Agro-Terrorism:
A Scenario Based Case Study for Brazil

by Dr. Guilherme G. Santana, National Petroleum Agency, Rio de Janeiro, Brazil and
Dr. Alexandre Rands, Federal University of Pernambuco, Brazil

          The use of chemical or biological weapons (CBW) targeting crops or livestock is not new. States around the world have developed capabilities for this purpose and there are recorded allegations, threats and confirmed incidents of deliberate use of CBW to destroy pre-harvest crops or livestock in all continents of the globe (some well-known “incidents” between 1915 and 2000 can be found at “Chronology of CBW Attacks Targeting Crops & Livestock 1915 — 2000” - http://cns.miis.edu/research/cbw/agchron.htm). These attacks obviously had the intention to cause economic damage. Table 1 (pg 20) displays States that have developed or are suspected of developing biological agents with anti-livestock or anti-crop properties (although zoonotic agents can infect both livestock and humans).

            This paper is extracted from a broad study that has the objective of demonstrating that agro-terrorism is an effective way to cause havoc in the long term, with the potential to jeopardize several sectors of the economy and inflict severe social consequences. There are some clear differences between terrorism acts that target people and those that target crops or livestock. For example, a single terrorist act in a city center making use of explosive has its impact felt immediately, trigging an immediate response (Madrid train attack in 2004 and London 7/7 — 2005, for instance). While an attack with pathogens in a crop is by no means lethal, it does not mean that there is no social impact. The effects are detrimental to related industries down the chain, which over time generate a series of adverse conditions that ranges from pure economic damage to potentially causing severe social disorder. This is particular true for developing nations that have their economy based on agriculture for both receipt generation and to sustain its domestic needs.

            The authors ponder that an agro-terrorism attack in Brazil could have a devastating effect over time, as the impact of such act drifts down the economic chain related to agriculture. For instance, the fact that China temporarily suspended the import of soybeans in 2004, due to an alleged fungus — strongly denied by the Brazilian government, meant a massive cancellation of machinery orders (close to a 100% of orders were cancelled), fertilizer, etc. As a consequence, hundreds of people lost their jobs, with devastating effects to their families. In this sense, with no economic means, people are literally denied access to a series of services and are, at least to a certain degree, marginalized. Another disturbing observation is that in many cases people are also uprooted and forced to move to other places in search of employment. Depending on the economic conditions, a certain number of people can always manage to get back to the economy, but the time and cost of re-employment are high. However, many do not. Therefore, although one may consider that there is no “loss of lives” in an agro-terrorism incident, it has to be acknowledged that there is a perverse element of “destruction of life” along the line. People may suffer severe consequences as a result of turning events that they are subjected to such as the ones mentioned above (loss of employment, reduction on family income, less access to quality services, psychological and sociological implications, “loss of dignity”, and so on).

            The reason why agro-terrorism is and will continue to be used effectively in the future lies on the facts that not only the deploying “technologies” are widely available, but also because the formal response to the perpetrators is relatively constrained. In addition, agro-terrorism pathogens constitute an appealing CBW to terror groups around the world due to their relatively cheap way of production and dissemination (Dando, 2005). Several experts have characterized agro-terrorism as low-tech, high consequence bioterrorism. A number of studies suggest that this kind of bioterrorism would require relatively little in the way of specialist knowledge, technical expertise or technology. Usually the diseases are highly contagious to the intended target (not necessarily to humans), spread rapidly, and would cost an enormous amount of money to eradicate. Of course other costs would also be incurred, as the ones mentioned above in relation to the impact on the agricultural chain, international trade and also would directly impact other counties or regions as they would be devoting considerable resources to protect their crops from the diseases.

            Another issue to be considered in the context of why agro-terrorism that might be attractive to attackers is that the response to an agro-terrorism attack by a state might be considered different from an attack in humans. Since the Geneva Convention (1925), there is a wide common understanding of the dramatic and traumatic consequences of the use of biological or chemical weapons in military or civilian use. A series of other conventions and bilateral agreements resulted from this common repugnance to CBW. However, when used in animal or plants, the sensitivity of the general public is considered reduced. Therefore, the issue is not “sensitive” enough to trigger a strong reaction and official response. It is also important to point out that there are several examples of “incidents” where state and no-state actors allegedly developed and employed anti-plant and anti-animal pathogens to sabotage food supply and transport (see “Chronology of CBW Attacks Targeting Crops & Livestock 1915 — 2000” - http://cns.miis.edu/research/cbw/agchron.htm).

