international approaches to the labeling of genetically

CHOICESThe magazine of food, farm and resource issues

A publication of theAmerican AgriculturalEconomics Association

Second Quarter 2003

The United States accounts for two thirds ofbioengineered crops produced globally. Othermajor suppliers include Argentina, Canada, andChina. More than 20% of the global crop acreageof soybeans, corn, cotton, and canola is now bio-tech varieties (International Service for the Acquisi-tion of Agri-biotech Applications, 2003). Inaddition, biotech ingredients and biotech processesare used in producing a wide selection of food andbeverage products such as meat, poultry, cheese,milk, and beer.

The United States first exported geneticallymodified (GM) food to Europe in 1996. It wastomato puree from California, and it was voluntar-ily labeled as genetically engineered. The productwas a big hit with consumers in Britain, because itwas cheaper than conventional tomato puree.However, when GM soybeans were imported intoEurope later that year, there was a huge backlashfrom environmental groups such as Greenpeace.The European Union (EU) was then quick tointroduce mandatory labeling, which took hold in1997, for GM foods. The US government viewedthe European mandatory labeling policy as a tradebarrier, and so began another US-EU agriculturaltrade dispute that is still ongoing.

This transatlantic trade dispute has spread tomany other countries. Governments around theworld are developing GM labeling requirements,and they are finding themselves caught between theUS and the EU approaches. Consumers are con-fused about what exactly GM foods are andwhether these foods are harmful. Scientists havedetermined that precommercialization assessmentprocedures of bioengineered food ensure that GMfood is nutritionally equivalent and as safe as con-

ventional food (e.g., the World Health Organiza-tion, 2000). However, in Europe consumers do notnecessarily trust scientists, especially after they weretold in the mid-1990s that humans could not getmad cow disease.

The GM labeling issue is not just about science.Rather, the politicians and environmental groups inEurope and elsewhere say GM labeling is aboutconsumer choice and consumer rights, and is noteven a health issue. The Europeans are clearly tak-ing a precautionary approach. Alternatively, theUnited States, Canada, and Japan are using science-based risk assessment procedures. The purpose ofthis article is to discuss international approaches toGM labeling and to shed some light on why theapproaches vary so much across nations.

Labeling IssuesAny decision on labeling of GM food presentsmajor challenges for policy makers. The most fun-damental problem relates to DNA detection,because the measurement of GM material becomesdifficult or impossible if the GM crop is highly pro-cessed. For example, products such as soybean oilor meat produced from GM feedstuffs do not con-tain any evident GM protein. In addition, biotech-nology that is used in certain food and beveragemanufacturing processes cannot easily be detectedin the final product. Most cheese, for example, ismade with genetically engineered enzymes.

Proponents of mandatory GM food labelingbelieve that consumers have the right to knowwhether they are eating GM foods. Opponents saythat such a label implies a nonexistent food safetyrisk. Trying to label something that is not detect-able invites fraud, and the fraud cannot even be

International Approaches to the Labeling of Genetically Modified FoodsBy Colin A. Carter and Guillaume P. Gruere

Second Quarter 2003 CHOICES 1

2

detected. Mandatory labeling would result inunnecessary marketing costs due to segregation andidentity preservation requirements, with no appar-ent offsetting consumer benefit. In addition, man-datory labeling requirements could inhibit furtherdevelopment of GM technology. Within the EU,there are concerns over a slowdown in biotech cropresearch and the long-run international competi-tiveness of EU agriculture (Mitchener, 2002).

Even with mandatory labeling, standards areinconsistent and consumers are not necessarily pro-vided with greater choice. In Japan and the EU(where GM labeling is mandatory) it is virtuallyimpossible to find food products on the shelflabeled as containing GM ingredients. Therefore,the approach taken by Japan and the EU is notreally giving consumers a choice. Furthermore, asubstantial amount of GM food eaten in the EUand Japan does not have to be labeled. These prod-ucts include cheese, soya sauce, vegetable oils,baked goods, and numerous manufactured foods.

The strict European regulations are also servingto delay the introduction of GM crops in poorcountries, such as vitamin A enriched rice (the so-called golden rice). Cotton remains the only GMcrop planted in South Africa or India, because GMcotton does not have labeling requirements any-where. Zambia’s refusal of GM corn as food aid inthe face of massive starvation caught the world’sattention and resulted in criticism of the EU.

International Approaches to GM LabelingInternationally, the Codex Alimentarius Commis-sion (http://www.codexalimentarius.net), an inter-national standards-setting body for food, has aCommittee on Food Labeling. Since 1990, Codexhas sought to develop guidelines for labeling bio-tech foods. So far, however, there is no agreementon the international standards. In all likelihood, afinal Codex standard on the labeling of biotechfoods will not occur for many years.

The approaches taken in different countriestowards GM food labeling differ greatly (Sheldon,2002) as shown on Table 1. The EU has very strictGM labeling guidelines. In contrast are the UnitedStates, Argentina, and Canada, the three big pro-ducers, whose governments do not believe in man-datory labeling. Japan, South Korea, China, andother countries have approaches that are betweenthe EU and the United States on this issue.

The US government’s lack of support for man-datory labeling reflects the scientific evidence bythe Food and Drug Administration (FDA) thatGM foods are nutritionally equivalent to non-GMfoods. As long as the food is safe from impurities,labeling is not needed.

This is not to say that a labeling debate does notexist in the United States. Antibiotech groups inOregon put the GM labeling issue on the state’s2002 ballot (Measure 27). If it had passed, Measure27 would have required the labeling of any productcontaining GM ingredients, or processed with GMmaterial or biotechnology. The antibiotech groupsplan a similar initiative for California in 2004.

In 2002, the EU’s Parliament approved newGM labeling proposals. If implemented, they couldjeopardize a large share of US food exports toEurope. These new guidelines were debated andmodified in the EU Commission and in the EUCouncil at the end of 2002. The European Parlia-ment is expected to deliberate over the proposal inthe first half of 2003, with final adoption by theend of the year.

The proposed EU regulations would require thelabeling of foods whose ingredients contain 0.9%or more of GM DNA or protein. The current toler-ance level is 1%. The new EU regulations would

Table 1. Sample of international guidelines for labeling GM foods

Labeling scheme

% threshold for unintended GM

material

Are some biotech foods and

processes exempt?

Canada Voluntary 5%c N/A

United States

Voluntary N/A N/A

Argentina Voluntary N/A N/A

Australia & New Zealand

Mandatory 1% Yes

European Union

Mandatory 0.9%a Yes

Japan Mandatory 5%b Yes

S. Korea Mandatory 3% b Yes

Indonesia Mandatory 5%c Yes

Notes: N/A—not applicable.Proposed threshold in the EU, lowered from 1%.Top 3 ingredients in Japan and top 5 ingredients in S. Korea.Not yet operational.Source: Personal interviews and various Attaché Reports from the USDA Foreign Agricultural Service (http://www.fas.usda.gov/itp/biotech/countries.html).

CHOICES Second Quarter 2003

require labeling of food and feed products contain-ing even nondetectable GM material. Importantly,only authorized GM material would be allowed infood and feed sold in the EU. Compared to the USor Japan, only a small number of GM crops areauthorized in the EU.

Partly in response to these new EU labelingproposals, the US government is threatening to filea World Trade Organization (WTO) trade actionagainst the EU for restraining trade. The tradeaction would cover the EU moratorium on anynew GM crops in Europe, in place since 1998, andthe European labeling regulations. The costs of thelabeling regulations in the EU are largely borne byUS exporters. The 0.9% tolerance level will becostly to implement for food processors. It seemsexcessively low for a label with no intended foodsafety purposes. Besides, the absence of labeled GMproducts at the retail level makes this labeling pol-icy a “political” moratorium.

Australia and New Zealand jointly adoptedmandatory labeling with a 1% threshold for theunintended presence of GM product, but vegetableoils, food additives, and food processing aids (suchas enzymes used in cheese and brewing) are exemptfrom labeling requirements.

Japan’s labeling regulations are much more rea-sonable than those in the EU. The Japanese govern-ment requires mandatory labeling when GMmaterial is present in the top three raw ingredientsand accounts for 5% or more of the total weight. Italso admits the presence of non-GM labels at thesame tolerance level, if produced with identity pres-ervation. So tofu can be made from non-GM soy-beans and be labeled as such or else it must belabeled as containing GM material. Exemptions toJapan’s labeling requirements include feedstuffs,alcoholic beverages, and processed foods such assoya sauce, corn flakes, and other vegetable oils.South Korea’s regulations are similar to Japan’s,except the tolerance level is 3% of the top fiveingredients. In the EU, the threshold applies toeach ingredient.

China leads the world in public biotech cropresearch (Huang, Rozelle, Pray, & Wang, 2002).Genetically modified crops in the field trial stageinclude rice, wheat, corn, soybeans, potatoes, cab-bage, and tobacco. Genetically modified cottonaccounts for about 30% of China’s cotton acreage.China has not yet announced a firm position on

GM labeling, but it has recently proposed restric-tions on GM crop imports. Outside China, this isviewed as a trade barrier that limits soybeanimports from the United States. China’s positiontowards biotechnology in agriculture appears to beheavily influenced by EU policy.

