
| Exports | 7,080 | −75 | 76 | 183 |
| GDP (at market prices) | 29,646 | −59 | 108 | 202 |
| Domestic activity | 48,561 | −114 | 210 | 404 |
| Cattle | 838 | −33 | 56 | 122 |
| Beef processing | 630 | −52 | 89 | 196 |
| Rest of economy | 47,093 | −29 | 65 | 86 |
The SAM permits tracing the impacts of these changes on returns to the factors of production by sub-sector, and subsequently on incomes for the different categories of households owning these factors (table 2). Based on these results, in the case of a beef-export ban, communal-area households suffer the largest proportional loss of their income rather than the large-scale commercial upper-income group, which has been the principal supplier of cattle to the export market and has benefited from its concentration in the FMD-free export zone.
| Population Group | Total Income-Base Value | Income Loss with Trade Ban | |
|---|---|---|---|
| Communal areas | 1,846 | 23.6 | 1.3% |
| Large-scale commercial—lower income | 100 | 0.6 | 0.6% |
| Large-scale commercial—upper income | 9,293 | 54.1 | 0.6% |
| Urban—lower income | 2,620 | 25.4 | 1.0% |
| Urban—upper income | 12,441 | 47.4 | 0.4% |
Disaggregated BCA
The results from the CGE/SAM simulations serve as inputs into the BCA by providing estimates of the changes in income as beef exports vary annually under the different FMD-control scenarios. Table 3 reports the aggregate results of the BCA, which confirm the economic utility of FMD control. While additional investment in improving control yields modest benefits, disinvesting through relaxed control and allowing FMD to become endemic generates much greater losses.2
| Scenario for FMD Control | Number of Trade Bans over 25-Year Horizon | Total Costs of FMD (Z$ billion) | Benefit-Cost Ratio |
|---|---|---|---|
| Continue current level | 10.8 | 17.4 | (Baseline) |
| Improved control | 3.5 | 14.5 | 1.54 |
| Relaxed control | 25.0 | 41.7 | −4.68 |
The disaggregated BCA provides estimates of the differential impacts on the various population groups. Table 4 reports the distribution of income gained under the scenario of improved FMD control, indicating that higher-income segments of the population capture two-thirds of the benefits.
| Population Group | Total Income Gain (Z$ Million) | Per Household (Z$) | |
|---|---|---|---|
| Communal sector | 547 | 16.1% | 360 |
| Commercial sector—upper income | 1,229 | 36.1% | 11,260 |
| Commercial sector—lower income | 14 | 0.4% | 50 |
| Urban and other—upper income | 1,031 | 30.3% | 2,460 |
| Urban and other—lower income | 580 | 17.1% | 690 |
Analysis of Price Transmission within Cattle Markets
Underlying the one-period CGE/SAM modeling is the assumption that access to lucrative beef-export markets not only generates added revenues, but also contributes upward pressure on prices in domestic cattle and meat markets, ultimately benefiting all producers and not just direct suppliers to the export chain. The analysis of price transmission confirmed that export market prices influence domestic prices in related markets. A simple correlation analysis tells a consistent story whereby changes in monthly prices received by Zimbabwe meat exports in the EU over 1998–2000 are associated with changes in cattle prices in the export zone after two to three months (maximum correlation after two months of r = 0.48 for commercial sales and r = 0.47 for communal sales); a little later to communal cattle sales in the other FMD-control zones (initial maximum correlation after three months of r = 0.35, with a second peak occurring after six months); and after four to five months to domestic retail meat markets (maximum correlation after five months of r = 0.59).
The series of econometric tests outlined above were applied to test hypotheses regarding price transmission between specific cattle markets over 1993–2000. The results are reported in detail in appendix 7 in Perry et al. The hypotheses trace back price transmission from changes in the cattle prices offered by the parastatal company responsible for beef exports. The first hypothesis examined the link between the company prices offered to commercial producers and those received by communal producers in auction markets for equivalent grade cattle within the FMD-free zone. These various tests provide clear evidence that increases in export abattoir prices for medium-grade cattle are fully transmitted to communal sellers of this quality of cattle in the long run—taking up to twelve months—whereas increases in abattoir prices for low-grade cattle are only partially transmitted.
The second hypothesis examined whether higher prices for good-quality cattle from communal farmers in the FMD-free zones lead to higher prices for inferior-quality animals in the same zone. In this case, price transmission was found to be complete and to occur within two months. And finally, the last hypothesis explored the link between prices received by communal cattle producers within the FMD-free zone and those in the other FMD-control zones, where cattle cannot be directly sold for export. These prices were found to move closely together, with price transmission complete and occurring within two to four months. Thus, if increased export demand raises prices received by commercial cattle producers, those selling high-quality cattle in communal areas in the FMD-free zone should soon see better prices, followed later by those selling inferior-quality animals in the same zone and by those selling animals in other FMD-control zones.
Conclusion
In this paper, we have outlined extensions of the traditional BCA approach applied to an analysis of the poverty-related impacts of FMD control in Zimbabwe. The traditional BCA confirms that FMD control—including further strengthening of control efforts—adequately responds to the economic efficiency criterion for public investment. However, many developing countries, including Zimbabwe, must now consider an additional poverty-alleviation criterion when evaluating such investments. On this count, the analysis of FMD control yields a less encouraging conclusion. Disaggregating the costs and benefits of FMD control by sector and by population groups with the help of a SAM/CGE model indicates that the majority of the impacts of FMD and the benefits from its control accrue to the commercial sector (commercial cattle producers, beef- processing and related input industries and services) rather than to the communal sector where the majority of cattle are kept. As benefits are passed on as returns to the owners of the underlying factors of production, a similar conclusion emerges: the higher-income segments of the population capture the majority of benefits, with lower-income households enjoying only a third of the income gains achieved. However, the poor do benefit and ultimately policy analysts will need to compare the absolute gains to the poor from investment in FMD with gains to the poor from alternative public investment options.3
Looking more closely at how access to export markets permitted by FMD control might differentially affect prices faced by different cattle producer groups, the analysis confirms that the high prices for Zimbabwean beef exported to the EU should translate into higher prices for cattle sold in both commercial and communal markets, whether inside or outside the FMD-free zone. However, again, the benefits of higher cattle prices are very limited for the majority of poor households because, as additional livelihood analyses reported in Perry et al. highlight, the lowest income households within the communal sector only occasionally keep cattle, and those that do, exhibit very low off-take rates (approximately 3%). Offsetting these modest benefits to poor cattle keepers are the much more extensive income effects on poor consumers; higher cattle prices contribute to increased retail meat prices in domestic markets, reducing the real incomes of all poor households, both rural and urban.
The case of investment in strengthening FMD control in Zimbabwe offers an excellent example of the trade-offs that policymakers face when responding to multiple policy objectives. Improving FMD control undeniably contributes to national economic growth in Zimbabwe, and inevitably will provide benefits to the lower-income groups. However, the magnitude of these benefits for the poor is small relative to the greater share that will be captured by the higher-income groups, and so such investment may ultimately exacerbate the inequitable distribution of wealth in Zimbabwe. Analysts will increasingly be called upon to extend the traditional BCA to evaluate such trade-offs.