            Although not the object of this study, animal husbandry is also particularly vulnerable to attacks. Brazil has a keen interest in this subject since it is one of the top producers of meat in the world and meat exports plays an important role in total exports. Among other reasons, these animals are vulnerable because they are often from very limited genetic stock so that a large percentage of them could succumb to a single strain of pathogen. Moreover, animals are reared in confined space. In addition, disease outbreaks that affect animal stocks are usually very virulent, such as the outbreak of foot-and-mouth disease in the UK in recent years and the avian influenza in Asia in 2005.

Potential Economic Impact of Agro-terrorism — A Brazilian Perspective

            From an economic perspective, there are two broad categories of impacts that can severely affect the agricultural GDP. The first one is simply generating the idea that there are false attributes to the Brazilian exported outputs. The second one is to try to harm Brazilian crops through introduction of pathogens and/or crop pests in crop areas, damaging local production, possibly for many years. There is no evidence of the occurrence of the second category taking place in Brazil. Nevertheless, there are two recent examples of market events that are very similar to the first kind mentioned above. The first one was, as mentioned previously, China’s suspension of soybean import in 2004, due to an alleged fungus — strongly denied by the Brazilian government, which had devastating effect on the agro-business chain (massive cancellation of machinery orders, fertilizers, etc.). The second example is the restrictions imposed by some European countries to imports of Brazilian beef, after an unconfirmed allegation of an outbreak of foot-and-mouth disease in the Northern area of the country. Both examples above weakened the market power of Brazilian exporters and consequently submitted them to lower exporting prices. The focus of this section will be on their economic impacts considered above, and their ability to generate lower income and welfare in the country, as there are no studies that fill up this gap in the literature.

            Agro-terrorism can have a sizable impact on Brazilian GDP and population welfare. Furthermore, its impact on poverty can be high, perhaps even higher than it is on GDP and welfare. Although most of the Brazilian GDP and employment comes from the service sector, as can be seen in Table 2 (pg 20), there are some estimations that agro-business reaches 33,8% of GDP in the country. Furthermore, in the Northeast and North, the poorest regions in the country, the share of agriculture on total GDP and employment is also higher than for the whole economy (compare the elements in Table 2). Poverty in rural areas, where most of the income comes from the primary sector, is also higher than in urban areas. These data already suggest that the impact of terrorist actions against the primary sector could be high in Brazil.

            Brazilian primary sector is composed of a huge number of commodities. Brazil is a very big country and has very diversified soil and climate conditions. There are since very dry regions in Northeast Caatingas to very humid regions in the Amazon Forest. There are temperate, tropical and equatorial climates within the same country. Therefore, it is a country prompted for agricultural diversification. In fact, its role in the many commodities markets, most of them as an important supplier, indicates that there is a good diversification in Brazilian primary sector. As most of agro-terrorism focuses on one or at most a very limited sample of products, this particular composition of the Brazilian primary sector works similarly to a portfolio diversification on finance theory. Therefore, such diversified composition of agriculture and animal raising output reduces substantially the economic risk of any terrorist attack.

          Nevertheless, although Brazilian agriculture is very diversified, there are some commodities that have a significant share on total output. Table 3 (pg 20) brings the share on total value of production of the major agricultural commodities in the country. Together, the top six products responded for 70% of the agricultural output of the top 25 products in 2003. The three major commodities respond for more than 48,28% of the top 25 products of the country (IBGE, 2004). Therefore, if agro-terrorism hits any of these major crops, they still can have sizable consequences for the Brazilian agriculture.