After 14 years of testing GM crops in experi-mental fields, the Mexican government has draftedlegislation authorizing the planting of GM cropsfrom March 2003. In February 2001, the MexicanSenate approved a draft regulation requiring thelabeling of GM food and feed, but the MexicanChamber of Deputies, under pressure from the US,did not ratify this proposition.

Why These Differences?So why do we observe the wide difference inapproaches to GM labeling across countries? Thereare several possible explanations. The EU andJapan have experienced domestic food scares inrecent years. Consumers in these countries do notbelieve scientists and politicians who say that GMfood is safe. Political pressure from environmentalgroups plays on this fear and raises unscientific con-cerns about GM food safety (Bernauer & Meins,2001).

Environmental and consumer groups supportmandatory labels for the sake of consumer choice.But mandatory labels (such as “this product con-tains GM ingredients”) are still perceived by con-sumers as warning signals. In this situation,processors prefer to avoid labeling, and thus do notexceed the threshold level. For some pressuregroups, labeling appears as a first step towards anoutright ban of GM products.

On a larger scale, labeling affects internationaltrade. Consequently, the European policy affectsthe choices of other agricultural exporting coun-tries. Australian GM policies were partiallydesigned to fit the labeling requirement for exportsto the EU or Japan. Eastern European countriesand Russia have probably decided to follow theEU’s 1% threshold labeling requirement for tradereasons. Those eastern European countries plan-ning to join the EU really have no choice.

The labeling of GM food may be soon disputedat the WTO. The United States has waited tolaunch a WTO dispute against the EU, becausethey feared it could drive a deeper wedge between


4

EU and US GM policy. The US administrationrecently decided to further delay the WTO actionto seek EU support for the Iraq war. US lawyers areconfident they would win the dispute (“US hints,”2003). To reduce the negative effect of an interna-tional dispute, the US needs allies in developingcountries. A dispute could actually help to generatea stable agreement on GM labels. According tointernational observers, a WTO jurisprudence onGM labeling regulations could be the only way toachieve an agreement on labeling based on Processand Production Methods (PPM) standards.

ReferencesBernauer, T., and Meins, E. (2001). Scientific revo-

lution meets policy and the market: Explaining cross-national differences in agricultural biotech-nology regulation (CIES Discussion Paper 0144). University of Adelaide.

Huang J., Rozelle S., Pray C., and Q. Wang. (2002). Plant Biotechnology in China. Science, 295, 674.

International Service for the Acquisition of Agri-biotech Applications. (2003). 2002 global GM crop area continues to grow for the sixth consecu-tive year at a sustained rate of more than 10%. Available on the World Wide Web: http://www.isaaa.org.

Mitchener, B. (2002, September 23). Europe has no appetite for modified food. The Wall Street Journal, p. B3.

Sheldon, I. (2002). Regulation of biotechnology: Will we ever ‘freely’ trade GMOs? European Review of Agricultural Economics, 29(1), 155-176.

US hints it will sue EU over altered crops. Com-plaint about food ban would go to WTO. (2003, January 9). The Washington Post.

World Health Organization. (2002). Safety aspects of genetically modified foods of plant origin (Report of a Joint FAO/WHO Expert Consul-tation on Foods Derived from Biotechnology). Geneva, Switzerland: WHO.

Colin A. Carter is Professor of Agricultural andResource Economics and a member of the GianniniFoundation, University of California, Davis. Guil-laume P. Gruere is a Ph.D. candidate in Agricul-tural and Resource Economics, University ofCalifornia, Davis. The authors acknowledge finan-cial support from the University of California Agri-cultural Issues Center as a part of the AgriculturalMarketing Resource Center project. [email protected].




Second Quarter 2003

With the events of September 11, 2001, theanthrax mail attacks, and the establishment of theOffice of Homeland Security to develop and coor-dinate the implementation of a comprehensivenational strategy to secure the United States fromattacks, many individuals have become aware of thethreat of biological weapons directed towards peo-ple. However, few realize how vulnerable the USagricultural infrastructure is, and has been, to pestsand disease outbreaks resulting from accidental ordeliberate introductions, and few are aware of theconstant battle that is being waged to prevent ormitigate the spread of invasive species. Over thepast 200 years or so, more than 50,000 foreignplant and animal species have become establishedin the United States. About one in seven hasbecome invasive,1 with damage and control costsestimated at more than $138 billon each year(United States Department of Agriculture Animaland Plant Health Inspection Service [APHIS],2001). The problem of invasive species has intensi-fied in the last few years, making it a serious chal-lenge to globalized trade. “Animals, plants, andmicrobes can now migrate across the planet to newhomes with unprecedented ease” (“New flora,”2000, p.118).

The problems of biological invasives and thedecision-making framework established to preventtheir introduction and spread have traditionally

been the domain of the biological scientific com-munity. However, as present management systemshave become overwhelmed by the increase in theintroduction and spread of invasives, the scientificcommunity is now calling for input by economicsand other social science disciplines to answer ques-tions and carry out strategic actions to address theproblems.

The economic dimension of the problem ofinvasive species is growing from at least two per-spectives. First, economics is central to the cause ofbiological invasiveness, and the consequences ofpest incursions go far beyond direct damages orcontrol costs. Most cases of invasiveness can belinked to the intended or unintended consequencesof economic activities (Perrings et al., 2002). Con-sequently, economic applications are essential tounderstand the problem and provide more accurateand comprehensive assessments of the benefits andcosts of control alternatives to increase the effective-ness and efficiency of publicly funded programs.

Second, modeling the economic and tradeimpacts of technical trade barriers is becomingmore important. Common among such barriers are

1. An "invasive species" is defined as a species that is nonnative (or alien) to the ecosystem under consideration and whose introduction causes or is likely to cause economic or envi-ronmental harm or harm to human health (Executive Order 13112;http://www.inva-sivespecies.gov). Invasive species can be plants, animals, and other organisms (e.g., microbes).

Economic Dimensions of Invasive SpeciesBy Edward A. Evans


6

those dealing with trade restrictions that can beimposed by a country in an attempt to prevententry of invasive species. Such measures are withinthe rights of a country and often can be justified onthe grounds of economic and social prosperity.However, they can also impose unnecessary socialcosts, thwart commercial opportunities, and reducecompetition and economic growth. The challengeis how best to incorporate economics in sanitaryand phytosanitary policy analyses to ensure that thebenefits of the measures enacted exceed their costs.

The purpose of this article is to highlightaspects of the economic dimensions of the problemof invasive species.

Evidence of Increased Incidence of Invasives and Cost ImplicationsThe increased spread of invasive species reflectsrapid globalization and trade liberalization. Thesedevelopments have spawned greater long-distancehitchhiking by invasive species of pests and dis-eases, especially in the trading of live animals, andhorticultural and raw animal products. The USAnimal and Plant Health Inspection Service(APHIS) has cited a dramatic increase in the inci-dence of invasive pests and diseases in the UnitedStates. Specifically, the study noted increased out-breaks of exotic fruit fly infestations in Californiaand Florida, entry of the Asian longhorn beetle intoNew York and Illinois, the introduction of theAsian gypsy moth in North Carolina and Oregon,and citrus canker infestations in Florida (APHIS,2001).

Invasive species can cause considerable damagesand costs for eradication and control for societies;however, the full extent of the costs of damagescaused by pest incursions has only recently receivedgreater appreciation. Invasive species can harm agri-cultural systems and native plants and animals, par-ticularly endemic species, because their naturalpredators and parasites in their native land are usu-ally not present in the new environment. Thus, aninvasive species that is not a pest in its native landcould cause significant damage in a new environ-ment. In the extreme, such damage could lead tothe loss of biodiversity. For example, the Asianlonghorn beetles that were first discovered in theUnited States (in New York in 1996 and Chicagoin 1998) are expected to damage millions of acres

of hardwood trees throughout US forests and sub-urban landscapes. The states of Illinois and NewYork City and local governments have alreadyinvested more than $30 million to eradicate thispest and protect 6.7 million trees in the infestedregions. Since 1996, the state of Florida has spentin excess of $300 million dollars trying to eradicatecitrus canker.

Invasive species can also adversely affect impor-tant environmental service flows such as croppingsystems, livestock grazing, and recreational uses.Water systems can be affected when pests clog riv-ers, irrigation systems, and shorelines. In addition,invasive species can have negative impacts on eco-logical services provided by one resource for otherresources or an entire ecological system (Evans,Spreen, & Knapp, 2002).

What Economics Has to Contribute towards Resolving the Problem of InvasivenessEconomics has traditionally been concerned withdecision making, particularly with what decisionsare made rather than how they are made—althoughto some extent the discipline has started to embracethe latter. The discipline has developed a set of ana-lytical capabilities that can aid decision makers inarriving at a set of rational and consistent decisions.The analytical capabilities, as they pertain to theproblem of invasives, include rationale decisionmaking over a range of pest threats and manage-ment interventions; monetary valuations; cost-ben-efit analysis as a tool to evaluate public interventionstrategies; allocation of scarce resources; and formalconsideration of risk and uncertainty. The disci-pline has also developed several methodologies toassess the value of nonmarketed environmental andhealth effects. With increasing demand for trans-parency in decision making, due to commitmentsto international agreements and pressure from vari-ous interest groups, effective and convincing com-munication is essential to implement desiredstrategies. When such communications are basedon sound economic analysis, efficiency in bargain-ing can be greatly enhanced.

Assessing the Economic Consequences of Invasive Pests and DiseasesConsiderable effort is being devoted to assessingthe full economic impacts of invasive pests and dis-


eases. The goal is to develop effective managementprograms to help prevent, control, or mitigate suchinvasions. Previously, the focus was on identifyingthe most cost-effective means of treatment for of anoutbreak. Now the emphasis is on the benefits andcosts of treatments to determine how best to man-age the particular pest or disease.

Assessing the impact is challenging and impre-cise. First, as noted earlier, the full range of eco-nomic costs of biological invasions goes beyond theimmediate impacts on the affected agricultural pro-ducers. Often included are secondary and tertiaryeffects, such as shifts in consumer demands,changes in the relative prices of inputs, loss ofimportant biodiversity, and other natural resourceand environmental amenities. The range of eco-nomic impacts can be broadly classified into twocategories: direct and indirect impacts (Bigsby &Whyte, 2001). The direct impacts reflect the effectsof the particular pest or disease on the host,whereas the indirect impacts are nonhost-specific.The latter would be the general effects that are cre-ated by the presence of a pest but not specific to thepest-host dynamics that could affect public healthissues (such as compromising key ecosystem func-tions); general market effects (including possiblechanges in consumers’ attitudes toward a givenproduct); research requirements; market accessproblem; and impacts on tourism and other sectorsof an economy.

Alternatively, six types of impacts can be identi-fied: (a) production; (b) price and market effects;(c) trade; (d) food security and nutrition; (e)human health and the environment; and (f ) finan-cial costs impacts (Food and Agricultural Organiza-tion, 2001).

Production Impacts. These are considered the mostdirect economic impacts associated with the host,resulting in the loss or reduced efficiency of agricul-tural production (such as yield decline). Eventhough such impacts may be relatively easy to iden-tify, they can be difficult to measure. Disease canhave lasting effects on the host in ways that are notalways obvious. In livestock, for example, therecould be delays in reproduction resulting in feweroffspring. Pesticides applied to treat a given pestcould pollute soil and surface water. Furthermore,distinguishing the impacts of the pests from otherimpacts (such as climate) could be difficult.

Price and Market Impacts. Outbreaks of pests and dis-eases can directly affect the quantities of a com-modity demanded or supplied. The exact impacton the market and the duration of the impactdepend on several factors, including the nature ofthe pests and diseases, market size, and the relativeelasticities of demand and supply. In cases whereconsumer health is involved, as in the recent out-break of bovine spongiform encephalopahy (BSE),consumer perceptions about an implicated productand the ability of a country to produce safe foodafter an outbreak or illness are usually slow torecover and can have a lasting influence on fooddemand and global trade. In addition, a range ofsecondary effects may result from the multipliereffect.

Trade Impacts. The introduction and/or spread ofinvasive species can have major trade implicationsthat could outweigh direct production losses. Suchtrade impacts will depend on a number of factors,including the policy response of trading partners tonews about outbreaks, the importance of tradedcommodities, the extent of the damage, and thedemand and supply elasticities. Important are theprospects of losing competitive advantage in anexport market and possibly the premium from sup-plying disease-free products. Such concerns are real,because unaffected countries will either prohibitentry of the commodities from the affected countryor establish a set of precautionary measures. Ineither case, the competitive trade advantage couldbe lost.

Food Security and Nutrition Impacts. The extent towhich invasive pests and diseases either reduce the


8

domestic supply of foods directly or restrict a coun-try’s international trade could harm its food secu-rity, especially for developing countries.

Human Health and the Environment Impacts. Assessingthe human health and environmental impacts ofinvasive pests and diseases are difficult, because inmany cases the impacts are not fully understood.Available evidence does suggest, however, that theincidence of invasive food-borne diseases is growingand that their health and socio-economic impactsare increasingly being felt in both developed anddeveloping countries.

Financial Costs Impacts. Measures taken at the indi-vidual, collective, and international levels to con-trol, eradicate, or mitigate invasive pests anddiseases may have budgetary implications. Suchcosts could include the costs of inspections, moni-toring, prevention, and response.

Estimating these economic impacts requires aconsiderable amount of biological and nonbiologi-cal information that involves considerable time andexpense. Most studies have easily calculatedimpacts such as costs of control, eradication, andprevention and the expected loss in productivity ofthe enterprise. However, such an approach is short-sighted, because in several cases the indirect effectsarising from (say) the trade impacts could easilyoutweigh production loss impacts. A recent USGeneral Accounting Office (GAO) report com-mented on the problem in its observation that:

The scope of existing studies on the economicimpact of invasive species in the United Statesrange from narrow to comprehensive, and mostare of limited use for guiding decision makersformulating federal policies on prevention andcontrol. Narrowly focused estimates includeanalyses of past damages that are limited to acertain commercial activities such as agricul-tural crop production and simple accounting ofthe money spent to combat a particular invasivespecies. These estimates typically do not exam-ine economic damage done to natural ecosys-tems, the expected costs and benefits ofalternative control measures, or the impact ofpossible invasions by other species in thefuture.... In general the more comprehensivethe approach used to assess the economicimpacts of invasive species, the greater itspotential usefulness to decision makers for iden-tifying potential invasive species, prioritizingtheir economic threat, and allocating resources

to minimize overall damages (US GAO, 2002,p. 3).

Valuing the nonmarket impacts can be chal-lenging. In this regard, economists are employingsuch tools as dynamic optimization and ex antesimulation analyses to assist decision makers(Evans, Spreen, & Knapp, 2002). Use is also beingmade of methods such as “contingent valuation”and “willingness-to-pay to obtain or avoid similarbenefits or losses.”

A more general measurement problem is theunavailability of data, especially when there is nodisease history. Complications also may arise fromthe uncertainty of the scientific evidence about theprobability of entry and establishment, rate ofspread, and the extent of damage. Closer collabora-tion between economists and biological scientists,as well as the increased availability of computersoftware programs (such as the Excel @RISK pro-gram that combines dynamic simulation proce-dures with probability distribution), allow analyststo combine actual, but limited, data with theoreti-cal modeling in determining potential impacts.

Modeling the Impacts of Sanitary and Phytosanitary RegulationsThe need for a government to protect its citizensand environment against imported externalities(such as invasive pests and diseases) is embraced bythe WTO Agreement,2 which promotes increasedtrade among countries. When legitimate externali-ties or other market failures are addressed throughtechnical trade barriers, for instance, in a commod-ity with the potential to introduce disruptive pestsand diseases, they can safeguard national welfare.However, when such measures are imposed to iso-late domestic producers from international compe-tition, they are welfare-decreasing. This dual natureof the SPS measures – providing externality-basedprotection versus economic-based protection –adds to the importance of comprehensive economicanalysis of the issues of invasive pests and diseases.

2. A separate agreement governing sanitary and phytosanitary issues, Agreement on the Appli-cation of Sanitary and Pytosanitary Measures, was negotiated during the 1986-1994 Uru-guay Round multilateral trade negotiations.


As a consequence, many economists are busytrying to develop a framework for assessing thetrade and welfare implications of trading a particu-lar commodity under different managementoptions when there is the potential for introductionof an invasive pest or disease. Developing such aframework, however, is far easier in theory than inpractice. Although not insurmountable, theinvolvement of externalities in the form ofunwanted pests and diseases, and specifically therisks and uncertainty associated with them, compli-cate the standard economic policy analysis.

Concluding RemarksThe invasive species problem is posing a seriouschallenge in an era of increased globalization andtrade liberalization. The problem has as much to dowith economics as with ecology. Any solutionsadvanced must be firmly grounded in both scienceand economics. Our economic discipline possessesthe capability of valuing various market and non-market impacts and provides a means for assessingimportant tradeoffs among various managementalternatives, which can improve greatly the deci-sion-making process for managing such risks. Inaddition, it can improve the transparency of thedecision-making process by providing justifica-tions for the measures implemented. The true valueof economics should therefore not be seen solely inthe precision of the numbers generated, albeit thisis important, but the extent to which the disciplineaids decision makers to formulate consistent andrationale decisions.

ReferencesBigsby, H. and C. Whyte. (2001). Quantifying

phytosanitary barriers to trade. In N.Hooker

and E. Murano (Eds.), Interdisciplinary Food Safety Research. New York, NY: CRC Press.

Evans, E., Spreen, T., and Knapp, J. (2002, November). Economic issues of invasive pests and diseases and food safety. The 2nd Interna-tional Agricultural Trade and Policy Conference, Gainesville, FL.

Food and Agricultural Organization. (2001). The state of food and agriculture 2001. Rome, Italy. Available on the World Wide Web: http://www.fao.org/docrep/003/x9800e/x9800e14.htm.

New flora and fauna for old. (2000, December 23). The Economist, 118-119.

Perrings, C., Williamson, M., Barbier, E., Delfino, D., Dalmazzone, S., Shogren, J., Simmons, P., and Watkinson, A. (2002). Biological invasion risks and the public good: An economic per-spective. Conservation Ecology 6(1), 1. Available on the World Wide Web: http://www.con-secol.org/vol6/iss1/art1.

United States Department of Agriculture Animal and Plant Health Inspection Service. (2001). APHIS strategic plan 2000-2005. Available on the World Wide Web: http://www.aphis.usda.gov/oa/aphissp/spextfaclhtml.