            Brazil is a very large country, with open borders with all countries in South America. Agriculture is mainly intensive and spreads over thousands of kilometers, making it extremely difficult for surveillance operations. In addition, there are some pressing issues that are currently worrying both producers and government agencies. The number of trading partners both regionally and globally has been increasing significantly in recent year. Trading activities have already introduced a series of species that caused severe environmental and economic damages to the country. One clear example has been the introduction of alien marine species into the Brazilian territory by ballast water from ships. One specie of crab from Asia, introduced in the Northeast of the country, prays on native species and reproduces three times faster than the Brazilian crab. Experts argue that the native specie will be wiped out within ten years. While the native crab is very much appreciated as a delicatessen by Brazilian, the introduced specie is not edible. Among other worrying issues that are contributing to increase the vulnerability of the sector have been the lack of resources devoted to surveillance, very reduced number of inspectors, the lack of training and equipment available to inspectors and the lack of political will to tackle these deficiencies. Brazil does have in place a series of very restricted legislation, but some of them have proved difficult to implement due to the reasons mentioned above. For example, there is hardly any border control with neighboring countries. While one can experience strict controls at major airports, no control at all is applied on some important land border points.

            An exercise was made to estimate the potential impact on the Brazilian agriculture of a possible terrorist attack against our major crops. The impact of an agricultural shock on GDP was estimated to calculate the dynamic loss of GDP as a consequence of falls in agriculture output. A dynamic linear model was used and simulations were made so that all the dynamic impact can be estimated.

The Impact of Agriculture on Brazilian GDP

            Most of the Brazilians lands are suitable for a reasonable number of crops. Technological developments have extended such natural options, as they have adapted crops for previously unsuitable soil. For example, grapes are now produced in the dry lands of Sertão where reasonably warm temperatures prevail, although they are originally from temperate weather. Therefore, if there is an adverse shock to a given crop, the short term consequences will often be higher than long term damages, as other crops may substitute for the formerly produced one within the same areas.

            Nevertheless, if there is shock to a given crop, which could arise either from changes in sales prices in the world market or simply from falls in production due to plagues or other disasters, there is not a full substitution in added values. If this damage is to last longer than a one year crop, the other products that capture the area affected will not have the same economic yields, otherwise they would substitute the previous crop before the disaster. Therefore, such disasters have both short- and long-term effects.

            These long term effects can be bigger than the short term effects, if there is increasing returns to scale, as often is the case in many industries, sometimes even in agriculture production. The tendency for agricultural activities to concentrate spatially is an indication of this. Often this is a consequence of their integration with agro-industry, which normally demands some scale. literature on clusters and clustering indicate there is this tendency to spatial concentrate economic activities (see Ellison and Glaeser, 1994, for an empirical identification of this tendency of cluster formations in economic activities in the United States). Therefore, even with some recovery of crop areas with other commodities, it is possible to have amplified long-term effects of shocks to agricultural production.

            It is also possible to have a decreasing long-term effect, as the crops that substitute the damaged one can have its productivity increased henceforward. In addition to the rise in its own productivity, it is also possible that other activities, directly linked to the new crop, also build their own producing cluster and, consequently, benefit from agglomeration effects. Under such circumstances, there is a phase out of the negative impact in the long term.

            The magnitude of dynamic effects is an empirical matter, which is the object of estimation in this section. Such estimation is necessary to have an idea of the economic impact of a fall in agricultural GDP as a consequence of disasters arising from potential terrorist attacks on the Brazilian agriculture.
Estimation of such an impact relied on a simple vector autoregression that can be represented as:

Where At and Yt are the growth rates of GDP for agriculture and the rest of the economy, respectively. Ct is the co-integrating variable, which introduces the notion that there is a long-term equilibrium between agriculture and the rest of the economy.

            Its introduction makes this equation an error correction mechanism, as analyzed by Salmon (1982) and Engle and Granger (1987). In this equation, aij are coefficients, which are assumed to be constant over the estimation period. Only the third equation in the system has contemporary dependence in this error correction mechanism. _0 is a 3x1 vector of coefficients and _(L) is a 3x3 polynomial in the lag operator. The innovations for each sector are represented by eat, eIt and eSt, with subscript determining the variable to which they refer.

            The data for estimation of equation (1) were from IBGE national account and they are for annual data from 1947 to 2004. Growth rates were simulated through the first difference of natural logarithm. This simplifies the estimation of impacts of shocks to the system in the level of the variables. The growth rate of other sectors was previously regressed against a dummy variable, with one for the period from 1980 to 2004. This was necessary to capture a structural change that led to a permanent change in the growth rate of the Brazilian economy in that period. Statistically, such procedure was necessary to make this series to be stationary.