United States General Accounting Office. (2002). Invasive species: Clearer focus and greater commit-ment needed to effectively manage the problem (Report GAO-02-1). Washington, DC: GAO.

Edward A. Evans ([email protected]) is anassistant professor and assistant director of the Inter-national Trade and Policy Center, located in theFood and Resource Economics Department, Univer-sity of Florida.


10



Second Quarter 2003

The potential impact of China’s future meat tradeon world markets, and in turn on US exports, hasexperienced considerable debate. Several articleshave appeared in CHOICES over the past fiveyears (Hayes, 1999; Hertel, Nin, Rae, & Ehui,2000) as well as in China (Guoda & Zhang, 2001;Qiao, 2001; Lu & Yang, 2002). Some analystsbelieve that China will become an important netimporter of livestock products (Beghin & Fabiosa,2001; Organisation for Economic Cooperation andDevelopment, 2000), while others argue that shewill become a major net exporter (Delgado, Roseg-rant, Steinfeld, Ehui, & Courbois, 1999). A thirdset of estimates stresses the wide range of possibleoutcomes depending on macro- and microeco-nomic uncertainty (Nin, Hertel, Foster, & Rae, inpress).

A clearer understanding of China’s imports andexports of livestock products is needed to addresssuccessfully the future of China’s trade. In this arti-cle, we will show the difficulties in assessing China’smeat trade, including the definition of nationalboundaries, the valuation of trade flows, and therole of smuggling and unreported imports intoChina. Although many studies assert that China iscurrently a net importer of poultry and pork prod-ucts, consideration of these difficulties yields a verydifferent picture.

Which China?Any discussion of China’s trade in livestock prod-ucts must determine: Which China? The Food andAgriculture Organization of the United NationsStatistical Database (FAOSTAT, http://apps.fao.org/default-c.htm) has options for GreaterChina, Macau, Hong Kong, and China+ (whichrefers to China plus Taiwan). Hong Kong andMacau are part of China, but have separate customs

territories with unique economiccharacteristics. They are largelytrading hubs with zero tariffs,which thrive on buying and sell-ing products as well as financialintermediation. Including them inthe trade data for China makesthe mainland economy look moreopen than it really is.

Taiwan is more complicated, because mainlandChina considers it a province, while the Taiwaneseoperate as an independent country. This politicaldispute is presumably why the FAO has not sepa-rated Taiwan from China in their data (eventhough separate customs data are available). YetTaiwan is a more highly developed economy, withper capita income roughly 13 times that of main-land China. In recent years, Taiwan’s growth ratehas also been much lower than mainland China’s.Separating the two when examining trade patternsand making future projections makes sense. Wefocus on mainland China, with data obtained fromChina’s Ministry of Agriculture through the Chi-nese language version of FAOSTAT.

Which Units: Tons or Dollars?The next key point in assessing China’s meat tradeis whether trade should be measured in physicalvolume (metric tons) or in value (e.g., $US mil-lions). Units of measurement can make a big differ-ence. Based on volume, China is a net importer ofpoultry meat (Figure 1). The net import status iswidely reported in the agricultural trade literature.Based on value of trade, however, China is a sub-stantial net exporter of poultry meat. The value ofexports is more than double import values. Why isthis? China tends to import low price products

The Puzzling State of China’s Meat TradeBy Yijun Han and Thomas Hertel


12

such as chicken feet and wings, while exportinghigher priced products such as deboned chickenbreast.

So, is China currently a net importer orexporter of poultry meat? This point is fundamen-tal to forecasting future trends. Value data are morerelevant. After all, the US trade deficit is notreported in terms of tons. It makes little sense tocombine diverse products without weighting themby their economic value. Adding tons of chickenbreast together with chicken feet makes no moresense than adding tons of coal and microchips. Theconclusion is that China is currently a substantialnet exporter of poultry meat products, with a tradesurplus of $US 530.8 million.

Which Products?In addition to fresh, chilled and frozen poultrymeat, trade in live animals, viscera, and cannedpoultry products is substantial. Although few ofChina’s imports fall into this category, the countryexported nearly $400 million of canned poultrymeat in 2001, a 100% increase over 1999. Whenall poultry products are combined (Figure 2),China is a substantial net exporter. China’s tradesurplus for all poultry products in 2001 amountedto nearly $1 billion. This dampens some of the

enthusiasm for China’s poultry market, based onlyon tons of meat traded.

Figure 2 suggests that imports and exports havebeen increasing strongly. This “intra-industry”trade characterizes an industry with considerableproduct differentiation. Producing the product forwhich imports are rising can be beneficial. How-ever, by-products (e.g., chicken feet) do little tobolster the total demand for poultry production.

Poultry meat has garnered the attention of agri-cultural trade experts, but pig meat remains thelargest consumption item in China. Figure 3reports the volume and value of China’s trade of pigmeat. When trade is measured in tons, China was anet importer of pig meat in 1999 and 2000. Likepoultry, however, China imports lower price itemsand exports higher price items, yielding a netexport status in value terms.

Over time, exports of pig meat have fallen whileimports have risen, especially since 1999. Thissame pattern is evident from Figure 4, whichincludes other pork products (live animals, skins,viscera, sausages, canned meat, etc.). Thus export-ing pork to China may be more promising, eventhough the volume of poultry imports into Chinaremains about four times as large as for pork.

Figure 1. China’s trade in poultry meat (FAOSTAT, 2003).

0

200000

400000

600000

800000

1000000

1200000

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 year

ton,

usd1

000

im quantity im value ex quantity ex value


Which Trade? The Role of Illegal TradeAny discussion of recent rends in China’s meattrade must consider smuggling and underinvoicingof imports. Substantial smuggling in meat products(Yu & Ma, 2000) occurred prior to 1999, when the

costs were low due to lax enforcement. The benefitswere high, as tariff rates were roughly 45% (andhigher in earlier years). In 1999, government crack-downs greatly increased the costs of smuggling.Meanwhile in November 1998, as part of its prepa-

Figure 2. China’s trade in all poultry products (FAOSTAT, 2003).

Figure 3. China’s trade in pig meat (FAOSTAT, 2003).

0

200000

400000

600000

800000

1000000

1200000

1400000

1600000

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 year

ton,

usd1

000


0

50000

100000

150000

200000

250000

300000

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 year

ton,

usd1

000



14

ration for entry into the World Trade Organization(WTO), China sharply reduced tariffs on meatproducts to 20%. Both of these changes madesmuggling into China less profitable, and led tosharp increases in reported imports from 1998 to1999. Further indirect evidence of the significanceof the crackdown on smuggling is provided by thefact that these changes coincided with a doublingof the price of chicken feet in China. This was themost popular item among smugglers (Wang,2002).

Distinguishing between true import increasesand reduced smuggling is difficult. One approachto measure underreporting of imports is to com-pare China’s bilateral import data with the exportdata of its trading partners. This method shouldpick up any systematic discrepancies in reportedimports and exports. Mark Gehlhar (2002), of theUSDA Economic Research Service, has formalizedthis approach, although differences in data classifi-cations hamper comparisons with the FAOSTATdata. Gehlhar’s estimates for nonruminant meatimports into China show only a 1.1% increasefrom 1998-1999, after correcting for underreport-ing of China’s imports. Thus, much of the previ-ously “missing” imports were reported in theexporting country. They simply evaded the Chinese

import authorities in order to avoid paying importduties. These findings suggest that the sharpincrease in meat trade from 1998-99 is misleadingand likely misinterpreted by some observers as thestart of an import boom.

Which Trade Partners?Who is trading with China? The US dominates thismarket, supplying 93% of the poultry products and75% of the pig products (China Customs, 2002;calculated by imports and exports data of poultryproducts in volume from January to September2002). About 70% of the US poultry exports toChina consist of chicken wings and chicken feet(i.e., byproducts) to meet the preferences of Chi-nese consumers. China’s major export partners forpoultry are Japan, Hong Kong, Saudi Arabia, andRussia, among others. Most exports to Hong Kongare live animals, whereas other countries mostlyreceived processed poultry products. Russia alsotakes two thirds of China’s pork exports — mostlyfrozen products. As with poultry, most of the HongKong imports from mainland China are live hogs.

Figure 4. China’s trade in total pig products (FAOSTAT, 2003).

0

100000

200000

300000

400000

500000

600000

700000

800000

1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000year

ton,

usd

1000



ConclusionEach measurement issue addressed above has likelyled to excessive optimism about China’s importpotential for meat products. First, using the GreaterChina, or China+ data from the FAOSTAT web-site, which includes Taiwan and perhaps HongKong and Macao, will erroneously suggest a netimporter position.

Second, most projections of China’s futuretrade are based on tons of product. CurrentlyChina is a net importer in tons, but a net exporterin dollars because it imports low-value by-products,such as chicken feet, and exports higher value pro-cessed meat products.

Third, most discussions of meat trade focus onfresh, chilled, and frozen meat, ignoring trade inlive animals and processed products. When the lat-ter products are factored in, China’s trade surplus inpoultry products increases to more than $1 billion.Finally, the huge jump in China’s reported meatimports from 1998 to 1999 disappears when previ-ous underreporting of imports is considered.

References Beghin, J., and Fabiosa, J. (2001). China’s accession

to the WTO: Effects on US pork and poultry. Iowa Ag Review, Summer 2001

China Customs. (2002). The imports and exports reports by monthly.1-9.

Delgado, C., Rosegrant, M., Steinfeld, H., Ehui, S., and Courbois, C. (1999). Livestock to 2020: The next food revolution (Food, Agriculture, and the Environment Discussion Paper No. 28). Washington, DC: International Food Policy Research Institute.

FAOSTAT. (2003). Chinese language version. Available on the World Wide Web: http://apps.fao.org/default-c.htm.

Gehlhar, M. (1996). Reconciling bilateral trade data for use in GTAP (GTAP Technical Paper No. 10). West Lafayette, IN: Purdue University Center for Global Trade Analysis. Available on the World Wide Web: http://www.gtap.age-con.purdue.edu/resources/tech_papers.asp.

Guoda, K., and Zhang, L. (2001). Analysis of com-parative advantages of China livestock products exports. China Rural Economy, 7.

Hayes, D. (1999). China’s role in world livestock and feed-grain markets. Choices, 1999(1).

Hertel, T.W., Nin, A., Rae, A., and Ehui, S. (2000). China: Future customer or competitor in live-stock markets? Choices, 2000(3).

Lu, K., and Yang, W. (2002). How to expand the exports of Chinese pork? China Rural Economy, 1.

Nin, A., Hertel, T.W., Foster, K., and Rae, A. (in press). Productivity growth, catching-up and uncertainty in China’s meat trade. Agricultural Economics.

Organisation for Economic Cooperation and Development. (2000). OECD agricultural out-look: 2000-2005. Paris: OECD.

Qiao, J. (2001). Analysis of the international com-petitiveness of China’s meat products. China Rural Economy, 7.

Wang, F. (2002, July 5). The trade of chicken feet runs into deadlock. Southern Weekend.

Yijun Han is an Agricultural Economist with theResearch Center for Rural Economics, Ministry ofAgriculture, P.R. China. Thomas Hertel ([email protected]) is Professor and Director, Centerfor Global Trade Analysis, Department of Agricul-tural Economics, Purdue University.


16



Second Quarter 2003

Calls for mandatory food traceability are makingnews in policy discussions ranging over homelandsecurity, country-of-origin labeling, Mad Cow dis-ease, and genetically engineered foods. A frequentunderlying assumption is that unlike Europe, theUnited States does not have food traceability. Herewe argue that although the United States does notmandate system-wide traceability, firms have anumber of motives for establishing traceability sys-tems; as a result, private-sector traceability systemsin the United States are extensive. The breadth,depth, and precision of private traceability systemsvary depending on the attributes of interest andeach firm’s traceability costs and benefits. Manda-tory traceability that fails to allow for variationacross firms may impose unnecessary costs on firmsalready operating efficient traceability systems.

Why Firms Have Traceability Systems Food suppliers have three motives for establishingproduct tracing systems: (a) to improve supply-sidemanagement; (b) to differentiate and market foodswith subtle or undetectable quality attributes; and(c) to facilitate traceback for food safety and qual-ity. Firms establish traceability systems to achieveone or more of these objectives. As a result, the pri-vate sector has a significant capacity for tracing.

Traceability for Supply-Side ManagementFirms have a strong incentive to establish product-tracing systems to manage production flows andtrack retail activity. Traceability systems help firmsreduce expensive overstocks, coordinate orders andshipments, and manage inventories. Electronicaccounting systems for tracking inventory, pur-chases, production, and sales are an integral part ofdoing business in the United States.

In addition to private systems, US firms mayalso use an industry-standard coding system. Thevast majority of packaged food products bear barcodes, as do a growing number of bulk foods, likebagged apples and oranges. Bar codes contain aseries of numbers reflecting type of product andmanufacturer (the UPC code) and a series of num-bers assigned by the manufacturer to nonstandardproduction or distribution details. While bar codeswere originally intended to facilitate tracking ofretail sales and food consumption trends and pat-terns, the codes are now used to track numerousproduct attributes.

The success of the original UPC system hascombined with technological advances and e-mar-keting to spur the development of integrated sys-tems that code, track, and manage wholesale andretail transactions. In some cases, buyers managethese systems to monitor supply flow. For example,a few big retailers such as Walmart and Target havecreated proprietary supply-chain information sys-tems that their suppliers must adopt. In other cases,firms establish systems to link suppliers and buyers.For example, UCCnet, which is a subsidiary of theUniform Code Council, has developed an inte-grated system to standardize and automate infor-mation systems across a supply chain.

Sophisticated tracking systems are not confinedto packaged products. The food industry has devel-oped a number of complex coding systems to man-age the flow of raw agricultural inputs and outputs.Vegetable and fruit farmers routinely tag their pro-duce crates to record location and date of harvest.This information aids in inventory management atthe packinghouse and in tracking shipments.Ranchers have been using electronic identificationeartags and corresponding data collection cards totrack information on an animal’s lineage, vaccina-

Traceability in the US Food Supply: Dead End or Superhighway?By Elise Golan, Barry Krissoff, Fred Kuchler, Ken Nelson, Greg Price, and Linda Calvin


18

tion records, and other health data. This allows forefficiency gains through sorting of individual cattlein feed yards, recording preconditioning and otherhealth regimes, and facilitating disease surveillanceand monitoring.

Traceability for Food Safety and Quality ControlProduct tracing systems are essential for good safetyand quality control systems. Traceability systemshelp firms isolate the source and extent of safety orquality control problems. The better and more pre-cise the tracing system, the faster a producer canidentify and resolve food safety or quality prob-lems.

A firm’s traceability system not only helps mini-mize potential damages for individual firms—italso helps minimize damages to the whole industryand to upstream and downstream industries. Con-taminated meat sales and foodborne illness out-breaks damage the reputation and sales of thewhole meat industry. Because of these spillovers,some product tracing systems are supported byindustry groups or buyers. For example, the Cali-fornia cantaloupe industry has incorporated trace-ability requirements in their marketing in order tomonitor food safety practices. Firms offering liabil-ity insurance may also require traceability systemsto ensure that insured firms have minimized risks.

If the failure of a firm’s quality control systemresults in sales of unsafe or defective products, atraceability system helps to track product distribu-tion and reduce the size and cost of recall. Most ifnot all voluntary recalls listed on USDA’s FoodSafety and Inspection Service website refer consum-ers to coded information on products’ packaging toidentify the recalled items.

Traceability for product safety and quality isbecoming a necessary part of doing business. Goodproduct tracing systems help minimize the produc-tion and distribution of unsafe or poor qualityproducts, thereby minimizing the potential for badpublicity, liability, and recalls.

Traceability to Differentiate and Market Foods with Credence AttributesAlthough the US food market successfully mass-produces homogenous commodities such as grainsand meats, it also offers products tailored to thetastes and preferences of segments of the consumerpopulation. Food producers differentiate products

for micromarkets over such attributes as taste, tex-ture, nutritional content, cultivation techniques,and origin. Consumers easily detect some qualityinnovations—green ketchup is hard to miss. How-ever, other differences involve credence attributes—characteristics that consumers cannot discern evenafter consuming the product. Consumers cannottaste or otherwise distinguish between food prod-ucts containing genetically engineered (GE) ingre-dients and those made with non-GE ingredients.

Credence attributes can be content or processattributes:

Content attributes affect the physical propertiesof a product, although they can be difficult for con-sumers to perceive. For example, consumers areunable to determine the amount of isoflavones in aglass of soymilk or the amount of calcium in a glassof enriched orange juice by drinking these bever-ages.

Process attributes do not affect final productcontent but refer to characteristics of the produc-tion process. Process attributes include country-of-origin, organic, free-range, dolphin-safe, shade-grown, earth-friendly, and fair trade. In general,


r coffee)

neither consumers nor specialized testing equip-ment can discern process attributes. No test con-ducted on a can of tuna, for example, could tellwhether the tuna was caught using dolphin-safetechnologies.

Traceability is an indispensable part of any mar-ket for process credence attributes—or contentattributes that are difficult or costly to measure.The only way to verify the existence of theseattributes is through a bookkeeping record thatestablishes their creation and preservation. Forexample, tuna caught with dolphin-safe nets canonly be distinguished from tuna caught using othermethods through the bookkeeping system that tiesthe dolphin-safe tuna to the observer on the boatfrom which the tuna was caught. Without trace-ability as evidence of value, no viable market couldexist for dolphin-safe tuna, fair trade coffee, non-biotech corn oil, or any other process credenceattribute.

Food producers have developed sophisticatedsystems for tracking and establishing value for cre-dence attributes. For example, farmers have begunusing Global Positioning Systems to create infor-mation to trace crops back to the precise locationwithin a field to determine cultivation practicessuch as pesticide use. For ranchers, the chain ofdocumentation generated by electronic eartagsenables them to more easily sell their cattle at aprice that reflects underlying quality.

Balancing Costs and Benefits in Private Traceability SystemsPrivate traceability systems are extensive, but whatkind of traceability do they entail? Are they ade-quate for tracking production from farm to fork?From seed to finished product? The characteristicsof a firm’s traceability system depends on the firm’sobjectives and the costs and benefits of traceability.Firms balance costs and benefits to determine thebreadth, depth, and precision of their individualtraceability systems.