            The co-integrating variable was estimated as the residual of a regression of the natural logarithm of the agricultural GDP against a constant and the natural logarithm of the GDP of the rest of the economy. This regression was estimated by ordinary least square, as suggested by Engle and Granger (1987). The assumption that there is co-integration between these two parts of the GDP was introduced. The generated series indicates that this is a reasonable assumption.

            Each equation of the system (1) was estimated by ordinary least square. The lag selection relied on the Akaike criteria. The possibility of difference on the number of lags of the two variables in each equation was introduced as a devise to save degrees of freedom. They indicate that there is a good fit of these equations, although the coefficients themselves do not tell much about the problem under analysis.
These equations were used to make two kinds of forecasts for the next twenty years, starting from 2005. The first one simply projected the two variables for the next twenty years based on the coefficients of the model and did not include any shock. The second forecasts these variables, but under the assumption that there is a negative shock of 5% in the agricultural GDP.

            The values of the variables, in level and not in natural logarithm were calculated in these two scenarios. The total GDP was estimated as the sum of the two variables in level, which is its exact value, by construction. The impact of these shocks was calculated from the differences on these two scenarios. Figure 1 (pg 20) brings the dynamic difference in the forecast values for Brazilian GDP under these two different hypotheses. Figure 2 (pg 20) brings the same loss, but as percentage of GDP.

Conclusions

            Two basic conclusions emerge from these figures. Firstly, it is seen that the loss of GDP reaches its maximum level only in the third year after the shock. At its maximum, it reaches 1.2% of total GDP in the country, although it was initially of 0.52% of GDP only. This implies in a loss of R$ 19.6 billion only this year, which should be compared to an initial shock of R$ 9.4 billion (2004 prices). There is a forecast of losses of R$ 10 billion for the 2005 Brazilian agriculture, as a consequence of area reduction and prices fall in the world markets. This indicates that such shock of R$ 9.4 billion is fully reasonable.
Secondly, there is a phase out of the impact after the third year, as the new substituting crops gain efficiency slowly and reduces the loss of generated income. In twenty-three years, the initial shock of 0.52% of GDP has a left impact of only 0.074% of GDP. These same results, on the other hand, indicate that the impact of such a shock is long lasting and consequently its real impact has to be measured dynamically.

            A 5% annual discount rate of the flow of losses presented in figure 1 gives a current value of R$ 120.4 billion, what represents 6.81% of GDP in the year 2004, when forecasts started. This means that a 0.52% negative shock because of any agricultural disaster have an estimated 6.81% negative impact on GDP, if future streams of losses are updated through a 5% discount rate. As this discount rate is real rate, since the values presented are in 2004 prices, it is a quite high discount rate. If a 3% real rate is used the current value of losses reaches R$ 136 billions and 7.7% of 2004 GDP. These data indicates that the total impact of a reasonably small fall in agricultural GDP may have a substantial final impact on net income produced in the country.

REFERENCES

- Engle, R., and C. Granger, “Co-Integration and Error Correction: Representation, Estimation, and Testing”, Econometrica, 55, 251—76, 1987.
- Ellison, G. and E. Glaeser, Geographic Concentration in U.S. Manufacturing Industries: A
Dartboard Approach,” NBER Working Paper, # W4840, August, 1994.
- Dando, M. (2005). Bioterrorism: What is the Real Threat? Science and Technology Report No.
3. www.carnegieendowment.org/static/npp/ST_Report_No _3.pdf (25/07/2005)
- IBGE — www.ibge.gov.br (11/07/2005)
- IPEADATA — www.ipeadata.gov.br (11/07/2005)
- Salmon, Mark H. (1982) Error Correction Mechanisms, Economic Journal, Royal Economic Society, vol. 92(367), 615-29.

• This paper was originally provided to the CBMTS-Industry IV and is reprinted here as an introduction to a CBMTS VI concern. AgroTerrorism is economics and little understood and rarely acted upon until after the fact.