Breadth Breadth is the amount of information the traceabil-ity system records. There is a lot to know about thefood we eat, and firms must decide which informa-tion is of value. A recordkeeping system catalogingall of a food’s attributes would be enormous and

unnecessary. The beans for a cup of coffee couldcome from any number of countries, be grown withnumerous pesticides or just a few, be grown onhuge corporate organic farms or small family-runconventional farms, be harvested by children or bymachines, be stored in hygienic or pest-infestedfacilities, or be decaffeinated using a chemical sol-vent or hot water. Even the most meticulous pro-ducer would not find it worthwhile to collect andmaintain information on all coffee attributes.

Given the huge number of attributes that coulddescribe any food product, full traceability is anunreachable goal. A traceability system that usedDNA to track beef back to information on an ani-mal’s lineage, vaccination records, and feedingregime would be incomplete if pasturage hours orplaytime were the attributes of interest. Only ahandful of attributes—as determined by supplymanagement requirements, consumer preferences,and food safety considerations—warrants theexpense of traceability recordkeeping. Firms bal-ance the costs and benefits of attribute informationto determine the efficient breadth of the producttracing system.

Depth The depth of a traceability system is how far backor forward the system tracks. Most businesses haveone-up, one-back traceability. Firms must knowwho their suppliers are if they pay their bills andthey must know who their buyers are if they cashtheir checks. Most businesses in the United States,and certainly all large businesses, maintain elec-tronic bookkeeping systems to track their accounts.The bulk of the US food supply system is therefore

Figure 1. The depth of a traceability system (fodepends on the attributes of interest.


20

monitored with electronic one-up, one-back trace-ability systems.

Whether product tracing goes beyond buyersand sellers depends on the objective of the sys-tem—and the attributes of interest to the produceror consumer. A traceability system for decaffeinatedcoffee would only extend back to the processingstage (Figure 1). A traceability system for fair tradecoffee would only extend to information on priceand terms of trade between coffee growers and pro-cessors. A traceability system for fair wage wouldextend to harvest; for shade grown, to cultivation;and for non-GE, to bean or seed. For food safety,depth of the traceability system depends on wherehazards and remedies can enter the food produc-tion chain.

PrecisionPrecision reflects the degree of assurance withwhich the tracing system can pinpoint a particularfood product’s movement. A precise traceabilitysystem would trace an apple to its orchard withhigh assurance, while a less precise system wouldonly trace a crate of apples to two or three orchardswith lower assurance.

The first decision a firm makes with respect toprecision involves the acceptable error rate. In ashipment of white corn, how many yellow kernelsare acceptable? In a shipment of non-GE soybeans,how many GE beans are acceptable? Error-ratespecifications (driven by quality requirements) willdetermine the strictness of the segregation systemwith which the traceability system is paired. Lowtolerances for yellow kernels in a shipment of whitecorn or for GE content in a shipment of non-GEsoybeans will require strict segregation systems andaccurate bookkeeping systems.

The second decision a firm makes with respectto precision is regarding the unit of analysis—con-tainer, truck, crate, day of production, or shift?Firms must determine the most efficient trackingunit for their objectives. Firms that choose a largeunit (such as feedlot or grain silo) for tracking pur-poses will have poor precision in isolating safety orquality problems. A smaller unit of analysis (such asindividual cow or crate) will allow greater precision.

Precision in traceability—as reflected in theaccuracy of the segregation system and the size of

unit of analysis—is more valuable the higher thelikelihood and cost of safety or quality breaches. Ifthe likelihood and cost of recall were high, a manu-facturer would quickly see the benefit in accuratelyreducing the size of the standard recall lot. How-ever, precision comes at a cost. In particular, theerror tolerance rate strongly affects the costs of seg-regation and traceability. The benefits of strictidentity preservation and product tracing will out-weigh the costs for some firms but not for others.The accuracy of the traceability system and level ofsegregation will vary widely depending on themotivations driving their development.

Firms are Building a Traceability Superhighway Technological advances are pushing improvementsin supply-side management and quality control sys-tems throughout the US food system. Electronicaccounting and traceability systems are standardiz-ing information on product attributes and synchro-nizing product tracking across the food system.Firms balance traceability costs and benefits so thatthe breadth, depth, and precision of private systemsreflect technological limits and consumer prefer-ences. In some cases, however, firms may not sup-ply the socially optimal amount of traceability, aswhen private and social traceability costs and bene-fits differ. In these cases, mandatory traceabilitymay be a policy option.

Paradoxically, the widespread voluntary adop-tion of traceability in the United States mayincrease firms’ cost of compliance with mandatorytraceability systems. Efficient traceability by firmsresults in systems with differing levels of breath,depth, and precision. Because government require-ments rarely allow for variation in process and out-come, firms may be required to make changes totheir traceability systems that do not improve effi-ciency. Mandatory traceability that allows for varia-tion or targets specific traceability gaps could bemore efficient than system-wide requirements.

The authors are economists with the United StatesDepartment of Agriculture Economic Research Ser-vice. The views expressed here are those of theauthors and do not necessarily reflect official USDApositions. Elise Golan ([email protected])




Second Quarter 2003

During the 1990s, an increasing share of UShouseholds became digitally connected to the vastamount of information available on the Internet.United States Bureau of Census data indicate thatbetween August 2000 and September 2001 alone,the percentage of all households with Internet con-nections increased dramatically (Figure 1). At thesame time, disparities in Internet access and useemerged among various segments of the popula-tion. Recent studies show that Whites have greateraccess to and use of the Internet than Blacks andnon-Hispanics show greater use than Hispanics(Compaine, 2001; National Telecommunicationsand Information Administration and EconomicsStatistics Administration [NTIA], 2002). Moreeducated and higher income individuals also showgreater Internet use (NTIA, 2002). Althoughhousehold use has increased regardless of location,the gap in use between metropolitan and nonmet-ropolitan areas has remained at about 13%.1 Thesedifferences in Internet use reflect the “digitaldivide.”

Characterizing the DivideConcerns exist that the metropolitan-nonmetropol-itan digital divide may widen regional differences in

household well-being (Drabenstott, 2001). As aresult, the digital divide has been the focus of sev-eral policy initiatives. The Rural Access Authorityin North Carolina, for example, was created to pro-vide local dial-up Internet access from every tele-phone exchange in the state. Other states (such asWashington and Virginia) have provided grants torural areas to promote high-speed Internet access.Other initiatives have supported infrastructureinvestments in low-density regions. These policieswill only stimulate home Internet use in nonmetro-politan areas if the current digital divide stems fromdifferences in infrastructure for Internet connectiv-ity between the two areas.

Digital infrastructure between metropolitanand nonmetropolitan areas can differ considerably.Greenman (2000) reports that less than one per-cent of towns with fewer than 10,000 persons havedigital subscriber line (DSL) or cable modem ser-vices. On the other hand, 86% of cities with morethan 100,000 persons have digital subscriber lines,and 72% of cities with more than 250,000 personshave cable modem services. The 2001 Current Pop-ulation Survey (CPS) data also suggest that infra-structure differences may be contributing to thedigital divide. For example, 9.9% of nonmetropoli-tan household Internet users had high-speed con-nections, compared to 20.9% of metropolitan users(Figure 2). 2 On the other hand, nonmetropolitanusers are only slightly more reliant on long-distancecarriers to obtain Internet access, suggesting mini-

1. This paper uses 1993 US Census designations of nonmetropolitan and metropolitan coun-ties to compare differences in home Internet use. Metropolitan counties generally have pop-ulations greater than 100,000 (75,000 in New England) or a town or city of at least 50,000. Nonmetropolitan counties are those counties not classified as metropolitan.

2. High-speed access is often more important for business applications than home applications.

Bridging the Metropolitan-Nonmetropolitan Digital Divide in Home Internet UseBy Brian E. Whitacre and Bradford F. Mills


22

mal differences in carrying costs between the twoareas.

Differences in education, income, and otherhousehold attributes, rather than technology, maydrive differences in metropolitan and nonmetropol-itan region use. In this case, closing the dividewould involve broader efforts to increase educationand income in nonmetropolitan areas. Ensuringschool children equal access to digital technology isalso essential to avoid passing digital knowledgegaps to the next generation. We next explore howdifferences in education, income, and other house-hold attributes influence the digital divide in homeInternet use.

The Role of Income and EducationNonmetropolitan areas trail metropolitan areas inboth household income and household head educa-tion (Figure 3). Income and education levels arealso higher in Internet-using households (Figure 4).Thus, income and education levels are likely toexplain much of the digital divide.

Our econometric analysis indicates that 63% ofthe digital divide is due to area differences in house-hold characteristics (Figure 5). The remaining 37%is linked to differences in the propensities of house-holds with otherwise similar characteristics to usethe Internet at home in metropolitan and nonmet-ropolitan areas.

Given the strong local composition of manyonline communities (Horrigan, 2001), a house-hold’s value of Internet access tends to increase asthe share of other connected households (and busi-nesses) in the same region increases. Addingregional density of home Internet use along withhousehold characteristics explains 91.2% of thedigital divide. Thus, household attributes andregional rates of household Internet use account formost of the digital divide.

Regional differences in Internet use coulddecline over time because of normal adoption pat-terns. The current findings, however, about theimportance of differences in income and educationon Internet use suggest greater persistence in thegap.

Policy ImplicationsPolicy options to address the metropolitan-non-metropolitan digital divide must be linked to thenarrowing of income and educational disparitiesrather than focusing solely on digital infrastructure.Programs to increase general access for underservedpopulations also are important. Public support forsuch initiatives is currently weak. The two majorfederal program initiatives (The TechnologyOpportunities Program and Community Technol-ogy Centers) to foster Internet use by underservedpopulations are losing their funding (Harris andAssociates, 2002).

Thanks to public investments, Internet accessin the nation’s schools is far more uniform acrossrace and income groups than use at home (New-burger, 2001; NTIA 2002). Similarly, the CPS dataindicate that the current rate of Internet use by

Figure 1. Household internet use for metropolitanand nonmetropolitan areas in 2000 and 2001 (Cur-rent Population Survey Internet and Computer UseSupplements, 2000-2001).

Figure 2. High speed and long distance access formetropolitan and nonmetropolitan areas (CurrentPopulation Survey Internet and Computer Use Sup-plements, 2000-2001)

0

10

20

30

40

50

60

Total U.S. Metro Non-Metro

Perc

ent

2000

2001

0

5

10

15

20

25

High SpeedAccess

LongDistanceAccess

Perc

ent

Metro

Non-Metro


Figure 3. Household head education and household income levels in metropolitan and nonmetropolitan areas(Current Population Survey Internet and Computer Use Supplements, 2000-2001).

Figure 4. Household head education and household income levels for home internet users and nonusers (Cur-rent Population Survey Internet and Computer Use Supplements, 2000-2001).

0

20

40

60

80

100

Metro Non-Metro Metro Non-Metro

Percent

Less than High

School

Less than High

School

High School

Diploma

High School

Diploma

Some College

Some College

College Degree or

More

College Degree or

More

Under $20K

$20K - $40K

$20K - $40K

$40K - $60K

Over $60K

$40K – $60K

Under $20K

Over $60K

Education Income

0

20

40

60

80

100

Metro Non-Metro Metro Non-Metro

Perc

ent

Less than High

School

Less than High

School

High School

Diploma

High School

Diploma

Some College

Some College

College Degree or

More

College Degree or

More

Under $20K

$20K - $40K

$20K - $40K

$40K - $60K

Over $60K

$40K – $60K

Under $20K

Over $60K

Education Income

0

20

40

60

80

100

Internet No Internet Internet No Internet

Perc

ent

Less than H.S.

Less than High

School

High School

Diploma

High School

DiplomaSome

College

Some College

College Degree or

More

College Degree or

More

Under $20K

$20K - $40K

$20K - $40K

$40K - $60K

Over $60K$40K – $60K

Under $20K

Over $60K

Education Income

0

20

40

60

80

100

Internet No Internet Internet No Internet

Perc

ent

Less than H.S.

Less than High

School

High School

Diploma

High School

DiplomaSome

College

Some College

College Degree or

More

College Degree or

More

Under $20K

$20K - $40K

$20K - $40K

$40K - $60K

Over $60K$40K – $60K

Under $20K

Over $60K

Education Income


24

children at school is higher in nonmetropolitanareas than in metropolitan areas. School access isconsidered essential to avoid transferring the digitaldivide to future generations. A similar commitmentto reduce disparities within the adult population isa policy choice.

For More Information:Compaine, B.M. (Ed.). (2001). The digital divide:

Facing a crisis or creating a myth? London: MIT Press.

Drabenstott, M. (2001). New policies for a new rural America. International Regional Science Review, 24(1), 3-15.

Greenman, C. (2000, May 18). Life in the slow lane: Rural residents are frustrated by sluggish web Access and a scarcity of local information online. New York Times, p. D1.

Harris and Associaties. (2002). Bringing a nation online: The importance of federal leadership. Washington DC: Leslie Harris and Associates.

Horrigan, J.B. (2001). Online communities: Net-works that nurture long-distance relationships and local ties. Pew Internet & American Life Project. Available on the World Wide Web: http://pewinternet.org.

National Telecommunications and Information Administration and Economics Statistics Administration. (2000). Falling through the net: Towards digital inclusion. Washington, DC: United States Department of Commerce.

National Telecommunications and Information Administration and Economics Statistics Administration. (2002). How Americans are expanding their use of the internet. Washington, DC: United States Department of Commerce.

Newberger, E.C. (2001). Home computers and inter-net use in the United States (Special Study P23-207). Washington, DC: United States Depart-ment of Commerce.

United States Department of Commerce Bureau of Census. (2001). Current population survey, Sep-tember 2001: Internet and computer use supple-ment [computer file]. Washington, DC: United States Department of Commerce.

Brian E. Whitacre is a Graduate Research Assistantand Bradford F. Mills ([email protected]) is an Asso-ciate Professor, respectively, of the Department ofAgricultural and Applied Economics at VirginiaTech University.

Figure 5. Decomposing the Digital Divide.

Decomposition of Digital Divide

0

10

20

30

40

50

60

70

80

90

100

Probability of Internet Use in 2001

0

10

20

30

40

50

60

70

80

90

100

Non – Metro (42%)

Metro (55%)

Household Location (37%)

Household Characteristics

(63%)

Decomposition of Digital Divide

0

10

20

30

40

50

60

70

80

90

100

Probability of Internet Use in 2001

0

10

20

30

40

50

60

70

80

90

100

Non – Metro (42%)

Metro (55%)

Household Location (37%)

Household Characteristics

(63%)

Perc

ent

Perc

ent




Second Quarter 2003

ng put multi-ts sales measure ich an itional e provid-, but the sually

A proposal being floated by a local economic devel-opment organization predicts that a new food pro-cessing plant will add 800 jobs and $100 million tothe local economy. Jobs and income will be gener-ated at the plant and in real estate, health care, andagriculture. The plant is also touted as having alocal economic multiplier of $6. As an industrypolicy analyst, you are asked to evaluate the pro-posal. To do so, you will need to understand whatthese values really mean, the assumptions underly-ing their estimation, and whether they are realistic.Your reply is important, because public and privatesector leaders, the general public, and even profes-sionals can misinterpret economic impact and mul-tiplier analysis. Worse yet, impact studies can beused to exaggerate the benefits of policies or pro-posals in some cases and their costs in others.

Outlined here are issues that should be consid-ered in conducting and interpreting impact andmultiplier analysis (see Checklist). These issuesshould influence choice of models and interpreta-tion of results by policy analysts. The emphasis ison the regional (multistate, state, substate, or local)level, where such studies are normally conducted.

Multiplier and Impact AnalysisAlthough a variety of methods can be used to gen-erate economic multipliers, the focus here is oninput-output (I-O) models as the most populartool for such analysis. This popularity has beenengendered by the growth of ready-made I-O mod-eling systems such as IMPLAN, where a basicknowledge of personal computers is sufficient forgenerating models, multipliers, and impacts.

Export base theory underlies the use of eco-nomic multiplier and impact analysis. It springsfrom the idea that a region must earn income tosurvive by producing a good or service that the out-

side world will purchase. The use of I-O models hascaused this idea to be extended to the sales gener-ated by any industry—whether export oriented ornot. The income injected into an economy byexports has a multiplier effect, as it is respentlocally. The level of respending is based on howmuch local businesses and consumers buy fromlocal businesses.

Impact analysis looks at the effects of a positiveor negative change in economic activity. Impactanalysis is based on economic multipliers, whichaccount for the total effect across the entire econ-omy of the event under study. For example, impactanalysis is often used to estimate the effects of anew local industry on jobs and incomes in all partsof the economy. It is also used to estimate policy orinvestment impacts and the total contribution ofan industry to an economy.

I-O Model Construction and AssumptionsI-O models examine the market flow of productsbetween industries, sales by industries to house-holds and other final users, and industry use of fac-tors of production (labor and capital). Such modelscan be very detailed, containing several hundredindustries.

Policy Uses of Economic Multiplier and Impact AnalysisBy David W. Hughes

Backward Linkages and I-O MultipliersSeveral different types of multipliers are generated usiregional I-O models. For a given local industry, the outplier measures the combined effect of a $1 change in ion the output of all local industries. All I-O multipliers the strength of backward linkages or the degree to whincrease in activity by a given local industry causes addpurchases from other local industries and local resourcers. The same relationships are used in impact analysisinitial change in output is much larger than $1 and is uspread across several local industries.


26

What are SA Social Athe econoexplicitly aernment wflows. A CGeconomy. product, cdata is usequently, elprecise mo

Multipliers are generated in I-O models basedon the key assumption of fixed-proportion produc-tion functions, where input use moves in lockstepfashion with production. For example, if a poultryprocessor doubles production, its use of each inputalso doubles. This production function is based ona completely elastic supply. That is, shifts indemand only result in changes in output, with nochanges in real (inflation removed) prices. Suchsupply curves are based on the assumption that allunits of a given input are equal in quality, and thereare no barriers to firms entering or exiting markets.

Similar assumptions are also made in I-O mod-els of regional economies. For example, if a localpoultry processor doubles production, its currentuse of regional inputs will also double. Householdspending (and implicitly population) is alsoassumed to move in a lockstep fashion with eco-nomic activity.

Limitations of Multiplier and Impact Analysis Including Possible Solutions Several issues can influence the interpretation ofresults in multiplier-based studies. These issues maylead policy analysts to do additional analysis or usealternative models. Such issues include investmentor project feasibility, employment impacts, effectson current residents, considerations about capital,impacts on local government, and accountingstance. Concerns about feasibility and profitabilitycan be particularly important in interpreting modelresults.

Profitability and Other Feasibility IssuesImpact analysis does not by itself address severalissues related to feasibility in project (investment)analysis—the most important being profitability. Alocal industry with a sizeable multiplier is not nec-

essarily profitable, and multipliers do not accountfor this lack of profitability. If profitability is nega-tive, then further development of the industry isnot feasible. For a particular study, it is preferableto include profitability analysis as a separate com-ponent. For multipliers in general, policy analystsshould be aware that by themselves multipliers donot speak to profitability.

Second, resource constraints are often ignoredin multiplier impact analysis. For example, a localcommunity may be considering a new paper mill,but the regional timber supply does not meet millneeds. In this case, the investment may be infeasi-ble. A solution is to expand the study explicitly toinclude resource impacts. Basic I-O model text-books contain examples of approaches to environ-mental and energy issues, which can be used inlooking at resource pressures.

Another issue related to feasibility is possibleimpacts on prices. In a growing economy, supplypressures may lead to big price increases oftenignored in multiplier-based studies. Rapid growthcould lead to upward pressure on local wages, forc-ing businesses to cut back on employment. Suchcutbacks would mute the expansion estimated withmultipliers based on the no change in pricesassumption. When price pressures are significant,policy analysts should be aware that more complexmodels are available (such as a computable generalequilibrium model, which allows for changes inprices). Alternatively, they can interpret the quan-tity change (such as a change in output) as an upperbound on the expected actual change. Sound judg-ment is required in deciding if an I-O model yieldsappropriate answers in such situations.

Employment ImpactsThe major concern in most impact studies is theeffect on local employment. Under a growth sce-nario, job impacts are generally based on theassumption that new jobs go to new residents,which leads to population growth. This in turnleads to increased consumer spending on localproducts. However, any number of factors couldbreak this chain of events. For example, the newjobs could go to current residents (the unemployed,job upgraders, or current out-commuters). Newjobs could also go to new in-commuters. In either

AMs and CGEs?ccounting Matrix (SAM) provides a detailed picture of my but in a more complete fashion than an I-O by ccounting for all market and nonmarket (such as gov-elfare payments to households) income and resource E also provides a complete and detailed picture of an

However, prices are free to change and thus impact onsumption, and trade relationships. Hence, more d in a complex set of nonlinear equations. Conse-iminating the fixed-price assumption may lead to less del estimates.


case, the increase in local population and spendingwould be less than expected.

For a decline impact scenario, job losses couldbe less than predicted. For example, those losingtheir jobs could commute to work elsewhere, withno loss in population and little decline in localspending.

An integrated I-O labor market model is onepossible solution to this limitation. These modelsuse other sources of information to help determinethe distribution of job changes between new andcurrent residents. If this approach is viewed as tooresource intensive, policy analysts should be awarethat the projected change in employment is anupper bound on the actual change with the caveatthat model predictions could be wide of the mark.

Other considerations should also be raised inevaluating employment impacts. For example:Under a growth scenario, will new jobs be perma-nent and full-time? Another consideration is thetype of occupations that will be generated (a keydeterminant of desirability in many rural areas). Inaddition, local workers may be unqualified for thenew positions, and in-migrants or in-commuterswill be employed instead.

Some of these issues can be examined with an I-O model. For example, wage estimates—an impor-tant part of desirability—are imbedded in suchmodels. An industry occupation table (matrix),showing the distribution of occupations by indus-tries, can be used to predict the types of generatedoccupations. The table translates employment esti-mates for each industry into a group of occupa-tions. For example, ten new jobs in a givenagricultural industry directly lead to one new farmmanagement and nine new farm laborer jobs. Cou-pled with information about the local labor force,the matrix can show if local individuals can fill thenew occupations. Other questions such as the per-manence of employment impacts can be evaluatedbased on knowledge of the economy and the issueat hand.

Financial and Physical Capital ConsiderationsAnother set of issues involves financial and physicalcapital. In evaluating a growth scenario, the level ofthe new capital investment and the residency ofinvestors (local or otherwise) may be important. Aregional SAM, which extends I-O by tracing all

market income flows, could be used to shed lighton these issues.

In evaluating a decline in economic activity,one should consider if the facilities involved (physi-cal capital) could have an alternative use. For exam-ple, the impacts of the closure of military bases areimportant concerns for local economies. Economicimpact analysis could indicate a major loss in localjobs and income when the base closes. However,the base is now available for other uses that maybenefit the local economy. The proper response isto also evaluate the likelihood of success and eco-nomic impact of such alternative uses.

Impact on Current Residents and ActivityThe effect on current residents and economic activ-ity is another set of issues often ignored in multi-plier-based studies looking at local economicgrowth. The value of the current housing stockmay increase, especially if the economy is alreadygrowing and the anticipated impact is large. If pop-ulation growth cannot be easily absorbed, surges ineconomic activity may create a tax burden for cur-rent residents by increasing property values.

Population growth can also place additionalpressures on other industries that should be consid-ered. In particular, local farms may close because ofsubdivision growth and other population-relatedimpacts. Environmental degradation from a newindustry could also have negative consequences forexisting industries.

Local Government ImpactsThe effect on government services and revenues isanother important consideration, especially at thelocal level. A new industry may place pressures onlocally provided public services. An impact study ofa proposed casino touted the projected increase inlocal tax yield, but ignored possible increases in thecost of public services (such as additional policeprotection). If population growth occurs, local gov-ernment may have to finance new roads, schools,and other infrastructure. Likewise, residents mayhave to endure crowding costs (such as increasedtraffic) if infrastructure development does not keepup with population growth.

An integrated public service I-O model canhelp shed light on these issues. Such models predictchanges in employment and population and then


28

indicate how changes in both lead to increases inthe cost of publicly provided services and govern-ment revenue. If using such a model is not possible,then the tax analysis should be eliminated from thestudy or at least tempered by indicating thatchanges in the cost of government services are notestimated.

Accounting StanceImproper accounting stance (comparison of applesand oranges) also occurs in impact studies. Forexample, a statement sometimes made concerningthe statewide impact of an institution of higherlearning is the following: “$3 of output are gener-ated in the state economy for every $1 that wereceive in state funding”. The comparison is one ofapples and oranges because output measures grosssales while state government revenue has some typeof income as its source.

Part of the solution to such accounting stanceissues is not to compare apples and oranges. Educa-tion and proper interpretation concerning the dif-ferent measures of economic activity estimated withI-O models should also help eliminate this prob-lem.

SummaryMultiplier and impact analysis indicate the level ofeconomic activity that may be generated by a givenindustry or event. Although useful, limitations ofsuch work should always be discussed. Policy ana-lysts should consider additional efforts to shed fur-ther light on critical issues when appropriate.

For Further Information:Hefner, F., and Morgan, J.M. (2001, April). The

economic impact of a university: A critical review of the issues. Paper presented at the annual meet-ing of the Southern Regional Science Associa-tion, Austin, TX.

Hughes, D.W. (1997). Military personnel and industry. In G.A. Goreham (Ed.), Encyclopedia of Rural America: The Land and the People (pp. 472-474). Santa Barbara, CA: ABC-CLIO Publishing Inc.

Miller, R., and Blair, P. (1985). Input-output analy-sis: Foundations and extensions. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Table 1. What are some reasonable values for local economy multipliers?

County employment size class Average multiplier Probable range

1,000-2,999 1.7 1.5-1.9

3,000-4,999 1.8 1.5-2.0

5,000-9,999 1.9 1.6-2.1

10,000-19,999 2.0 1.8-2.2

20,000-49,999 2.2 2.0-2.4

50,000 and over 2.2 2.0-2.5

All things else equal, multipliers will tend to be higher where: (a) the community is larger with a more diverse economy; (b) the com-munity is a substantial distance from competitive retail/service centers; and (c) the per capita income is low. Any output multiplier larger than 2.5 should be especially examined!Source: Mulkey, 1978.

Table 2. Multiplier and impact analysis checklist (concerns and solutions).

Concerns Solutions

Feasibility:

Profitability Include profitability in analysis;warn that profitability is not addressed

Resource constraint Include resource impacts in analysis;warn that resource availability is assumed

Price impacts Use price change model;interpret quantity changes as upper bounds

Employment impacts:

Who gets job Use integrated I-O labor market model;interpret local resident job changes as upper bound

Type of job Include industry-occupation analysis;use knowledge of the situation to interpret results

Capital considerations:

Financial capital Use model that traces capital flows;use knowledge of the situation to interpret results

Physical capital Determine likelihood of success of alternative uses

Current versus new residents:

Housing stock Use knowledge of situation to interpret results

Pressure on other industries

Include resource impacts;include declines in other industries in analysis

Local government impacts

Use integrated public service model;omit tax analysis;indicate public service impacts not considered

Accounting stance Do not compare apples and oranges;properly interpret different variables


Mulkey, D. (1978). The Journal of the Community Development Society, 9(2), 85-93.

Sumners, G.F. (1976). Small towns beware: Indus-try can be costly. Planning, 42(4), 20-21.

David W. Hughes ([email protected]) is anAssociate Professor and Extension Economist, WestVirginia University.


30

international approaches to the labeling of genetically

Documents